INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial transposases are an interesting class of multifunctional DNA-binding proteins that catalyze the process of transposition. They bind specifically to sequences at the transposon ends and catalyze the DNA cleavage events that are the first step of transposition. The transposase also recognizes the target DNA into which the transposon integrates. The recognition and interaction properties of the transposase with the target DNA differ from those with the transposon ends in two ways. First, the transposase binds specifically to the ends of the transposon; in contrast, for most transposable elements, transposition is rarely targeted to a specific sequence, thus implying involvement of nonspecific interactions between the transposase and target DNA. The most notable exception to this is Tn7, which inserts site specifically into a unique site on the Escherichia coli chromosome, attTn7 (7). Second, the cleavage event at the target site is different from the precise strand cleavages that must occur for transposon excision. The 3' ends of the transposon are joined to the target in a staggered fashion, and the resulting gaps are filled in to generate the characteristic target site duplications found on either side of the transposon. The length of the duplication is characteristic for each transposon and can vary from 2 to 13 bp (13). For example, IS903 and IS10 generate 9-bp duplications, while Mu and Tn7 generate 5-bp target duplications. The variation in the size of target duplication indicates that the sites of DNA cleavage for each strand, and the transposase residues that mediate the cleavage, are in very different positions along the DNA backbone and suggests that each transposase interacts with the DNA with a precise, but different, geometry.

Selection of a target site is an important part of the transposition process (reviewed in reference 6). Although it is chronologically perceived to be one of the final steps in transposition, it is clear that for some transposons the target DNA is an important and integral component of the earlier steps in transposition. Intermolecular transposition of Mu is enhanced by the addition of MuB protein and target DNA before any cleavage steps have occurred (25, 28), and for Tn7, it has been demonstrated that cleavage at the transposon ends, considered the first full commitment to transposition, will occur only in the presence of target DNA (1). The requirement for a target interaction prior to initiation of transposition is thought to coordinate and regulate transposition and thereby prevent the formation of aberrant transposition products. This is in contrast to the situation observed with IS10, in which transposon excision must occur before target interaction can take place (20, 31). The results obtained with IS10 suggest that the flanking donor DNA must be released to allow binding of target DNA into the same "binding pocket" of the transposase end complex.

The choice of target can have important biological and evolutionary consequences for both the host and the transposon (6, 33). Transposition can be a mutagenic process resulting in insertion into essential genes and deletion or inversion of genomic segments. Therefore, the ability to direct transposition events to certain regions or segments of DNA will increase the fitness of both the host and the transposon. By inserting into a unique site on the chromosome, Tn7 ensures that it does not inactivate essential host genes. In addition, intramolecular rearrangements can destroy the integrity of the transposon if a target is chosen within the element itself. Certain classes of transposons avoid self-insertion by a process called immunity, which favors insertion into distal sites or intermolecular targets (24). For Tn10, in vivo experiments suggest that the interaction of the transposon end with the target can be channeled in a precise manner to promote intramolecular inversions, which result in the formation of new mobile elements (20, 35).

In the few examples for which target selection has been systematically examined, it is clear that there are a variety of ways a target is selected. IS10 has a preferred target site, the selection of which clearly involves transposase-target interactions. Sequence analysis of multiple IS10 insertions has identified a symmetrical 6-bp consensus sequence that is found within the 9-bp target duplication (4, 16). In addition, the context of the flanking DNA has been shown to have a dramatic influence on the use of these preferred target sites, suggesting that there is a structural component to site selection (4). In retroviral integration (a process mechanistically equivalent to transposition), in vitro experiments have shown that there is a preference for insertion into nucleosomal DNA and that it is the bend induced in the DNA by the nucleosome that favors integration (27).

Protein-protein interactions can also target transposon insertion. Insertion of the Saccharomyces cerevisiae retrotransposon Ty3 is targeted immediately upstream of genes transcribed by polymerase III. In vitro experiments have demonstrated that this targeting involves protein-protein interactions between the integrase and the transcription factor TFIIIB or TFIIIC (19). Tn7 is directly targeted to its unique chromosomal site by the binding of the transposon-encoded TnsD protein to a site adjacent to attTn7 (1, 38). TnsD then recruits Tn7 to the target site by interacting directly with TnsC, which in turn interacts with the transposase proteins that are bound to the transposon ends (7). Tn7 can also transpose via an alternative pathway mediated by the TnsE protein. In this case, transposition is not directed to attTn7 but occurs at a low frequency to other, unrelated sites (7). Recently, it has been shown that the TnsE pathway prefers conjugative plasmids as targets (40). Through a series of elegant genetic experiments, it was shown that conjugal functions were required for this targeting, implying that transposition occurred during transfer. Furthermore, transposition into the conjugative plasmids occurred with a preferred orientation and with a distinct preference for the leading region of transfer. It was proposed that conjugal DNA synthesis in the recipient was a preferred target for transposition, and the process of replication provided the polarity to dictate the orientation of insertion.

Little is known about the target selection of most transposable elements. Either a limited number of insertions have been examined or a relatively small plasmid has been used as a target, which restricts both the availability of a preferred site and the size of the target that can be used and still give rise to a viable insertion. This paper describes an analysis of the insertion sites selected by IS903. We undertook this analysis for two reasons. First, there has been no comprehensive analysis of insertion preference for this element. Indeed, data accumulated over the years suggested that IS903 inserts randomly (9). Second, we wished to identify a target that had been used in vivo for our in vitro transposition studies. We reasoned that a DNA substrate containing a target site used by IS903 in vivo would be more likely to work in vitro. In addition, other information obtained concerning target selection might also help optimize such an in vitro system.

We show that IS903 inserts into discrete regions of the F plasmid derivative pOX38 and that within these regions there are a few sites that are used multiple times, thus allowing us to identify preferred in vivo hot spots for IS903 transposition. Alignment of the insertion sites showed that there are no preferred sequences within the target duplication but that there are distinct symmetrically located nucleotide preferences in the DNA flanking the target, implying that the transposase interacts with the target in a symmetrical fashion. IS903 also shows a very strong orientation bias for insertion into a second conjugative plasmid, pUB307. We have examined the plasmid features that are responsible for this and have shown that reversing the direction of plasmid transfer abolishes this orientation bias, suggesting that the process of conjugation directly influences IS903 transposition.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Media and standard procedures. Luria-Bertani (LB) medium was used for growth of all E. coli strains. The following antibiotics were used at the indicated concentrations: chloramphenicol, 20 µg/ml; kanamycin, 50 µg/ml; nalidixic acid, 40 µg/ml; ampicillin, 100 µg/ml; tetracycline, 10 µg/ml; streptomycin, 200 µg/ml. Restriction enzymes and DNA-modifying enzymes were purchased from New England Biolabs and Boehringer Mannheim and used as recommended. Oligonucleotide synthesis and DNA sequencing were performed at the molecular genetics core facilities of the Wadsworth Center. Plasmid DNA manipulations were carried out by using standard procedures (32). PCR colony screening was performed as described previously (37).

Bacterial strains and plasmids. The strains and plasmids used in this study are listed in Table 1. Segments of pOX38 used to assay target selection in pUB307 were generated by PCR, using the oligonucleotide primers (oKD94 to oKD97) listed in Table 2. These PCR fragments were cloned into the HindIII site of pUC118 before being recloned into the HindIII site of pUB307. Derivatives of pUB307 were isolated with each pOX38 DNA segment inserted in both orientations. The constructs were confirmed by restriction enzyme analysis, PCR analysis using appropriate pairs of fragment- and pUB307-specific primers, and in many cases DNA sequencing. A 71-bp deletion of the traY promoter (traYp) in region I (traY to traL) was made by digesting the pUC118 derivative pUCF1 with MluI and BstEII, end filling, and religating to generate pUCF2. These two sites are unique in that plasmid and flank the traYp. The 1.0-kb HindIII fragment of pUCF2 was subcloned into the HindIII site of pUB307 to generate pUBF36. To eliminate traYp activity without affecting the length of region I, point mutations were introduced in the 35 and 10 regions of the promoter in pUCF1 to generate pUCF9. The primer oKD151 was used to introduce multiple point mutations in the 35 and 10 hexamer sequences, which were predicted to eliminate transcription and to create BamHI and SphI recognition sites, respectively. The mutations were confirmed by restriction analysis and DNA sequencing before the HindIII fragment was subcloned into pUB307 to generate pUBF38. pUBF381 was generated from pUBF38 by introducing the omega fragment of pHP45 (30) into the HindIII site closest to the Km^r promoter of pUBF38. This is a 2-kb fragment that encodes resistance to spectinomycin and is flanked by transcriptional terminators.

Transposition and conjugation assays. Transposon insertions into the target plasmids were obtained by a mating-out assay using E. coli NG135 as the recipient (10). To ensure that all insertion events were unique, each mating was carried out with a single colony from independent transformations of the donor strain. RR1023 was used as a donor to isolate independent insertions into pOX38 as previously described (10). In the transposition analysis using pUB307 and its derivatives as a target, DH1 was used as the donor (17). Donors and recipients were grown overnight with selection for resident plasmids and inoculated in fresh medium at a 1:20 dilution. The cells were grown to mid-log phase, and donor (0.1 ml) and recipient (1 ml) strains were mixed and resuspended in 20 µl of broth and then deposited on an LB plate for 2 h at 37°C. The mating mixture was scraped from the plate and resuspended in 0.5 ml of LB medium. Appropriate amounts of mating mixture were spread on media selective for transposition events. Conjugation and transposition frequencies were determined from at least three matings and are expressed as the number of transconjugants per donor.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

IS903 prefers to insert into distinct regions of plasmid pOX38. To determine whether IS903 exhibits a target preference, we have mapped the locations of independent insertions into the plasmid pOX38, which is a transfer-proficient deletion derivative of the F plasmid (14). It provides an ideal target for transposition since insertions can be selected by use of a mating-out assay, it lacks insertion sequences, it provides a large target (55 kb) that includes many nonessential genes, and the majority of its sequence is known (12, 18, 23). We used pKD100 (10) to deliver an IS903 derivative that includes genes encoding resistance to chloramphenicol (Cm^r) and ampicillin (Ap^r) and IS903 transposase and carries the pBR322 origin of replication (Fig. 1A).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   (A) Linear diagram of the transposon donor plasmid pKD100 (7.0 kb). The inverted repeats (IR) are represented by triangles. The direction of transcription of genes is indicated by wavy arrows. The transposon in pKD100 carries transposase (Tnpase), genes encoding resistance to ampicillin (Apr) and chloramphenicol (Cmr), and the pBR322 origin of replication (ori). The gene for kanamycin resistance (Kmr) is present on the plasmid backbone. Primers oKD78 and oKD79, indicated by half arrows, were used to sequence the target duplications and to identify the orientation of insertion by PCR analysis. (B) Transposon insertion sites in pOX38. The open box represents a linear map (to scale) of 55-kb pOX38, showing three regions: the RepFIA replication region containing the origin of vegetative replication (oriV), the leading region, and the transfer (tra) region. The arrowheads indicate the direction of vegetative DNA replication from oriV and the direction of conjugal DNA transfer from oriT. Arrowed lines in the polycistronic tra region indicate transcription initiating at the indicated promoters; dotted segments signify uncertainty in the extent of the transcript (12). Transposon insertion sites were determined by DNA sequencing using both primers oKD78 and oKD79. The numbers below the physical map of pOX38 represent the first nucleotide of each IS903 insertion. The target sites used more than once are boxed, and an underline indicates an insertion in the opposite orientation. Nucleotide coordinates are designated with respect to the published sequence of the tra operon which begins at the BglII site (marked B) (accession no. U01159, 33.6 kb). The leading region (accession no. M97768) upstream of oriT has been assigned negative numbers beginning from the BglII site. The thick line represents an enlargement of region IV. The arrowed boxes represent coding regions and directions of transcription of the genes in region IV.

Alignment of target sites. IS903 generates a 9-bp duplication, which is likely a result of a 9-bp staggered cleavage made by the transposase during integration of the transposon. The 63 unique duplications sequenced in this study were aligned to determine if there was a preferred consensus sequence for insertion. As can be seen in Fig. 2A, there are no preferred sequences found within the target duplication; however, certain nucleotides are not favored at several positions, implying that they negatively influence the selection of that target. In contrast, in the 10 bp flanking the target site duplication, there are distinct nucleotide preferences symmetrically located about the target site. The consensus for this region of symmetry is 5'-T/A-T-T/A-Py-A-3' and extends from positions 2 to 6 on either side of the duplication. This preference is highly significant (greater than 99% confidence level) as determined by the chi-square test based on an AT content of 48% for the pOX38 plasmid. The least significant position in this consensus is at position +6, which occurred only within confidence limits of 95%. The symmetrical location of these sequences suggests either that the transposase makes base-specific symmetrical contacts with these nucleotides or that this sequence forms a unique structure on either side of the target which predisposes it to recognition, cleavage, and integration.

View larger version (61K): [in this window] [in a new window]   FIG. 2.   Alignment of target sites used by IS903. (A) Sixty-three target sites and the flanking DNA sequences have been aligned to generate the matrix shown. The sequences compiled are those adjacent to the left end of IS903 (1 to 10) and those adjacent to the right end of IS903 (+1 to +10). The central 9 nucleotides are those that would be duplicated on insertion. The consensus sequence derived from the outlined numbers is shown below the matrix. Positions at which a nucleotide bias occurs with greater than 99% confidence levels as determined by a chi-square test are indicated by a "+" symbol. An A(T) at position +6 in the consensus occurs with 95% confidence limits. The parentheses around a nucleotide designation indicate that the preference for that base is not as strong. (B) Target and flanking sequences of sites used by IS903 more than once are shown. The data are compiled from insertions mapped in Fig. 1 and 3. Boxed nucleotides are those that match the consensus sequence in panel A and also include the preference for a G/C nucleotide at positions 1 and 9 of the target duplication. Note that three sites contained insertions in both orientations and so each orientation has been included. Numbering at the top is as for panel A.

Insertion preference is reproducible with a second IS903 derivative. To rule out the possibility that target preference was influenced by internal sequences of the IS903 derivative in pKD100, we used a second transposon derivative carried on pKD1 (39). This transposon carries a Km^r gene flanked by 180 bp and 113 bp of DNA from the left and right ends of IS903, respectively (Fig. 3A). Thirty-eight independent transposition events were isolated by the mating-out assay, and all but four insertions mapped to the same four clusters seen previously (Fig. 3B).

View larger version (21K): [in this window] [in a new window]   FIG. 3.   (A) Linear diagram of the transposon donor plasmid pKD1. The transposon carries the Kmr gene, and the transposase gene (Tnpase) is located immediately adjacent to one end of the transposon. The direction of transcription of genes is indicated by wavy arrows. The annealing sites of oligonucleotides oKD20 and oKD21 are indicated by half arrows; they were used to determine the site of transposon insertion. IR, inverted repeat. (B) Insertion sites of IS903 from pKD1 into pOX38. Only the tra region and leading region of pOX38 are shown. Thirty-eight transposon insertion sites were determined. The majority of insertions were mapped to the same four regions shown in Fig. 1; these are indicated above the map. The numbers below the physical map of pOX38 represent the first nucleotide of each IS903 insertion. The target sites used more than once are boxed, and an underline indicates an insertion in the opposite orientation.

Transposition onto pOX38 does not occur at high frequencies during vegetative growth. All of the experiments described above used a mating-out assay to select for transposition events. This assay prevented us from determining whether the process of conjugal transfer or transfer functions played a direct role in targeting. Therefore, we developed a second transposon delivery system that made use of a temperature-sensitive RK2-based plasmid that carries the Km^r transposon from pKD1. Transposition events could be directly selected by plating cells on media containing kanamycin at the nonpermissive temperature for the transposon donor plasmid. By using this transposon delivery system, we have monitored IS903 transposition in the presence and absence of pOX38. Transposition was not stimulated by the presence of pOX38 in the cell (data not shown). Furthermore, of 20 Km^r colonies screened, none contained insertions in pOX38, suggesting that pOX38 was not used efficiently as a target under these growth conditions. We have also been unable to detect efficient transposition onto a pUC118 derivative containing a wild-type oriT. Thus, at present, we are unable to directly test the role of conjugation in target selection by using transfer-deficient F derivatives.

IS903 insertion into pUB307 is not targeted but does occur with a unique orientation. To determine if oriT sites in general, or proteins associated with them, were functionally "attractive" to IS903, we have examined insertion preference into a second conjugative plasmid, pUB307. pUB307 is a deletion derivative of RP1, a member of the IncP family of plasmids, and is unrelated to F (5, 29). This plasmid was chosen because conjugation could be used to select for transposition events, the entire nucleotide sequence is known (29), the conjugal transfer system is well characterized and is different from that of pOX38 (22), and the presence of a defined oriT and transfer region would allow us to make comparisons with the regional targeting seen in pOX38.

View larger version (31K): [in this window] [in a new window]   FIG. 4.   IS903 insertion sites in pUB307. Physical and genetic maps of RP1/pUB307 and the insertion sites of IS903 transposition are shown. The first nucleotides of each insertion site with respect to the RP1 sequence (Genbank database accession no. L27758) are marked around the map. The region of RP1 deleted in pUB307 (including the ampicillin-resistant transposon Tn1) is indicated by vertical bars. Genetic elements are indicated in the inner circle and include the following: the two transfer regions (Tra); genes encoding resistance to tetracycline (Tcr) and kanamycin (Kmr); the partition locus par; and the insertion sequence IS21. The unique HindIII site used for cloning pOX38 fragments is shown within the Kmr gene. The AseI and BglII sites used for inverting the oriV region are also indicated. Black boxes represent oriV and oriT; the directions of DNA replication from oriV and conjugal transfer from oriT are indicated by arrows.

All signals for targeting are present on a 1.1-kb traY-to-traL pOX38 DNA fragment. Since IS903 insertions in pUB307 were well distributed, the plasmid was used as a vector to test the regional preference for pOX38 hot spots in a different DNA context. PCR-amplified segments of pOX38 DNA were subcloned into the unique HindIII site in the Km^r gene of pUB307 (Fig. 4). Segments of pOX38 DNA that included sites that had been used multiple times as a target were amplified from each of the preferred regions. Independent transposition events into the pUB307 derivatives containing the different pOX38 segments were generated by using a mating-out assay. Insertions into the pOX38 fragment were screened by a PCR assay that allowed insertions into the cloned fragment to be detected and their orientation to be determined. This was achieved with combinations of primers that were specific to the Km^r gene and each end of the transposon.

View larger version (25K): [in this window] [in a new window]   FIG. 5.   IS903 insertion specificity and orientation preference in region I from pOX38 when cloned into pUB307. A simple map of the Kmr gene in pUB307 is indicated in line 1. The direction of transcription of the Kmr gene is designated by the wavy arrow. Oligonucleotides oKD125 and oKD126 were used in a PCR analysis to identify IS903 insertions into the region. To determine the orientation of those insertions by PCR, oKD125 and oKD126 were used in combination with the transposon-specific primers oKD78 and oKD79 (Fig. 1). Region I and derivatives of it were cloned into the unique HindIII (H) site of pUB307 in both orientations (lines 2 to 12). The DNA present in each plasmid is shown by the solid arrow. The tra genes and their coordinates with respect to the BglII site in the pOX38 transfer region are indicated. The vertical dashed lines indicate the end points of subclones made from the traY-to-traL fragment. Filled and open boxes represent the wild-type and mutant 35 and 10 hexamers of the traY promoter, respectively. pUBF36 contains a 71-bp deletion of the traY promoter. The  fragment present in pUBF381 contains a gene encoding spectinomycin resistance (Spr) and is flanked by transcriptional terminators (T). Insertion specificity is defined as the number of insertions in region I divided by the total number of insertions. The orientation of IS903 insertion is indicated by small arrows ( and ). Line 1 shows that 31 insertions were mapped in the same orientation in pUB307 (data from Fig. 4). Transposition frequency was determined from at least three independent mating-out assays.

Dissection of the traY-to-traL gene fragment. The 1.1-kb fragment containing the first part of the tra operon (traY to traL) is clearly a preferred target for IS903, irrespective of its orientation when cloned into pUB307. In addition, target sites used within this region have been used before. For example, insertions at position 2265 of the tra region were isolated both in pOX38 and in pUB307, indicating that this is a true hot spot for IS903 insertion. To further define the critical regions required for insertion, this DNA fragment was divided into three almost equal segments. Primers were designed to amplify individual segments as well as combinations of them, and these were then subcloned into the HindIII site of pUB307. Independent transposition events into each pUB307 derivative were generated and then screened by the PCR assay for insertions into the pOX38 fragment. The PCR assay was also used to show that the majority of insertions had inserted in the same relative orientation.

Orientation of insertion is influenced by oriT. One of the more striking results found was that IS903 insertion has a strong orientation bias (Fig. 4 and 5). This bias must result from the transposase sensing a strong polarity in the target. For pUB307, DNA replication from both oriV and oriT is unidirectional, and we reasoned that either one or both of these processes might explain the strict orientation of IS903 insertion. A simple test of this model would be to flip the orientation of oriT or oriV to see whether it affected the orientation of insertion. The ability to clone and manipulate DNA in pUB307 also made it feasible to carry out these flips in the native vector rather than with a high-copy or miniplasmid derivative. In addition, we could take advantage of the high targeting preference for the pOX38 hot spot to use a PCR-based assay to simplify the determination of the direction of transposon insertion.

View larger version (11K): [in this window] [in a new window]   FIG. 6.   A schematic representation of the pairs of plasmids used to monitor the effect of reversing the direction of either oriV or oriT. The native orientations of oriV and oriT are shown in pUBF12. All plasmids contain the 1.1-kb pOX38 fragment inserted in the HindIII site; this is the target for IS903 insertion.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Target preference. Despite the availability of a large target (>55 kb) for transposon insertion, we have shown that IS903 prefers to insert into very discrete regions within pOX38. About half of the insertions were found in three small regions in the tra operon. This was a surprising result, in that our selection for transposition relied on transfer proficiency, and suggests that many of the events occurred during, or immediately prior to, DNA transfer since many of the insertions result in a transfer-deficient phenotype (data not shown). The remaining insertions were all located in the leading region of pOX38 transfer, immediately adjacent to oriT. These same regions were used as targets in multiple experiments using different IS903 derivatives. This rules out influences of transposon sequences and implies that these regions are intrinsically attractive as IS903 targets. In particular, region I was used efficiently as a target in both pOX38 and pUB307, implying that it contains all the information necessary for targeting. The fact that a precise region for targeting could not be defined by a simple deletion analysis (Fig. 5) suggests that there are multiple sequence components throughout the region that all contribute to targeting. The lack of an obvious association of host-factor binding sites with insertion sites in this region and the insensitivity to plasmid backbone context suggest that region I is capable of forming a structure or domain that makes it more accessible to transposition. In addition, by eliminating transcription through this region in pUB307, we demonstrated that it does not play a major role in targeting to region I.

A preferred target site. The availability of a large collection of insertions allowed us to examine the target sites for a consensus sequence. Although no consensus sequence could be found for the 9-bp target duplication, assessment of the flanking sequences showed that there is a preferred consensus sequence found on either side of the target duplication (5'-T/A-T-T/A-Py-A-3'). A similar consensus detected on both sides of the duplication was maintained when sites that had been used more than once were aligned (Fig. 2B). The most significant difference observed when the multiply used sites were compared with all the target sites was the very strong preference for G/C nucleotides at the first and last positions of the target duplication. The arrangement of symmetrical sequences is consistent with a dimer or multimer of transposase, in a complex with the ends of the transposon, recognizing a target by making specific, symmetrical contacts outside the target duplication. These flanking sequences could be recognized in two ways: (i) by base-specific contacts with the transposase or (ii) as a specific structural conformation in the DNA favored by those sequences. We note that there are two potential matches to the consensus sequence within the inverted repeat of IS903 (8), which, if significant, would provide a rational explanation for the presence of this sequence around insertion sites. In preliminary band shift experiments, we have been unable to detect binding of transposase to a segment containing a preferred target (data not shown). However, this might simply be because it is only recognized by the transposase when in the form of an active transpositional integration complex. Clarification of the significance of these sequences will require more detailed genetic and especially biochemical analyses using an in vitro transposition system.

Orientation of insertion. One of the more striking observations made in this work was that there was a distinct orientation preference for insertions in pUB307. A similar effect has been seen with Tn7 insertions into both pOX38 (40) and the parent of pUB307, RP1 (2, 3). To generate such a profound bias, there has to be an asymmetry in both the transposon donor and target; otherwise, insertion would occur equally in both orientations. For IS903, the asymmetry could arise if one end was presented in a more-or-less active form. For example, transcription across one end may reduce its activity, or delivery of the cis-acting transposase (or host factor) to one end may place that end in a more active conformation relative to the other end (11).