Efficient transposition of the Tol2 transposable element from a single-copy donor in zebrafish

Abstract

The Tol2 transposable element is a powerful genetic tool in model vertebrates and has been used for transgenesis, insertional mutagenesis, gene trapping, and enhancer trapping. However, an in vivo transposition system using Tol2 has not yet been developed. Here we report the in vivo Tol2 transposition system in a model vertebrate, zebrafish. First, we constructed transgenic zebrafish that carried single-copy integrations of Tol2 on the genome and injected transposase mRNA into one-cell stage embryos. The Tol2 insertions were mobilized efficiently in the germ lineage. We then mobilized an insertion of the Tol2 gene trap construct in the nup214 gene, which caused a recessive lethal mutant phenotype, and demonstrated that this method is applicable to the isolation of revertants from a transposon insertional mutant. Second, we constructed transgenic fish carrying the transposase cDNA under the control of the hsp70 promoter. Double-transgenic fish containing the transposase gene and a single-copy Tol2 insertion were treated with heat shock at the adult stage. We found that transposition can be induced efficiently in the male germ cells. We analyzed new integration sites and found that the majority (83%) of them were mapped on chromosomes other than the transposon donor chromosomes and that 9% of local hopping events mapped less than 300 kb away from the donor loci. Our present study demonstrates that the in vivo Tol2 transposition system is useful for creating genome-wide insertions from a single-copy donor and should facilitate functional genomics and transposon biology in vertebrates.

The Medaka fish Tol2 transposable element is active in a variety of vertebrates. When a donor plasmid containing a Tol2 construct and the transposase activity, in the form of either an expression plasmid DNA or in vitro synthesized mRNA, is introduced into cells, the Tol2 construct transposes from the donor plasmid to the genome efficiently in human, mouse, chicken, frog, and zebrafish (1). Genetic methods using Tol2 have been developed most extensively in a popular vertebrate model, zebrafish (2–4). For instance, a highly efficient transgenesis method has been developed, and transgenic fish that express fluorescent reporter genes or cloned genes under the control of specific promoters have been created; the gene trap and enhancer trap methods have been developed, and a number of fish expressing GFP in temporally and spatially restricted fashions have been created (5, 6); insertional mutations in important developmental genes have been identified (7); and recently the Gal4 gene trap and enhancer trap methods that enabled targeted expression of reporter and effector genes in the desired time and place were developed (8–10).

To facilitate genetic studies using a transposon tool, development of an in vivo transposition system is important. In Drosophila, a single-copy P element on the genome is mobilized by crossing a transgenic fly carrying a P insertion with a fly line expressing the transposase gene (11). In mice, in vivo transposition systems were developed by using the Sleeping Beauty (SB) transposable element (12–14). In these systems, transgenic mice carrying concatemeric transposon integrations were crossed with transgenic mice expressing the SB transposase. It has been reported that the SB construct transposes efficiently from such concatemeric donors in both germ cells and somatic cells and tends to transpose on the same chromosome as the donor loci, a phenomenon called “local hopping” (12, 15, 16). In zebrafish, it has been shown that the transposition efficiency of Tol2 is higher than that of SB (5, 17). Also, Tol2 belongs to the hAT family of transposons, a different family from the Tc1/mariner family to which SB belongs (18, 19). Therefore, it is important to develop an in vivo transposition system using Tol2 in a vertebrate and to investigate how it behaves in comparison with SB.

Results

Mobilization of Tol2 by Microinjection of Transposase mRNA.

First, to mobilize Tol2 insertions in the genome, we injected the Tol2 transposase mRNA into the XIG8A and the SAGp22A homozygous fish at the one-cell stage. These fish carried single-copy insertions of T2KXIG and T2KSAG in the pax6b gene and the hoxc cluster, respectively (Fig. 1 A and B). We analyzed the injected embryos by PCR using primers that hybridized to genomic sequences adjacent to the transposon insertions. In 100% of theinjected embryos (n = 14 for both experiments), excision products were amplified, indicating that Tol2 at these two different loci can be excised by the transposase efficiently. Second, to test whether excision can occur in the germ lineage, 12 XIG8A and 12 SAGp22A fish injected with the transposase mRNA were raised to adulthood and were crossed with XIG8A and SAGp22A homozygous fish, respectively. We analyzed 12 embryos (F1) from each cross by PCR. In 8.3% to 42% of the embryos from three XIG8A- and two SAGp22A-injected fish, the excision products were detected, indicating that the Tol2 insertions can be excised in the germ lineage efficiently (Fig. 1C). Finally, to determine whether the excised Tol2 is integrated in new loci, we crossed the same injected fish with wild-type fish, and analyzed 12 progeny fish (F1) from each cross by Southern blot hybridization. In 8.3% to 100% of the F1 fish from the three XIG8A and four SAGp22A fish, new bands were detected (Fig. 1D). In total, seven and six new bands were detected in the F1 progeny from the XIG8A and SAGp22A fish, respectively (Fig. 1D). We cloned and sequenced genomic DNA surrounding all these insertions and confirmed that these insertions indeed represented integrations at new loci (supporting information [SI] Table S1). From these results, we concluded that a single-copy Tol2 insertion can transpose efficiently in the germ lineage when the transposase activity is supplied.

Fig. 1.

Mobilization of Tol2 insertions by microinjection of transposase mRNA. (A) The structures of the XIG8A and SAGp22A insertions. Arrows indicate directions of genes. Arrowheads indicate primers used to detect excision. (B) A scheme for mobilization of Tol2. Transposase mRNA is injected into fertilized eggs. Excision and integration at new loci occur in the germ lineage of the injected fish, and the original insertion, excision footprints (either precise or imprecise), and/or new insertions are transmitted to the progeny. (C) PCR analysis of excision of Tol2. XIG8A and SAGp22A homozygous fish injected with transposase mRNA were crossed with respective homozygous fish, and the progeny (lanes 1–12) were analyzed by PCR. M, maker; N, negative control (no DNA); P, positive control (wild-type fish). Arrowheads indicate PCR products created upon excision of Tol2 from the XIG8A and SAGp22A locus. (D) Southern blot analysis of new Tol2 insertions by using the ³²P-labeled GFP probe. XIG8A and SAGp22A homozygous fish injected with transposase mRNA were crossed with wild-type fish, and the progeny (lanes 1–12) were analyzed. ori, parental fish that carry the original insertion. Arrowheads indicate the positions of the original insertions.

The SAGp22A gene trap line expresses GFP in somites. In the course of crosses between the injected SAGp22A fish and wild-type fish, we found embryos that showed different GFP expression patterns. These patterns were thought to result from gene trap events caused by integrations of T2KSAG at new loci. To confirm this notion, we crossed 34 SAGp22A-injected fish with wild-type fish, observed 3967 F1 embryos, and identified embryos with nine new GFP expression patterns. We analyzed these fish and confirmed that these patterns were generated by integrations at new loci (Table S1). In addition, we found that 8.5% (337/3967) of the F1 embryos showed no GFP expression. We raised 13 of these embryos and analyzed the 13 embryos by Southern blot hybridization and identified a loss of the original insertion and six new bands that were thought to be silent (data not shown). We cloned four of them (Table S1).

Isolation of Reversions from the nup214 Insertional Mutation.

Because we found a single-copy Tol2 can be excised efficiently by injection of transposase mRNA, we aimed to apply this method to create revertants. We performed a screen by using the T2KgSAG gene trap construct and isolated the gSAG37B insertion that was located within an intron of a zebrafish homolog of the mammalian nup214 gene, which encodes a component of the nuclear pore complex (NPC) (Fig. 2A) (20). The gSAG37B insertion reduced the nup214 transcript (data not shown), and the homozygous embryos showed a recessive lethal phenotype (Fig. 2F). Because the mutant phenotype became obvious 4 days post fertilization (dpf), it was difficult to prove the causality of the phenotype either by inhibition of Nup214 by microinjection of morpholino or by rescue of the mutant phenotype by injection of the nup214 mRNA into fertilized eggs. We injected transposase mRNA into embryos obtained from heterozygous gSAG37B parents (F0), raised five GFP-positive F1 fish, and crossed these five GFP-positive F1 fish with heterozygous gSAG37B fish to obtain GFP-positive F2 progeny. If the gSAG37B insertion was excised in the germ lineage of F1 fish, the F2 fish should carry either a wild-type allele or a Tol2-excision allele at the nup214 locus. Previously, we showed that Tol2 executed both precise and imprecise excision, and imprecise excision was observed in half of the cases (21). Also, in the excision experiment above, imprecise excision was observed in the progeny from one of the three XIG8A founder fish (data not shown). Because precise excision is indistinguishable from the wild-type sequence, we designed a primer that matched perfectly with the wild-type sequence of the integration site to detect imprecise excision (Fig. 2B). In 3%–14% of F2 embryos from three of the five F1 fish, the PCR product was not amplified, suggesting that imprecise excision occurred in these embryos (Fig. 2 C and D). We analyzed these PCR products by DNA sequencing and identified a complete loss of the gSAG37B insertion and addition of several nucleotides that featured imprecise excision (Fig. 2B). We raised and crossed the F2 fish with the imprecise excision allele (rev2). Homozygous rev2 embryos were normal (Fig. 2G). From these results, we concluded that the embryonic lethality observed in the gSAG37B homozygous embryo was indeed caused by the transposon insertion. Thus, mobilization of a Tol2 insertion can be used to prove the causality of a mutant phenotype.

Fig. 2.

Isolation of reversions from the nup214 insertional mutation. (A) The structure of the gSAG37B insertion. An arrow indicates the direction of transcription. Arrowheads indicate primers to detect imprecise excision (gSAG37BL and gSAG37BEX). (B) DNA sequences of the wild-type gSAG37B insertion allele and reversions created upon excision of the gSAG37B insertion. An arrow indicates the direction and position of the gSAG37BEX primer. A box indicates Tol2. Bold characters indicate the 8-bp target site duplication and excision footprints. (C) PCR analysis to detect reversions. The gSAG37B fish injected with transposase mRNA were crossed with heterozygous gSAG37B fish, and the F2 progeny (lanes 1–16) were analyzed by PCR using gSAG37BL and gSAG37BEX. In lane 5, the PCR product was not amplified (rev2). N, negative control (no DNA); P, positive control (wild type). (D) The numbers of PCR-negative fish per F2 fish analyzed. From #2 and #5 F1 injected fish, rev2 and rev5-1 and -2 were identified, respectively. (E–G) Side views of embryos at 5 dpf. (E) The wild-type embryo. (F) gSAG37B homozygous embryo. (G) rev2 homozygous embryo.

Construction of the Fuji Transgenic Fish Carrying the Heat-Shock Inducible Transposase Gene.

To supply the transposase activity in vivo, we constructed transgenic fish lines, Fuji28 and Fuji70. These lines carried the transposase cDNA under the control of the hsp70 promoter and the DsRed expression cassette as a dominant marker (Fig. 3 A and B). First, we treated the heterozygous Fuji embryos with heat shock at the 70% epiboly stage or at 24 hours post fertilization (hpf). The transposase mRNA was increased upon heat shock (data not shown). Second, heterozygous Fuji70 embryos were injected with a plasmid containing a Tol2 construct T2KXIGΔin (22) at the one-cell stage and were treated with heat shock at the 70% epiboly stage. DNA prepared from these embryos at 24 hpf were analyzed by PCR to detect excision of T2KXIGΔin from the plasmid. PCR products were amplified from all of the heat-shocked embryos (n = 12) but not from embryos that were not treated with heat shock (n = 2) or from wild-type embryos injected with the T2KXIGΔin plasmid (n = 13). Third, we tested whether a Tol2 insertion on the genome can be excised by heat shock. For this purpose, we treated double-transgenic embryos homozygous for the SAGp22A insertion and heterozygous for the Fuji70 transgene with heat shock at 30 hpf and analyzed them by PCR. Excision of Tol2 from the hoxc locus was detected in a pool of 25 DsRed-positive embryos but not in a pool of 25 DsRed-negative embryos or in a pool of 25 DsRed-positive embryos that were not treated with heat shock. We further analyzed 40 DsRed-positive embryos individually by PCR and detected excision in all of them. From these results, we concluded that the transposase activity is very low at the basal level and can be induced upon heat shock in the Fuji transgenic fish.

Fig. 3.

Heat-shock induced transposition in the Fuji;HG6D double-transgenic fish. (A) The structure of the hspTP/efDsRed construct used to create Fuji transgenic fish. The arrows indicate DsRed gene and transposase cDNA. (B) The DsRed expression in the Fuji70 fish at 1 dpf. (C) The GFP expression pattern in the HG6D enhancer trap line at 5 dpf. (D) A scheme for induction of transposition. The Fuji70;HG6D double heterozygous fish were incubated in a water bath and mated, and the progeny were analyzed for GFP expression. (E, F) New GFP expression patterns in the progeny from heat-shocked Fuji70;HG6D parents. (G) Southern blot analysis using the ³²P-labeled GFP probe. Lanes 1–12: F1 progeny with new GFP expression patterns from the heat-shocked Fuji70;HG6D fish. ori, the HG6D fish. An arrowhead indicates the original HG6D insertion.

Heat Shock-Induced Transposition in the Germ Lineage.

To mobilize Tol2 in vivo, we constructed double-transgenic fish heterozygous for the transposase transgene and a single-copy insertion of the T2KHG enhancer trap construct (HG6D or HG21B, Fig. 3C) (7). It is expected that different GFP expression patterns will be detected when T2KHG is transposed to new loci and the GFP gene is activated by nearby enhancers. First, the Fuji70;HG6D and Fuji70;HG21B embryos at the 70% epiboly stage were treated at 38 °C for 30 min and were raised and crossed with wild-type fish. In total, we observed 1108 embryos. Of these embryos, 670 were GFP positive, but none showed new patterns. Then the Fuji28;HG21B embryos at 24 hpf were treated at 37 °C for 3 h, 6 h, or 12 h and were raised and crossed with wild-type fish. We observed 874 embryos (348, 121, and 405 for the 3-h-, 6-h- and 12-h heat shock treatments, respectively). Of these embryos, 597 were GFP positive, but none showed new patterns. These results indicate that the heat-shock treatment at the embryonic stages can induce Tol2 excision in somatic cells but cannot induce transposition in the primordial germ cells at detectable levels. Supply of a large amount of mRNA at the early cleavage stages, as has been done by microinjection to fertilized eggs, may be required for the creation of transposon insertions in the germ lineage during embryogenesis. Therefore, we needed to develop another strategy.

We tested whether the heat-shock treatment at the adult stage can cause transposition in the germ cells (Fig. 3D). Five pairs (pairs 1–5, Table 1) of the Fuji70;HG6D heterozygous fish were heat-shocked repeatedly up to 26 times over 60 days (Table 1). During this period, the pairs were crossed on days 9, 17, 30, 31, 37, 45, and 60 after the first heat shock, and their embryos were analyzed for GFP expression (Table 1). In a total of 1036 embryos obtained on the day 9 (after six heat-shock treatments), we could not find any new GFP expression patterns (Table 1). However, in a total of 387 embryos obtained on the day 17 (after 11 heat-shock treatments), we found five new patterns that differed from the original HG6D pattern (Fig. 3 C, E, and F and Table 1). Furthermore, embryos with new patterns were always detected in all the crosses after day 30 (i.e., after 19 heat-shock treatments) (Table 1). On days 45 and 60, 6.2% (21/339 and 20/323, respectively) of the progeny showed new patterns (Table 1).

Table 1.

Progeny with new patterns after multiple heat shocks

hs*	6	11	19	20	22	25	26
days†	9	17	30	31	37	45	60
Pair 1‡	0/200	3/93		3/66		5/60	4/85
Pair 2‡	0/180	0/80		4/96	6/23	3/93	5/78
Pair 3‡	0/190	0/69	1/51			2/87	4/54
Pair 4‡	0/204	1/98	3/82			11/99	7/106
Pair 5‡	0/262	1/47

*Number of heat shock treatment.

^†Number of days after the first heat shock treatment.

^‡Number of embryos with new patterns per total number of embryos observed.

To prove that these new patterns indeed resulted from integrations at new loci, we analyzed 34 fish with new patterns by Southern blot hybridization. All of them had new bands (Fig. 3G). In total, 40 new bands were detected. Eighteen fish had single insertions, indicating that these insertions were responsible for the observed GFP expression patterns. We analyzed genomic DNA surrounding 24 of these insertions, including 13 single insertions, and confirmed integrations at new loci (Table S1). Although insertions derived from the same heat-shocked fish were analyzed, none of them were the same, indicating that the germ cells of the heat-shocked fish were highly mosaic.

Transposition Occurred Predominantly in the Male Germ Cells.

Did the heat shock induce transposition in the germ cells only in male fish, only in female fish, or in both? To address this question, the four males and four females treated with heat shock 26 times were crossed with wild-type fish of the opposite sex. We analyzed 485 and 503 embryos for males and females, respectively, and identified 38 new patterns. Ninety-five percent (36/38) of the new patterns were found in the progeny from the males, indicating that transposition occurred predominantly in the male germ cells (Table 2). We analyzed genomic DNA from 15 fish with new patterns and identified 18 new integration sites (Table S1), all of which were different.

Table 2.

Progeny with new patterns from males and females

Pair	Male*	Female*
1	8/130	0/50
2	9/132	1/204
3	12/141	0/118
4	7/82	1/131
Total	36/485	2/503

*Number of embryos with new patterns per total number of embryos observed.

Finally, we tested whether a single heat-shock treatment can cause transposition in the male germ cells. We treated four Fuji70;HG6D heterozygous males with heat shock once and crossed them with wild-type females. In the crosses performed 3 days after the heat shock, no new patterns were detected (0/429; Table 3). In the crosses performed 10 days after the heat shock, a new pattern was detected (1/604; Table 3). We also treated four Fuji70;HG6D double-heterozygous females with heat shock once and found no new patterns in their progeny. We analyzed 11 fish with new patterns and identified 11 new insertions (Table S1). These results showed that a single heat-shock treatment can induce transposition in the male germ cells, but the frequency of transposition was lower than with multiple heat shocks.

Table 3.

Progeny with new patterns after a single heat shock

Days*	3	10	15	20–51 (total)
Male 1†	0/108	0/80	1/257	4/869
Male 2†	0/120	0/76	0/183	0/622
Male 3†	0/117	1/201	0/40	4/677
Male 4†	0/84	0/247	0/179	1/924

*Days after the heat shock treatment.

^†Number of embryos with new patterns per total number of embryos observed.

Global Hopping vs. Local Hopping.

In the present study, we cloned and sequenced 82 new integration sites: 29 for injection of transposase mRNA, 42 for multiple heat shocks, and 11 for a single heat shock. To characterize further the distribution of integration sites on the genome, we treated three male fish with heat shock repeatedly, crossed them with wild-type fish, collected embryos with new patterns, and cloned and sequenced 26 integration sites. In these sequences, we found for the first time that two different embryos from the same heat-shocked fish had the same insertion. Thus, we cloned a total of 107 integration sites and analyzed them by using the Ensembl database (http://www.ensembl.org). Sixty-five sites were mapped on the zebrafish genome (Zv7); 11 (17%) were mapped on the same chromosome as the donor chromosomes, and 54 (83%) were mapped on different chromosomes (Table S1).

Two F1 fish, HG6D-SH-m1–2 and HG6D-SH-m1–3, which were obtained from the Fuji70;HG6D male fish treated with a single heat shock and crossed with wild-type female fish, carried new insertions located 260 kb and 1.8 Mb away from the original locus, respectively, and retained the original HG6D insertion. To test whether these local hops were located on the donor chromosome, we crossed these F1 fish with wild-type fish and analyzed their progeny. Forty-nine percent (66/135) of F2 embryos from HG6D-SH-m1–2 were GFP positive, 65 showed both the original and new patterns, and one showed only the new pattern. Fifty-six percent (24/43) of embryos from HG6D-SH-m1–3 were GFP positive; 100% (24/24) showed both the original and new patterns. In these fish, the new and original insertions were tightly linked, indicating that these transposon insertions were located on the same chromosome.

Discussion

Mobilization of Tol2 by Microinjection of Transposase mRNA.

In the present study, we showed that Tol2 insertions at two different loci, pax6b (XIG8A) and hoxc (SAGp22A), can be mobilized by microinjection of transposase mRNA. This finding indicates that Tol2 can transpose efficiently from a single-copy donor irrespective of its chromosomal location. In one case, all 12 of the progeny fish analyzed had the same new insertion, suggesting that transposition may have occurred at the one-cell stage, immediately after injection of the transposase mRNA. We propose that this method can be used to isolate revertants from an insertional mutation. When Tol2 is excised, imprecise excision occurs in half of the cases (21). We demonstrated here that it is possible to design PCR primers that do not work for the sequence created by imprecise excision. Also, when a polymorphic marker is available near the integration site, it also is useful to distinguish an excision allele from a wild-type allele. It is noteworthy that, although in the course of this analysis we designed primer pairs to detect deletions of ∼ 1 kb around the integration site and analyzed ∼ 100 embryos injected with the transposase mRNA (data not shown), we could not detect any such deletions, suggesting that Tol2 excision rarely causes deletions.

The nup214 gene encodes a component of the vertebrate NPC (20). In humans, the CAN/Nup214 gene was identified as an oncogene (23), and in mouse a knockout mutation caused embryonic lethality (24). Our present study proved the zebrafish mutant phenotype was indeed caused by the transposon insertion, confirming an essential role of the nup214 gene in embryogenesis. In zebrafish, another component of the NPC, nup205, was disrupted by retroviral insertional mutagenesis (25). It is interesting that the nup205 mutant phenotype is similar to nup214. These mutants will be useful in studying the roles of NPC during vertebrate development.

Mobilization of Tol2 by the Heat-Inducible Transposase.

We demonstrated that the heat-shock treatment of the Fuji70;HFG6D double-transgenic fish at the adult stage can induce transposition in germ cells very efficiently. Although we detected the basal transposase transcription in transgenic fish by RT-PCR before heat shock (data not shown), we could not identify any new insertion events in the progeny without heat shock (0 of 3447 embryos analyzed in this study), indicating that this system is highly regulatable. It should be noted that this estimate is based mainly on the detection of new trap patterns, and it is possible that a small number of new transposon insertions that did not cause GFP expression or that showed GFP expression patterns that overlapped with the original pattern may have been uncounted.

After multiple heat shocks, ∼ 6% to 7% of the progeny from males showed new patterns that corresponded to transposition of the Tol2 enhancer trap construct to new loci. By using the same construct, we previously had estimated that ∼ 70% of insertions gave rise to specific GFP expression patterns (7). Therefore it is estimated that 100 embryos obtained from one heat-shocked Fuji transgenic male will contain transposon insertions at 10 new loci. We also had estimated previously that one fish injected with a Tol2-donor plasmid and the transposase mRNA transmits on average approximately three different insertions to the progeny (5). Thus, our present system can create a larger number of insertions more efficiently and enables a small laboratory to perform a large-scale insertional screen.

Moreover, this system has the following merits. First, the germ cells of heat-shocked males are highly mosaic, and they continuously generate new insertions. Second, the number of insertions detected in single transgenic progeny was small, 1.5 on average, and more than half of the progeny (53%; 18/34) carried single insertions, whereas ∼ 2.8 insertions per fish were found in single progeny created by the co-injection method (5). This method will ease the establishment of fish lines carrying single insertions and the cloning of integration sites by PCR-based techniques. A possible disadvantage of this approach is that each pattern would be expected to manifest in only one fish because of the high mosaicism. Therefore, it is necessary to keep and raise carefully F1 fish with patterns of interest.

Transposition in the Male Germ Cells.

We found that transposition occurred predominantly in male germ cells. Why is the transposition activity high in the male? In other words, why is it low in the female? At present, the reason for this observation is unknown. We observed a six- to eightfold increase of the transposase mRNA upon heat shock in both the ovary and testis (data not shown). Therefore, we speculate that differences in expression levels may not be the cause of the difference in transposition activity; rather, transposition itself may be suppressed in female germ cells by an unknown mechanism, or female germ cells may be more susceptible to DNA damage that might be caused by transposition.

Does transposition in germ cells affect target sites of Tol2? By using the same enhancer trap construct that we used in this study, we previously had created genomic integrations by co-injection of a donor plasmid and the transposase mRNA into fertilized eggs (7). With this method, transposition is thought to occur in early cleavage stages, before zygotic transcription starts. In the earlier study, 22 of 53 integration sites (43%) were located within transcribed regions. In the present study, the same construct was mobilized in germ cells, and 19 of 44 mapped integration sites (43%) were located within transcribed regions (Table S1). Thus, the frequency of transposition was similar in the two studies. Currently the numbers are too small to determine whether specific types of genes are enriched in the different transposition methods (e.g., whether genes expressed in germ cells are enriched in the present method), but it will be interesting to address this question when more data are available.

We found two cases in which F1 fish contained a new insertion together with the original insertion on the same chromosome. Because Tol2 transposes by a cut-and-paste mechanism, this finding indicates that the new Tol2 insertion was transposed from a sister chromatid. Transposition from a replicated chromatid has been described in the case of Ac, another transposon of the hAT family, in plant cells (26). Also, we found one case in which two fish derived from one heat-shocked male had the same insertion, indicating cell division occurred after transposition. Taking these findings together, we think that transposition is likely to occur in spermatogonia. Furthermore, we could not detect any new transposition events in the progeny born before the day 9 after the first heat-shock treatment, suggesting that it may take ≈ 9 days for such transposition-competent spermatogenic cells to differentiate into the sperm in zebrafish. When male fish were mutagenized with N-ethyl-N-nitrosourea (ENU), mutations in premeiotic germ cells were detected in their progeny 2–4 weeks after the ENU treatment (27). When mitotic male germ cells were treated with androgen in vitro, flagellated sperm appeared 9 days after the treatment (28). Our results are consistent with these observations.

Transposition from a Single-Copy Donor.

We demonstrated that Tol2 can transpose efficiently from a single-copy donor. This finding contrasts with the SB transposition system in which transposon donor loci contained concatemers of 20–100 copies of the SB construct (12–14). It has been shown that the transposition frequencies of SB from such concatemeric donors are high, whereas those from single-copy donors are very low (29, 30). An explanation for this difference might be that the SB transpose preferentially catalyzes transposition from methylated donors (31) and the concatemeric donors are likely to be heavily methylated. Because our present study revealed that Tol2 has a distinctive feature, it should be a suitable tool for developing an in vivo transposition system from a single-copy donor. It has also been reported that transposition from concatemeric donors often causes complicated rearrangements at the donor loci together with unwanted mutant phenotypes known as “donor-site effects” (32, 33). Transposition from a single-copy donor should overcome these problems.

Local Hopping vs. Global Hopping.

In the present study, we found that 17% of new integration sites of Tol2 were located on the donor chromosomes (local hopping). Thus, transposition of Tol2 is rather global in comparison with the greater frequency of local hopping (60%–80%) observed in the in vivo SB transposition systems in mice (12, 15, 34). It can be argued that the reduced local hopping in our system may be a consequence of multiple transposition events caused by repeated heat-shock treatments. To test this possibility, we carried out a mobilization experiment using a single heat-shock treatment and found that Tol2 still showed less local hopping (20%; 2/10). Although it is possible that multiple transposition events occurred during the single heat-shock, we have observed lower frequencies of local hopping (14%–20%) throughout the present study, regardless of donor loci (XIG8A, SAGp22A, and HG6D) or methods used to supply the transposase (microinjection of mRNA, multiple heat shocks, or a single heat shock). Several factors might contribute to the frequency of local hopping: the transposon itself (Tol2 vs. SB), the structure of the donor locus and the difference in methylation status (single copy vs. concatemers), or host factors involved in transposition (zebrafish vs. mouse). It is interesting to note that local hopping from a single-copy Ac donor introduced in tobacco culture cells by transferred DNA (T-DNA) mediated transgenesis occurred with varying frequency (20%–70%) in different cell lines (35). This finding suggests that chromosomal positions, transposon structures, methylation status, or combination of these factors may affect the frequency of local hopping. Further studies along this line will open a new field in transposon biology in vertebrates. Although Tol2 is suitable for creating genome-wide insertions in vivo, the Tol2 transposition system also will allow us to perform region-specific mutagenesis, because we found that 6 of 11 local hops were located very close to the donor locus, within 300 kb.

A number of different types of Tol2 transposon vectors for gene trapping and enhancer trapping, including these encoding the Gal4 transcription activator, have been developed. The heat-shock inducible Fuji transposition system should facilitate the generation of a large number of insertions of such constructs. Also, our present study revealed unique and useful features of the in vivo Tol2 transposition system. We believe that Tol2 will be a useful tool for developing in vivo transposition systems in other animals as well.

Materials and Methods

Fish Lines and Primers.

The XIG8A, SAGp22A, HG6D, and HG21D fish were constructed by Tol2-mediated transgenesis and contain single-copy insertions of T2KXIG (GFP expression construct), T2KSAG (gene trap construct), and T2KHG (enhancer trap construct), respectively (Figs. 1A and 2A) (5, 7). The gSAG37B fish was isolated from a gene trap screen using T2KgSAG that contained a splice acceptor from the Zebrafish gata6 gene and the GFP gene. TL and TAB were used as wild-type fish.

Microinjection of Transposase mRNA, Excision Assay, and Primers.

Transposase mRNA was synthesized in vitro by using pCS-TP as described previously (5). Approximately 1 nl of 25 ng/μl mRNA was injected to fertilized eggs. To detect excision and reversions, PCR was carried out as described previously (36) by using primers listed in Table S2. To detect excision at the XIG8A or SAGp22A locus, XIG8AL and XIG8AR or SAGp22AL and SAGp22AR were used. gSAG37BL and gSAG37BEX were used to detect imprecise excision. gSAG37BL and gSAG37BR were used to amplify the empty site. TYR1 and BS1 were used to detect excision of T2KXIGΔin.

Construction of the Fuji Transgenic Fish Expressing Transposase.

The hspTP/efDsRed construct contains the zebrafish hsp70 promoter (a kind gift of Dr. Kuwada), the transposase cDNA, and the DsRed expression cassette (Fig. 3A). The linearized hspTP/efDsRed plasmid DNA was injected into fertilized eggs. Seventy injected fish were raised and crossed, and DsRed-positive progeny were identified from four injected fish. We named the transgenic fish “Fuji” for “Furo (hot bath) utilizing jump inducer.” Two lines, Fuji28 and Fuji70, both carrying two copies of hspTP/efDsRed at single loci as revealed by Southern blot analysis, were established. In addition to these fish, we constructed two types of transgenic fish carrying the transposase cDNA downstream of the CAG promoter and the EF1α-DsRed expression cassette and carrying the transposase cDNA downstream of the EF1α promoter and the DsRed gene downstream of the CMV promoter. However, the DsRed expression in these transgenic fish was silenced after the F2 generation by unknown causes. In contrast, the DsRed expression in the Fuji28 and Fuji70 lines has been observed from generation to generation. Therefore we used Fuji transgenic fish in this study.

Heat-Shock Treatment of Embryos and Adults.

Embryos at the 70% epiboly stage were incubated in 500 μl of E3 buffer at 38 °C for 30 min on Block Incubator BI-516S (Astec). Embryos at 24 dpf were incubated in 10 ml E3 buffer at 37 °C for 3 h, 6 h, or 12 h in an air incubator (MIR-162, Sanyo). After the heat-shock treatment, the embryos were transferred to E3 buffer at 28 °C. The heat-shock treatment of the adult fish was performed by warming fish in a water bath (SJ-10R,TAITEC) at 37 °C for 1 h. The adults were treated by heat shock no more than once per day.

Southern Blot Hybridization and Adaptor-Ligation PCR.

Southern blot hybridization using ³²P-labeled GFP DNA probe was carried out as described (36). Junction fragments of genomic DNA and Tol2 were cloned by adaptor-ligation PCR (22) and sequenced by using BigDye terminator v3.1 cycle sequencing kit and ABI PRISM 3130 (Applied Biosystems).