| Edited by Etsuko Matsuura. Akihiko Koga: Corresponding author. E-mail: koga@pri.kyoto-u.ac.jp |
DNA-based transposable elements form one major class of mobile DNA. They are transposed mainly in a cut-and-paste process that is catalyzed by an enzyme known as transposase. Elements in this class occur either as autonomous copies, which contain a functional gene for the transposase, or as nonautonomous copies, which are incapable of producing the enzyme and thus are not mobile unless the transposase is supplied by an autonomous copy coexisting in the same cell. Nonautonomous copies are generated from autonomous copies; one cause of this transition is nucleotide substitutions in the transposase gene resulting in alteration or truncation of the amino acid sequence of the protein product (Medhora et al., 1991; Yang et al., 2008). Another cause, which appears to be more frequent among many DNA-based elements, is deletion of internal regions of the element (Fedoroff et al., 1983; O’Hare and Rubin, 1983; Streck et al., 1986). The underlying mechanism for internal deletion is considered to be premature interruption of gap repair (Engels et al., 1990; Gloor et al., 1991; Rubin and Levy, 1997).
Gap repair is a process of restoring a gap region generated around the double-strand break that is left after excision of an element. According to a widely accepted model, gap repair involves synthesis-dependent strand annealing (SDSA) using the corresponding region of the homologous chromosome or sister chromatid as a template. Within the element region, strand synthesis proceeds inward from the ends. When the two newly synthesized strands extend sufficiently to overlap each other, the entire element is restored at the donor site of excision. If the extension is aborted prematurely, nonhomologous endjoining occurs at short repeats on the two newly synthesized strands, resulting in the lack of a particular internal region.
The transposase enzyme cuts out the entire element that is being transposed, but it only recognizes the nucleotide sequences of the terminal regions of the element. Because of this property, it can transpose nonautonomous copies as long as their terminal regions are retained. Thus the transposition efficiency per copy is not expected to differ between autonomous and nonautonomous copies when they coexist in a single genome. Though natural selection favors neither autonomous nor nonautonomous copies, nonautonomous copies are thought to be better tolerated, for two main reasons (Hartl et al., 1997). The first reason is that multiplication of an autonomous copy may lead to an increase in the total amount of transposase, while multiplication of a nonautonomous copy does not. The second is that some nonautonomous copies may produce nonfunctional transposases that compete with, and thus weaken the power of, the functional transposase.
The relatively strong natural selection against autonomous copies and the virtual unidirectionality of the transition from autonomy to nonautonomy cause the incidence of nonautonomous copies to increase over time. This process can be regarded as the decay of autonomous copies, and the period of their decay may overlap that of their amplification. The progression of decay is expected to lead eventually to a complete loss of autonomous copies. Stochastic change in copy number is considered to be another important factor causing the loss of autonomous copies as well as autonomous and nonautonomous copies simultaneously (Lohe et al., 1995). Once autonomous copies are lost, any nonautonomous copies present in the genome are no longer mobile.
Several species of plants and invertebrate animals contain DNA-based elements that have not yet reached the point of immobilization. Maize, for example, is the host species for Activator and En/Spm (Fedoroff et al., 1983; Pereira et al., 1986), Drosophila is the host for P, hobo, and mariner (Bingham et al., 1982; McGinnis et al., 1983; Jacobson et al., 1986), and Caenorhabditis elegans is the host for Tc1 and Tc3 (Emmons et al., 1983; Collins et al., 1989). In vertebrates, in contrast, only one element has yet been clearly demonstrated to be active. This is Tol2 in the medaka fish (Fig. 1), which causes various mutations by insertion (Koga et al., 1996; Iida et al., 2004) and excision (Iida et al., 2005; Koga et al., 2006). We have previously shown that its copies are highly homogenous in structure. No internally deleted copies were found among 24 samples randomly collected from one natural population (Koga and Hori, 1999) or among 12 samples from different populations (Koga et al., 2000). There was virtually no sequence variation among Tol2 copies, even at the third nucleotides of codons of the transposase gene and in introns. We inferred from these observations that Tol2 has undergone a recent and rapid amplification in the genome and habitat of medaka.
 View Details | Fig. 1 Structure of autonomous copies of Tol1 and Tol2. Abbreviations are as follows: TSD, target site duplication; TIR, terminal inverted repeat; IIR, internal inverted repeat. Tol1 does not have IIRs. ProA-ProB indicate regions covered by these probes for Southern blot analysis. The following regions of the GenBank file AB288091 were amplified by PCR: ProA, nt 1–425 (425 bp); ProB, nt 937–1476 (540 bp); ProC, nt 1817–2379 (563 bp); ProD, nt 2917–3416 (500 bp); ProE, nt 3864–4355 (492 bp). |
In addition to Tol2 in medaka, there may be one other mobile element that is active in natural populations: the Tol1 element in the same species. A 1.9-kb-long copy of Tol1 was found more than ten years ago; it was obviously nonautonomous because it lacked a meaningful open reading frame (Koga et al., 1995). Extensive subsequent searches failed to find an autonomous copy of Tol1, suggesting that this element had already been immobilized (Koga et al., 2002). Recently, however, we identified a 4.4-kb-long autonomous Tol1 copy (Fig. 1) in a laboratory medaka strain (Koga et al., 2007a). This copy carries a transposase gene consisting of three exons. The 1.9-kb copy previously observed was revealed to be an internally deleted version. Since a nucleotide sequence of internal regions of the autonomous Tol1 copy is now available, we can examine whether autonomous copies still survive in natural populations of medaka by Southern blot analysis using the internal regions as probes. We surveyed fish samples originally collected at 32 locations, representing the entire habitat of medaka, and obtained results indicating that full-length Tol1 copies reside in the genomes of 10 of the 32 samples. Further analyses of some of these copies provided evidence that they are capable of producing a functional transposase.
It should be noted that the distinction between autonomous and nonautonomous elements is, in a widely accepted concept, made on the assumption that the elements are themselves substrates for the transposase. An "autonomous element" is simply defined there as a copy that can produce the transposase. We here use this definition unless otherwise noted. When we need to consider a copy that is capable of producing the transposase but has lost signals to be recognized by the transposase, we describe the copy as such in each case.
Our primary purpose was to determine the genomic organization of Tol1 copies. BLAST search against the sequence database of medaka is one choice for this purpose, but we relied on genomic Southern blot analysis. It is known that repetitive sequences at least several kilobase pairs long tend to be underrepresented in sequence databases. In fact, when we conducted a BLAST search with the entire Tol1 sequence as a query, many contigs were found to be truncated in the internal regions of Tol1. Thus, genomic Southern blot was considered to be more powerful for our primary purpose, especially for internal regions. On the other hand, terminal regions of Tol1 can be presumed to be accurately reflected in the database because their flanking regions are expected to be helpful in determining their linkage to other sequence reads. For this reason, we chose BLAST search as a method to determine the copy number of Tol1.
Survey by PCR is another choice for the purpose of determining the genomic organization of Tol1 copies, but this is also less powerful than Southern hybridization. We had an experimental strain which contained a full-length Tol1 copy, in addition to defective copies, in its genome. We tried a PCR to amplify the transposase gene region from genomic DNA of this strain, using primers that represented the terminal regions of the gene, but this was not successful (data not shown). A probable reason is that shorter fragments are amplified more efficiently in PCR and that defective copies were much more abundant than the full-length copy. Thus, we performed our survey not by PCR but by Southern hybridization.
The medaka fish inhabits East Asia, including China, Korea and Japan. Fish lines originally collected at more than 60 geographical locations are maintained as mass-mating stocks at Niigata University. For the present study, we chose the 32 lines shown in Fig. 2. In addition to samples from these lines, we included samples from two laboratory strains, HNI and Hd-rR, in our analyses.
 View Details | Fig. 2 Original collection sites of the 32 fish lines used in the present study. The sites are as follows: 1, Higashidori; 2, Honjo; 3, Ojiya; 4, Nanao; 5, Ine; 6, Kumihamanagae; 7, Toyooka; 8, Hamasaka; 9, Hanamaki; 10, Kawachi; 11, Odawara; 12, Fuji; 13, Saori; 14, Kumano; 15, Aki; 16, Hikone; 17, Saigo; 18, Iwakuni; 19, Saito; 20, Ashibe; 21, Sato; 22, Ginoza; 23, Sokcho; 24, Daebu; 25, Guoje; 26, Jindo; 27, Guhang; 28, Misan; 29, Shimcheon; 30, Ilan; 31, Kunming; 32, Kwangi. |
Studies based on isozyme frequencies and nucleotide sequences of mitochondrial DNA have concluded that the species consists of four local populations (Sakaizumi, 1986; Sakaizumi et al., 1987). The local populations and the line numbers included in each are "Northern Japan" (1–5), "Southern Japan" (9–22), "East Korea" (23–26), and "China and West Korea" (27–32). In addition, a relatively small area where fish possess characteristics intermediate between those of the first two populations is known to exist on the northwest coast of the largest island of Japan. Lines 6–8 belong to this population. The HNI and Hd-rR strains are highly inbred strains originating from the "Northern Japan" and "Southern Japan" populations, respectively. The latter is the strain used as the DNA source for sequence data published in the medaka section of the Ensembl genome browser.
The purpose of our Southern blot analysis was to examine the copy numbers of various parts of the 4.4-kb-long autonomous copy of Tol1 in each of our 34 fish samples. To accomplish this, we prepared five probes (ProA-ProE) representing different regions of the Tol1 copy; these are illustrated in Fig. 1. The probes were 425–563 bp in length and distributed with roughly the same intervals on the element.
Genomic DNA was extracted from whole adult fish by a standard method consisting of homogenization in EDTA-containing buffer, SDS and Proteinase K treatments, salt sedimentation, and then ethanol precipitation. Each DNA pellet was dissolved in 10 mM Tris (pH8.0) at a final concentration of 0.2 μg/μl.
For each of the 34 fish samples, we digested 50 μg of DNA with 500 units of HindIII restriction enzyme. The 4.4-kb Tol1 copy carries a site for this enzyme, but the site is not inside the regions for the five probes. After ethanol precipitation, we divided the digested DNA into five tubes. We then conducted five sets of electrophoresis using 130-mm-long 1.0% agarose gels. Because the maximum number of sample lanes contained in a gel was 12, we prepared and used three gels for each set ([1–12], [13–24], [25–32, HNI, Hd-rR]). After the electrophoresis, DNAs were transferred to nylon membranes, fixed by UV irradiation, and hybridized with alkaline-phosphatase-labeled probes. We used the AlkPhos Direct Labeling and Detection System (GE Healthcare, Chalfont St. Giles, UK) for probe labeling, hybridization, washing and signal detection, following the manufacturer’s instructions. The hybridization temperature was 59°C, which is presumed to give a medium stringency.
The transposition reaction of DNA-based elements consists of two processes: excision and insertion. The former is easy to detect because loss of a specific copy can be readily manifested using PCR. Detection of the latter requires a complex approach because no one can anticipate where the element will land. In the present work, we only examined excision, because the Tol1 transposase has already been demonstrated to catalyze both excision and insertion (Koga et al., 2008).
pDon[Tol1] is a plasmid used in our previous study (Koga et al., 2008). It carries the terminal regions of Tol1 with a green fluorescent protein (GFP) gene reporter cassette between them (Fig. 3). The modified Tol1 element in this plasmid is a nonautonomous copy because it does not carry the transposase gene.
 View Details | Fig. 3 Structure of pDon[Tol1]. The plasmid contains a GFP gene reporter cassette between the 157-bp left and 106-bp right Tol1 terminal regions, which are flanked by 8-bp TSDs. The plasmid backbone is pUC19. The 4.4-kb-long Tol1 copy is also shown for reference. Black triangles are primers for PCR analyses. Sequences of primers located on pUC19 are the following nucleotide blocks of GenBank file L09137: Pri1L, nt 1921–1950; Pri1R, nt 2348–2319; Pri2L, nt 244–273; Pri2R, nt 758–729; Pri3L, nt 338–367; Pri3R, nt 650–621). Those on the Tol1 element are the following parts of AB288091: Pri4L, nt 1817–1846; Pri4R, nt 2379–2350. |
Assay to detect excision of the modified Tol1 element from its vector portion was conducted using medaka culture cells. We first crossed a male fish of each line to be examined with female fish of the HNI strain. The reason for using HNI females was the ease in collecting females in the reproductive stage. We collected fertilized eggs, incubated them at 25°C for five days, and then homogenized single embryos in PBS. After two rounds of centrifugation and resuspension of cell pellets in PBS, we put the cells in 1.0 ml of culture medium (L-15 with 15% fetal bovine serum and antibiotics) in a well of a 24-well microplate. We allowed the cells to propagate on the bottom surface (1.9 cm2) of the well by incubating them at 28°C in an air incubator and changing half of the volume of the medium every two days. When cells had propagated sufficiently to cover 20–40% of the bottom surface (5000–20000 cells), we introduced 500 ng of the DNA of pDon[Tol1] into the cells using the Lipofectamine LTX transfection reagent (Life Technologies Corp., Carlsbad, CA, USA). Cells were exposed to the transfection medium for 6 hours, and incubated in the regular medium for a further 24 hours. After the cells were washed twice with PBS, we extracted DNA from them. Finally, the DNA was dissolved in 200 μl of 10 mM Tris (pH8.0). We then conducted three kinds of PCR analysis, each of which is described below, using ExTaq polymerase (Takara Bio Inc., Otsu, Japan). In all analyses, the reaction mixture was 25 μl in total volume, containing 5 μl of template DNA, 0.4 μM of each primer, 2.0 μM MgCl2, 1x buffer, and 0.1 μl of the polymerase. The PCR conditions were: [120 s at 94°C], n cycles of [20 s at 94°C, 15 s at 64°C, 30 s at 72°C], and [60 s at 72°C]. After the completion of PCR, 5 μl of the reaction solution was electrophoresed on a 90-mm-long 1.4% agarose gel.
The first PCR analysis was intended to confirm recovery of the DNA of pDon[Tol1] from the cells. The PCR primers, Pri1L and Pri1R, were intended to amplify a 0.5-kb portion of the plasmid backbone of pDon[Tol1] (Fig. 3). The cycle number was set at n = 30.
The second analysis was intended to detect Tol1 excision events that may have occurred while the cells were being cultured. Because only a small fraction of DNA molecules of the recovered plasmid, if any, were expected to have undergone excision, we conducted two rounds of PCR, using nested primer pairs (Fig. 3). The first round was with primers Pri2L and Pri2R; the cycle number was set at n = 20. We diluted the reaction mixture to 1/500 with water, then conducted a second round of PCR with primers Pri3L and Pri3R and a cycle number of n = 30.
The third PCR analysis was intended to determine whether a putative autonomous Tol1 copy had been inherited by each embryo. We used primers Pri4L and Pri4R (Fig. 3), which had served as primers for the preparation of ProC (0.6-kb-long central region of the autonomous Tol1 copy). PCR was conducted with the cycle number set at n = 40.
Searches for sequences containing the Tol1-terminal region and its flanking regions were conducted using the BLAST search program of Ensembl (http://www.ensembl.org/). The query sequence was 50 nucleotides from the left end of the first-found Tol1 copy (GenBank accession number D42062).
We conducted Southern blot analysis on fish samples from 32 lines and two inbred strains, using five probes (ProA-ProE) representing different regions of the 4.4-kb Tol1 copy (Fig. 4). An obvious feature of the organization of Tol1 is that the internal regions (ProB-ProD) exhibit smaller copy numbers than the terminal regions (ProA and ProE), the central region (ProC) having the smallest copy number. These results indicate that the Tol1 element has undergone a decay process driven by internal deletion, as observed in many DNA-based elements. Among the 32 samples from different collection sites, 22 samples exhibited no hybridization signal with ProC, and the other 10 samples exhibited between 1 and 4 bands. The total number of bands for this probe was 17. Many of the elements responsible for these bands are likely to be hemizygous because the fish stocks have been maintained by mass mating. Even if they are all homozygous, however, the average copy number per diploid genome is about 1 (2×17/32). Furthermore, this is probably an overestimate in that more bands than the actual number of responsible Tol1 copies may appear when an element has a recognition site for the restriction enzyme used (HindIII) in the probe region.
 View Details | Fig. 4 Southern blot analysis of the 32 fish lines and 2 inbred strains for 5 regions of the Tol1 element. Probes are shown on the left side of each row. |
On the autoradiographs of the Southern blot analysis, the numbers of bands for ProC and ProD were easily determined, but those for the other probes were too large to count. As another means of estimating the copy number of Tol1, we conducted a BLAST search against the medaka sequence database, using the 50 nucleotides from the left end of Tol1 as a query. We found 174 sequence files exhibiting E values of less than 10–6. We then compared their Tol1-flanking regions, obtaining more than 100 clearly distinct sequences (data not shown). Thus, the copy number of the left terminal region of Tol1 is more than 100 per haploid genome, Hd-rR being a highly inbred strain. In the panel for ProA in Fig. 4, the intensity of the hybridization signals in the Hd-rR lane appears to be approximately average compared to that in the other fish samples. Thus a minimum estimate of the copy number of Tol1 in an average fish is 100 per haploid genome, or 200 per diploid genome.
We chose two lines (4 and 9) from the ten exhibiting more than one band for ProC in our Southern blot analysis. We crossed male fish of these lines to females of the HNI strain by pair matings. A pair of HNI fish was also prepared; note that this strain does not exhibit a band for ProC (Fig. 4). We cultured cells originating from single embryos obtained from these crosses and introduced pDon[Tol1], which was a plasmid carrying a defective Tol1 copy, into the cells. Using DNAs extracted from the cultured cells, we conducted our three kinds of PCR analysis (Fig. 5). Cultured cells were analyzed from 24 embryos from the [HNI × line 4] cross, 24 embryos from [HNI × line 9], and 2 embryos from the HNI pair.
 View Details | Fig. 5 Results of excision assay. Results from 24 embryos of the [HNI × line 4] cross are shown here. The [HNI × line 9] cross gave similar results. PCR primers are shown on the left of each row. 1 pg of DNA was used as the template in PCR reactions for pDon[Tol1] and pUC19. pUC19 is the vector portion of pDon[Tol1], the structure of which is almost the same as that of excision products from pDon[Tol1]. Five μl of DNA extracted from culture cells was used for fish embryos. |
The results of PCR with primers Pri1L/Pri1R (upper panel) indicate that plasmid DNA was successfully recovered from all embryos. A 0.3-kb-long PCR product that appeared first with Pri2L/Pri2R and then with Pri3L/Pri3R (middle panel) indicates the excision of the Tol1 portion from its vector plasmid. The particular embryos that inherited a Tol1 copy carrying the ProC region were identified by the appearance of a 0.6-kb-long PCR product appearing in response to primers Pri4L/Pri4R (lower panel). Table 1 shows the distribution of embryos with and without PCR products in the last two PCR analyses. The Fisher’s exact test rejected, at high levels of significance, the null hypothesis that the occurrence of excision is not affected by the inheritance of the ProC-carrying Tol1 copy. These results indicate that a functional Tol1 transposase was present in cells that developed from embryos which inherited a ProC-carrying Tol1 copy.
 View Details | Table 1 Distribution of embryos exhibiting PCR bands (Yes or No) and results of Fisher’s exact test |
Since our discovery over ten years ago of a 1.9-kb-long nonautonomous copy of Tol1 (Koga et al., 1995), we have been trying to find its autonomous copy by various methods, including database searches, PCR from genomic DNA, and RT-PCR from RNA. The apparent lack of any autonomous copy led us to suppose that Tol1 had already been completely inactivated. Recently, however, a lucky event indicated the presence of an autonomous copy in a laboratory medaka strain. This event was the incidental coexistence, in a single fish, of a nonautonomous copy inserted in a pigmentation gene and an autonomous copy. Their encounter caused a visible, highly frequent reversion mutation of an albino phenotype. We succeeded in identifying the 4.4-kb-long autonomous copy by analyzing this fish (Koga et al., 2007a, 2007b).
In the present study we used the sequence information of the autonomous copy to analyze the genomic organization of Tol1 and the distribution of various portions of the autonomous Tol1 copy among samples of natural medaka populations. It is now clear that the decay of Tol1 through internal deletion has progressed to an advanced stage. A very rough picture of this genomic organization of Tol1 can be obtained by database searches, but the Southern blot analysis that we performed in this study is far more powerful than database searches because repetitive sequences tend to be underrepresented in databases.
Another significant finding from our survey is that a small number of copies carrying the central region of the autonomous Tol1 copy (corresponding to ProC) are still present in natural medaka populations. It is not known whether the sequences of such copies are identical or similar to those of the 4.4-kb autonomous copy. The number of such copies per fish was estimated to be 1 or less, while more than 200 nonautonomous copies reside in the fish genome. The frequency of full-length copies among all Tol1 copies is thus 0.5% or less. It is also not clear whether all the copies carrying the central region of the autonomous Tol1 copy are themselves autonomous. The results of our excision assay, however, indicate that at least two of those copies are autonomous. Thus Tol1 has not reached the immobilization point in natural populations of medaka. The term "autonomous" indicates here that an element is capable of producing a functional transposase, as described in the INTRODUCTION section. It is possible that the two copies are not "autonomous copies in a strict sense": they may have suffered a decay in their terminal regions. Even if this is the case, our findings still have significance because the elements can supply other copies with the transposase.
Further analysis of the results of our Southern blots provided interesting information about differences in the genomic organization of Tol1 among local populations. Ten lines (3, 4, 9, 10, 11, 13, 14, 16, 19, and 22) were found to carry the central region, of which 2 belong to the Northern Japan population and 8 to Southern Japan. Such copies were not observed in the 22 lines from the other two populations. The difference in the distribution of the copies between Japan (the Northern Japan, hybrid, and Southern Japan populations) and Continent (the other two populations) was statistically significant (Table 2). The decay of Tol1 through internal deletion has progressed to an advanced stage, as discussed above, but the speed may differ among local populations.
 View Details | Table 2 Distribution of lines that were found to carry the ProC region and results of Fisher’s exact test |
The genomic organization of Tol1, revealed in the present work, is in sharp contrast with that of Tol2. The latter element is highly homogenous in structure (Koga and Hori, 1999; Koga et al., 2000); virtually no sequence variation was found, suggesting that the time span that has passed since the proliferation of Tol2 in natural medaka populations has been too short for variation to emerge. The situation of Tol1, on the other hand, is more typical of the known DNA-based elements in various other organisms. Both Tol1 and Tol2, however, are exceptional among vertebrates in that autonomous copies still survive.
Tol1 and Tol2 are both members of the hAT transposable element family, and their autonomous copies are similar in structure, as shown in Fig. 1. The two elements are independent transposase-substrate systems, however: the Tol1 transposase does not catalyze transposition of Tol2, and the Tol2 transposase does not serve as an enzyme for Tol1 (Koga et al., 2008). If they were not independent of each other, the conspicuous difference in their evolutionary stages observed in the present study and our previous studies would not have been generated.
Transposition activity of transposable elements is generally controlled by various factors (Lohe et al., 1995), and there are many examples of sudden or gradual increase in the transposition frequency caused by changes in the status of such factors. For example, the Drosophila P element is activated in specific types of crosses that lead to lack or dilution of repressor proteins originating from the element itself (Misra and Rio, 1990). The transposition frequency of this element is also greatly affected by temperature: high temperature leads to an increase in the transposition frequency (Ronsseray et al., 1984). A converse case exists in the snapdragon Tam3 element (Hashida et al., 2003). Another factor affecting transposition frequency is DNA methylation: the CACTA element of Arabidopsis was reported to be activated by mutations on a gene that participates in the control of the DNA methylation level (Miura et al., 2001).
Increase in the transposition frequency often gives rise to increase in the mutation frequency of host genes due to insertions or imprecise excisions of the elements, such as the striking example of hybrid dysgenesis by the Drosophila P element (Kidwell, 1985). A sudden increase in transposition frequency, as well as its effect on a host gene, has already been observed with Tol2 (Koga et al., 2006), although the mechanisms by which the transposition frequency is increased are yet to be examined. The present study suggests that a similar phenomenon could happen with Tol1 in natural populations of medaka because autonomous copies still exist.
A likely future for Tol1, after a certain amount of time passes, is the complete loss of autonomous copies. One unresolved matter of interest is the time scale on which such evolutionary transitions occur in natural populations. This may be discovered or estimated by comparing the results of the present study and those of future continuous surveys.
Accelerated expansion of genome sequence information has revealed that DNA-based transposable elements exhibited intense activity in the period of mammalian radiation (Pace and Feschotte,, 2007), and that DNA-based elements have been active until recently in some lineages of mammals, such as the bat, bushbaby, opossum and mouse (Ray et al., 2007, 2008; Pace et al., 2008). Among the several known transposable element families, the hAT family is central to the results in these reports. Considering that Tol1 and Tol2 are both elements of the hAT family, data on their evolutionary changes are expected to contribute much to the understanding of effects of DNA-based elements on vertebrate genomes. A unique feature of these elements is that their changes are occurring now rather than in the past.
We are grateful to Samuel S. Chong and Hiroshi Hori for helpful discussion. The fish samples were obtained from the National BioResource Project of Japan. This work was supported by grant no. 19570003 to A. K. and A. S. from the MEXT of Japan, and by a grant from the Yamada Foundation to A. K.
|
Index