Genetic analysis of RET, GFRα1 and GDNF genes in Spanish families with multiple endocrine neoplasia type 2A

Authors


Abstract

Multiple endocrine neoplasia type 2A (MEN 2A) is associated with specific germline missense mutations in the RET proto-oncogene. This locus encodes a receptor tyrosine kinase whose activation requires the formation of a multimeric receptor complex including GDNF as a ligand and GFRα1 as a coreceptor. In order to explore the role of RET, GFRα1 and GDNF genes in the variation of phenotypes observed in MEN2A families, we analysed germline mutations of these genes in 4 unrelated Spanish MEN2A families (23 cases studied). We found 2 novel variants corresponding to a single change in position + 47 (intron 12) of RET and position +22 (intron 7) of GFRα1. Furthermore, we observed strong co-segregation between 2 polymorphisms of RET [G691S (exon 11) and S904S (TCC-TCG, exon 15) (100%, Fisher's exact test, p< 0.001)]. More interestingly, we found that these polymorphisms occurred at a significantly high frequency in patients with age at onset < 20 years old (Kruskal-Wallis's and Fisher's exact test, p = 0.007). These findings suggest that the G691S and S904S variants of RET may somehow play a role on the age of onset of MEN 2A. © 2002 Wiley-Liss, Inc.

Multiple endocrine neoplasia type 2 (MEN 2) is an inherited cancer syndrome with 3 clinically distinct forms: MEN 2A, MEN 2B and familial medullary thyroid carcinoma (FMTC).1 All forms are transmitted as an autosomal dominant trait with a high degree of penetrance and variable clinical expression. The presence of medullary thyroid carcinoma (MTC) is a common clinical feature. In patients with FMTC, only the thyroid gland is affected, while patients with MEN 2A may develop phaeochromocytoma and primary hyperparathyroidism. Furthermore, MEN 2B is characterised by mucosal neuromatosis, ganglioneuromas of the intestinal tract and a marfanoid habitus.

It has been shown that specific germ line point mutations of the RET gene are responsible for the 3 forms of MEN 2.2–4 The RET proto-oncogene encodes a tyrosine kinase receptor implicated in neural crest tissue development and differentiation. The glial cell line-derived neutropic factor (GDNF) has been recognised as a ligand for RET.5 RET activation requires the formation of a multimeric receptor complex that includes GDNF as ligand and a glycosylphosphatidyl inositol (GPI)-anchored protein termed GDNF receptor-α (GFRα1) that functions as coreceptor.6, 7 Independently of GNDF and GFRα1 the multiprotein RET-signalling complex can contain other members, such the ligand neurturin,8 persephin9 or artemin10 and the coreceptor GFRα2,11–13 GFRα310 or GFRα4.14

The majority of MEN 2A cases have germline missense RET mutations involving 1 of 6 highly conserved cysteines of the extracellular cysteine-rich domain.2, 4, 15–19 Interestingly, loss of function germline RET mutations were discovered in Hirschsprung disease,20, 21 which is characterised by the absence of the intramural ganglia of Meissner and Auerbach in the hindgut and results in functional intestinal obstruction.22 Although GDNF and GFRα1 mutations do not appear to have a major role in the pathogenesis of familial or sporadic phaeochromocytomas, allelic variation at the GDNF locus may modify phaeochromocytoma susceptibility.23 Furthermore, both genes are strongly expressed in human glioma.24

Within each type of MEN 2, there are variations between members of the same family regarding the presence of 1 or several of the features of the disease and the age at onset.25, 26 Variations in phenotypes within the same family suggest a role for genetic modifiers, which may also work through quantitative effect.26 We undertook the present study to explore the potential differential role of GDNF, GFRα1 and RET genes in the variety of phenotypes detected within the MEN 2A families.

MATERIAL AND METHODS

Patients and DNA preparation

We identified 4 independent Spanish families as having individuals affected by MEN 2A. Venous blood was obtained for DNA studies, after informed consent, from 23 family members, 13 with MEN 2A and 10 unaffected. Diagnosis of MTC and phaeochromocytoma was based on documented pathological examination. DNA was extracted from blood lymphocytes according to standard procedures.27 Likewise, genomic DNA was obtained as control from 40 unrelated healthy individuals to check the frequency of G691S / S904S haplotype.

Genomic DNA was amplified by Taq DNA polymerase (Dynazyme, Madrid, Spain). Primers used in amplification of exons 10–16, 1–10 and 1–3 of RET, GFRα1 and GDNF genes, respectively, were described previously.21, 28–33 Additionally, 2 sets of primer sequences were designed for exons 10 and 11 of RET: exon 10 (F: 5′AGGAGGCTGAGTGGGCTA CGT 3′, R: 5′ GGACCTCAGATGTGCTGT 3′) and exon 11 (F: 5′ TGCCAAGCCTCACACCAC CC 3′, R: 5′ AGGGGATCTTCCCTGCCCGCA 3′).

SSCP, RFLP and DNA sequence analysis

The variants were detected using single-strand conformation polymorphism (SSCP).34 Some specific RET and GFRα1 polymorphisms (G691S, S904S and T366A) were screened in all samples by restriction fragment length polymorphism (RFLP) using digestion of amplified DNA with Ban I, Rsa I and Fnu4HI (New England Biolabs, Beverly, MA), respectively.

The nucleotide sequences of amplified products that show an abnormal electrophoretic mobility on SSCP analysis were determined by direct sequencing using the ABI Prism Dye Primer Kit (Perkin-Elmer, Oak Brook, IL) and analysed on an Applied Biosystems automated DNA sequencer (Perkin-Elmer).

Statistical methods

The relationship between type of RET mutation and presence of a given polymorphism was assessed using the Fisher's test. The same test was applied to judge the statistical significance of the joint distribution of a pair of polymorphisms. For co-segregation analysis, we checked whether the percentage of co-segregation was significantly different from expected, according to the binomial probability distribution. Furthermore, the difference between the percentage of co-segregation among MEN2A patients and their healthy relatives was evaluated applying the Fisher's test. Using the Wilcoxon's test, we discerned whether the number of positive alleles for RET or GFRα1 polymorphisms in MEN2A patients were significantly different from those found in healthy relatives.

The relationship between number of positive alleles and age of onset of the disease was considered, by first taking the age as a continuous variable and secondly using a cut-off of 20 years. Statistical significance was verified using Kruskal-Wallis and Fisher's tests.

RESULTS

DNA mutations of RET, GFRα1 and GDNF genes

In our study, we analysed germline DNA mutations of RET, GFRα1 and GDNF genes in 4 unrelated Spanish MEN2A families, with a total of 23 members studied. Thirteen of them were MEN 2A patients (9 MTC and 4 MTC and phaeochromocytoma). Figure 1 shows the pedigree of these families and the nucleotide changes detected. The MEN2A patients studied did not display hyperparathyroidism and at least 1 member of each family presented phaeochromocytoma. Therefore, they were classified as MEN2A-2.22

Figure 1.

Pedigrees of the 4 MEN2A Spanish families. We ascertained 4 independent Spanish families as having individuals affected by MEN 2A. Venous blood was obtained, after informed consent was given, from 23 family members (denoted with asterisk), 13 with MEN 2A and 10 unaffected, for DNA studies. Diagnosis of MTC and phaeochromocytoma was based on documented pathological examination. Black symbol indicates MTC + phaeochromocytoma, grey symbol MTC only and white symbol corresponds to unaffected individual. Circles and squares denote females and males, respectively. The small number below the patient symbol indicates the age of onset (in years) and the symbol + highlights homozygous polymorphisms. The arrows highlighted the probands.

Specifically, we screened exons 10–16 of RET since they are the most related to MEN 2A pathologies. In particular, the oncogenic RET mutation in exon 11 (codon 634) was identified in all MEN 2A cases. Seven of them, corresponding to families A and D, displayed the C634R mutation and the other 6 (families B and C) the C634Y. Furthermore, we found 2 previously described polymorphisms of RET gene,35, 36 G691S (exon 11) and S904S (TCC-TCG, exon 15). The frequencies of both alleles were similar in MEN 2A patients and healthy relatives (23 and 20%, respectively). A new single change (A-T) at position + 47 (i47AT, intron 12) of RET was detected in 3 families (A, B and D) (Fig. 1). Again, the frequency of this allele was similar in MEN 2A patients and healthy relatives (19 and 10%). In contrast, we did not detect any nucleotide change in the exons 10, 12–14 and 16 of RET. Specifically, the polymorphism of RET, L769L (CTT-CTG, exon 13), that has been described with a frequency of 26% in healthy control population37 was not detected in any sample analysed.

We carried out the analysis of GFRα1 gene focusing on exons 1–10, since different mutations related to other pathologies, have been described in those exons.28, 38 In our study, we detected 4 different polymorphisms of GFRα1 gene. Three of them were previously reported,28 Y85N (exon 3), N184N (AAC-AAT, exon 6) and T366A (exon 9) and the frequencies of these alleles were similar in MEN 2A patients and healthy relatives. We also found a new variant (G-A) at position +22 (i22GA, intron 7) but only in 2 samples (Fig. 1). We did not detect any nucleotide change in exons 1, 2, 4, 5, 7, 8 and 10 of this gene. Similarly, we screened the exons of GDNF32 in all samples (MEN 2A and normal controls) of our study. However, we did not detect any mutations in this gene.

Co-segregation of the polymorphisms

Owing to its possible application as markers in molecular epidemiology studies, we were interested in knowing the relationship among the above-described variants. Table I shows the analysis of co-segregation of polymorphisms detected in RET and GFRα1 genes. Interestingly, we observed a strong co-segregation between 2 RET variants: G691S (exon 11) and S904S (TCC-TCG, exon 15) (100%, Fisher's exact test, p < 0.001). We also found a strong negative correlation, although not statistically significant (0%, p = 0.083), between 2 polymorphisms of GFRα1: N184N (AAC-AAT, exon 6) and i22GA (the nucleotide change G-A at position +22 of intron 7). When we compared the co-segregation in MEN2A patients vs. healthy relatives, we detected only 1 case of different distribution, a frequency close to statistical significance (4/6 in MEN2A vs. 0/6 in healthy, p = 0.061). This situation corresponds to the polymorphisms i47AT (change A-T at position + 47 of intron 12) and T366A (exon 9) of RET and GFRα1 genes, respectively.

Table I. Relationship Between RET and GFRα1 Polymorphisms1
Variant 1 vs. Variant 2Agreement2Fisher's p-valueCo-segregation3Co-segregation5Fisher's p-value
+% observed% expectedp-valueMEN2AHealthy
  • 1

    For the co-segregation analysis, we check whether the percentage of co-segregation was significantly different from what was expected, according to the binomial probability distribution. Furthermore, the difference between the percentage of co-segregation among MEN2A patients and their healthy relatives was evaluated applying the Fisher's test.

  • 2

    Number of concordant samples over the total number analysed.

  • 3

    Percentage of positive samples for both polymorphisms among number of them that was positive for at least one. Confidence interval according with the binomial distribution.

  • 4Binomial probability test comparing observed and expected values.

  • 5

    Percentage of positive samples for both polymorphisms among number of them that was positive for at least one in MEN2A vs. healthy relatives.

G691Si47AT10/233/231.00023% (3/13)37%0.39543% (3/7)0% (0/6)0.192
G691SS904S14/239/23<0.001100% (9/9)100%100% (5/5)100% (4/4)
i47ATS904S10/233/231.00023% (3/13)37%0.39543% (3/7)0% (0/6)0.192
N184Ni22GA5/230/230.0830% (0/18)10%0.2490% (0/11)— (0/0)
N184NT366A3/235/230.36325% (5/20)36%0.36025% (3/12)25% (2/8)1.000
N184NY85N4/232/230.14211% (2/19)22%0.2809% (1/11)13% (1/8)1.000
i22GAT366A14/232/230.14222% (2/9)22%1.00025% (1/4)50% (1/2)1.000
i22GAY85N17/231/230.39517% (1/6)28%1.00033% (1/3)0% (0/1)1.000
T366AY85N10/231/230.6118% (1/13)27%0.20617% (1/6)0% (0/1)0.462
G691S
S904SN184N5/237/230.65739% (7/18)44%0.81336% (4/11)43% (3/7)1.000
G691S
S904Si22GA12/230/230.5020% (0/11)15%0.3870% (0/6)0% (0/5)
G691S
S904ST366A10/235/230.38338% (5/13)48%0.58543% (3/7)40% (2/6)1.000
G691S
S904SY85N9/230/230.1160% (0/14)23%0.0510% (0/7)0% (0/7)
i47ATN184N3/233/230.13715% (3/20)28%0.31715% (2/13)14% (1/7)1.000
i47ATi22GA15/231/230.52613% (1/8)22%1.00020% (1/5)0% (0/3)1.000
i47ATT366A11/234/230.36333% (4/12)44%0.56967% (4/6)0% (0/6)0.061
i47ATY85N14/233/230.14233% (3/9)43%0.74117% (1/6)67% (2/3)0.226

The G691S / S904S haplotype of RET may somehow play a role in the age of onset of MEN 2A

The genetic variations within families with the same oncogenic mutation provide a potentially interesting tool for the screening of modifier genes and, for the analysis of its effects on different aspects of MEN2A pathology. In this regard, the samples of our study correspond to the second and third generation (the majority of cases) of MEN2A families and the patients analysed were diagnosed when they exhibited clinical symptoms. All patients displayed MTC, although only 4 of them exhibited MTC and phaeochromocytoma (Fig. 1). Likewise, the age of patients (at onset of symptoms) was different. Given the small sample size, we decided to categorise the age at onset into 2 levels, taking the 33rd percentile of the observed distribution as cut-off. Therefore, we can classify the patients in 2 groups, namely, age at onset < 20 years old (4 patients) and ≥ 20 years old (9 patients). Furthermore, we observed that some MEN2A patients exhibited low number of RET and GFRα1 polymorphisms, whereas other exhibited high numbers of them. This difference was significant (Kruskal-Wallis's and Fisher's exact test, p = 0.01) to compare both groups of age (see Table II) and specifically was due to RET polymorphisms (p < 0.01). Thus, the group of age at onset ≥ 20 years old contained lower number of these polymorphisms than the group < 20 years old. This outcome was due to a different distribution (p = 0.007) of the variants G691S/S904S between both groups of age (see Table III). The age at onset was also analysed as a continuous variable (see Table III) and its distribution across G691S/S904S groups pointed to the same conclusion.

Table II. Relationship Among Number of RET and GFRα1 Polymorphisms vs. MEN2A and Age at Onset1
PolymorphismsHealthy people (n = 10) MEN2A patients (n = 13)Age at onset ≥ 20 (n = 9) Age at onset < 20 (n = 4)
MeanMed1Min–MaxMeanMed2Min–MaxMeanMed2Min–MaxMeanMed2Min–Max
  • 1

    The relationship between number of polymorphisms in RET or GFRα1 vs. the age of onset was considered using a cut-off of 20 years.

  • 2

    Med = Median value.

  • 3

    p-value according with Wilcoxon's test.

RET1.0010–21.3110–40.5600–33.0032–4
p-value30.74<0.01
GFRα11.5010–31.6221–21.5621–21.7521–2
p-value30.560.73
Total2.5030–42.9231–62.1121–54.7554–6
p-value30.780.01
Table III. Relationship Between RET and GFRα1 Polymorphisms vs. Age at Onset1
PolymorphismsAge at onsetKruskal-Wallis' p-valueAge at onset < 20Fisher's p-value
nMedianMin–Maxn%
  • 1

    The relationship was considered taking firstly the age as a continuous variable and secondly using a cut-off of 20 years. The statistical significance was verified using Kruskal-Wallis's and Fisher's tests.

  • 2

    Categories + and ++ collapsed.

G691S/S904S
 0826.521–370.079200%0.0072
 +417.014–43375%
 ++111.01100%
i47AT
 0824.011–280.379225%1.000
 +530.014–43240%
N184N
 0330.014–370.236133%0.266
 +726.016–43114%
 ++318.011–22267%
i22GA
 01224.011–430.181433%1.000
 +137.000%
T366A
 0824.011–300.714225%1.000
 +526.014–43240%
Y85N
 01122.011–430.374436%1.000
 +228.026–3000%

We have analysed the allelic frequency of G691S/S904S in a set of 40 controls (unrelated healthy individuals) and we did not observe a significant difference in the allelic frequency between controls, MEN2A patients and their relatives (see Table IV). We detected only 1 homozygote G691S/S904S among MEN2A patients, but we can not conclude that the distribution is different to the rest of groups, since the 95% confidence intervals completely overlap.

Table IV. G691S/S904S Allelic Frequency Among Controls, MEN2A Patients and Healthy Relatives
GroupNumberAllelic frequencyp-value1Percentage of heterozygous (95% C.I.2)Percentage of homozygous (95% C.I.2)
  • 1

    Comparisons with the control group using Fisher exact test.

  • 2

    95% Confidence Interval according to the binomial distribution.

Controls4016.3%32.5%0.0%
(18.6–49.1)(0.0–8.8)
MEN2A1323.1%0.55630.8%7.7%
(9.1–61.4)(0.2–36.0)
Relatives1020.0%0.74240.0%0.0%
(12.2–73.8)(0.0–30.8)

Taken together, these results suggest that early appearance of symptoms in MEN2A patients could be related to the presence of polymorphisms G691S/S904S of RET and these polymorphisms could be considered as genetic modifiers. In accordance with this possibility, the youngest patient analysed (11 years old), 3-III-Family C (Fig. 1), contained the C634Y oncogenic mutation associated with the homozygous polymorphisms G691S / S904S, suggesting a genetic dose effect.

DISCUSSION

In our study, we identified an oncogenic RET mutation in exon 11 (codon 634) in all MEN2A cases. Seven of them, corresponding to families A and D, displayed the C634R mutation and the other 6 (families B and C) the C634Y. Note that the results of a study carried out by the International RET Mutation Consortium showed that the most frequent mutation at this codon was C634R (52%) followed by C634Y (26%).15 However, other study denoted high prevalence (73% of cases) of the C634Y mutation in MEN 2A Spanish families.19 Two new variants were detected at position +47 (i47AT, intron 12) of RET and at position +22 (i22GA, intron 7) of GFRα1 gene, respectively, although the frequencies of these alleles were similar in MEN 2A patients and healthy relatives.

Since several reports suggested that allelic variation in GDNF gene could influence the phenotype caused by RET mutations,23, 39 we screened the exons of GDNF32 in all samples (MEN2A and normal controls) of our study. However, we have not detected any mutations in this gene, suggesting, in accordance with previous reports,40 that GDNF is not playing a critical role in MEN2A pathology.

We observed that 2 variants of RET, G691S (exon 11) and S904S (TCC-TCG, exon 15), co-segregated together as haplotype (100%, Fisher's exact test, p < 0.001), suggesting that these polymorphisms are in linkage disequilibrium with each other. Interestingly, this RET haplotype (G691S and S904S) displayed a different distribution (p = 0.007) according the age at onset of MEN 2A patients.

RET polymorphisms G691S and S904S were previously described35, 36 and until now it is not known whether these germline variants play other interacting, predisposing or modifying roles in the pathogenesis of MEN2A. The precise mechanism by which these polymorphisms may affect the age of onset is unknown and open to speculation. The conservative amino acid substitution G691S (exon 11) occurs close to the extracellular cysteine rich domain of RET and, although it is not considered an oncogenic mutation, we cannot exclude a functional role based on quantitative effects. Since S904S (TCC-TCG, exon 15) does not lead to an amino acid alteration, it is difficult to imagine how this conserved polymorphism may affect RET activity. A plausible explanation is that, due to the 100% of co-segregation, the results obtained with S904S could be interpreted as a founder effect lacking influence as genetic modifier and the critical mutation would be G691S. However, it is possible that the sequence variant S904S has influence on RET expression. In this regard, it has been shown that polymorph variants can lead to production of different amounts of mRNA.41 Likewise it has been suggested that the rare RET sequence variant S836S (AGC-AGT) may play a role in the genesis of sporadic MTC.37 In addition, it has been reported that 2 polymorphisms of RET, A45A (GCG-GCA; exon 2) and L769L (CTT-CTG; exon 13), exhibit a strong association with Hirschsprung disease.42, 43

In light of these results, our hypothesis is that the G691S/S904S haplotype of RET may somehow influence the age of onset in MEN2A patients and could be considered as genetic modifier of this pathology. Interestingly, these 2 polymorphisms appear to be under-represented in Hirschsprung patients compared with controls, suggesting that they might protect against the development of that disease in low penetrance manner.43 If not pathogenic/modifier effect can be associated with the G691S/S904S polymorphisms of RET, an alternative hypothesis is a linkage disequilibrium with an unknown functional variant.44 Nevertheless, our results are based on a small sample size, so we do not know to what extent they could be generalised to the majority of patients with this disease. Despite the limited statistical power of our study, the associations highlighted here might deserve further study in a broader sample of patients. It remains for further large-scale molecular epidemiology studies to determine whether this is a phenomenon more widespread for different populations and for functional analyses to demonstrate that these variants indeed affect RET activity.

Acknowledgements

This work was supported by grants to JMR fromRatiopharm España S.A., CAM (08.6/0005/1997) and FIS (98/1336). LG and BA were recipient of fellowships from Instituto de Salud Carlos III (Spain) and FIS, respectively.

Ancillary