Subfunctionalization of Duplicate mitf Genes Associated With Differential Degeneration of Alternative Exons in Fish

Corresponding author: Jean-Nicolas Volff, Biocenter (Theodor Boveri Institute), University of Würzburg, Am Hubland, D-97074 Würzburg, Germany., volff{at}biozentrum.uni-wuerzburg.de (E-mail)

Communicating editor: D. J. GRUNWALD

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The microphthalmia-associated transcription factor (MITF) exists in at least four isoforms. These are generated in higher vertebrates using alternative 5' exons and promoters from a single gene. Two separate genes (mitf-m and mitf-b), however, are present in different teleost fish species including the poeciliid Xiphophorus, the pufferfishes Fugu rubripes and Tetraodon nigroviridis, and the zebrafish Danio rerio. Fish proteins MITF-m and MITF-b correspond at both the structural and the expression levels to one particular bird/mammalian MITF isoform. In the teleost lineage subfunctionalization of mitf genes after duplication at least 100 million years ago is associated with the degeneration of alternative exons and, probably, regulatory elements and promoters. For example, a remnant of the first exon specific for MITF-m is detected within the pufferfish gene encoding MITF-b. Retracing the evolutionary history of mitf genes in vertebrates uncovered the differential recruitment of new introns specific for either the teleost or the bird/mammalian lineage.

GENE duplications are important events in evolution (OHNO 1970 ). One major advantage is that one of the versions of the ancestral genes is free to diverge independently of the selective forces that maintain the original function. This may lead to one of three scenarios. First, and obviously most frequently, one copy may be inactivated by accumulation of random mutations and become nonfunctional ("defunctionalization"). Second, one of the genes may diverge to a state where it acquires a new function ("neofunctionalization"). Third, both genes may be altered in a way such that the combined activity of the two genes together fulfills the task of the ancestral gene in a complementary fashion ("subfunctionalization"; FORCE et al. 1999 ).

Fish are a group of organisms especially suited for the study of such events, because it became evident that they have more duplicated genes than do the other groups of vertebrates (WITTBRODT et al. 1998 ; MEYER and SCHARTL 1999 ). Within the three consequences of gene duplication, subfunctionalization has attracted a lot of interest and paradigmatic cases have been described of genes where two genes in fish perform together the function of their common tetrapod ortholog (FORCE et al. 1999 ).

Two different isoforms of the microphthalmia-associated transcription factor (MITF) that appear to fulfill the criteria of subfunctionalization (LISTER et al. 2001 ) were recently described in zebrafish. MITF is a member of the basic helix-loop-helix leucine zipper (bHLH-Zip) protein family (HODGKINSON et al. 1993 ; HUGHES et al. 1993 ). It plays a central role in the differentiation and/or survival of melanocytes, which is reflected in the correlation between a number of mutant pigmentation phenotypes and specific mutations in the mitf gene (STEINGRIMSSON et al. 1994 ; LISTER et al. 1999 ; HALLSSON et al. 2000 ). Different mutations in the human gene can lead to Waardenburg syndrome type 2 (TASSABEHJI et al. 1994 ) or Tietz syndrome (SMITH et al. 2000 ), both of which are associated with pigmentary changes.

In mammals and birds, several isoforms of MITF that are encoded by a single gene are known. The different proteins are generated by the use of alternative promoters and 5' exons (YASUMOTO et al. 1998 ; UDONO et al. 2000 ). The zebrafish data, on the other hand, indicate the presence of different genes for at least two isoforms in this organism (LISTER et al. 1999 , LISTER et al. 2001 ).

The mammalian and avian MITF-m isoforms have a unique N terminus of 11 amino acids not found in the other variants a, c, and h. These last three isoforms have a common exon, B1b, directly upstream of exon 2, but differ in their N terminus, encoded by the exons a, c, and h, respectively. Differences are not limited to the primary structure of the protein, but are also found at the transcriptional level. Whereas mitf-m is expressed exclusively in melanocytes and melanoma cells, the other isoforms are expressed in a much wider spectrum of cells (AMAE et al. 1998 ; UDONO et al. 2000 ). Convergent evidences from different organisms including human indicate that MITF-m is the isoform involved in melanocyte differentiation (VACHTENHEIM and NOVOTNA 1999 ; YAJIMA et al. 1999 ). The MITF-a isoform most likely exerts its major function in the retinal pigment epithelium (RPE).

In zebrafish two cDNAs coding for MITF have been isolated, which differ not only in their 5' terminal exons but over the entire length. These cDNAs originate from two separate genes. One, mitf-a is required for the development of neural crest-derived pigment cells, while the other, mitf-b, is expressed in the RPE and can rescue MITF-a function in the eye, but not in the neural crest-derived pigment cells. This situation clearly demonstrates the use of two mitf genes in zebrafish as an alternative genomic strategy to the subfunctionalization through alternative promoter/5' exon usage of birds and mammals.

For a full evolutionary understanding of this peculiar situation, in addition to the gene function and cDNA sequence data, information on the structure of the corresponding genes is required, which is not available from fish so far. Another important question that was evoked from the zebrafish data concerns the origin and fate of the two mitf genes. It has been proposed that the source for most of the duplicated genes in fish was an ancient whole genome duplication in the lineage leading to modern day teleosts (AMORES et al. 1998 ; POSTLETHWAIT et al. 2000 ; TAYLOR et al. 2001 ). This is suggested by the presence of large syntenic segments on different chromosome pairs in zebrafish. Zebrafish mitf genes, located respectively in linkage groups 6 and 13, are apparently not included in such large regions of synteny (LISTER et al. 2001 ). Although additional genomic rearrangements might have disrupted syntenic regions after genome duplication, this result is at least equally consistent with a more local event of gene duplication. Thus, the duplication and subfunctionalization of mitf genes could be an exceptional situation in zebrafish but not a mechanism of general importance in fish.

To perform a broader evolutionary and functional survey of this pair of duplicate genes in the teleost lineage, we have analyzed mitf genes in the poeciliid Xiphophorus, which comes from a totally different branch of the phylogenetic tree of teleosts. The genomic structure of mitf genes was also determined from the almost completely sequenced genome of Fugu (Takifugu) rubripes and from an advanced version (6x coverage) of the genome of the related freshwater pufferfish Tetraodon nigroviridis.

We show the presence of two mitf genes in all these different teleost species, indicative of a duplication at least 100 million years ago. We find that the mitf genes of Xiphophorus are differentially expressed, supporting the interpretation that these genes have experienced an ancient subfunctionalization associated with differential degeneration of specific regulatory sequences and use of alternative exons in each of the duplicate genes.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Experimental animals and cell lines:
All fish used for this study were bred under standard conditions (KALLMAN 1975 ). Xiphophorus maculatus strain I is homozygous for the golden mutation; strain II represents wild-type X. maculatus, origin Rio Jamapa, Mexico, and is derived from the Jp163A strain; and strain III stands for ornamental X. maculatus. X. helleri is a wild-type strain from Rio Lancetilla, Guatemala.

The embryonic epithelial cell line A2 (KUHN et al. 1979 ) and the melanoma cell line PSM (WAKAMATSU 1981 ), both from Xiphophorus, were cultured as described previously (BAUDLER et al. 1997 ).

DNA and RNA experiments:
Fish genomic DNA was extracted according to SCHARTL et al. 1995 . Standard PCRs were performed with 200 ng genomic DNA as template. For RT-PCR, 1 µg of total RNA isolated with TRIzol (GIBCO BRL, Gaithersburg, MD) was reverse transcribed with Super Script II (GIBCO BRL) in the presence of random hexamers. Xmitf-m-specific cDNA was amplified in RT-PCR using primers MiEx1 for (5'-TGCTGGAGATGCTGGAATACAG-3') and MiEx9rev3 (5'-GACGGTGAGACCGTGAGCC-3'); Xmitf-b-specific cDNA was amplified with MiExBfor3 (5'-CGCTATCAATGTCAGTGTCCC-3') and MiEx9rev3. To obtain full-length cDNA sequences, the 5' and 3' rapid amplification of cDNA ends (RACE) systems from GIBCO BRL were used. PCR fragments were cloned into pUC18 using the SureClone kit (Amersham Pharmacia Biotech) or into pCR II-TOPO (Invitrogen, San Diego). DNA sequencing was done using the ThermoSequenase fluorescent labeled primer cycle sequencing kit with 7-deaza-dGTP and run on an ALFexpress automated laser fluorescent sequencer (Amersham Pharmacia Biotech), or using the CEQD TCS dye terminator cycle sequencing kit and run on a CEQ 2000XL DNA sequencing system (Beckman-Coulter).

Southern blotting was done according to standard protocols. Hybridization was performed in 50% formamide at 42° and the filters were washed with 0.1x SSC/1% SDS at 68°. Screening of the PSM cDNA phage library was performed using a PCR-based phage selection procedure (ISRAEL 1993 ). The cDNA-containing plasmid was excised in vivo (Stratagene, Lambda-ZAP Express library).

Sequence analysis:
Multiple sequence alignments were generated using "PileUp" of the GCG Wisconsin package (Version 10.0, Genetics Computer Group, Madison, WI). Phylogenies were determined with PAUP* (SWOFFORD 1989 ) by bootstrap analysis using maximum parsimony (100 replicates) and neighbor joining (SAITOU and NEI 1987 ; 1000 replicates). Maximum-likelihood analysis was performed by quartet puzzling (10,000 puzzling steps; STRIMMER and VON HAESELER 1996 ) with PAUP*. The pairwise number of synonymous (K_s) and nonsynonymous (K_a) substitutions per site between two aligned sequences was estimated with "Diverge" of GCG using an unambiguous alignment of DNA sequences [390 nucleotides (nt) in length]. Gene structure was analyzed using programs available on the NIX server (http://menu.hgmp.mrc.ac.uk/menu-bin/Nix).

Isolation of Xiphophorus mitf cDNAs:
To obtain Xiphophorus sequences coding for MITF, RNA from the melanoma cell line PSM was reverse transcribed and the resulting cDNA was used as template in a touchdown PCR using degenerate primers MiEx8for (5'-TGMGSTGGAAYAARGGMACC-3') and MiEx8rev (5'-CTGWAYWCKGAGCARSARRTG-3'). These primers were derived from the region encoding the helix-loop-helix domain, which is highly conserved between the known vertebrate MITF proteins. A product of 140 bp was obtained, which unequivocally codes for a part of MITF different from all other vertebrate MITFs and, thus, most likely codes for Xiphophorus MITF. A 902-bp cDNA fragment encoding parts of a different isoform (XMITF-b) was obtained from the amplification of liver cDNA with the primers MiExBfor3 (5'-CGCTATCAATGTCAGTGTCCC-3') and MiEx9rev3 (5'-GACGGTGAGACCGTGAGCC-3').

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Duplicate mitf genes in Xiphophorus:
To isolate mitf cDNA(s) from Xiphophorus, nondegenerate primers derived from the highly conserved small fragment within the helix-loop-helix region (see MATERIALS AND METHODS) were used to screen a PSM melanoma cell line cDNA library by PCR. A cDNA encompassing a complete MITF open reading frame was obtained, which encodes the homolog of the mammalian melanocyte-specific isoform MITF-m (XMITF-m) as indicated by the presence of the melanocyte-specific first exon.

Using a degenerate primer derived from the mammalian exon B1b (AMAE et al. 1998 ) in combination with a primer from exon 9 common to all isoforms, we obtained from liver RNA a mitf-cDNA distinct from the Xmitf-m cDNA, which was designated Xmitf-b in analogy to the corresponding gene from zebrafish. 5' and 3' RACE were used to complete the sequence information. Differences between Xmitf-m and Xmitf-b were not limited to the 5' exons, but were found along the whole sequence (Fig 1). The cDNAs showed only 72% identity at the nucleotide level (data not shown), indicating that they originated from separate genes. Southern blot analyses with probes from both cDNAs confirmed this interpretation (Fig 2).

View larger version (110K): In this window In a new window Download PPT slide   Figure 1. Comparison of vertebrate MITF proteins. The conceptual translation products of fish mitf genes are compared to other vertebrate MITFs. Identical residues are white type on black and conservative substitutions are black type on gray (drawn with MacBoxshade). The acidic activation domain (aad), the basic region, the helix-loop-helix domain, and the leucine residues (L) structuring the Zip region are shown. Accession numbers: MITF-m zebrafish, AAD48371; MITF-b zebrafish, AAK95588; MITF mouse, AAF81266; MITF human, I38024; MITF chicken, AAF67467.

View larger version (51K): In this window In a new window Download PPT slide   Figure 2. Genomic structure (A) and expression (B) of mitf genes in Xiphophorus. (A) Genomic DNAs from several X. maculatus and X. helleri genotypes (see MATERIALS AND METHODS) were hybridized with Xmitf-m- and Xmitf-b-specific probes. The blot was first hybridized with the full-length Xmitf-m cDNA (top), stripped, and rehybridized with a Xmitf-b cDNA fragment (bottom). Similar results were also obtained in experiments using the same DNAs digested with BamHI and with DNAs from seven other genotpyes (data not shown). (B) The presence of Xmitf transcripts was analyzed by RT-PCR, in which primer combinations MiEx1for/MiEx9rev3 for Xmitf-m (top) and MiExBfor3/MiEx9rev3 for Xmitf-b (bottom) were used for the amplification of specific cDNA of similar length. For the amplification of Xmitf-b transcripts five times more cDNA was used than for the amplification of Xmitf-m. RNAs were prepared from normal tissues of backcross hybrids between X. maculatus and X. helleri and from benign (ben. mel.) and malignant (mal. mel.) melanoma. In addition to tissue RNAs, RNAs from the Xiphophorus melanoma cell line PSM and the embryonic cell line A2 were included in the analysis. M is a DNA size marker and water a control reaction without template.

The predicted proteins show all the hallmarks of their counterparts from higher vertebrates (Fig 1). The basic helix-loop-helix region in the central part is identical to the mammalian protein with the exception of a single lysine-to-arginine substitution in a position where a methionine residue is found in the pufferfish MITF-m proteins. The leucine zipper motif directly adjacent serving as a dimerization interface is also well conserved. Mapping studies had revealed the presence of two activation domains in the mammalian protein, one strong acidic activator region in the N-terminal half of the molecule and one threonine-rich region close to the C terminus with modest transactivation properties (TAKEDA et al. 2000 ). While the first region is nearly identical in the fish proteins with all acidic residues present, the second region is poorly conserved.

Expression of the Xmitf isoforms in tissues from adult Xiphophorus fish and in two established cell lines was analyzed by RT-PCR (Fig 2). Xmitf-b seems to be expressed ubiquitously, as specific transcripts were present in all cell types analyzed. The actual concentrations are fairly variable with highest levels found in muscle. Xmitf-m RNA, however, was detected only in tissues containing melanocytes, like eyes and skin, and in melanomas as well as in the melanoma cell line PSM, thus paralleling the expression pattern of mitf-m in mammals. In addition, the levels of Xmitf-m RNA in the melanoma-derived samples are considerably higher than levels of Xmitf-b transcripts.

Duplicate mitf genes in pufferfish genomes:
Two different mitf genes could be identified in sequences generated by the Japanese pufferfish F. rubripes genome project (ELGAR et al. 1996 , ELGAR et al. 1999 ; http://www.jgi.doe.gov/programs/fugu/fugu_mainpage.html, http://fugu.hgmp.mrc.ac.uk/blast/): mitf-b (accession FT:T000263 scaffold_263, length 82968 nt), and mitf-m (accession FT:T008421 scaffold_8422 length 10919 nt). Both mitf genes could be reconstructed from the freshwater pufferfish T. nigroviridis as well (http://www.genoscope.cns.fr, CROLLIUS et al. 2000 ; FISCHER et al. 2000 ; sequences available on request).

In both pufferfishes, the nucleotide identity between exonic sequences from the two mitf genes was again only 72%. All four genes contain introns at similar positions, indicating that the duplication of the mitf gene occurred at the genomic level rather than through retrotransposition of mitf mRNA (Fig 3). The prediction of intron placement and exon sizes in the pufferfishes match the gene structure of mitf-m in Xiphophorus, which was determined from the size and/or sequence of genomic PCR products (data not shown).

View larger version (16K): In this window In a new window Download PPT slide   Figure 3. Genomic structure of Fugu and mammalian mitf genes. Exons are numbered according to the homologous regions of the human mitf gene. The approximate size of Fugu introns is given in kilobases.

Phylogeny and evolution of MITF sequences:
In contrast to the situation observed in higher vertebrates, phylogenetic analyses supported the presence of two distinct groups of orthologous MITF sequences in teleost species having diverged at least 100–120 million years ago: the MITF-b group, including proteins homologous to the MITF-b proteins from the zebrafish Danio rerio (LISTER et al. 2001 ) that contain sequences encoded by exons 1a and 1b (structurally related to the MITF-a isoform from higher vertebrates), and the MITF-m group, including proteins homologous to the zebrafish MITF-m (LISTER et al. 1999 , LISTER et al. 2001 ) that contain exon 1m and thus are structurally more related to the MITF-m isoform from mammals and birds (Fig 1 and Fig 4). Both groups contain bona fide MITF proteins that do not correspond to fish versions of the related TFE3 transcription factor (REHLI et al. 1999 ). Maximum parsimony (bootstrap value 86%) but neither neighbor joining nor maximum-likelihood analyses supported the existence of a fish-specific group containing both MITF-b and MITF-m proteins.

View larger version (25K): In this window In a new window Download PPT slide   Figure 4. Phylogenetic analysis of MITF-related proteins. Protein regions encoded by exons 1a/1b/1m have been removed for analysis. The tree (neighbor joining) is unrooted. Bootstrap values (%) using neighbor joining (1000 replicates, first value) and maximum parsimony analyses (100 replicates, second value), as well as the reliability values (%) for maximum-likelihood analysis (quartet puzzling, 10,000 puzzling steps, third value) are given. Branches with <50% support have been collapsed. Only the maximum parsimony supported the presence of a fish-specific group containing the MITF-b and MITF-m sequences (bootstrap value 86%). Accession numbers: MITFm D. rerio, AAD48371; MITFb D. rerio, AAK95588; MITF Mus musculus, AAF81266; MITF Homo sapiens, I38024; MITF Gallus gallus, AAF67467; TFE3a D. rerio, AAK95589; TFE3 H. sapiens, P19532; TFE3 M. musculus, Q64092; TFEB H. sapiens, CAB54146; TFEB M. musculus, NP_035679; TFEC M. musculus, AF077742; TFEC H. sapiens, D43945; TFE Caenorhabditis elegans, AAB37997.

Since duplicated gene copies without evolutionary constraints can frequently evolve as pseudogenes, we compared the number of synonymous (K_s) vs. nonsynonymous (K_a) substitutions in the mitf genes. In all comparisons, the number of synonymous (silent) substitutions was considerably greater than the number of nonsynonymous substitutions (min.–max. of K_s/K_a rate): 18.1–22.7 between mitf-b fish sequences, 7.6–20.4 between mitf-m fish sequences, 8.5–23.2 between mitf-b and mitf-m fish sequences, 13.2–32.6 between mitf-b fish sequences and sequences from higher vertebrates, 11.6–28.4 between mitf-m fish sequences and sequences from higher vertebrates, and >40 between mitf sequences from higher vertebrates. This strongly suggests that all fish mitf-b and -m genes are not pseudogenes but on the contrary mostly evolved under purifying selection. The fact that the K_s/K_a ratio seems to be smaller in comparisons involving fish mitf sequences than between sequences from higher eukaryotes might be due to relaxed selective constraints after duplication (LYNCH and CONERY 2000 ).

Differential exon recruitment or degeneration during mitf evolution in vertebrates:
Comparative analysis of mitf cDNA sequences from fish and higher vertebrates revealed that the alternatively spliced bird/mammalian exon 3 was absent from all fish cDNAs described so far (Fig 1). Analysis of the intervening sequence (maximal size 80 nt) between exons 2 and 4 (according to the mammalian nomenclature) in the mitf-b and mitf-m genes from both Japanese and freshwater pufferfishes showed that this exon is not present in the mitf genes (Fig 3). An additional exon (termed 5b) was found in cDNA and genomic sequences in fish but not in mammals (Fig 1 and Fig 3). Neither intron 3 nor intron 5b could be detected in the human tfe3 gene (Fig 3). This strongly suggests the acquisition of intron 3 and intron 5b in higher vertebrates and fish, respectively, after divergence of the mammalian/bird lineage from the teleost lineage.

According to the phylogenetic analysis, both F. rubripes and T. nigroviridis mitf-m genes contain a typical 1m exon about 700 nt upstream from exon 2 (Fig 1, Fig 3, and Fig 5). This exon is predicted by half of the gene structure analysis programs available on the NIX server (http://menu.hgmp.mrc.ac.uk/menu-bin/Nix). The 5' splice site of intron 1 in both mitf-m genes (CAG/GTGAGA) is conserved between both pufferfishes and is closely related to the consensus of 5' canonical splice sites in mammals (M₇₀A₆₀G₈₀/GTR₉₅A₇₁G₈₁T₄₆; BURSET et al. 2001 ). Interestingly, 1m-exon-like sequences were also detected in the same region in both F. rubripes and T. nigroviridis mitf-b genes (Fig 3 and Fig 5). However, these sequences were degenerated. Their putative translation products lack amino acid residues that are highly conserved in MITF-m proteins from fish and higher vertebrates (Fig 5). While the 1m-like sequence of the Fugu mitf-b gene was still predicted as an exon by several programs from the NIX package, this sequence was not identified as an exon in the Tetraodon mitf-b gene. This might be due to a less favorable splicing context (GAG/GTATGG), in which the A present in 71% of 5' splice sites at position +4 (BURSET et al. 2001 ) has been replaced by a T, found in only 9.3% of 5' splicing sites in mammals. Sequences present in the Tetraodon database did not allow the complete reconstruction of the mitf-b gene and the identification of exons 1a and 1b, probably because of the relatively large size of introns 1 and 2 as observed in the Fugu mitf-b gene (Fig 3). No exons 1a and 1b could be detected within 4.5- and 2.2-kb genomic sequences upstream from the 1m exon of mitf-m genes in F. rubripes and T. nigroviridis, respectively, but the presence of more distant degenerated exon 1a/b sequences in these genes cannot be excluded.

View larger version (29K): In this window In a new window Download PPT slide   Figure 5. Exon 1m-related sequences in pufferfish mitf genes. Horizontal arrow shows the putative ATG start codon, vertical arrow shows the position of the predicted 5' splice site of intron 1 in mitf-m. Amino acid residues that are identical within the sequence encoded by all bona fide m exons in vertebrates are boxed.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Melanocyte development as well as postnatal proliferation and survival is dependent on MITF, the gene product of the microphthalmia locus. In mammals and birds a single gene within this locus, mitf, encodes several proteins with different N termini. Use of alternative promoters and first exons results in different mRNAs encoding these MITF isoforms. One of them, MITF-m, is of critical importance for the neural crest-derived melanocytic cell lineage. In addition to its function in normal pigmentation, MITF-m might also play a role in transformed pigment cells, as it has been detected in melanomas in contrast to many other melanocytic markers, which are lost during the transformation process (KING et al. 1999 ).

In contrast to the situation in higher vertebrates, we have now demonstrated the presence of at least two distinct mitf genes in the small aquarium fish Xiphophorus, an animal model for melanoma formation with well-established tumor genetics (SCHARTL 1995 ), and in both pufferfishes F. rubripes and T. nigroviridis. These results, added to the recent report of the presence of two mitf genes in zebrafish (LISTER et al. 2001 ), show the presence of duplicate mitf genes in the teleost fish lineage for at least 100–120 million years (minimal age of divergence between the zebrafish and the Xiphophorus/pufferfish lineages). This is supported by the fact that fish MITF-m and MITF-b proteins form two distinct monophyletic groups. According to the analysis of K_s/K_a ratios in four different fish species, the duplicate mitf genes are not pseudogenes but rather evolved under purifying selection.

While the phylogenetic analysis using maximum parsimony supports a duplication of the mitf genes within the fish lineage after its separation from the bird/mammalian lineage, the neighbor joining and maximum-likelihood methods did not allow resolving the relative position of both fish MITF protein groups. A hypothesis alternative to fish-specific duplication would imply that the mitf gene was duplicated before the separation between the bird/mammalian and fish lineages. While both copies would have been maintained in teleosts, one duplicate would have been lost in higher vertebrates. The remaining copy should be then more related to either the mitf-b or the mitf-m fish gene.

Comparative structural analysis of vertebrate mitf cDNA and genomic sequences showed that intron 3 (according to the mammalian nomenclature) is present in birds and mammals but not in teleosts, while intron 5b is found in both mitf-b and mitf-m genes in fish but not in the unique mitf gene of higher vertebrates. Since both introns are absent from the mitf-related tfe3 genes, they have probably been recruited during evolution of the mitf genes in vertebrates. If we consider the scenario of duplication of mitf before separation of fish and birds/mammals, intron 5b should have been present in mitf before duplication (since intron 5b is present in both mitf-m and mitf-b in fish) and should have been lost in the remaining copy of higher vertebrates. On the other hand, if we consider the hypothesis of a fish-specific duplication of mitf, intron 5b might have been recruited within the fish lineage before the duplication event.

The fish-specific duplication hypothesis, suggested by the maximum parsimony analysis, implies only one event of gene duplication. Although we cannot exclude that the mitf gene was duplicated before the separation between fish and birds/mammals, this alternative hypothesis would involve additional events (loss of one mitf duplicate and loss of intron 5b in the remaining copy in higher vertebrates). Hence, the fish-specific duplication hypothesis appears to be the most parsimonious one. Absence of strong support for this hypothesis in the neighbor joining and maximum-likelihood analyses might be explained by modified evolutionary rates due to relaxed selective constraints after duplication, as suggested by the analysis of the K_s/K_a ratios.

The fish-specific duplication hypothesis fits accumulating evidence that during the course of evolution teleost fishes underwent a whole genome duplication since they separated from the last common ancestor that they shared with amphibians, reptiles, birds, and mammals (AMORES et al. 1998 ; GATES et al. 1999 ; MEYER and SCHARTL 1999 ; TAYLOR et al. 2001 ; VAN DE PEER et al. 2001 ; but see ROBINSON-RECHAVI et al. 2001 ). As a consequence in many instances fish have two similar genes where mammals have only one. This genome duplication might have arisen as early as 300–450 million years ago (TAYLOR et al. 2001 ; VAN DE PEER et al. 2001 ). The divergence between mitf-m and mitf-b nucleic acid and protein sequences, even at synonymous sites, is always higher than the divergence between chicken and mammal sequences (divergence 300 million years ago). This is consistent with an ancient event of duplication of the mitf genes. It has been estimated that 20% of the duplicate genes, particularly genes encoding DNA-binding proteins (like MITF), have been maintained in teleost genomes by neofunctionalization of a duplicate and/or by subfunctionalization (FORCE et al. 1999 ; POSTLETHWAIT et al. 2000 ; VAN DE PEER et al. 2001 ). This last phenomenon could arise as a consequence of the differential decay of specific sequences in each gene copy and the subsequent need for the presence of both duplicates with complementary activities to achieve the function(s) of the original unique gene (FORCE et al. 1999 ). We present evidence for a subfunctionalization where two alternative transcripts of the same gene with different functions have been replaced by two distinct genes. This is associated with the degeneration of at least one alternative exon: the 1m exon specific for the m-form of MITF in higher vertebrates shows a degenerated coding sequence in the mitf-b gene of both pufferfishes. This might be associated, at least in Tetraodon, with the degeneration of the flanking splicing site. As observed in other cases of subfunctionalization of duplicate fish genes reported to date (for examples, see CHIANG et al. 2001A , CHIANG et al. 2001B ; SERLUCA et al. 2001 ), fish mitf-m and mitf-b show different expression patterns during embryonic development in zebrafish (LISTER et al. 2001 ) and in adult tissues in Xiphophorus. Hence, subfunctionalization of mitf genes in fish is also probably associated with the degeneration of specific promoters and/or regulatory sequences. The order of degenerative events remains to be determined: mutations may have first altered the coding sequence and/or the splicing site of an alternative exon in duplicate A (for example exon 1m in mitf-b) but not in duplicate B. In this way the selection for functionality on the promoter driving the expression of the transcript containing the alternative exon in duplicate A would have been abolished. Alternatively, alteration of promoter or specific regulatory sequences necessary for the expression of an alternative transcript in duplicate A but not in duplicate B would allow accumulation in duplicate A of nonconservative mutations in sequences that are specific for this alternative transcript.

We cannot exclude that the duplication of the mitf genes was a single gene duplication event that happened very early in the teleost lineage. More data are needed from other fish genomes to understand the genomic mechanism and the evolutionary significance underlying the duplication of the mitf gene in fish genomes.

	FOOTNOTES

Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AF475090 and AF475091.
¹ These authors contributed equally to this work.
² Present address: Molecular Hematology, University of Frankfurt Medical School, D-60590 Frankfurt, Germany.

	ACKNOWLEDGMENTS

We thank U. Hornung and E. Geissinger for RNAs and Jim A. Lister for sharing the zebrafish nacre sequence with us prior to publication. We are grateful to Cécile Fischer, Alain Bernot, Jean Weissenbach, and the other members of the Tetraodon Genome project (Genoscope, Evry, France), as well as to the International Fugu Sequencing Consortium, for making sequencing data available to the public. Fugu data have been provided freely by the Fugu Genome Consortium for use in this publication only. This work was supported by grants to M.S. supplied by the Deutsche Forschungsgemeinschaft through SFB 465 ("Entwicklung und Manipulation pluripotenter Zellen") and the Fonds der Chemischen Industrie. J.-N.V. is supported by the BioFuture program of the German Bundesministerium für Bildung und Forschung (BMBF).

Manuscript received December 13, 2001; Accepted for publication February 22, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AMAE, S., N. FUSE, K. YASUMOTO, S. SATO, and I. YAJIMA et al., 1998  Identification of a novel isoform of microphthalmia-associated transcription factor that is enriched in retinal pigment epithelium. Biochem. Biophys. Res. Commun. 24:710-715.

AMORES, A., A. FORCE, Y. L. YAN, L. JOLY, and C. AMEMIYA et al., 1998  Zebrafish hox clusters and vertebrate genome evolution. Science 282:1711-1714[Abstract/Free Full Text].

BAUDLER, M., J. DUSCHL, C. WINKLER, M. SCHARTL, and J. ALTSCHMIED, 1997  Activation of transcription of the melanoma inducing Xmrk oncogene by a GC-box element. J. Biol. Chem. 272:131-137[Abstract/Free Full Text].

BURSET, M., I. A. SELEDTSOV, and V. V. SOLOVYEV, 2001  SpliceDB: database of canonical and non-canonical mammalian splice sites. Nucleic Acids Res. 29:255-259[Abstract/Free Full Text].

CHIANG, E. F., C. I. PAI, M. WYATT, Y. L. YAN, and J. POSTLETHWAIT et al., 2001a  Two sox9 genes on duplicated zebrafish chromosomes: expression of similar transcription activators in distinct sites. Dev. Biol. 231:149-163[Medline].

CHIANG, E. F., Y. L. YAN, Y. GUIGUEN, J. POSTLETHWAIT, and B. CHUNG, 2001b  Two Cyp19 (P450 aromatase) genes on duplicated zebrafish chromosomes are expressed in ovary or brain. Mol. Biol. Evol. 18:542-550[Abstract/Free Full Text].

CROLLIUS, H. R., O. JAILLON, C. DASILVA, C. OZOUF-COSTAZ, and C. FIZAMES et al., 2000  Characterization and repeat analysis of the compact genome of the freshwater pufferfish Tetraodon nigroviridis.. Genome Res. 10:939-949[Abstract/Free Full Text].

ELGAR, G., R. SANDFORD, S. APARICIO, A. MACRAE, and B. VENKATESH et al., 1996  Small is beautiful: comparative genomics with the pufferfish (Fugu rubripes). Trends Genet. 12:145-150[Medline].

ELGAR, G., M. S. CLARK, S. MEEK, S. SMITH, and S. WARNER et al., 1999  Generation and analysis of 25 Mb of genomic DNA from the pufferfish Fugu rubripes by sequence scanning. Genome Res. 9:960-971[Abstract/Free Full Text].

FISCHER, C., C. OZOUF-COSTAZ, H. ROEST CROLLIUS, C. DASILVA, and O. JAILLON et al., 2000  Karyotype and chromosome location of characteristic tandem repeats in the pufferfish Tetraodon nigroviridis.. Cytogenet. Cell Genet. 88:50-55[Medline].

FORCE, A., M. LYNCH, F. B. PICKETT, A. AMORES, and Y. L. YAN et al., 1999  Preservation of duplicate genes by complementary, degenerative mutations. Genetics 151:1531-1545[Abstract/Free Full Text].

GATES, M. A., L. KIM, E. S. EGAN, T. CARDOZO, and H. I. SIROTKIN et al., 1999  A genetic linkage map for zebrafish: comparative analysis and localization of genes and expressed sequences. Genome Res. 9:334-347[Abstract/Free Full Text].

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].

HODGKINSON, C. A., K. J. MOORE, A. NAKAYAMA, E. STEINGRIMSSON, 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].

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].

ISRAEL, D. I., 1993  A PCR-based method for high stringency screening of DNA libraries. Nucleic Acids Res. 21:2627-2631[Abstract/Free Full Text].

KALLMAN, K. D., 1975 The platyfish, Xiphophorus maculatus, pp. 81–132 in Vertebrates of Genetic Interest, edited by R. C. KING. Plenum Press, New York.

KING, R., K. N. WEILBAECHER, G. MCGILL, E. COOLEY, and M. MIHM et al., 1999  Microphthalmia transcription factor. A sensitive and specific melanocyte marker for melanoma diagnosis. Am. J. Pathol. 155:731-738[Abstract/Free Full Text].

KUHN, C., U. VIELKIND, and F. ANDERS, 1979  Cell cultures derived from embryos and melanoma of poeciliid fish. In Vitro 15:537-544[Medline].

LISTER, J. A., C. P. ROBERTSON, T. LEPAGE, S. L. JOHNSON, and D. W. RAIBLE, 1999  nacre encodes a zebrafish microphthalmia-related protein that regulates neural-crest-derived pigment cell fate. Development 126:3757-3767[Abstract].

LISTER, J. A., J. CLOSE, and D. W. RAIBLE, 2001  Duplicate mitf genes in zebrafish: complementary expression and conservation of melanogenic potential. Dev. Biol. 237:333-344[Medline].

LYNCH, M. and J. S. CONERY, 2000  The evolutionary fate and consequences of duplicate genes. Science 290:1151-1155[Abstract/Free Full Text].

MEYER, A. and M. SCHARTL, 1999  Gene and genome duplications in vertebrates: the one-to-four (-to-eight in fish) rule and the evolution of novel gene functions. Curr. Opin. Cell Biol. 11:699-704[Medline].

OHNO, S., 1970 Evolution by Gene Duplication. Springer-Verlag, New York.

POSTLETHWAIT, J. H., I. G. WOODS, P. NGO-HAZELETT, Y. L. YAN, and P. D. KELLY et al., 2000  Zebrafish comparative genomics and the origins of vertebrate chromosomes. Genome Res. 10:1890-1902[Abstract/Free Full Text].

REHLI, M., N. DEN ELZEN, A. I. CASSADY, M. C. OSTROWSKI, and D. A. HUME, 1999  Cloning and characterization of the murine genes for bHLH-ZIP transcription factors TFEC and TFEB reveal a common gene organization for all MiT subfamily members. Genomics 56:111-120[Medline].

ROBINSON-RECHAVI, M., O. MARCHAND, H. ESCRIVA, and V. LAUDET, 2001  An ancestral whole-genome duplication may not have been responsible for the abundance of duplicated fish genes. Curr. Biol. 11:458-459.

SAITOU, N. and M. NEI, 1987  The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406-425[Abstract].

SCHARTL, M., 1995  Platyfish and swordtails: a genetic system for the analysis of molecular mechanisms in tumor formation. Trends Genet. 11:185-189[Medline].

SCHARTL, M., B. WILDE, and J. PARZEFALL, 1995  Evolutionary origin of a parthenoform, the Amazon Molly Poecilia formosa, on the basis of a molecular genealogy. Evolution 49:827-835.

SERLUCA, F. C., A. SIDOW, J. D. MABLY, and M. C. FISHMAN, 2001  Partitioning of tissue expression accompanies multiple duplications of the Na+/K+ ATPase alpha subunit gene. Genome Res. 11:1625-1631[Abstract/Free Full Text].

SMITH, S. D., P. M. KELLEY, J. B. KENYON, and D. HOOVER, 2000  Tietz syndrome (hypopigmentation/deafness) caused by mutation of MITF. J. Med. Genet. 37:446-448[Abstract/Free Full Text].

STEINGRIMSSON, E., K. J. MOORE, M. L. LAMOREUX, A. R. FERRE-D'AMARE, 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].

STRIMMER, K. and A. VON HAESELER, 1996  Quartet puzzling: a quartet maximum likelihood method for reconstructing tree topologies. Mol. Biol. Evol. 13:964-969.

SWOFFORD, D. L., 1989 PAUP: Phylogenetic Analysis Using Parsimony. Smithsonian Institution, Washington, DC.

TAKEDA, K., C. TAKEMOTO, I. KOBAYASHI, A. WATANABE, and Y. NOBUKUNI et al., 2000  Ser298 of MITF, a mutation site in Waardenburg syndrome type 2, is a phosphorylation site with functional significance. Hum. Mol. Genet. 9:125-132[Abstract/Free Full Text].

TASSABEHJI, M., V. E. NEWTON, and A. P. READ, 1994  Waardenburg syndrome type 2 caused by mutations in the human microphthalmia (MITF) gene. Nat. Genet. 8:251-255[Medline].

TAYLOR, J. S., Y. VAN DE PEER, I. BRAASCH, and A. MEYER, 2001  Comparative genomics provides evidence for an ancient genome duplication event in fish. Philos. Trans. R. Soc. Lond. B Biol. Sci. 356:1661-1679[Medline].

UDONO, T., K. YASUMOTO, K. TAKEDA, S. AMAE, and K. WATANABE et al., 2000  Structural organization of the human microphthalmia-associated transcription factor gene containing four alternative promoters. Biochim. Biophys. Acta 1491:205-219[Medline].

VACHTENHEIM, J. and H. NOVOTNA, 1999  Expression of genes for microphthalmia isoforms, Pax3 and MSG1, in human melanomas. Cell. Mol. Biol. 45:1075-1082.

VAN DE PEER, Y., J. S. TAYLOR, I. BRAASCH, and A. MEYER, 2001  The ghost of selection past: rates of evolution and functional divergence of anciently duplicated genes. J. Mol. Evol. 53:436-446[Medline].

WAKAMATSU, Y., 1981  Establishment of a cell line from the platyfish-swordtail hybrid melanoma. Cancer Res. 41:679-680[Abstract/Free Full Text].

WITTBRODT, J., A. MEYER, and M. SCHARTL, 1998  More genes in fish? Bioessays 20:511-515.

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., S. AMAE, T. UDONO, N. FUSE, and K. TAKEDA et al., 1998  A big gene linked to small eyes encodes multiple Mitf isoforms: many promoters make light work. Pigm. Cell Res. 11:329-336[Medline].

This article has been cited by other articles:

				M. Semon and K. H. Wolfe Preferential subfunctionalization of slow-evolving genes after allopolyploidization in Xenopus laevis PNAS, June 17, 2008; 105(24): 8333 - 8338. [Abstract] [Full Text] [PDF]

				I. Braasch, W. Salzburger, and A. Meyer Asymmetric Evolution in Two Fish-Specifically Duplicated Receptor Tyrosine Kinase Paralogons Involved in Teleost Coloration Mol. Biol. Evol., June 1, 2006; 23(6): 1192 - 1202. [Abstract] [Full Text] [PDF]

				Z. Su, J. Wang, J. Yu, X. Huang, and X. Gu Evolution of alternative splicing after gene duplication Genome Res., February 1, 2006; 16(2): 182 - 189. [Abstract] [Full Text] [PDF]

				F. S. J. de Souza, V. F. Bumaschny, M. J. Low, and M. Rubinstein Subfunctionalization of Expression and Peptide Domains Following the Ancient Duplication of the Proopiomelanocortin Gene in Teleost Fishes Mol. Biol. Evol., December 1, 2005; 22(12): 2417 - 2427. [Abstract] [Full Text] [PDF]

				T. R. Pacheco, A. Q. Gomes, N. L. Barbosa-Morais, V. Benes, W. Ansorge, M. Wollerton, C. W. Smith, J. Valcarcel, and M. Carmo-Fonseca Diversity of Vertebrate Splicing Factor U2AF35: IDENTIFICATION OF ALTERNATIVELY SPLICED U2AF1 mRNAs J. Biol. Chem., June 25, 2004; 279(26): 27039 - 27049. [Abstract] [Full Text] [PDF]

				J. H. Hallsson, B. S. Haflidadottir, C. Stivers, W. Odenwald, H. Arnheiter, F. Pignoni, and E. Steingrimsson The Basic Helix-Loop-Helix Leucine Zipper Transcription Factor Mitf Is Conserved in Drosophila and Functions in Eye Development Genetics, May 1, 2004; 167(1): 233 - 241. [Abstract] [Full Text] [PDF]

				S. Rombauts, K. Florquin, M. Lescot, K. Marchal, P. Rouze, and Y. Van de Peer Computational Approaches to Identify Promoters and cis-Regulatory Elements in Plant Genomes Plant Physiology, July 1, 2003; 132(3): 1162 - 1176. [Abstract] [Full Text] [PDF]

				C. Winkler, M. Schafer, J. Duschl, M. Schartl, and J.-N. Volff Functional Divergence of Two Zebrafish Midkine Growth Factors Following Fish-Specific Gene Duplication Genome Res., June 1, 2003; 13(6): 1067 - 1081. [Abstract] [Full Text] [PDF]

				A. Locascio, M. Manzanares, M. J. Blanco, and M. A. Nieto Modularity and reshuffling of Snail and Slug expression during vertebrate evolution PNAS, December 24, 2002; 99(26): 16841 - 16846. [Abstract] [Full Text] [PDF]