Transforming ability of MEN2A-RET requires activation of the phosphatidylinositol 3-kinase/AKT signaling pathway.

The RET gene codes for a receptor tyrosine kinase that plays a crucial role during the development of both the enteric nervous system and the kidney. Germ line missense mutations at one of six codons specifying extracytoplasmic cysteines are responsible for two related cancer disorders as follows: multiple endocrine neoplasia type2A (MEN2A) and familial medullary thyroid carcinoma (FMTC). MEN2A and FMTC mutations result in a constitutive catalytic activity and as a consequence convert RET into a dominantly acting transforming gene. Although it has been shown that RET-MEN2 mutants activate several transduction pathways, their respective contribution to the neoplastic phenotype remains poorly understood. Over the past few years, it has become increasingly clear that the transforming ability of several viral and cellular oncoproteins depends on their capacity to activate phosphatidylinositol 3-kinase (PI3K). We now report that RET carrying a representative MEN2A mutation at Cys-634 (termed RET-MEN2A) activates PI3K and its downstream effector, the serine/threonine kinase AKT/protein kinase B. Previous studies have demonstrated that mutation of Tyr-1062, which is the intracellular docking site for Shc and Enigma on RET, abolishes the RET-MEN2A transforming activity. We provide evidence that mutation of Tyr-1062 abrogates the binding of the p85 regulatory subunit of PI3K to RET-MEN2A and the subsequent stimulation of the PI3K/AKT pathway. Furthermore, infection of rat fibroblasts with a retrovirus expressing a dominant-interfering form of PI3K suppresses RET-MEN2A-dependent transformation, whereas overexpression of AKT enhances the RET-MEN2A oncogenic potential. In summary, these data are consistent with the notion that RET-mediated cell-transforming effect is critically dependent on the activation of the PI3K/AKT pathway.

addition to RET, this complex contains glycosylphosphatidylinositol-anchored receptors called GFR␣s that directly interact with the different ligands and induce homodimerization of RET, thereby leading to the activation of the tyrosine kinase (2). To date, four homologous GFR␣s have been identified, and the ligand specificity of the multisubunit receptor is dictated by the nature of the glycosylphosphatidylinositol-bound protein since GFR␣-1, -2, -3, and -4 exhibit a higher affinity for GDNF, neurturin, artemin, and persephin, respectively (2). Gene knock out studies have revealed that null mice lacking either RET, GFR␣-1, or GDNF die soon after birth and share strikingly analogous phenotypes including renal agenesis and absence of the enteric nervous system in the intestine (2). Evidence has been further provided that GDNF is both a neurotrophic and a urotrophic factor capable of both protecting enteric neurons from apoptotic cell death and promoting growth and branching of the ureteric bud, respectively (2).
Germ line mutations of RET are responsible for multiple endocrine neoplasia type 2 (MEN2), a dominantly inherited cancer syndrome that encompasses three related clinical subtypes, each typified by the involvement of a specific spectrum of endocrine tissues involved. MEN type 2A (MEN2A) is characterized by the association of medullary thyroid carcinoma (MTC), a malignant tumor arising from calcitonin-secreting C-cell, pheochromocytoma, and more occasionally parathyroid hyperplasia or adenoma. MEN type 2B (MEN2B) is a more rare and a more severe syndrome defined by the presence of MTC, pheochromocytoma, mucosal neuromas, hyperplasia of the enteric nervous plexuses, and skeletal abnormalities. Finally, in the familial form of MTC (FMTC), thyroid cancer is the sole clinical manifestation (see Ref. 3 for review). Missense mutations at one of six cysteines codons (Cys-609, -611, -618, -620, -630, and -634), which result in the substitution of any one of these extracytoplasmic cysteines by a different amino acid, are responsible for the majority of MEN2A and a large fraction of FMTC cases (3)(4)(5)(6). Moreover, FMTC and rare cases of MEN2A have been also ascribed to five mutations singly affecting amino acids within the tyrosine kinase domain (3). MEN2B is exclusively caused by two distinct mutations involving residues located within the kinase catalytic domain and resulting in the replacement of either Ala-883 for a Phe or Met-918 for a Thr, the latter mutation being found the most frequently (3).
All MEN2 mutations tested so far have been shown to induce a ligand-independent constitutive activation of RET (7,8). Cysteine mutations result in the aberrant RET homodimerization via the formation of inter-chain disulfide bridges (7)(8)(9)(10), whereas MEN2B mutations are thought to alter the substrate specificity of the tyrosine kinase (8,11,12). In addition, transgenic mice expressing an MEN2A/FMTC allele in the thyroid C-cell developed C-cell hyperplasia and MTC (13), whereas mice expressing MEN2B allele under the control of the dopamine ␤-hydroxylase promoter displayed ganglioneuromas and renal malformations (14). Taken together, these results demonstrate that MEN2 mutations transform RET into a dominantly acting oncogene and suggest that the uncontrolled stimulation of signaling pathways downstream of RET accounts for its transforming ability. Upon RET activation, the autophosphorylated tyrosine residues Tyr-905, Tyr-1015, Tyr-1062, and Tyr-1096 act as docking sites for the transduction effectors Grb7/Grb10/Grb14, phospholipase C␥, Shc/Enigma, and Grb2, respectively (15)(16)(17)(18)(19)(20)(21)(22)(23). The Tyr-1062 residue is an essential site since its mutation impairs the oncogenic potential of all MEN2 forms of RET tested so far (16,24). Furthermore, it is known that RET-MEN2 relays an intracellular signal by activating several signaling cascades including the Ras/mitogen-activated protein kinase and the Jun and Src kinases (25)(26)(27)(28)(29). However, evidence that the sustained stimulation of these respective pathways contributes to the RET-MEN2 transforming capacity is still limited.
Phosphoinositide 3-kinases (PI3K) are crucial transducers that control multiple biologic responses, such as mitogenesis, protection from programmed cell death, and cellular motility (reviewed in Refs. 30 and 31). Class I of PI3K is a heterodimeric enzyme comprising a regulatory subunit of 85 kDa, p85, and a catalytic component of 110 kDa, p110. The p85 subunit contains two SH2 domains that can interact with phosphotyrosine. PI3K binds activated receptor tyrosine kinases either directly via the interaction of p85 to phosphorylated tyrosines or indirectly through the association of p85 with other transduction effectors possessing tyrosine docking sites, such as IRS-1, Gab-1, or c-Cbl (30,31). The translocation of PI3K to the inner face of the plasma membrane locates the enzyme closer to upstream regulators (such as Ras) and recruits PI3K in proximity to its lipid substrates. Once activated, class I PI3Ks catalyze the phosphorylation of phosphoinositide on position 3 of the inositol ring and generate PtdIns(3,4)P 2 and PtdIns(3,4,5)P 3 . These lipid products act as binding sites for transduction effectors, and this interaction regulates their subcellular localization and modulates their activity. The serine/ threonine kinase protein kinase B/AKT has been identified as a crucial downstream effector that mediates the effects of PI3K on cell survival, cell proliferation, and on the response to insulin (32,33).
We now report that Rat1 cell line expressing an MEN2A form of RET displays a constitutive PI3K activity. Furthermore, mutation of Tyr-1062 blocked the RET-mediated activation of PI3K, and expression of a dominant-interfering mutant form of PI3K drastically reduced the transforming ability of RET-MEN2A.

Construction of Vectors and Production of Ecotropic Retroviruses-
The construction of the pBabe Puro retroviral vectors containing the cDNAs coding for the human RET9 and RET9-MEN2A was described previously (34,35). The USE Mutagenesis Kit (Amersham Pharmacia Biotech) was used to introduce the mutations Y981F and Y1062F with the following primers: 5Ј-CTGCAGCGAGGAGATGTTCCGCCTGAT-GCTGCAATGCTG-3Ј and 5Ј-GAAAACAAACTCTTTGGTAGAATTTC-CCAT-3Ј.
The vectors expressing ⌬p85 and HA epitope-tagged AKT were described previously (36,37). The cDNA coding for HA-AKT was subcloned in the pLXSN retroviral vector.
Ecotropic retroviruses were obtained by transient transfection of the retroviral vectors into the BOSC 23 packaging cell line according to the protocol described by Pear et al. (38). Forty-eight hours after transfection, the supernatants were harvested and filtered on 0.45-m pore size filters.
Cell Culture and Retroviral Infections-The Rat1 fibroblast cell line and the human embryonic kidney epithelial Bosc 23 cell line were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics. For retroviral infections, 2 ϫ 10 6 cells were seeded on 100-mm culture plates. The following day, cells were incubated for 7 h with 1 ϫ 10 6 infectious retroviral particles in the presence of 8 g/ml Polybrene. Forty-eight hours after infection, Rat1 cells were trypsinized and seeded at a dilution of 1:50 to 1:100 in complete medium containing 2 g/ml puromycin. Then, cells expressing LacZ, RET wt, or RET mutants were single cell-cloned.
For the experiments with ⌬p85 and AKT, Rat1 cell clones expressing either LacZ, RET wt, or RET-MEN2A were infected with recombinant retrovirus carrying the neo selectable marker and expressing the genes of interest. Mass population of antibiotic-resistant cells was selected with 400 g/ml of G418 (Life Technologies, Inc.). After 1 week of selection, G418-resistant cells were analyzed for protein expression and seeded in soft agar to assess their capacity to grow without anchorage.
Antibodies and Protein Analyses-The polyclonal anti-RET serum used in this study was described previously (35). The monoclonal antibody specific for antiphosphotyrosine (4G10) and the anti-p85 polyclonal antibody were purchased from Upstate Biotechnology, Inc. Detection of the phosphorylated form of AKT was performed by immunoblotting using a phospho-specific antibody that recognizes AKT only when phosphorylated at Ser-473 (New England Biolabs). A control antibody that recognizes AKT (phosphorylation-state independent) was also used as a control (New England Biolabs). All these commercially available reagents were used according to the manufacturers' recommendations. Phosphorylated FKHRL1 (39) was detected by Western blot analysis using antiphospho-specific antibodies that recognize FKHRL1 phosphorylated either on Thr-32 (diluted 1:500) or Ser-253 FKHRL1 (diluted 1:350). Western blot analysis using antibody FKHRL1 (diluted 1:500), which detects both phosphorylated and unphosphorylated forms of FKHRL1, was used as a control (39).
GST Fusion Affinity Precipitation-Cleared cell lysates containing 400 g of total cell proteins were incubated for 2 h at 4°C with 100 l of glutathione-Sepharose 4B (Amersham Pharmacia Biotech) and 3 g of GST p85/N-SH2 protein (Upstate Biotechnology, Inc.). The GST protein complexes were washed four times with lysis buffer, and the beads then were resuspended in 25 l of Laemmli buffer (62.5 mM Tris-HCl (pH 6.8), 20% glycerol, 2% SDS, 1.4 mM of ␤-mercaptoethanol, 20 g of bromphenol blue per ml) and then fractionated on SDS-7% polyacrylamide gel. Immunoblots were incubated with the anti-RET polyclonal serum previously described (diluted 1/100), incubated with an anti-rabbit conjugated to horseradish peroxidase (Amersham Pharmacia Biotech) and revealed by an enhanced chemiluminescence detection kit (ECL, Amersham Pharmacia Biotech).
Immunokinase Assay-The AKT catalytic activity was assayed by an in vitro immunokinase assay using histone 2B as a substrate. Comparable amounts of proteins were immunoprecipitated for 2 h at 4°C with the anti-AKT polyclonal serum (New England BioLabs). The kinase assay was essentially carried out according to Cheng et al. (40) with the following modifications. Immunoprecipitates were washed twice with lysis buffer and twice with kinase buffer (20 mM Hepes-NaOH, 10 mM MnCl 2 , 10 mM MgCl 2 , 20 mM ␤-glycerophosphate (pH 7.4)) and then incubated in 30 l of kinase buffer containing 10 Ci of [␥-32 P]ATP, 1 M cold ATP, and 20 g of histone 2B (Roche Molecular Biochemicals). After incubation at 30°C for 30 min, samples were fractionated on 12% acrylamide SDS-polyacrylamide gel electrophoresis, transferred onto polyvinylidene difluoride membrane (Immobilon P, Millipore), and then autoradiographed at Ϫ80°C. In vitro PI3K assays were performed as described previously (41) except that cells at confluence were starved for 4 h in Dulbecco's modified Eagle's medium without fetal calf serum.
Transformation Assay-The transformed phenotype was assessed by the capacity of cells to grow without anchorage. Rat1 cells were trypsinized, and 2.5 ϫ 10 3 cells were seeded into culture medium containing 10% fetal calf serum and 0.3% agar. Colonies were scored 3 weeks after plating.

PI3K Is Constitutively Activated in Fibroblasts Expressing
RET-MEN2A-Three RET isoforms containing 1114, 1106, and 1072 amino acids (referred as RET9, RET43, and RET51) that diverge in the carboxyl termini have been characterized (reviewed in Ref. 3). Rat 1 fibroblasts were infected with recombinant ecotropic retroviruses expressing either the short RET isoform (RET9) or an MEN2A form of RET (RET9 -634R) (Fig.  1A), and individual clones of puromycin-resistant cells were isolated. Two representative clones, one expressing a high level and the other one a lower level of exogenous RET wild type (RET wt) or RET-MEN2A, were expanded and further analyzed (Fig. 1B). Two RET species migrating at 150 and 170 kDa, which are known to correspond, respectively, to the incompletely glycosylated form present in the endoplasmic reticulum and to the fully glycosylated RET species (42), were detected in each cell clone (Fig. 1B). Consistent with previous reports, RET-MEN2A was constitutively active and exhibited a clear increase of the phosphotyrosine content compared with the RET wt (Fig. 1B, upper panel) (7,8). We also established several cell clones stably expressing RET-MEN2A carrying a mutation at Tyr-981 (Y981F) or Tyr-1062 (Y1062F) (Fig. 1A). As shown in Fig. 1B, in two of these independent cell clones the RET mutant protein was correctly synthesized, and the level of phosphorylation on tyrosine residues was approximately comparable to RET-MEN2A (Fig. 1B). As previously reported (16), Rat1 fibroblasts expressing RET-MEN2A or RET-MEN2A/ 981F displayed the morphologic changes characteristic of a transformed phenotype: cells were refractile, rounded, piled up, formed foci, and had the tendency to detach from the plates to grow without anchorage (data not shown). On the contrary, Rat1 cells expressing RET wt or RET-MEN2A/1062F formed the typical monolayer of fibroblasts and did not grow in soft agar (data not shown).
To examine the lipid kinase activity of PI3K in Rat1 cells expressing either RET wt or RET-MEN2A, proteins were immunoprecipitated with an antiphosphotyrosine monoclonal antibody, and the immunocomplex was subjected to an in vitro PI3K assay. In this assay the levels of the products of PI3K, i.e. PtdIns(3,4)P 2 and PtdIns(3,4,5)P 3 , were determined by thin layer chromatography. To reduce the basal level of PI3K activity due to growth factors contained in the serum of the culture medium, cells were serum-starved during 4-h prior lysis. As shown in Fig. 2, a marked increase of PI3K activity was detected in cell clones expressing RET-MEN2A compared with RET wt. Quantification of this effect revealed that PI3K activity was enhanced 4-and 2-fold in clones 1 and 2, respectively, and this difference correlated with the level of RET-MEN2A expression.
RET-MEN2A Interacts with the p85 Regulatory Subunit of PI 3-Kinase; Mutation of Tyr-1062 Disrupts RET-p85 Interaction-To determine whether RET forms a complex with PI3K, Rat1 cell clones were lysed with mild conditions of detergent to preserve protein-protein interactions. Cell lysates were subjected to an immunoprecipitation with anti-RET polyclonal serum and then Western-blotted with an anti-p85 antibody. By using these conditions, p85 was detected in RET-MEN2A immunoprecipitates but not or very weakly in RET wt and LacZ immunoprecipitates (Fig. 3).
To test whether the SH2 domains of p85 were involved in the RET-MEN2A-p85 interaction, we performed a "GST-pull down assay" using glutathione-Sepharose coated with a GST fusion protein containing the amino-terminal SH2 domain of p85 (GST-p85/N-SH2, Fig. 4A). Total cell lysates from cells expressing RET wt or RET-MEN2A were incubated with the GST-p85 fusion protein, and bound proteins were fractionated on an SDS-polyacrylamide gel and analyzed by Western blot using an anti-RET polyclonal antibody. As shown in Fig. 4B, RET-

PI3K Involvement in RET-MEN2 Oncogenic Activity
MEN2A specifically bound to GST-p85/N-SH2, whereas RET wt was unable to interact with the GST fusion protein. Similar results were obtained with a GST recombinant protein containing the carboxyl-terminal SH2 domain of p85, although the interaction appeared weaker than with N-SH2 (data not shown). The levels of exogenously expressed RET were comparable in the cell lysates tested (Fig. 4D).
It has been previously reported that mutation of Tyr-1062 drastically impairs the transforming activity of MEN2A and MEN2B forms of RET (16,24). Since activation of the PI3K transduction pathway appeared to be a critical event in RET-MEN2 oncogenic ability (see below), we hypothesized that mutation of Tyr-1062 might prevent activation of PI3K. Consistent with this assumption, the GST-p85/N-SH2 fusion protein was unable to bind RET-MEN2A carrying a mutation at Tyr-1062 (Fig. 4B), whereas the protein was readily detected in total cell lysates (Fig. 4D).
It has been shown that both the amino-and carboxyl-terminal SH2 domains of p85 preferentially interact with peptides containing the sequence Tyr(P)-Met-Val-Ile-Glu-Xaa-Met (43). Although RET lacks a tyrosine residue embedded in a peptidic sequence closely related to the canonical p85-binding site, Tyr-981 is followed by a Met at position ϩ3, hence there is a possibility that this tyrosine might act as docking site for p85. To test this hypothesis, the GST-p85/N-SH2 fusion protein was incubated with cell lysates prepared from two distinct Rat1 clones expressing RET-MEN2A/981F. As shown in Fig. 4C, this RET mutant was capable of binding to p85, and the intensity of the band was comparable to those obtained with RET-MEN2A, thus indicating that mutation of this tyrosine residue does not significantly impede the binding of p85.
PI 3 Kinase-dependent Activation of AKT in Cells Expressing RET-MEN2A-The protein kinase AKT/protein kinase B is implicated in PI3K-mediated control of apoptosis and cell proliferation (32,33). AKT is activated upon binding of PI3K lipid products to its pleckstrin domain and by phosphorylation of Thr-308 and Ser-473 (44) by two kinases, PDK1 and PDK2. We therefore tested whether endogenous AKT was activated in cells expressing RET-MEN2A. For that purpose, we used an antiphospho-AKT antibody which detects AKT only when the enzyme is phosphorylated on Ser-473. This technique revealed that AKT was phosphorylated in cells expressing RET-MEN2A but not in cells expressing RET wt (Fig. 5A). In vitro kinase assay of the AKT immunoprecipitates using histone 2B as a substrate (37) confirmed that AKT was constitutively active in Rat1 cells expressing RET-MEN2A and not in cells expressing RET wt (Fig. 5B). To determine whether activation of AKT induced by RET-MEN2A was dependent of the PI3K pathway, we examined the effect of two potent inhibitors of PI3K, wortmannin and LY294002 (45,46); these two compounds are known to block the PI3K activity at nanomolar and micromolar concentrations, respectively. As shown in Fig. 5C, pretreatment of cells with LY294002 (10 M) and wortmannin (100 nM) for 10 min resulted in a complete inhibition of AKT activity in cells expressing RET-MEN2A, thus confirming that this effect was PI3K-dependent. Surprisingly, treatment with rapamycin, a drug known to inhibit the catalytic activity of p70S6 kinase, a downstream effector of PI3K, led to a reproducible increase of phosphorylation of AKT (Fig. 5C).
Mutation of Tyr-1062 but Not of Tyr-981 Impairs RET-MEN2A-mediated Activation of AKT-We next investigated the effects of mutations of Tyr-981 or Tyr-1062 on the RET-MEN2A-mediated activation of the PI3K pathway. Since, we had established that AKT is activated in a PI3K-dependent . B and C, cell lysates prepared from cells stably expressing RET WT or RET-MEN2A were subjected to GST "pull-down assay" using beads covered with GST-p85/N-SH2 fusion proteins or GST alone. Proteins bound to the GST fusion proteins were then analyzed by Western blotting using the anti-RET polyclonal antibody (WB: ␣RET). D, comparable amount of proteins was subjected to Western blot analysis with an anti-RET antibody to verify that RET was expressed at the same level in each cell line. IP, immunoprecipitation.Numbers 1 and 2 refer to two independent clonal cell lines. manner in cells expressing RET-MEN2A, we examined the activity of endogenous AKT in cells expressing this mutant form of RET. Since, AKT becomes phosphorylated on Thr-308 and Ser-473 upon stimulation of PI3K, we used antiphospho antibodies (antiphosphoserine 473) to determine whether AKT was activated. As shown in Fig. 6, AKT was phosphorylated in Rat1 fibroblasts expressing RET-MEN2A but not in cells expressing RET-MEN2A/1062F, whereas comparable amounts of AKT were present in each lane (Fig. 6). By contrast, AKT was found to be phosphorylated in Rat1 cells expressing RET-MEN2A/981F (Fig. 6), thus confirming that mutation of this tyrosine residue does not have a significant impact on the binding of p85 and the subsequent activation of the PI3K pathway.

The Forkhead Transcription Factor FKHRL1 Is Phosphorylated in Cells Expressing RET-MEN2A; Mutation of Tyr-1062
Disrupts This Phosphorylation-It has been shown recently that three transcription factors of the forkhead family, FKHR, FKHRL1, and AFX, are direct targets of AKT-catalyzed phosphorylation (39,47). Phosphorylation of these proteins by AKT inhibits their transactivation function (39,47), and this inhibition is caused by the association of the phosphorylated forms of forkhead family transcription factors with 14-3-3 proteins that subsequently lead to their cytoplasmic retention (39). We focused our analysis on FKHRL1 since it has been reported previously that this protein is the most ubiquitously expressed of the three forkhead family members (39). Brunet and coworkers (39) have determined that Thr-32 and Ser-253 are the two target sites phosphorylated by AKT. To determine whether FKHRL1 was phosphorylated in cells expressing RET-MEN2A and other RET mutants, we took advantage of phospho-specific FKHRL1 antibodies that specifically detect the phosphorylated forms of Thr-32 and Ser-253 (39). Western blot experiments revealed that FKHRL1 was phosphorylated on both residues in Rat1 cells expressing RET-MEN2A or RET-MEN2A/981F but not or very weakly in cells expressing either RET wt or RET-MEN2A/1062F (Fig. 7A) depending of the extracts tested; comparable amounts of FKRHL1 were present in each lane (Fig.  7B). Taken together, these results confirm that RET-MEN2A promotes the stimulation of the PI3K/AKT pathway that converges on the phosphorylation, and likely inhibition, of FKHRL1 and also underscore the crucial function of Tyr-1062 in the triggering of this signaling cascade.

Expression of a Dominant-negative Mutant of PI 3-kinase
Inhibits RET-MEN2A-mediated Transformation-It has been previously reported that a mutant form of p85 (called ⌬p85), which lacks the binding site of p110, acts as a PI3K dominantnegative mutant (36). To determine whether activation of PI3K was required for the RET-MEN2A transforming ability, Rat1 cell clones were infected with a retrovirus expressing ⌬p85. As shown in Fig. 8A, immunoblotting with an anti-p85 polyclonal serum readily detected a band in the molecular mass range of 85 kDa only in the cells infected with the retrovirus LXSN-⌬p85. On a longer exposure of the membrane, the endogenous p85 could be detected in all the lanes (data not shown). A second protein with a size in the range of 75-80 kDa which reacted with the anti-p85 antibody was detected but only in Rat1 cells expressing RET-MEN2A (Fig. 8A). As anticipated, overexpression of ⌬p85 completely inhibited the activation of AKT in Rat1 cells transformed with RET-MEN2A (Fig. 8B). We next tested whether ⌬p85 was capable of impairing the capacity of Rat1 cells expressing RET-MEN2A to grow without anchorage, a characteristic feature of transformed cells. As reported in Fig. 8C, Rat1 fibroblasts expressing RET-MEN2A formed colonies when seeded in soft agar, and the colonyforming ability correlated with the level of RET-MEN2A expression with clone 1 giving rise to 3-fold more colonies than clone 2. Remarkably, expression of ⌬p85 caused a dramatic FIG. 5. RET-MEN2A activates AKT in a PI3K-dependent manner. A, equivalent amounts of total cell proteins (50 g) were Western-blotted with an antibody that recognizes specifically the phosphorylated form of Ser-473 on AKT (pAKT, top panel). The same filter was then incubated with an antibody that reacts with both unphosphorylated and phosphorylated forms of AKT. B, immunoprecipitated AKT was assayed for AKT activity using histone 2B as a substrate in the presence of [␥-32 P]ATP before analysis by SDS-polyacrylamide gel electrophoresis/autoradiography. The position of phosphorylated histone 2B is indicated. 50 g of the same protein extracts were Western-blotted with an antiphospho-AKT (anti-Ser(P)-473; pAKT) and with an antibody that recognizes AKT in a phosphorylation-independent state (AKT). C, Rat1 cell clones were pretreated with wortmannin (100 nM), LY294002 (10 M), or rapamycin (20 ng/ml) for 10 min at 37°C and then immunoblotted with an antiphospho-AKT (pAKT) or an anti-AKT (AKT) antibody. Numbers 1 and 2 refer to two independent clonal cell lines. reduction in the number of colonies (a decrease of approximately 70 -80% of the number of colonies), thus supporting the notion that sustained activation of PI3K contributes to the RET-MEN2A-transforming ability.
Overexpression of AKT Enhances the Transforming Ability of RET-MEN2A-We next sought to ascertain whether AKT activity was required for the RET-MEN2A transforming capacity. We reasoned that the level of AKT might be limiting, and in this case overexpression of wild type AKT should enhance the RET-MEN2A oncogenic ability but should not have any effect on Rat1 cells expressing RET wt. Rat1 cell clones were infected with recombinant retroviruses expressing either ␤-galactosidase or an HA-tagged form of AKT (Fig. 9A), and then mass populations of puromycin-resistant cells were amplified. Western blot analysis of proteins isolated from these resistant cells with the anti-HA monoclonal antibody (12CA5) confirmed that the AKT protein was expressed at a comparable level in the various cell extracts (Fig. 9B). Clonogenic assays in soft agar revealed that exogenous expression of AKT induced a 2-fold increase in the number of colonies produced by RET-MEN2A, whereas AKT did not exhibit any significant effect in cells expressing RET wt (Fig. 9C). Taken together, these data indi-cate that AKT is a limiting factor and point to a direct role of the PI3K/AKT pathway to the RET-MEN2A oncogenic activity.

DISCUSSION
Several recent studies have revealed that dysregulation of the PI3K/AKT pathway is a frequent occurrence during tumorigenesis (reviewed in Refs. 30, 31, and 48). In this study, we found that PI3K and AKT were constitutively activated in rat fibroblasts expressing a MEN2A mutant form of RET. We further show that interference with the signal transduced through PI3K inhibited the RET-MEN2A transforming capacity measured by the ability of cells to grow without anchorage.
Mechanisms of PI3K Activation by RET-MEN2A-The recruitment of p85 to activated tyrosine kinase receptors is a crucial event for the activation of PI3K (reviewed in Ref. 31). With regard to this mechanism, we showed that RET-MEN2A bound in vitro to a recombinant GST protein fused to the amino-terminal SH2 domain of p85 (GST-p85) and also, although to a lesser extent, to the carboxyl-terminal SH2 domain of p85 (data not shown). Furthermore, p85 was found either phosphorylated on tyrosine residues or associated in a multimolecular complex with proteins containing phosphotyrosine in FIG. 7. RET-MEN2A induces phosphorylation of FKHRL1; mutation of Tyr-1062 of RET-MEN2A disrupts this phosphorylation. A, protein extracts prepared from each Rat1 cell clones were subjected to Western blot analysis using antiphospho-specific antibodies that react with the phosphorylated forms of Thr-32 (pT32) and Ser-253 (pS253) on FKHRL1. B, the same filter was probed with an antibody that recognizes both phosphorylated and unphosphorylated forms of FKHRL1.Numbers 1 and 2 refer to two independent clonal cell lines.

FIG. 8. Expression of a dominantnegative form of PI3K suppresses RET-MEN2A transforming activity.
A, Rat1 cell lines were infected with retroviruses expressing a dominant-negative form of p85 (LXSN ⌬p85) or with control retroviruses that only express the neoselectable marker (LXSN). Mass populations of infected cells were selected with G418 for 1 week. Comparable amount of proteins were subjected to Western blotting analysis with an anti-p85 polyclonal serum. B, Western blot analysis performed with an antiphospho-specific antibody (pAKT) or an anti-AKT antibody (AKT). C, mass population of resistant Rat1 cell clones were seeded in soft agar, and colonies were scored 3 weeks after plating. Results represent the scoring of colonies obtained in two independent series of infection and selection; each mass population of G418-resistant cells was seeded three times in duplicate.Numbers 1 and 2 refer to two independent clonal cell lines. cells expressing RET-MEN2A (data not shown). These results indicate that p85 might either directly recognize a phosphotyrosine docking site on RET or indirectly contact RET through its association with bridging signaling molecules. Our results support the latter hypothesis as follows: (i) mutation of Tyr-1062 in RET-MEN2A prevents the binding of p85 and considerably reduces the activation of the PI3K/AKT pathway; (ii) mutation of Tyr-981, which is the RET tyrosine residue surrounded by the peptidic sequence the closest to the optimal p85-binding site, does not significantly alter either the RET-MEN2A-p85 interaction or the activation of AKT. It is known that Shc and Enigma interact with Tyr-1062 (16,18,21,22), and therefore it is possible that either one of these two proteins recruit p85. Alternatively, additional adaptor proteins, yet to be characterized, might recognize Tyr-1062 and associate with p85. Interestingly, Murakami and co-workers (49) have recently shown that the Grb2-associated binder-1 protein (Gab-1) is phosphorylated on tyrosine in cells expressing RET-MEN2. Gab-1 contains several docking sites for p85 and is known to mediate PI3K activation (50,51), thus raising the possibility that Shc connects RET to PI3K via Grb2 and Gab-1. Although we cannot formally exclude that RET is also able to tether p85 directly, our data indicate that Tyr-981 is not a genuine p85binding site. These findings are corroborated by the demonstration that mutation of Tyr-981 does not impair the transforming activity of RET-MEN2A and RET-MEN2B (16).
It is now well established that activated Ras interacts with p110 and contributes to the activation of PI3K (reviewed in Refs. 30 and 52). Accordingly, the failure of RET-MEN2A Y1062F mutant to trigger the PI3K pathway might also reflect its inability to activate Ras via the formation of the Shc-Grb2-SOS complex. This idea is sustained by a number of recent experiments that showed that the expression of a dominantinterfering form of Ras (Asn-17 Ras) partially inhibits RET/ GDNF-mediated induction of AKT (53). It is likely that both the recruitment of p85/p110 in the vicinity of RET and the activation of Ras synergize to stimulate PI3K. In this case, Tyr-1062 would be a crucial site for both molecular mechanisms that converge on the activation of PI3K. A similar situation has been documented for the TrkA receptor tyrosine kinase that possesses an Shc-docking site and whose role is essential for the activation of Ras and PI3K (54).
Activation of AKT Is Required for RET-MEN2A Transforming Activity-It is known that PI3K transduces intracellular signals through multiple effector pathways that contribute to a plethora of biological responses (30,31). The protein kinase AKT is a crucial downstream target that delivers mitogenic and anti-apoptotic signals (32,33). Consistent with these data, AKT was found to be activated in cells expressing RET-MEN2A, and this activation was PI3K-dependent. Furthermore, overexpression of AKT markedly enhanced the transforming effect of RET-MEN2A, thus indicating that AKT is a limiting effector for RET-MEN2A oncogenic potential. In addition, expression of dominant-negative forms of AKT was recently found to inhibit RET-MEN2A oncogenic activity (data not shown). 2 Previous studies have revealed that the mechanisms by which AKT prevents cell death require its ability to phosphorylate and subsequently inactivate pro-apoptotic factors such as BAD (55,56), caspase 9 (57), and transcriptions factors of the forkhead family homologous to FKHR (39,47). In accordance with these findings, we obtained evidence that FKHRL1, one member of the FKHR family characterized as a substrate of AKT, is phosphorylated in cells transformed by RET-MEN2A and thus excluded from the nucleus (data not shown). 2 This exclusion is likely to be due to the sequestration of the phosphorylated form of FKHRL1 by 14-3-3 proteins, as demonstrated by Brunet and co-workers (39) in a different experimental context. Furthermore, AKT phosphorylates and suppresses the catalytic activity of the glycogen synthase kinase 3␣ (GSK-3 ␣), an enzyme involved not only in the regulation of glucose homeostasis but also implicated in the regulation of both apoptosis and mitogenesis via its control of ␤-catenin and cyclin D1 turnover (Ref. 58 and references therein). Finally, several recent reports have shown that AKT increases the activity of endothelial nitric-oxide synthase and NF-kB thus leading to the production of nitric oxide and to the transcription of genes coding for anti-apoptotic proteins (59 -62). Future experiments that aim at analyzing the respective roles of these downstream targets of AKT should enable us to 2 C. Segouffin-Cariou and M. Billaud, unpublished data.
FIG. 9. Overexpression of AKT potentiates RET-MEN2A transforming ability. A, schematic representation of the HA epitope-tagged form of AKT; PH, pleckstrin domain. The two residues (Thr-308 and Ser-473) that are phosphorylated by phosphoinositide-dependent protein kinase are also indicated. B, Rat1 cell clones were infected with the ecotropic retroviruses that express either ␤-galactosidase or AKT, and mass populations of resistant cells were selected with G418. Comparable amounts of total cellular proteins were Western-blotted (WB) with an anti-HA monoclonal antibody (12CA5). C, mass populations of resistant Rat1 cell clones were seeded in soft agar, and colonies were scored 3 weeks after plating. Results represent the scoring of colonies obtained in two independent series of infection and selection; each mass population of G418-resistant cells was seeded three independent times in duplicate.Numbers 1 and 2 refer to two independent clonal cell lines. elucidate the molecular mechanisms presumably involved in the etiology of MEN2 tumors.
Role of the Dysregulation of PI3K during RET-MEN2-mediated Transformation-Expression of ⌬p85 was found sufficient to reduce the oncogenic ability of RET-MEN2A, thus raising a question concerning the biological role of PI3K/AKT signaling during cellular transformation. The signal transmitted via PI3K/AKT is known to promote cell survival and to control the G 1 to S phase transition of the cell cycle in response to different growth factors (63). However, in our system it is unlikely that RET-MEN2A mediated its transforming effect by blocking cell death, since fibroblasts expressing RET wild type or RET-MEN2A were cultured in the presence of serum and did not display apoptotic features. It appears more likely that protracted activation of PI3K via RET-MEN2A results in cellular changes that characterize tumor growth, such as anchorageindependent growth. This is consistent with a study that reported that chronic stimulation of PI3K in combination with serum causes cellular modifications specific to transformed cells, whereas activation of PI3K in the absence of serum leads to apoptosis (63). Yet, it remains possible that in vivo both the PI3K-mediated protection from programmed cell death and the mitogenic effect are required in RET-MEN2 oncogenic activity, a question we are currently exploring. It has been recently demonstrated that the platelet-derived growth factor induces two waves of PI3K activity, the second wave being required for platelet-derived growth factor-dependent DNA synthesis (64). Therefore, it is plausible that aberrant accumulation throughout the cell cycle of PI3K-derived lipid products may account for the capacity of RET-MEN2 to promote cell cycle abnormalities and anchorage-independent growth. While this work was being completed, Murakami and co-workers (49) reported that RET-MEN2B is a more potent activator of PI3K than RET-MEN2A, thus suggesting that high levels of activity of this lipid kinase might contribute to the pathogenesis of clinical features specific to MEN2B.
In conclusion, these results provide evidence that the prolonged activation of the PI3K/AKT signaling pathway is a key event that accounts for the oncogenic ability of the MEN2A form of RET. It now becomes essential to investigate whether mutant forms of RET chronically activate the PI3K pathway in endocrine cells that are affected in MEN2 and to determine if this activation is required for the development of tumors.