Positional Cloning of the Gene for Multiple Endocrine Neoplasia-Type 1

See allHide authors and affiliations

Science  18 Apr 1997:
Vol. 276, Issue 5311, pp. 404-407
DOI: 10.1126/science.276.5311.404


Multiple endocrine neoplasia–type 1 (MEN1) is an autosomal dominant familial cancer syndrome characterized by tumors in parathyroids, enteropancreatic endocrine tissues, and the anterior pituitary. DNA sequencing from a previously identified minimal interval on chromosome 11q13 identified several candidate genes, one of which contained 12 different frameshift, nonsense, missense, and in-frame deletion mutations in 14 probands from 15 families. The MEN1gene contains 10 exons and encodes a ubiquitously expressed 2.8-kilobase transcript. The predicted 610–amino acid protein product, termed menin, exhibits no apparent similarities to any previously known proteins. The identification of MEN1 will enable improved understanding of the mechanism of endocrine tumorigenesis and should facilitate early diagnosis.

Familial cancer syndromes have attracted widespread interest over the past decade, in part because of their potential to shed light on the general mechanisms of carcinogenesis. Positional cloning methods have led to the precise identification of the responsible gene for more than a dozen such disorders (1). In keeping with the hypothesis originally articulated by Knudson for retinoblastoma (2), most of the responsible genes are of the tumor suppressor type. In such a circumstance, affected individuals have inherited one altered copy of the responsible gene from an affected parent, but the tumors have lost the remaining copy (the wild-type allele) as a somatic event. Thus, the inheritance pattern is dominant, but the mechanism of tumorigenesis is recessive. The importance of gene discovery often extends beyond affected pedigrees, as the same tumor suppressor gene is often found to play a role (by mutation of both alleles) in sporadic cases of the same neoplasm.

Multiple endocrine neoplasia–type 1 (MEN1) (OMIM *131100) appears to be a compelling example of this paradigm, with prevalence estimates ranging from 1 in 10,000 to 1 in 100,000 (3, 4). Affected individuals develop varying combinations of tumors of parathyroids, pancreatic islets, duodenal endocrine cells, and the anterior pituitary, with 94% penetrance by age 50 (4). Less commonly associated tumors include foregut carcinoids, lipomas, angiofibromas, thyroid adenomas, adrenocortical adenomas, angiomyolipomas, and spinal cord ependymomas. Except for gastrinomas, most of the tumors are nonmetastasizing, but many can create striking clinical effects because of the secretion of endocrine substances such as gastrin, insulin, parathyroid hormone, prolactin, growth hormone, glucagon, or adrenocorticotropic hormone.

Nine years ago MEN1 was mapped (5) to chromosome 11q13 by linkage analysis (Fig. 1A). Subsequent investigation of a large number of pedigrees by many groups revealed no evidence of locus heterogeneity (6, 7). The identification of critical recombinants recently led to the conclusion that the candidate interval is bounded by marker D11S1883 on the centromeric side and marker D11S449 on the telomeric side (7) (Fig.1B).

Figure 1

Steps in the positional cloning of the MEN1 gene. Initial linkage to chromosome 11q13 (A) led to finer mapping by meiotic recombination and tumor loss of heterozygosity (LOH) analysis (B). Nearly complete bacterial clone coverage of the most likely candidate interval (PYGM to D11S4936) was achieved with BACs b137C7 and b79G17 and cosmids cSRL116b6, 23c9, and 114g4 (16), which could be assembled into two sequence contigs, C1 and C2 (C). DNA sequencing revealed several candidate genes, one of which (D) was found to harbor mutations in 14 of 15 probands. The arrow indicates the direction of transcription.

In a concerted effort to identify MEN1, we developed 18 new polymorphic markers in the MEN1 region of 11q13 (8) and constructed a fully overlapping 2.8-Mb contig map of yeast, bacteriae, and P1 artificial chromosome (YAC, BAC, and PAC) clones and P1 clones (9). We then carried out an intensive search for transcripts, which resulted in the identification of 33 candidate genes (10). To focus the search more precisely, we also took advantage of the observation that tumors arising in MEN1 patients are frequently found to have somatically lost the wild-type allele of markers in the vicinity of the gene (5, 11). Interstitial deletions or mitotic crossing-over events of this sort provide information on candidate interval boundaries. We used tissue microdissection to separate tumor cells from stroma (12) in a large number of familial MEN1 tumors and sporadic gastrinomas, and we found an entirely consistent minimal interval (Fig. 1B) bounded centromerically by marker PYGM (12-14) and telomerically by marker D11S4936 (14).

We analyzed the sequence of two BACs (b137C7 and b79G17) covering most of this interval (Fig. 1C) (15), as well as publicly available sequence of a few cosmids just telomeric to b79G17 (16). A total of eight transcripts were identified by comparison with expressed sequence tag (EST) databases and computer analysis for the likely presence of exons. Each of these transcripts was considered a possible candidate for MEN1.

One of these eight candidates, originally designated mu, was first identified by PowerBLAST matches (17) between shotgun sequence assemblies derived from b137C7 and 44 different ESTs in the dbEST database. Twenty-six of these ESTs were human clones isolated from seven different tissues; the remaining 18 ESTs were derived from mouse or rat libraries. Interestingly, 20 of the human ESTs had previously been assembled into a UniGene cluster and placed on the transcript map between markers D11S913 and D11S1314 (18).

These 26 human ESTs constituted a 1.9-kb cDNA contig. Northern (RNA) blotting (10) identified a transcript of 2.8 kb that was expressed in roughly equivalent amounts in all adult tissues tested, including pancreas, adrenal medulla, thyroid, adrenal cortex, testis, thymus, small intestine, stomach, spleen, prostate, ovary, colon, and leukocytes. Screening of a leukocyte cDNA library yielded an apparently full-length 2.8-kb clone whose sequence was then fully determined on both strands (Fig. 2). Comparison of the cDNA sequence with genomic sequence from b137C7 revealed that the mu gene contains 10 exons (with the first exon untranslated) and extends across 9 kb (Fig. 1D).

Figure 2

Predicted amino acid sequence of the protein encoded by the MEN1 gene, as derived from an apparently full-length leukocyte cDNA clone. The first methionine is associated with an excellent Kozak (26) consensus sequence (GCCATGG), and no other in-frame ATG codons are found upstream. The GenBank accession numbers for the cDNA (2772 bp) and genomic (9181 bp) sequences are U93236 and U93237, respectively.

Primers designed from intronic sequence were used to amplify exons from genomic DNA of affected members of 15 typical MEN1 families (19), and mutations were sought by the dideoxy fingerprinting (ddF) method (20). Two examples of abnormal ddF patterns are shown in Fig. 3, A and B (exons 2 and 9). Sequencing of polymerase chain reaction (PCR)–amplified material (Fig. 3E), or in some instances cloned products (Fig. 3C), was used to identify the nature of the abnormality. For 10 different mutations for which other affected family members were available for study (all except E363del and W436X), we confirmed that the observed alteration was inherited concordantly with the MEN1 phenotype (Fig. 3D) (21).

Figure 3

Detection of frameshift and nonsense mutations. (A) Analysis of exon 2 in a MEN1 patient and a normal control, using ddF to reveal pattern differences (arrows) indicative of a possible mutation (20). (B) Abnormal ddF pattern in exon 9 from a different patient. (C) Identification of a single nucleotide deletion by sequencing of a cloned exon 2 PCR product from the patient whose ddF pattern is shown in (A). The sequence shown is of the antisense strand; the mutation is 512delC. (D) This frameshift mutation was confirmed by detecting the presence of a new Afl II site in PCR-amplified exon 2 from this patient and two affected relatives. (E) Direct sequencing of the exon 9 PCR product from (B), revealing the presence of a heterozygous C → T substitution. Again the sequence is of the antisense strand; the mutation creates a stop codon (TGG → TAG or W436X).

A total of five frameshift mutations, three nonsense mutations, two in-frame deletions, and two missense alterations were identified (Fig.4). Two mutations (416delC and 512delC) were encountered twice in families not known to be related. None of these mutations were observed in an analysis of 71 normal DNA samples. Four relatively common polymorphisms—R171Q (CGG/CAG), L432L (CTG/CTA), D418D (GAC/GAT), and A541T (GCA/ACA)—were also encountered and were observed in 1.4%, 0.7%, 42%, and 4% of normal chromosomes, respectively (n = 142).

Figure 4

Summary of mutations identified in 15 unrelated MEN1 patients. The locations of the five frameshift mutations are shown above a diagram of the MEN1 gene, with the exons numbered; cross-hatched areas are untranslated. Two in-frame deletions of a single amino acid, three nonsense mutations, and two missense mutations are shown below the gene diagram. The 416delC and 512delC mutations were each encountered twice. Mutation abbreviations follow standard nomenclature (27).

The identification of mutations in 14 of 15 unrelated affected individuals leaves little doubt that the MEN1 gene has been identified. We propose the name menin for the 610–amino acid predicted protein product. Sequence analysis provides few clues to its normal function. There is no signal peptide, and, although there are four moderately hydrophobic regions in the NH2-terminal half of the protein, these are not likely to represent transmembrane domains. Three leucine-rich regions match the PROSITE signature for leucine zippers (22), but these regions are not amphipathic and have no strong coiled-coil potential, and this signature is known to generate many false positive matches. Nuclear localization signatures are absent. The protein sequence has several regions of low compositional complexity, including a very hydrophilic mixed-charge cluster between residues 446 and 491 (23). There is no detectable homology to the complete genomic sequence ofSaccharomyces cerevisiae.

The observation that many of the mutations detected (Fig. 4) would most likely result in loss of function of the protein product is consistent with a tumor suppressor mechanism. Such a mechanism distinguishes MEN1 from the related disorder multiple endocrine neoplasia–type 2, where activating mutations of the RET oncogene are responsible (24). Although, in the absence of examples of complete gene deletion, we cannot rule out the possibility of a dominant negative effect of the truncated menin protein product, the observation of mutations in which as few as 82 amino acids would be left intact (357del4, Fig. 4) makes this mechanism unlikely. It will be of great interest to determine whether, as predicted by the Knudson model (2), somatic mutations in the MEN1 gene are responsible for sporadic endocrine tumors, including the common parathyroid adenomas, which occur at an annual incidence of 154 per 100,000 in individuals over age 60 (25).

Now that the MEN1 gene has been cloned, it will be important to study the role of MEN1 gene diagnostics in younger at-risk individuals so as to assess the value of identifying or excluding the presence of a mutation before the onset of symptoms. Moreover, the application of a broad and powerful repertory of molecular genetic, cell biological, and animal model approaches can now be initiated to pursue an understanding of the molecular basis of this disorder, with the eventual goal of developing better therapeutic strategies.

  • * To whom correspondence should be addressed. E-mail: fc23a{at}

  • Present address: Department of Medicine, University of Sydney, Sydney NSW 2006, Australia.


View Abstract

Navigate This Article