In this study, we analyzed a genome-wide reintegration of the non-autonomous transposable element Tol2 in zebrafish when it was remobilized from two different donor sites. We showed that the genomic Tol2 copy can be remobilized upon injection of transposase mRNA into the germlines of up to 48% of founders. Since we selected only those Tol2 reintegrations in germlines that caused changes in GFP expression, we in fact measured the "apparent transposition rate"; the actual germline transposition frequency during in vivo remobilization would be higher.
We analyzed Tol2
integration sites with respect of their chromosomal distribution, integration into intragenic regions and insertion site sequence specificity. Although novel integration sites were found on different chromosomes, Tol2
reintegration was not random. Almost 39% of transposon integrations were found within known or predicted genes. Most of them were found within introns, as expected in view of the high intron/exon ratio in the zebrafish genome. If we consider insertions into the regulatory regions adjacent to transcriptional initiation and termination sites, the rate of Tol2
transposition into genes was even higher. Since we used the TAIL-PCR method to isolate transposon inserts, we could not recover all possible Tol2
insertions in the genome. However, despite of using enhancer trap approach, the frequency with which Tol2
was integrated within intragenic regions was similar to that found for the SB
transposon in human (39%) and mouse (31%) cells [31
] and for Ac
in rice (30%) [32
] and Arabidopsis
]. Therefore, Tol2
is as prone to integrate into transcriptional units as other DNA transposons.
Our results further demonstrated that about 15% of Tol2
reintegrations from a specific donor site were linked to the same chromosome. Such behavior was also noticed in [11
], where about 18% of the mapped integration sites (6/34) were located on the donor chromosome. We found that about one third of intrachromosomal reintegrations were located within 1 Mb of the donor site. However, since we selected reintegrations on the basis of new GFP expression patterns, this number is likely to be lower than the actual number of such transpositions (for example, closely-linked transpositions may retain the GFP expression pattern of the donor). This local hopping phenomenon has been described for other DNA transposons [19
]. For example, SB
is mostly re-integrated within 3 Mb of the donor site [19
] and the local hopping interval of the P
element is within 100 kb [36
]. Local hopping was also found for the hAT family. In this case, more than half of Ac
transposon reintegrations occurred within 1.7 Mb of the donor site [37
]. Overall, a linked reintegration property of the Tol2
system might be beneficial for setting a region-specific saturation ET screen.
Interestingly, our analysis revealed that some chromosomes other than the donor were somewhat preferred targets for Tol2
integration; still others appeared to be disfavored. Such transposon behavior may reflect the spatial chromosomal architecture within the nucleus, if multiple non-adjacent chromosome segments are closely juxtaposed at the nuclear interior or periphery [41
]. There are many examples of correlations among the intranuclear positions of genes, their clusters and genetic activities, whereas the relative positioning of chromosomes seems to be maintained (reviewed in [44
]). Therefore, such a property of transposons may potentially be used for analyzing the spatial organization of the genome.
We found that Tol2
differentially targeted the different classes of endogenous repetitive elements. For example, it more frequently targeted DNA transposons than retrotransposons and tandem repeats. The latter tendency contrasts with the profound preference of Tc1/mariner
transposons for TA-containing microsatellite DNA [31
]. Expansion of such repeats during replication slippage can cause repeat instability and increased recombination rates (reviewed in [46
]), suggesting that these transposons may use the recombination machinery during integration. Like the SB
also avoided retrotransposons containing LTRs. The differences in targeting of endogenous repetitive elements may reflect differences in the copy numbers of each class of repeats, as well as differences among the mechanisms of integration utilized by each transposon family.
We also found that Tol2
was prone to integrate into AT-rich DNA regions and that a target site contained a weak palindrome-like consensus sequence. An AT-rich palindromic consensus has previously been found in the target sequence of Tc1/mariner
transposons such as SB
in human and mouse cells [31
] and the Tc1
element in worms [47
]. However, in contrast to the strict preference of SB
for TA dinucleotide targets, Tol2
has no such preference at the nucleotide level. There is some evidence that distinct preferred transposon integration sites may not necessary match consensus sequences, but rather share similar structural patterns [48
]. DNA structural characteristics such as bending and protein-induced deformability play an important role in directing DNA integration [28
]. DNA bending can lead to changes in the width and depth of the major and minor grooves, affecting a protein's access [49
]. AT-rich palindromes are particularly susceptible to local melting and have been experimentally shown to adopt a bendable DNA structure [50
]. In addition, palindromic sequences have the potential to form cruciform configurations, which are an efficient target for RAG-mediated transposition [51
]. Also, AT-rich palindromic repeats are known to be double-strand break hotspots. It has been proposed that DNA bending plays a role in the integration specificity of the hobo
transposable element from the hAT family, but hobo
has no strict preference for targeting at nucleotide level [39
]. This suggests the likelihood that the target site selection of Tol2
is primarily determined at the level of DNA structure, not sequence.
element transposes by "cut-and-paste" mechanism, which involves the excision and re-integration of the transposon from one site to another, creating an 8-bp duplication of the integration site [3
]. Previously, we found that in one third of Tol2
excision events, reparation of donor site results in different footprints [13
]. In our experiments we used extremely high amounts of transposase mRNA (around 9 × 107
molecules per a single copy of transposon), therefore it may be reasonable to expect multiple "cut-and-paste" events before transposon will finally settle. Such multiple hops could generate double-strand breaks and, as a consequence, the footprints. Our analysis of DNA sequences flanking the integration/target site revealed no signs of the footprints, at least, at the vicinity (up to 900 bp) of new integration sites. All DNA sequence modifications found at these regions exhibited DNA sequence polymorphism between zebrafish strains (data not shown). However, we could not rule out the possibility that the footprints left after multiple "cut-and-paste" events may be found far away from integration sites.