Non-redundant role of Shc in Erk activation by cytoskeletal reorganization.

We have shown previously that cytoskeletal reorganization (CSR) induced by pharmacological reagents such as colchicine or cytochalasins can up-regulate the urokinase-type plasminogen activator (uPA) gene via the Ras/Erk signaling pathway. In this present study using the small interfering RNA technique, we have found that ShcA adapter proteins play a rather active role in CSR-induced Erk activation, contrary to their mostly redundant role in other signaling pathways, e.g. growth factor-induced Erk activation, where Grb2 can bind directly to the receptor tyrosine kinase and activate Erk in the absence of ShcA. ShcA knockdown abolished CSR-induced activation of both Erk and the uPA promoter. Expression of small interfering RNA-escaping silent mutants of p52 or p46 but not p66 ShcA isoform efficiently rescued CSR-induced Erk activation. Moreover, we have shown that phosphorylation of either Tyr-239/Tyr-240 or Tyr-313 in p52(ShcA) can mediate CSR-induced Erk activation equally well. In a quest for molecules upstream of ShcA in this signaling, we found that CSR-induced ShcA tyrosine phosphorylation, its association with Grb2, Erk activation, and uPA gene expression were all dependent on Rho kinase, p38 mitogen-activated protein kinase, and Src. In summary, we have found a novel, non-redundant role for ShcA in contrast to its redundant role in many other signaling pathways.

rived from a single gene through differential usage of transcription/translation initiation sites and alternative splicing, which differ in their amino-terminal sequence. The three proteins share an amino-terminal phosphotyrosine binding, a carboxyl-terminal Src homology 2, and a central CH1 domain (7,10). p66 Shc has an additional amino-terminal collagen homology-like domain denoted as CH2. Although all three isoforms of ShcA contain three conserved tyrosine residues (Tyr-239/Tyr-240 and Tyr-317) within the CH1 domain that are phosphorylated by activated tyrosine kinases and serve as docking sites for the Grb2-Sos complex (3,11), we have reported recently that they differ from each other with respect to their serine/threonine phosphorylation patterns (12). Ubiquitous expression of p52 Shc /p46 Shc and the presence of p66 Shc in most cells other than those of hematopoietic lineage suggest a distinct biological role for each isoform (13).
ShcA gene knockout in mice results in death at day 11.5 of embryogenesis, demonstrating an important role for ShcA in development (14). Using two different genetic approaches, inducible expression of a phosphorylation-defective mutant of Shc in transgenic mice and conditional knockout of Shc in thymocytes, Zhang et al. (15) have shown that Shc plays an essential and non-redundant role in T cell development. Some studies have suggested isoform-specific functions. Overexpression of p52 Shc /p46 Shc exerted a positive effect on growth factor-induced c-fos promoter activity, but p66 Shc expression showed a negative effect (13). Isoform-specific gene knockout of p66 Shc increased mouse longevity, most likely through suppressing oxidative stress-induced apoptosis (16). Although expression of the largest isoform is not essential for development (16), the benefit to the organism of this particular isoform remains to be elucidated. Likewise, it is not clear how important ShcA is in the regulation of growth factorinduced Ras/Erk signaling. In several systems, Grb2 has been shown to be recruited directly to the activated receptor tyrosine kinases, leading to Ras activation (17)(18)(19)(20)(21). In line with this, growth factor-induced Erk activation seems to proceed normally in ShcA Ϫ/Ϫ fibroblasts (14). In this system, however, ShcA seemed to be required for sensitizing cells and giving full induction of Erk activity at low concentrations of growth factor (14). Based on these observations, it has been suggested that p52 Shc /p46 Shc act as amplifiers of receptor tyrosine kinase-mediated signaling in pathways leading to Ras activation and involving a Grb2-Sos complex (10).
The cytoskeleton, consisting of actin, microtubule, and intermediate filaments, is a highly organized architectural entity that regulates cell shape and size and is associated with diverse functions depending on cell type (22). Cells respond to changes in cytoskeletal networks by inducing expression of specific genes through specific signaling cascades (23)(24)(25)(26)(27)(28)(29). Of particular interest among the signaling molecules activated by cy-toskeletal reorganization (CSR) 1 are the members of the mitogen-activated protein (MAP) kinases, i.e. Erk, c-Jun NH 2terminal kinase, and p38 (25, 30 -32), which are involved in a variety of cellular processes including cell differentiation, cell movement, cell division, and cell death (33). The mechanisms of activation of these molecules, however, are poorly understood. We have shown previously in LLC-PK 1 non-transformed epithelial cells that reorganization of actin and microtubule filaments by cytochalasins and colchicine, respectively, induces urokinase plasminogen activator (uPA) through the Ras/Erk signaling pathway (25). Induction was observed also in suspension cells, making it less likely that focal adhesions are involved, where various signaling molecules co-localize, including ShcA (25). ShcA tyrosine phosphorylation and its association with focal adhesion kinase (FAK) and Src after CSR were also observed, suggesting a role for these proteins in CSR-induced Erk activation and subsequent uPA induction (28). However, it has not been demonstrated that ShcA is essential to this process as Grb2 can be recruited directly to FAK (34). Also undetermined is the possible relative contribution of each ShcA isoform toward Erk activation.
It has been rather difficult to study the role of individual isoforms in a clean background (i.e. without the influence of other isoforms) when, as is usually the case, a gene of interest is expressed in multiple isoforms in a given cell. Recently, we developed a system, termed knockdown-in (35), in which isoform-specific down-regulation and expression are achieved utilizing siRNA-mediated RNA interference. In this method, a specific isoform is expressed from an expression vector with silent mutations within the region where siRNA targets the endogenous gene. Using this approach, we have determined the indispensability of ShcA for CSR-induced Erk activation. We show that either p52 Shc or p46 Shc alone is enough to mediate Erk activation after CSR but that p66 Shc does not play a role in this regulation, and we show that phosphorylation of either Tyr-313 (equivalent to Tyr-317 in human ShcA) or Tyr-239/ Tyr-240 in the CH1 domain of p52 Shc is required for this pathway. We also show signaling molecules upstream of ShcA in CSR-induced Erk signaling.

MATERIALS AND METHODS
Materials-Cytochalasin D (CytD) was purchased from Sigma. Luciferin was from Chemie Brunschwig AG. TPA, horseradish peroxidaseconjugated anti-mouse and anti-rabbit antibodies, ECL reagents, and protein A-and G-Sepharose were from Amersham Biosciences. Anti-ShcA polyclonal and antiphospho-Src (Tyr-416) polyclonal antibodies were obtained from Transduction Laboratories. Antiphospho-Erk, antiphospho-p38 MAP kinase, and antiphospho-Src (Tyr-416) polyclonal antibodies were from Cell Signaling, and anti-Erk polyclonal and anti-RhoA monoclonal antibodies were obtained from Santa Cruz. Mouse monoclonal antibodies against hemagglutinin (HA; 12CA5) and against phosphotyrosine (4G10) were purified on a protein A-Sepharose column. Anti-Src mouse monoclonal antibody (clone 327) was a gift from Dr. Kurt Ballmer. SB203580 and CPG77675 were kindly provided by E. Blum (Novartis AG).
Cells and Transfections-LLC-PK 1 cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% (v/v) fetal calf serum (AMIMED, Allschwil, Switzerland), 0.2 mg/ml streptomycin, and 50 units/ml penicillin at 37°C in a humidified incubator with 5% CO 2 . For serum starvation, cells were incubated in Dulbecco's modified Eagle's medium containing 0.1% fetal calf serum. Cells were transfected with siRNA as described previously (35). Briefly, a day before transfection, cells (10 5 /well) were plated in 6-well plates in medium without antibiotics, and the next morning siRNAs were transfected into cells using the LipofectAMINE reagent (Invitrogen) according to the manufacturer's instructions, with 10 l of 20 M siRNA and 5 l of transfection reagent/well.
Analysis of Reporter Gene Expression-One day after siRNA transfection, cells were replated in 6-well plates. On the next day, cells were co-transfected with a reporter plasmid and the Renilla control plasmid using the calcium phosphate precipitation method (Amersham Biosciences). After 6 h of transfection, cells were starved for 16 h followed by treatment with CytD for 6 h, and luciferase expression was measured as described previously (36) and normalized against Renilla expression.
Generation of Stable Cell Lines-Stable cell lines were generated as described previously (12). Briefly, a plasmid encoding mouse wild-type p46 Shc , p52 Shc , or p66 Shc or mutant p66 Shc S36A, p66 Shc T29A, p66 Shc S36A/T29A, p52 Shc Y313F, or p52 Shc Y239F/Y240F was co-transfected with the plasmid pX343 (37) expressing a hygromycin resistance gene in LLC-PK 1 cells at a ratio of 9:1 by the calcium phosphate method. All ShcA-encoding plasmids contained two silent mutations at the site of targeting siRNA (see below). Stable cell lines were selected by culturing cells with 3 mg/ml hygromycin B. Clones were isolated and screened for ShcA expression by immunoblotting with anti-HA antibody. The parent cell line transfected with pcDNA3 plus pX343 was used as a control.
siRNAs-The following 21-mer oligoribonucleotide pairs were used: all Shc siRNA nucleotides 677-697 (in the protein tyrosine binding domain), 5Ј-CUA CUU GGU UCG GUA CAU GGG-3Ј, and 5Ј-CAU GUA CCG AAC CAA GUA GGA-3Ј. Nucleotide numbering is based on the sequence of human p66 Shc mRNA (GenBank TM /EBI accession number HSU7377), and each pair has a 3Ј overhang of two nucleotides on each side. The siRNA recognition sequence was conserved between human and mouse, and its uniqueness was confirmed by blasting against the GenBank TM /EBI data base. Annealing was as described previously (38). An siRNA corresponding to nucleotides 753-773 of the firefly luciferase mRNA or a scrambled siRNA with the sequence 5Ј-GUA CCU GAC UAG UCG CAG AAG-3Ј was used as a negative control. They were obtained from Qiagen.
Immunoprecipitation and Western Blot Analysis-LLC-PK 1 cells (2 ϫ 10 6 ) were plated in 10-cm plates. The next day, cells were starved for 16 h, treated with different stimuli for the indicated times, washed with phosphate-buffered saline (PBS), and lysed with radioimmune precipitation assay buffer. Cell lysates were immunoprecipitated for 2 h at 4°C with anti-ShcA, anti-FLAG, anti-HA, or anti-Src antibodies; pulled down by protein A-or G-Sepharose beads; and washed twice with TNET (TNE ϩ 1% Triton X-100) and once with TNE (50 mM Tris⅐HCl, pH 7, 140 mM NaCl, and 5 mM EDTA). The beads were boiled in 2ϫ SDS sample buffer, and eluates were analyzed by Western blotting using anti-ShcA, antiphosphotyrosine, anti-Grb2, anti-FAK, and anti-Src antibodies. An enhanced chemiluminescence detection method (ECL, Amersham Biosciences) was employed, and the membrane was exposed to Kodak X-Omat LS or Biomax MR film.
Erk Kinase Assay-LLC-PK 1 cells were transfected with siRNA as described above. On the second day after transfection, cells were retransfected with HA-Erk2 plasmid using LipofectAMINE according to the manufacturer's instructions in a serum-free medium; 5 h later serum was added to 10%. After 10 h of transfection, cells were starved overnight in 0.1% fetal calf serum. Cells were then stimulated with CytD and lysed, and HA-Erk2 was immunoprecipitated as described above. The immunoprecipitated HA-Erk2 was used for a kinase assay as described previously (39).
Immunofluorescence-Cells were transfected with siRNA as de-scribed above, and 2 days later they were starved in 0.1% FBS for 16 h and stimulated with CytD for 30 min. Cells were then washed with PBS, fixed in 1 ml of prewarmed 3% paraformaldehyde in PBS for 20 min at room temperature, permeabilized with 0.5% Triton X-100 in PBS for 10 min, blocked with 5% normal goat serum for 20 min, and incubated with a polyclonal antiphospho-Erk (1:100) antibody in PBS containing 1% goat serum for 2 h. Cells were washed twice with PBS (10 min/wash), incubated with secondary Alexa 568 -anti-rabbit goat antibody (1:100, Molecular Probes) and phalloidin (1:100) for 40 min, and washed three times with PBS (10 min/wash). To visualize nuclei, DAPI (4Ј,6-diamidino-2-phenylindole, 1:5000) was added during the last wash. Coverslips were mounted on glass slides with Fluoromount (Serva). Fluorescence was visualized with a Zeiss Axioplan 2 fluorescence microscope, and all images were captured at ϫ600 magnification.

CSR-induced ShcA Tyrosine Phosphorylation, Grb2
Association, and Erk Activation-LLC-PK 1 cells treated with varying concentrations of CytD were analyzed for ShcA tyrosine phosphorylation, its association with Grb2, and Erk phosphorylation. As shown in Fig. 1A, CytD induced all these events at concentrations as low as 0.25 M and exhibited optimal effects at 1 M. In a kinetic analysis, Erk phosphorylation was observed after 15 min of CytD treatment, with a maximum at 30 min. This level was maintained for 1 h of treatment but de-clined slowly thereafter. Enhanced Erk phosphorylation was detected even after 4 h of treatment (Fig. 1B). Tyrosine phosphorylation of ShcA and its association with Grb2 occurred as early as 5 min, reached a maximum at 15 min, and stayed at high levels until 4 h after treatment (Fig. 1B).
ShcA Is Essential for CSR-induced Erk Activation-ShcA has been considered an important component of receptor tyrosine kinase-mediated signaling leading to Erk activation. However, based on results with ShcA Ϫ/Ϫ fibroblasts, it has been suggested recently that ShcA proteins are dispensable but act as amplifiers in growth factor-induced signaling (10). Although we showed previously that CSR induces ShcA tyrosine phosphorylation just like growth factor treatment (28), the role of ShcA in CSR-induced Erk activation was not determined. Here we used siRNA to address this question and found that knockdown of ShcA completely abolished CSR-induced Erk activation ( Fig. 2A). The suppression of Erk activation was not due to a nonspecific effect of RNA interference on the Ras/Erk signaling pathway as TPA-and growth factor-induced Erk activation was not affected ( Fig. 2A and data not shown). The result suggests an essential role for ShcA in CSR-induced Erk activation. To be sure that inhibition of CSR-induced Erk activation by ShcA siRNA is a genuine effect of RNA interference, we performed rescue experiments using a modified knockdown-in protocol (35). In this experiment, we first prepared stable LLC-PK 1 cell lines expressing ShcA isoforms from vectors with silent mutations at the targeting site of siRNA and then downregulated endogenous ShcA by transfecting with ShcA siRNA. As shown in Fig. 2B, knockdown of ShcA in control cells completely abolished CSR-induced Erk phosphorylation/activation. This could be rescued effectively by expressing silent mutants of either p52 Shc (Fig. 2, B and C) or p46 Shc (Fig. 2C) but not of p66 Shc (Fig. 2B) as shown by Western blot analysis and in vitro Erk kinase assays. Further analysis showed, in accordance with these results, that CSR-induced Erk localization to the nucleus was suppressed by transfection with Shc siRNA and that this suppression could be overcome in cells expressing silent mutants of p46 Shc and p52 Shc but not of p66 Shc (Fig. 2D). Thus, CSR-induced phosphorylation, activation, and nuclear localization of Erk are strictly dependent on the presence of either of the two shorter isoforms of ShcA.
Involvement of ShcA Tyrosine Phosphorylation in CSR-induced Erk Activation-ShcA is phosphorylated by receptor or non-receptor tyrosine kinases on three tyrosine residues, Tyr-317 (Tyr-313 in mouse) and the twin Tyr-239/Tyr-240 (11). Because phosphorylation at these residues has been attributed to different functions of ShcA (40,41), we looked at the relative importance of these tyrosine residues in CSR-mediated Erk activation by a rescue experiment similar to that described above. Stable cells lines expressing silent mutants of p52 Shc Y313F, p52 Shc Y239F/Y240F, and p52 Shc Y239F/Y240F/Y313F (3YF) were prepared. Suppression of CSR-induced Erk activation by ShcA siRNA was rescued by the expression of either p52 Shc Y313F or p52 Shc Y239F/Y240F mutants (Fig. 3A) but not by the triple tyrosine mutant p52 Shc 3YF (Fig. 3B). These results indicate that CytD treatment can induce ShcA tyrosine phosphorylation at either of these residues, which contributes equally well to the mediation of Erk activation. To confirm this, we examined tyrosine phosphorylation of these mutants and their association with Grb2. As shown in Fig. 3C, CytD treatment induced tyrosine phosphorylation of the p52 Shc WT, Y313F, and Y239F/Y240F mutants but not of 3YF. This tyrosine phosphorylation pattern correlated with the ability to associate with Grb2 (Fig. 3C, bottom).
ShcC Can Replace ShcA in CSR-induced Erk Activation-Two other members of Shc family, ShcC and ShcB, play an adaptor role similar to ShcA in the neural system (42), and their importance in neural development has been well demonstrated in vivo (43). ShcC, however, was found to be less efficient than ShcA in mediating nerve growth factor-induced Erk activation in PC12 cells because it has only one high affinity binding site for Grb2 compared with two such sites in ShcA (44). Therefore, we investigated whether ShcC is as efficient as ShcA in mediating CSR-induced Erk activation in our non-

FIG. 2. Shc indispensability in CytD-mediated Erk activation.
A, ShcA knockdown blocked CytD-induced Erk activation. Cells were transfected with buffer, Shc or control siRNA as described under "Materials and Methods," starved for 14 h, and left untreated or treated with 3 M CytD or 100 ng/ml TPA for 30 or 10 min, respectively. Equal amounts of total cellular proteins were resolved by SDS-PAGE, and Western blot analysis was performed using specific antibodies against Erk, phospho-Erk, and Shc. unstim., unstimulated. B, rescue of CytD-induced Erk activation by p52 Shc ; Western blot analysis. Stable cell lines expressing empty vector or silent mutants of HA-p52 Shc or HA-p66 Shc were prepared. Cells were transfected with Shc siRNA, control siRNA, or buffer alone. After starvation for 14 h, cells were treated with 3 M CytD and collected. Erk activation and Shc knockdown were analyzed by Western blot using specific antibodies. sm, silent mutant. C, rescue of CytD-mediated Erk activation by p46 Shc and p52 Shc ; in vitro kinase assay. Cells were transfected with siRNAs as described above, and HA-Erk2 was transfected 2 days later as described under "Materials and Methods." After starvation for 14 h, cells were stimulated with 3 M CytD for 30 min and lysed. HA-Erk2 was immunoprecipitated (IP) using anti-HA antibodies. The immunoprecipitates were incubated with myelin basic protein (MBP) in the presence of [␥-32 P]ATP, resolved by SDS-PAGE, and autoradiographed. HA-Erk2 was also analyzed by Western blotting using anti-Erk and antiphospho-Erk antibodies. D, rescue of CytD-mediated Erk activation by p46 Shc and p52 Shc ; indirect immunofluorescence. The stable cell lines indicated were transfected with siRNA, starved, and stimulated with CytD as described above. Cells were fixed and permeabilized, and the actin cytoskeleton and phospho-Erk were visualized by indirect immunofluorescence using phalloidin and antiphospho-Erk antibodies, respectively, as described under "Materials and Methods." neuronal LLC-PK 1 cell system. For this we prepared stable cell lines expressing either of the mouse ShcC isoforms p55 ShcC and p69 ShcC (45) and addressed the question by further rescue experiments. As shown in Fig. 4A, both isoforms of ShcC were able to rescue ShcA siRNA-suppressed CSR-induced Erk activation. Interestingly, in contrast to p66 ShcA (ϭ p66 Shc ), the p69 ShcC isoform did rescue Erk activation (Fig. 4A, cf. Fig. 2B).
If p69 ShcC could mediate CSR-induced Erk activation, why could p66 ShcA not? Serine/threonine phosphorylation of p66 ShcA has been attributed to its inability to associate with activated EGF receptor and thus to mediate Erk activation (46). We showed previously that Thr-29 and Ser-36 in the CH2 domain of p66 ShcA are phosphorylated upon TPA and growth factor treatment (12). To see whether possible serine/threonine phosphorylation of p66 ShcA is responsible for its inability to mediate CSR-induced Erk activation, we mutated these serine and threonine residues and made stable cell lines expressing the double mutant p66 ShcA S36A/T29A. As shown in Fig. 4B, this double mutant was still unable to rescue Erk activation, indicating that it is not the serine/threonine phosphorylation of p66 ShcA that prevents it from mediating CSR-induced Erk activation. Because tyrosine phosphorylation is necessary for Shc to recruit the downstream Grb2-Sos complex, we next sought the difference in tyrosine phosphorylation between p66 ShcA and p69 ShcC . As shown in Fig. 4C, EGF induced tyrosine phosphorylation in both p66 ShcA (with either wild type or the S36A/ T29A mutant) and p69 ShcC , but CSR induced tyrosine phosphorylation only in p69 ShcC . These results suggest that tyrosine phosphorylation is important for Erk activation and that the CH2 domain of p66 ShcA specifically interferes with the CSR-induced mechanism responsible for ShcA tyrosine phosphorylation.

Involvement of Src, p38 MAP Kinase, and Rho Kinase Upstream of ShcA in CSR-induced Erk Activation-Src and p38
MAP kinase have been shown to be involved in Erk activation and gene expression induced by microtubule or actin filamentdisrupting drugs (32,47). In addition, we have shown that a dominant negative mutant of Src inhibits CSR-induced uPA promoter activation in LLC-PK 1 cells (28). Consequently, we examined whether Src and p38 MAP kinase are involved in CSR-induced ShcA tyrosine phosphorylation and Erk activation. As shown in Fig. 5A, CGP77675 and SB203580, Srcspecific and p38 MAP kinase-specific inhibitors (48,49), respectively, attenuated CSR-induced Erk activation. The two inhibitors also inhibited ShcA tyrosine phosphorylation and its association with Grb2 (Fig. 5B), suggesting that both Src and p38 MAP kinase act upstream of ShcA in CSR-induced Erk activation. Rho proteins are important for regulating the actin

FIG. 3. Requirement of Shc tyrosine phosphorylation for CytD-mediated Erk activation. A, both Tyr-313 and
Tyr-239/Tyr-240 in ShcA are involved in CSR-induced Erk activation. Stable cell lines expressing silent mutants of HA-p52 Shc Y313F or HA-p52 Shc Y239F/Y240F (YYFF) were transfected with Shc siRNA, control siRNA, or buffer alone. Cells were starved for 14 h, stimulated with 3 M CytD for 30 min, and lysed. Equal amounts of proteins were resolved by SDS-PAGE and analyzed for Erk activation and Shc knockdown by Western blotting using specific antibodies. unstim., unstimulated. B, a triple tyrosine mutant of ShcA, HA-p52 Shc 3YF, did not rescue CytD-induced Erk activation in ShcA knockdown cells. Stable cell lines expressing pcDNA or silent mutants of HA-p52 Shc 3YF or HA-p52 Shc WT were used. ShcA was knocked down by siRNA transfection as described above, and Erk phosphorylation was analyzed by Western blot. sm, silent mutant. C, HA-p52 Shc 3YF was not tyrosine-phosphorylated or associated with Grb2 after CytD treatment. Cell lines expressing silent mutants of HA-p52 Shc WT, Y313F, Y239F/Y240F (YYFF), and 3YF were starved for 14 h, stimulated with 3 M CytD for 30 min, and lysed. Equal amounts of total cellular proteins were used for immunoprecipitation (IP) with anti-HA antibodies. The immunoprecipitates were resolved by SDS-PAGE and analyzed by Western blotting using antiphosphotyrosine, anti-Shc, and anti-Grb2 antibodies. cytoskeleton, and RhoA has been shown to be activated after treatment with CytD in Swiss 3T3 cells (50). Therefore, we tested the involvement of Rho kinase in Erk activation using a specific Rho kinase inhibitor, Y27632 (49). Treatment with Y27632 completely inhibited both Erk activation and ShcA tyrosine phosphorylation (Fig. 5C), indicating a potential role for Rho kinase in CSR-induced signaling.
Src, p38 MAP Kinase, and RhoA in CSR-induced Erk Activation-We examined whether these molecules are actually activated by CSR and, if so, how they are related to each other. Fig. 6A shows that CSR enhanced Src activation more that 2-fold (top), p38 MAP kinase 1.8-fold (middle), as measured by their phosphorylation status, and RhoA 3-fold (bottom), as measured by RhoA bound to GST-RBD as described previously (50). Interrelationships between the kinases were assessed by the effects of specific inhibitors. Note that these inhibitors inhibit the enzymatic activities of the target kinases and should not inhibit their own phosphorylation unless they are autophosphorylated directly or indirectly. CSR-induced Src phosphorylation was attenuated by its own inhibitor CGP77675, Rho kinase inhibitor Y27632, and the p38 MAP kinase inhibitor SB203580 (Fig. 6B), suggesting that Src is involved in its own activation and that Rho kinase and p38 MAP kinase are upstream of Src. However, activation of p38 MAP kinase was also inhibited by Src inhibitor, indicating a possible interdependence between these two molecules for activation (Fig. 6C). SB203580 also inhibited p38 MAP kinase activation, suggesting also the involvement of p38 MAP kinase autophosphorylation, directly or indirectly, as for Src in response to CSR. Rho kinase activation was not inhibited by Src and p38 MAP kinase inhibitors, showing that Rho is upstream of both Src and p38 MAP kinases (data not shown). We also showed previously that CytD causes Shc-FAK and Src-FAK association (28). Therefore, we asked whether Shc-FAK association is also affected by inhibition of these upstream kinases. As shown in Fig. 6D, treatments with inhibitors of Rho kinase, p38 MAP kinase, and Src all blocked Shc association with FAK.
Requirement of Shc, Rho, and p38 MAP Kinase for CSRinduced uPA Promoter Activation-To examine the biological relevance of the involvement of ShcA and Rho in CSR-induced Erk activation, we examined whether CSR-induced uPA gene expression involved ShcA and Rho. We showed previously that CSR induces uPA gene expression by activating the Ras/Erk signaling pathway (25). Consequently, the involvement of ShcA FIG. 4. ShcC can also mediate CytDinduced Erk activation. A, ShcC rescued CytD-induced Erk activation in ShcA knockdown cells. Stable cell lines expressing p55 ShcC or FLAG-p69 ShcC were prepared. Cells stably transfected with pcDNA or silent mutant HA-p52 ShcA were used as negative and positive controls, respectively. Cells were transfected with Shc siRNA, control siRNA, or buffer. After 2 days of transfection, cells were starved in 0.1% FBS for 14 h, stimulated with 3 M CytD for 30 min, and lysed. Equal amounts of cellular proteins were resolved by SDS-PAGE and immunoblotted with antiphospho-Erk, anti-Erk, and anti-Shc antibodies. unstim., unstimulated. B, serine/threonine phosphorylation-defective mutant of p66 ShcA did not rescue CytD-induced Erk activation in ShcA knockdown cells. Cells expressing a silent mutant of p66 ShcA S36A/T29A were transfected with siRNAs as described above. Cells were starved in 0.1% FBS, stimulated with CytD for 30 min, and lysed. Erk activation and ShcA knockdown were analyzed as described above. C, p69 ShcC but not p66 ShcA or p66 ShcA S36A/T29A was tyrosine-phosphorylated after CytD treatment. Cells expressing silent mutants of HA-p66 ShcA WT or HA-p66 ShcA S36A/T29A or FLAG-p69 ShcC were starved for 14 h in 0.1% FBS and stimulated with 3 M CytD for 30 min or 50 ng/ml EGF for 10 min. Cells were lysed and immunoprecipitated (IP) with either anti-HA antibodies or anti-FLAG antibodies. Immunoprecipitates were analyzed by Western blotting using antiphosphotyrosine, anti-HA, or anti-FLAG antibodies.
in CSR-induced uPA gene expression was examined by transient transfection assays. CSR induction of uPA promoter activity was completely suppressed by ShcA siRNA pretreatment (Fig. 7A), but the control siRNA also showed some inhibitory effect (about 20%), which could be a nonspecific effect of siRNA on the uPA promoter because we did not notice any such effect on CytD-induced Erk activation. uPA mRNA induction by CytD was partially suppressed by SB203580 (20%) and Y27632 (27%) (Fig. 7B). This indicates that p38 MAP kinase and Rho kinase play a role in CSR-induced uPA gene expression. CSR induction of the uPA promoter assessed by transient transfection assays was efficiently inhibited (71%) by pretreatment with Y27632. This is comparable with the inhibition by PD98059, a potent mitogen-activated protein kinase/extracellular signalregulated kinase kinase inhibitor (Fig. 7C). DISCUSSION In this work, we have shown that Shc plays an indispensable role in CSR-induced Erk activation, unlike growth factor-induced signaling, where Shc or Grb2 transduces signaling equally well (20,21). Knockdown of ShcA by siRNA completely abolished Erk activation after CytD treatment but not after TPA ( Fig. 2A) or EGF treatment (data not shown), and this inhibition was suppressed by the expression of silent mutants of different isoforms of ShcA. Moreover, the inhibitory effect of ShcA knockdown on Erk activation and its rescue by either of Cells were lysed, and equal amounts of total cellular proteins were resolved by SDS-PAGE and immunoblotted with anti-Erk and antiphospho-Erk antibodies. B, Src and p38 MAP kinase inhibitors blocked CytD-induced ShcA tyrosine phosphorylation and association with Grb2. Cells pretreated with inhibitors were treated with CytD for 30 min and lysed, and lysates were immunoprecipitated (IP) with anti-Shc antibodies. The immunoprecipitates were resolved by SDS-PAGE and analyzed by Western blotting using anti-Shc, antiphosphotyrosine, and anti-Grb2 antibodies. C, Rho kinase inhibitor blocked CytD-induced Erk activation and Shc tyrosine phosphorylation. Cells starved for 14 h in 0.1% FBS were pretreated with 10 M Y27632 for 45 min and then treated or not with CytD for 30 min. After lysis, equal amounts of proteins were used either for SDS-PAGE and Western blotting with anti-Erk and antiphospho-Erk antibodies (top) or for immunoprecipitation with anti-Shc antibodies followed by immunoblotting with antiphosphotyrosine, anti-Shc, and anti-Grb2 antibodies (bottom). the p52 Shc /p46 Shc isoforms but not by p66 Shc were monitored in three different ways to make sure that Erk activity after CSR is really controlled in an ShcA-dependent manner, namely Western blotting using phospho-specific antibodies, in vitro kinase assay, and immunocytochemistry for nuclear localization (Fig. 2, B-D). Because Erk activation is involved in CSRinduced uPA gene expression, we expected and observed uPA gene down-regulation by ShcA knockdown. The inability of p66 Shc to rescue CSR-induced Erk activation in ShcA knockdown-in experiments may correspond to the previous reports showing that p66 Shc is a negative regulator of EGF-induced Erk activation and c-fos promoter activation (13,46). The negative regulatory effect of p66 Shc was attributed to its serine phosphorylation because serine-phosphorylated p66 Shc associated with Grb2 but was unable to associate with tyrosinephosphorylated receptor, thus acting in a dominant-interfering manner (46). In our case, however, p66 Shc did not act in a dominant-interfering manner; in contrast to EGF stimulation, p66 Shc was not phosphorylated on tyrosine residues by CytD treatment and did not compete with other isoforms for Grb2 binding. Furthermore, a serine/threonine mutant of p66 Shc in the CH2 domain was also unable to rescue Erk activation, suggesting that p66 Shc is not involved, either positively or negatively, in CytD-induced Erk activation (Fig. 4B).
In a quest for signaling molecules that link CSR to ShcA activation/phosphorylation, we identified several essential molecules, including Rho kinase, p38 MAP kinase, Src, and FAK. CSR brought about by CytD or colchicine treatment has been shown to lead to Src activation and subsequent downstream signaling (29,47,51). RhoA and p38 MAP kinase have been shown to be required for CSR-induced cyclooxygenase 2 or matrix metalloproteinase gene expression (32,52). Overexpression of dominant negative RhoA inhibited taxol-and CytDinduced COX2 gene expression (32). Taking into consideration these results and previous reports from our lab (25,28), the present work has generated a CSR-induced signaling pathway leading to uPA gene expression shown schematically in Fig. 8 and discussed in the following two paragraphs.
ShcA can be phosphorylated on three conserved tyrosine residues in the CH1 domain, Tyr-239, Tyr-240, and Tyr-317, Src was immunoprecipitated using anti-Src antibodies, and its activation status was analyzed as described above. C, p38 MAP kinase and Src are mutually dependent in CytD-induced signaling. Cells were transfected with FLAG-tagged p38 MAP kinase, starved, pretreated with 10 M SB203580 or 5 M CGP77675 for 45 min, stimulated with 3 M CytD for 30 min, and lysed. p38 MAP kinase was immunoprecipitated and analyzed for its activation as described above. D, FAK-ShcA association was disrupted by inhibitors of Rho kinase, Src, and p38 MAP kinase. Cells were starved for 14 h; pretreated with 10 M SB203580, 10 M Y27632, or 5 M CGP77675 for 45 min; and then stimulated with 3 M CytD for 30 min. Cells were lysed, and ShcA was immunoprecipitated. Co-immunoprecipitated FAK was analyzed by anti-FAK antibodies.
which, once phosphorylated, are recognized by the Src homology 2 domain of Grb2. It has been shown that Tyr-239/Tyr-240 and Tyr-317 are phosphorylated by different kinases (53,54), and they are suggested to have different functions in cellular signaling, being linked to c-Myc and Erk MAP kinase activation, respectively (40). In rescue experiments with tyrosine mutants of p52 Shc , we found that both Tyr-239/Tyr-240 and Tyr-313 are phosphorylated in response to CytD and can mediate Erk activation independently of each other. However, as suggested previously, the upstream kinases for these tyrosines may be different (19,54). Schlaepfer et al. (19) reported that FAK phosphorylated Shc-Tyr-317 in vivo after fibronectin receptor integrin stimulation and in vitro. They also showed in vitro phosphorylation of Tyr-239/Tyr-240 of ShcA by Src. Preferential phosphorylation of Tyr-239/Tyr-240 by Src was also reported by van der Geer et al. (11). Similarly, using the Src family-specific inhibitor SU6656, Blake et al. (54) demonstrated that Tyr-239/Tyr-240 are phosphorylated by Src and Tyr-317 by platelet-derived growth factor receptor. Furthermore, Tyr-239/Tyr-240 phosphorylation was necessary for Srcmediated Myc induction and DNA synthesis. Based on our previous report that CSR induces FAK association with Src and ShcA (28) as well as our current data showing (1) phosphorylation of all three tyrosines upon CytD treatment, which could be suppressed by Src inhibitor, and (2) suppression of ShcA-FAK association by the same inhibitor, we propose that Src phosphorylates ShcA at Tyr-239/Tyr-240 as well as FAK in the activation loop after activation by CytD treatment. FAK in turn phosphorylates ShcA at Tyr-313 (Fig. 8). The fact that ShcA knockdown completely blocked CytD-induced Erk activation rules out the possibility that Src phosphorylates FAK at Tyr-925, a binding site for Grb2. Grb2 binding to Tyr-925 in FAK can also mediate fibronectin-induced Erk activation (19), which is not the case for CytD-induced Erk activation.
Interestingly, p38 MAP kinase and Src were mutually dependent for activation, and both activities were required for CSR-induced activation of ShcA as well as FAK. As CSR induces the association of Src and FAK (28), it is also possible that FAK is involved in the mutual dependence on CSR-induced activation. With these observations, we would like to introduce a new concept of "system field," in which the potential of the field is elevated as a result of mutual activation of its components; the system as a whole acts as an upstream regulator (Fig. 8). The interaction between the components of this system is currently under investigation. In our scheme, the question remains of how CSR is coupled to RhoA activation. As RhoA activation triggers polymerization of actin filaments, disruption of the filaments may induce feedback regulation involving RhoA activation to restore the actin cytoskeleton. The immediate signals induced by CSR still remain to be identified.
ShcB and ShcC, the other Shc family members, are involved in receptor tyrosine kinase signaling in the neural system. Mouse ShcC encodes two isoforms, p55 ShcC and p69 ShcC , which correspond to p52 ShcA (ϭ p52 Shc ) and p66 ShcA (ϭ p66 Shc ), respectively (45). Recently, Nakamura et al. (44) compared the efficacy of ShcA and ShcC in transducing NGF-induced Erk activation in PC12 cells and found that ShcA was more efficient than ShcC. This was due to the presence of only one high affinity Grb2 binding site in ShcC, compared with two such sites in ShcA. ShcC, however, had novel tyrosine phosphorylation sites that interacted with Crk in a phosphorylation-dependent manner. Keeping in mind the difference in efficacy of these two Shc family members in growth factor signaling, we also compared the actions of ShcA and ShcC in transducing CSR-induced Erk activation by means of rescue experiments. The efficiency of both isoforms of ShcC in transducing signals for Erk activation was comparable with ShcA. This suggests that although ShcC behaves differently from ShcA in growth factor-induced signaling in the neural system, it could play a role similar to that of ShcA in CSR-induced signaling. It would be interesting to see whether externally caused changes in cell morphology, such as those after injury, induce Erk activation in an ShcC-dependent manner in neural cells. Also interesting is the difference between the longer isoforms of ShcA and ShcC. The two molecules are similar in structure with an additional CH2 domain at their amino termini, but they differ in capacity to mediate CSR-induced signals. p66 ShcA did not mediate CSRinduced signaling for Erk activation, whereas p69 ShcC did as efficiently as p52 Shc /p46 Shc . This difference could be attributable to the different tyrosine phosphorylation patterns of the two molecules; p69 ShcC was tyrosine-phosphorylated by either EGF or CytD treatment, whereas p66 ShcA was phosphorylated only by EGF treatment (Fig. 4C). The CH2 domain of p66 ShcA likely has some information that suppresses CSR-induced phosphorylation, whether through steric hindrance or association with accessory proteins. Clearly, it does not involve Thr-29 or Ser-36 phosphorylation, which is specific to the p66 ShcA isoform. This difference in behavior also suggests that p69 ShcC plays a role in the neural system different to that of p66 ShcA in other tissues, i.e. to mediate oxidative stress response.