Kinase–Akt Cascade by the ETV6-NTRK3 Chimeric Tyrosine Kinase Requires c-Src

The ETV6-NTRK3 (EN) chimeric tyrosine kinase, a potent oncoprotein expressed in tumors derived from multiple cell lineages, functions as a constitutively active protein-tyrosine kinase. ETV6-NTRK expression leads to the constitutive activation of two major effector pathways of wild-type NTRK3, namely, the Ras-mitogen-activated protein kinase (MAPK) mitogenic pathway and the phosphoinositide-3-kinase (PI3K)-Akt pathway mediating cell survival, and both are required for EN transformation. However, it remains unclear how ETV6-NTRK3 activates Ras-Erk1/2 and/or PI3K-Akt cascades. Here, we define some aspects of the molecular mechanisms regulating ETV6-NTRK-dependent Ras-Erk1/2 and PI3K-Akt activation. We show that ETV6-NTRK3 associates with c-Src, and that treatment with SU6656, a c-Src inhibitor, completely blocks ETV6-NTRK-transforming activity. Treatment of NIH3T3 cells expressing ETV6-NTRK3 with SU6656 attenuated the activation of Ras-Erk1/2 and PI3K-Akt. Suppression of c-Src by RNA interference in NIH3T3-ETV6-NTRK3 cells resulted in markedly decreased expression of cyclin D1 and suppression of activation of Ras-Erk1/2 and PI3K-Akt. However, in Src-deficient cells, the ETV6-NTRK3 failed to activate the PI3K-Atk pathway, but not the Ras-Erk1/2 pathway. Therefore, these data indicate that ETV6-NTRK3 induces the PI3K-Akt cascade through the activation of c-Src.

The mechanism of ETV6-NTRK3-mediated oncogenesis is not well characterized.ETV6-NTRK3 (EN) is capable of homodimerization or heterodimerization with endogenous ETV6 via the helixloop-helix domain, leading to constitutive NTRK3 signaling.However, ETV6-NTRK3 fails to interact with adapter molecules known to associate with wild-type NTRK3 (17,18).These include Src homology and collagen and growth factor receptor binding protein 2 (Grb2), which link NTRK3 to the Ras-Erk1/2 mitogen-activated protein (MAP) kinase (MAPK) pathway involved in mitogenesis or differentiation (16,19) or the p85 subunit of phosphoinositide-3kinase (PI3K) linking NTRK3 with the PI3K-Akt neuronal survival pathway (16,19,20).ETV6-NTRK3 does bind peritoneal lymphocytes, another known NTRK3 interactor that activates protein kinase C, but PTK-active mutants unable to bind phospholipase C (PLC) did not show defects in the transformation activity (17).We have recently shown that the distal COOH-terminal sequence of EN interacts specifically with the phosphotyrosine binding (PTB) domain of insulin receptor substrate-1 (IRS-1), and that this interaction is essential for EN transformation (21).However, it remains to be determined whether the interaction with the downstream signaling cascades is direct through IRS-1 or indirect via other molecules.
Expression of the ETV6-NTRK3 fusion protein is associated with constitutively high levels of phosphorylated Mek1 and Akt, even in the absence of serum.Moreover, ETV6-NTRK3-expressing cells show serum-independent elevation of cyclin D1 protein.Inhibition of either the Ras-Erk1/2 MAP kinase or the PI3K-Akt pathway alone completely blocks colony formation of ETV6-NTRK3-expressing cells in soft agar assays.Furthermore, the constitutive expression of cyclin D1 protein in ETV6-NTRK3-expressing cells can be transiently down-regulated by MEK1 or PI3K inhibition.However, only the inhibition of the Ras-Erk1/2 pathway led to persistent down-regulation of cyclin D1 levels in cells expressing ETV6-NTRK3 (15).c-Src is known to activate PI3K-Akt and the MAP/extracellular signal-regulated kinase (ERK) kinase (MEK)/ERK cascades through both focal adhesion kinase (FAK)-dependent and FAK-independent
Plasmids.Three small interfering RNA-coding oligos against mouse c-Src were designed and verified to be specific to c-Src by Blast search against the mouse genome.To construct hairpin-type single RNAi vectors, 5 AL (100 mmol/L) of the synthesized sense and antisense oligonucleotides (Table 1) were combined with 1 AL of 1 mol/L NaCl and annealed by incubation at 95jC for 2 min, followed by rapid cooling to 72jC, and ramp cooling to 4jC over a period of 2 h.The c-Src-siRNA-1 and c-Src-siRNA-2 insert were subcloned into the XbaI/XhoI sites of the pFG12 lentivirus vector.A control siRNA, which does not match any known mouse coding cDNA, was used as control.
Immunoblotting and immunoprecipitation. 293T cells were used for the detection of protein-protein interaction in vivo.293T cells were transiently transfected with the indicated plasmids.After a 24-h transfection, cells were switched to 0.2% serum overnight.Cells were lysed in a buffer containing 25 mmol/L HEPES (pH, 7.5), 150 mmol/L NaCl, 1% Triton X-100, 10% glycerol, 5 mmol/L EDTA, and protease inhibitor mixture (Complete, Roche, Gipf-Oberfrick, Switzerland).Extracts were separated by SDS-PAGE, followed by electrotransfer to poly(vinylidene diflouride) membranes and probed with polyclonal or monoclonal antisera, followed by horseradish peroxidase-conjugated anti-rabbit, anti-mouse immunoglobulin G (IgG), respectively, and visualized by chemiluminescence according to the manufacturer's instructions (Pierce, Rockford, IL).For immunoprecipitation, the cell lysates were incubated with the appropriate antibody for 1 h, followed by incubation with g-bind beads (GE Healthcare Bio-Sciences Corp., Piscataway, NJ) for 1 h at 4jC.Beads were washed four times with the buffer used for cell solubilization.Immune complexes were then eluted by boiling for 3 min in 2Â Laemmli buffer (pH, 6.8), and then extracts were analyzed by immunoblotting as described above.
Soft agar assays.Soft agar assays were done according to established protocols (24).NIH3T3 cells infected with recombinant retroviruses carrying either the empty vector, ETV6-NTRK3, or NIH3T3-EN cells carrying siRNA-coding oligos against mouse c-Src were seeded in triplicate at a concentration of 8 Â 10 3 cells per well in a six-well plate.Bottom layers were made up of 0.4% agar in 9% FBS and DMEM.Cells were resuspended in a top layer of 0.2% agar in 9% FBS and DMEM.Cells were fed every other day by placing two drops of medium on the top layer.After 3 weeks at 37jC, colonies were then stained with p-iodonitrotetrazolium violet (1 mg/mL stock diluted at 1:500) for 16 h, and macroscopic colonies were counted (in quadruplicate).Each determination was done in triplicate.
Viral production and infection of target cells.293T cells were transfected with the transfer vector plasmid pCAG (empty) or pCAG-ETV6-NTRK3 plasmid with the envelope-encoding plasmid pHCMVG, the packaging plasmid pMDLg/pRRE, and the Rev-expression plasmid pRSV-Rev by using the calcium phosphate method.The supernatants were harvested 48 and 72 h after transfection, pooled, passed through a 0.45-Am filter, ultracentrifuged for 2 h at 100,000 Â g in a SW28 rotor, resuspended in 100 AL of 0.1% bovine serum albumin in PBS, and the lentiviral stocks were stored in a small aliquot at À80jC for titration and cell infection.SYF and SYF-Src cells were plated in six-well plates (1 Â 10 5 cells per well) and were cultured overnight.Lentiviruses were diluted in 2 mL DMEM containing polybrene (8 g/mL) and then centrifuged for 30 min at 1,500 rpm and after 24 h infection, polybrene-DMEM was replaced with fresh DMEM medium, and the cells were cultured for other assays.

Results
EN directly binds to c-Src.We examined the possibility that EN may interact directly with c-Src in vivo.We used V5-tagged EN to screen for potential interactions between EN and Src.293T cells were transfected with HA-tagged Src and V5-tagged EN.Total cell extracts were subjected to immunoprecipitation with the anti-HA or anti-V5 antibody, followed by immunoblotting using anti-HA or anti-V5 antibodies.As shown in Fig. 1A, EN interacted strongly with c-Src.We next determined whether endogenous Src interacts with EN.To this end, we did immunoprecipitation experiments in NIH3T3 cells, which were infected with recombinant retroviruses carrying either ETV6-NTRK3 cDNA or the empty MSCVpac vector as a negative control.Endogenous c-Src was able to interact with EN (Fig. 1B).The kinase dead EN-K380N mutant, which fails to autophosphorylate and completely lacks transformation activity (17), as well as a mutant in which all three tyrosines were converted to phenylalanine (EN-Yx3F), failed to interact with c-Src (Fig. 1B).We also examined the effect of K252a 43, an inhibitor of the Trk tyrosine kinases, to determine whether NTRK3 kinase activity of EN is required for its ability to interact with c-Src.As shown in Fig. 1C, pretreatment with K252a markedly reduced levels of c-Src bound to EN in NIH3T3 cells expressing EN.This indicates that NTRK3 (TrkC) kinase activity is required for the interaction between c-Src and EN.To further investigate whether interaction of EN with c-Src results in the activation of c-Src kinase, we examined the phosphorylation of c-Src in lysates of NIH3T3 cells expressing EN as well as EN-K380N and EN-Yx3F.EN expression increased phosphorylation of c-Src-Tyr 416 .However, c-Src phosphorylation was not observed in NIH3T3 cells expressing EN-K380N and EN-Yx3F (Fig. 1D).These results suggest that phosphorylation of EN kinase may result in a conformational change, exposing its interaction domain to c-Src.Identification of EN-Src complexes in human primary tumors.We next wished to confirm that the above findings are relevant to human tumors expressing EN.We therefore screened for the presence of EN-v-Src complexes in human primary tumors positive for the ETV6-NTRK3 gene fusion.These included four cellular mesoblastic nephromas (CMN) occurring in 5-month-and 5-week-old infants and one congenital fibrosarcoma (CFS) occurring in a 15-month-old infant (Fig. 1E and Supplementary Fig. S1).Tumors were diagnosed using standard histopathologic criteria.All cases were screened for ETV6-NTRK3 fusion transcripts using established methods (1), but only the CMN and CFS cases were fusion positive (data not shown).We screened 12 primary tumor samples (4 CFSs and 8 CMN), but only 5 samples were of sufficient integrity for protein studies.Therefore, we examined the interaction between c-Src and EN using these five samples.Total tissue extracts were isolated from primary tumors and subjected to immunoprecipitation with the anti-c-Src antibody, followed by immunoblotting with the anti-TrkC antibody.As shown in Fig. 1F and G, EN-c-Src complexes were found in the CMN and CFS cases, but not in control liver samples.EN expression was associated with a strong increase in phosphorylation of c-Src and Akt (S473; Fig. 1F).These results clearly show that the interaction between EN and c-Src occurs in vivo in human primary cancer tissues expressing ETV6-NTRK3 gene fusions.

SU6656 treatment suppresses EN transformation activity.
Cells infected with ETV6-NTRK3 exhibited a dramatically transformed phenotype compared with negative control cells.We examined the effect of SU6656 (25), an inhibitor of c-Src kinase, on EN transformation activity by assessing the cell morphology.EN transformation activity was almost completely abrogated by the addition of SU6656 (Fig. 2A).Treatment of NIH3T3 cells expressing EN with even 1 Amol/L SU6656 exhibited a normal phenotype and failed to form microscopic or macroscopic colonies in soft agar when scored after 28 days (Fig. 2A and B).Most tumor cell lines suppress anoikis and form multicellular spheroids through cell-cell adhesion when grown in nonadhesive agar-coated dishes.Under these conditions, nonmalignant cells generally fail to do so and undergo massive cell death in culture (26).We observed that when placed in nonadherent cultures, parental NIH3T3 cells poorly formed cell aggregates and existed as dead or dying cells under these conditions (Fig. 2C).In contrast, EN-expressing cells readily form multicellular spheroids within 6 to 24 h that are stable indefinitely in culture (27).However, SU6656 treatment of EN-expressing cells suppressed formation of multicellular spheroids (Fig. 2C).The fact that NIH3T3-EN cells do not form stable spheroids in the presence of SU6656 indicates that this reagent is blocking the ability of cells to suppress anoikis.were used.Cell extracts were immunoprecipitated using anti-TrkC (NTRK3) antibody and g-bind beads (Amersham Pharmacia Biosciences), followed by immunoblotting with anti-Src antibody.C, NTRK3 kinase activity is required for the EN interaction with c-Src.EN-expressing NIH3T3 cells or vector control cells were pretreated with K-252a for 2 h.Cell extracts were immunoprecipitated using anti-TrkC (NTRK3) antibody, followed by immunoblotting with anti-Src antibody.D, EN increased phosphorylation of c-Src in vivo.E, identification of EN-c-Src complexes in human primary tumors.Primary tumor tissue extracts from EN-positive CMN and CFS were subjected to immunoprecipitation using anti-cSrc antibody or IgG as a control, followed by immunoblotting with anti-TrkC (NTRK3) antibody.Whole-tissue lysates from primary tumors also were probed for EN, c-Src, phospho-c-Src, and h-actin as a loading control.

Cancer Research
Cancer Res 2007; 67: (7).April 1, 2007 Characterization of the c-Src/EN interaction.c-Src is a multidomain protein containing a unique NH 2 -terminal domain: UD (residues 1-80), SH3 (residues 83-144), SH2 protein interaction domains (150-247), a catalytic region (269-522), and a negativeregulatory tyrosine located near the COOH terminus (residue 523; ref. 28; Fig. 3A).To identify the functional domain of c-Src responsible for the interaction with EN, we used a series of deletion constructs.The c-Src mutant lacking the UD region, SH3 and SH2 still interacted with EN (Fig. 3B).This indicates that the c-Src COOH-terminal region is directly involved in this interaction.Next, we tested the interaction of EN with a series of COOH-terminal deletion mutants.The c-Src DC, which does not have a negative regulatory tyrosine in the COOH-terminal region, still interacted with EN.Although the deletion of amino acid residues 361-535 (DKD-1 and DKD-2) also interacted with EN, the deletion of amino acid residues 275-360 abrogated the EN interaction (Fig. 3C).These results clearly indicate that the region including amino acids 275-360 of Src, which includes the ATP binding domain, is required for EN interaction.
The Fyn tyrosine kinase, a member of the c-Src kinase family, is known to associate with IRS-1 in vitro and in vivo (29).Because other members of the c-Src kinase family have similar SH2 domains, these kinases may also associate with IRS-1 and contribute to the overall response in various cellular backgrounds.Because IRS-1 functions as the adaptor protein linking EN to downstream signaling pathways, we examined whether the interaction between EN and c-Src is mediated through IRS-1.293T cells were transfected with Myc-tagged c-Src and HA-tagged IRS-1 with or without the V5-tagged EN.IRS-1 interacts with either EN or c-Src (Fig. 4A and B).To determine whether c-Src could associate with EN directly, 293T cells were transiently transfected with HA-tagged c-Src and either the V5-tagged construct or V5-tagged EN-D614 construct which fails to associate with IRS-1 (21).Lysates were immunoprecipitated with HA antibodies followed by Western blot analysis with V5 antibodies.As shown in Fig. 4C, EN and EN-D614 were pulled down in c-Src lysates.Equal expression of c-Src was confirmed by Western blotting of lysates using HA antibodies (Fig. 4C).These data indicate that c-Src directly interacts with EN.
c-Src expression in c-Src-deficient MEF cells enhances the EN-IRS-1 complex formation.To test whether EN binding to IRS-1 requires c-Src, we used SYF cells that were derived from Src, Yes, and Fyn triple knock-out mouse embryos (30).Because Src, Yes, and Fyn are three ubiquitously expressed members of the Src family tyrosine kinases, no Src family tyrosine kinase activity was detected in these SYF cells (30).We infected SYF and SYF-c-Src cells with the lentiviral vector transducing the ETV6-NTRK3 gene.As shown in Fig. 4D, expression of EN protein in these cells was verified by immunoblotting using the anti-TrkC (C-14) antibody.We next determined whether EN-IRS-1 complex formation is increased in the presence of c-Src.IRS-1 immunoprecipitated with EN was detected in c-Src-deficient SYF cells.However, the level of IRS-1 immunoprecipitated with EN was markedly enhanced in SYF cells expressing c-Src compared with that in SYF cells (Fig. 4E).Interestingly, expression of EN markedly decreased the level of IRS-1 protein in SYF cells.In SYF-c-Src cells, the level of IRS-1 is much lower than that in SYF cells, and expression of EN further reduced the level of IRS-1.It is possible that when cells are transformed by an oncoprotein such as EN, IRS-1 cellular levels must remain below a certain threshold to avoid inducing apoptosis in these cells.However, this hypothesis requires further experiments to validate.We next examined whether endogenous c-Src plays a critical role for EN and IRS-1 interaction and EN-mediated IRS-1 phosphorylation.To do this, the RNA interference approach was employed to knockdown endogenous c-Src in NIH3T3 cells expressing EN.We constructed lentiviral vectors expressing two different c-Src siRNAs (Si-c-Src 1 and 2).To establish cell lines expressing these c-Src siRNA genes, two cycles of infection and selection were done.The c-Src gene silencing activity of c-Src RNAi in NIH3T3 cells expressing EN was confirmed by immunoblotting with c-Src antibodies.Expression of c-Src siRNAs (Si-c-Src 1 and 2) resulted in more than 80% decreases in the expression of the endogenous c-Src (Supplementary Fig. S2).The luciferase RNAi nucleotide (Si-Lu) was used as a control.We examined whether the EN and IRS-1 interaction and EN-mediated IRS-1 phosphorylation are affected when endogenous c-Src is silenced in NIH3T3-EN cells.The level of IRS-1 bound to EN was markedly reduced in c-Srcsilenced NIH3T3-EN cells (Fig. 4F).Interestingly, EN-mediated IRS-1 phosphorylation seemed to be increased in c-Src-silenced NIH3T3-EN cells.However, when IRS-1 phosphorylation levels are normalized with endogenous IRS-1 levels, this induction effect is negligible (data not shown).We are currently investigating how c-Src reduces expression of IRS-1.Taken together, these results suggest that c-Src facilitates the EN-IRS-1 interaction.However, these data do not exclude the possibility that EN interacts directly, although more weakly, with IRS-1, or that an unidentified protein may also be involved in the EN-IRS-1 complex formation.
Treatment with SU6656 attenuates Akt Activation in ENexpressing NIH3T3 cells.ETV6-NTRK3 seems to be linking to both the Ras-Erk1/2 and the PI3K-Akt pathways via adaptors other than those used by wild-type NTRK3 (15).Recently, we have shown a role for IRS-1 binding in constitutive activation of the Ras-MAPK and PI3K-Akt pathways in EN-transformed cells.Therefore, we examined whether c-Src also plays a role in ENinduced activation of the Ras-MAPK and PI3K-Akt pathways and in constitutive overexpression of cyclin D1 by using an inhibitor of c-Src kinase.To assess this, we examined the activation states of MEK1/2 and Akt, as well as cyclin D1 levels, in NIH3T3 cells expressing EN by treating these cells with SU6656.Levels of phosphorylated (activated) MEK1/2 and Akt as well as cyclin D1 were reduced markedly in EN-expressing cells after treatment with SU6656 (Fig. 5A).Previous studies have shown that the PI3K inhibitor LY294002 blocks soft agar colony formation of EN-transformed NIH3T3 fibroblasts (27).To determine that PI3K-Akt is a downstream effector of the c-Src kinase, we treated EN-expressing NIH3T3 cell lines with the PI3K-Akt inhibitor LY294002 and examined the phosphorylation of Src.LY294002 treatment caused a reduction in phosphorylation of Akt, whereas it has no effect on c-Src phosphorylation (Fig. 5B), indicating that PI3K-Akt acts downstream of c-Src in EN-expressing NIH3T3 cell lines.Taken together, our findings show that c-Src is essential for the ability of EN to constitutively activate the Ras-MAPK and the PI3K-Akt pathways and to induce cyclin D1 overexpression.
Effects of c-Src-targeted siRNA on EN-transforming activity.To further prove the requirement of c-Src for EN transformation activity, we examined the EN-induced MEK1/2 and Akt-1 activation, cyclin D1 expression, and the EN transformation activity in c-Src-silenced NIH3T3-EN cells and control NIH3T3-EN cells (si-Luc).As shown in Fig. 5C, EN-induced MEK1/2 and Akt-1 activation and cyclin D1 expression were significantly reduced in c-Src knockdown cells.Expression of c-Src RNAi nucleotides (Si-c-Src) failed to form microscopic or macroscopic colonies in soft agar when scored after 28 days in NIH3T3 cells expressing EN (Fig. 5D).Taken together, our data suggest that c-Src plays a role in the EN-induced MEK1/2 and Akt-1 activation and cyclin D1 expression and mediates the EN transformation activity.
c-Src is required for EN-induced Akt activation in c-Srcdeficient MEF cells.To genetically test the role of c-Src in EN signaling to the MEK1/2 and PI3K-Akt-1 pathways, we investigated the activation of MEK1/2 and PI3K-Akt-1 by the EN in SYF and SYF-c-Src cells.SYF cells and SYF-Src cells with or without the EN were serum starved overnight and then incubated with serum.Ras-Erk1/2 and PI3K-Akt-1 activation as well as cyclin D1 expression were not observed in SYF cells.Constitutive serumindependent phosphorylation of Akt Ser 473 was observed in SYF cells expressing c-Src, but EN expression in these cells markedly increased Akt phosphorylation (Fig. 6A).Interestingly, stable expression of EN in SYF cells did not lead to increased activity of Akt, suggesting that EN-induced PI3K-Akt activation requires c-Src (Fig. 6B).In both SYF and SFY-c-Src cells, expression of EN increased the Mek1 phosphorylation even in the absence of serum (Fig. 6B).Interestingly, SYF-Src cells had constitutively high levels of Mek1 phosphoryation, but addition of serum decreased levels of Mek1 phosphoryation.SYF cells expressing EN exhibited constitutively high levels of cyclin D1 even after overnight serum starvation (Fig. 6A).SYF-Src cells responded to EN in cyclin D1 elevation only in the presence of serum (Fig. 6B).In NIH3T3-EN cells, silencing c-Src expression blocked EN-mediated MEK phosphorylation (Fig. 5C).In contrast, EN-mediated MEK phosphorylation remained unaffected in SYF cells.Because SYF cells are also deficient of Yes and Fyn in addition to c-Src, Yes and Fyn may also affect EN-mediated MEK phosphorylation via unknown mechanisms.It is also possible that in the absence of the three Src family members, the cells use another mechanism to induce Ras-MAPK and MEK activation.
Previous studies have shown that the Mek1 inhibitor U0126 markedly decreased levels of cyclin D1 levels in EN-expressing cells.We also confirmed this finding in SYF and SRF-c-Src cells expressing EN (Fig. 6C and D).These genetic data showed that c-Src is required for EN-induced PI3K-Akt activation, but not for the Ras-Erk1/2 pathway.These results further supported by our previous finding that only the inhibition of the Ras-Erk1/2 pathway led to persistent down-regulation of cyclin D1 expression (15).

Discussion
The EN chimeric product functions as a constitutively active protein-tyrosine kinase with potent transforming activity in multiple cell lineages (15,17).Activation of the EN protein-tyrosine kinase is linked to the induction of downstream signaling pathways of wild-type NTRK3, including Ras-MAPK and PI3K-Akt, leading to elevated cyclin D1 expression and aberrant cell cycle progression (15).However, the mechanism of EN-mediated oncogenesis remains for the most part unknown.In this study, we show that c-Src activation is necessary for the transformation activity of EN in NIH3T3 cells.Inhibition of either c-Src kinase activity or c-Src expression blocked anchorage-independent growth and formation of microscopic or macroscopic colonies in soft agar in NIH3T3 fibroblasts expressing EN.
The transforming properties of EN are associated with the constitutive activation of NTRK3 signaling pathways, namely, the Ras-MAPK and PI3K-Akt cascades (15).However, other than IRS-1 (ref.27; see below), the adaptor proteins that link EN to activation of these pathways have yet to be identified.EN does not directly associate with known NTRK3-associating molecules, including Src homology and collagen, Grb2, PI3K p85, ABL, SH2B, or Ship2 (15,17).The unique ability of ETV6-NTRK3 to activate both the Ras-Erk1/2 and PI3K-Akt pathways may be key to its oncogenic activity.An association between the aberrant activation of the Ras-Erk1/2 pathway and cellular transformation is well described in the literature (31,32).Alterations in the PI3K-Akt pathway have also been observed in humans tumors, including the amplification of the Akt h isoform in breast and ovarian carcinomas (33), amplification of the PI3K regulatory subunit in ovarian carcinoma (34), and mutations of the PTEN tumor suppressor gene leading to the constitutive Akt activation (reviewed in ref. 35).Although these and other studies (reviewed in ref. 36) illustrate the oncogenic potential of each individual pathway, it is becoming increasingly clear that a synergistic effect exists between Ras-Erk1/2 and PI3K-Akt cascades in transformation.Activated Ras mutants that are incapable of activating PI3K are unable to transform NIH3T3 cells unless an activated (nontransforming) viral form of Akt is coexpressed (37).Moreover, the transformation of rat intestinal epithelial cells by oncogenic Ha-Ras not only leads to coactivation of PI3K-Akt but is blocked by PI3K inhibitors (38).Recent studies suggest that depletion of endogenous c-Src levels by means of siRNA or expression of dominant-negative c-Src significantly reduced cell proliferation in vitro and led to a clear decrease in Akt phosphorylation, placing c-Src upstream of the PI3K-Akt signals.Therefore, our data indicate that EN-activated c-Src tyrosine kinase exerts its transforming activity by activating PI3K-Akt.
Recent studies show that EN binds the IRS-1 protein, and that EN-IRS-1 complex localization to the membrane leads to the activation of the PI3K-Akt cascade (27).Expression of EN in insulin-like growth factor-IR (IGF-IR)-null R À mouse embryo fibroblasts is specifically associated with a defect in Akt activation, whereas these cells retain intact Ras-MAPK pathway activation and cell A, MSCV or ETV6-NTRK3-transformed NIH3T3 cells were grown for 48 h in 9% calf serum-DMEM and then incubated with or without the c-Src family tyrosine kinase inhibitor SU6656 (2 and 5 Amol/L) or DMSO (0) for 1 h.Total cell lysates were prepared from the untreated and treated cells, and Western blotting was done using antibodies against P-MEK, P-Akt, and cyclin D1.B, MSCV or ETV6-NTRK3expressing cells were grown for 48 h in 9% calf serum-DMEM and then grown in 9% calf serum-DMEM with or without LY294002 (25 Amol/L) for 1 h.Whole cell lysates were prepared, and Western blotting was done using phospho-Src and phospho-Akt antibody.C, suppression of c-Src expression by stable siRNA decreases elevated levels of phosphorylated Akt in NIH3T3 cells stably expressing ETV6-NTRK3.ETV6-NTRK3, c-Src-siRNA-expressing or control NIH3T3 cells were serum starved overnight in 0.5% serum and then stimulated with 9% CS-DMEM for 1 h.Whole cell lysates were prepared for Western blotting and probed with antibodies against the phosphorylated forms of Mek1/2 (Ser 217/221 ), phosphorylated Akt (Ser 473 ), and cyclin D1.D, soft agar colony-forming assay.Soft agar colony-forming assay of EN control-siRNA, EN c-Src-siRNA-1, and EN c-Src-siRNA-2.Monolayer cells were trypsinized, washed, and plated in medium containing 0.2% agar to assess anchorage-independent growth.Results are presented as the number of macroscopic colonies formed at 3 wks after plating.cycle progression.R À EN cells undergo detachment-induced apoptosis (anoikis) under anchorage-independent conditions.This report showed that soft agar colony formation, spheroid formation, and tumorigenesis induced by EN require Akt activation, and IGF-IR-mediated IRS-1 membrane localization is essential for Akt activation during EN transformation.We found that either inhibition of c-Src kinase activity or knockdown of c-Src expression in NIH3T3 fibroblasts expressing EN suppresses Akt activation.These results suggest that c-Src is involved in the EN-IRS-1induced Akt activation.One possibility to explain our findings is that c-Src is also recruited to EN-IRS-1 complexes and is essential to link these complexes to downstream signaling cascades such as PI3K-Akt and Ras-Erk1/2.IGF-I and IGF-II are known to promote direct association of c-Src with IGF-IR and to phosphorylate c-Src (39) and PP2, a specific c-Src-like kinase inhibitor, blocks IGF-I-stimulated Akt activation (39).Interestingly, IGF-IR expression was reported to be up-regulated in the PANC-1 human pancreatic cancer cell line when cells were stably transfected with active c-Src (40).This increased expression of IGF-IR may also enhance the Akt activation.IGF-IR is frequently overexpressed in several types of human malignancy and is associated with invasion and metastasis of tumor cells.It will be interesting to investigate whether IGF-IR is also overexpressed in EN-expressing tumors.Interestingly, we recently showed that Ras-ERK pathway signaling, cyclin D1 induction, and cell cycle progression may be independent of EN/IRS-1 complex localization to the plasma membrane (27).Whereas PI3K-Akt signaling occurs almost exclusively at the plasma membrane, Ras-ERK signaling can take place on Golgi and endoplasmic reticulum membranes (41).EN showed a perinuclear expression pattern in IGF-IR null R À cells, which is indicative of localization to internal membranes.It is therefore possible that Ras-ERK signaling is activated predominantly on endomembranes in EN-expressing cells (27).Taken together, our observations establish an essential role for c-Src kinase in the activation of the PI3K-Akt cascade in EN-expressing fibroblasts.
c-Src is a proto-oncogene belonging to the nonreceptor protein tyrosine kinase family, which plays a prominent role in carcinogenesis (42,43).c-Src kinase activity is significantly increased in a number of human cancers, including lung, skin, colon, breast, ovarian, endometrial, and head and neck malignancies, and this elevated c-Src activity is correlated with poor metastasis-free survival (42,44,45).Because c-Src is a critical component of so many different processes that promote cancer progression, it is becoming recognized as a valid chemotherapeutic target.c-Src and its family members are critical mediators of multiple signaling pathways that regulate all stages of cancer progression ( from initiation to metastasis) in multiple cell types, and thus, one can envision the use of c-Src inhibitors in a wide range of malignancies at all stages of disease.Recent studies provided compelling evidence that in vivo pharmacologic treatment with a c-Src inhibitor could be very promising for the treatment of human cancers and their metastatic complications (46,47).Interestingly, human primary cellular tumors expressing the ETV6-NTRK3 gene exhibit elevated c-Src protein levels and catalytic activity.Given the difficulty in the clinical management of CFS and cellular CMN, the possibility that c-Src inhibitors can be used alone or in combination for the treatment of human tumors expressing ETV6-NTRK3 is of great interest.
In summary, we have shown that EN binds to and activates c-Src, and inhibition of c-Src activation blocks EN transforming activity and activation of PI3K-Akt.In c-Src-deficient SYF cells, it seems that there is no Akt activation in response to serum, and that expression of EN cannot overcome this.Because a recent report showed that c-Src is directly associated with Akt (48), our data suggest that EN requires c-Src to activate Akt.

Figure 1 .
Figure1.EN interacts with c-Src.A, V5-tagged EN was cotransfected into 293T cells with HA-tagged c-Src.Cell extracts were subjected to immunoprecipitation using anti-HA antibody or anti-V5 antibody and g-bind beads (Amersham Pharmacia Biosciences), followed by immunoblotting with anti-V5 antibody or anti-HA antibody.The expression of EN and c-Src was monitored as indicated.B, schematic representation of ETV6-NTRK3 mutant constructs (top ).c-Src does not associate with the kinase-dead EN-K380N and EN-Yx3F mutant protein.NIH3T3 cell lines stably expressing EN, EN-K380N, and EN-Y3XF as well as control NIH3T3 cells (MSCV) were used.Cell extracts were immunoprecipitated using anti-TrkC (NTRK3) antibody and g-bind beads (Amersham Pharmacia Biosciences), followed by immunoblotting with anti-Src antibody.C, NTRK3 kinase activity is required for the EN interaction with c-Src.EN-expressing NIH3T3 cells or vector control cells were pretreated with K-252a for 2 h.Cell extracts were immunoprecipitated using anti-TrkC (NTRK3) antibody, followed by immunoblotting with anti-Src antibody.D, EN increased phosphorylation of c-Src in vivo.E, identification of EN-c-Src complexes in human primary tumors.Primary tumor tissue extracts from EN-positive CMN and CFS were subjected to immunoprecipitation using anti-cSrc antibody or IgG as a control, followed by immunoblotting with anti-TrkC (NTRK3) antibody.Whole-tissue lysates from primary tumors also were probed for EN, c-Src, phospho-c-Src, and h-actin as a loading control.

Figure 2 .
Figure 2. Inhibition of c-Src activation blocks soft agar colony formation of ETV6-NTRK3-expressing NIH3T3 cells.A, The NIH3T3 cell line stably expressing EN as well as control NIH3T3 cells (MSCV ) were grown in medium containing 9% serum and were treated with various doses of SU6656, vehicle control (DMSO) for 12 h.Cells were then photographed in time at Â200 magnification.B, soft agar colony forming assay of NIH3T3 cells stably expressing EN.Monolayer cells were trypsinized, washed, and plated in medium containing 0.2% agar to assess anchorage-independent growth.Results are presented as the number of macroscopic colonies formed at 3 wks after plating.C, anchorage-independent spheroid growth of NIH3T3 cells stably expressing EN as well as control NIH3T3 cells.

Figure 3 .
Figure 3. Immunoprecipitation analyses of overexpressed EN and its interaction with c-Src proteins in 293T cells.A, top, a schematic representation of full-length and truncated c-Src proteins.B and C, immunoprecipitation and immunoblot analyses of HA-tagged c-Src protein (B) or Myc-tagged c-Src protein (C) interacting with V5-tagged EN.

Figure 4 .
Figure 4. EN interacts with c-Src and IRS-1.A, V5 tagged EN was cotransfected into 293T cells with the HA-tagged IRS-1 and Myc-tagged c-Src constructs.Cell extracts were subjected to immunoprecipitation using anti-HA antibody and g-bind beads (Amersham Pharmacia Biosciences), followed by immunoblotting with anti-TrkC(NTRK3) antibody or anti-V5 antibody.The expression of c-Src, IRS-1, or EN was monitored as indicated.B, HA-tagged IRS-1 was cotransfected into 293T cells with the Myc-tagged c-Src constructs.C, END614 associates with the c-Src protein.HA-tagged c-Src was cotransfected into 293T cells with V5-tagged EN or the EN mutant END614 construct.D, generation of SYF and SYF-Src cells stably expressing V5-tagged EN.Expression of EN protein was examined by immunoblotting in SYF and SYF-Src cells expressing V5-tagged EN.E, c-Src expression in c-Src-deficient MEF cells enhances the EN-IRS-1 complex formation.Cell extracts were subjected to immunoprecipitation using anti-V5 antibody and g-bind beads (Amersham Pharmacia Biosciences), followed by immunoblotting with anti-IRS-1 antibody.The expression of V5-EN or IRS-1 was monitored as indicated.F, suppression of c-Src expression by stable siRNA decreases the level of IRS-1 bound to EN in NIH3T3 cells stably expressing EN.ETV6-NTRK3, c-Src-siRNA-expressing or control NIH3T3 cells were serum starved overnight in 0.5% serum and then stimulated with 9% calf serum-DMEM for 1 h.Cell extracts were subjected to immunoprecipitation using anti-V5 antibody or anti-IRS-1, followed by immunoblotting with anti-IRS-1 antibody or anti-phospho-tyrosine antibody.The expression of V5-EN or IRS-1 was monitored as indicated.

Figure 5 .
Figure5.The SU6656 c-Src inhibitor, but not LY294002 Akt inhibitor, reduces the level of MEK phosphorylation, Akt phosphorylation, and cyclin D1 in ETV6-NTRK3-expressing NIH3T3 cells.A, MSCV or ETV6-NTRK3-transformed NIH3T3 cells were grown for 48 h in 9% calf serum-DMEM and then incubated with or without the c-Src family tyrosine kinase inhibitor SU6656 (2 and 5 Amol/L) or DMSO (0) for 1 h.Total cell lysates were prepared from the untreated and treated cells, and Western blotting was done using antibodies against P-MEK, P-Akt, and cyclin D1.B, MSCV or ETV6-NTRK3expressing cells were grown for 48 h in 9% calf serum-DMEM and then grown in 9% calf serum-DMEM with or without LY294002 (25 Amol/L) for 1 h.Whole cell lysates were prepared, and Western blotting was done using phospho-Src and phospho-Akt antibody.C, suppression of c-Src expression by stable siRNA decreases elevated levels of phosphorylated Akt in NIH3T3 cells stably expressing ETV6-NTRK3.ETV6-NTRK3, c-Src-siRNA-expressing or control NIH3T3 cells were serum starved overnight in 0.5% serum and then stimulated with 9% CS-DMEM for 1 h.Whole cell lysates were prepared for Western blotting and probed with antibodies against the phosphorylated forms of Mek1/2 (Ser 217/221 ), phosphorylated Akt (Ser 473 ), and cyclin D1.D, soft agar colony-forming assay.Soft agar colony-forming assay of EN control-siRNA, EN c-Src-siRNA-1, and EN c-Src-siRNA-2.Monolayer cells were trypsinized, washed, and plated in medium containing 0.2% agar to assess anchorage-independent growth.Results are presented as the number of macroscopic colonies formed at 3 wks after plating.

Figure 6 .
Figure 6.c-Src is required for EN-induced Akt activation in c-Src-deficient MEF cells.A, EN-expressing SYF cells exhibit elevated levels of phosphorylated Mek1 and cyclin D1, but not phosphorylated Akt.EN-expressing or control SYF cells were serum starved overnight in 0.5% serum and then stimulated with (+) and without (À) 9% calf serum-DMEM for 1 or 6 h.Data are representative of three experiments.B, EN-expressing SYF-Src cells exhibit elevated levels of phosphorylated Mek1, phosphorylated Akt, and cyclin D1.SYF cells and EN-expressing or control SYF-Src cells were serum starved overnight in 0.5% serum and then stimulated with (+) and without (À) 9% calf serum-DMEM for 1 or 6 h.Whole cell lysates were prepared for Western blotting and probed with antibodies against the phosphorylated forms of Mek1/2 (Ser 217/221 ), phosphorylated Akt (Ser 473 ), cyclin D1, and h-actin (as a loading control).C and D, the inhibitor U0126 reduces cyclin D1 levels in EN-expressing SYF and SYF-c-Src cells.Control and EN-expressing cells were serum starved overnight in 0.5% calf serum-DMEM and then grown in 9% calf serum-DMEM F U0126 (10 mmol/L) for 24 h.Whole cell lysates were prepared, and immunoblotting was done using a cyclin D1 antibody.Equal loading was determined by probing the blots with an antibody against h-actin.