Report

Pituitary Adenoma Predisposition Caused by Germline Mutations in the AIP Gene

See allHide authors and affiliations

Science  26 May 2006:
Vol. 312, Issue 5777, pp. 1228-1230
DOI: 10.1126/science.1126100

Abstract

Pituitary adenomas are common in the general population, and understanding their molecular basis is of great interest. Combining chip-based technologies with genealogy data, we identified germline mutations in the aryl hydrocarbon receptor interacting protein (AIP) gene in individuals with pituitary adenoma predisposition (PAP). AIP acts in cytoplasmic retention of the latent form of the aryl hydrocarbon receptor and also has other functions. In a population-based series from Northern Finland, two AIP mutations account for 16% of all patients diagnosed with pituitary adenomas secreting growth hormone and for 40% of the subset of patients who were diagnosed when they were younger than 35 years of age. Typically, PAP patients do not display a strong family history of pituitary adenoma; thus, AIP is an example of a low-penetrance tumor susceptibility gene.

Pituitary adenomas are common benign neoplasms, accounting for approximately 15% of intracranial tumors. Most common hormone-secreting pituitary tumor types oversecrete prolactin or growth hormone (GH); the oversecreted hormones, together with local compressive effects, account for substantial morbidity. Oversecretion of GH causes acromegaly or gigantism. Acromegaly is characterized by coarse facial features, protruding jaw, and enlarged extremities. Because of the slow development of the potentially severe symptoms of untreated acromegaly, including cardiac manifestations, the condition is difficult to diagnose early (1). Gigantism refers to excessive linear growth that occurs as a result of GH oversecretion when epiphyseal growth plates are still open, in childhood and adolescence. Genetic predisposition to pituitary tumors is believed to be rare (2).

We detected three clusters of familial pituitary adenoma in Northern Finland [supporting online material (SOM) text]. The most notable cluster displayed three cases of acromegaly or gigantism. Genealogy data reaching back to the 1700s had been generated by family members from the publicly available official population registries. Two first clusters could be linked by genealogy (Fig. 1A), and the third appeared separate (Fig. 1B). We hypothesized that a previously uncharacterized form of low-penetrance pituitary adenoma predisposition (PAP) would contribute to the disease burden in Northern Finland. We had previously characterized a population-based cohort of 54 patients diagnosed with GH-secreting pituitary adenoma (somatotropinoma) between 1980 and 1999 in Oulu University Hospital (OUH) (3). We identified pituitary adenoma patients in Northern Finland by using data on this cohort, patient interviews, and a computerized search for all cases with archived samples of pituitary adenomas at the OUH from 1978 to 2000. These data were linked to the pedigree information to identify additional affected relatives. Altogether, 11 affected individuals in family 1 were identified (Fig. 1A). The PAP phenotype—very-low-penetrance susceptibility to somatotropinoma and prolactinoma—did not fit well to any of the known familial pituitary adenoma syndromes, including multiple endocrine neoplasia type 1 (MEN1), Carney complex (CNC), isolated familial somatotropinoma (IFS), and familial isolated pituitary adenoma (1, 2, 4). These syndromes are familial, and the low penetrance of PAP appeared unique. By low penetrance, we refer to hereditary predisposition that relatively rarely leads to actual disease but which may cause much more effect on population level than high-penetrance disease susceptibility, which typically is very uncommon.

Fig. 1.

Pedigrees of two families with pituitary adenoma. (A) Family 1. (B) Family 2. Individuals available for the study are indicated by A2, A5, etc. Numbers within diamonds indicate numbers of children. Circles, females; squares, males; slashes through symbols, deceased. Generations are indicated by Roman numerals on the left. Generation I is from the 18th century. Pedigrees are modified for confidentiality.

To identify the PAP locus, we performed whole-genome single-nucleotide polymorphism genotyping for 16 individuals from family 1 (Fig. 1A) (5) (SOM text). Before any linkage information was obtained, we opted to perform two alternative affected-only analyses. One analysis considering only individuals with acromegaly or gigantism (somatotropinoma or mixed adenoma) as affected (high stringency), and the other considered individuals with any pituitary adenoma as affected (low stringency). We did this because the number of phenocopies for acromegaly and gigantism is much lower than that for prolactinoma.

Linkage analysis using high-stringency criteria provided evidence for linkage in chromosome 11q12–11q13 (5) (fig. S1), a region previously implicated in isolated familial somatotropinoma and including the MEN1 gene (611). Notably, no MEN1 mutations were detected in our sample set, compatible with published reports on familial pituitary adenoma (6, 1217). We genotyped this candidate locus using 36 markers in families 1 and 2 (table S1). The added maximum logarithm of the odds (LOD) score for these two families was 7.1 with high-stringency criteria. Families 1 and 2 shared the linked haplotype, which segregated perfectly with acromegaly, providing unambiguous evidence for disease locus identification. The linked region was between 61.7 and 69.0 megabases (Mb) (Ensembl, version 36, December 2005), harboring 295 genes. Although analysis with low-stringency criteria also showed linkage at this locus (5), two individuals with prolactinoma appeared to represent phenocopies (A9 and A10). The data derived using the high-stringency criteria was considered the cornerstone of subsequent gene identification efforts.

To detect genes with aberrant expression in blood samples of PAP patients and carriers, we obtained expression profiles for 16 individuals (nine PAP carriers from families 1 and 2, and seven controls) (5). There were 172 probe sets that mapped in the linked region. The two lowest P values were obtained for the two separate probe sets representing AIP (also known as XAP2 and ARA9, GenBank no. U78521.1) (P = 0.00026 and P = 0.00114) (table S2). Thus, AIP was chosen as the prime candidate for mutation analysis. One other gene, galectin-12 (LGALS12), was also chosen on the basis of decreased expression (table S2) and an association of galectin-3 to pituitary tumorigenesis (18). No difference was detected in MEN1 expression. The coding region of AIP was sequenced from normal tissue DNA. A nonsense mutation Q14X (where Q is Gln), perfectly segregating with the GH secreting adenoma phenotype in families 1 and 2, was identified (Fig. 2A). The mutation was absent in 209 local blood donors. LGALS12 analysis was negative.

Fig. 2.

AIP germline mutations found in Finnish and Italian PAP cases. (A) A Finnish nonsense mutation Q14X in exon 1. (B) A Finnish splice site mutation IVS3-1G>A. (C) An Italian nonsense mutation R304X in exon 6. Each mutated base is indicated by an arrow. The respective wild-type sequences are depicted on the right and shown for comparison.

To evaluate the contribution of AIP in the population-based material (3), we had DNA available from 45 of the 54 acromegaly patients belonging to the study cohort, including four cases from families 1 and 2. Out of 45 patients from the population-based cohort, 6 displayed Q14X, and one displayed IVS3-1G>A, affecting the splice acceptor site of exon 4 (Fig. 2B). We screened 219 local blood donors for the latter change, with negative results. The age at diagnosis, sex, and size of adenoma were compared between PAP (n = 7) and AIP mutation–negative (n = 38) patients. Differences in tumor size or sex distribution were not observed. PAP patients were significantly younger than mutation-negative patients (24.7 ± 10.7 versus 43.6 ± 11.9 years, P = 0.0003). For identification of PAP patients, young age at onset is a useful indicator; six out of the fifteen patients diagnosed under 35 years of age (40%) in the population-based series had PAP. In addition, we screened 10 unselected Finnish sporadic acromegaly patients, from which DNA and appropriate authorization was available, and found Q14X in 2 of them, a result compatible with findings in the population-based cohort.

Loss of heterozygosity analysis was possible in eight tumors from mutation-positive individuals, including five somatotropinomas, one mixed-type tumor, and two prolactinomas; loss of the wild-type allele was detected in all cases, showing that these tumors were null with respect to AIP (fig. S2). This finding strengthened the notion that PAP is associated with predisposition to both prolactinomas and somatotropinomas and indicated that AIP is likely to act as a tumor suppressor.

The possible role of AIP in pituitary adenoma predisposition in other populations was studied in three families with two affected individuals. Normal DNA from one German (15) and one Turkish familial somatotropinoma case, as well as two Italian siblings with somatotropinoma (16) were analyzed. Whereas no mutation was detected in the German and Turkish sample, the Italian siblings displayed a nonsense mutation R304X (where R is Arg) in exon 6 (Fig. 2C). The change was absent in 203 Caucasian controls from the United Kingdom and the Centre d′Etude du Polymorphisme Humain (CEPH) (5), as well as in 52 local (Treviso) blood donors. The phenotype in the siblings resembled that seen in Finns: young age at onset and no visible evidence of dominant transmission (table S6).

These data strongly associate loss-of-function mutations of AIP to PAP. AIP was identified by its interaction with the hepatitis B virus X protein (19). AIP forms a complex with the aryl hydrocarbon receptor (AHR) and two 90-kD heat-shock proteins (HSP90) (20) (Fig. 3). The R304X mutation removes the AHR binding region (21, 22). AHR is a ligand-activated transcription factor that regulates a variety of xenobiotic metabolizing enzymes (23). Dioxin-like chemicals display high affinity to AHR, which mediates most of the toxic responses of these agents. AHR also participates in cellular signaling pathways (24). AIP modulates the subcellular localization of AHR and prevents the AHR from undergoing nucleo-cytoplasmic shuttling (25). It also binds to and attenuates the activity of PDE4A5—a phosphodiesterase that modulates cyclic adenosine monophosphate (cAMP) signaling—as well as PPARα (26, 27). The mechanisms by which AIP exerts its tumor-suppressive action in the pituitary remain to be determined. Further work on the functional role of AIP should prove informative in revealing key cellular processes involved in genesis of pituitary adenomas, including potential drug targets.

Fig. 3.

Schematic figure of AIP. FK506-binding protein (FKBP)-–homology region and tetratricopeptide repeats (TPRs) are shown as colored boxes. The regions necessary for interaction with AHR and HSP90 proteins are shown with black lines. The exon boundaries of the AIP gene are marked with dashed lines. Identified germline AIP mutations in PAP patients are indicated by black triangles.

It has not been previously realized that genetic predisposition to pituitary adenoma, in particular the GH-oversecreting type, can account for a substantial proportion of cases. Our study not only reveals this aspect of the disease but also provides molecular tools for efficient identification of predisposed individuals. Without preexisting risk awareness, the patients are typically diagnosed after years of delay, leading to substantial morbidity. Simple tools for efficient clinical follow-up of predisposed individuals are available, underlining the importance of our findings.

Our results suggest that inherited tumor susceptibility may be more common than previously thought. The identification of the PAP gene indicates that it is possible to identify the causative genetic defects in the low-penetrance conditions even in the absence of a strong family history.

Supporting Online Material

www.sciencemag.org/cgi/content/full/312/5777/1228/DC1

Materials and Methods

SOM Text

Figs. S1 and S2

Tables S1 to S6

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article