Interallelic Complementation at the Mouse Mitf Locus

Corresponding author: Eiríkur Steingrímsson, Faculty of Medicine, University of Iceland, Vatnsmýrarvegur 16, 101 Reykjavík, Iceland., eirikurs{at}hi.is (E-mail)

Communicating editor: C. KOZAK

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mutations at the mouse microphthalmia locus (Mitf) affect the development of different cell types, including melanocytes, retinal pigment epithelial cells of the eye, and osteoclasts. The MITF protein is a member of the MYC supergene family of basic-helix-loop-helix-leucine-zipper (bHLHZip) transcription factors and is known to regulate the expression of cell-specific target genes by binding DNA as homodimer or as heterodimer with related proteins. The many mutations isolated at the locus have different effects on the phenotype and can be arranged in an allelic series in which the phenotypes range from near normal to white microphthalmic animals with osteopetrosis. Previous investigations have shown that certain combinations of Mitf alleles complement each other, resulting in a phenotype more normal than that of each homozygote alone. Here we analyze this interallelic complementation in detail and show that it is limited to one particular allele, Mitf^Mi-white (Mitf^Mi-wh), a mutation affecting the DNA-binding domain. Both loss- and gain-of-function mutations are complemented, as are other Mitf mutations affecting the DNA-binding domain. Furthermore, this behavior is not restricted to particular cell types: Both eye development and coat color phenotypes are complemented. Our analysis suggests that Mitf^Mi-wh-associated interallelic complementation is due to the unique biochemical nature of this mutation.

MUTATIONS at the mouse microphthalmia locus (Mitf) affect the development of several different cell types. Common to all the mutations are defects in the neural-crest-derived melanocytes, resulting in reduction or lack of pigmentation in the coat, eye, and inner ear. Most of the mutations also affect pigmented epithelial cells of the eye and mast cells, while only a few mutations result in osteoclast defects. Approximately one-half of the Mitf alleles are semidominantly inherited and show white spotting and/or coat color dilution when heterozygous with wild type. The remaining alleles are recessive and exhibit a phenotype only in homozygous condition. The Mitf^mi-spotted (Mitf^mi-sp) mutation is unusual in that it displays a visible phenotype only in combinations with other mutations at the locus (reviewed by MOORE 1995 ).

In 1953, Grüneberg noted the paradoxical relations of the semidominant Mitf^{microphthalmia} (Mitf^mi) and Mitf^Mi-wh mutations (GRUNEBERG 1953 ). While the Mitf^mi mutation exhibits a weak phenotype in the heterozygous condition (occasional head blaze and belly streaks) and a very strong phenotype in the homozygous condition (microphthalmia, white coat, and osteopetrosis), the Mitf^Mi-wh mutation shows a severe heterozygous phenotype (large belly spots and coat color dilution) and an intermediate homozygous phenotype (white coat and partial microphthalmia). Thus, the relative effects of these two mutations are different in the hetero- and homozygous conditions. Although many more Mitf mutations have been isolated since, Grüneberg's observations still hold true; the semidominant phenotype of the Mitf^Mi-wh mutation is the most severe phenotype associated with any mutation at the locus while its homozygous phenotype is only intermediate. KONYUKHOV and OSIPOV 1968 analyzed the relationship between these two alleles further and showed that heteroallelic animals show interallelic complementation with respect to effects on the eye: Mitf^Mi-wh/Mitf^mi animals have normal eye size while homozygous littermates exhibit severe (Mitf^mi/Mitf^mi) or intermediate (Mitf^Mi-wh/Mitf^Mi-wh) microphthalmia. Similar analysis with a few other Mitf alleles also showed interallelic complementation. For example, SCHAIBLE 1963 showed that a combination of the Mitf^Mi-wh allele with the Mitf^{mi-black-eyed white} (Mitf^mi-bw) mutation resulted in white animals with pale yellow spots (on the C3HxB6 background they have pigmented spots on the back). Homozygotes for both mutations are completely white. Similarly, heteroallelic combinations of Mitf^Mi-wh with Mitf^{mi-white spot} (Mitf^mi-ws) produced animals with dark eyes of normal size and a spotted, checkerboard-like coat with yellowish-brown to gray colors (HOLLANDER 1964 ). Homozygotes for the Mitf^mi-ws mutation are white with pink eyes. This suggests that interallelic complementation is a common phenomenon at the Mitf locus and that both eye and coat color defects are affected. Although this phenomenon has never been characterized systematically in detail, these observations suggest that interallelic complementation reveals an important aspect of the nature of the Mitf locus.

The MITF protein is a member of the MYC supergene family of basic-helix-loop-helix-leucine zipper (bHLH-Zip) transcription factors and is most closely related to the TFE3, TFEB, and TFEC proteins (HODGKINSON et al. 1993 ; HUGHES et al. 1993 ). Like other members of the bHLH-Zip family, MITF has been shown to bind the CANNTG E-box sequence in vitro as either a homodimer or a heterodimer with TFE3, TFEC, and TFEB (HEMESATH et al. 1994 ). The basic domain is the DNA-binding domain of the protein while the HLH and Zip domains are responsible for dimerization. Consistent with its role as a regulator of gene expression, MITF is primarily located in the nucleus (TAKEBAYASHI et al. 1996 ) where it can activate expression from pigment cell, mast cell, and osteoclast specific promoters (BENTLEY et al. 1994 ; YASUMOTO et al. 1994 ; MORII et al. 1996 ; YASUMOTO and SHIBAHARA 1997 ; MOTYCKOVA et al. 2001 ).

The molecular and biochemical defects associated with most of the Mitf alleles have been determined (HODGKINSON et al. 1993 ; HUGHES et al. 1993 ; HEMESATH et al. 1994 ; STEINGRIMSSON et al. 1994 , STEINGRIMSSON et al. 1996 ; YAJIMA et al. 1999 ; HALLSSON et al. 2000 ). The semidominant mutations characterized to date affect either the DNA-binding or the transcriptional activation domains of the protein while dimerization domains are unaffected (HODGKINSON et al. 1993 ; STEINGRIMSSON et al. 1994 , STEINGRIMSSON et al. 1996 ). The mutant proteins cannot bind DNA; however, they can still dimerize with proteins such as TFE3 and thereby interfere with DNA binding of the partner (HEMESATH et al. 1994 ; STEINGRIMSSON et al. 1996 ). The dominant-negative behavior of these mutant proteins in vitro thus accounts for the phenotype seen in heterozygous mice. Consistent with this, the recessive mutations affect either the dimerization domain of the MITF protein or the transcription of the Mitf gene, resulting in little or no MITF production (HODGKINSON et al. 1993 ; HUGHES et al. 1993 ; STEINGRIMSSON et al. 1994 ; YAJIMA et al. 1999 ).

Despite the detailed molecular and biochemical analysis of the Mitf mutations, no satisfactory explanation for the interallelic complementation has emerged. Here, we perform a detailed genetic analysis of the interallelic complementation at the Mitf locus and show that it is restricted to the Mitf^Mi-wh mutation, a mutation with unique characteristics. Our analysis suggests that the nature of the Mitf-associated interallelic complementation is due to neomorphic action of the Mitf^Mi-wh mutation.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The following Mitf mutants were used in this study: C57BL/6J-Mitf^Mi-wh, C57BL/6J-Mitf^mi-sp, C57BL/6J-Mitf^{mi-red-eyed white} (Mitf^mi-rw), C57BL/6J-Mitf^{mi-eyeless white} (Mitf^mi-ew), C57BL/10-Mitf^{mi-black and white spot} (Mitf^mi-bws), C57BL/6J-Mitf^mi, 82UT-Mitf^{Mi-oak ridge} (Mitf^Mi-or), 82UT-Mitf^mi-brownish (Mitf^mi-b), and mixed [C3H/C57BL/6J]-Mitf^mi-vga-9 (Table 1). These strains are maintained at the National Cancer Institute in Frederick, Maryland, and at the National Institute of Neurological Disorders and Stroke, National Institutes of Health, in Bethesda, Maryland. The mice were mated systematically to generate the different allelic combinations. At least three different independent crosses were set up for generating each combination and multiple offspring (>25) were analyzed from each cross. The phenotypes of the resulting animals were visually inspected, the animals were photographed at 6 weeks of age (or earlier), and eyes, skin, and Harderian gland were dissected for histologic analysis. All tissue specimens were fixed in Bouins fixative, sectioned, and then stained with hemotoxylin and eosin. For the evaluation of melanin in Harderian glands and in the retinal pigment epithelium (RPE) of the eye, tissues were stained with Fontana-Masson.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Heteroallelic combinations of Mitf mutations:
To perform a systematic study of the Mitf-associated interallelic complementation, available Mitf alleles were crossed to each other in all possible combinations. The phenotypes of the resulting heteroallelic offspring were studied by visual inspection of coat color and eye size. The alleles used in this study are described in Table 1 and range in phenotype from very mild (e.g., Mitf^mi-sp) to severe (Mitf^mi) and in mode of inheritance from recessive (Mitf^mi-sp, Mitf^mi-rw, Mitf^mi-ew, Mitf^mi-vga9) to semidominant (Mitf^Mi-wh, Mitf^Mi-or, Mitf^Mi-b, Mitf^mi). In all the crosses made, the number of heteroallelic progeny was according to Mendelian ratios (data not shown) and the phenotypes were consistent among the different progeny and litters of each cross.

The phenotypes of the resulting combinations are described in Table 2 and in Fig 1 Fig 2 Fig 3. In most cases where the homozygous phenotypes of the two alleles are similar to each other, the phenotypes of the heteroallelic combinations are the same as each of the homozygotes. For example, Mitf^Mi-or/Mitf^mi animals are white and microphthalmic and have severe osteopetrosis just like the respective homozygotes (Mitf^Mi-or/Mitf^Mi-or and Mitf^mi/Mitf^mi; Table 2). Similarly, Mitf^mi-ew/Mitf^mi-vga9 animals have a white coat and severe microphthalmia but no osteopetrosis, just like the Mitf^mi-ew/Mitf^mi-ew and Mitf^mi-vga9/ Mitf^mi-vga9 homozygotes (Table 2). In most cases in which the homozygous phenotypes are different, however, the heteroallelic animals exhibit an intermediate phenotype. For example, Mitf^mi-ew/Mitf^mi-sp animals have white feet, head, and belly while the rest of the coat is gray (Table 2 and Fig 1A). This phenotype is intermediate between the white microphthalmic Mitf^mi-ew/Mitf^mi-ew (Table 2) and near normal Mitf^mi-sp/Mitf^mi-sp (Table 2, Fig 1B) animals. Similarly, Mitf^Mi-b/Mitf^mi-rw animals are white with large pigmented spots (Table 2 and Fig 1C), a phenotype intermediate between the white Mitf^Mi-b homozygotes (Table 2) and Mitf^mi-rw/Mitf^mi-rw animals, which are white with a pigmented spot on the head (Table 2, Fig 1D). In all these cases, one of the alleles encodes a partially functional protein. Interestingly, the combination of Mitf^mi-ew and Mitf^mi results in white, severely microphthalmic animals (Fig 1E); this phenotype is identical to that of Mitf^mi-ew homozygotes while Mitf^mi homozygous animals also exhibit osteopetrosis.

View larger version (44K): In this window In a new window Download PPT slide   Figure 1. Phenotypes resulting from Mitf mutations and heteroallelic combinations. (A) Mitfmi-sp/Mitfmi-ew. (B) Mitfmi-sp/Mitfmi-sp: Note the apparently normal appearance of this animal. (C) MitfMi-b/Mitfmi-rw: The pigmented spots are agouti due to the C3H background of the MitfMi-b mutation. (D) Mitfmi-rw/Mitfmi-rw: Note the black head spots. (E) Mitfmi-ew/Mitfmi: This animal lacks pigment and shows severe microphthalmia but no osteopetrosis. (F) MitfMi-wh/Mitfmi-sp. (G) Mitfmi-rw/Mitfmi: Note the severe microphthalmia. (H) MitfMi-wh/Mitfmi-rw: Note the faintly pigmented patch on their heads (dotted line).

View larger version (60K): In this window In a new window Download PPT slide   Figure 2. Phenotypes associated with combinations with the Mitfmi-vga9 loss-of-function mutation. (A) Mitfmi-vga9/Mitfmi-vga9: Note the severe microphthalmia. (B) Mitfmi-ew/Mitfmi-vga9. (C) MitfMi-wh/Mitfmi-vga9: Note the normal eye size in the compound heterozygote (arrow). (D) Mitfmi-rw/Mitfmi-vga9. (E) Mitfmi-sp/Mitfmi-vga9. (F) Mitfmi-bws/Mitfmi-vga9.

View larger version (49K): In this window In a new window Download PPT slide   Figure 3. Interallelic complementation at the Mitf locus. (A) MitfMi-wh/+ (pigmented, in the back) and MitfMi-wh/MitfMi-wh (white and microphthalmic, in the front) animals: Note the intermediate microphthalmia. (B) Mitfmi/Mitfmi: Note the severe microphthalmia and small size of the 3-week-old animals. (C) MitfMi-wh/Mitfmi. Eyes are of normal size (arrows) although eye pigment is somewhat reduced compared to normal. (D) MitfMi-wh/Mitfmi-ew. The eyes are of normal size (arrows). (E) MitfMi-or/MitfMi-or (front) and MitfMi-wh/MitfMi-or (back) animals. Note the normal eye size in the compound heterozygote (arrow). (F) MitfMi-wh/ MitfMi-b and MitfMi-b/MitfMi-b animals. The two compound heterozygotes (leftmost animal and the animal on top) show light-brown coat color while the two MitfMi-b homozygotes (bottom) are completely white. Neither genotype affects eye size. (G) Mitfmi-bws/Mitfmi-bws: Note the black spots on otherwise white background. Eye size is normal. (H) MitfMi-wh/Mitfmi-bws. The phenotype is intermediate between each homozygote (compare to A and G) and no complementation is observed.

All combinations involving the Mitf^mi-sp mutation fall into the intermediate group; the resulting phenotypes are in direct relation to the severity of the allele to which Mitf^mi-sp is crossed (Table 2). This mutation was originally found in a colony of Mitf^Mi-wh animals (WOLFE and COLEMAN 1964 ). While animals homozygous for the Mitf^mi-sp mutation cannot be distinguished from their wild-type littermates, Mitf^Mi-wh/Mitf^mi-sp heterozygotes are tan with white spots and normal-size eyes (WOLFE and COLEMAN 1964 ; Table 2, Fig 1F). Although Mitf^mi-sp homozygotes have a normal appearance, a reduction in the activity of the tyrosinase enzyme has been detected in their skin (WOLFE and COLEMAN 1964 ). All heteroallelic combinations involving Mitf^mi-sp show a more severe defect in pigmentation than these mutations show in heterozygous combinations with wild type, albeit less than what is observed in homozygotes for the other allele (Table 2). For example, Mitf^mi-rw/Mitf^mi-sp animals have a large belly spot (Table 2), a phenotype not seen in either Mitf^mi-rw/+ or Mitf^mi-sp/Mitf^mi-sp animals. Also, Mitf^Mi-b/Mitf^mi-sp animals exhibit a diluted coat color phenotype while Mitf^Mi-or/Mitf^mi-sp animals are white with spots that gradually lose their pigmentation with age (Table 2). This gradual depigmentation is similar to that observed for the Mitf^mi-vitilago (Mitf^mi-vit) mutation (LERNER et al. 1986 ; LAMOREUX et al. 1992 ).

It is interesting to compare the phenotypes of Mitf^mi-sp homo- and heterozygotes with the phenotypes of heteroallelic combinations involving the loss-of-function Mitf^mi--vga9 and the recessive Mitf^mi-ew mutations. As already explained, no phenotype is visible in Mitf^mi-sp/Mitf^mi-sp animals, suggesting that the mutant protein has some activity. However, one dose of the Mitf^mi-sp mutant protein is clearly not sufficient for proper melanocyte development since heteroallelic combinations involving Mitf^mi-vga9 and Mitf^mi-ew result in defective pigmentation (Table 2, Fig 1 and Fig 2). Interestingly, although the homozygous phenotypes of the Mitf^mi-vga9 and Mitf^mi-ew mutations are almost identical, the heteroallelic combinations differ in that Mitf^mi-vga9/Mitf^mi-sp animals have large pigmented spots while Mitf^mi-ew/Mitf^mi-sp animals have a uniformly gray coat color over most of their body; head and feet are white. This reflects the fact that the Mitf^mi-vga9 mutation is a loss-of-function mutation; the heteroallelic combination reveals the function of a single copy of the Mitf^mi-sp protein. The Mitf^mi-ew/Mitf^mi-sp combination, on the other hand, reveals the partial dominant-negative nature of the Mitf^mi-ew protein. Although the Mitf^mi-ew mutation behaves in a recessive fashion on its own and is classified as such in this study, in vitro studies suggest that the mutant protein has strong dominant-negative activity (HEMESATH et al. 1994 ). However, the Mitf^mi-ew protein is unable to enter the nucleus efficiently (TAKEBAYASHI et al. 1996 ; NAKAYAMA et al. 1998 ), resulting in a dominant-negative cytoplasmic protein. Since the protein needs to enter the nucleus to realize its effects, this may result in only minor dominant-negative effects in vivo, at least in some of the cell types affected. This idea has been further supported by genetic studies (STEINGRIMSSON et al. 2002 ; see below).

The Mitf^mi-rw mutation also has unique features. Homozygotes for this mutation are white with pigmented spots of somewhat variable size on the head (Fig 1D); occasional pigmented spots are observed in other body regions, including the rump. Although the eyes are smaller than normal, appear red, and lack pigment altogether, they are not as severely affected as animals carrying the loss-of-function Mitf^mi-vga9 mutation or in Mitf^Mi-wh animals (Table 2; Fig 3A). The Mitf^mi-rw mutation is caused by the lack of a large portion of the Mitf regulatory region, resulting in the absence of the 5' exons 1h and 1b and in aberrant expression of the gene (STEINGRIMSSON et al. 1994 ; HALLSSON et al. 2000 ). Most heteroallelic combinations involving Mitf^mi-rw result in white microphthalmic animals: Combinations with the Mitf^mi, Mitf^Mi-or, and Mitf^mi-ew mutations all produce white microphthalmic animals (Table 2; Fig 1G) and Mitf^mi-rw/Mitf^mi-vga9 animals are white and microphthalmic and most have a small pigmented head spot (Table 2; Fig 2). In all these cases the microphthalmia observed is more severe than that in Mitf^mi-rw homozygotes. However, combinations of the Mitf^mi-rw mutation with Mitf^Mi-b and Mitf^Mi-wh result in somewhat milder phenotypes: Mitf^mi-rw/Mitf^Mi-b animals are white with an occasional pigmented spot and normal eye size (Table 2) while Mitf^mi-rw/Mitf^Mi-wh animals are white with a tan head spot; eye size is near normal and lacks pigment (Fig 1H).

In all the cases described above, coat color is most sensitive to mutations at Mitf, indicating that melanocytes have the highest requirement for Mitf function. Eye color follows, then eye size, and only a few of the mutations show osteopetrosis. Thus, the pigment cells of the retinal epithelium of the eye have a lower requirement for Mitf protein than do melanocytes, and osteoclasts seem to be able to function with very low amounts (if any) of Mitf in the cell.

Interallelic complementation at Mitf:
In contrast to the results described above, most of the combinations involving the Mitf^Mi-wh allele show interallelic complementation in which the resulting phenotype is more normal than that of each of the homozygotes alone. This is true for combinations involving the semidominant mutations Mitf^mi, Mitf^Mi-or, and Mitf^Mi-b, as well as the recessive Mitf^mi-ew mutation and the Mitf^mi-vga9 loss-of-function mutation. All these combinations, except Mitf^Mi-wh/Mitf^Mi-b, result in white animals with normal eye size (Table 2; Fig 3, A–E). The normal eye size is in sharp contrast to the severe microphthalmia observed in each of the Mitf^Mi-wh, Mitf^mi, Mitf^Mi-or, Mitf^mi-ew, and Mitf^mi-vga9 homozygotes (Table 2; Fig 2 and Fig 3). The heteroallelic combination of Mitf^Mi-wh and Mitf^Mi-b results in light-tan animals with occasional white spots (Fig 3F); while eye size is normal, the RPE layer contains some pigment (data not shown). Although the eye size of this heteroallelic combination is more normal than that of Mitf^Mi-wh homozygotes, Mitf^Mi-b/Mitf^Mi-b animals have normal eye size. Thus, the eye size of the heteroallelic combination is similar to that of one of the homozygotes so no complementation is observed with respect to this feature. However, coat color pigmentation is more normal in the heteroallelic combination than in each of the homozygotes, suggesting complementation with respect to this phenotype. Clearly, the Mitf^Mi-wh mutation can complement both the microphthalmia and the coat color phenotypes of the different Mitf mutations.

Interestingly, three recessive mutations are not complemented by Mitf^Mi-wh (Table 2). As explained above, the heteroallelic combination Mitf^Mi-wh/Mitf^mi-sp results in tan animals with occasional white spot. This phenotype is intermediate between the two homozygotes and therefore no complementation is observed (Table 2). Mitf^Mi-wh/Mitf^mi-rw animals also have an intermediate phenotype although some complementation may be observed with respect to eye development. In these animals coat color is white with tan head spots while eye size varies from microphthalmic to near normal (Fig 1H); the difference in eye size can be bilateral with one eye near normal and the other microphthalmic. Perhaps most interestingly, the Mitf^Mi-wh/Mitf^mi-bws compound heterozygotes have an intermediate phenotype showing tan spots and normal eye size (Fig 3H). In the homozygous condition the Mitf^mi-bws mutation results in large, heavily pigmented spots (Table 1, Fig 3G) and normal eye size. Thus, the phenotype of the heteroallelic combination is not complemented. While two of the recessive mutations that fail to be complemented by Mitf^Mi-wh (Mitf^mi-bws and Mitf^mi-sp) have rather mild phenotypes in homozygous condition, the third (Mitf^mi-rw) is more severely affected; all three have milder phenotypes than that of Mitf^Mi-wh in homozygous condition. Both semidominant and recessive mutations are complemented while only recessive mutations fail to be complemented (Table 1 and Table 2).

This is further supported by previously reported heteroallelic combinations involving Mitf^Mi-wh. For example, Mitf^mi-bw/Mitf^Mi-wh are white with some pale yellow spots, which depigment with age, whereas homozygotes for the recessive Mitf^mi-bw mutation are white with black eyes (SCHAIBLE 1963 ). The Mitf^mi-ws mutation is semidominant; heterozygotes have a white belly spot while homozygotes are white, most with normal eye size. Heteroallelic combinations of this mutation with Mitf^Mi-wh result in spotted animals with yellowish-brown-to-gray spots that depigment with age (HOLLANDER 1968 ). Thus, coat color is complemented. The Mitf^mi-di and Mitf^mi-ce mutations have very similar phenotypes in homozygous condition (white coat and small eyes) and, in fact, the molecular defect involved is identical (HALLSSON et al. 2000 ). Consistent with this, heterozygous combinations of Mitf^mi-di and Mitf^mi-ce with Mitf^Mi-wh result in normal eye size, showing that both mutations are complemented (WEST et al. 1985 ; M. L. LAMOREUX, unpublished observations).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The allelic series at the Mitf locus shows that the three major cell types affected by mutations at the locus, melanocytes, osteoclasts, and retinal pigment epithelial cells of the eye, have different requirements for the Mitf protein: Most of the Mitf mutations affect pigmentation of the coat, many affect eye development, and only a few result in osteopetrosis. Clearly, the requirement for Mitf function is very different in the cell types in which the highest requirement is observed in melanocytes and the lowest in osteoclasts. Osteopetrosis is observed only in homozygotes for the Mitf^mi and Mitf^Mi-or mutations as well as in compound heterozygotes of these mutations (Table 2). No osteopetrosis is observed in Mitf^Mi-wh homozygotes or in any of the other mutations tested, except Mitf^mi-ew, which shows mild hyperosteosis in the homozygous condition (STEINGRIMSSON et al. 2002 ). Although the Mitf^mi-ew mutation is classified as recessive in this study, the mutant protein has partial dominant-negative activity, at least in osteoclasts. Recent studies have shown that although osteoclasts are normal in Mitf null mice as well as in mice carrying a loss-of-function mutation in the Mitf-related gene Tfe3, the combined loss of both genes results in severe osteopetrosis (STEINGRIMSSON et al. 2002 ). Thus, the Mitf-associated osteopetrosis is seen only upon simultaneous loss of Mitf and Tfe3 activity in osteoclasts. Only dominant-negative Mitf alleles that are able to interfere with Tfe3 function show this phenotype. The osteoclast phenotype never shows interallelic complementation.

The phenotype of Mitf^Mi-wh mutant animals is unusual in that heterozygotes are severely affected while homozygotes have an intermediate phenotype compared to that of the other Mitf alleles: In heterozygotes coat color is severely affected while eye development is not; in homozygotes, coat color is still severely affected while eye development has an intermediate phenotype. This suggests different effects of the mutation in the two pigment cells affected, melanocytes and RPE cells. Furthermore, it suggests that the dominant-negative nature of the Mitf^Mi-wh mutation has much more serious effects in melanocytes than in RPE cells.

In the eye, complementation is observed in many different combinations with Mitf^Mi-wh, even in Mitf^Mi-wh/Mitf^mi-vga9 animals in which the Mitf^Mi-wh allele is paired with a loss-of-function mutation. Coat color is complemented only when Mitf^Mi-wh is paired with hypomorphic mutations such as Mitf^Mi-b and Mitf^mi-ws, which already have normal eye development (HOLLANDER 1964 ). This suggests that the complementation occurs in a reverse allelic series, again reflecting the activity requirement for Mitf in the different cell types. The threshold for the Mitf requirement is lower in RPE cells than in melanocytes and therefore the eye phenotype is more easily complemented than the coat color phenotype. Interestingly, no complementation is observed in combinations with alleles with a phenotype more normal than that of Mitf^Mi-wh in homozygous condition.

To date, two main models have been proposed to explain interallelic complementation. The first model involves transvection in which one allele affects the expression of a second allele on the homologous chromosome. Generally, these involve combinations of a regulatory mutation with a mutation in the coding region. Transvection depends on chromosomal pairing and chromosomal rearrangements have been shown to interfere with the complementation. The best examples of this involve the Ultrabithorax (Ubx) gene in Drosophila where pairing of certain alleles results in partial complementation of the phenotype (reviewed by PIRROTTA 1999 ). The second model for interallelic complementation involves dimer formation between protein monomers expressed by two different alleles. The different alleles affect different functional domains and the resulting dimer is therefore partially functional and complementation is observed. Examples of this include the Egfr locus in Drosophila (RAZ et al. 1991 ) and the let-23 gene in Caenorhabditis elegans (AROIAN et al. 1994 ).

In the case of Mitf reported here, neither of these two models clearly apply. Transvection is a highly unlikely explanation for Mitf-associated interallelic complementation. The only allele that results in complementation (Mitf^Mi-wh) does not detectably affect regulation of the gene and is a mutation in the protein-coding region (STEINGRIMSSON et al. 1994 ). Among the alleles that fail to be complemented by Mitf^Mi-wh are mutations that affect regulatory regions as well as mutations that affect protein-coding regions. The common theme among the mutations that fail to be complemented is that they have milder effects in the homozygous condition than Mitf^Mi-wh and the mutations that are complemented by Mitf^Mi-wh. This does not fit the criteria of transvection.

Despite the fact that the MITF protein can form both homo- and heterodimers, it is difficult to see how the Mitf-associated complementation could be due to the formation of functional dimers by two different mutant proteins. Three of the mutations that show interallelic complementation with Mitf^Mi-wh (Mitf^mi-ew, Mitf^mi-mi, and Mitf^Mi-or) all affect the DNA-binding basic domain, the very same domain affected by the Mitf^Mi-wh mutation (Table 1; STEINGRIMSSON et al. 1994 ). The fourth mutation to show interallelic complementation is Mitf^mi-b, whose molecular defect is in the loop of the HLH domain and also affects DNA-binding abilities of the protein (Table 1; STEINGRIMSSON et al. 1996 ). Thus, all these mutations affect DNA binding. Most interestingly, the loss-of-function mutation Mitf^mi-vga9 can complement the Mitf^Mi-wh mutation, suggesting that for complementation to take place it is better to have no or very little Mitf activity with the Mitf^Mi-wh mutation than to have partial function. These facts are difficult to reconcile with the model of complementation by active dimers.

Our observations lead us to propose an alternative dose-dependent model of interallelic complementation (Fig 4). According to this model, the phenotype is determined by the level of the Mitf^Mi-wh protein in combination with a cell-type-specific level of sensitivity to this protein. Earlier studies have shown that like wild-type Mitf, the Mitf^Mi-wh protein comes in two distinct isoforms, which differ in the presence or absence of six residues upstream of the basic domain (STEINGRIMSSON et al. 1994 ). The alternative six amino acids are the result of alternative splice acceptor sites in exon 6; all tissues analyzed so far contain both splice forms (HALLSSON et al. 2000 ). In vitro, the Mitf^Mi-wh protein lacking the alternative six amino acids acts in a dominant-negative fashion; the protein can dimerize with partner proteins but cannot bind DNA and interferes with the DNA binding of the related Tfe3 protein in vitro (HEMESATH et al. 1994 ). However, although the Mitf^Mi-wh protein containing the alternative six amino acids cannot bind DNA as a homodimer, it can bind DNA as a heterodimer with Tfe3 at levels similar to the wild-type Mitf protein (HEMESATH et al. 1994 ). Thus, the mutation produces two different proteins with critically altered function: a dominant-negative protein [Mitf^Mi-wh(-6)] and a protein that cannot bind DNA as a homodimer, yet can bind DNA as a heterodimer [Mitf^Mi-wh(+6)] (Fig 4A) with its partners. Due to the intermediate phenotype of Mitf^Mi-wh homozygous animals, one of these proteins must be partly functional.

View larger version (21K): In this window In a new window Download PPT slide   Figure 4. The biochemical behavior of the MitfMi-wh protein provides a model for the interallelic complementation at Mitf. (A) The biochemical behavior of the two different MitfMi-wh proteins. (B) A model depicting the effects of the mutant proteins on the phenotype. Yellow bars represent levels of normal Mitf function while red bars indicate levels of neomorphic activity. The dotted lines indicate threshold requirements for each type of activity. The thresholds can be different in the different cell types affected. Three different situations are shown: wild type, MitfMi-wh homozygotes, and the MitfMi-wh/ Mitfmi-vga9 heteroallelic combination. In wild type (left), two different messages are expressed in all tissues analyzed, leading to the synthesis of two different Mitf proteins: The Mitf(-6) protein has 20% less stability in a complex with DNA than the Mitf(+6) protein (HEMESATH et al. 1994 ). In MitfMi-wh homozygotes (middle), the MitfMi-wh(+6) protein is more or less normal as a heterodimer while the MitfMi-wh(-6) protein cannot bind DNA as a homo- or heterodimer. In addition, it has acquired new (neomorphic) properties, which negatively affect the phenotype. The resulting phenotype is a trade-off between the two different effects and differs in cell types. In the MitfMi-wh/Mitfmi-vga9 heteroallelic combination (right) the dose of neomorphic activity has been reduced by one-half such that it is below the threshold required to see effects on the phenotype. At the same time, the normal activity of the MitfMi-wh(+6) protein is sufficient to allow normal function and complementation is observed.

The nature of the two different forms of the Mitf^Mi-wh protein is likely to provide the explanation for both the interallelic complementation associated with this mutation and the fact that Mitf^Mi-wh has the most severe phenotype in heterozygous condition. We therefore propose that the genetic behavior of this mutation is due to the relative effects of homo- and heterodimers involving this mutation where the Mitf^Mi-wh(-6) protein is very efficient as a dominant-negative protein and where Mitf^Mi-wh(+6) is at least partly functional. The two different versions of the Mitf^Mi-wh protein may not be able to form DNA-binding dimers between themselves but may be able to dimerize with other partner proteins in the cell. Furthermore, we propose that one of the two isoforms of the Mitf^Mi-wh protein (or both) has acquired a new (neomorphic) function, resulting in negative effects in the cell. This new action may be the result of dominant-negative action of the Mitf^Mi-wh (-6) protein against essential proteins or pathways in the cell. It may also be due to a novel action of the Mitf^Mi-wh(+6) protein, e.g., binding the wrong promoter sequence or activating the wrong set of genes with subsequent negative effects. For the following discussion we assume that the new (neomorphic) activity is associated with the Mitf^Mi-wh(-6) protein. Finally, we propose that the different activities of the two forms of the Mitf^Mi-wh protein each have their own threshold requirements in the different cell types affected. A model depicting the Mitf-associated complementation is shown in Fig 4B.

From the mutant phenotype it is clear that the effects of the Mitf^Mi-wh mutation are relatively more serious in the neural-crest-derived melanocytes than in RPE cells or osteoclasts. Coat color is severely affected in Mitf^Mi-wh homo- and heterozygotes while only homozygotes show intermediate microphthalmia; bone development is normal in both cases. This suggests a tissue-specific difference in the effects of the Mitf^Mi-wh mutation. Perhaps this is an indication that the mutant protein interacts with one or more melanocyte-specific proteins. Thus, in Mitf^Mi-wh/+ animals, the strong dominant-negative action of the Mitf^Mi-wh(-6) protein may interfere with partner proteins in melanocytes. Alternatively, the new (neomorphic) activity may negatively affect the function of melanocyte-specific factors or processes. Together, these effects result in the severe coat color phenotype observed in heterozygotes. Although the RPE cells of Mitf^Mi-wh/+ and Mitf^Mi-wh/Mitf^Mi-wh animals also contain the neomorphic activity (since they also express the mutant Mitf gene), RPE cells are not as severely affected since the activity threshold is different (Fig 4B) and the RPE cells do not express the melanocyte-specific protein(s) against which the dominant-negative Mitf^Mi-wh (-6) protein acts.

The intermediate eye phenotype observed in Mitf^Mi-wh homozygotes supports the idea that the Mitf^Mi-wh(+6) protein has partially normal function in RPE cells. The partially normal function of the Mitf^Mi-wh(+6) protein is unaffected by the dominant-negative activity of the Mitf^Mi-wh(-6) protein since Mitf^Mi-wh(+6) is unable to form DNA-binding homodimers with other mutant Mitf^Mi-wh proteins (HEMESATH et al. 1994 ). In the different heteroallelic combinations, the new (neomorphic) action of Mitf^Mi-wh has been reduced by half as compared to that of Mitf^Mi-wh homozygotes, thereby reducing the effects of neomorphic activity below a certain threshold level and allowing more normal development to proceed (Fig 4B). Thus, the heteroallelic combination of Mitf^Mi-wh and Mitf^mi-vga9 uncovers the partially normal function of the Mitf^Mi-wh(+6) protein, which is sufficient to generate eyes of normal size. In heteroallelic combinations with milder Mitf mutations that still retain some Mitf activity, the dominant-negative Mitf^Mi-wh(-6) protein interferes with the function of the mutant proteins and an intermediate phenotype is observed. The neomorphic activity is still effective, resulting in no complementation.

Our model predicts that gene expression is affected differently by the two versions of the Mitf^Mi-wh protein than by the corresponding versions of wild-type Mitf proteins. Gene expression studies using gene arrays and/or studies on the effects of the different Mitf proteins on well-defined promoter elements in the different cell types can therefore be used to test the model in vitro.

	FOOTNOTES

² Present address: 8255 Sandy Point Road, Bryan, TX 77807.

	ACKNOWLEDGMENTS

We thank Debbie Swing, Joanne Dietz, and Fran Dorsey for expert technical assistance and the staff of the Histopathology Laboratory, Frederick, Maryland, for help with histology. This work was supported by the National Cancer Institute (N.G.C. and N.A.J.), the National Institute for Neurological Disorders and Stroke, DHHS (H.A.), and the Icelandic Research Council (E.S. and J.H.H.).

Manuscript received July 30, 2002; Accepted for publication October 15, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AROIAN, R. V., G. M. LESA, and P. W. STERNBERG, 1994  Mutations in the Caenorhabditis elegans let-23 EGFR-like gene define elements important for cell-type specificity and function. EMBO J. 13:360-366.[Medline]

BENTLEY, N. J., T. EISEN, and C. R. GODING, 1994  Melanocyte-specific expression of the human tyrosinase promoter: activation by the microphthalmia gene product and role of the initiator. Mol. Cell. Biol. 14:7996-8006.[Abstract/Free Full Text]

GRÜNEBERG, H., 1953  The relations of microphthalmia and white in the mouse. J. Genet. 51:359-362.

HALLSSON, J. H., J. FAVOR, C. HODGKINSON, T. GLASER, and M. L. LAMOREUX et al., 2000  Genomic, transcriptional and mutational analysis of the mouse microphthalmia locus. Genetics 155:291-300.[Abstract/Free Full Text]

HEMESATH, T. J., E. STEINGRÍMSSON, G. MCGILL, M. J. HANSEN, and J. VAUGHT et al., 1994  microphthalmia, a critical factor in melanocyte development, defines a discrete transcription factor family. Genes Dev. 8:2770-2780.[Abstract/Free Full Text]

HODGKINSON, C. A., K. J. MOORE, A. NAKAYAMA, E. STEINGRÍMSSON, and N. G. COPELAND et al., 1993  Mutations at the mouse microphthalmia locus are associated with defects in a gene encoding a novel basic-helix-loop-helix-zipper protein. Cell 74:395-404.[Medline]

HOLLANDER, W. F., 1964  Mouse News Lett. 30:29.

HOLLANDER, W. F., 1968  Complementary alleles at the mi-locus in the mouse. Genetics 60:189.

HUGHES, M. J., J. B. LINGREL, J. M. KRAKOWSKY, and K. P. ANDERSON, 1993  A helix-loop-helix transcription factor-like gene is located at the mi locus. J. Biol. Chem. 268:20687-20690.[Abstract/Free Full Text]

KONYUKHOV, B. V. and V. V. OSIPOV, 1968  Interallelic complementation of microphthalmia and white genes in mice. Genetika 4:65-76.

LAMOREUX, M. L., R. E. BOISSY, J. E. WOMACK, and J. J. NORDLUND, 1992  The vit gene maps to the mi (Micropthalmia) locus of the laboratory mouse. J. Hered. 83:435-439.[Abstract/Free Full Text]

LERNER, A. B., T. SHIOHARA, R. E. BOISSY, K. A. JACOBSON, and M. L. LAMOREUX et al., 1986  A mouse model for vitiligo. J. Invest. Dermatol. 87:299-304.[Medline]

MOORE, K. J., 1995  Insight into the microphthalmia gene. Trends Genet. 11:442-448.[Medline]

MORII, E., T. TSUJIMURA, T. JIPPO, K. HASHIMOTO, and K. TAKEBAYASHI et al., 1996  Regulation of mouse mast cell protease 6 gene expression by transcription factor encoded by the mi locus. Blood 88:2488-2494.[Abstract/Free Full Text]

MOTYCKOVA, G., K. N. WEILBAECHER, M. HORSTMANN, D. J. RIEMAN, and D. Z. FISHER et al., 2001  Linking osteopetrosis and pycnodysostosis: regulation of cathepsin K expression by the microphthalmia transcription factor family. Proc. Natl. Acad. Sci. USA 98:5798-5803.[Abstract/Free Full Text]

NAKAYAMA, A., M.-T. T. NGUYEN, C. C. CHEN, K. OPDECAMP, and C. A. HODGKINSON et al., 1998  Mutations in microphthalmia, the mouse homolog of the human deafness gene MITF, affect neuroepithelial and neural crest-derived melanocytes differently. Mech. Dev. 70:155-166.[Medline]

PIRROTTA, V., 1999  Transvection and chromosomal trans-interaction effects. Biochim. Biophys. Acta 1424:M1-M8.[Medline]

RAZ, E., E. SCHEJTER, and B. Z. SHILO, 1991  Interallelic complementation among DER/flb alleles: implications for the mechanisms of signal transduction by receptor-tyrosine kinases. Genetics 129:191-201.[Abstract]

SCHAIBLE, R. H., 1963 Developmental genetics of spotting patterns in the mouse. Ph.D. Thesis, Iowa State University, Ames, IA.

STEINGRÍMSSON, E., K. J. MOORE, M. L. LAMOREUX, A. R. FERRÉ-D'AMARÉ, and S. K. BURLEY et al., 1994  Molecular basis of mouse microphthalmia (mi) mutations helps explain their developmental and phenotypic consequences. Nat. Genet. 8:256-263.[Medline]

STEINGRÍMSSON, E., A. NII, D. E. FISHER, A. R. FERRÉ-D'AMARÉ, and R. J. MCCORMICK et al., 1996  The semidominant Mi-b mutation identifies a role for the HLH domain in DNA binding in addition to its role in protein dimerization. EMBO J. 15:6280-6289.[Medline]

STEINGRÍMSSON, E., L. TESSAROLLO, B. PATHAK, L. HOU, and H. ARNHEITER et al., 2002  Mitf and Tfe3, two members of the Mitf-Tfe family of bHLH-Zip transcription factors, have important but functionally redundant roles in osteoclast development. Proc. Natl. Acad. Sci. USA 99:4477-4482.[Abstract/Free Full Text]

TAKEBAYASHI, K., K. CHIDA, I. TSUKUMOTO, E. MORII, and H. MUNAKATA et al., 1996  The recessive phenotype displayed by a dominant negative mirophthalmia-associated transcription factor mutant is a result of impaired nuclear localization potential. Mol. Cell. Biol. 16:1203-1211.[Abstract]

WEST, J. D., G. FISHER, J. F. LOUTIT, M. J. MARSHALL, and N. W. NISBET et al., 1985  A new allele of microphthalmia induced in the mouse: microphthalmia-defective iris (mi di). Genet. Res. 46:309-324.[Medline]

WOLFE, H. G. and D. L. COLEMAN, 1964  Mi-spotted: a mutation in the mouse. Genet. Res. 5:432-440.

YAJIMA, I., S. SATO, T. KIMURA, K. YASUMOTO, and S. SHIBAHARA et al., 1999  An L1 element intronic insertion in the black-eyed white (Mitf ^mi-bw) gene: the loss of a single Mitf isoform responsible for the pigmentary defect and inner ear deafness. Hum. Mol. Genet. 8:1431-1441.[Abstract/Free Full Text]

YASUMOTO, K. and S. SHIBAHARA, 1997  Molecular cloning of a cDNA encoding a human TFEC isoform, a newly identified transcriptional regulator. Biochim. Biophys. Acta 1353:23-31.[Medline]

YASUMOTO, K., K. YOKOYAMA, K. SHIBATA, Y. TOMITA, and S. SHIBAHARA, 1994  Microphthalmia-associated transcription factor as a regulator for melanocyte-specific transcription of the human tyrosinase gene. Mol. Cell. Biol. 14:8058-8070.[Abstract/Free Full Text]

This article has been cited by other articles:

				O. Puk, J. Loster, C. Dalke, D. Soewarto, H. Fuchs, B. Budde, P. Nurnberg, E. Wolf, M. H. de Angelis, and J. Graw Mutation in a Novel Connexin-like Gene (Gjf1) in the Mouse Affects Early Lens Development and Causes a Variable Small-Eye Phenotype Invest. Ophthalmol. Vis. Sci., April 1, 2008; 49(4): 1525 - 1532. [Abstract] [Full Text] [PDF]

				K. Bismuth, S. Skuntz, J. H. Hallsson, E. Pak, A. S. Dutra, E. Steingrimsson, and H. Arnheiter An Unstable Targeted Allele of the Mouse Mitf Gene With a High Somatic and Germline Reversion Rate Genetics, January 1, 2008; 178(1): 259 - 272. [Abstract] [Full Text] [PDF]

				P. A. Leegwater, M. A. van Hagen, and B. A. van Oost Localization of White Spotting Locus in Boxer Dogs on CFA20 by Genome-Wide Linkage Analysis with 1500 SNPs J. Hered., August 3, 2007; (2007) esm022v2. [Abstract] [Full Text] [PDF]

				A. Goldraij, L. J. Beamer, and J. C. Polacco Interallelic Complementation at the Ubiquitous Urease Coding Locus of Soybean Plant Physiology, August 1, 2003; 132(4): 1801 - 1810. [Abstract] [Full Text] [PDF]