The non-long terminal repeat retrotransposons Rex1 and Rex3 were identified in 13 species of Antarctic fishes from five families of the suborder Notothenioidei. Partial reverse transcriptase gene sequences were characterized for Notothenia coriiceps, Trematomus newnesi and Dissostichus mawsoni (Nototheniidae), and Gymnodraco acuticeps (Bathydraconidae). Rex1 and Rex3 both formed a notothenioid-specific monophyletic group compared to the corresponding elements from other fishes. They globally evolved under purifying selection, showing their activity during notothenioid evolution. Fluorescence in situ hybridization analysis of the chromosomal distribution of Rex1 and Rex3 was performed for several notothenioid fish species. Rex1 was generally less abundant than Rex3, which was widely scattered on the chromosomes with more intense hybridization patterns in some specific zones. Particularly, Rex3 accumulated in Chionodraco hamatus in pericentromeric areas, short arms of some pairs as well as in an intercalary band in the long arm of the Y chromosome similarly to a previously described DNA transposon. Such pattern similarities Suggest the presence of autosomal and gonosomal regions of preferential accumulation for different types of repeated elements in notothenioid genomes. To the best of our knowledge, this report is the first description and analysis of retrotransposable elements in Antarctic fish genomes.

The Xmrk oncogene involved in melanoma formation in the fish Xiphophorus was formed relatively recently by duplication of the epidermal growth factor co-orthologue egfrb. In the platyfish X. maculatus, Xmrk is located close to the major sex-determining locus in a subtelomeric region of the X and Y sex chromosomes that frequently undergoes duplications and other rearrangements. This region accumulates repetitive sequences: more than 80% of the 33-kb region 3′ of Xmrk is constituted by retrotransposable elements. The high degree of nucleotide identity between X- and Y-linked sequences and the rarity of gonosome-specific rearrangements indicated that the instability observed was not a manifestation of gonosome-specific degeneration. Seven other duplicated genes were found, all corresponding, in contrast to Xmrk, to pseudogenes (nonfunctionalization). Functional persistence of Xmrk in a highly unstable region in divergent Xiphophorus species suggests a beneficial function under certain conditions for this dispensable and potentially injurious gene.

In contrast to the situation observed in mammals and birds, a switching between different sex determination systems frequently occurred during the evolution of the teleost fish lineage. This might be due to a frequent turnover of sex-determining signals at the top of the sex determination cascade (master sex-determining genes in the case of genetic sex determination). Alternatively, different variations of a same master gene might decide the sex of individuals in different sex determination systems. In the medaka Oryzias latipes, a Y-specific copy of the putative transcription factor gene dmrt1 very likely corresponds to the master sex-determining gene inducing male formation [Nature 417 (2002) 559; Proc. Natl. Acad. Sci. U. S. A. 99 (2002) 11778]. This gene, alternately called dmY and dmrt1Y, has been formed by duplication of the autosomal dmrt1. In order to determine if an orthologue of dmrt1Y was also located in the sex determination region of the related platyfish Xiphophorus maculatus, Southern blot analysis was performed on genomic DNA from XY and YY males and WY and XX females using both medaka and human dmrt1 cDNA as probes. Using different restriction enzymes, no evidence for sex-specific dmrt-containing genomic fragments could be found in the platyfish. No dmrt gene could be detected by low-stringency Southern blot analysis of genomic inserts from 60 bacterial artificial chromosome (BAC) clones linked to the sex-determining locus on the X and Y chromosomes. Six different groups of BAC clones containing dmrt genes were isolated from a platyfish genomic library. X. maculatus dmrt1 gene and cDNA were characterised. In adult, expression of dmrt1 was detected only in testis. Phylogenetic analysis indicated that the duplication of dmrt1 that led to the formation of dmY/dmrt1Y within the medaka lineage clearly occurred after its divergence from the platyfish lineage. Hence, a dmY/dmrt1Y orthologue generated by this particular event of duplication does not correspond to the master sex-determining gene in the platyfish.

The fish retrotransposable element Zebulon encodes a reverse transcriptase and a carboxy-terminal restriction enzyme-like endonuclease, and is related phylogenetically to site-specific non-LTR retrotransposons from nematodes. Zebulon was detected in the pufferfishes Tetraodon nigroviridis and Takifugu rubripes , as well as in the zebrafish Danio rerio . Structural analysis suggested that Zebulon , in contrast to most non-LTR retrotransposons, might be able to retrotranspose as a partial tandem array. Zebulon was active relatively recently in the compact genome of T. nigroviridis , in which it contributed to the extension of intergenic and intronic sequences, and possibly to the formation of genomic rearrangements. Accumulation of Zebulon together with other retrotransposons was observed in some heterochromatic chromosomal regions of the genome of T. nigroviridis that might serve as reservoirs for active elements. Hence, pufferfish compact genomes are not evolutionarily inert and contain active retrotransposons, suggesting the presence of mechanisms allowing accumulation of retrotransposable elements in heterochromatin, but minimizing their impact on euchromatic regions. Homologous recombination between partial tandem sequences eliminating active copies of Zebulon and reducing the size of insertions in intronic and intragenic regions might represent such a mechanism.

A high level of genetic variability affects pigmentation, melanoma formation, sex determination and sexual maturity in the fish Xiphophorus. Gene loci encoding these traits have been mapped to a subtelomeric region of the sex chromosomes that very frequently undergoes DNA rearrangements including duplications, amplifications, deletions and transpositions. This conspicuous genomic fluidity might generate new sex-linked gene regulations and functions, and be involved in the initiation of gonosome differentiation around the major sex-determining locus in Xiphophorus.

Despite the major importance of sex determination in aquaculture, no master sex-determining gene has been identified so far in teleost fish. In the platyfish Xiphophorus maculatus, this master gene is flanked by two receptor tyrosine kinase genes, the Xmrk oncogene responsible for melanoma formation in some Xiphophorus interspecific hybrids, and its proto-oncogenic counterpart. Both Xmrk genes, which have already been characterised at the molecular level, delimit a region of about 1 Mb that contains other gene loci involved in sexual maturity, pigmentation and melanoma formation. We have constructed a genomic bacterial artificial chromosome (BAC) library of X. maculatus with a tenfold coverage of the haploid genome and walked on both X and Y sex chromosomes starting from both Xmrk genes. This led to the assembly of BAC contigs from the sex-determining region covering approximately 950 kb of the X and 750 kb of the Y chromosome. To our knowledge, these are the largest contigs reported so far for sex chromosomes in fish. Molecular analysis suggests that the sex-determining region of X. maculatus frequently undergoes retrotranspositions and other kinds of rearrangements. This genomic plasticity might be related to the high genetic variability observed in Xiphophorus for sex determination, sexual maturity, pigmentation and melanoma formation, which are encoded by gene loci located in the sex-determining region.

The fish non–long-terminal-repeat (non-LTR) retrotransposon Rex3 has recently been isolated from the platyfish Xiphophorus maculatus (Volff et al. 1999). Complete versions of Rex3 encode a reverse transcriptase (RT) and an apurinic/apyrimidinic endonuclease (fig. 1). Rex3 belongs to the RTE family of non-LTR retrotransposons (Malik and Eickbush 1998; Volff et al. 1999). From all autonomous fish retrotransposons reported to date, Rex3 has the widest distribution observed in teleosts and is present in fish species having diverged 150–200 MYA. We report here a large PCRand Southern blot–based survey of Rex3 evolution including 21 representative teleost species (fig. 1) and 115 Rex3 partial reverse transcriptase sequences (fig. 2). The species chosen include a panel of economically important fishes (salmon Salmo salar, trout Oncorhynchus mykiss, carp Cyprinus carpio, sturgeon Acipenser sturio, mandarin fish Siniperca chuatsi) and several small aquarium teleosts used as models for developmental biology, cancer research and evolutionary studies (zebrafish Danio rerio, medakafish Oryzias latipes, platyfish Xiphophorus maculatus, and other Poeciliidae), as well as the genome project fish, the Japanese pufferfish Fugu rubripes. Rex3 partial RT-encoding sequences could be amplified by PCR from the majority of fish species with at least one of the nine primer combinations tested, but not from the more divergent nonteleost Acipenser sturio (sturgeon) and not from both salmonid species tested (O. mykiss and S. salar) (fig. 1). Using cloned Rex3 probes from X. maculatus and from Anguilla anguilla (European eel) in Southern blot experiments, no specific signal could be detected in A. sturio, O. mykiss, and S. salar even under low-stringency conditions (not shown), confirming the PCR analysis results. According to classical morphological and molecular fish phylogenies (fig. 1; Nelson 1994; Forey et al. 1996; Orti and Meyer 1996), the distribution of Rex3 in teleosts is discontinuous. Hence, Rex3 was lost (or, alternatively, diverged extremely rapidly) in the Oncorhynchus/Salmo lineage after its divergence from the Esox lineage. There are about 1,000 Rex3 copies in the haploid genome of Xiphophorus species (Volff et al. 1999). All other Poeciliidae species included in this study and the related Fundulus displayed a high level of Rex3 reiteration as well (data not shown). Rex3 is present in high copy numbers in the genomes of O. latipes, Oreochrom-

Jule is the second complete long-terminal-repeat (LTR) Ty3/Gypsy retrotransposon identified to date in vertebrates. Jule, first isolated from the poeciliid fish Xiphophorus maculatus, is 4.8 kb in length, is flanked by two 202-bp LTRs, and encodes Gag (structural core protein) and Pol (protease, reverse transcriptase, RNase H, and integrase, in that order) but no envelope. There are three to four copies of Jule per haploid genome in X. maculatus. Two of them are located in a subtelomeric region of the sex chromosomes, where they are associated with the Xmrk receptor tyrosine kinase genes, of which oncogenic versions are responsible for the formation of hereditary melanoma in Xiphophorus. One almost intact copy of Jule was found in the first intron of the X-chromosomal allele of the Xmrk proto-oncogene, and a second, more corrupted copy is present only 56 nt downstream of the polyadenylation signal of the Xmrk oncogene. Jule-related elements were detected by Southern blot hybridization with less than 10 copies per haploid genome in numerous other poeciliids, as well as in more divergent fishes, including the medakafish Oryzias latipes and the tilapia Oreochromis niloticus. Database searches also identified Jule-related sequences in the zebrafish Danio rerio and in both genome project pufferfishes, Fugu rubripes and Tetraodon nigroviridis. Phylogenetic analysis revealed that Jule is the first member of the Mag family of Ty3/Gypsy retrotransposons described to date in vertebrates. This family includes the silkworm Mag and sea urchin SURL retrotransposons, as well as sequences from the nematode Caenorhabditis elegans. Additional related elements were identified in the genomes of the malaria mosquito Anopheles gambiae and the nematode Ascaris lumbricoides. Phylogeny of Mag-related elements suggested that the Mag family of retrotransposons is polyphyletic and is constituted of several ancient lineages that diverged before their host genomes more than 600 MYA.

Long-terminal-repeat (LTR) retrotransposons from the Ty3 / Gypsy superfamily have been detected in vari-ous eukaryotic taxa, including some vertebrate lineages bony amphibians, Nevertheless, molecular and database screenings failed to detect such elements in the genome of mammals. Considering the of sequence on mammalian genomes and transcriptomes, this suggested that Ty3 / Gypsy retrotransposons either have been or are present at an extremely low copy number in mammals.Byexamination of we / -like similarities the of / Gypsy retrotransposons /

All autonomous non-long terminal repeat (non-LTR) retrotransposons reported to date in vertebrates encode an apurinic/apyrimidinic endonuclease-like enzyme necessary for target sequence cleavage and subsequent target-primed reverse transcription. We describe here vertebrate non-LTR retrotransposons encoding another type of endonuclease more related to type IIS restriction enzymes. Such retrotransposons have been detected until now only in trypanosomes, nematodes, and arthropods. The retrotransposon Rex6 was identified in the genome of several teleost fish including Xiphophorus maculatus (platyfish), Oryzias latipes (medakafish), Oreochromis niloticus (Nile tilapia), and Fugu rubripes (Japanese pufferfish). Rex6 encodes a reverse transcriptase and a putative restriction enzyme-like endonuclease and is a member of the R4 family of non-LTR retrotransposons containing the Dong and R4 elements found in nematodes and insects. Rex6 was active in many species during teleost evolution and underwent several bursts of retrotransposition (some of them being relatively recent) leading to a high copy number of Rex6 in the genome of numerous fish. Extremely truncated Rex6-related sequences were detected by database screening in reptiles, including the snake Trimeresus flavoviridis and the lizard Anolis carolinensis, but not in sequences from the human genome project, suggesting that this element might have been lost from certain vertebrate lineages.

Melanoma formation in certain interspecific hybrids of the genus Xiphophorus (Teleostei: Poeciliidae) is associated with the overexpression of the Xmrk receptor tyrosine kinase oncogene. The Xmrk oncogene arose by duplication of the pre-existing Xmrk protooncogene in a highly unstable subtelomeric region of the X and Y sex chromosomes undergoing frequent rearrangements including duplications, deletions, amplifications, and transpositions. Some of these rearrangements are likely to be responsible for the overexpression of the Xmrk oncogene in melanoma. The oncogene itself is very unstable in Xiphophorus and is frequently removed by deletion or disrupted by transposable elements. The Xmrk oncogene region displays a high concentration of retroelements not observed in the corresponding Xmrk protooncogene region. Particularly, a retrovirus long terminal repeat-like sequence was amplified in the proximity of the Xmrk oncogene. Additional genes, some of them also duplicated copies, were detected in this region and might be involved in modulating the melanoma's phenotype.

Small aquarium fishes become increasingly important in the study of normal vertebrate development and disease. Differential DNA methylation might play a role in these processes. In the teleost Xiphophorus, a well-established animal model for melanoma formation, tumour-specific hypomethylation of the melanoma-inducing gene ONC-Xmrk has been observed. We have isolated a cDNA for the DNA-(cytosine-5)-methyltransferase XDNMT-1 from this organism, which encodes the first full-length protein from a fish species. Linkage analysis showed that Xdnmt-1 is different from the Xiphophorus tumour suppressor R, which is involved in the transcriptional repression of the ONC-Xmrk melanoma oncogene in healthy fish. As methylation has been implicated in the regulation of ONC-Xmrk expression, XDNMT-1 might play a role by acting up- or downstream of R. Expression analysis demonstrated that the Xdnmt-1 transcript is present in all adult tissues and cell lines tested. However, developing embryos show a spatially and temporally regulated expression pattern suggesting that the enzyme might play a role during development in fish.

The platyfish (Xiphophorus maculatus), in which sex chromosomes are evident from stable and predictable inheritance of sex, is one of the best-studied lower vertebrates with respect to sex determination. In order to identify the structural equivalent for this in the karyotype, which does not contain heteromorphic pairs of chromosomes, two sex-linked molecular probes were used for fluorescent in situ hybridization analysis. One probe, derived from the melanoma oncogene locus ONC-Xmrk, stained both the X and the Y chromosome. This cytogenetic analysis mapped the sex-determining locus to the subtelomeric region of a medium-sized telocentric chromosome. Another probe, a repetitive element (XIR), specifically labeled the Y chromosome in metaphase spreads and in interphase nuclei. The sex chromosomes of X. maculatus can be considered to be at an early stage of evolution of gonosomes. Expansion of the XIR repeat is obviously one of the earliest of the molecular events that lead to divergence of the Y chromosome and recombinational isolation of the sex-determining locus.

Rex1, together with the related BABAR: elements, represents a new family of non-long-terminal-repeat (non-LTR) retrotransposons from fish, which might be related to the CR1 clade of LINE elements. Rex1/BABAR: retrotransposons encode a reverse transcriptase and an apurinic/apyrimidinic endonuclease, which is very frequently removed by incomplete reverse transcription. Different Rex1 elements show a conserved terminal 3' untranslated region followed by oligonucleotide tandem repeats of variable size and sequence. Phylogenetic analysis revealed that Rex1 retrotransposons were frequently active during fish evolution. They formed multiple ancient lineages, which underwent several independent and recent bursts of retrotransposition and invaded fish genomes with varying success (from <5 to 500 copies per haploid genome). At least three of these ancient Rex1 lineages were detected within the genome of poeciliids. One lineage is absent from some poeciliids but underwent successive rounds of retrotransposition in others, thereby increasing its copy number from <10 to about 200. At least three ancient Rex1 lineages were also detected in the genome project fish Fugu rubripes. Rex1 distribution within one of its major lineages is discontinuous: Rex1 was found in all Acanthopterygii (common ancestor in the main teleost lineage approximately 90 MYA) and in both European and Japanese eels (divergence from the main teleost lineage about 180 MYA) but not in trout, pike, carp, and zebrafish (divergence 100-120 MYA). This might either result from frequent loss or rapid divergence of Rex1 elements specifically in some fish lineages or represent one of the very rare examples of horizontal transfer of non-LTR retrotransposons. This analysis highlights the dynamics and complexity of retrotransposon evolution and the variability of the impact of retrotransposons on vertebrate genomes.

Rex3, the first reverse transcriptase (RT)-encoding retrotransposon isolated from the melanoma fish model Xiphophorus, is a non-long-terminal-repeat element related to the RTE family. The essential features of Rex3 are (1) an endonuclease and a reverse transcriptase, (2) 5' truncations of most of the copies, (3) a 3' tail consisting of tandem repeats of the sequence GATG, and (4) short target site sequence duplications of variable length. Compilation of Rex3 sequences from the pufferfish genome project suggested that, as observed for other members of the RTE family, no additional large open reading frame was present upstream of the endonuclease/reverse transcriptase open reading frame. There are about a thousand copies of Rex3 in the haploid genome of Xiphophorus, some of them probably resulting from recent retrotransposition events. Rex3 RNA was detected by RT-PCR in melanoma and in nontumorous tissues, as well as in melanoma-derived and embryonic cell lines. Rex3 is present in a broad panel of teleost species and was found in the promoter region and in introns of various genes. To our knowledge, Rex3 is the first autonomous retrotransposon described to date which is widespread in teleosts. This wide distribution and occasional association with coding sequences may confer on Rex3 a predisposition to play a role in genome evolution in teleosts.