ret is the only gene known to be involved in the inherited predisposition to MEN 2 and the only gene known to cause a predisposition to MTC. Germ-line mutations in other genes may predispose to pheochromocytoma in VHL syndrome (see Chap. 41) and neurofibromatosis type 1 (see Chap. 22). It is not clear whether there is an additional syndrome of site-specific pheochromocytoma or whether all these families will fall within VHL. Germ-line mutations in the Menin gene in MEN 1 predispose to parathyroid tumors.4-6
Mutations of Ret in MEN 2
Identification of Ret as a Proto-Oncogene.
The ret proto-oncogene is a cell-surface glycoprotein that is a member of the receptor tyrosine kinase (RTK) family.58 The name ret is an acronym for “rearranged during transfection,” reflecting the original identification of ret as a chimeric oncogene formed by rearrangement during transfection assays using DNA from human lymphomas and gastric tumors.59 Three different rearranged versions of ret have since been described in vivo, specifically in papillary thyroid carcinomas (which arise from thyroid follicular epithelial cells and are therefore distinct from the C-cell-derived MTC).60-62 These rearranged versions are termed ret PTC-1, -2, and -3. In each case, the effect of the rearrangement, which occurs as a somatic event, is to fuse the tyrosine kinase region of ret with different activating sequences that are expressed in thyroid epithelial cells. The fused activating genes contribute a new N-terminal portion to the ret protein that is capable of dimerization, leading to activation of the tyrosine kinase domain independent of any ligand. Mutations of the ret-PTC type are not seen in MEN 2-related tumors.
Identification of Ret in MEN 2.
ret lies in the region of the MEN 2 locus defined by linkage analysis and was therefore a candidate gene.63,64 At that time, ret was known as a proto-oncogene, whereas all the tumor-predisposing genes identified to that point acted as suppressor genes. The plausibility of ret as a candidate was, however, strengthened by the first reports that the ret knockout mouse had a phenotype that resembled HSCR 65 and the known association of MEN 2 and HSCR in some families. Mutation analysis of ret in MEN 2 families revealed mutations in the extracellular domain of the gene in MEN 2A and FMTC 66-68 and subsequently in the tyrosine kinase domain in MEN 2B.69-71
The coding sequence consists of 21 exons in a genomic sequence of approximately 55 kb.72,73 The protein exists in three main 39 alternatively spliced forms of 1072 to 1114 amino acids.74 There is a cleavable signal sequence of 28 amino acids; an extracellular domain, which is glycosylated and has a conserved75,76 cysteine-rich region close to the cell membrane and a region of cadherin homology further out77 ; a transmembrane domain; and a tyrosine kinase domain with a short interkinase region of 27 amino acids (Fig. 42-4). Further details of the structure are given in references72-74 and 78–80.
The main features of the protein encoded by the ret proto-oncogene and the sites of the mutations in the different clinical varieties of MEN 2.
Ligands for the Ret Receptor.
Three ligands have so far been identified: glial cell line-derived neurotrophic factor (GDNF), 81 neurturin, 81a and persephin.81b These are structurally related secreted proteins that are widely expressed in the nervous system and in other tissues and promote the survival of neurons during development.81c, 81d Each signals through a multicomponent receptor consisting of ret and one of the GFRα family of glycosyl-phosphatidylinositol (GP1)-linked proteins.81e, 81f, 81g, 81h, 84,85 GFRα1 is the preferred ligand-binding protein for GDNF; GFRα2, for neurturin; and GFRα4, for persephin.81e, 81f, 81h Mouse knockouts of the GDNF gene have a phenotype similar to that of ret knockouts.82,83 The distinct and overlapping roles of the different ret-ligand combinations in development are currently under study.
Germ-Line Ret Mutations in MEN 2.
A summary of the mutations is given in Table 42-3 and Fig. 42-4. With one exception, an in-frame insertion, 85a all MEN 2 germ-line mutations so far identified are point mutations that lead to amino acid substitution.
Mutations in MEN 2A and FMTC.
The majority of mutations in MEN 2A and FMTC lie in one of five cysteine codons in the cysteine-rich region of the extracellular domain11,38 and result in substitution of the cysteine by another amino acid. A few mutations in families with MTC have been found in exons 13, 14, and 15 of the intracellular domain, 86,87 and a single MEN 2A family has been reported with mutation in codon 790 (exon 13)55a (see Figs. 42-4 and 42-5). Figure 42-5 shows clearly that there is a correlation between the codon involved in the mutation and the MEN 2 phenotype.11 In families with MEN 2A with both pheochromocytoma and parathyroid involvement, almost all the mutations are in codon 634; in MEN 2A families lacking pheochromocytoma and in FMTC, codons 609, 611, 618, and 620 are more frequently involved. This correlation is highly significant; 160 of 186 families with at least one proven case of pheochromocytoma have mutations in codon 634, compared with 18 of 43 families with no evidence of pheochromocytoma (p < 0.0001).11, 11a There also may be an effect not only of the position of the cysteine codon but of the particular amino acid substitution involved. Mutations at codon 634 seen in MEN 2 include all the possible amino acid substitutions allowed by the coding sequence. The most common changes are cysteine to arginine (C634R; TGC→CGC) and cysteine to tyrosine (C634Y; TGC→TAC), which may reflect the known frequency of T→C and G→A changes rather than the particular biologic significance of these substitutions. Nevertheless, it is intriguing that while C634R was present in 88 of 169 MEN 2A families with a codon 634 mutation, none of 9 FMTC families with a codon 634 mutation had this change.11 Furthermore, Mulligan et al.38 found a highly significant association between the C634R mutation, compared with all other 634 mutations, and the presence in the family of parathyroid disease. This, however, has not so far been replicated in an independent study.11a Four families (three meeting the criteria for FMTC and one “other”) have been identified with a glu→asp mutation in codon 768 (exon 13), 86,87 and two families with MTC have been placed in the “other” category with a leu→val mutation in codon 804 (exon 14).86 Mutations in codons 790 and 791 have been reported in several small families with MTC and one with MEN 2A.55a Thus mutations in this region of the intracellular domain seem mostly to be specifically associated with MTC rather than with pheochromocytoma or parathyroid disease, although the MEN 2A family with codon 790 mutation indicates that this correlation may not be absolute.
Proportion of mutations in different codons of ret in different phenotypic subtypes of MEN 2A, FMTC, and other MTC families. Based on data from the International ret Mutation Consortium summarized in Table 42-3.
Some 95 percent of MEN 2B families reported to date have an identical mutation: methionine → threonine in codon 918 of exon 16.11,69-71,88 Each of four families lacking this mutation had typical and well-documented phenotypic features.89 In four families without M918T mutation, a mutation of A883F recently has been reported.89a, 89b
Mutations in Families with MEN 2 and HSCR.
Each of the 17 or so families reported has a mutation in either cys 609, cys 618, or cys 620.33,37,90, 90a, 90b, 90c No other mutation has been found after careful examination of the remainder of the gene in these families, and so the conclusion must be that the same mutation can result in apparently contrasting phenotypes in the same individual. Families with HSCR alone with no evidence of MEN 2 also have been reported to have missense mutations in cysteine codons 609 and 620.35,36
Expression of Ret in Development.
Three of the tissues principally involved in MEN 2—thyroid C cells, adrenal medulla, and intestinal autonomic ganglia—are derived from neural ectoderm.91 The parathyroids are derived from the endoderm of the third and fourth pharyngeal pouches. The lineage relationships between C cells and other cells of neuroectodermal origin are still unclear, but the origin of C cells from vagal neural crest and the biochemical similarities with enteric neurons92 suggest that they share a common precursor with enteric neurons and ultimately with the sympathoadrenal progenitor that is the precursor of chromaffin cells and sympathetic neurons.93
In situ hybridization studies during mouse and rat development show that ret is expressed in the neural crest-derived cells that migrate from the region of the hindbrain into the posterior pharyngeal arches and from there to form the thyroid C cells and the vagal neural crest that gives rise to the intestinal autonomic nerves.94,95 ret is also expressed in migrating cells derived from the trunk neural crest as they coalesce alongside the aorta to form the sympathetic ganglia and the chromaffin cells that will form the adrenal medulla and in the endoderm of the pharyngeal pouches that give rise to the parathyroids.94,96 The expression of ret is therefore consistent with a role in the development and differentiation of the tissues that are involved in MEN 2. It is perhaps surprising that ret homozygous knockout mice appear at birth to have absent intestinal autonomic ganglia but normal C cells and adrenal medulla.65 The mice die at this stage, probably of respiratory or kidney failure resulting from other developmental defects, and so the possible role of ret expression in postnatal development cannot be assessed. However, the tentative conclusion must be that while disordered ret expression can lead to tumor formation, normal ret expression is not necessary for C-cell or adrenal medullary development up to the time of birth. A caveat is that while the C cells and adrenal medulla may appear grossly normal, the development of the cells may have been perturbed in some way that is not readily apparent. There is a further possibility, with some evidence to support it, that the C-cell population is heterogeneous, with only some C cells expressing ret. 95,97 It may be, therefore, that the C cells that are seen in the knockout mice are only one component of the population, with the other component being absent.
In the later stages of embryogenesis in the mouse and in rodent and human thyroid and adrenal medullas after birth, there appears to be only weak and patchy expression of ret by the criteria of in situ hybridization and immunohistochemistry.95,97 In most MTCs and pheochromocytomas, by contrast, ret is expressed at high levels.98,99 At present, nothing is known about the role of ret in C-cell or adrenal medullary development or in the adult glands. The mechanism and significance of the apparent increase in the expression in tumors are uncertain but may in part have a trivial explanation in terms of stabilization of the ret mRNA or protein as a result of the mutation.100
Function of Ret at the Cellular Level.
ret is a receptor tyrosine kinase (RTK). Binding of ligand results in dimerization of the receptor, activation of the tyrosine kinase, and initiation of onward signaling pathways.101 Evidence is slowly accumulating, 102-106, 106a, 106b but there is no coherent picture of the signaling events that follow ret activation. Analysis is complicated by the three 39 alternative splice forms of ret, which might be predicted from the sequence context of their tyrosines to differ in the affinity with which they bind different signaling molecules and may therefore signal through different pathways, and by the likelihood (supported by some evidence106) that the pathways of signaling are specific for different cell types, which implies that studies should be done in cells that resemble as closely as possible those involved in MEN 2.
Consequences of Ret Mutations.
Transfection experiments101,104 have shown that both the MEN 2A (cys 634 arg) and the MEN 2B (met 918 thr) mutations lead to activation of ret tyrosine kinase. The evidence is of two types: (1) biologic, in which transfection of mutant but not wild-type ret induces transformation of NIH 3T3 cells and differentiation of rat PC12 (pheochromocytoma) cells, and (2) biochemical, in which the ret protein becomes phosphorylated on tyrosine and acquires tyrosine kinase activity against added substrates.
Extracellular Domain Cysteine Mutations.
The cysteine mutations activate ret by inducing covalent dimerization101,104,107,108 (Fig. 42-6). The genotype-phenotype correlation observed with different cysteine mutations is probably explained by quantitative differences in signaling. The different cysteine mutants have been shown to differ both in their efficiency of dimerization and in their maturation to the fully glycosylated form, which is necessary for insertion into the plasma membrane. Either or both of these differences may be responsible for differences in the level of ret activation.108a, 108b
Constitutive activation of ret as a result of mutation of a cysteine in the extracellular domain.
The Met 918 Thr MEN 2B Mutation.
This mutation has proved to be of considerable interest because one effect of the mutation is to convert the substrate specificity of the ret tyrosine kinase from that typical of a receptor tyrosine kinase (RTK) to that typical of a cytoplasmic tyrosine kinase. It also confers some activity on the receptor, independent of dimerization (see below).
Residue 918 is predicted from modeling studies to lie at the base of a pocket in the protein that is involved in substrate binding.58,69 The substitution of threonine for methionine is predicted to alter the dimensions of the pocket and thus the sub-strate specificity. Tyrosine kinases fall into two classes: RTKs and cytoplasmic tyrosine kinases. Almost all RTKs have methionine at the equivalent position to codon 918, whereas almost all cytoplasmic tyrosine kinases have threonine (Fig. 42-7).58 Songyang et al.109 used degenerate peptide libraries to demonstrate that whereas RTKs prefer hydrophobic amino acids at positions 11 and 13 downstream of the target tyrosine in their substrate, cytoplasmic tyrosine kinases prefer a hydrophilic residue at 11 and a hydrophobic residue at 13. These different amino acid contexts flanking the tyrosine provide different preferred substrates for different groups of SH2 domains on signaling molecules and hence the possibility of different pathways of downstream signaling.
Amino acid sequence of consensus receptor and cytoplasmic tyrosine kinases in the substrate binding region of Hanks domain VIII, showing the mutation characteristic of MEN 2B.
When wild-type (equivalent to MEN 2A) and MEN 2B ret tyrosine kinases were compared for their ability to phosphorylate model substrates, a clear shift toward the specificity characteristic of a cytoplasmic tyrosine kinase was seen with the MEN 2B mutation.109 The inference that the MEN 2B mutation has altered the pathway of downstream signaling is supported by the observation that activated MEN 2B ret differs from activated wild-type ret both in the pattern of tyrosine phosphorylation of the ret protein itself110 and in the patterns of tyrosine phosphorylations seen in cell extracts.104
The MEN 2B mutation does not lead to covalent dimerization of ret. 104 However, there is in vitro evidence that it does lead to activation of the receptor through an intramolecular mechanism.101,104,105 Activation by this mechanism may be further enhanced by ligand binding, and the combined result may be a higher activity than is conferred by the cysteine MEN 2A mutations. It is unclear whether this increased activity, or the altered substrate specificity, is responsible for the characteristic features of the MEN 2B phenotype.
The 768 and 804 Mutations.
The glu 768 asp mutation86,87 involves a residue that is highly conserved in different receptor tyrosine kinases (RTKs). Modeling suggests a possible effect both on ATP binding and on substrate specificity, each of which is being tested experimentally. The val 804 leu mutation86 also affects a conserved residue, but no modeling studies have been reported.
Genes Involved in Somatic Mutations
The genetic events in the progression of MEN 2 tumors and in the initiation and progression of the related nonhereditary tumors are largely unknown. Baylin et al.111 showed that the tumors are clonal. Several candidate oncogenes have been screened for mutations or altered expression (N-ras, Ha-ras, N-myc, c-myc, 1-myc, c-mos, β nerve growth factor, and the low-affinity nerve growth factor receptor).112,113 Apart from one report of overexpression of N-myc in 6 of 21 MTCs analyzed by in situ hybridization, 114 all the studies were negative.
The role of mutations of ret in sporadic tumors is of some interest. In the inherited cancer syndromes, nonhereditary tumors often have mutations in the same gene that shows germ-line mutations in hereditary cases. In MEN 2 this is partly true, with an interesting twist. Among 157 apparently sporadic MTCs in the literature, 39 percent are reported to have a somatic mutation in ret codon 918 (the MEN 2B mutation), but mutations of the cysteine codons characteristic of MEN 2A and FMTC are uncommon, probably on the order of 1 to 3 percent.28,67,115, 115a, 115b Somatic mutations in the intracellular domain are seen a little more frequently, e.g., codon 768 in 4 of 72 sporadic MTCs87, 87a, 116 and codons 790 and 791 in 11 tumors55 and codon 883 in exon 15 in 4 of 111 MTCs116,117 (Eng et al., unpublished data). In pheochromocytomas, the picture is slightly different. Six of 112 (5 percent) sporadic pheochromocytomas had a mutation in the ret codon 918, 118,119 2 of 112 had proven somatic mutations of codon 634, 119,120 and a further 3 tumors have been reported with novel somatic mutations affecting the 39 splice acceptor site of exon 9, codons 632 through 633, and codon 925.118,121 None of 32 non-MEN 2–associated parathyroid lesions was found to have ret mutations.122 No ret mutations have been found in sporadic or hereditary neuroblastoma.122a
Detailed study of codon 918 mutations in sporadic MTCs by PCR analysis of microdissected portions of primary tumors and metastases has shown that the tumors are most often mosaic for the mutation; i.e., there are mutant and nonmutant clones, implying that the 918 mutation is not the initiating event in formation of sporadic MTC.123 One tumor was found to have different areas with codon 918 and codon 883 mutations. Mosaicism for codon 918 mutation also was found in two of three MEN 2A tumors studied, in which a germ-line codon 634 also was present.
Synthesis: Speculation on How Ret Mutations Result in Tumor Formation
Carlson et al.124 showed that induction of raf-1 signaling in TT cells (an MTC cell line bearing a codon 634 mutation)125 results in both differentiation and silencing of ret expression. Possibly, during development, C cells move from a “predetermined” to a terminally differentiated state in which ret expression is reduced and the potential for proliferation is lost.124 Inappropriate continued activation of ret by a cysteine mutation may override the differentiation program, allowing continued proliferation and the hyperplasias seen in MEN 2. In this scheme, the effects of mutations in the different cysteine codons may depend on the degree of activation that resulted. The combined HSCR/MEN 2 phenotype occasionally seen with mutations of cys 609, cys 618, and cys 620 may (speculatively) be a result of these mutations causing inappropriate activation of ret sufficient to result in tumor formation but insufficient to sustain the development of enteric neurons. If these mutations also impaired ligand binding, any reinforcement of ret activation by that means would be reduced. The lack of developmental abnormalities in MEN 2A and FMTC compared with MEN 2B may be the case because although the timing of ret signaling is inappropriate, the pathways of signaling are normal and the degree of activation of ret signaling is probably less. Similarly, the preponderance of MEN 2B-type somatic mutations in tumors could occur because once they are past a certain stage of differentiation and development, thyroid C cells and adrenal medullary cells are no longer susceptible to transformation by increased ret activity alone but are susceptible to the altered pathways of signaling induced by the MEN 2B mutation, or because MEN 2B mutations are associated with a higher level of activation. The lack of parathyroid involvement in MEN 2B may be the case either because the MEN 2B-specific signaling pathways are not present in parathyroid cells or because the MEN 2B mutations, unlike MEN 2A, still retain partial dependence on ligand binding for activity and the ligand may not be present in parathyroid tissue. Evidence is not currently available to distinguish any of these alternatives but may become available through studies of transgenic models in which the design of both mutant ret and ligand can be modulated.
Implications for Diagnosis
Regular screening by biochemical testing and imaging, followed by surgery where necessary, has been shown to be effective in preventing mortality and morbidity in MEN 2 families.21,126-128
In families with MEN 2A or FMTC, it is generally recommended that screening be started at about 4 to 5 years of age, continued annually until about age 20, and then possibly (this is controversial) continued at rather wider intervals until age 35. Screening should consist of measurement of plasma calcitonin after stimulation with intravenous pentagastrin, calcium, or both combined21,127,128 ; blood pressure measurement; urinary or plasma catecholamines; possibly imaging of the adrenals; and serum calcium levels. Thyroid surgery should consist of total thyroidectomy and central node dissection with conservation of normal parathyroid in situ or autotransplantation to the forearm.129 It is controversial whether after the detection of a unilateral adrenal abnormality one or both adrenals should be removed.130 The risks of serious problems from pheochromocytoma developing subsequently in the remaining adrenal must be balanced against the inconvenience and possible dangers of life after adrenalectomy and hormone replacement. Current opinion generally favors bilateral adrenalectomy. There is no consensus about screening for MEN 2B. Because of the sometimes early development and aggressive nature of the thyroid tumor, screening for MTC in a ret mutation-positive child of a known MEN 2B patient probably should start by age 1 year.131,132 Normal calcitonin levels may be high in infants, and screening results may be difficult to interpret. In this situation, some clinicians have advocated thyroidectomy on the basis of the MEN 2B phenotype alone. Greater certainty can now be provided by DNA testing for the MEN 2B mutation.
Although generally effective, biochemical screening has several problems. The stimulated calcitonin tests are somewhat unpleasant, and compliance is not always good. Occasional false-positive results have been obtained, and results on successive tests that fluctuate at and just above the normal level are common and a frequent cause of anxiety to physicians and family members. Finally, there is the problem of whether to initiate a program of screening for the families of apparently sporadic cases of one of the MEN 2 component tumors. Although many physicians are reluctant to burden a family with these tests, not following such a screening program will inevitably lead to missing some opportunities for early diagnosis.
The number of distinct ret mutations that occur in MEN 2 is small. One of these mutations can be detected in over 90 percent of patients. DNA testing is therefore relatively simple and is now in routine clinical use.53
In a family known to have MEN 2 in which the causative mutation can be identified, DNA testing of unaffected family members at risk will eliminate those who do not have the mutation from the need for biochemical screening and simplify the decision to have surgery in those whose screening results are equivocal. Increasingly, opinion is moving toward a recommendation for thyroidectomy in childhood on the basis of DNA testing alone, without waiting for biochemical testing to show abnormalities indicative of C-cell hyperplasia. This may seem surprising given that the chances of presentation with symptomatic disease in MEN 2A are only on the order of 50 percent by age 5024 and that a good biochemical screen for early disease is available. It probably reflects a mixture of concern about continued compliance with the biochemical screening and the possibility that equivocation over borderline results may lead to surgery being carried out too late; there is also the view that thyroidectomy even in young children has low morbidity and that the best thing for the child and family is to deal with the problem and put it behind them. Because of the lower probability of adrenal disease and the much greater morbidity of adrenalectomy, prophylactic removal of the adrenals is not advised except when there is evidence of abnormality.
The genotype-phenotype correlations outlined above11 may provide some indication of the probability of adrenal or parathyroid involvement, but at present the data are too few and the overlaps too great to recommend that mutation data be used to exclude families from adrenal or parathyroid screening.
DNA Testing in an Apparently Sporadic Case.
DNA testing plays an important role in determining which apparently sporadic cases have heritable disease. All apparently sporadic cases of MTC should be offered DNA testing. If the patient is dead, normal or tumor tissues from pathology specimens may be tested. The limited range of mutations makes this technically feasible in most cases, and since MEN 2A-type mutations are uncommon as somatic events, they can be interpreted as probably germ line in origin even if they are found in tumor (the same does not, of course, apply to MEN 2B mutations). If there is no family history on careful review and no evidence of C-cell hyperplasia on the thyroidectomy specimen, the probability of hereditary disease will vary according to the age at diagnosis of the index case but is almost certainly well below 10 percent.27,55 In this case, failure to find a mutation in exons 10, 11, 13, 14, and 16 of ret excludes MEN 2A with 99 percent probability11 and, if there is no abnormal phenotype, MEN 2B as well. The residual probability of FMTC or “MTC only” familial disease is a little higher, since up to 10 percent of such families have no detectable ret mutation, but it is still probably below 2 percent. It remains a matter of clinical judgment, according to the circumstances and perceptions of each family, whether to pursue biochemical screening of family members at these levels of risk. On the one hand, one does not wish to lose any possibility of early diagnosis and treatment of a potentially lethal and unpleasant cancer (but possibly the FMTC, which is most likely to be missed, could be treated satisfactorily at clinical presentation); on the other hand, one does not want to run the risk of unnecessarily “medicalizing” a family over a period of many years.
Direct mutation testing of unselected cases of apparently sporadic pheochromocytomas suggests that roughly 5 percent may carry a ret mutation and that another 5 percent will have a VHL mutation119 (see Chap. 41). Mutation screening of apparently sporadic cases of pheochromocytoma for MEN 2A and VHL mutations is probably worthwhile. The incidence of occult MEN 2 among apparently sporadic cases of parathyroid hyperplasia and adenoma appears to be low, 122 and unless there are other suggestive features, DNA screening probably is not justified. Children presenting with Hirschsprung disease should be tested for ret mutation and, if a mutation is found in exons 10 or 11, for thyroid C-cell tumor.