Spectrum and Prevalence of FP/TMEM127 Gene Mutations in Pheochromocytomas and Paragangliomas FREE

Pheochromocytomas and paragangliomas are chromaffin cell tumors of neural crest origin that arise from the adrenal medulla or extra-adrenal sympathetic paraganglia, respectively, and are frequently catecholamine secreting.¹ These tumors are usually benign and can occur as a single entity or as part of various hereditary tumor syndromes. Genetically, pheochromocytomas and paragangliomas are heterogeneous, with at least one-third of cases resulting from germline but not somatic mutations in 1 of several independent genes: RET, VHL, NF1, and succinate dehydrogenase (SDH) subunit B, C, and D genes.^2- 5 More recently, other candidate susceptibility genes have also been reported, including KIF1Bβ,^6- 7EgIN1/PHD2,⁸SDHAF2,^9- 10 and SDHA,¹¹ although these findings remain restricted to 1 or 2 occurrences of the reported mutations. Despite this broad spectrum of susceptibility genes, the molecular basis for the majority of pheochromocytomas and paragangliomas, including most of the sporadic and rare familial cases, remains unknown. These observations support the existence of additional pheochromocytoma susceptibility genes, which may account for some of the genetically undefined cases.^12- 13

We recently identified TMEM127¹⁴ (NM_017849.3) as the pathogenic target of the familial pheochromocytoma (FP) locus, which we previously mapped to chromosome 2q11.¹³ Germline FP/TMEM127 mutations were found both in familial and sporadic-appearing pheochromocytomas, with loss of the wild-type allele in tumor DNA consistent with its role as a tumor suppressor gene. FP/TMEM127 encodes a highly conserved 3-spanner transmembrane protein that localizes to multiple intracellular organelles and is linked to regulation of the mTORC1 signaling complex, a critical control node for protein synthesis and cell survival.¹⁴ All but 1 of the 7 initially discovered mutations were truncating and resulted in markedly reduced expression of the FP/TMEM127 gene in the evaluable tumors, suggesting that the mutations result in loss of FP/TMEM127 function in the tumor tissue.

In recent years, specific genotype-phenotype associations in pheochromocytomas due to mutations in the various susceptibility genes have spurred the development of guidelines for genetic screening of these patients, with a goal of improving the outcomes of affected and at-risk individuals.^3,5,15- 16 To define the prevalence and genotype-phenotype correlations in FP/TMEM127 -mutated cases, we sequenced the FP/TMEM127 gene in 990 pheochromocytoma and paraganglioma patients and examined 545 of these samples for germline deletions of the gene. We also determined the subcellular localization of 5 novel mutations by in vitro confocal microscopy.

A total of 990 samples from patients with pheochromocytomas and paragangliomas without mutations in RET, VHL, SDHB, SDHC, and SDHD and who had no clinical features of neurofibromatosis type 1 were studied. In addition, 319 of these samples were negative for SDHAF2 and 52 for KIF1Bβ gene mutations. A cohort of 898 samples were recruited from a multi-institutional collaborative effort derived from the International Familial Pheochromocytoma Consortium, which encompassed referral centers based in the United States (San Antonio, Texas, and Ann Arbor, Michigan), Italy (Padova, Florence, and Brescia), Spain (Madrid), England (Birmingham and London), Brazil (São Paulo), Belgium (Brussels), and Argentina (Buenos Aires). Written informed consent was obtained from patients in accordance with institutional review board–approved protocols from each center. Eighteen samples were obtained from the Cancer Therapy and Research Center Pathology Tumor Bank at the University of Texas Health Science Center at San Antonio. Diagnosis of pheochromocytoma and/or paraganglioma, including tumors of both sympathetic (thoracic or abdominal) and parasympathetic (head and neck) origin, was established following conventional procedures (including clinical, biochemical, and imaging tests) and the diagnosis was confirmed histologically in every case. Tumors were considered familial when more than 1 affected individual was identified in the family. The study was performed between January 2009 and June 2010.

We previously reported FP/TMEM127 sequence variations in a cohort of 103 patients with pheochromocytoma and/or paraganglioma.¹⁴ In the present study, 92 of these cases were used for genotype-phenotype associations with the 898 new cases described above. Eleven cases from the original cohort that had germline mutations in other known pheochromocytoma susceptibility genes were excluded.¹⁴ The main features of each series of the entire cohort are shown in Table 1.

Of the new series of 898 samples, DNA was isolated from blood in 774 cases and from tumor tissue in 124 cases, following standard procedures.¹⁴

A control group composed of samples of Europeans, South Americans (heterogeneous population mainly of Latino origin but also including non-Hispanic whites, Asians, African Americans, and indigenous Native Americans), North Americans (including both Hispanic and non-Hispanic whites and African Americans), and Asians totaling 1064 alleles was used as reference samples, as reported.¹⁴ In addition to these, 718 new alleles were screened for exon 3 of the FP/TMEM127 gene only. This group comprised 318 alleles of white ancestry and 404 alleles of South American origin similar in racial/ethnic composition to those described above.

To define the prevalence of FP/TMEM127 mutation in a large, multi-institutional cohort of pheochromocytomas and paragangliomas, all 4 exons of FP/TMEM127 were amplified by polymerase chain reaction (PCR) and directly sequenced as previously reported.¹⁴ Primer sequences are listed in eTable 1.

The accession numbers for nucleotides (NM_017849.3) and respective protein variations (NP_060319.1) detected in this study were deposited in the National Center for Biotechnology Information (NCBI) single-nucleotide polymorphism (SNP) database (http://www.ncbi.nlm.nih.gov/SNP/tranSNP/VarBatchSub.cgi).

Variations that included nonsynonymous sequence substitutions, insertions, deletions, duplications, or splice site or nonsense changes and that were absent in the control group were considered potentially pathogenic in the present analysis. Synonymous substitutions, variants affecting intronic regions not immediately adjacent to the intron-exon border, or changes detected in the control group or in the SNP database (dbSNP, NCBI) were not considered of pathogenic relevance for the purposes of this study. We also used 3 prediction software modules to evaluate the pathogenic potential of the identified variants. Two of them, PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/)¹⁷ and SIFT (http://sift.jcvi.org/),¹⁸ are algorithms that predict the possible effect of an amino acid substitution on the structure and function of a human protein based on physical/structural considerations (eg, amino acid properties) and sequence homology. The third program, NetStart 1.0, predicts initiation of translation sites in eukaryotes based on a training data set of mRNA derived from genomic sequences with known start sites (http://www.cbs.dtu.dk/services/NetStart/).¹⁹

To assess larger deletions that might disrupt the FP/TMEM127 gene and that would not be detected by direct sequencing, a quantitative multiplex PCR of short fluorescent fragments method was designed. The assay comprised 6 sets of primers spanning the 5′ UTR, exons 1 through 4, and 3′ UTR of the FP/TMEM127 gene (eTable 1). Each forward primer was labeled with a 6-FAM-5′ fluorescent tag, and fragments were amplified in a multiplex PCR as previously described.²⁰ In addition, amplicons spanning distinct chromosomal regions (chromosomes 1, 11, or 17) were included in the multiplex reaction as internal controls. Two to 3 normal DNA samples were run within each batch to serve as references for normal, 2-copy gene pattern. A total of 545 germline samples with high-quality, high-molecular-weight DNA were successfully processed through this assay. Results were analyzed using Peak Scanner software, version 1.0 (Applied Biosystems, Foster City, California) as previously reported.²⁰ Reliability of the assay was validated by the detection of monoallelic FP/TMEM127 loss in tumor DNA from previously reported mutant samples, which were known to carry loss of the wild-type TMEM127 allele¹⁴ and, thus, represented models of single-copy TMEM127 loss (eFigure 1).

Loss of heterozygosity (LOH) analysis is a classic method to test whether a gene involved in tumor susceptibility has features of a tumor suppressor gene. In classic tumor suppressor genes, LOH consists of loss of the wild-type allele in tumor DNA, or the “second hit” (the first hit being the germline mutation in hereditary tumors), which results in loss of functional copies of the putative tumor suppressor gene in the tumor tissue. To identify the second hit of FP/TMEM127 inactivation in samples carrying a mutation, LOH analysis was performed in 4 cases in which tumor DNA was available. Loss of heterozygosity was identified when the wild-type–mutant allele ratio was less than 0.5 based on sequence data from germline and tumor DNA, as described.¹⁴ Loss of heterozygosity data were available from 6 previously reported samples.¹⁴

To demonstrate the effect of the c.409 + 1G>T mutation on splicing of the TMEM127 transcript, RNA was obtained from peripheral lymphocytes of the patient carrying this mutation. Complementary DNA was prepared by reverse transcription using standard procedures¹⁴ and used for a PCR spanning nucleotides 180 to 677 of the FP/TMEM127 open reading frame (primer sequences listed in eTable 1), followed by direct sequencing of the products.

To begin to determine the effects of FP/TMEM127 mutations in vitro, we generated clones carrying 5 novel substitution mutations identified in the present series. Four of the mutations (c.280C>T, p.Arg94Trp; c.208G>A, p.Asp70Asn; c.419G>A, p.Cys140Tyr; and c.418T>C, p.Cys140Arg) were obtained using site-directed mutagenesis (Phusion, New England Biolabs, Ipswich, Massachusetts) from wild-type FP/TMEM127 sequence cloned into Hin dIII/ Eco RI sites of the pEGFP-C2 vector (Clontech, Mountain View, California), as previously described.¹⁴ Additionally, the c.3G>T (p.Met1?) mutation was generated by PCR, using as the start codon an in-frame methionine at position 85 and cloned as above. All constructs were verified by sequencing. The remaining constructs that led to truncating products had previously shown to be unstable for protein detection.¹⁴ Each construct, as well as the wild-type version, was transfected into HEK293T, HeLa, or MPC9/30 (mouse pheochromocytoma cell line²¹) cells. After 24 hours, cells were fixed and imaged by confocal microscopy as previously reported.¹⁴ Green fluorescent protein (GFP) fluorescence of transfected cells was examined for subcellular localization as punctate (wild-type) or diffuse patterns.

Statistical analyses were carried out using SPSS software, version 17.0 (SPSS Inc, Chicago, Illinois). Because FP/TMEM127 mutations were detected exclusively in patients with pheochromocytoma, only this group was used for statistical studies. Differences between pheochromocytoma mutation carriers and non–mutation carriers were assessed for sex, bilaterality, familial history, and malignancy using a χ² test or the Fisher exact test when appropriate. The Mann-Whitney test was applied for testing age differences. Nominal 2-sided P <.05 was considered statistically significant. A modified Bonferroni-corrected nominal threshold of P = .05/N* was used to correct for multiple hypothesis testing where N* is the number of independent comparisons. We further evaluated age at disease onset of the current series in comparison with 2 previously published pheochromocytoma/paraganglioma cohorts (the Spanish series⁴ and the Italian series⁵). A proportions-difference test for 2 independent samples was applied to determine the degree of heterogeneity between these series. This test showed that the cohorts were not heterogeneous and, thus, could be combined for the comparative analysis.

A total of 44 distinct FP/TMEM127 variants were detected in 990 samples from pheochromocytoma or paraganglioma patients (Figure 1, Table 2, and eTable 2). Of these, 19 mutations found in 20 patients were considered of potential pathogenic significance (2.02%) (Table 2). Thirteen of these variants were novel changes, while the remainder had been previously reported.¹⁴ Approximately half of the mutations (10/19) comprised small deletions, duplications, or nonsense or splice site substitutions that led to truncation or extension (1 mutation) of the predicted FP/TMEM127 product. The other 9 variants were nonsynonymous missense mutations affecting conserved codons of the FP/TMEM127 sequence (Table 2). None of these 19 sequence variants was detected in an ethnically matched control group comprising 1064 alleles (or, in the case of exon 3, 718 additional alleles). The constitutive (germline) nature of the mutation could be determined in all but 1 case in which only somatic tissue was available for the analysis (c.208G>A, p.Asp70Asn) (eFigure 1A). Loss of heterozygosity of the wild-type allele was detected in 4 new cases in which tumor DNA was available (Table 2 and eFigure 1A). None of the 545 germline samples analyzed for larger FP/TMEM127 deletions showed evidence of partial or complete deletion of the gene.

The mutations spanned all 3 coding exons of FP/TMEM127 (Figure 1), and 13 of them affected 1 or more transmembrane domains of the predicted protein (Figure 1, Table 2, and eFigure 2), highlighting the relevance of these domains for FP/TMEM127 function. All of the missense mutations were considered to be potentially pathogenic by at least 1 of the 3 prediction programs used to preliminarily evaluate the likelihood of pathogenicity of the novel missense variants (Table 2 and eFigure 2) and/or by in vitro assays, as described herein.

Next, we examined the association of FP/TMEM127 mutations with various clinical variables, including age at onset, tumor location, and familial history of pheochromocytoma. The mean age at development of FP/TMEM127- mutated tumors was 42.8 years (95% confidence interval [CI], 36.44-49.25 years) and the median was 41.5 years. This age at onset is similar to the mean age of nonmutated cases in this series, 43.2 years (95% CI, 41.88-44.52 years) (median, 45 years) and to the reported average diagnostic age for sporadic pheochromocytomas (47.01 years; 95% CI, 45.42-48.6 years).^4- 5 This is in contrast with patients with hereditary pheochromocytomas due to mutations in other susceptibility genes, in whom the disease has an earlier manifestation (eTable 3). The single exception are individuals with RET mutations (eTable 3), who can usually be distinguished by their unique clinical presentation.^2,4- 5

Furthermore, FP/TMEM127 mutation frequency in patients presenting with unilateral pheochromocytomas without a family history or other syndromic features (11/547 cases [2%])—ie, truly sporadic-appearing cases—is similar to that reported for mutations of the VHL and SDHB genes and higher than that described for other pheochromocytoma susceptibility genes (Table 3).^4- 5 In the group diagnosed after age 45 years, FP/TMEM127 showed the highest frequency of mutations (1.1%) compared with all other susceptibility genes (Table 3). When all 20 probands and other affected members from 4 different families were combined (n = 29), the age at diagnosis of pheochromocytoma was older than 40 years in two-thirds of these individuals and older than 50 years in almost one-third of the cases (Figure 2). In a single family in which multiple affected individuals were available for analysis, 7 of 11 individuals carrying the c.410-2A>C mutation had clinical disease by age 55 years, while 1 mutation carrier remained free of disease at age 60 years.¹⁴

All mutated tumors arose from the adrenal medulla. No mutations were detected among the 189 extra-adrenal tumors, including 60 head and neck paragangliomas. Overall, one-third of patients with a mutation had bilateral tumors (7/20; P = 2.7 × 10⁻⁴). FP/TMEM127 mutations were detected in 11% (7/65) of all bilateral pheochromocytomas without other genetic cause. None of the 9 bilateral cases diagnosed in the pediatric group (younger than 20 years) had an FP/TMEM127 mutation. A clear family history of pheochromocytoma was present in only a quarter of patients carrying a mutation (5/20; P = .005).

Most tumors with an FP/TMEM17 mutation were benign. However, 1 patient carrying a missense mutation (c.419G>A, p.Cys140Tyr) had multiple vertebral metastases. In another patient (c.116_119delTGTC, p.Ile41ArgfsX39), features of a more aggressive histological profile (capsular and vascular tumor invasion) were reported; however, stringent criteria for malignancy were not met.

All tumors with FP/TMEM127 mutations were catecholamine secreting, with no preferential production of either norepinephrine or epinephrine. Moreover, no recurrent clinical manifestations other than pheochromocytoma were detected in the FP/TMEM127- mutated cases, although other neoplasias were reported in 3 of these patients (Table 2).

Enforced expression of GFP fusion constructs of 5 novel mutations in HeLa cells revealed that these variants had a subcellular distribution distinct from that of wild-type FP/TMEM127, which shows a typical plasma membrane and punctate cytoplasmic localization pattern corresponding to endomembrane organelles (endosome, lysosome, and Golgi body) (Figure 3eFigure 3). The mutant proteins were localized diffusely within the cytoplasm (Figure 3). Similar results were obtained with 2 independent cell lines, HEK293T and MPC9/30 (eFigure 3), indicating that these effects were not cell-specific.

In this study, we assessed the prevalence of mutations of FP/TMEM127, a recently identified pheochromocytoma susceptibility gene, in a large, multi-institutional, ethnically diverse cohort of 990 individuals with pheochromocytomas and paragangliomas (Table 1). Overall, 19 mutations detected in 20 independent families were considered potentially pathogenic (2%) based on a combination of sequence conservation data, prediction algorithms, LOH, clinical segregation of the mutation in families, and in vitro studies.

The main phenotypic associations that could be drawn from our analysis suggest the following pattern in association with FP/TMEM127: (1) mutations were detected only in patients with tumors of adrenal localization (pheochromocytomas) but not with paragangliomas; (2) age at onset of tumors in patients with mutations was similar to that of patients with sporadic disease; (3) while combined bilateral and familial cases accounted for almost half of the mutant cases, more than one-third of the mutation carriers presented with a sporadic-appearing, benign pheochromocytoma after age 40 years; (4) malignancy was rarely found among FP/TMEM127 mutation carriers; and (5) no recurrent pattern of manifestations other than pheochromocytoma was detected among the affected individuals or families, although the presence of other neoplastic manifestations in a few patients, including medullary carcinoma of the thyroid, supports the idea that clinical disease associated with FP/TMEM127 mutations may mimic other pheochromocytoma-associated syndromes (eg, multiple endocrine neoplasia type 2A).

A clear familial history of pheochromocytoma was observed in only a quarter of cases, suggesting low penetrance of FP/TMEM127 mutant alleles. However, a comprehensive assessment of age-related penetrance of FP/TMEM127 -associated disease was not available in the current families and awaits further analyses. The preliminary genotype-phenotype associations reported in the present study can be used to further refine genetic testing priorities in patients with pheochromocytomas and suggest that FP/TMEM127 mutation screening may be recommended for patients presenting at an older age with adrenal tumors, especially but not exclusively those with bilateral disease.

Loss of the wild-type FP/TMEM127 allele was detected in all informative cases, suggesting a classic mechanism of tumor suppressor gene inactivation similar to other pheochromocytoma susceptibility genes. However, in contrast with a minority of patients with SDHB, SDHC, SDHD, or VHL gene deletions,^4- 5,22 our investigation of alternative mechanisms of primary gene disruption via germline deletions of the entire gene or specific exons did not reveal any abnormalities in more than 500 samples tested. These results suggest that small mutations are the main mechanism of FP/TMEM127 disruption, although it cannot be excluded that methylation of the gene promoter, which has not yet been characterized, can also occur in some cases.

Finally, the aberrant subcellular distribution of newly identified FP/TMEM127 mutations strongly suggests that intracellular distribution within the endomembrane system is relevant for FP/TMEM127 function. Future studies should determine quantitative intracellular effects of individual variants. We previously reported an effect of FP/TMEM127 on mTOR (mammalian target of rapamycin) signals.¹⁴ The mislocalization of mutated proteins raises the possibility that FP/TMEM127 might affect mTOR accessibility to its regulators and offers a useful platform to test functions of FP/TMEM127 in protein trafficking and mTOR signaling in vitro.