Pheochromocytomas are tumors originating from chromaffin cells, the large majority of which are sporadic neoplasms. The genetic and molecular events determining their tumorigenesis continue to remain unknown. On the other hand, RET germ-line mutations cause the inheritance of familial tumors in multiple endocrine neoplasia (MEN)-2 diseases, which account for a minority of pheochromocytomas. We investigated the expression of the RET gene in 14 sporadic tumors harboring no activating mutations. A subset of highly RET-expressing tumors(50%) could be distinguished. They showed RET transcript, protein amounts as well as Ret-associated phosphotyrosine levels similar to those measured in MEN-2A-associated pheochromocytomas. We also determined the GDNF and GDNF family receptor α (GFRα)-1 transcript levels in tumors and in normal tissues. Whereas the GFRα-1 transcripts were detected at similar levels in normal tissues and in tumors, GDNF was frequently found expressed in sporadic tumors at levels several times higher than in controls. These results led us to propose the existence of an autocrine or paracrine loop leading to chronic stimulation of the Ret signaling pathway, which could participate in the pathogenesis of a number of sporadic pheochromocytomas.
Pheochromocytomas are tumors originating from chromaffin cells,80–85% of which are sporadic neoplasms (1, 2, 3) that most often develop from the adrenal gland and typically follow a benign course (3). The remaining 15–20% of pheochromocytomas are of familial origin associated with the MEN4-2 diseases, the von Hippel-Lindau disease, or the type 1 neurofibromatosis, which present germ-line mutations in the RET,VHL, and NF-1 genes, respectively (1, 2, 4, 5).
The proto-oncogene RET is a receptor-like protein tyrosine kinase (6). Four distinct ligands for the Ret protein have recently been identified. All are polypeptide growth factors belonging to the glial cell line-derived neurotrophic factor family (GDNF,neurturin, persephin, and artemin). Ret association to any of the ligands is mediated by the presence, in the same molecular complex, of distinct glycosyl-phosphatidylinositol anchored proteins, the GFRα-1–4 (Ref. 7 and references therein). Germ-line mutations of RET cause the inheritance of the MEN-2 syndromes (for review, see Ref. 8). Mutations in cysteine residues of the extracellular domain (exons 10 and 11) are the most frequent causative genetic events of familial medullary thyroid carcinoma and MEN-2A syndromes. A single-point mutation that results in a Thr-for-Met substitution at codon 918 (exon 16) within the Ret catalytic domain is responsible for the MEN-2B syndrome(9). These mutations cause chronic induction of the tyrosine kinase and convert RET into a dominant oncogene (10, 11, 12).
Contrary to familial pheochromocytomas, knowledge of the genetic and biochemical events involved in the pathogenesis of the sporadic pheochromocytomas is still lacking. Mutations in the GDNFgene do not seem to play a major role in the pathogenesis of these tumors (13, 14). Moreover, sporadic pheochromocytomas are only rarely associated with somatic activating mutations of RET. Substitution of methionine 918 occurs in 10–15% of total sporadic pheochromocytomas, and mutations in the Ret extracellular cysteine-rich domain in less than 5% of these tumors(5). On the other hand, the wild-type RET gene is frequently expressed in pheochromocytomas, but few data are available on the Ret protein activity in these tumors(15, 16, 17). Indeed, the presence of Ret raises the question of whether stimulation of wild-type Ret activity in sporadic pheochromocytomas might be implicated in determining the neoplastic phenotype.
The causal function played by RET mutants in pheochromocytomas of familial origin has raised the hypothesis that, in addition to the Ret protein, other partners of the Ret ligand-receptor complex might be expressed in sporadic pheochromocytomas. This might then proceed to a persistent stimulation of the Ret signaling pathway. To test this hypothesis we asked: (a) whether sporadic pheochromocytomas presented enhanced RET expression levels;(b) whether Ret enzymatic activity was stimulated;(c) whether other Ret partners, i.e., GDNF and GFRα-1 were also expressed in these tumors; and (d)whether Ret expression and activity levels were compatible with a contribution of Retwt to a multistep process resulting in the tumorigenesis of chromaffin cells.
Here we show the persistent stimulation of the Ret protein in a number(50%) of sporadic adrenal pheochromocytomas harboring no activating mutation, and we discuss the possible effects of the observed enhanced RET and/or GDNF transcript levels in these tumors.
MATERIALS AND METHODS
Human Tissue Samples and DNA Analysis
The adrenal RNA sample a.g.1 is a pool of total RNAs extracted from six whole adrenal gland specimens (purchased from Clontech); a frozen postmortem a.m. was obtained from Dr. P. F. Plouin (Hôpital Broussais, Paris, France) and two frozen postmortem s.n. were obtained from Dr. E. Hirsch (Höpital Pitié Salpêtrière,Paris, France). Frozen samples of human pheochromocytomas were obtained from the Réseau Comète (Hôpital Broussais). Tumor samples were frozen immediately after surgery and kept in liquid nitrogen. All of the tumors were adrenal tumors. Malignancy was defined by histological evidence of distant metastases (3). Genomic and tumor DNA were assayed for RET-activating mutations situated in exons 10, 11, and 16 (4, 5, 18) for all of the patients. Nineteen tumors, numbered 2 to 20, were analyzed. Germinal mutations were found in four cases: (a) tumor 2(MEN-2A, C634R); (b) tumor 3 (MEN-2A, C634P); (c)tumor 4 (MEN-2A, C634R); and (d) tumor 5 (MEN-2B, M918T);one tumor DNA mutation was found for patient tumor 6 (M918T), but no germinal mutation was found in this case. All of the other tumor DNA samples (tumors 7 to 20) as well as the corresponding genomic DNAs presented no activating mutation.
For all of the samples, total cellular RNA was extracted from frozen tissues or tumor samples with RNAzol B (Bioprobe) according to the manufacturer’s instructions.
In Vitro RNA Synthesis
DNA templates for transcription of labeled antisense RNA probes and unlabeled sense RNAs were obtained by PCR including the sequence of the T7 promoter. Antisense RNA probe and the corresponding sense RNA corresponded to the following regions: (a) GDNF, exon 2 G378 to G545 and T352to G564; (b) GFRα-1, G1044 to G1182 and G908 to G1257; (c) RET, exon 19 T31 to G129 and exon 18 T23 to exon 20 G58; (d) tyrosine hydroxylase, exon 14 C61 to G140 and exon 13 T102 to exon 14 C213; and (e) β-actin,exon 5 G1 to T64 and exon 4 C342 to exon 5 C58. Labeled antisense RNA probes were synthesized in 40 mm Tris-HCl (pH 7.5), 10 mm NaCl, 6 mmMgCl2, 2 mm spermidine, 6 mm DTT, 140 mm ATP, 140 mm GTP, 140 mm CTP, 3.5 mm UTP, 0.5 mm[α-32P]UTP (3000 Ci/mmol) for the RET probe,1.05 mm [α-32P]UTP for the GDNF and GFRα-1 probes, 2 pmoles of template DNA, 50 u of T7 RNApolymerase (TEBU), in a total volume of 10 μl, incubated for 30 min at 37°C, then incubated for 30 min with 10 units of DNase I (Boehringer Mannheim) and purified on 5% denaturing acrylamide gel. Sense RNA was synthesized in the same conditions with 0.6 mm ATP, 0.6 mm GTP, 0.6 mm CTP, 0.5 mm UTP and 0.17μ m [α-32P]UTP. RNA probes and synthetic RNAs were resuspended in H2O containing 2 ng/ml tRNA (Escherichia coli extract).
Cellular RNAs (or synthetic mRNAs) and 0.3–1 fmol of radiolabeled RNA probe were mixed, lyophilized, resuspended in 80% formamide,0.4 m NaCl, 40 mm PIPES, and 1 mmEDTA in a total volume of 7.5 μl; heated for 5 min at 85°C;incubated for 12 h at 60°C; digested by the addition of 87.5μl of 300 mm NaCl, 10 mm Tris (pH 7.5), 5 mm EDTA, 50 μg/ml RNase A, and 120 units/ml RNase T1 (Sigma); incubated for 30 min at 30°C; then treated with 0.2 mg/ml proteinase K and 1% SDS for 30 min at 37°C; treated with phenol and ethanol; and analyzed on 7% acrylamide denaturing gels. Densitometric measurements were performed with a PhosphorImager. The intensities of the bands were obtained after subtraction of the background. RNA-protected bands that were obtained with known amounts of synthetic RNAs were used as standards to calculate the absolute amounts of the corresponding mRNA species contained in the cellular RNA samples. In all of the experiments, the intensities of the bands of interest increased linearly with the amount of cellular RNA (see Fig. 1 b).
The comparison of the distributions of the RET transcript levels of tumors to that of normal tissues (a.m. and a.g.1) was performed by the Student t test with the following data. One mean value (of at least three measurements) was used for each tumor. Because a.m. is only one specimen and a.g.1 is a pool of six adrenal glands,one mean value was used for the a.m. sample and six different measurements were used for the a.g.1 sample; the resulting P-value was 0.0031.
Ret Protein Analysis
Frozen tumor fragments (0.05–0.1 g) were crushed in liquid nitrogen,resuspended in 1 ml of ice-cold lysis buffer [10 mmTris-HCl (pH 8), 150 mm NaCl, 0.4 mm EDTA, 1%NP40, 10 mm NaF, 10 mmNa2H2P2O7,2 mg/ml aprotinin, 2 mg/ml leupeptin, 100 mg/ml AEBSF, and 2 mm Na3Va04],incubated for 10 min on ice, and clarified by centrifugation(19). Four mg of proteins were immunoprecipitated with 10μl of anti-Ret antibody C-19 (Santa Cruz) as described previously(19) and then were fractionated on 8% SDS-PAGE gel and transferred to nitrocellulose membrane and probed for antiphosphotyrosine with antibody 4G10 (Santa Cruz; 19). To detect the Ret protein, the membrane was incubated in 100 ml of stripping buffer [62.5 mm Tris-HCl (pH 6.8), 2% SDS, 100 mm β-mercaptoethanol] during 30 min at 55°C, treated with T-TBS 5% nonfat milk, and probed with anti-Ret antibody C-19 (Santa Cruz) in 20 ml T-TBS for 2 h at room temperature. Detection was performed by an antirabbit secondary antibody (ImmunoPure) and chemiluminescence reagents (SuperSignal,Pierce).
We first analyzed the RET, GDNF, and GFRα-1 transcript levels in nineteen adrenal pheochromocytomas. All of the tumors were characterized and tested for the presence of RET-activating mutations in exons 10, 11, and 16 as described in “Materials and Methods.” Fourteen sporadic tumors presented no activating mutation 4 were MEN-2-associated tumors and 1 was sporadic harboring a RET-activating mutation. Several normal tissues (a.g.1 and an a.m. sample) were studied as controls; two samples of s.n. were also examined as additional RET- expressing tissues. To measure the absolute amounts and proportions of transcripts, we performed quantitative RPAs. These were carried out with total cellular RNA samples and in parallel with known amounts of purified RET synthetic RNAs used as standards (see “Materials and Methods”).
Labeled antisense RNA probes specific for the RET, GDNF, and GFRα-1 transcripts were first tested separately and then mixed to allow more reliable measurements (see “Materials and Methods”). In vitro transcribed RNAs corresponding to these three genes were also tested separately and mixed to be used as controls. Fig. 1 a shows a typical RPA experiment, in which separated and mixed control RNAs exhibit the protected RNA bands specific for the RET, GDNF, and GFRα-1 transcripts (synthetic RNAs Lanes); a short artifactual band migrating slightly above the RET specific band can be identified in these control Lanes. Fig. 1 billustrates the linearity of the measurements.
Expression of the RET Gene
We found that all of the tumors expressed RET transcripts in highly variable amounts (mean = 10.1, SD = 9.85 atmol/μg, Fig. 2,a). These RET values differed strongly from those found in normal tissues (0.74, 1.42, and 0.42 atmol/μg for s.n, a.m., and a.g.1 samples, respectively; Fig. 2,a). To determine whether these differences were statistically significant, the distributions of the RET tumor values were compared with the RET values for the a.m. and a.g.1 samples with the Student t test (see “Materials and Methods”). This showed that the two distributions were significantly different (the probability that they were similar was P = 0.0031). Interestingly, the distribution of the RET transcript levels among the tumors presented a bimodal aspect, revealing two different subsets of sporadic tumors (Fig. 2,b). Half of the sporadic tumors expressed RET transcripts at high levels (mean = 18.1, SD = 7.1 atmol/μg) ranging from 10 to 32 times the level found in the control tissues (mean = 0.9 atmol/μg). On the other hand, in the remaining half of sporadic tumors the RET transcript levels(mean = 2.1, SD = 1.7 atmol/μg) were dispersed around the values found in normal tissues. In addition, the three MEN-2A associated pheochromocytomas (tumors 2–4) as well as the sporadic tumor harboring a RET mutation (tumor 6) showed RET levels(12–29 atmol/μg) similar to those found in the highly RET-expressing sporadic tumors (Fig. 2 b). We note that no correlation could be observed between malignancy of tumors and the RET transcript levels.
Phosphorylation of the Ret Protein
Because stimulation of the Ret protein results in autocatalytic tyrosine kinase activity, we investigated whether Ret products were phosphorylated on tyrosine residues in sporadic tumors. We immunoprecipitated protein extracts with anti-Ret antibodies, followed by immunoblot with antiphosphotyrosine (Fig. 3,a, lower panel) and anti-Ret antibodies (Fig. 3,a, upper panel). A neuroblastoma cell line,Neuro-2A, transfected with the GFRα-1 receptor was used as control. The relative amounts of Ret protein in the tumors analyzed paralleled the amount of transcripts. In neuroblastoma cells, immunoreactivity of Ret with the antiphosphotyrosine monoclonal antibodies depended as expected on stimulation by the GDNF ligand (Fig. 3 a, Lanes 1 and 2).
In agreement with the RPA data, Ret was present in pheochromocytoma 3 and strongly reacted with antiphosphotyrosine antibodies, consistently with the fact that it is a MEN-2-associated tumor. In tumors 18 and 7,Ret was expressed at high levels and also strongly reacted with antiphosphotyrosine antibodies (Fig. 3,a). This is a surprising result because these tumors did not harbor any known Ret activating mutation. Interestingly, the amount of tyrosine-phosphorylated Ret in tumors 18 and 7 was comparable to that found in tumor 3, raising the possibility that, in these tumors, Ret may also play a role in tumorigenesis (Fig. 3,a). More generally, in all of the other sporadic tumors analyzed that presented high RET transcript levels (tumors 8, 10, and 14) Ret reacted with antiphosphotyrosine antibodies to levels similar to that of tumor 3(Fig. 3 b), raising the possibility that, in these tumors,Ret may also play a role in tumorigenesis. In low-Ret-expressing tumors 20 and 17, the amount of phosphorylated Ret was much less than that found in the MEN-2A tumor 3.
Finally, we investigated the expression of the tyrosine hydroxylase gene. This gene is a well- established marker of differentiated chromaffin cells. We found as expected that, in most of the tumors analyzed, TH was expressed at higher levels than in normal tissues (Fig. 4). On the other hand, we could not observe any correlation between RET and TH transcript levels.
Expression of the GDNF and GFRα-1Genes
The finding of tyrosine-phosphorylated Retwt in sporadic tumors suggested that either an autocrine or a paracrine stimulation could take place in these tumors. Because Ret activity is physiologically stimulated by growth factors of the GDNF family and requires the presence of the GFRαs (1, 2, 3, 4) as components of the same cell surface complex (7), we asked whether Ret-persistent-tyrosine phosphorylation might be induced by the presence of GDNF and GFRα-1 in the same tumor samples. We, thus,examined the GDNF and GFRα-1 absolute transcript levels in sporadic pheochromocytomas.
The GDNF transcripts were not detected in the a.m. sample but were detected in the a.g.1 to a level of 0.03 atmol/μg (Fig. 5). All of the benign (tumors 7 to 17) and two of three malignant tumors expressed GDNF. Most of them presented enhanced GDNF levels, up to 14 times larger than that found in the a.g.1 tissue sample, but no strong correlation was observed between the GDNF and RET levels. Two malignant tumors (tumors 18 and 19) showed high RET levels and in one of them(tumor 18) no GDNF transcript could be detected. Three MEN-2 associated pheochromocytomas (3 to 5) as well as the sporadic tumor 6 (harboring a RET mutation) showed enhanced GDNF levels.
The GFRα-1 transcript levels among the sporadic tumors did not present strong variations except for tumor 18 (this tumor showed no detectable GDNF transcript). The GFRα-1 mean value for sporadic tumors, 0.08 atmol/μg, was similar to the values observed in control tissues.
Whether the levels of GDNF and GFRα-1 transcripts reflect the corresponding protein levels remains to be investigated, but this is suggested by the analysis of several poly(A)+tumor RNA fractions, which showed GDNF and GFRα-1 levels similar to those found in the total RNA samples (data not shown). The overexpression of GDNF observed in most tumors, thus, supports the hypothesis that a paracrine or autocrine stimulation may occur in the process of Ret activation.
Ret-activating mutations cause different types of tumors,including medullary thyroid carcinoma and pheochromocytoma, of either sporadic or familial origin. Here we report that the normal Ret products are expressed at high levels and also are phosphorylated on tyrosine residues in a number of sporadic adrenal pheochromocytomas. Moreover, we find that the amount of Retwt phosphorylation in sporadic pheochromocytomas is similar to that found in a tumor where the RET gene is mutated and, therefore, constitutively active. Furthermore, the presence in these tumors of GDNF and/or GFRα-1 transcripts suggests the existence of an autocrine or paracrine stimulation involving the Ret signaling pathway that may contribute to the maintenance and/or development of a large subset of sporadic pheochromocytomas.
The finding that in a number of sporadic pheochromocytomas, Retwt is highly expressed and phosphorylated raises the question of its implication in the determination of the final neoplastic phenotype of these tumors. Indeed, enhanced expression of receptor tyrosine kinases and polypeptide growth factors are common features of many tumor and tumor-derived cell lines that are likely to contribute to proliferation or invasion (20, 21, 22). However, because of the causal role played by the Ret signaling in MEN-2 pheochromocytomas, the finding of enhanced expression of Retwt (in sporadic pheochromocytomas) assumes a more important significance. Indeed, the MEN-2A-like RETmutations induce chronic activation of Ret tyrosine kinase by forming stable disulphide bonds between Ret monomers, thus mimicking ligand-induced dimerization. On the other hand, among the sporadic pheochromocytomas with no activating RET mutations, 50% of tumors exhibited Ret protein expression and phosphorylation at levels close to those found in MEN-2A tumors. Thus, it seems reasonable to infer that the overall signaling triggered by Ret in these sporadic tumors may be quantitatively, and likely qualitatively, comparable to that triggered by Ret-2A in the familial ones. In this respect, Ret would cause similar biological effects in both tumor types (Ret-2A in familial versus Retwt in sporadic tumors). A main difference between sporadic and familial tumors may consist in the developmental stage at which Ret becomes chronically stimulated. In the case of the inherited mutated allele, we may assume that Ret becomes active as soon as expressed. However, in the case of Retwt in sporadic tumors, we have no indication about the stage at which its expression becomes deregulated and, thus, whether its activation is implicated in the initial steps of tumor progression. We cannot exclude the possibility that distinct genetic events determine the tumor formation, which in turn would induce the expression of the Ret ligand/receptor complexes.
As previously reported, in the a.m., Ret either is undetectable during rat embryogenesis or is expressed in only a small number of chromaffin cells in the adult tissue (23). Thus the high Ret levels observed in sporadic adrenal tumors may be explained by a clonal expansion of the highly Ret-expressing cells of the original tissue. However, the small number of Ret-positive cells detected in the normal tissue cannot lead to the observed large proportion (50%) of Ret-overexpressing tumors. A more likely possibility is that the Ret-positive chromaffin cells have a growth advantage; some of these cells would then be preferentially selected to give rise to tumor formation. This scenario is in good agreement with our hypothesis that the Ret activity plays an essential role in the neoplastic processes leading to the Ret-overexpressing sporadic pheochromocytomas.
Until now, four different ligands for Ret have been isolated, each of which activates the Ret tyrosine kinase in association with a member of the membrane-bound receptor family (GFRα-1–4). In this study we analyzed the expression of three of the components of the Ret ligand/receptor complex, i.e., Ret, GFRα-1, and GDNF. GFRα-1 was present at similar levels in all of the sporadic tumors and normal tissues analyzed, thus providing the necessary receptor for GDNF signaling. GDNF was expressed in all of the tumors but one, and frequently its expression was highly enhanced as compared with control tissues. The presence of GFRα-1 and GDNF strongly indicates that both participate in stimulating Ret tyrosine kinase activity in these tumors by an autocrine mechanism. The only exception was tumor 18; in this sample, Retwt was highly phosphorylated in tyrosine residues, even in the absence of GDNF. A likely possibility is that other ligands, either described or still unidentified, may replace GDNF to stimulate Ret in these tumors, or, alternatively, ligands may be provided by surrounding cells, thus stimulating Ret by a paracrine mechanism. Furthermore, even though unlikely, we cannot exclude the possibility that ligand-independent dimerization of Ret may take place in some of these tumors.
An alternative interpretation of these results is that Retwt signaling may contribute to the maintenance of the differentiated phenotype rather than to the neoplastic progression of these tumors. Indeed, the effects of Ret activity in neuroendocrine cells still remains controversial. In vitro experiments indicate that Ret causes differentiation (in some cases, even terminal differentiation)of neuroectodermal cells, including PC12 (24, 25, 26),neuroblastoma (27), and primary cultures from human pheochromocytomas (28). This is true either if the Ret kinase is activated by an oncogenic mutation or if it is induced by ligand stimulation (29). However, these in vitro observations seem at least in part, difficult to reconcile with the in vivo data. In fact: (a) the Ret oncogene causes tumor formation in MEN-2 syndromes;(b) transgenic expression of the Ret oncogene in C-cells of mice thyroid causes proliferation and tumors(30); (c) cells from either sporadic or MEN-2A tumors showed mitotic activity in vivo. However, when grown as primary cultures in vitro, they lack the ability to incorporate bromodeoxy-uridine, observed in the tumors from which they originate; moreover, in these cells, GDNF stimulation induces neurite outgrowth but not bromodeoxy-uridine incorporation (28);and (d) in our experiments, neither the expression of the RET gene nor the Ret protein activity correlate with the expression levels of the TH gene, a marker of chromaffin cells; the TH transcript levels appeared to be similar in all of the tumors analyzed, either familial or sporadic.
High expression levels of GDNF and RET genes were also observed in the MEN-2 tumors. Although it is conceivable that enhanced expression of mutated RET alleles would confer a growth advantage to MEN-2-associated pheochromocytomas, the presence of high levels of GDNF in these tumors was quite unexpected and difficult to reconcile with the ligand-independent activation of the RetC634 mutants. In tumors with a RetM918T mutation (tumors 5 and 6), it is conceivable that the ligand availability can further stimulate the receptor kinase activity (31). In contrast, mutations of the Ret cysteine C634 are believed to be sufficient to induce complete activation of the Ret receptor (32), and it seems unlikely that it can be further stimulated by binding of the ligand. However,our findings are consistent with the recently described in vitro effects of GDNF (28). Indeed, primary cultures from human MEN-2-associated pheochromocytomas with a Ret C634 mutant protein still respond to GDNF, which likely stimulates the wild-type Ret encoded by the nonmutated allele (28). Whether in these MEN-2A tumors, the acute stimulation of the wild-type Ret can participate to define the final phenotype remains to be determined.
In conclusion, we report an analysis of the expression levels of the genes that code for three elements of the Ret receptor complex in the normal adrenal tissues as well as in adrenal pheochromocytomas. High levels of phosphotyrosine-containing Ret molecules in one-half of sporadic pheochromocytomas indicate that a persistent stimulation of the Ret activity takes place in these tumors. This suggests that the Ret signaling pathway may be implicated in the pathogenesis of sporadic tumors which represent the large majority of pheochromocytomas. A more extensive analysis of sporadic pheochromocytomas is in progress to determine whether the presence of high RET expression may be relevant to further define a subset of these tumors.
Supported by the Centre National de la Recherche Scientifique (CNRS); the Institut National de la Santé et de la Recherche Médicale (INSERM); the Ligue Nationale Contre Le Cancer(LNCC); the Association Pour La Recherche Sur Le Cancer (ARC); and in part by PHRC Grant AOM95201 for the COMETE Network; the Associazione Italiana per la Ricerca sul Cancro (AIRC); the Consiglio Nazionale delle Ricerche, Target Project on Biotechnology; the Fondazione Telethon Grant A.097. H. L. H. was supported by fellowships from LNCC and ARC; N. C. B. was supported by a fellowship from the French Ministère de l’Education Nationale et de la Recherche; and V. d. F. was supported by EC Grant BIO4-CT97-5078.
The abbreviations used are: MEN, multiple endocrine neoplasia; RPA, RNases protection assay; atmol, attomole;GFRα, GDNF family receptor α; a.g.1, a pool of six total adrenal glands; Retwt, Ret wild type; a.m., adrenal medulla; s.n., substantia nigra.
We thank Anne Julien for technical assistance in DNA mutation analysis. We thank Prof. G. Vecchio for helpful discussions.