The results of B-PCR of genomic DNA from eight fishes are shown in Figure 1. This PCR was performed with only one primer, corresponding box B of the tRNA-Leu-like region. The number of resulting bands varied from three to six among the fish genomes (Figure 1A). T. kammalensis gave the fewest bands, where one was intense and two were weak. The shortest band was ~300 bp, and was present in C. nasus, C. mystus, C. grayi, and B. pectinirostris. The longest fragment was ~2000 bp, observed in P. fulvidraco. The amplified patterns in the B-PCR of these fishes may be due to genomic fragments that included at least two “tail-to-tail” oriented SINE copies inserted not far from each other.

C. nasus was selected as a representative species for further study. All the bands produced from this species by PCR were separately excised from the gel and purified, then propagated in E. coli DH5α cells. Positive clones with inserted fragments of the expected sizes were selected randomly and sequenced. Six of 15 sequenced clones revealed that they contained a box A (Figure 2). Three of the cloned target DNAs were identical and were therefore not included into the alignment shown in the Figure 2.

Only the A primer to box A of tRNA-Leu-like region was used to perform PCR amplification of SINE copies from genomic DNA. The band patterns of PCR products are shown in Figure 1. Multiple bands were produced from one genome, suggesting that genomic DNA fragments contained SINE copies oriented in “head-to-head” arrangement, separated from each other by linker sequences of variable length. For further detailed study, the PCR bands obtained from the genome of C. nasus were excised from the gel, cloned, and sequenced. The sequence dataset revealed that these clones have characteristics of SINE elements (Figure 2), such as box B sited at the 5′ region, a tail region containing poly(A), A + T richness, or a 5-bp repeated motif (TGTAA)_n in the 3′ region. This tandem motif may be produced by a tandem duplication event. The tRNA-related region was highly conserved between these clones obtained by A- and B-PCR. In the tRNA-unrelated region, however, these clones showed significant variety, especially in a large fragment inserted in the region of two clones (CnAE10 and CnAF5). Although the inserts are, respectively, 37 bp and 55 bp long, they share a common central domain (26/29 bp) (Figure 3). The conserved sequences indicate that the inserts originated from the same DNA sequence and subsequently evolved through mutation.

A total of seven clones that contained SINEs were isolated from C. nasus genomic DNA, using probe bound to magnetic particles. Alignment of these clones’ sequences showed that they have a 75-bp tRNA-related tract. In addition, a tRNA-unrelated region and a ~40-bp-long tail with an A/T-rich tract were also found in these clones, indicating that they are copies of a type of SINE (Figure 4). We used BLAST to search for sequences homologous to the SINEs from C. nasus in the NCBI databases [19]. No significant similarity was found except that a 386-bp fragment (position 415 to 800) of clone CnAE7 is homologous to long interspersed nuclear elements (LINEs) in the pufferfish, Tetraodon nigroviridis (position 913 to 524, GenBank accession: AJ312227) (data not shown). These clones thus belong to a new family of SINEs, which we name Cn-SINEs (SINEs from C. nasus). The members of this family are 180–360 bp long. We have deposited their sequences in the NCBI databases (GenBank accession: JQ083280–JQ083297).

The BLAST homology search revealed that the tRNA-related region of the consensus Cn-SINE family was most similar to tRNA^Leu in Xenopus laevis (62%, not counting the extra arm-loop region) [20], and showed high sequence conservation (91%) in box A and box B relative to tRNA^Leu in X. laevis and Drosophila melanogaster. It is also 83% identical to the 5′ end of the Hpa1SINE isolated from Parahucho perryi [21]. The predicted secondary structure of the tRNA-derived region of Cn-SINE family has the common characteristics of tRNAs such as 5'pG, 3'CCA, and the arm and loop structures. These new structures lack an extra arm present in tRNA^Leu from X. laevis. However, the acceptor arm, 3-pair-bp-DHU arm, 5-pair-bp-TψC arm, and TψC loop of Cn-SINE are highly similar to that of the tRNA^Leu gene (Figure 5). This conservation of secondary structure suggests that the tRNA-related region of the Cn-SINE family likely evolved from the tRNA^Leu gene.

Analysis of the full-length SINEs isolated from C. nasus revealed direct repeats flanking each terminus of the SINEs in this family. A “TTT” sequence that acts as a strong polymerase III terminator in tRNA genes, including the tRNA^Leu gene [20], was found in the tail region, suggesting that transcription in SINE elements should terminate at this locus. The putative secondary structure of the tail region of Cn-SINE was as previously described [22]. Although two types (I and II) of loop structures were inferred from the sequences of the different clones, a hairpin structure was found in the tail with a string of A’s in the DNA template strand and the U’s in the RNA product (Figure 6). The rU-dA base pairs are exceptionally weak. This led polymerase III to pause at the rU-dA base pairs locus, allowed the RNA to dissociate from the DNA template strand and ensured the complete transcription of the SINEs. Interestingly, an imperfect inverse-repeated tract corresponding to box A was found in Cn-SINEs (Figure 4). It likely results from a head-to-head orientation of SINEs. This feature allowed us to isolate copies of these SINEs by use of a simple, single-primer PCR method.

In the present study, we exploited the structural features of SINEs to capture novel SINE families from the genome of C. nasus. In general, the known SINEs have the conserved boxes A and B, which serve as internal transcription promoters of polymerase III [12]. In addition some head-to-head/tail-to-tail-oriented members of SINE families are separated by short spacer sequences, although the majority of SINEs are widely scattered in the genome [13]. Members of the tRNA-derived SINE family may exist in distantly related fishes, arranged in head-to-head or tail-to-tail orientation. In our study, the presence of box A and box B sequences within the clones initially obtained via A/B-PCR does not prove that SINEs exist in the genome. Clones with the spacer sequence between box A and B of SINEs could be used as templates for biotinylated probes. This is an important step in using probes linked to magnetic particles to capture SINEs as described [14]. In the previous method, inverse PCR was used to obtain the sequence used as the template for making the biotin-labeled probe, with subsequent capture of genomic DNA fragments containing target sequence by the bead-linked probe. Our approach omitted inverse PCR, and only two PCRs were performed to obtain probe template for isolation of full-length SINEs, without genomic library construction or screening.

The Cn-SINE family that we isolated in this study has not been reported previously, and belongs to a novel family. This success in finding previously undescribed SINEs validates our approach. The box A and B sequences of Cn-SINEs are 91% identical to that of the tRNA^Leu genes of Drosophila and Xenopus [20,23], indicating that Cn-SINEs may have originated from tRNA^Leu genes. BLAST of the consensus tRNA-related sequence of Cn-SINEs against the NCBI databases shows 83% identity to the 5′-terminal section of Hap1 SINE (D49862, AB002416) obtained using in vitro synthesis of labeled RNA from P. perryi and Oncorhynchus masou DNAs [21], suggesting that SINEs derived from tRNA^Leu genes are likely to be present in various species. Successfully isolation of Cn-SINEs from C. nasus demonstrates that our method is a simple and rapid way to isolate tRNA^Leu-related SINEs from the unsequenced genomes of these species. Borodulina and Kramerov [12], taking advantage of the conserved box A and box B, developed the “A-B” PCR approach to isolate SINEs from bat genomes, and other authors have used this method to isolate SINEs from various genomes [9,14,17]. We tried to isolate SINEs from C. nasus by “A-B” PCR, but failed to detect any (data not shown). To our knowledge, the B(A)-PCR method described in this paper is the first to discover a SINE family derived from a tRNA^Leu gene or a tRNA^Leu-like gene.

More than 100 SINE families have been described from the genomes of various eukaryotes [24]. Of the previously known SINE families, most have a tRNA-derived tract in the 5′-end region of the repeated elements, and the tRNA genes of origin are commonly one of eight tRNA genes (Ala, Arg, Ile, Tyr, Gly, Ser, Lys, and Glu) [5]. For example, the CHRS family originated from tRNA^Glu in the genomes of artiodactyls and cetaceans [7]; the salmonid SINE families derived from a tRNA^Lys gene [25]; and a tRNA^Arg-related region of the P.k.SINE family was described in bats [17]. The novel Cn-SINEs isolated from the C. nasus genome have a tRNA^Leu-related region at their 5′ end, and their consensus tRNA-related sequence contains the conserved 11-bp box A and 11-bp box B separated by a 33-bp spacer similar to other SINE families [8]. The tRNA-unrelated region of Cn-SINEs varies in length mainly due to nucleotide indels, but high sequence identity (~90%) was found among these members of the Cn-SINE family, excluding the large inserted fragment (Figure 4; identical nucleotides are denoted by asterisks). A high degree of similarity was observed between the insertions of clones CnAE10 and CnF5, though they differed in length (Figure 3). These results suggest that the tRNA-unrelated regions of Cn-SINEs may be of common origin from DNA fragments, subsequently subjected to numerous mutational events during SINE evolution. There are other cases of a common central domain in the tRNA-unrelated region existing in various SINE families [26].

The Cn-SINEs have a 3′-terminal tail with poly(A), or an AT-rich region. The length of the tail region is 17–48 bp. Two types of secondary structures were inferred from the sequence of the tail region of SINEs in different clones (Figure 6), and both have a hairpin structure. This is required for efficient release of the transcript of DNA with SINE fragments. The hairpin structure is followed by a string of A’s in the tail. As the DNA template of SINE is transcribed, the poly(A) would let the RNA dissociate from the template and terminate transcription of SINE through a string of A-U base pairs formed between the A’s in the SINE template strand and the U’s in the transcription. Thus, transcription of the full SINE would probably be confined to special DNA fragments. The poly-A tail of a transcribed SINE will attach to the free “TTTT” site of the target DNA fragment, and act as a primer for reverse transcriptase to synthesize a new SINE insert in the target site [3]. Because the two strands of the “new” DNA are cut at staggered sites, the inserted SINE is flanked by small gaps which, when filled in by a host enzyme, lead to short target-site duplications (TSDs) at the inserted sites [27]. The Cn-SINEs have a perfect or irregular TSD flanking each SINE locus (Figure 4), revealing that in Cn-SINEs, amplification and integration, which are dependent on enzymes derived from the host genome and LINEs, occurred during evolution. SINEs and LINEs have similar 3′ tails, and SINEs acquire retropositional activity from reverse transcriptase generated by LINEs [14,17,28]. In our present study, one clone, CnAE7, has a 3′ terminal tail similar to LINEs in the pufferfish Tetraodon nigroviridis, implying that the 3′ tail of Cn-SINE family members was derived from LINEs. These and further studies may reveal the influence of these novel Cn-SINEs on the evolution of fish genomes as well as help explain how this type of retrotransposition originates from tRNA^Leu genes.

Genomic DNA (~1 μg) was completely digested for 2 h at 37 °C with Bsp1431 (Takara, Dalian, China) in a total volume of 100 μL. The fragments were separated by 1% agarose gel electrophoresis. The fragments between ~400 bp and ~2000 bp were purified from the gel using a gel DNA purification kit and suspended in 50 μL of distilled water.

The adaptor is a partially double-stranded DNA composed of a long strand annealed to a short strand to yield Bsp1431-compatible cohesive ends (a 3′ 4-bp overhang). The adaptor sequence was as follows: 5′-GTAATACGACTCACTATAGGGCCGAGGT-3′, 3′-CCCGGCTCCACTAG-5′.

Excess double-stranded adaptor was added to the above 50 μL of prepared Bsp1431-digested DNA fragments in a 100 μL reaction mixture containing buffer and 20 units T4 DNA ligase (Takara, Dalian, China). The reaction proceeded overnight at 16 °C, and was then purified on a UNIQ-10 column, and finally dissolved in 50 μL of distilled water.

The adaptor primer used for PCR amplification of genomic DNA corresponds to the 5′-end sequence of the long strand of the adaptor. The primer sequence was 5′-GTAATACGACTCACTATA GGGC-3′. The adaptor-ligated DNAs were amplified by PCR in a 20 μL volume composed of 15.6 μL distilled water, 2.5 μL 10× PCR Buffer, 0.5 μL of 10 μM primer, 0.2 μL of 10 mM dNTPs, 0.2 μL of Taq DNA polymerase, and 1 μL of the adaptor ligation product. The PCR began at 72 °C for 5 min to extend the 3′ end of the adaptor fraction, followed by 17 cycles of 94 °C for 30 s, 58 °C for 45 s, 72 °C for 90 s; final extension was 72 °C for 10 min. Five separate PCR products were pooled to 100 μL and purified using a column purification kit and finally suspended in 100 μL of distilled water.

The forward (F) and reverse (R) PCR primers used to synthesize biotin-labeled probes were designed by Primer premier 5.0 [32] based on an internal region of a SINE from the clone CnAG2 obtained by B-PCR (Figure 2). The F primer sequence was 5′-AGTGGTGAGGGAGTTGGTCTT-3′. The R primer sequence was 5′-GGGTTTCAGTTACAGGGGTTAG-3′. Biotin was added to the 5′ end of the F primer. A biotinylated probe was obtained via PCR with the biotin-labeled F and R primers and clone CnAG2 as the template. PCR was performed and products were purified as described above, except that the annealing temperature was 56 °C.

The denatured biotinylated probe above was added to the tube containing the denatured genomic fragments with adaptor at their ends. These were annealed at 55 °C for 2 h then incubated at room temperature until completely cooled to allow the biotin-labeled probe to hybridize with the adaptor DNA fragments.

A volume of 0.6 mL of MagneSphere magnetic particles was washed three times with 0.5× SSC (300 μL per wash), and the wash solution was carefully removed using the magnetic stand. The washed magnetic particles were resuspended in 100 μL of 0.5× SSC, added to the prepared probe-adaptor DNA hybrid solution, and incubated at room temperature for 30 min so that the biotin present in 5′ end of the adaptor DNA-hybrid probe specifically attached to the magnetic particles.

The supernatant containing DNA strands not attached to the magnetic beads was removed by washing four times with 0.1× SSC (300 μL per wash). The magnetic stand was used to separate the magnetic beads from the supernatant. Finally, the probe-target DNA complexes were eluted from the magnetic particles in 50 μL of distilled water at 94 °C for 5 min.

The target DNA fragments were used as a template for PCR with the adaptor primer. The PCR volume and profile were as described above. The PCR products were inserted into E. coli DH5α cells for propagation. The insert sizes in many single bacterial clones were directly determined by PCR with the S7 and R47 pMD-19 T vector primers. The clones having inserts longer than 300 bp were chosen randomly and sequenced.