DNA-based transposable elements

Four major classes of transposable elements are known: viral family retrotransposons, long interspersed nuclear elements (LINEs), short interspersed nuclear elements (SINEs) and DNA-based transposable elements. The last class, also called terminal-inverted-repeat elements, is different from the others in that elements move directly from DNA to DNA, that is, without forming RNA intermediates. The enzyme that is central or solely required for their transposition reaction is called a transposase and is considered to act mainly on their terminal and/or subterminal regions. Transposable elements are present in multicopies in host genomes and, in many cases, albeit not all, autonomous copies exist including a gene for their own transposase. Copies that do not carry the transposase gene or carry an imperfect gene are called non-autonomous copies, these being thought to be produced from autonomous copies by errors during transposition or subsequent mutational sequence changes. As long as the transposase cannot distinguish autonomous from nonautonomous copies, which so far appears to be true, the evolutionary scenario they would be expected to follow is first increase in frequency of nonautonomous elements, subsequent loss of transposition activity, and ultimate extinction from the host genome (cf. Lohe et al., 1995; Hartl et al., 1997).

Several DNA-based transposable elements have been identified in vertebrates, analogous to many other organisms. However, autonomous copies are exceedingly rare, which suggests the situation where most vertebrate DNA-based elements are in the final or nearly the final stage of the scenario described above. Inferences from the results of the human genome sequencing project are consistent with this view (cf. International Human Genome Sequencing Consortium 2001). For studying transposition mechanisms, as well as evolution, the few elements in teleost fishes demonstrated or suggested to be still active offer unique materials. The authors have been working on identification and examination of active elements of the medaka fish O. latipes. The present review describes results so far obtained on various aspects of medaka fish transposable elements, such as structure, transposition mechanisms, evolution and application to biotechnology.

Medaka fish

The medaka fish is a freshwater teleost native to Eastern Asia, including China, Japan and Korea. It is, together with zebrafish, a superior model animal for vertebrate genetics and developmental biology because of its small genome size (800 Mb), small body size (4 cm), short generation time (3 months), high reproductive rate (10 to 20 eggs per female every day) and transparent egg chorion (Iwamatsu 1997; Ishikawa 2000; Packer 2001; also see Medakafish Home-page at  http://biol1.bio.nagoya-u.ac.jp:8000/). In addition, a strain having completely transparent bodies has recently been generated by accumulating deficient alleles at several pigmentation loci (Wakamatsu et al., 2001). Another remarkable advantage is the in-depth knowledge of closely related fishes. Fourteen species are so far identified in the genus Oryzias and their detailed phylogenetic relationship is known (Naruse et al., 1993; Naruse, 1996; Koga et al., 2000). This provides a strong basis for evolutionary studies comparative to that in the melanogaster species subgroup of Drosophila. A great amount of genetic variation is known to exist not only among but also within species (Sakaizumi et al., 1983) and many inbred strains reflecting geographic differences have been prepared (Hyodo-Taguchi and Sakaizumi, 1993). In relation to identifying active transposable elements, a useful infrastructure is provided by the collection of various body color mutant lines available mainly due to the efforts of Hideo Tomita of Nagoya University. His professional experience and patience have allowed identification of various genetic variants of the medaka fish and establishment of mutant lines. These are all spontaneous mutations and still maintained in Nagoya and at other institutions (Tomita, 1975, 1992).

The Tol1 element

There are several different albino mutants in Tomita's collection. He was convinced (personal communication) that some of them are due to defective tyrosinase genes. Tyrosinase is the key enzyme in melanin biosynthesis. Because structural analysis of spontaneous mutant genes has long been the main procedure for identifying new transposable elements, we first cloned the wild-type tyrosinase gene of the medaka fish (Inagaki et al., 1994) and then examined its structure in candidate albino strains. The Tol1 element (transposable element of Oryzias latipes, 1; GenBank accession number  D42062) was found as a result of this analysis. It had been inserted in the first exon of the tyrosinase gene, making the gene nonfunctional (Koga et al., 1995).

The particular Tol1 copy found in the tyrosinase gene was denoted Tol1-tyr. It is 1.9 kb in length, carries terminal inverted repeats of 14 bp, and is flanked by an 8-bp target site duplication (Fig. 1.). No meaningful open reading frame (ORF) is evident in Tol1-tyr. Mainly for this reason, it is not clear which other transposable elements Tol1 is related to. We have estimated the copy number of Tol1 to be about 80 per diploid genome (Koga et al., 1995).

Fig. 1

Nucleotide sequences of Tol1 and Tol2 terminal regions. Tol1-tyr and Tol2-tyr are specific copies of Tol1 and Tol2, respectively, inserted in the tyrosinase gene. Nucleotide sequences of their terminal regions are shown in uppercase letters. Tol1 has terminal inverted repeats of 14 bp each and Tol2 carris terminal inverted repeats of 17 bp and 19 bp, indicated by underlining. The sequences of the tyrosinase gene are shown in lowercase letters. Both elements are flanked by 8-bp target site duplications indicated by boldface letters.

While the human tyrosinase gene is more than 70 kb in length (Ponnazhagan et al., 1994), that in the medaka fish is as small as 5 kb, primarily because of the smaller sizes of introns (Inagaki et al., 1998). The entire defective gene of the albino strain, containing the 1.9-kb Tol1 copy, is within the carrying capacity of a lambda phage vector. We obtained such a clone and replaced a small part of the first exon, in which Tol1 is included, with the corresponding region of the wild-type gene. Introduction of this modified clone into fertilized eggs of the albino strain resulted in melanin pigmentation in the skin of hatched fish. There was no nucleotide sequence difference in the manipulated region between the wild-type and the defective genes, except for the Tol1 insertion and its flanking target site duplication (unpublished). These results indicate that the particular Tol1 insertion event in the tyrosinase gene is not so old as to lead to accumulation of other mutations on the already defective tyrosinase gene. In other words, Tol1 is, or was until recently, active with respect to transposition.

More direct evidence for its transposition activity has recently been obtained. We exposed embryos of the albino strain to UV, and conducted PCR with a pair of primers encompassing the Tol1 insertion point. The length of a major PCR product was longer by 1.9 kb than the product with a wild-type fish. However, in some cases, minor products of a length identical, on agarose gels, to the wild-type form were observed. Cloning and sequencing of these products revealed excision footprints, remnants of Tol1 terminal regions or loss of nucleotides originally belonging to the tyrosinase gene (unpublished). Such footprints are observed with many DNA-based transposable elements (cf. Pohlman et al., 1984; Bryan et al., 1990).

From the above results, an autonomous copy that is presumed to carry a transposase gene would be expected to be present somewhere in the genome. However, no such copy has been detected among more than 20 copies we have so far cloned and sequenced.

Tol1 occurs in seven of the nine Oryzias species we have examined by genomic Southern blot analysis (Fig. 2). The probe used was the 1.9-kb entire Tol1-tyr copy cloned from the medaka fish O. latipes. Hybridization signals became weaker with the phylogenetic distance from O. latipes, with the exception of O. mekongensis. This result suggests that Tol1 was present in the common ancestor of the Oryzias species. We have also conducted more analytical Southern blot experiments, in which different parts of the 1.9-kb Tol1 copy were used separately as probes. We observed several species-specific structures. For example, a central 0.4-kb region of Tol1-tyr hybridized to virtually all Tol1 copies in O. latipes but did not produce any hybridization signals with O. javanicus, indicating that all the Tol1 copies of the latter species lack this portion (unpublished). These results suggest that, although Tol1 appears to be an ancient resident in the genus Oryzias, its transposition occurred after species divergence.

Fig. 2

Distribution of Tol1 among species. Phylogenetic trees with nucleotide sequence data were constructed by Naruse et al. (1993) and Naruse (1996). The scheme was redrawn from these references for the nine species examined. Genomic Southern blot hybridization to examine the distribution of Tol1 among species was conducted as described (Koga et al., 1995), using the entire Tol1-tyr element as probe. The species names are abbreviated with their first three letters. Hybridization signals are not evident in the the lane for O. mekongensis on this autodiagram, but longer exposure revealed about ten bands. Along the left margin are indicated the sizes and mobilities of the size marker DNA fragments.

The Tol2 element

Tol2 was identified by the same approach, as adopted for Tol1, in a different albino strain, that is, a different tyrosinase allele (Koga et al., 1996; GenBank  D84375). It had been inserted in the last, fifth exon of the tyrosinase gene. It is 4.7 kb in length, carries terminal inverted repeats of 17 bp and 19 bp, and is flanked by an 8-bp target site duplication (Fig. 1). This element belongs to the hAT transposable element family (Calvi et al., 1991; Atkinson et al., 1993) that includes hobo of Drosophila, Activator of maize and Tam3 of snapdragon. Tol2 was directly demonstrated to be active soon after its discovery. We carried out PCR with primers encompassing the Tol2 insertion point, and PCR products with the same length as the product from the wild-type fish were readily obtained without any treatment such as exposure to UV. In these PCR products, the element was found to have been removed and footprints were observed.

Tol2 has four ORFs that roughly correspond to four exons making up the mRNA for the Tol2 transposase (Koga et al., 1999; Koga and Hori, 2000), which catalyzes both excision and insertion of Tol2. Microinjection of fish eggs with in vitro synthesized mRNA for the transposase triggers transposition of a co-injected Tol2 clone (Koga and Hori, 2000; Kawakami et al., 2000). This opened a way to the development of a gene transfer vector. We embedded the green fluorescent protein (GFP) gene in Tol2 and injected this clone into fertilized eggs together with the transposase mRNA. The GFP gene was successfully introduced into the fish genome and transmitted to subsequent generations. Sequencing analysis demonstrated the entire Tol2 element, including the GFP gene, to have been integrated into the chromosomes through its transposition (Koga et al., 2001). The Tol2 clone, modified to carry the GFP gene and other necessary components, was 9.0 kb in length. Therefore, Tol2 still retains transposition activity even when it is as large as 9.0 kb. Larger versions are now being tested to determine the limitation in size.

Regarding variation among copies in the genome and the distribution among species, Tol2 presents a striking contrast to Tol1. We examined the presence/absence of Tol2 in ten Oryzias species and found it only in two, the medaka fish O. latipes and the Hainan medaka fish O. curvinotus, the latter being an inhabitant of Southern China. These two species are relatively closely related to each other but nucleotide sequence variations exist on their chromosomes. Despite this variation, Tol2 copies exhibit virtually no nucleotide sequence differences between the species. From results of quantitative analysis, we have proposed that Tol2 was recently introduced into one of these two species from the other or into both of them from an unknown source (Koga et al., 2000).

An LINE-like element

Nonautonomous copies of DNA-based elements are frequently shorter than autonomous copies due to deletion of internal regions. Therefore, a longer copy, if found, can be taken as a candidate for an autonomous copy. We therefore examined length variation of Tol1 copies randomly collected from genomic libraries of O. latipes and O. curvinotus. Although no candidate has yet been obtained, a by-product of this search was an LINE-like element inserted in Tol1. One of the Tol1 copies we examined was disrupted by an extra fragment of about 5 kb and this fragment proved to contain ORFs. The longest frame was for 388 amino acids and their sequence exhibited a similarity to that of the reverse transcriptase gene of the human L1 (LINE-1) element (Fig. 3). This insertion sequence is present in multiple copies in the genomes of O. latipes and O. curvinotus. Because the ORF is not sufficiently long, the sequence is not likely to be a complete copy of LINE. An LINE element called Swimmer 1 has already been identified in the medaka fish genome (Duvernell and Turner, 1998). The newly found LINE-like element, however, seems to differ from Swimmer 1.

Fig. 3

The newly found LINE-like element. The upper dot matrix is for nucleotide sequences (nt) between the genomic fragment containing a Tol1 copy of O. curvinotus (abscissa) and the first found Tol1 copy of O. latipes, Tol1-tyr (ordinate). The criterium for matching was a 80% match over a window of 20 nucleotides. The Tol1 copy of O. curvinotus contains an insertion sequence, which is represented by the shaded box. The arrows in the box are included ORFs. The lower dot matrix is for amino acid sequences (aa) between the longest ORF (abscissa) and part of the second ORF of the human L1 element (ordinate). The criterium for matching was a 30% match over a window of 20 amino acids. The human L1 element is quoted from GenBank  X52234, and the part used here corresponds to its reverse transcriptase gene region.

Future studies

Evidently Tol1 and Tol2 are active with respect to transposition. Therefore, these elements can be expected to provide useful genetic tools applicable to vertebrates. In particular, Tol2 belongs to the hAT transposable element family, members of which are widely distributed among organisms, including vertebrates (Esposito et al., 1999; Rubin et al., 2001). We have already demonstrated that Tol2 works as a gene transfer vector in fish (Koga et al., 2001). We are now trying to develop other types of genetic tools, such as an insertional mutagenesis agent and a cloning tag, first in fish and subsequently in other vertebrates.

The elements are also useful in studying the evolution of transposable elements because Tol2 seems to be in a very early stage and Tol1 in a relatively late stage of the probable scenario for transposable elements, and because a strong background for evolutionary studies has already been generated for the medaka fish.