Conservation of Palindromic and Mirror Motifs within Inverted Terminal Repeats of mariner-like Elements

doi:10.1016/j.jmb.2005.05.006

J. Mol. Biol. (2005) 351, 108–116

Conservation of Palindromic and Mirror Motifs within Inverted Terminal Repeats of mariner-like Elements Yves Bigot*, Benjamin Brillet and Corinne Auge´-Gouillou* Laboratoire d’Etude des Parasites Ge´ne´tiques, Universite´ Franc¸ois Rabelais, E.A.3868 UFR des Sciences et Techniques Parc de Grandmont, Avenue Monge, 37200 Tours, France

The transposase of the mariner-like elements (MLEs) specifically binds as a dimer to the inverted terminal repeat of the transposon that encodes it. Two binding-motifs located within the inverted terminal sequences (ITR) are therefore recognized, as previously indicated, by biochemical data obtained with the Mos1 and Himar1 transposases. Here, we define the motifs that are involved in the binding of a MLE transposase to its ITR by analyzing the nucleic acid properties of the 5 0 and 3 0 ITR sequences from 45 MLEs, taking into account the fact that the transposase binds to the ITR, using its CRO binding domains and the general characteristics of the cro binding sites so far investigated. Our findings show that in all the MLE ITRs, the outer half was better conserved than the inner half. More interestingly, they allowed us to characterize conserved palindromic and mirror motifs specific to each “MLE species”. The presence of the palindromic motifs was correlated to the binding of the transposase dimer, whereas the properties of the mirror motifs were shown to be responsible for the bend in each ITR that helps to stabilize transposase–ITR interactions. q 2005 Elsevier Ltd. All rights reserved.

*Corresponding authors

Keywords: transposon; DNA binding; DNA bending; repeat; evolution

Introduction A significant proportion of prokaryotic and eukaryotic genomes consist of transposable elements, which are short DNA sequences that are able to move around within the genome. They are divided into two main classes, depending on their mechanism of transposition. Class-I contains the retrotransposons, which use an RNA intermediate to move, and encode at least a reverse transcriptase. Class-II consists of the transposons, which code for a transposase (Tnp), and use a DNA intermediate to transpose. The mariner-like elements (MLEs) are class-II transposons that belong to the Tc1-mariner superfamily.1 They move within genomes by means of a “cut and paste” mechanism that duplicates a TA dinucleotide at the insertion site. mariner elements consist of a DNA fragment of about 1280 base-pairs (bp) that contains a single open reading frame with no intron that encodes the transposase. This protein is able to carry out all the transposition

Abbreviations used: ITR, inverted terminal repeat; MLE, mariner-like elements; Tnp, transposase. E-mail addresses of the corresponding authors: [email protected]; [email protected]

steps, and is therefore both necessary and sufficient for mobility in the absence of host factors.2,3 The open reading frame coding the Tnp is flanked by two short inverted terminal sequences (ITRs) of about 28–30 bp that are imperfectly conserved, most of them displaying the 5 0 -YYAGRT-3 0 consensus at their outer extremities. Functional analyses of the mariner Mos1 Tnp have revealed that it contains a DNA-binding domain located within its first 120 amino-terminal residues,4,5 which very specifically binds the ITR.6 In silico analyses of this domain have shown that it has a helix-turn-helix (HTH) structure between positions 83–109. This HTH structure is related to that found within the cro-like proteins.7–9 Most of the softwares used to predict secondary structures also indicate that this domain contains three a-helices located between positions 19–35, 44–52 and 72–81.4,10 The Mos1 Tnp carboxy-terminal domain contains a conserved triad [D, D34D] that forms the catalytic core of the DNA cutting and strand-transfer reactions necessary to ensure the transposition process.11 The Tnp also contains three regions involved in the dimerization and the tetramerization of the active Tnp bound to the ITRs.12 These regions are localized within the first 143 amino acid residues.

0022-2836/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.

109

Sequence Organization of mariner ITR

Figure 1. Patterns of conservation of the nucleic acid sequence, and the presence of palindromic and mirror motifs within the DNA binding sites of the l and cro repressors ((a) and (b)), and some of the CRP proteins ((c) and (d)). Majority rules with a significance threshold of at least 75% were used to verify the consensus sequences of both binding sites. In palindromic and mirror motifs, strictly conserved residues are white typed and boxed in black, whereas those conserved as a purine or a pyrimidine nucleotide are just boxed in gray. (I) Indicates the locations of the most important residues involved in motifs used as DNA binding sites by l and cro repressors ((a) and (b)) or the CRP protein ((c) and (d)). In (a) and (c), arrows indicate the nucleotides corresponding to the symmetry axis of the palindromic motifs. In (b) and (d), single or double arrows indicate the nucleotides or dinucleotides corresponding to the anti-symmetry axis of the mirror motifs. The name of each binding-site is indicated on the left: OR 1–3 and OL 1–3 refer to six operator sequences contained in the l phage genome. nZA, C, G or T; yZT or C, rZA or G.

We have recently shown that the initiation of the mariner Mos1 transposition depends on the elaboration of a nucleoprotein complex, known as the synaptic complex.13 This complex consists of two ITRs held together by a tetramer of Tnp, and is designated PEC2. The assembly of PEC2 requires the preliminary formation of a simpler complex, containing one ITR and two Tnp molecules, which is designated SEC2. In the light of the formation of SEC2 and PEC2, we were able to demonstrate the presence of two binding sites for the Mos1 Tnp within a single ITR. We also observed that the sequence of the Mos1 ITR contains overlapping palindromic and mirror motifs, which may explain why two Tnp molecules are bound “side by side” on the same ITR. The occurrence of two Tnp binding sites within a single 30 bp ITR is not a specific characteristic of Mos1. Using an experimental approach involving copper phenanthroline footprints, Lipkow et al.14 have recently demonstrated the presence of two protected 6–7 bp areas within the Himar1 ITR that correspond to two Tnp binding sites centered around positions 10 and 21. These binding sites had hitherto remained undetected, due to the lack of sensitivity of the methods previously used.2,4,6 Using UV crosslinking assays and electrophoretic mobility shift assays involving truncated ITR, we have recently obtained similar results with the Mos1 ITR.13 Here we show that MLE Tnps not only have a crolike DNA binding domain (the HTH structure), but also share a cro-like DNA binding pathway: a protein dimer that interacts with a “bipartite” DNA target. Like the extensively studied DNA binding sites of three cro-like proteins, the l repressor, the CRO repressor and the Escherichia coli cAMP receptor protein (CRP), MLE ITRs showed similar structural properties, although

their nucleic acid sequences were completely different from each other. They contain a palindromic motif that is the binding target of the Tnp dimer, each monomer interacting with half of the palindrome. They also display a mirror structure that stabilizes the Tnp–ITR interactions. Overall, our results show that the organization of the information in the MLE ITR is entirely different from those found in the TLE ITRs.1

Results Properties of cro-like protein DNA binding sites In an attempt to improve our understanding of the conservation rules within the cro-binding sites, the sequences of the l and CRO repressors9 were extracted from the literature, and that of CRP from the web site†. They were aligned, and then analyzed in terms of nucleotide conservation and of the presence of palindromic and mirror motifs. As previously described,9 searches for a conserved palindromic organization within the DNA binding sites of the l and cro repressors have shown that only six of the 21 positions were conserved (Figure 1(a); nucleotide boxed in black and typed in white). However, our analysis revealed here that eight of the 15 “degenerated” nucleotide positions contained relaxed purine-pyrimidine conservation, that reinforced the general palindromic properties of these binding sites (Figure 1(a); nucleotide boxed in gray). Our searches also revealed that a second kind † http://www.clunet.edu/BioDev/omm/cap/ frames/captx.html

110

Sequence Organization of mariner ITR

Figure 2 (legend after part (d))

of conserved motif occurred within these sequences. Indeed, we found that mirror motifs were present with an anti-symmetry axis located at position 12, on the 3 0 side of the nucleotide that corresponded to the symmetry axis of the palindrome (Figure 1(b)). The analyses done with the consensus sequence of the CRP DNA binding sites and those known as LacP1 and GalP1 confirmed that palindromic motifs

were conserved, some of their positions being strictly conserved, whereas some of the others displayed relaxed purine-pyrimidine conservation (Figure 1(c)). The analysis of certain CRP DNA binding sites,15 such as LacP1, had also previously shown that mirror motifs were significantly conserved, whereas other sites, such as Gal P1, do not demonstrate this property (Figure 1(d)).

111

Sequence Organization of mariner ITR

Figure 2 (legend after part (d))

112

Sequence Organization of mariner ITR

Figure 2. Patterns of conservation of the nucleic acid sequence ((a) and (b)), and intrastrand inverted (c) and mirror (d) motifs within the outer 28 bp of 5 0 and 3 0 ITR from 45 MLEs belonging to five different sub-families. The ITRs are separated by —. For each sequence, the name of the MLE to which the ITR belongs and its databank accession number are indicated on the right in (a). P.D. indicates personal data, and C.P. indicates data kindly supplied by David J. Witherspoon (Salt Lake City, Utah, USA). Majority rules with a significance threshold of at least 75% in (a), and of at least 50% in (b), (c) and (d), were used to design the consensus sequences ((a) and (b)) and the conserved positions in the palindromic and mirror motifs within each ITR ((c) and (d)). In (a), residues strictly conserved in different ITRs for each sub-family are typed in white and boxed in black, whereas those conserved as a purine or a pyrimidine nucleotide are just boxed in gray. In (b), typing rules similar to those used in (a) were used to compare the 5 0 and 3 0 consensus ITRs of each sub-family. In (c) and (d), strictly conserved residues within inverted or mirror motifs contained in each ITR are typed in white and boxed in black, whereas those conserved as a purine or pyrimidine nucleotide are just boxed in gray. In (c), (I) and areas boxed in pale green locate the two areas described as Himar1 or Mos1 Tnp dimer binding sites.12,14

Sequence Organization of mariner ITR

In accordance with previously published data,9,15 our observations indicate that a general trait of the cro binding sites is the presence of palindromic motifs, each repeat ranging from eight to nine nucleotides in length. In these palindromic motifs, the nucleotide sequence is more or less conserved, depending on the binding site. We have also confirmed that numerous sites contained a second conservation level, corresponding to anti-symmetric mirror motifs that are frequently centered just beside the nucleotide in the middle of the palindrome. These mirror motifs have the same sequence conservation properties as the palindromic motifs. Finally, we observed that the conservation features of the palindromic and mirror motifs are specific to each site. Thus, they are better conserved within each site than within the consensus sequence. This indicates that the nucleic acid sequence of each binding site has been subjected to a concerted evolutionary process that has led to sequence differentiation between sites, but has maintained the palindromic and mirror conservations within each site. At the CRP binding sites, the conservation of these two kinds of motif was related to their function. Indeed, each repeat of the palindrome was shown to be bound by one CRP protein, and the mirror motifs were shown to play a role in DNA bending at these sites.15 Conserved motifs within MLE ITRs Taking into account the properties of the cro binding sites, sequence analyses were carried out in an attempt to define the features of the nucleic acid motifs putatively present within 45 MLE ITRs. Three kinds of analysis were developed. First, we analyzed the nucleic acid sequence conservation of the ITR within each MLE sub-family (Figure 2(a)). We then carried out a manual search for the best conserved palindromic (Figure 2(c)) and mirror (Figure 2(d)) motifs, keeping in mind that the ITR had to contain symmetry and anti-symmetry axes for both kinds of motif between positions 12–20. The analysis of the sequence conservation (Figure 2(a)) revealed that about 20 of the outer nucleic residues in the 5 0 and 3 0 ITRs were well conserved in MLEs belonging to the same subfamily. Conservation of the ITR sequence appeared to be less marked in two members of the irritans

113 sub-family, Bytmar1 and Hsmar2. On the basis of the analysis of the sequence similarities of their Tnp and ITR, we have recently shown that the irritans sub-family is composed of at least three different lineages.16,17 The first lineage comprises Ahmar1, Armar1, Cpmar1, Damar1, Himar1, Mpmar1 and Ocamar1, the second comprises Bytmar1 and Xtmar1 and the third Hsmar2 and numerous relatives present in the genome of various vertebrate species. The existence of these lineages is consistent with our present results. Nucleic residues located within the inner region of the ITRs were less well conserved, except for the ITR of the MLE belonging to the irritans sub-family. We also observed that some nucleic positions were differently conserved in the two ITRs of each element in the five MLE sub-families, as previously reported.5,18 These observations indicate that the ITR inner region of the MLEs belonging to the mauritiana, mellifera/capitata and elegans/briggsae sub-families, and the ITR outer region of those belonging to the irritans sub-family contain enough information to make it possible to different “MLE species” to differentiate within each sub-family.7 They also showed that only nine residues are strictly conserved in both the 5 0 and 3 0 ITR of the MLEs analyzed here (Figure 2(b)), four of these positions being located within the cardinal consensus 5 0 -YYAGRT-3 0 , that corresponds to the cleavage signal at the outer extremities of the ITR. Although sequence comparison of ITRs within and between mariner sub-families can provide evolutionary information, it fails to provide any information about how the binding site works. Our investigations of the conservation of palindromic and/or mirror motifs were more informative in this regard. Firstly, we detected palindromic motifs with a more or less strictly conserved sequence within the 5 0 and 3 0 ITRs of every MLE (Figure 2(c)), apart from Mcmar1. This may result from a different process of Tnp binding. Indeed, Mcmar1 has large ITRs of 355 bp that are highly conserved in terms of sequence, and have motifs located in the inner region of the ITR that might be involved in the Tnp binding. The palindromes detected in the other MLEs are centered at position 16, apart from the ITRs of the MLEs belonging to the irritans and elegans-briggsae

Arrows indicate the nucleotides corresponding to the symmetry axis of the palindromic motifs. In (d), single or double arrows indicate the nucleotides or dinucleotides corresponding to the anti-symmetry axis of the mirror motifs. In (c) and (d), nucleotides at these symmetry and anti-symmetry axes are also boxed in yellow. In (d), nucleotides boxed in green correspond to the single nucleotide insertion found in one of the mirror repeats of every ITR. In (c) and (d), Cs corresponds to the consensus ITR sequences defined within each sub-family in (a). Residues boxed in black and gray in Cs correspond to consensus positions that are conserved in the inverted or mirror motifs contained in the ITRs of each sub-family. These consensus positions are known as CIR in inverted repeats in (c) and as CMR in mirror repeats (d). It should be noted that there are great dissimilarities between the conservation patterns of Cs, CIR and CMR in each subfamily, indicating that the ITRs of each element have been subject to differing selection pressures. nZA, C, G or T; yZT or C, rZA or G. In (a), (c) and (d), the lower case letters in the Acmar1 sequences correspond to the nucleotides located at the outer extremities of the untranslated terminal regions of this element that has a 19 bp ITR. It should be pointed out that the conserved palindromic and the mirror motifs located at the extremities of Acmar1 are spanned from the beginning of the 5 0 and 3 0 untranslated terminal regions.

114 sub-families, for which the symmetry axis is located at position 15. It must be noted that there is a high degree of coincidence between the location of the best-conserved parts of the Himar1 palindrome (irritans sub-family) and the two 6–7 bp areas bound by the tranposase within the Himar1 ITR.14 Moreover, a similar coincidence is also observed for most of the MLE ITRs. Overall, this indicates that the short inverted repeats within palindromic motifs of MLE ITRs are the main motifs bound by the MLE Tnp dimers;9,15 even though the MLE palindromes are not as well conserved as those of cro binding sites, they are as well conserved as most of the CRP DNA binding sites. Secondly, mirror repeats were detected within all the ITR sequences (Figure 2(d)). Their sequence was not strictly conserved, but showed relaxed purinepyrimidine conservation, as previously reported for cro binding sites. Our results also revealed that these mirror repeats have three distinctive features: (1) most of the repeats span the entire ITR; (2) in all ITRs, one of the two mirror repeats contains a onenucleotide insertion; (3) mirror sequences are specific to each MLE species, and their patterns are specific to the 5 0 and 3 0 ITR in each MLE species. Finally, the occurrence of these mirror motifs indicates that they have evolved under selection pressure, indicating that they have a functional role that is probably to stabilize the interaction between the Tnp dimer and the ITR. Connecting with biochemical data: the case of Mos1 ITR In an attempt to analyze the mirror and/or palindrome of the MLE ITRs, and to clarify the impact of the differences observed between the 5 0 and 3 0 ITR, we focused our attention on the Mos1 ITRs, because these are the only ones for which documented biochemical data are available. Indeed, our previous work has demonstrated that the 3 0 ITR is at least ten times more efficiently bound by the Mos1 Tnp, and also more effectively ensures that the full transposition process occurs.4,18,19 In our studies, we have also demonstrated that position 16 contains information that accounts for most of the differences between the 5 0 and the 3 0 ITRs.17 Here, our data show that the mirror and palindromic repeats in the 3 0 ITR are better conserved than those of the 5 0 ITR (Dmmar1 in Figure 2(c) and (d)). The palindrome found in Mos1 ITR is centered on position 16 (Figure 3(a)). It contains four strictly conserved positions, and one pyrimidine conserved position in each half part of the 3 0 ITR. Although the 5 0 ITR is strictly well conserved, the palindrome involves the same number of residues. Thus, it is not obvious that the less efficient binding of the Tnp to the 5 0 ITR is due in part to weaker palindrome conservation. It is interesting to note that we have previously shown17 that the most important position for Tnp binding, the position 16, lies at the center of the palindrome. A sequence modification at this

Sequence Organization of mariner ITR

Figure 3. In silico analyses of the natural bendability of Mos1 ITRs. (a) Nucleic acid sequences of the ITR used for the calculations are developed in (b). (a) Nucleotides added at the 5 0 end are shown in lower case, and the TA target site dinucleotide in bold type. The residues at position 16 within the ITR are in bold cyan-blue type. The ITR5 0 -G16T has the nucleic acid sequence of a 5 0 ITR and the G at position 16 is substituted by T, similarly to the nucleotide that occurs at this position in a 3 0 ITR. (b) Only the sugar backbone of each ITR has been drawn. The colors used in (a) were also used in (b) to identify each ITR, the TA dinucleotide, and then to locate position 16.

point has no effect on the palindrome conservation, but could alter the mirror conservation. As previously mentioned, the mirrors modify the bending properties of each CRP DNA binding site and, in consequence, modulate the stability of each of the CRP proteins bound to these sites.15 According to this hypothesis, the bending properties of the Mos1 ITR were modelled by the DNaseI-based trinucleotide using the BEND algorithm of Goodsell Dickerson20 at the web site†. PDB files were then visualized using Swiss-PDBViewer software, version 3.7 (Figure 3). Three ITR sequences were used for the calculations: those of the 5 0 and 3 0 ITR and that of the 5 0 ITR, substituted by a T at position 16. The resulting model (Figure 3(b)) showed that all three naked ITRs are naturally bent. However, they are bent to different extents, with 3 0 ITR, which is bent at an angle close to 208, exhibiting the greatest bending. These calculations also show that the bending of 5 0 ITR substituted by a T at position 16 is similar to that of 3 0 ITR. Overall, these findings suggest that the natural bendability of an ITR is a factor that may modulate the stability of the Tnp dimer bound to the ITR. They also show that the natural bending of an ITR can be correlated to the quality of its mirror repeats. Finally, they also indicate that if the bendability of an ITR is a criterion for efficient binding of the Tnp, this can easily be checked by means of a simple base substitution, as exemplified by position 16 in the Mos1 ITR. † http://hydra.icgeb.trieste.it/~kristian/dna/bend_it. html

115

Sequence Organization of mariner ITR

Discussion Our sequence analyses and models show that the binding sites of the MLE Tnp within the MLE ITR display all the properties of cro binding sites, apart from Mcmar1. These observations confirm the previous suggestion13 that the palindromic motifs centered at position 16 are the main binding sites for the Tnp dimer. They also show that the MLE ITRs contain a mirror structure that is very probably involved in their natural bending. Taken together, these findings indicate that at least three kinds of highly concentrated information are sent to the Tnp via the ITR. The first two originate from the palindromic and mirror motifs as indications for the binding of a Tnp dimer, whereas the third concerns the cleavage at the outer ITR extremity, and seems to be encoded by the cardinal motif 5 0 -YYAGRT-3 0 (5 0 -YYRGGT-3 0 in Figure 2(b)). Comparing the conservation pattern of the palindromic and mirror repeats helps us to understand the specificity of the Tnp–ITR interaction for each MLE species. Indeed, biochemical analyses4,6 have shown that closely related MLE Tnp, such as those of Mos1 and Botmar1 that both belong to the mauritiana sub-family, are unable to cross-bind to the Botmar1 and Mos1 ITR,6 respectively (personal data). The results of our sequence analyses also indicate that the sequence evolution of an ITR does not occur randomly and neutrally in parallel with that of the Tnp within each MLE species. It probably results from the impact of at least three interdependent selection pressures. The first two are directly due to the dual function of the ITR, which implies that ITRsequence evolution must maintain both efficient Tnp–ITR interactions and accurate ITR cleavage. In this context, the outer and inner regions of each ITR need to co-evolve in order to maintain functional palindromic and mirror motifs. Finally, a third source of selection pressure results from the fact that each MLE has to keep two ITRs that are able to work together to ensure that the transposition process proceeds correctly. The complexity of this last point must not be underestimated. Indeed, some sequence differences are always observed between most of the MLE ITRs. These differences may play an important part in initiating the transposition process by orientating the binding of the Tnp dimer to the first ITR and then to the second, or by negatively modulating the transposition efficiency.18 It has been shown that the different nucleic and protein components3,6,21,22 contained within or encoded by a class-II transposon, including MLE, can be improved by molecular engineering in an attempt to increase the transposition efficiency. The complexity of the sequence conservation and evolution of the MLE ITRs presented here strongly suggest that attempting to produce hyper-efficient ITRs by a molecular selection process would probably be very difficult to design and to achieve. Modelling from biochemical data and in silico

analyses on the functioning of the MLE ITR might therefore offer an alternative strategy for designing hyper-efficient ITRs, the impact on the transposition efficiency of which could be assayed in vivo.

Materials and Methods Sequences The nucleic acid sequences of 5 0 and 3 0 ITR from 45 MLEs belonging to the five MLE sub-families, mauritiana, cecropia, melifera/capitata, irritans and elegans/briggsae23 were recovered from previous sequence extractions in databanks as described.16 In agreement with the most recent classification of the Tc1-mariner elements,24 the ITR of Tc1-like elements (TLE) and the maT-like elements (maTLE) were not included in the analysis. Indeed, PAIRED-like and LACI-like domains are, respectively, on the origins of the DNA binding domains of TLE and maTLE transposases. These domains bound DNA target that are fully different in structure and sequence. Only three of them correspond to ITR sequences belonging to elements with an ability to trigger transposition that was verified in vitro and/or in vivo, Dmmar1 (Mos1), Himar1 and Famar1. The 42 other sequences corresponded to consensus or to MLE elements with uninterrupted ORF (Bytmar1, Ccmar1, Mbmar1, Mcmar1, Momar1 and Tvmar1), but for which the activity has so far not been confirmed. Our data set was completed by ITR sequences from MLEs found in the Caenorhabdidtis elegans and C. briggsae genomes25 that were kindly communicated by David J. Witherspoon (Salt Lake City, Utah, USA). Sequence analyses The sequences were aligned using ClustalW. Searches for palindromic and mirrors motifs were done using the facilities at the website†. A palindrome was defined as a nucleic acid region presenting interstrand sequence similarities between motifs located on both sides of a symmetry axis. A mirror region was defined by the presence of two intrastrand motifs separated by a few or no nucleotides and that have sequence similarities when they are read from the center (anti-symmetry axis) outwards. The results were then verified and improved by hand. The shape of the ITR was calculated using the facilities at the website‡. The results were recovered as PDB files. Their shapes were visualized using the SwissPDBViewer version 3.7 software and X3DNA.26

Acknowledgements This work was funded by the Centre National de la Recherche Scientifique, the Ministe`re de l’Education, de la Recherche et de la Technologie, the Groupement de Recherche 2157, the Institut Fe´de´ratif de Recherche 136, the Ligue Nationale † http://www.infobiogen.fr ‡ http://hydra.icgeb.trieste.it/~kristian/dna/bend_it. html

116

Sequence Organization of mariner ITR

contre le Cancer and the Association Contre le Cancer (grant number 7684). B. B. is funded by a PhD grant from the Re´gion Centre. The English text was revised by Monika Ghosh.

References 1. Plasterk, R. H. A. & Van Luenen, G. A. M. (2002). The Tc1/mariner family of transposable elements. In Mobile DNA II (Craig, N. L., Craigie, R., Gellert, M. & Lambowitz, A., eds), p. 519, ASM Press, Washington, DC. 2. Lampe, D. J., Churchill, M. E. A. & Robertson, H. M. (1996). A purified mariner transposase is sufficient to mediate transposition in vitro. EMBO J. 15, 5470–5479. 3. Tosi, L. R. O. & Beverley, S. M. (2000). cis and trans factors affecting Mos1 mariner evolution and transposition in vitro, and its potential for functional genomics. Nucl. Acids Res. 23, 784–790. 4. Auge´-Gouillou, C., Hamelin, M. H., Demattei, M. V., Periquet, G. & Bigot, Y. (2001). The ITR binding domain of the Mariner Mos1 transposase. Mol. Gen. Genet. 265, 58–65. 5. Zhang, L., Dawson, A. & Finnegan, D. (2001). DNAbinding activity and subunit interaction of the mariner transposase. Nucl. Acids Res. 29, 3566–3575. 6. Lampe, D. J., Kimberley, K., Walden, O. & Robertson, H. M. (2001). Loss of transposase-DNA interaction may underlie the divergence of mariner family transposable elements and the ability of more than one mariner to occupy the same genome. Mol. Biol. Evol. 18, 954–961. 7. Auge´-Gouillou, C., Notareschi-Leroy, H., Abad, P., Periquet, G. & Bigot, Y. (2000). Evolution of DNAbinding and catalytic domains in mariner-like transposases: consequences on their phylogenetic analyses. Mol. Gen. Genet. 264, 514–520. 8. Rosinski, A. A. & Atchley, W. R. (1999). Molecular evolution of helix-turn-helix proteins. J. Mol. Evol. 49, 301–309. 9. Kim, J. G., Tadeka, Y., Matthews, B. W. & Anderson, W. F. (1987). Kinetics studies on Cro repressoroperator DNA interaction. J. Mol. Biol. 196, 149–158. 10. Richardson, J. M., Zhang, L., Marcos, S., Finnegan, D. J., Harding, M. M., Taylor, P. & Walkinshaw, M. D. (2004). Expression, purification, and preliminary crystallographic studies of a single-point mutant of Mos1 mariner transposase. Acta Crystallog. sect. D, 60, 962–964. 11. Lohe, A. R., De Aguiar, D. & Hartl, D. L. (1997). Mutations in the mariner transposase: the D,D(35)E consensus sequence is nonfunctional. Proc. Natl Acad. Sci. USA, 94, 1293–1297. 12. Auge´-Gouillou, C., Brillet, B., Germon, S., Hamelin, M.H., Bigot, Y. Mariner Mos1 transposase dimerizes prior to ITR binding. J. Mol. Biol. In the press.

13. Auge´-Gouillou, C., Brillet, B., Hamelin, M. H. & Bigot, Y. (2005). Assembly of the mariner Mos1 synaptic complex. Mol. Cell. Biol. 25, 2861–2870. 14. Lipkow, K., Buisine, N., Lampe, D. J. & Chalmers, R. (2004). Early immediate transposition: catalysis without synapsis of the transposon ends suggests a novel architecture of the synaptic complex. Mol. Cell. Biol. 24, 8301–8311. 15. Shwu-Hwa, L. & Lee, J. C. (2003). Determinants of DNA bending in the DNA-cyclic AMP receptor protein complexes in Escherichia coli. Biochemistry, 42, 4809–4818. 16. Halaimia-Toumi, N., Casse, N., Demattei, M. V., Renault, S., Pradier, E., Bigot, Y. & Laulier, M. (2004). The GC-rich transposon Bytmar1 from the deep-sea hydrothermal crab, Bythograea thermydron, may encode three transposase isoforms from a single ORF. J. Mol. Evol. 59, 747–760. 17. Sinzelle, L., Chesneau, A., Bigot, Y., Mazabraud, A. & Pollet, N. (2005) The mariner transposons belonging to the irritans subfamily were maintained in chordate genomes by vertical transmission. J. Mol. Evol. In the press. 18. Auge´-Gouillou, C., Hamelin, M. H., Demattei, M. V., Periquet, G. & Bigot, Y. (2001). The wild-type conformation of the Mos1 inverted terminal repeats is suboptimal for transposition in bacteria. Mol. Gen. Genet. 265, 51–57. 19. Pledger, D. W., Fu, Y. Q. & Coates, C. J. (2004). Analyses of cis-acting elements that affect the transposition of Mos1 mariner tranposons in vivo. Mol. Gen. Genet. 272, 67–75. 20. Goodsell, D. S. & Dickerson, R. E. (1994). Bending and curvature calculations in B-DNA. Nucl. Acids Res. 22, 5497–5503. 21. Lampe, D. J., Akerley, B. J., Rubin, E. J., Mekalanos, J. J. & Robertson, H. M. (1999). Hyperactive transposase mutants of the Himar1 mariner transposon. Proc. Natl Acad. Sci. USA, 96, 11428–11433. 22. Wiegand, T. W. & Reznikoff, W. S. (1992). Characterization of two hypertransposing Tn5 mutants. J. Bacteriol. 174, 1229–1239. 23. Robertson, H. M. (2002). Evolution of DNA transposons in eukaryotes. In Mobile DNA II (Craig, N. L., Craigie, R., Gellert, M. & Lambowitz, A., eds), p. 1093, ASM Press, Washington, DC. 24. Claudianos, C., Brownlie, J., Russell, R., Oakeshott, J. & Whyard, S. (2002). maT—a clade of transposons intermediate between mariner and Tc1. Mol. Biol. Evol. 19, 2101–2109. 25. Whiterspoon, D. J. & Robertson, H. M. (2003). Neutral evolution of ten types of mariner transposons in the genome of Caenorhabditis elegans and C. briggsae. J. Mol. Evol. 56, 751–769. 26. Lo, X. J. & Olson, W. K. (2003). 3DNA: a software for the analysis, rebuilding and visualization of threedimensional nucleic acid structures. Nucl. Acids Res. 31, 5108–5121.

Edited by J. O. Thomas (Received 17 December 2004; received in revised form 14 April 2005; accepted 2 May 2005) Available online 23 May 2005