Structure and Function of MITF
Once the human MITF cDNA had been isolated,65 exon-intron boundaries were defined by sequencing clones from vectorette libraries and PCR primers designed to amplify each exon.66 The genomic structure is similar to mouse mi, with the coding sequence contained in nine exons (Fig. 244-7). At least in the mouse, an alternative promoter and 5′ exon are used in the heart, and there are two alternative starts for exon 6, separated by 18 nucleotides, giving isoforms with or without the six amino acids TACIFP just upstream of the basic domain.49
MITF gene structure and mutations. The gene comprises 9 exons; the basic DNA-binding domain in exons 6 and 7, and the helix-loop-helix and leucine zipper dimerization domains in exons 7 and 8 and 8 and 9 are marked. The distribution of truncating (a 1-bp deletion and several splice-site and nonsense mutations) and nontruncating (an in-frame deletion and several missense) mutations are shown. The question mark indicates that R203K, an amin-oacid substitution found in an affected person, may not be pathogenic because it does not track with WS2 in the family.
MITF encodes a member of the well-known basic helix-loop-helix leucine zipper (bHLH-ZIP) protein family. These proteins form homo- or heterodimers through their HLH and ZIP domains, and the dimers bind DNA through their basic domains. The DNA targets of bHLH-ZIP proteins have a consensus core sequence (the E-box) CANNTG. A specific target of mi, the M-box (AGTCATGTGCT), has been identified in the promoter region of several melanocyte-specific genes such as tyrosinase and tyrosinase-related protein 1.122 Cotransfection experiments show that mi protein activates transcription from promoters containing an M-box.121 It has been suggested that mi may be a master gene for melanocyte differentiation, and in support of this, mouse 3T3 fibroblast cells take on a melanocyte-like appearance when transfected with mi and start to express tyrosinase and TRP1.123 As discussed below, it has been shown recently that expression of MITF is controlled by PAX3.124
Mutation Analysis of MITF.
Relatively few families have so far proved to carry MITF mutations. The mutations found to date, all in heterozygous form, include missense, nonsense, and splice-site point mutations (Table 244-8). The missense mutations mostly affect residues in the known functional domains, such as N278D, which places a polar amino acid in the mating face of the leucine zipper. An interesting family was reported by Morell et al.126 An MITF mutation, 944del(1), caused WS2 in a family. A sequence variant in the tyrosinase gene, R402Q, also was segregating in the family. This variant is a common polymorphism that encodes an enzyme with reduced catalytic activity. When present in isolation, it has no phenotypic effect, but in compound heterozygotes with a mutant TYR allele, it produces an autosomal recessive ocular albinism. Individuals in the family who had just the tyrosinase variant were normal, as expected, but when they had both the MITF mutation and the tyrosinase variant, they had WS2 plus ocular albinism.
Table 244-8: MITF Mutations |Favorite Table|Download (.pdf) Table 244-8: MITF Mutations
|Mutation ||Exon and Position ||Phenotype ||Reference |
|Truncating mutations || |
|153+1 G→A ||E1 splice donor ||WS2 (3 families) ||66, 33, unpub. |
|562–2 A→C ||E5 splice acceptor ||WS2 (2 families) ||66, unpub. |
|R214X ||E7 Basic ||WS2 ||125 |
|R259X ||E8 HLH ||WS2 ||127 |
|944del(A) ||E8 Zip ||WS2 ||126 |
| Missense and other nontruncating mutations || |
|R203K ||E6 Basic ||WS2 or nonpathogenic ||33 |
|N207K ||E6 Basic ||Tietz ||128 |
|Del(R214) ||E7 Basic ||Tietz ||127 |
|S250P ||E8 (HLH helix 2) ||WS2 ||33 |
|Y253C ||E8 ||WS2 ||Unpub. |
|N278D ||E8 Zip ||WS2 ||33 |
|S298P ||E9 ||WS2 ||33 |
Mi is one of the few loci where at the moment the molecular pathology is richer in the mouse than in humans. Many mutant alleles are known in the mouse, some of which are recessive and others semidominant, and some compound heterozygotes show allelic interactions (see Table 244-5). In humans, all the known alleles are manifest in heterozygotes—probably in the mouse they would be classed as semidominant because no doubt homozygotes, if we ever saw them, would be much more dramatically affected. Heterozygous mice, when affected at all, usually show spotting of the coat to varying degrees or occasionally uniform dilution of pigmentation (in Mi b/+ mice, for example). No hearing loss is reported for heterozygous mice, although homozygotes for several alleles including mi, Miwh , and mice are affected.46 As with Splotch mice, humans are evidently more sensitive to hearing loss, but otherwise, the phenotypic range in humans is rather similar to that in heterozygous mice, with variable pigmentary disturbances. The range of phenotypes in homozygous mice is quite wide, including microophthalmia, osteopetrosis, and defective mast cell function. We have tested individuals with various candidate severe recessive syndromes, such as the Yemenite deaf-blind-hypopigmentation syndrome or black locks-albinism-deafness syndrome,12 but have so far failed to discover any MITF homozygotes. A deaf and depigmented patient described by Hultén et al.129 is certainly homozygous for some pigmentary defect, but more probably KIT than MITF.
Because mi protein binds its DNA target as a dimer, mutant forms that can dimerize but then prevent the dimer from binding DNA can show dominant-negative effects. Thus it is predicted that null alleles and mutants affecting the HLH-ZIP dimerization domains would be either recessive, or dominant if there is haploinsufficiency, whereas mutants affecting the DNA-binding basic domain have the potential to be dominant via a dominant-negative effect. These predictions have been tested in gel mobility assays in vitro.122,130 The products of four basic-domain mutants mi, Miwh , Mior , and miew not only failed to bind their DNA target but also interfered with binding of the wild-type protein. In vivo, however, only two of the four alleles show a significant phenotype in heterozygotes. The likely cause is failure of the mi and miew products to localize to the cell nucleus130 ; probably the basic domain also contains the nuclear localization signal. At least one HLH-domain mutant, Mib , also shows defective DNA binding,50 reminding us that dividing a linear diagram of the amino acid sequence into a series of neatly defined domains does not necessarily reflect the three-dimensional reality.
The activity of mi protein is regulated by phosphorylation of serines, at least in part through the Erk-2 protein kinase, which in turn is regulated by the KIT oncogene.131 This may be the mechanism through which KIT mutations produce piebaldism. One WS2 mutation, S298P, replaces a serine that is a target for phosphorylation (see ref143 & M. Tachibana personal communication 1997), in this case almost certainly by a mechanism independent of the KIT/Erk-2 system. Other complications in the molecular pathology of mi mutants include the possible role of heterodimers (mi protein forms heterodimers with the closely related bHLH-ZIP proteins TFEB, TFEC, and TFE3122) and possible differential functions of the isoforms produced by use of alternative splice acceptors at the start of exon 6. The only molecular effect of the misp mutant is to prevent use of one of these sites,49 but the mutation has a phenotype (see Table 244-5).
In humans, mutations in MITF can produce two different phenotypes in heterozygotes. The majority of mutations cause the relatively mild and variable WS2 syndrome. These mutations include splice-site mutants that are presumably null alleles and HLH or zipper mutants that are likely to be defective in dimerization, as well as the phosphorylation-defective S298P mutation. In two families, however, MITF mutations produced a much more severe and consistent phenotype. We found the del(R214) mutation in a mother and son.127 The mother was congenitally deaf, and she had blue eyes, red hair, which went gray at age 16, and generalized skin hypopigmentation with numerous orange freckles. Her baby son was deaf, had auburn hair, blue eyes (at 18 months), and generalized hypopigmentation. Deafness with uniform pigmentary dilution rather than patchy depigmentation is labeled Tietz-Smith syndrome (MIM 103500). This syndrome was described originally by Tietz132 in a six-generation family where 14 people showed profound sensorineural hearing loss and uniform pigmentary dilution (partial albinism), except for the irides, which were blue with no nystagmus or photophobia. Recently, this family was shown to carry the basic domain mutation N207K in MITF.128 It was reported that there was also a “CG to GC transversion adjacent to the splice junction” in exon 5.
Clearly, the MITF mutations in Tietz families are having a different, more severe, and more consistent effect in heterozygotes than the MITF mutations seen in WS2 families, and it seems likely that they do this through a dominant-negative effect. Del(R214) is identical to the original mouse mi mutation. As mentioned earlier, the mouse mi protein shows a dominant-negative effect in vitro, but the effect was limited in the cells studied because the mutant protein localized poorly (but partially) to the nucleus. We have tested several other families with a Tietz phenotype and not found other MITF mutations.
Comparing those WS2 families with known MITF mutations with those where no mutation was found, no clear clinical difference emerges. All the MITF families have classic and clearly inherited WS2, apart from those with Tietz-Smith syndrome; we have not found MITF mutations in any of the many “WS2 penumbra” cases tested. The incidence of hearing loss is high in MITF families but not dramatically higher than in other WS2 families.
EDNRB, EDN3, and SOX10 in WS4
As described earlier, WS4, or Shah-Waardenburg syndrome, was a clinical label given to a probably highly heterogeneous group of patients who had Hirschsprung disease with deafness and/or pigmentary disturbances. Research on the mouse mutants, piebald-lethal, lethal spotted, and Dominant megacolon led to identification of mutations in the EDNRB, EDN3, and SOX10 genes in a small number of patients (Table 244-9). (See Chap. 251 for other details of Hirschsprung disease.)
Table 244-9: Mutations in the EDN3, EDNRB, and SOX10 Genes Associated with Type 4 Waardenburg Syndrome |Favorite Table|Download (.pdf) Table 244-9: Mutations in the EDN3, EDNRB, and SOX10 Genes Associated with Type 4 Waardenburg Syndrome
|Mutation ||Exon and Position ||Phenotype ||Reference |
| EDN3 mutations || || || |
|C159F (homozygous) ||E3 (downstream of ET-3 sequence) ||Total colonic aganglionosis, bilateral profound hearing loss, hypoplastic blue eyes ||135 |
|262GC → T (homozygous) ||E2 (frameshift upstream of ET-3 sequence) ||Total colonic aganglionosis, bilateral hearing loss, pale blue eyes, white skin patches ||136 |
| EDNRB mutations || || || |
|W276C (homozygous) ||E4 ||Variable: HSCR and/or WS2 (see text) ||72 |
|A183G (homozygous) ||E2 (3rd transmembrane domain) ||HSCR, deaf, white forelock, heterochromia ||134 |
| SOX10 mutations || || || |
|E189X (heterozygous) ||E4 (3′ of HMG domain) ||Short-segment HSCR, bilateral profound hearing loss, white hair, blue eyes, depigmented skin patches ||138 |
|Y83X (heterozygous) ||E3 (5′ of HMG domain) ||HSCR, profound bilateral hearing loss, fair hair, blue eyes. ||138 |
|482ins(6) (heterozygous) ||E4 (in-frame insertion within HMG box) ||Short segment HSCR, deaf ||138 |
|1076del(GA) (heterozygous) ||E5 (frameshifting deletion 3′ of HMG box) ||Variable; see text ||138 |
Endothelins are a family of three peptide hormones (endothelin-1, -2, and -3) that act through two transmembrane receptors (A and B), mainly in the control of vascular tone. The system is of considerable interest as a drug target, and mouse knockouts of all five genes have been produced. Unexpectedly, the ednrb and edn3 knockouts identified the piebald-lethal (sl ) and lethal spotted (ls) loci, respectively. The phenotype of both these mutants was megacolon with spotting of the coat. The precise role of the endothelin system in neural crest function has yet to be defined.
EDNRB maps to 13q22 in humans, a location already known through chromosomal deletions to affect neural crest development and to which a WS4-like phenotype had been mapped in one family (see above). Puffenberger et al.72 identified an EDNRB mutation, W276C, in affected family members. Many other patients with isolated Hirschsprung disease and some with WS2 or WS4 have been tested for EDNRB mutations, and a few have proved positive.
EDN3 maps to 20q13. The locus encodes a large inactive precursor, preproendothelin 3, from which the active 21-amino-acid peptide (ET3) is released by proteolytic cleavage. Although there was no prior linkage evidence in humans for a neurocristopathy locus at this location, there is a case report133 of a child with a balanced de novo 7;20 translocation with features of WS4. Again, testing a range of patients revealed heterozygous EDN3 mutations in a few patients with isolated Hirschsprung disease. One family reported by Hofstra et al.135 is particularly interesting (Fig. 244-8). The two affected children were homozygous for the C159F mutation in EDN3 and had classic WS4 with total colonic aganglionosis, profound bilateral hearing loss, hypoplastic blue eyes (in Pakistani children), and pigmentary disturbances. The parents, who were first cousins, were phenotypically normal but both heterozygous for C159F. Five first-degree relatives of the parents had hearing loss (2), pigmentary disorders (2), and pigmentary disorder plus hypoplastic blue eyes (1). That part of the pedigree looks remarkably like the “WS2 penumbra” families discussed earlier. Nevertheless, we have tested many such families for mutations in EDNRB and EDN3 with uniformly negative results.
A Pakistani family with two children with WS-Hirschsprung disease who are homozygous for the EDN3 mutation C159F. Note the five relatives with hearing loss or pigmentary abnormalities of the hair and/or eyes, thus resembling the “WS2 penumbra” families described in the text. (Used with permission from Hofstra et al.135 ) HSCR, Hirschsprung disease, SNHL, sensorineural hearing loss.
The Sox10 gene underlying the Dominant megacolon (Dom) mouse mutant was isolated recently by positional cloning by two groups.55,56 It encodes a member of the SOX family of transcription factors, characterized by the presence of a highly conserved HMG domain that is believed to bind double-stranded DNA and bend it, creating sites for other DNA-binding proteins to act. A human SOX10 cDNA was isolated by screening a brain cDNA library with a rat sox10 cDNA probe, and the genomic structure was established mainly by database searching. Primers were designed to amplify each exon, and genomic DNA samples from a range of patients were analyzed. No mutations were found in 34 patients with uncomplicated Hirschsprung disease, but mutations were found in four families with a WS4 phenotype.138 Three of the mutations are truncating, and the fourth inserts two amino acids within the HMG domain.
Genotype-phenotype relationships are not at all straightforward with EDNRB or EDN3 mutations, but a WS4 phenotype is most likely to be seen in homozygotes, and in this respect, they fit the original definition of Shah-Waardenburg syndrome as recessive. In the Mennonite family studied by Puffenberger et al.,72 rather than simply determining a WS4 phenotype, the EDNRB mutation W276C behaved as one component of a polygenic determinant. Hirschsprung disease was seen in 17 of 23 W276C homozygotes and 17 of 82 heterozygotes but also 5 of 45 people homozygous for the normal allele. Some people homozygous for the mutation were deaf or had a white forelock without Hirschsprung disease, some had Hirschsprung disease without any WS features, and some had varying combinations of both.
SOX10 mutations are dominant, although since many will be de novo and no doubt cases of parental mosaicism eventually will be found, these families can appear recessive. Expression is variable; in the only familial case reported,138 a woman and her two children had the frameshifting mutation 1076del(GA). Both children had WS4 with Hirschsprung disease, but the mother showed only deafness and a white forelock. Unlike EDNRB and EDN3, SOX10 mutations do not appear, on the present limited evidence, to be found in patients with uncomplicated Hirschsprung disease.
It is unlikely that mutations in these three genes are the only cause of WS4. Although rare, WS4 is probably highly heterogeneous. Phenotypes falling within the general definition of WS4 may result from any number of rather generalized disturbances of early neural crest function.