Ret-mediated Mitogenesis Requires Src Kinase Activity¹ Free

The RET proto-oncogene encodes a TK³ transmembrane receptor, Ret (1). Inactivating mutations of RET are responsible for Hirschsprung’s disease (2, 3, 4, 5). It has been demonstrated that transforming growth factor-β-related neurotrophic factors, including GDNF and neurturin, stimulate TK activity of Ret. A family of GPI-linked proteins, including GFR α-1, GFR α-2, GFR α-3, and GFR α-4, mediates high-affinity ligand binding and ligand-induced Ret phosphorylation (6, 7, 8, 9). Oncogenic activation of RET can occur through multiple mechanisms. Gene rearrangements leading to the fusion of its TK-encoding domain with the 5′ terminal region of other genes, generate the RET/PTC oncogenes in human thyroid papillary carcinomas (10). Specific point mutations of RET are responsible for the MEN syndrome types 2A and 2B and familial medullary thyroid carcinoma. MEN2A mutations, which involve cysteine residues of the extracellular domain of the protein, induce ligand-independent dimerization and constitutive activation of the Ret receptor. MEN2B mutation is a Met-918→Thr substitution, which likely activates Ret through a conformational change of its catalytic core and through altering its substrate specificity. Other mutations in the TK domain of Ret have been found that are responsible for a small number of MEN2B and familial medullary thyroid carcinoma cases (11). Activated RET isoforms can transform NIH3T3 and thyroid cells and induce the appearance of a neuronal-like phenotype in neuroectodermal cells (12, 13, 14, 15, 16, 17).

Activation of growth factor receptors leads to the binding and activation of several molecules that are involved in the signal transduction cascade (18). Among the effector proteins that bind directly or indirectly to Ret are PLC-γ (19, 20), the Shc and Grb2 adaptor (21, 22), the Grb7 and Grb10 proteins (23, 24), and the Enigma protein (25, 26). Ret is able to activate the ras/mitogen-activated protein kinase pathway by phosphorylation and recruitment of the adaptors Grb2 and Shc to the plasma membrane (21, 22). Furthermore, activated Ret stimulates Jun kinases in certain cell lines (27).

The Src cytoplasmic TKs can bind to and are activated by membrane receptors. Most of the members of this family are expressed in a cell type-restricted fashion, whereas three of them (Src, Fyn, and Yes) are ubiquitous (28). Several studies show that Src family kinases are implicated in signal transduction mediated by transmembrane receptors, which are devoid of catalytic activity, such as T-cell receptor and FcgRII receptor (29). In addition, receptors with intrinsic TK activity (EGFR, c-erbB2/Neu, PDGF receptor, and colony-stimulating factor-1 receptor) have been shown to associate and/or activate Src, Fyn, and Yes through the SH2 domains of Src, Fyn, and Yes (30, 31, 32, 33). Activation of Src kinase is required for PDGF- and EGF-mediated signaling (34, 35, 36).

Here, we show that different cell lines expressing oncogenic forms of Ret possess high Src kinase activity levels. Ligand stimulation of Ret resulted in activation of Src. Src kinase was found to coprecipitate in vivo with activated Ret. The binding between the two molecules is mediated by Src SH2 domain because an isolated Src SH2 was able to bind to Ret in vitro. Finally, a dominant negative form of c-Src was able to inhibit the mitogenic activity of Ret when it was microinjected in NIH3T3 and PCCl3 cells, indicating that Src (or Src-related) kinase activity is required for Ret-mediated mitogenic signaling.

NIH3T3, NIH RET/MEN2A, NIH RET/MEN2B (13), NIH RET/PTC1 ((37)), EGFR/RET (19), NIH RET (38), and v-Src-expressing fibroblasts (39) have been described previously. They were grown in DMEM (Life Technologies, Inc.) supplemented with 10% calf serum. PCCl3 and PCCl3 RET/PTC1 (12) were maintained in Coon’s modified F12 medium supplemented with 5% calf serum (Life Technologies, Inc.) and six growth factors (thyrotropin, hydrocortisone, insulin, transferrin, somatostatin, and glycyl-histidyl-lysine; Sigma Chemical Co.). The long terminal repeat RET/MEN2A D100 was generated by recombinant PCR as described (5). A forward primer mapping on the RET sequence upstream the BglII site (position: 2690–2710) was used (5′-CCATGGGCGACCTCATCT-3′). The downstream primer was designed to delete the COOH-terminal residues of Ret, including Tyr-1015, Tyr-1029, and Tyr-1062 (5′-ACGCGTCTAGTCTCTGCCTCTTAA-3′) and contained a MluI site for the cloning in the long terminal repeat vector. The cell line NIH3T3 RET/MEN2A D100 was obtained by calcium phosphate coprecipitation transfection technique. Transfected cells were subjected to selection by the addition of mycophenolic acid, as described elsewhere (19). The levels of Ret/MEN2A D100 protein in the mass population were similar to those of the wild-type Ret/MEN2A-expressing cells (data not shown). The cell lines F212 T1 and F212 T23 were obtained from two mammary tumors arising in RET/PTC1 transgenic mice (40). They were cultivated in Coon’s modified F12 medium supplemented with 5% calf serum (Life Technologies, Inc). The HC11 is a normal mouse mammary epithelial cell line (41). The plasmids pSG5 Src and pSG5 Src K- were kind gifts of S. Courtneidge (Sugen, Inc., Redwood City, CA).

Confluent cells were serum starved for 12–24 h and, when indicated, stimulated with 100 ng of EGF or GDNF as described (19). After two washes with cold PBS, cells were lysed in a buffer containing 50 mm Tris-HCl (pH 8.0), 150 mm NaCl, 1% (v/v) NP40, 1 mm EDTA, 50 mm NaF, 20 mm sodium pyrophosphate, 1 mm sodium vanadate, 2 mm phenylmethylsulfonyl fluoride, and 0.2 mg/ml each aprotinin and leupeptin. Lysates were centrifuged at 10,000 × g for 30 min and immunoprecipitated by incubating 1 mg of total proteins with anti-Src (Ab-1; Oncogene Science) antibodies for 60 min at 4°C. Immunocomplexes were recovered by incubation with protein G-Sepharose beads (GammaBind G Sepharose; Pharmacia) on a rotating platform at 4°C for 60 min. After three washes with lysis buffer, the immunoprecipitates were washed with kinase buffer [20 mm Tris-HCl (pH 7.0)-5 mm MgCl₂] and resuspended in 30 ml of kinase buffer, 10 mCi of [γ-³²P]ATP (>10,000 Ci/mmol, Amersham), and 10 mm cold ATP. After 30 min of incubation at room temperature, the beads were washed twice with lysis buffer, and the reaction was terminated by adding an equal volume of SDS-gel loading buffer [62.5 mm Tris-HCl (pH 6.8), 2% SDS, 5% glycerol, 0.7 m 2-mercaptoethanol, and 0.25% bromphenol blue]. When indicated, 5 mg of acid-denatured enolase were added to the reaction (Boehringer Mannheim Biochemicals). The samples were electrophoresed on SDS-10% polyacrylamide gel. After the run, gels were incubated three times for 30 min in a fixing solution (20% methanol-10% acetic acid), dried, and processed for autoradiography with phosphor screens.

Total cellular proteins were quantitated by a modified Bradford assay (Bio-Rad). To evaluate Src expression, equal amounts of proteins were immunoprecipitated with the anti-Src antibodies (Ab-1; Oncogene Science). Antibodies purchased from Santa Cruz were used to evaluate p27^kip1 (C-19), cyclin E (M-20), and D3 (C-16) expression. Proteins were separated by a 10% SDS polyacrylamide gel and transferred onto a PVDF membrane (Millipore). After a 60-min incubation in 1% nonfat dry milk blocker (Bio-Rad), the membrane was probed with primary antibodies. Detection of the immunocomplexes was obtained by an enhanced chemiluminescence system (ECL; Amersham), following manufacturer’s instructions.

In vivo association experiments were conducted by immunoprecipitating 1 mg of total cell lysate with the anti-Src antibody. The immunocomplexes were electrophoresed on a 7.5% SDS acrylamide gel and transferred onto a PVDF membrane (Millipore), which was then incubated with rabbit polyclonal antibodies directed against the TK domain of the Ret (19). Proteins were then visualized by an enhanced chemiluminescence system (ECL; Amersham). For in vitro binding assays, cell lysates were incubated either with GST or with GST-SH2-Src protein (5 mg) linked to agarose beads. Following incubation for 3 h at 4°C on a rotating platform, the complexes were washed several times in lysis buffer, resuspended in SDS loading buffer, separated on a 7.5% SDS acrylamide gel, and immunoblotted with the anti-Ret antibodies, as described above. The plasmid expressing the GST-SH2-Src recombinant protein was provided by G. Superti-Furga and S. Gonfloni (EMBL, Heidelberg, Germany).

NIH EGFR/Ret cells were seeded on glass coverslips in DMEM containing 10% calf serum and grown to 60% confluence. The medium was then replaced with DMEM supplemented with 0.5% calf serum, and the cells were incubated for another 30–48 h. The purified plasmids were injected into cell nuclei at a concentration of 25–50 ng/ml using an automated microinjection system (AIS; Zeiss). Six h later, cells were stimulated with EGF (Upstate Biotechnology) at 300 ng/ml to induce DNA synthesis. BrdUrd (Sigma) was added to the culture medium to a final concentration of 100 mm, and the labeling procedure was carried out for 18–20 h. Cells were then fixed for immunostaining. Codetection of Src and BrdUrd in the cells was essentially performed as described previously (42, 43). In brief, cells were fixed in paraformaldehyde, permeabilized with 0.1% Triton X-100, and incubated with a primary rabbit antiserum directed against c-Src. After several washes in PBS, rhodamine-conjugated antirabbit IgG was added to identify the Src expressing cells. A fluorescein-conjugated anti-BrdUrd mouse monoclonal antibody (Boehringer Mannheim Biochemicals) was used to detect the fraction of cells that were in S phase. All coverslips were finally washed in PBS containing Hoechst 33258 (final concentration: 1 mg/ml; Sigma), rinsed in water, and mounted in Moviol on glass slides. The fluorescent signal was visualized with an epifluorescent microscope (Axiovert 2; Zeiss). BrdUrd incorporation was measured in injected and uninjected cells, stimulated or not with EGF. In each experiment, at least 60 Src K+ cells were counted and compared to 400 nonmicroinjected cells from the same coverslip. The results shown are the means of three independent experiments. Variations were <25% of the mean. Microinjection of RET/PTC1-expressing NIH3T3 and PCCl3 RET/PTC1 was performed as described above with a few modifications. SrcK− and SrcK+ plasmids were microinjected in proliferating cells maintained in DMEM or in Coon’s modified F12 medium, respectively, supplemented with 5% calf serum (Life Technologies, Inc.). The day after, BrdUrd was added for 1 h. The cells were then fixed, permeabilized, and stained both for Src expression and BrdUrd incorporation as described above. BrdUrd incorporation was evaluated in the injected and uninjected cells. In each experiment, at least 100 Src K+ cells were counted and compared to 400 nonmicroinjected control cells from the same coverslip. The results shown are the mean of two independent experiments.

The GDNF is a functional ligand for Ret. GDNF binds to GFR α-1, a GPI-linked cell surface molecule, which, in turn, activates Ret. We have shown that NIH3T3 coexpressing Ret and GFR α-1 undergo DNA replication upon GDNF triggering (38). To evaluate whether stimulation of Ret with its physiological ligand was able to induce Src activation, Src kinase was immunoprecipitated using specific antibodies from unstimulated or GDNF-treated NIH-Ret cells and an in vitro kinase assay was performed. As shown in Fig. 1,A, GDNF triggering resulted in Src activation. We have previously shown that an EGFR/Ret chimeric receptor, when exogenously expressed in NIH3T3 cells, is able to transduce mitogenic and transforming signals upon EGF triggering (19). To determine whether c-Src was involved in Ret signaling, Src kinase was immunoprecipitated using specific antibodies from quiescent wild-type and EGFR/Ret-expressing NIH3T3 cells (NIH EGFR/Ret), stimulated or not with EGF. Both Src autophosphorylation and its ability to phosphorylate an exogenous substrate, enolase, were measured. As shown in Fig. 1,B, EGF stimulation was able to activate Src in NIH EGFR/Ret but not in the parental cells, which express negligible levels of endogenous EGFR (19). Immunoprecipitation of the same lysates with a preimmune serum gave negative results. Subsequently, time course evaluation of Ret and Src activation was performed in NIH EGFR/Ret cells (Fig. 1,C). Cells were treated with EGF and harvested at different times during stimulation. Src (Fig. 1,C, top) and Ret (Fig. 1,C, bottom) activation were measured. Ret tyrosine phosphorylation reached a peak at 5 min and started decreasing at 15 min when Src reached its maximal activation. Furthermore, we studied molecular markers of S-phase entry upon EGF triggering of NIH EGFR/Ret cells. In particular, we quantitated the amounts of a cyclin-dependent kinase inhibitor, p27^kip1, cyclin D3, and cyclin E. As shown in Fig. 1 D, p27^kip1 levels were high in quiescent cells, decreased after 6 h of stimulation with EGF, and remained low thereafter. Cyclin D3 accumulation peaked at 90 min, remained stable until 12 h, and then decreased. Cyclin E levels peaked at 12 h and were low after 24 h. These kinetics indicated that Ret-mediated activation of Src kinase precedes the early G₁ events required for cell cycle progression.

To test whether different oncogenic forms of RET activate Src kinase, endogenous c-Src activity was measured in immunocomplex kinase assays in serum-starved NIH3T3 cells expressing RET/MEN2A(Cys634Tyr), RET/MEN2B(Met918Thr) (13), and RET/PTC1(H4-Ret) (37). Positive control fibroblasts expressing v-Src were used. Autophosphorylation of c-Src was higher in NIH3T3 cells expressing the activated Ret isoforms than in parental cells (∼10-fold; Fig. 2,A). The increase in Src autophosphorylation activity was paralleled by an analogous increase in the phosphorylation of the exogenous substrate enolase (data not shown). As shown by the immunoblot reported in Fig. 2,A, bottom, this effect was not due to increased expression levels of the endogenous Src kinase, which were, indeed, comparable in all of the cell lines tested. We have reported previously that the exogenous expression of RET/PTC1 in a rat thyroid cell line, PCCl3, has a mitogenic effect (12). To evaluate whether Src activation occurred also in RET/PTC1 transformed PCCl3 (PCCl3 RET/PTC1), Src kinase activity was measured. Src autophosphorylation was higher in PCCl3 RET/PTC1 cells than in untransfected parental cells (Fig. 2,B, top). The expression levels of endogenous Src were comparable in the cell lines tested, as shown by the immunoblot reported in Fig. 2,B, bottom. PCCl3 cells expressing a constitutively active v-Src oncogene were used as a positive control (39). Transgenic mice expressing the RET/PTC1 oncogene under the transcriptional control of the H4 gene promoter were generated in our laboratory. Despite the fact that the RET/PTC1 transgene was expressed in several tissues, the animals developed a restricted pattern of neoplasias: mammary adenocarcinomas and adnexial tumors (40). Two cell lines, F212T1 and F212T23, derived from two independent mammary adenocarcinomas that occurred in RET/PTC1 transgenics, were used to investigate if RET/PTC1 tumor induction in vivo correlated with an increase in endogenous c-Src activity. Levels of c-Src activity in the F212T1 and F212T23 cells were evaluated and compared to those of normal mammary cells and mammary tissue from nontransgenic littermates. As shown in Fig. 2 C, c-Src autophosphorylation was higher in the tumor-derived cell lines than in normal mammary cells (HC11 cells) and in normal mammary tissue, and this event was not due to increased expression levels of endogenous c-Src, as demonstrated by Western blot analysis.

Some TK receptors have been shown to directly interact with and activate members of the Src family (28). To test if this was also the case of Ret, equal amounts of proteins were immunoprecipitated with anti c-Src antibodies from quiescent NIH EGFR/Ret cells stimulated or not with EGF. Immunocomplexes were then separated on a denaturing polyacrylamide gel, transferred to a PVDF membrane, and probed with antibodies directed against the COOH-terminal portion of the Ret protein (19). Association between c-Src and Ret was readily detectable (Fig. 3,A). A preimmune serum was not able to coprecipitate the Ret protein from stimulated cells (Fig. 3,A). Src associated only with tyrosine-phosphorylated Ret because unstimulated Ret was not coprecipitated by the anti-Src antibody (Fig. 3,A). The requirement of Ret phosphorylation for association with Src, suggested that this interaction was mediated by the Src SH2 domain. To test this hypothesis, we used a GST-SH2-Src (GST-SH2-Src) recombinant protein. Fig. 3,B shows that the GST-SH2-Src, but not the GST alone, was able to associate to activated Ret in vitro. As expected, unstimulated Ret was unable to bind to Src SH2 (Fig. 3 B).

The Ret COOH-terminal tail contains three of these autophosphorylation sites: Tyr-1062, Tyr-1029, and Tyr-1015. Tyr-1062 is the binding site for Shc (44, 45, 46), whereas Tyr-1015 is the docking site for PLC-γ (14, 20). On the basis of the optimal consensus sequence for Src SH2 binding (47), Tyr-1015 (YLDL) and, less likely, Tyr-1029 (YDDG) could represent candidates for Src binding. We then investigated whether these tyrosines were involved in Src binding and activation by using the mutant Ret/MEN2A Δ100. In this mutant, the last three tyrosine residues of Ret/MEN2A have been deleted. As expected, this mutant was unable to bind to Shc and PLC-γ (data not shown). Ret/MEN2A Δ100 was still able to coimmunoprecipitate Src (Fig. 4,A), bind to GST-SH2-Src (Fig. 4,B), and activate Src kinase (Fig. 4 C).

To investigate whether c-Src activity was required for Ret-mediated mitogenesis, we applied a microinjection technique. Serum-starved NIH EGFR/Ret cells, when stimulated with EGF, enter S phase, as measured by thymidine incorporation (19). Serum-deprived NIH EGFR/Ret cells were stimulated or not with EGF in the presence of BrdUrd, and S-phase entry was measured by counting cells stained with anti-BrdUrd-specific antibodies. Arrested NIH EGFR/Ret cells showed only a very low level of BrdUrd incorporation. Upon stimulation with EGF, almost 50% of the cells entered S phase (Fig. 5,A). Arrested NIH EGFR/Ret cells were microinjected with a dominant negative form of Src (Src K−; Ref. 43). After 6 h, they were stimulated with EGF for 18 h. The average results of three independent experiments, in which at least 60 Src K− microinjected cells were counted, are shown in Fig. 5,A. The entrance into S phase was inhibited by Src K−; only 8% of injected cells incorporated BrdUrd. This effect was specific because serum-induced BrdUrd incorporation was not affected by the expression of Src K− (data not shown). Moreover, microinjection of a plasmid expressing wild-type Src (Src K+) did not affect the mitogenic response to EGF. An example of these assays is reported in Fig. 5,B. Cells expressing Src K− were identified by immunostaining with anti-Src (Src). These cells did not incorporate BrdUrd, whereas surrounding cells did, as shown by the staining with anti-BrdUrd antibody (Fig. 4 B, BrdU). These experiments strongly support the concept that Src kinase activity is required for Ret-induced mitogenic response.

To investigate the involvement of Src in mitogenic signaling mediated by oncogenic Ret proteins, we used NIH3T3 and PCCl3 cells expressing the Ret/PTC1 oncoprotein. These experiments were carried on asynchronously growing cells: cells were microinjected with either Src K− or Src K+ plasmids, and BrdUrd incorporation was measured. As shown in Fig. 6, 50% of the cells entered S phase in normal growing conditions. Microinjection of Src K− but not Src K+ strongly inhibited BrdUrd incorporation of both cell types. Microinjection of Src K− in asynchronously growing parental NIH3T3 and PCCl3 did not affect the fraction of cells that entered S phase.

Here, we showed that Ret triggering is able to stimulate the activity of pp60 c-Src. A 2–3-fold increase of c-Src activity following Ret triggering was reproducibly measured; this is consistent with the increases observed for activated PDGFR, colony-stimulating factor-1 receptor, and erbB2/Neu (30, 31, 32, 33). We also show that, in NIH3T3, thyroid and mammary cell lines stably expressing activated forms of RET, levels of c-Src kinase activity are elevated in comparison to wild-type cells. The induction of c-Src activity correlated with the ability of Src to interact with tyrosine-phosphorylated Ret, as shown by coimmunoprecipitation experiments. Activated Ret was also able to bind c-Src in vitro, and the SH2 domain of c-Src alone was sufficient for this association.

Time-course experiments indicated that Src activation follows the peak of Ret activation. Moreover, Ret-dependent Src activation occurs before the first events of G₀-G₁ transition. This is consistent with the possibility that Src activation may be necessary for the induction of Ret-mediated cell cyle progression. Indeed, in PDGF-stimulated NIH3T3 fibroblasts, the block of Src activation inhibits cyclin E accumulation.⁴ Accordingly, a dominant inhibitory form of c-Src was able to inhibit the mitogenic activity of Ret when it was microinjected in EGF-stimulated NIH EGFR/Ret cells and of Ret/PTC1 in NIH3T3 and PCCl3 cells.

The SrcK− mutant used in our study is probably also active on other members of Src kinase family because its SH2 domain could theoretically compete with the binding to the receptor mediated by SH2 domains of closely related Src-like kinases. For this reason, we cannot exclude that the activity of other Src-like kinases may be required for Ret-mediated biological activity. In support of this hypothesis, we have observed that the EGFR/Ret chimera is also an efficient inducer of the c-Fyn kinase.⁵ Similarly, in addition to Src, other Src-like kinases are required for PDGF- and EGF-mediated signaling in NIH3T3 fibroblasts. Indeed, microinjection of antibodies directed against Src and Fyn or plasmids expressing dominant negative Src and Fyn inhibited PDGF-induced S-phase entry in NIH3T3 cells. On the other side, microinjection of Src dominant-negative plasmids or of neutralizing antibodies does not interfere aspecifically with cell functions necessary for S-phase entry. Indeed, serum stimulation of NIH3T3 cells is unaffected by the block of Src kinase activity (data not shown). In fact, only some growth factors require functional Src family kinases to transmit mitogenic responses in these cells: lysophosphatidic acid- and bombesin-induced DNA synthesis, for instance, is unaffected by anti-Src-neutralizing antibodies (48, 49).

Deletions or substitutions of specific tyrosine residues in the intracellular domain of Ret have been performed. Substitution of Tyr-1062 with phenylalanine has been shown to severely impair mitogenic and transforming activity of RET/PTC2, RET/MEN2A, and RET/MEN2B by decreasing the ability of Ret to associate with Shc adaptor protein and to activate the ras/mitogen-activated protein kinase pathway⁶(45, 46). Replacement of Tyr-1015, identified as the binding site for PLC-γ, with phenylalanine, reduced oncogenic activity of Ret/PTC2 (20, 50). We show here that a deletion mutant, Ret/MEN2A Δ100, in which these tyrosines have been abolished, retains its ability to bind and activate c-Src. Thus, tyrosines of the COOH-terminal tail are not involved in binding Src, and the ability of Ret to bind Src can be separated from its ability to associate with PLC-γ and Shc. Other tyrosine residues mapping either in the catalytic core or in the juxtamembrane domain of Ret have been demonstrated to be phosphorylated and/or relevant for Ret-mediated mitogenesis as Y687, Y826, Y900, and Y905 (23, 24, 44, 50); their potential role in mediating Src-binding needs to be explored.

Elevated levels of c-Src activity have been detected in many types of human tumors, including carcinomas of the breast and colon, whereas normal tissues have low activity (51, 52, 53, 54, 55). Furthermore, Src kinase levels are significantly increased in highly metastatic cell lines in comparison to primary tumors. It has been reported that, in these tumors, receptor TKs participate to activation of c-Src above basal levels. In breast cancer, c-Src is found in association with c-erbB2/Neu, and it has been proposed that this activation might be linked to the development of the metastatic phenotype (32, 56). Our data demonstrate that RET/PTC1-induced mammary tumors possess an elevated c-Src activity. Similarities between Ret and c-erbB2/Neu have been reported in previous studies; for instance, the EGFR/Ret chimera and an activated erbB2/Neu kinase displayed a poor ability to induce survival and proliferation of the 32D hematopoietic cells and a strong transforming ability in NIH3T3 cells (19, 57, 58). Our data strengthen the hypothesis that Ret and c-erbB2 couple with similar pathways and that one of this pathways involves the activation of c-Src.

We thank Nina A. Dathan for her early contribution to the expression of RET/MEN2A mutants and generation of cell lines, Giovanni Santelli for his help, and Giulio Superti-Furga and Stefania Gonfloni for providing the GST-SH2-Src plasmid. We also thank F. D’agnello and M. Berardone for the artwork and F. Sferratore for excellent technical assistance.