The Sensitivity of Activated Cys Ret Mutants to Glial Cell Line-Derived Neurotrophic Factor Is Mandatory To Rescue Neuroectodermic Cells from Apoptosis

Abstract

Hirschsprung's disease (HSCR), a frequent developmental defect of the enteric nervous system is due to loss-of-function mutations of RET, a receptor tyrosine kinase essential for the mediation of glial cell-derived neurotrophic factor (GDNF)-induced cell survival. Instead, gain-of-function Cys mutations (e.g., Cys⁶⁰⁹, Cys⁶²⁰, and Cys⁶³⁴) of the same gene are responsible for thyroid carcinoma (MEN2A/familial medullary thyroid carcinoma) by causing a covalent Ret dimerization, leading to ligand-independent activation of its tyrosine kinase. In this context, the association of Cys⁶⁰⁹- or Cys⁶²⁰-activating mutations with HSCR is still an unresolved paradox. To address this issue, we have compared these two mutants with the Cys⁶³⁴ Ret variant, which has never been associated with HSCR, for their ability to rescue neuroectodermic cells (SK-N-MC cells) from apoptosis. We show here that despite their constitutively activated kinase, the mere expression of these three mutants does not allow cell rescue. Instead, we demonstrate that like the wild-type Ret, the Cys⁶³⁴ Ret variant can trigger antiapoptotic pathways only in response to GDNF. In contrast, Cys⁶⁰⁹ or Cys⁶²⁰ mutations, which impair the terminal Ret glycosylation required for its insertion at the plasma membrane, abrogate GDNF-induced cell rescue. Taken together, these data support the idea that sensitivity to GDNF is the mandatory condition, even for constitutively activated Ret mutants, to rescue neuroectodermic cells from apoptosis. These findings may help clarify how a gain-of-function mutation can be associated with a developmental defect.

The RET receptor tyrosine kinase (RTK) provides one of most interesting and documented models of human diseases caused by mutations within a single gene (for a review see reference 37). Its germ line mutations have been associated with Hirschsprung's disease (HSCR), a frequent defect (1 in 5,000 births) of development of the enteric nervous system (ENS) characterized by aganglionosis of the distal digestive tract. RET mutations are also involved in tumor formation: somatic RET chromosomal rearrangements are implicated in papillary thyroid carcinoma, and germ line RET mutations are responsible for the development of three inherited endocrine carcinomas: Familial medullary thyroid carcinoma (FMTC) and multiple endocrine neoplasia types 2A (MEN2A) and 2B (MEN2B).

Detailed biochemical analysis of RET mutations in the MEN2A/FMTC and HSCR families has shown how different mutations can lead to such opposite diseases. Indeed, RET encodes an RTK, expressed in tissues derived from the neural crest such as ENS, thyroid, and adrenal medulla (34, 54). Its ligand is a complex composed of the survival factor glial-cell line-derived neurotrophic factor (GDNF) (30, 40, 43) and the glycosylphosphatidyl inositol (GPI)-linked protein GFRα1 (for “GDNF family receptor α1”) (9, 17, 28, 52). It has been demonstrated that the GDNF/GFRα1 binding to Ret elicits its dimerization, which is a prerequisite for the activation of the Ret TK activity and the downstream signaling pathways essential for the survival of enteric crest-derived precursors (51). Consistently, most HSCR patients are affected by loss-of-function mutations that result in the reduction of Ret expression at the membrane or the abrogation of its intrinsic TK, thereby impairing the transduction of GDNF-induced survival signaling pathways (19, 36, 38).

Conversely, RET mutations which are responsible for medullary thyroid carcinoma (MEN2A/FMTC) are gain-of-function substitutions that specifically replace one of the six cysteines (Cys⁶⁰⁹, Cys⁶¹¹, Cys⁶¹⁸, Cys⁶²⁰, Cys⁶³⁰, and Cys⁶³⁴) present in the Ret juxtamembrane domain (15, 32, 33). Of these cysteines Cys⁶³⁴ is the most frequently mutated residue in families with MEN2A whereas mutations of Cys⁶⁰⁹, Cys⁶¹⁸, and Cys⁶²⁰ are often detected in FMTC (16, 32). MEN2A and FMTC have the same clinical features, but MEN2A is a more severe syndrome, characterized by the additional occurrence of pheochromocytoma and hyperparathyroidism. Consistently, in vitro experiments conducted with fibroblasts have shown that the Cys⁶³⁴ substitutions result in a stronger expression, TK activity, and transforming power than the other Cys mutations do (11, 13, 24). Taken together, these findings have led to the notion that RET loss-of-function mutations cause a developmental defect in the ENS while gain-of-function mutations promote medullary thyroid carcinoma.

Nevertheless, this concept does not fit the puzzling observations that several families presenting HSCR harbor one RET mutation of the MEN2A type at Cys⁶⁰⁹ (1) and that patients presenting a double HSCR/MEN2A phenotype carry a unique Cys⁶¹⁸ or Cys⁶²⁰ mutation (14, 39, 42). In contrast, the Cys⁶³⁴ mutations have never been associated with HSCR. This implies that among the activating Cys mutations, some substitutions (Cys⁶⁰⁹, Cys⁶¹⁸, and Cys⁶²⁰) can lead to impaired development of the ENS. In an attempt to resolve this apparent paradox, it has been proposed that a critical threshold level of Ret activity is necessary to promote neural crest survival (for a review, see reference 49). In this model, the Cys⁶³⁴-mutated Ret (Ret⁶³⁴) is believed to be, like the GDNF-stimulated Ret^WT, fully activated and thus able to promote cell survival, whereas the lower kinase activity of the Ret⁶⁰⁹ or Ret⁶²⁰ mutants would not be sufficient to reach the threshold level necessary to escape apoptosis (11, 13, 24). Very little information is currently available about the capacity of these Ret^Cys mutants to protect neuroectodermic cells from apoptosis, because most of the studies carried out so far have focused on the capacity of these variants to transform fibroblasts (2, 8, 11, 13, 24, 44). However, this latter model does not take into consideration the facts that expression of Ret in vivo is restricted to cells of neuroectodermic origin and that enteric crest-derived precursors coexpress Ret and GFRα1 (58).

Consistently, we have analyzed the capacity of these different Ret^Cys mutants to promote cell survival using the human neuroectodermic SK-N-MC cell line (55). We provide evidence that the simple expression of any of the Ret⁶⁰⁹, Ret⁶²⁰, or Ret⁶³⁴ mutants fails to rescue SK-N-MC cells from apoptosis, regardless of their constitutive TK activity. Instead, we found that the Ret⁶³⁴ variant, but not the Ret⁶⁰⁹ no Ret⁶²⁰ mutants, remains sensitive to the GDNF/GFRα1 complex. Interestingly, our results point to the absolute requirement for Ret⁶³⁴ to bind GDNF/GFRα1 in order to allow neuroectodermic cells to evade apoptosis. Possible implications of these findings for the understanding of the origin of the neurocristopathies observed in some patients carrying activating Cys⁶⁰⁹ or Cys⁶²⁰ mutations are discussed.

MATERIALS AND METHODS

Plasmids.

The human cDNA coding for the short isoform of RET (Ret9, 1,072 amino acids) was cloned into the pAlter vector (Promega) and site-directed mutagenesis of Cys⁶⁰⁹→Trp (TGC→TGG), Cys⁶²⁰→Arg (TGC→CGC), Cys⁶³⁴→Arg (TGC→CGC), and Met⁹¹⁸→Thr (ATG→ACG) was performed, as already described (36). The wild-type (WT) as well as the mutated RET cDNAs were then subcloned into the XbaI site of the pRc/CMV expression vector (In Vitrogen). Mutation of the desired codon was confirmed by complete sequence analysis. The Cys⁶⁰⁹-, Cys⁶²⁰-, Cys⁶³⁴-, and Met⁹¹⁸-mutated Ret proteins are referred to throughout this paper as Ret⁶⁰⁹, Ret⁶²⁰, Ret⁶³⁴, and Ret^2B, respectively. The myc tag was added to the C-terminal end of RET⁶³⁴ cDNA by PCR.

Cell culture, transfections, and treatments.

The human neuroectodermic SK-N-MC cell line, which constitutively expresses GFRα1 but not Ret (55), was cultured in Dulbecco's modified Eagle's medium (Gibco BRL) supplemented with 10% fetal calf serum (Gibco BRL), 2 mM l-glutamine, 100 U of penicillin per ml, and 100 μg of streptomycin per ml (referred as complete medium). To generate cell lines expressing either the WT or the mutated Ret, the SK-N-MC cells were stably transfected with 10 μg of the corresponding cDNA using the polyethyleneimine (Sigma) precipitation method. At 48 h later, transfected cells were selected in complete medium containing 0.8 mg of G418 (Geneticin; Gibco BRL) per ml. After 2 weeks, at least 100 G418-resistant clones for each mutant were individually picked, expanded, and assayed for Ret expression by anti-Ret Western blotting (C19; Santa Cruz Biotechnology).

Cell treatments.

For all experiments, cells were grown to 70% confluency and were serum starved for 6 to 16 h in fresh Dulbecco's modified Eagle's medium supplemented with 0.1% bovine serum albumin (A7030; Sigma) before being subjected to stimulation with GDNF (100 ng/ml; Genentech). When phosphatidylinositol-specific phospholipase C (PI-PLC) (1 U/ml; Oxford Glycosciences) was used to release GPI-linked proteins from the cell surface, it was added to the starvation medium for 90 min. The cells were then washed three times and incubated with GDNF and soluble GFRα1 (sGFRα1; 0.75 μg/ml [R & D Systems]).

Cell lysis, immunoprecipitation, and Western blotting.

Cells were washed with phosphate-buffered saline (PBS) and solubilized in NP-40 lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1% Nonidet P-40 [NP-40], 1 mM Na₃VO₄, 10 mM β-glycerophosphate, 10 mM NaF, 2 mM EDTA, 1 μM aprotinin, 25 μM leupeptin, 1 μM pepstatin, 2 mM phenylmethylsulfonyl fluoride). A 1-mg portion of precleared whole-cell lysates (WCL) was immunoprecipitated with 2 μg of either anti-Ret or anti-p85^PI3K antibodies (a gift of J. F. Tanti, EPI 99-11, Nice, France), as described previously (4). WCL or immunoprecipitates were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (9% polyacrylamide) electroblotted onto a polyvinylidene fluoride membrane (Immobilon-P; Millipore), and incubated overnight at 4°C with either anti-Ret (1:2,000), biotinylated antiphosphotyrosine (anti-Ptyr; Upstate Biotechnology Inc.) (1:10,000), anti-phospho-Thr²⁰²/Tyr²⁰⁴ ERK (1:2,000), or anti-phospho Ser⁴⁷³ Akt (1:1,000) antibodies (New England Biolabs), as described previously (4).

Ret kinase assay.

To determine the effect of the Cys mutations on the RTK activity, the WT and Cys-mutated Ret proteins were immunoprecipitated from cell lysates that were treated previously with GDNF (100 ng/ml) for 15 min or left untreated. Then, their kinase activity was assessed in vitro by measuring their ability to phosphorylate myelin basic protein (MBP; Sigma) (5 μg). After a 15-min incubation at 25°C in the presence of [γ-³²P]ATP (Amersham) (4 μCi), the reaction was stopped by the addition of reduced Laemmli buffer and the products were heated at 100°C for 5 min and resolved by SDS-PAGE (5 to 15% polyacrylamide). The radiolabeled proteins were then transferred to polyvinylidene difluoride membranes and visualized by autoradiography. The intensity of the bands corresponding to phosphorylated MBP was quantified by PhosphorImager (Molecular Dynamics) analysis and was expressed as the fold increase relative to unstimulated Ret^WT, which was given the arbitrary value of 1. Immunoprecipitation of equal amounts of Ret was checked by anti-Ret immunoblotting. Control reactions, in which Ret immunoprecipitates were omitted, showed no MBP phosphorylation.

Expression of Ret⁶³⁴ and GFRα1 in COS-7 cells.

Plates (diameter, 100 mm) of COS-7 cells were transiently transfected with expression plasmids for GFRα1, a myc-tagged RET⁶³⁴, alone or in combination (10 μg; ratio 1:10) by the DEAE-dextran method, as previously described (4). Two days after transfection, the cells were depleted for 6 h and treated with GDNF (100 ng/ml) before being subjected to cell lysis with NP-40 buffer supplemented with 1% Brij 96 (Fluka). The myc Ret⁶³⁴ proteins were then immunoprecipitated with a Sepharose-conjugated anti-myc polyclonal antibody (Clontech).

Endo-H digestion.

WCL (50 μg) from transfected SK-N-MC cells were denatured at 95°C for 5 min (in a reaction buffer containing 0.2 M sodium citrate [pH 5.5], 0.5% SDS, 1 M β-mercaptoethanol, and 0.5% phenylmethyl sulfonyl fluoride) before the addition of Endoglycosidase-H (Boehringer Mannheim). After an overnight incubation at 37°C, the reactions were stopped by the addition of an equal volume of reduced SDS sample buffer (125 mM Tris-HCl [pH 6.8], 10% glycerol, 2% SDS, 5% β-mercaptoethanol, 0.01% bromophenol blue) and the reaction products were resolved by SDS-PAGE and analyzed by anti-Ret Western blotting.

Immunofluorescence staining.

Cells seeded on a glass coverslip were fixed with methanol for 7 min at −20°C and permeabilized for 5 min at room temperature with saponin buffer (0.5 % saponin, 2.5% goat serum, 1% bovine serum albumin, and 0.2% gelatin in PBS). Using this saponin buffer throughout the experiment, the cells were incubated overnight with anti-Ret antibodies at 4°C (5 μg/ml), washed, and revealed with fluorescein isothiocyanate-conjugated anti-rabbit (Dako) (1:50) antibodies for 1 h at room temperature. Pictures were taken with a 63× magnification lens using a confocal microscope (Leica). The cells incubated with a control rabbit immunoglobulin G showed no staining.

Induction and detection of apoptosis.

The indicated cell lines were treated with GDNF or left unstimulated for 1 h before being incubated with the apoptotic agonist anisomycin (10 μg/ml) (Sigma). At 3 h later, both adherent (recovered after trypsinization) and nonadherent (present in the culture medium) cells were combined and washed in PBS. Apoptosis was then measured by using flow cytometry (FACScan Becton Dickinson apparatus) to score the percentage of living cells with low transmembrane mitochondrial potential (Ψ_M) among 10,000 gated events (using CellQuest software [Becton Dickinson]). For this purpose, the cells were stained with the mitochondrial fluorochrome 3,3′-dihexylocarbocyanine Iodide (DiOC₆) (40 ng/ml) (Molecular Probes) for 15 min at 37°C. Propidium iodide (PI) (5 μg/ml) (Sigma) was added to stain the dead cells. The cell population with low fluorescence intensity with both DiOC₆ (FL-1) and PI (FL-2) was defined as consisting of the apoptotic cells since compared to the living cells, they display a lower fluorescence intensity with the DIOC₆ probe due to the loss of Ψ_M and they exclude PI (intact membrane).

RESULTS

GDNF increases the tyrosine phosphorylation of Ret⁶³⁴ in SK-N-MC cells.

We first compared in SK-N-MC cells the signal transduction pathways of Ret⁶³⁴ mutant with those of the GDNF-stimulated Ret^WT. As shown in Fig. 1, the two types of transfectants expressed comparable levels of the 170- and 150-kDa Ret isoforms, corresponding to the mature cell surface receptor and the intracellular precursor, respectively (56). We checked by reverse transcriptase PCR that transfection of RET⁶³⁴ cDNA into SK-N-MC cells did not induce the expression of the endogenous RET^WT gene (data not shown). Analysis of anti-Ret precipitates by anti-Ptyr Western blotting indicated that the Ret^WT was not phosphorylated in unstimulated cells while addition of GDNF induced a marked increase in its autophosphorylation level (Fig. 1A). In agreement with previous studies (2, 8, 44), the p170 isoform of Ret⁶³⁴ exhibited a constitutive level of activation, as evidenced by its autophosphorylation and the pattern of phosphorylated cellular substrates (Fig. 1B). However, exposure of SK-Ret⁶³⁴ cells to GDNF induced a considerable enhancement of Ret⁶³⁴ autophosphorylation, providing strong evidence that despite its constitutive activation, Ret⁶³⁴ mutant has retained its sensitivity toward its ligand.

FIG. 1.

The Ret⁶³⁴ protein, when expressed in the neuroectodermic SK-N-MC cells, remains sensitive to GDNF. (A) Expression and autophosphorylation of Ret^WT and Ret⁶³⁴ proteins in SK-N-MC cells. The SK-N-MC cell line was stably transfected with WT or Cys⁶³⁴ →Arg-mutated RET9 isoform cDNA. Serum-starved cell lines were left untreated (−) or incubated with GDNF (100 ng/ml) (+) for 15 min and solubilized in buffer containing 1% NP40. Ret was immunoprecipitated (Ip), and its tyrosine phosphorylation was then assessed by anti-Ptyr Western blotting. Membranes were subsequently stripped and reprobed with anti-Ret (lower panels). Black arrows indicate the mobilities of mature p170^Ret and immature p150^Ret forms. (B) Comparison of phosphoprotein patterns in Ret^WT and Ret⁶³⁴ transfectants. Anti-Ptyr Western blotting of the WCL indicated that the SK-Ret⁶³⁴ cells constitutively elicited several phosphorylated proteins (open arrows) that were observed in GDNF-stimulated SK-Ret^WT cells. The right panel was exposed longer than the left panel to allow detection of the GDNF-induced substrates in SK-Ret^WT lysates. GDNF stimulation of the SK-Ret⁶³⁴ cells further increased the phosphorylation of p170^Ret and of p46 and p52, which represent two Shc isoforms (indicated by asterisks) (data not shown). As expected, GDNF stimulation of the untransfected cells did not result in protein phosphorylation. The positions of the molecular weight markers are indicated on the left in thousands.

Ret⁶³⁴ requires GFRα1 to bind and respond to GDNF.

Given that activation of the Ret^WT by GDNF depends on the presence of the GPI-anchored protein GFRα1 (28, 52), we sought to determine whether Ret⁶³⁴ also needs the expression of GFRα1 to bind GDNF. To this end, Ret⁶³⁴ was tagged with a myc epitope at its C terminus and cotransfected with the GFRα1 cDNA into COS-7 cells. After 48 h, transfected cells were left untreated or incubated with GDNF and lysed. Ret⁶³⁴ was then immunoprecipitated with an anti-myc antibody and the presence of GDNF and GFRα1 in the immunopellets was investigated by Western blotting. As shown in Fig. 2, Ret⁶³⁴ coimmunoprecipitated with GDNF from cells cotransfected with mycRET⁶³⁴ and GFRα1. In contrast, no trace of the ligand was detectable in anti-myc precipitates when cells were transfected with mycRET⁶³⁴ alone. This indicates that the interaction of GDNF with Ret⁶³⁴ was not direct but occurred through GFRα1. Accordingly, GDNF stimulation resulted in the coprecipitation of GFRα1 with Ret⁶³⁴. These results demonstrate that the concomitant presence of Ret⁶³⁴, GFRα1 and GDNF was necessary to stabilize the complex. Consistently, the ability of GDNF to increase Ret⁶³⁴ autophosphorylation, as revealed by anti-Ptyr Western blotting, was observed only when cells coexpressed GFRα1. These results provide the first evidence that, similarly to the WT form (28, 52), Ret⁶³⁴ can bind and respond to GDNF, provided that GFRα1 is coexpressed.

FIG. 2.

Ret⁶³⁴ and GFRα1 form a functional receptor for GDNF. To reconstitute the GFRα1/Ret⁶³⁴ complex, the COS-7 cells were transiently transfected with the cDNA of GFRα1, a myc-tagged RET⁶³⁴ alone or in combination (“mycRet⁶³⁴ + GFRα1,” 1:10 ratio). Two days after transfection, the cells were treated with GDNF for 30 min (100 ng/ml) (+) before being subjected to cell lysis with the NP-40 buffer supplemented with 1% Brij 96 to solubilize both Ret⁶³⁴ and GPI-anchored proteins. The mycRet⁶³⁴ proteins were then immunoprecipitated (Ip) with an anti-myc antibody. The phosphorylation of Ret⁶³⁴ and the coprecipitation of GFRα1 and GDNF in the myc immunoprecipitates were detected by anti-Ptyr, anti-GFRα1, and anti-GDNF Western blotting, respectively. The positions of the p170Ret⁶³⁴, GFRα1 (45 and 60 kDa), and GDNF proteins are indicated by arrowheads.

The activated Ret⁶³⁴ mutant needs to bind the GDNF/GFRα1 complex to rescue SK-N-MC cells from apoptosis.

Given the critical role played by GDNF in promoting neuroectodermic cell survival via the activation of Ret TK (51), we were interested in determining whether GDNF was still required for the survival of neuroectodermic cells expressing the constitutively activated Ret⁶³⁴ mutant. To investigate this, we treated, the SK-Ret^WT and SK-Ret⁶³⁴ cells with GDNF or left then unstimulated for 1 h and then incubated them for 3 h with the apoptotic agonist anisomycin. Apoptosis was then assessed by scoring the percentage of living cells exhibiting a collapsed mitochondria potential (Ψ_M), after DiOC₆ staining. As shown in Fig. 3A, GDNF preincubation protected the SK-Ret^WT cells from anisomycin-induced apoptosis whereas the sole expression of the Ret⁶³⁴ mutant failed to rescue SK-N-MC cells from apoptosis, despite its constitutive activation. In accordance with the above-demonstrated ability of Ret⁶³⁴ to bind and respond to GDNF, SK-Ret⁶³⁴ cells were rescued from anisomycin-induced apoptosis by GDNF. This prompted us to investigate whether Ret⁶³⁴ could activate Akt and extracellular signal-regulated kinase (ERK), two serine/threonine kinases that are involved in cell survival (for a review, see reference 26). As shown in Fig. 3B, the Ret⁶³⁴ mutant was not or barely constitutively coupled to these pathways, while addition of GDNF induced the activation of both ERK and Akt in SK-Ret^WT as well as in SK-Ret⁶³⁴ cells.

FIG. 3.

Signal transduction pathways induced by GDNF in SK-Ret⁶³⁴ cells. (A) GDNF protected the SK-Ret⁶³⁴ cells from anisomycin-induced apoptosis. The indicated SK-N-MC cells were treated with GDNF (100 ng/ml) or left unstimulated for 1 h before being incubated with the apoptotic agonist anisomycin (10 μg/ml). At 3 h later, both adherent and nonadherent cells were recovered and stained with the mitochondrial fluorochrome DiOC₆. Indeed, compared to the living cells, the apoptotic cells display a lower fluorescence intensity with the DiOC₆ probe, due to the loss of transmembrane mitochondria potential (Ψ_M). The percentage of apoptotic cells is indicated in each panel. (B) GDNF-induced activation of the ERK and Akt kinases in SK-Ret⁶³⁴ cells. WCL from control and GDNF-stimulated cells were analyzed by Western blotting using anti-phospho-Akt (anti-PAkt) or anti-phospho-ERK (anti-PERK) antibodies. The positions of the phosphorylated Akt, p42^ERK2, and p44^ERK1 are shown. As expected, GDNF stimulation of the untransfected cells did not result in cell survival or in ERK and Akt phosphorylation.

To ensure that GDNF needs to bind to GFRα1 to mediate its antiapoptotic effect, SK-Ret⁶³⁴ cells were pretreated with phosphatidylinositide-specific phospholipase C (PI-PLC) to remove GPI-linked proteins from the cell membrane. As shown in Fig. 4, this treatment markedly diminished concomitantly GDNF-induced Ret⁶³⁴ autophosphorylation (Fig. 4A) and Akt and ERK activation (Fig. 4B) and abrogated the rescuing effect of GDNF on anisomycin-treated SK-Ret⁶³⁴ cells (Fig. 4C). Importantly, addition of soluble GFRα1 (sGFRα1) in combination with GDNF restored all these responses, indicating that, as established for Ret^WT (28, 52), Ret⁶³⁴ did have to form a ternary complex with GFRα1 and GDNF to mediate the antiapoptotic signaling pathways.

FIG. 4.

Signaling induced by GDNF in SK-Ret⁶³⁴ cells depends on the presence of GFRα1. To remove GPI-linked molecules from the cell surface, the SK-Ret⁶³⁴ cells were pretreated with PI-PLC. The cells were then incubated in the absence or presence of GDNF and soluble GFRα1 (sGFRα1; 0.75 μg/ml) for either 30 min before NP-40 cell lysis (A and B) or 1 h before the addition of anisomycin (C). The responses to GDNF were examined by Western blotting of WCL (with anti-Ptyr, anti-PERK, and anti-PAkt [A and B]) as well as by scoring the cell survival upon anisomycin treatment (C), as described in the legend for Fig. 3. The positions of the phosphorylated p170^Ret, p46^Shc, p52^Shc, Akt, and ERK proteins are indicated by arrowheads.

GDNF does not protect SK-Ret⁶⁰⁹ and SK-Ret⁶²⁰ cells from apoptosis.

Germ line RET mutations at Cys⁶⁰⁹ or Cys⁶²⁰ were found in several families who develop HSCR in addition to MEN2A/FMTC (14, 31, 42). In an attempt to explore the molecular defects engendered by this type of mutation, Cys⁶⁰⁹ →Trp or Cys⁶²⁰ →Arg mutated RET cDNA was transfected into SK-N-MC cells. As shown in Fig. 5, the Ret⁶⁰⁹ and Ret⁶²⁰ mutants exhibited an autophosphorylation level (Fig. 5A) and an in vitro kinase activity (Fig. 5B) significantly higher than those of the unstimulated Ret^WT. However, at variance with Cys⁶³⁴ mutation, these two mutants were weakly activated and were expressed mainly in a p150 form instead of the normal p170/p150 doublet (Fig. 5A), confirming previous results obtained with fibroblasts and PC12 cells (11, 13, 24, 38). Indirect immunofluorescence in SK-N-MC cells has allowed us to localize the p150^Ret precursor in the reticulum and the p170^Ret receptor at the plasma membrane (56). To ascertain that the p150 forms of Ret⁶⁰⁹ and Ret⁶²⁰ corresponded to the immature form and not to a proteolytic fragment of p170^Ret, cell lysates were treated with Endo-H and the products were analyzed by anti-Ret Western blotting. Endo-H can digest the N-linked immature oligosaccharides of the precursor before they are processed in the Golgi apparatus. As shown in Fig. 5C, Endo-H completely digested the p150 forms of Ret⁶⁰⁹ and Ret⁶²⁰ into a 120-kDa band that comigrated with the digested product of the p150 form of Ret⁶³⁴, strongly suggesting that the p150 forms of Ret⁶⁰⁹, Ret⁶²⁰ and Ret⁶³⁴ indeed corresponded to the same intracellular precursor. At that stage, it was of interest to address whether these different Ret^Cys mutants presented the same subcellular localization. To this end, the cells were fixed and permeabilized before being stained with anti-Ret antibodies and ultimately analyzed by confocal microscopy. As shown in Fig. 5D, both the Ret^WT and Ret⁶³⁴ mutants preferentially exhibited a cell surface localization whereas we failed to detect any trace of membrane staining on SK-Ret⁶⁰⁹ and SK-Ret⁶²⁰ cells. Instead, these latter variants localized in a perinuclear zone, probably corresponding to the endoplasmic reticulum.

FIG. 5.

The Cys⁶⁰⁹ and Cys⁶²⁰ mutations exert a dual effect on Ret. (A) Expression and tyrosine phosphorylation of the Ret⁶⁰⁹ and Ret⁶²⁰ mutants in SK-N-MC cells. The SK-N-MC cells were stably transfected with Cys⁶⁰⁹ →Trp- and Cys⁶²⁰ →Arg-mutated RET9 cDNA. Ret was immunoprecipitated (Ip) from control and GDNF-stimulated cell lysates, and its tyrosine phosphorylation was then assessed by anti-Ptyr Western blotting. The membranes were subsequently stripped and reprobed with anti-Ret (lower panel). (B) Kinase activity of the Ret^Cys mutants. The kinase activity of Ret immunoprecipitates was assessed in vitro in the presence of [γ-³²P]ATP by the phosphorylation of MBP. The radiolabeled proteins were resolved by SDS-PAGE, electroblotted onto PVDF membranes, and visualized by autoradiography. The intensity of the bands corresponding to phosphorylated MBP was quantified by PhosphorImager analysis and expressed as the fold increase relative to unstimulated Ret^WT, which was given the arbitrary value of 1 (indicated under each lane). Control reactions, in which Ret immunoprecipitates were omitted, showed no MBP phosphorylation (data not shown). Immunoprecipitation of equal amounts of Ret was verified by anti-Ret Western blotting (lower panel). (C) WCL from the indicated cell lines were incubated without (−) or with (+) Endo-H before being analyzed by anti-Ret Western blotting. The positions of the mature glycosylated (170-kDa), the partially glycosylated (150-kDa), and the digestion product (120-kDa) Ret forms are shown on the left. (D) Subcellular localization of Ret^Cys mutants. The indicated cells were fixed and permeabilized with methanol before being stained with anti-Ret antibodies and analyzed by confocal microscopy. Cells incubated with a control rabbit immunoglobulin G showed no staining (data not shown).

We therefore investigated the consequences of the expression of Ret⁶⁰⁹ and Ret⁶²⁰ mutants on the survival of neuroectodermic cells. We found that the simple expression of Ret⁶⁰⁹ and Ret⁶²⁰ did not induce SK-N-MC cell death (Fig. 6B) or protect these cells from apoptosis, strengthening the idea that these constitutively activated Ret^Cys mutants were unable to trigger antiapoptotic signaling pathways. Furthermore, addition of GDNF to SK-Ret⁶⁰⁹ or to SK-Ret⁶²⁰ cells failed to enhance the phosphorylation level (Fig. 5A) and the kinase activity (Fig. 5B) of these two Ret variants, nor did it induce the activation of Akt and ERK (Fig. 6A), in contrast to what we observed in SK-Ret⁶³⁴ cells (Fig. 5 and 6). Consistently, GDNF did not protect the SK-Ret⁶⁰⁹ or SK-Ret⁶²⁰ cells from anisomycin-induced apoptosis (Fig. 6B). These data strongly suggest that the intracellular retention of Ret⁶⁰⁹ and Ret⁶²⁰ prevented these mutated forms from interacting with the GDNF/GFRα1 complex, thus precluding the activation of the antiapoptotic pathways.

FIG. 6.

The SK-Ret⁶⁰⁹ and SK-Ret⁶²⁰ cells are insensitive to GDNF. SK-Ret⁶⁰⁹ and SK-Ret⁶²⁰ cells were incubated in the presence of GDNF for 15 min prior to NP-40 cell lysis (A) or for 1 h before the addition of anisomycin (B). The responses to GDNF were examined by Western blotting of WCL (with anti-phospho-ERK and anti-phospho-Akt) (A) as well as by scoring the cell survival upon anisomycin treatment (B), as described in the legend for Fig. 3. The positions of the phosphorylated Akt and ERK proteins are indicated by arrows. An overexposure of the blot was required to detect a faint constitutive phosphorylation of ERK in the SK-Ret⁶⁰⁹, SK-Ret⁶²⁰, and SK-Ret⁶³⁴ cells.

Activation of the PI3K/Akt pathway correlates with the ability of the MEN2B Ret mutant to rescue SK-N-MC cells from apoptosis.

The Met⁹¹⁸ →Thr substitution within the Ret TK domain is responsible for MEN2B, an inherited cancer syndrome defined, like MEN2A, by the presence of medullary thyroid carcinoma and pheochromocytoma (12, 22). However, in the ENS, MEN2B differs from MEN2A by the development of ganglioneuroma. It is therefore of interest to investigate the possibility that Met⁹¹⁸-mutated Ret (referred as Ret^2B) triggered both the activation of Akt and ERK and cell rescue in SK-N-MC cells. As shown in Fig. 7, the Ret^2B mutant was expressed under the p170^Ret and p150^Ret isoforms. The high autophosphorylation level of p150^Ret2B revealed the constitutive activation of Ret^2B that, in turn, elicited the phosphorylation of several cellular substrates. Nonetheless, Ret^2B, which was shown by confocal microscopy (Fig. 7B) to be expressed at the plasma membrane, retained its sensitivity toward GDNF since GDNF significantly enhanced the autophosphorylation level of the p170^Ret2B form. However, in contrast to Ret⁶³⁴, Ret^2B was able by itself to constitutively activate Akt bur not ERK, whose activation was strictly dependent on the presence of GDNF (Fig. 8A). In agreement with what is known about the upstream regulation of Akt, p85^PI3K was constitutively recruited by Ret^2B along with several phosphorylated proteins of 46, 52 and 120 kDa (Fig. 8B). This is in accord with recent data showing the assembly of a multimolecular complex including Ret, p46^Shc, p52^Shc, p120^Gab, and p85^PI3K (3, 21). Conversely, coprecipitation of p85^PI3K with either Ret^WT- or Ret⁶³⁴-containing complex did require ligation to GDNF. Interestingly, one should notice that the mere expression of Ret^2B allowed cell rescue in the absence of GDNF. Exposure to GDNF did not further protect SK-Ret^2B cells to anisomycin-induced apoptosis. Taken together, these findings reinforce the idea that the PI3K/Akt pathway is the key step that must be activated by Ret mutant to protect the cell from apoptosis.

FIG. 7.

The Ret^2B mutant is sensitive to GDNF. (A) Expression and autophosphorylation of Ret^2B in SK-N-MC cells. The SK-N-MC cell line was stably transfected with Met⁹¹⁸ →Thr-mutated RET9 cDNA. Serum-starved cell lines were left untreated (−) or incubated with GDNF (+) for 15 min before being subjected to NP-40 lysis. Anti-Ptyr Western blotting of WCL indicated that the SK-Ret^2B cells elicited constitutively several phosphorylated proteins (open arrowheads) that were observed in SK-Ret⁶³⁴ and GDNF-stimulated SK-Ret^WT cell lysates. GDNF stimulation of the SK-Ret^2B cells further increased the phosphorylation of p170^Ret and p52 proteins (arrowheads and ∗). Anti-Ret Western blotting of this membrane is shown in the lower panel. The positions of the molecular weight markers are indicated on the left in thousands. (B) The subcellular localization of Ret^2B was analyzed as described in the legend for Fig. 5D.

FIG. 8.

Ret^2B is able to constitutively activate the PI3K/Akt pathway and to protect SK-N-MC cells from apoptosis. SK-Ret^2B cells were incubated in the presence of GDNF for 15 min before NP-40 cell lysis (A and B) or for 1 h before the addition of anisomycin (C). The responses to GDNF were examined by Western blotting of WCL (with anti-phospho-ERK and anti-phospho-Akt) (A) as well as by scoring cell survival upon anisomycin treatment (C), as described in the legend for Fig. 3. (B) Coimmunoprecipitation (Ip) of p85^PI3K with Ret. Anti-Ptyr Western blotting of anti-p85^PI3K immunoprecipitates indicated that GDNF induced in both SK-Ret^WT and SK-Ret⁶³⁴ cells the coprecipitation of p85^PI3K with several phosphorylated proteins including p170, p110, p52, and p46 (arrowheads). While no such complex could be detected in unstimulated SK-Ret^WT and SK-Ret⁶³⁴ cells, the coprecipitation of p85^PI3K with Ret^2B did not require ligation to GDNF. After stripping, the presence of p170^Ret and p150^Ret was detected by reprobing the same blot with anti-Ret (data not shown). The positions of the molecular weight markers are indicated on the left in thousands.

DISCUSSION

Throughout development and life, the Ret RTK is essential for the mediation of the GDNF survival signals. Like all RTK, Ret exists as an inactive monomer until it binds its ligand, which drives its dimerization, required for the activation of its intrinsic TK activity. Consistently, RET gain-of-function mutations are involved in tumor formation while loss-of-function mutations are associated with developmental defects (37). Among the activating mutations, much effort has been devoted to delineating how Cys substitutions could result in Ret activation and thereby in the transformation of fibroblasts. It turned out that these mutations result in the dimerization of the mutated Ret via the formation of an intermolecular disulfide bond and hence in the permanent activation of its intrinsic TK (2, 8, 44). Based on these features, the Cys mutants seem locked in a dimeric activated state that mimics the ligand-occupied receptor. This has led to the notion that these mutants are ligand-insensitive oncogenes that support by themselves cell transformation but also, by inference, support all the ligand-induced effects. Intriguingly, several families with ENS defect (HSCR) harbor one RET gain-of-function mutation at Cys⁶⁰⁹, Cys⁶¹⁸, or Cys⁶²⁰ (41, 49). To address this issue, we have compared the consequences of these substitutions in terms of cell survival with those of the Cys⁶³⁴ mutation, which has never been associated with HSCR.

The explanation proposed for the normal development of the RET⁶³⁴-carrying patients is that Ret⁶³⁴ shares with GDNF-stimulated Ret^WT the capacity for promoting the signaling required for neuroectodermic cell survival (for a review, see reference 49), but this hypothesis still lacks experimental support. To this end, the RET^WT or the RET⁶³⁴ cDNA was transfected into the neuroectodermic SK-N-MC cell line, which constitutively expresses GFRα1 (55). We found that GDNF stimulation of Ret^WT in this cellular model induced phosphorylation of several cellular substrates, activation of the ERK and Akt pathways, and protection of SK-N-MC cells from anisomycin-induced apoptosis. We then confirmed that the mere expression of Ret⁶³⁴ resulted in phosphorylation of the same substrates as those lying downstream of the GDNF-stimulated Ret^WT. However, this mutant failed to activate ERK and Akt and to protect these cells from apoptosis. These data indicate that even though the constitutive activity of Ret⁶³⁴ mutant is sufficient for fibroblast transformation (2, 8, 24, 44), it was not able to protect neuroectodermic cells from apoptosis. This is in apparent contrast to a recent report showing that Ret⁶³⁴ constitutively activates the PI3K/Akt pathway in fibroblasts (47). An attractive hypothesis to explain this discrepancy might be to consider that the signaling generated by the Ret⁶³⁴ mutant depends critically on the cell type, as previously reported (57).

Recent cross-linking experiments have evidenced that Ret⁶³⁴ could interact with GDNF (53). However, other experiments have shown that GDNF does not increase the proliferation of NIH 3T3 fibroblasts that have been cotransfected with Ret⁶³⁴ and GFRα1 (10), supporting the idea that the oncogenic power of Ret⁶³⁴ is independent of GDNF. This prompted us to investigate whether GDNF might trigger survival signaling pathways in neuroectodermic Ret⁶³⁴-expressing cells. We demonstrated that exposure of SK-Ret⁶³⁴ cells to GDNF produced a dramatic enhancement of Ret⁶³⁴ autophosphorylation, the subsequent induction of ERK and Akt activities, and a concomitant protection of SK-Ret⁶³⁴ cells from apoptosis. We found that all these GDNF-dependent events required the coexpression of GFRα1. Consistently, Ret⁶³⁴ retained its capacity to form a ternary complex with GDNF and GFRα1. In this regard, it is of interest that in PI-PLC-treated SK-Ret⁶³⁴ cells, addition of soluble GFRα1 allowed GDNF to exert its protective effect. However, it was less efficient in activating Akt, consistent with the notion of an Akt threshold requirement. This strongly suggests that the recruitment of Ret⁶³⁴ to lipid raft via the GPI-anchored protein GFRα1 is an important step for the full activation by GDNF of Ret⁶³⁴ signaling, as recently demonstrated for the Ret^WT (50). These findings point to the important notions that activation of Ret⁶³⁴ by GDNF is absolutely required for neuroectodermic cell survival and that the signaling pathways induced by GDNF in the SK-Ret⁶³⁴ cells are indistinguishable from those induced in the Ret^WT-expressing cells.

In the RET⁶⁰⁹- or RET⁶²⁰-carrying patients, ENS development can be severely impaired (1, 14, 31, 42). In an attempt to explain these puzzling observations, it has been proposed that the Ret⁶⁰⁹ and Ret⁶²⁰ mutants display a TK activity under the threshold required to promote neuroectodermic cell survival (49). Alternatively, it has been assumed that these two mutants trigger a signaling pathway that commits the cell to death (37, 46). We found that when expressed in SK-N-MC cells, the Ret⁶⁰⁹ and Ret⁶²⁰ variants exhibited a constitutive TK activity higher that those of the Ret^WT. However, at variance with Cys⁶³⁴ mutations, these two mutations were weakly activating. In contrast to the above-mentioned hypotheses, the expression of Ret⁶⁰⁹ and Ret⁶²⁰ did not by itself cause the death of SK-N-MC cells. Furthermore, addition of GDNF did not increase Ret⁶⁰⁹ and Ret⁶²⁰ phosphorylation levels or induce the activation of Akt and ERK. Consistently, GDNF could not protect the SK-Ret⁶⁰⁹ nor the SK-Ret⁶²⁰ cells from anisomycin-induced apoptosis. This was correlated with the intracellular accumulation of these mutants under a 150-kDa incompletely glycosylated form. In this regard, evidence of retention of misfolded and incompletely glycosylated Ret molecules in the endoplasmic reticulum (6, 13) is of particular interest, suggesting that a blockade in the glycosylation process of Ret⁶⁰⁹ and Ret⁶²⁰ mutants prevents their insertion into the plasma membrane.

Taken together, these data demonstrate that the activated Ret^Cys mutants do not function as originally expected: none of the three Ret^Cys variants was able per se to rescue neuroectodermic cells from apoptosis. Only the GDNF binding to Ret^WT and Ret⁶³⁴ was capable of generating antiapoptotic pathways. Indeed, we could demonstrate that the inability of the Ret^Cys mutants to trigger these antiapoptotic pathways was independent of their level of expression (data not shown), their subcellular localization, and their extent of activated state. In fact, Ret⁶³⁴, which was expressed at the cell surface and displayed a TK activity twice that of GDNF-stimulated Ret^WT, was unable to ensure a survival signaling. These findings definitively invalidate the “threshold” model and, rather, support a model in which the Ret⁶³⁴ mutant is not only sensitive to but also entirely dependent on its ligation to GDNF for triggering neuroectodermic cell survival. This implies that ligand binding is able to activate signals that are qualitatively different from that resulting from the covalent dimerization imposed by the Cys mutations. In support of this hypothesis, recent studies have evidenced that the unliganded receptor can adopt inactive dimeric conformations that are different from those of the activated receptor complexed with ligand (for a review, see reference 27). This poses the question of what “activating mutation” means. To address this issue, we have studied the consequences on cell survival of the Met⁹¹⁸→Thr RET substitution (Ret^2B), another gain-of-function mutation responsible for MEN2B that has never been associated with HSCR. This substitution leads to Ret activation by forcing its catalytic domain into a constitutively activated conformation (8, 12, 22, 44). When expressed in SK-N-MC cells, Ret^2B was constitutively activated, was expressed at the cell surface, and retained its sensitivity towards GDNF. However, in the absence of GDNF, Ret^2B could at the same time induce cell rescue and activation of the PI3K/Akt pathway. This is in contrast to what we observed for the Ret^2B-mediated activation of ERK, which still necessitated GDNF. Constitutive activation of the PI3K/Akt pathway may be in relation to the reported changes in Ret^2B autophosphorylation sites (25, 29) and substrate specificity (4, 35, 45, 48), compared to Ret^WT, and Ret⁶³⁴. This is a fundamental point that distinguishes Ret^2B from Ret⁶³⁴ in neuroectodermic cells and may underlie the differences in the phenotypes of patients carrying these respective germ line mutations. Indeed, MEN2B is defined, as is MEN2A, by the occurrence of medullary thyroid carcinoma and pheochromocytoma. However, MEN2B displays a more complex and severe phenotype characterized by rapid disease progression and interestingly, ganglioneuromas in the intestinal tract. Concerning the Ret⁶⁰⁹ and Ret⁶²⁰ mutants, our data support the idea that these variants failed to ensure cell survival because their intracellular location impede their interaction with the membrane-associated GDNF/GFRα1 complex. These observations may help clarify why the RET⁶⁰⁹- and RET⁶²⁰-carrying patients have impaired ENS development. The fact that the HSCR phenotype is present in only a small number of families with recurrence of FMTC and in few carriers within these families has to be ascribed to the intrinsic complexity of the HSRC etiology. Indeed, compelling studies indicate that the expression of the HSRC phenotype does not arise solely from the presence of a mutation but, rather, involves the contribution of converging components such as the genetic background and the action of environmental factors or modifying genes (5, 7, 18, 20, 23). An exciting challenge for future studies would be to characterize these factors and to verify whether the completion of the glycosylation process of Ret⁶⁰⁹ and Ret⁶²⁰ mutants could restore their expression at the cell surface and hence their sensitivity to GDNF.