Self-Inflicted Wounds, Template-Directed Gap Repair and a Recombination Hotspot: Effects of the mariner Transposase

Corresponding author: Daniel L. Hartl, Department of Organismic and Evolutionary Biology, 16 Divinity Ave., Cambridge, MA 02138., dhartl{at}oeb.harvard.edu (E-mail)

Communicating editor: M. J. SIMMONS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Aberrant repair products of mariner transposition occur at a frequency of ~1/500 per target element per generation. Among 100 such mutations in the nonautonomous element peach, most had aberrations in the 5' end of peach (40 alleles), in the 3' end of peach (11 alleles), or a deletion of peach with or without deletion of flanking genomic DNA (29 alleles). Most mariner mutations can be explained by exonuclease "nibble" and host-mediated repair of the double-stranded gap created by the transposase, in contrast to analogous mutations in the P element. In mariner, mutations in the 5' inverted repeat are smaller and more frequent than those in the 3' inverted repeat, but secondary mutations in target elements with a 5' lesion usually had 3' lesions resembling those normally found at the 5' end. We suggest that the mariner transposase distinguishes between the 5' and 3' ends of the element, and that the 5' end is relatively more protected after strand scission. We also find: (1) that homolog-dependent gap repair is a frequent accompaniment to mariner excision, estimated as 30% of all excision events; and (2) that mariner is a hotspot of recombination in Drosophila females, but only in the presence of functional transposase.

ALMOST all mariner transposable elements isolated from natural populations are nonautonomous because of insertions, deletions, or missense mutations. The only naturally occurring mariner elements known to produce active transposase are closely related to the Mos1 element, which was originally identified in Drosophila mauritiana using a genetic screen (BRYAN et al. 1987 ). The inability of most mariner elements to encode active transposase is evident even in genomes that have a large number of copies (ROBERTSON and MACLEOD 1993 ; ROBERTSON and LAMPE 1995 ). Despite the identification of hundreds of copies of mariner elements from a wide variety of invertebrates (ROBERTSON 1993 ; GARCIA-FERNANDEZ et al. 1995 ), plants (JARVIK and LARK 1998 ), and the human genome (AUGEGOUILLOU et al. 1995 ; MORGAN 1995 ; OOSUMI et al. 1995 ; HARTL 1996 ), few if any copies contain an open reading frame (HARTL et al. 1997A ). The existing literature does not address the issue of what proportion of naturally occurring mariner elements is still capable of being mobilized in trans, because the use of PCR approaches has not allowed the inverted terminal repeats (ITRs) of most naturally occurring elements to be examined.

The evolutionary "natural history" of mariner elements is proposed to begin with horizontal transmission of an active element that invades the germline of a susceptible host species (HARTL et al. 1997B ). Invasion is followed by a period of increasing copy number that ensures the dispersal of active elements throughout the genome and, through sexual reproduction, eventually the entire population. Attainment of a high copy number of mariner elements is proposed to increase the chance of horizontal transmission to yet other hosts, so that an increase in copy number is an essential part of the long-term survival strategy of mariner. Subsequent inactivation of a high proportion of autonomous mariner elements by mutation may be essential to minimize the detrimental effects of transposon mutagenesis in the host genome that result from continuing mobilization. Mariner elements are active in the soma as well as in the germline, and a decrease in fitness of organisms containing autonomous elements has been detected under experimental conditions (NIKITIN and WOODRUFF 1995 ).

Some transposable elements, in Drosophila most notably P (ENGELS 1989 ) and hobo (BLACKMAN and GELBART 1989 ), produce deletion derivatives at appreciable frequencies owing to abortive excision reactions. We undertook the present experiments to assess the extent to which mariner produces such aberrations and to compare the molecular nature of the aberrations with those produced particularly by the P element. The peach element used in the present experiments is a mariner element that is inserted into the 5' untranslated region of the white gene in the w^pch allele. The insertion changes the phenotype from bright red to peach (JACOBSON et al. 1986 ). Although the peach element has an open reading frame, the element is nonautonomous: it produces either no transposase or a transposase incapable of excising peach. Nevertheless, peach can still be excised efficiently in response to active transposase supplied in trans. Somatic activity of Mos1 in flies carrying the w^pch allele results in a mosaic eye color of red sectors on a peach background. Germline transposase activity results in excision of peach and restoration of the wild-type white gene function in the progeny (BRYAN et al. 1987 ). When under control of the dual hsp70::Mos1 promoter, Mos1 transposase is strongly expressed in the germline and soma even in the absence of heat shock. In flies carrying the w^pch allele, hsp70::Mos1 induces high levels of germline reversion of peach and extreme mosaicism in the eyes (GARZA et al. 1991 ; LOHE et al. 1995 ).

A mutation screen for reduced levels of somatic mosaicism in w^pch; Mos1 flies has been used to isolate mutations in the Mos1 transposase gene (LOHE and HARTL 1996 ; LOHE et al. 1997 ). Here we use a similar screen to identify mutations in w^pch with reduced somatic mosaicism due to lesions in the peach element that impair its ability either to be recognized by transposase or to be excised. In the absence of active transposase, such excision-defective mutations arise at a frequency no greater than the spontaneous mutation rate (LOHE et al. 1997 ). In the presence of active transposase, the frequency of excision-defective mutations in peach is ~0.2% per exposed w^pch allele per generation. Most of the molecular defects can be interpreted in terms of either aberrant mismatch repair or aberrant template-directed gap repair (TDGR). We call these lesions "self-inflicted wounds" because they are initiated by transposase-directed strand scission at one or both ends of peach that are aberrantly repaired using host functions. The major types of peach aberrations are completely different from those isolated in comparable experiments with the P element. In the case of P element, most lesions are internal deletions, and defects overlapping the last 15–17 bp of the ITRs are rare (STAVELEY et al. 1995 ). In the case of peach, the majority of mutations with defective excision have damage to either the 5' or 3' ITR. Only a few of the peach mutations have internal deletions. Moreover, these were large internal deletions, quite unlike the small deletions that characterize nonautonomous mariner elements in natural populations.

One excision-defective peach element afforded an opportunity to estimate the frequency of TDGR in the mariner system. We found that the ratio of homolog-directed gap repair to excision is ~1:2. We also demonstrate here that peach is a hotspot of recombination in Drosophila females, but only in the presence of functional transposase. This finding is of some interest in view of data suggesting that a human mariner-like element in chromosome 17p11.2–p12 is a hotspot of recombination (KIYOSAWA and CHANCE 1996 ; REITER et al. 1996 ).

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Transgenic stocks and genetic markers:
The wpch transgene P[w^pch, ry⁺], described below, contains a nonautonomous mariner element designated peach (JACOBSON et al. 1986 ). The peach element was identified originally as an insertion into the 5' region of the white gene of D. mauritiana (JACOBSON et al. 1986 ), yielding the w^pch allele; the peach insertion is at a position corresponding to coordinate 3643 in the D. melanogaster physical map of LEVIS et al. 1982 (see BRYAN et al. 1990 ). A peach-containing BamHI restriction fragment was isolated from the w^pch allele and was used to replace the corresponding restriction fragment of the white gene of D. melanogaster. The resulting chimeric gene was inserted into a P-element germline transformation vector, creating pP[w^pch, ry⁺], which was used to transform D. melanogaster (GARZA et al. 1991 ). The P[w^pch, ry⁺] transgene is inserted into the X chromosome at map position 27.0 between singed and lozenge; we designate this insert as the wpch transgene to distinguish it from the original w^pch allele, which is at the white locus. In the experiments reported here, the genetic background of all stocks contained the w¹¹¹⁸ deletion at the white locus. In the absence of mariner transposase, flies of genotype w¹¹¹⁸ wpch have peach-colored eyes; in the presence of transposase, the eyes are heavily mosaic owing to the excision of peach from multiple pigment-cell lineages, restoring wild-type pigmentation (GARZA et al. 1991 ). Typical mosaics have >200 red spots in each eye, in an otherwise peach-colored background.

To isolate excision-defective mutations in wpch, we used either of two transposase-producing transgenes. The Mr182 transgene (LOHE et al. 1995 ) is a transposase-producing P[hsp70::Mos1, ry⁺]-182 transgene inserted in chromosome 2. Mr182 includes the heat shock 70 promoter (hsp70) fused to the Mos1 promoter at nucleotide position 58–59 [numbering as in JACOBSON et al. 1986 ] and all of the transposase open reading frame. The Mos1 element is an autonomous mariner element, but it cannot be mobilized in this transgene owing to the absence of the 5' ITR in P[hsp70::Mos1, ry⁺]-182. The promoter in the hsp70::Mos1 transgene has high activity even in the absence of heat shock (LOHE et al. 1995 ). The MrX transgene, located at map position 30.2 on the X chromosome, was derived from Mr182 by P-element mobilization of P[hsp70::Mos1, ry⁺]-182 and selection for an X chromosome encoding high transposase activity (data not shown).

The sn (singed) and lz (lozenge) alleles employed as markers in some of the experiments were sn³ and lz^77a7. Genetic crosses were carried out at 25° on standard cornmeal-molasses medium.

Isolation of mutants:
Excision-defective derivatives of wpch arise in stocks containing both wpch and a source of functional transposase. Such mutants are recognized by absent, or markedly reduced, levels of somatic mosaicism. Some of the mutants also have altered eye color, but the primary basis of the screen was degree of somatic mosaicism. Approximately half of the mutant wpch alleles reported here were obtained from exposure of wpch to the transposase source Mr182 on chromosome 2, which was crossed out of each excision-defective stock immediately after isolation. The rest were obtained from exposure to the transposase source MrX in X chromosomes carrying wpch MrX. Flies containing excision-defective mutations were crossed for one generation to replicate the X chromosome and the progeny were frozen at -70° for subsequent analysis by PCR and sequencing. The same kinds of wpch mutations were found, at statistically indistinguishable frequencies, from both sources of transposase; hence the mutations are grouped by type rather than by origin.

In the continued presence of transposase, excision-defective wpch alleles are more stable genetically than wpch, probably owing to impaired transposase binding to the mutant allele. We were, however, able to obtain secondary mutations in some wpch MrX stocks by selecting for derivatives in which somatic mosaicism was absent or markedly reduced relative to the level found in the original mutant allele. These secondary derivatives are denoted by appending the letter S to the original mutant allele designation.

Molecular analysis:
The peach element present in each mutant wpch allele was characterized by PCR amplification of genomic DNA from single flies using two pairs of primer oligonucleotides, each pair consisting of one primer that anneals to flanking white DNA and one primer that anneals to an internal region of peach. Amplified mutant peach elements were cloned with a TA cloning kit (Invitrogen, Carlsbad, CA) or sequenced directly (LOHE et al. 1996 ). The 5' half of peach was amplified with the primer pair 5'-GCCCATTGCCGGGATTTTTCA-3' and 5'-AGCTCATAGTAAATGACACCG-3', which also amplifies 134 bp of flanking white sequence. The 3' half of peach was amplified with the primer pair 5'-TTGCGAGAGATGGGAAAGATT-3' and 5'-CTGTTTGCCTCCTTCTCTGTC-3', which also amplifies 155 bp of flanking white sequence. DNA sequencing was carried out first to examine the extreme 5' and 3' ends of peach, including the ITRs. The complete mutant element was sequenced in all cases in which the ~350 bp at each end was found to be intact.

Verification of template-directed gap repair:
We used the peach mutant 41R because of its characteristic rosy-like phenotype, reduced excision rate, and unique CCAGTA sequence at the 5' end (see Table 2 below). Putative wpch alleles that had been converted to 41R by TDGR were isolated and analyzed by PCR using primers 5'-TACTTGTACACCTGCCAGTA-3' and 5'-GCCCATTGCCGGGATTTTTCA-3'. These primers are specific to the lesion in 41R, and they amplify a characteristic 149-bp fragment only from the 41R mutant allele.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Preliminary experiments indicated that, in the absence of mariner transposase, the wpch transgene containing the nonautonomous element peach is genetically stable. Although no attempt was made to estimate the spontaneous mutation rate of wpch to forms that exhibit reduced mosaicism when exposed to functional transposase, the spontaneous mutation rate is certainly less than 10^-4 and perhaps much less (LOHE et al. 1997 ). However, we did note that in the presence of mariner transposase, derivatives of wpch showing absent or reduced somatic mosaicism were recovered at appreciable frequencies. We call the lesions in these derivatives self-inflicted wounds because their production appears to be the result of aberrant excision, initiated by at least one transposase-induced scission of one or both DNA strands, followed by repair using host functions. Accordingly, systematic experiments were carried out to recover a large set of such excision-defective wpch transgenes in order to characterize the molecular nature of the lesions. To isolate such mutations, we used either the Mr182 transgene in chromosome 2 or the MrX transgene in the X chromosome, both of which produce high levels of transposase even in the absence of heat shock. (These transgenes have a dual hsp70::Mos1 promoter.)

Two eye-color phenotypes predominate in progeny of flies that carry both the wpch transgene and a source of functional transposase. Most progeny have a strong mosaic eye color like their parents, but some progeny have a uniform bright red eye color that results from germline excision of peach. In addition, males with an exceptional phenotype appeared with a frequency of ~1/500 per generation. In these exceptions, the peach element showed either no somatic excision or a much reduced level of somatic excision. In these flies, the number of red spots per eye was usually less than ~10, substantially smaller than the >200 spots observed in mosaic siblings. Some of the exceptions lacking mosaicism also showed various shades of eye color including white, yellow, orange, or rosy. Each of the mutant lines was retested in the presence of either Mr182 or MrX to confirm the mutant phenotype of the excision-defective wpch allele. Subsequent PCR analysis and DNA sequencing confirmed that, in each of the mutant alleles, either the peach element was missing from its normal position near wpch, or else there were molecular lesions in the elements to which the low rate of excision could be attributed.

We report in detail two classes of mutations that are recovered frequently. They demonstrate not only some of the sequence requirements for efficient excision, but they also appear to illuminate the first steps in the mariner excision reaction. These classes are: (1) aberrations in or near the 5' end of peach (40 alleles) and (2) aberrations in or near the 3' end of peach (11 alleles). We also examine six secondary mutations derived from mutant alleles with 5' lesions, since these serve as important controls against bias in the phenotypic selection based on absent or much reduced somatic mosaicism.

Other types of mutations were also found that warrant a brief description, although they will not be discussed in detail in this report. The most numerous classes had either an excision of peach accompanied by an unusual footprint (11 alleles) or an excision or deletion of peach accompanied by a deletion of flanking white DNA (18 alleles). As might be expected, the latter class had no somatic mosaicism and a bleached-white eye-color phenotype. These kinds of lesions have been reported previously (BRYAN et al. 1990 ). Together with the aberrations at either the 5' or the 3' end of peach mentioned above, these classes account for 80% of the mutant alleles analyzed. The remaining types of lesions included a variety of insertions, inversions, apparent local hops into nearby white sequences, and internal deletions. None of these classes was sufficiently frequent to infer general patterns.

Alteration in or near the 5' end of peach:
A total of 40 alleles in this class were recovered (Table 1). Most of the phenotypes were weakly mosaic in the presence of transposase, as shown by the counts of number of mosaic spots per eye. Nucleotide sequencing revealed alterations in or adjacent to the 5' ITR of the peach element. The 3' ITR and flanking DNA were sequenced and were found to be unaltered. The 5' lesions consisted of either a simple deletion (21 alleles) or of a deletion accompanied by the insertion of additional DNA between the deletion endpoints (19 alleles). In mutant alleles with a simple deletion, the deletion was 1–14 bp in length and it always included nucleotides 2 or 3 of the 5' ITR. The most common mutation was a 1-bp deletion of nucleotide 3 (alleles 22R, 40R, 46R, and 48R). Although 4 other alleles contained an identical 9-bp deletion (1R, 45R, 68R, and 105R), this deletion could have been generated in three different ways, depending on how one aligns the sequences.

The 21 alleles grouped at the top of Table 1 are all simple deletions. The 7 alleles at the very top retain the 5' TA duplication but they have a small deletion at or near the end of the 5' ITR. A deletion of even 1 bp in the 5' ITR is sufficient to reduce somatic excision by a substantial amount (alleles 110R, 22R, 40R, 46R, and 48R), with a greater effect observed for 2–3-bp deletions at the 5' end of the ITR (alleles 64R and 53R). The alleles 69R through 36R imply that some ability to be excised is retained even by wpch alleles that are missing up to 6 bp of the 5' ITR along with the 5' TA and some flanking white sequences.

The 19 mutant alleles grouped at the bottom of Table 1 all have a deletion and an accompanying insertion. Except for allele 66R, the deletions range in size from 1 to 20 bp, and, except for allele 23R, the insertions range in size from 1 to 25 bp. The largest insertion (149 bp in allele 23R) accompanies the smallest deletion (1-bp deletion of nucleotide 4 in peach); at least 143 bp of the inserted DNA in this allele is derived from nucleotides 14,151–14,293 of the white gene, ~7 kb upstream from the insertion point of the peach element in white. In contrast, the smallest insertion (1 bp in allele 11R) accompanies a relatively large deletion of 13 bp. In the remaining 17 mutant alleles, a sequence 3–25 bp in length was present in the region that had been deleted. In general, the sequence of the inserted DNA bears no relationship to the sequence of the deleted DNA.

In virtually all of the alleles with a deletion and insertion, some or all of the inserted DNA appears to originate by replication of either white or peach sequences flanking the deletion endpoint, usually at a distance of 0–10 bp. The underlined sequences in Table 1 are identical between the inserted DNA and either white or peach flanking DNA. The replicated sequence can be derived from either strand, which is indicated in Table 1 by roman font when it is derived from the strand shown and by italics when it is derived from the complementary strand. In the insertions in alleles 65R and 114R, a sequence appears to have been copied more than once to generate tandem repeats. Such "filler" sequences have been observed in P-element derivatives and are presumed to originate by replication slippage (O'HARE and RUBIN 1983 ; KURKULOS et al. 1994 ; STAVELEY et al. 1995 ).

In three mutant alleles (41R, 54R, and 96R), the repair process by chance restored the exact sequence of the 5' ITR, but the TA dinucleotide was altered to either GG (allele 41R), GT (54R), or TC (96R). The phenotype of each of these mutants was a weak mosaic of 10–30 spots per eye. Therefore, the TA dinucleotide immediately adjacent to the 5' ITR is not essential for excision of peach, but the efficiency of excision is greatly enhanced if the TA dinucleotide is present.

The allele 54R was studied further to determine the actual site of strand scission. To excise the peach element in this allele, the transposase must recognize either the GT dinucleotide or a TA dinucleotide, of which the nearest is 7 bp upstream in white or 7 bp downstream in peach. Germline revertants had the footprint 5'-GTACAAGTGTXXXTAAACGC-3', where XXX indicates the extreme 3 bp from either the 5' or 3' end of peach (data not shown). The 5' flanking sequence is the insertion in 54R, and the 3' flanking sequence is that of the white gene at the 3' end of peach. This result indicates that, in allele 54R, the transposase initiates strand scission at the GT dinucleotide adjacent to the 5' ITR rather than at either of the nearby TA dinucleotides.

Among the mutations with a deletion plus insertion, the alleles yielding the greatest levels of somatic excision (114R and 115R) have a sequence resembling the 5' ITR as well as the adjacent TA dinucleotide. In particular, both 114R and 115R have a repaired sequence at the 5' end of peach that reads 5'-TACAAGG...-3', which is equivalent to a C A transversion at position 2 of the 5' ITR with the 5' TA dinucleotide left intact.

Only six alleles in Table 1 had nonmosaic eyes in the presence of transposase. This result would be expected of alleles 23R and 66R, both of which have a substantial part of the 5' ITR replaced with unrelated sequence. The lesions in 65R and 104R are also substantial: 65R differs in 8 of the first 10 nucleotides in the 5' ITR, 104R differs in 4 of the first 6 nucleotides in the 5' ITR, and both alleles are flanked by a GT dinucleotide. Comparing wpch with 30R and 69R reveals smaller differences. In the wpch allele, the 5' dinucleotide and ITR are 5'-TACCAGG-3'. In allele 30R this sequence reads 5'-AAGTAGG-3' and in allele 69R it reads 5'-TGGCGGG-3'. Evidently the altered flanking dinucleotide, along with the multiple substitutions in the first few nucleotides of the 5' ITR, is enough to prevent excision, although this finding does not identify whether the defect is in transposase recognition or strand scission.

Alteration in or near the 3' end of peach:
In 11 alleles there was no alteration in the 5' ITR or flanking white DNA, but there was a deletion at the 3' end of the element (Table 2). With the exception of allele 58R, in which very weak somatic mosaicism was observed, none of the alleles evidenced somatic mosaicism in the presence of transposase. Allele 58R was also exceptional in having a small 3' deletion (8 bp), in contrast to the other alleles. In relative position with respect to the nearest end of peach, overall size, and response to transposase, allele 58R more nearly resembles the 5' deletions 2R and 33R in Table 1 than the other 3' deletions recovered.

The rest of the 3' deletions were large (73–1069 bp) and extended internally into the element from the 3' ITR terminus. The deletions rarely included the 3' TA dinucleotide and flanking white sequences. Allele 10R was the only allele with a deletion accompanied by an insertion; in this case the 4-bp sequence TAAA was inserted at the site of deletion, which may have been copied from the flanking DNA to generate a 4-bp tandem duplication.

Although the alleles in Table 2 were isolated based on reduced or absent somatic mosaicism, their eye colors also differed from wpch, in general according to the length of the deletion. Thus, the smallest deletion was 73 bp in mutant 101R, and the phenotype was dull red. The next smallest deletion was 375 bp in 7R, and this phenotype was orange. All other deletions ranged from 385 to 1069 bp in length and the eye phenotype was pale yellow.

Secondary mutations at the 3' end:
The dramatic difference between the 5' and 3' lesions was quite unexpected. One possibility is that there is a bias in the identification or recovery of 3' lesions, such that primarily large 3' deletions are isolated. To assess this possibility, we reexposed some alleles with 5' lesions and weak somatic mosaicism to active transposase from either Mr182 or MrX and selected from each a derivative in which somatic mosaicism was reduced even more or was absent altogether. The alleles chosen were 86R, 18R, 48R, 100R, 22R, and 1R (Table 2). Secondary mutations were recovered in each of the lines. DNA sequencing showed that the original aberration within the 5' ITR remained in each of the secondary mutations, but that each had an additional deletion at or near the 3' ITR. The alleles with secondary mutations are denoted 86RS, 18RS, and so forth, and the 3' lesions are shown in the bottom part of Table 2. All secondary alleles except 22RS show no somatic mosaicism.

Unexpectedly, five of the six secondary mutations at the 3' end were small deletions of 1–9 bp. They were quite different from the large primary deletions at the 3' end shown in the upper part of Table 2. In fact, the secondary 3' deletions were similar in size and position to the deletions in the 5' ITRs in Table 1, except that they occurred at the 3' end instead of at the 5' end. For example, in the secondary allele 86RS, nucleotide 1284 is deleted. This nucleotide is 3 bp from the 3' terminus of the element, and the deletion is similar in pattern to alleles 22R, 40R, 46R, and 48R in Table 1, in which nucleotide 3 is deleted at the 5' terminus. Similarly, the secondary allele 22RS, which has a very weak mosaic phenotype of one to two spots per eye, has a 3-bp deletion at the 3' terminus accompanied by insertion of the single nucleotide C; the overall result is to change the dinucleotide flanking the 3' ITR from TA to CA, explaining the lower level of mosaicism. The secondary allele 1RS was exceptional in including a relatively large 39-bp deletion spanning 18 bp of the 3' ITR, the TA dinucleotide, and 19 bp of flanking DNA of the white locus.

Template-directed gap repair:
Many of the aberrant repair products in Table 1 have sequences duplicated from elsewhere in the white gene or the peach element, as would be expected of TDGR. To quantify how frequently TDGR accompanies transposase scission of peach, we made use of the unusual sequence of the 41R allele in Table 1. This allele has a 1-bp deletion of the A in the 5' TA dinucleotide, which is replaced with a 6-bp insertion TACTGG. The net effect is that 41R has a 4-bp insertion at the 5' end of peach, which results in reduced mosaicism and a characteristic rosy-like phenotype. In a wpch/41R heterozygote, any peach scission events in wpch that are accompanied by TDGR from the 41R-bearing homolog would yield the 41R type of inverted repeat and the rosy-like phenotype. The unique 5' sequence of the 41R element also enables molecular verification of the repair event (equivalent to a gene conversion), since the CCAGTA sequence at the 5' end of 41R (Table 1) can be used to design 41R-specific PCR primers that will not amplify peach or any other mariner elements.

Accordingly, crosses were carried out between females of genotype sn⁺ wpch lz⁺ MrX/sn 41R lz and males of genotype w^-. The markers sn, wpch, lz, and MrX are at map positions 21.0, 27.0, 27.7, and 30.2, respectively (data not shown), and across the region sn-wpch-lz there is complete interference (data not shown). Among the male progeny of the cross were 225 sn⁺ w⁺ lz⁺, resulting from excision of peach, and 93 sn⁺ lz⁺ with an eye-color phenotype indistinguishable from that of 41R. Genomic DNA from all 93 41R-like males or their progeny was tested by PCR with 41R-specific primers, and all yielded the expected 41R-specific amplification product. Consistent with the occurrence of TDGR, we had also observed occasional sn wpch lz male progeny in the original cross, in which the 41R allele had become converted into an allele indistinguishable from wpch.

Transposase-induced hotspot of recombination:
To test whether the wpch transgene becomes a recombinational hotspot in the presence of transposase, we carried out the experiments reported in Table 3. The symbol , depending on its location in the genotypic formula, indicates absence of either the wpch or the MrX transgene. The experimental cross is in the top row. One type of control is the same as the experimental genotype but lacking a source of active transposase. The recombination fraction in the sn-lz interval in the control was 1.0 map unit smaller than that observed in the experimental crosses, which is statistically highly significant (P < 0.01). The other control lacked the wpch transgene as well as a source of active transposase. In this case the map distance was 1.6 map units smaller than that in the experimental genotypes, and this result has even greater statistical significance (P << 0.01). On the other hand, the map distances observed in the control crosses do not differ significantly from each other. Taken together, these results imply that wpch does act as a hotspot of recombination, but only in the presence of active transposase.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have exploited the somatic activity of the Mos1 mariner transposase to screen for mutations in a nonautonomous target element peach that have an impaired ability to be excised. In the absence of transposase, the wpch transgene, which contains a peach insertion into the white gene, is genetically stable and the eyes are a uniform peach color. In the presence of transposase, the eyes are mottled red due to somatic excision of peach, but mutant wpch alleles with reduced mosaicism are recovered at a frequency of ~1/500 per generation. In the absence of transposase, mutations of this type are not recovered at any appreciable frequency. The screen is apparently very sensitive, since even 1-bp deletions in the 5' ITR were recovered. Altogether we isolated 100 mutant alleles and characterized them by DNA sequencing. Among these, 40 had aberrations in or near the 5' end of peach, 11 had aberrations in or near the 3' end of peach, and 29 had either an excision of peach leaving an unusual footprint or an excision or deletion of peach accompanied by deletion of flanking white DNA. The remaining 20 alleles were an eclectic collection of insertions, inversions, apparent local hops into nearby white sequences, and internal deletions. Most of the mutations appear to represent aberrant products from host-mediated repair of intermediates formed in the excision reaction.

In about half of the 5' mutations, the only alteration was a small deletion, averaging ~6 bp, that always included the beginning of the 5' ITR (Table 1). Most of the other mutant alleles had a small deletion (8 bp average size) accompanied by the insertion of DNA sequences. The inserted DNA appears to have been copied from a template flanking the deletion endpoints, typically 0–10 bp distant from the site of insertion, and often shows evidence of replication slippage leading to two or more direct repeats of a short sequence (O'HARE and RUBIN 1983 ; STAVELEY et al. 1995 ). A single-stranded overhang long enough to promote such events could result from host exonuclease acting on one or both ends of the transposase-induced scission. However, in one allele (23R), 143 bp had been copied from genomic DNA ~7 kb upstream from the peach element.

Deletions were also recovered in the 3' ITR (Table 2), but they were much less frequent (ratio 1:4) and larger than deletions in the 5' ITR (average 669 vs. 6 bp). One possible explanation of this finding is that small deletions at the 3' end are not recovered because they have a negligible effect on the level of somatic mosaicism. To test this possibility, we chose six mutant alleles with small 5' deletions, each of which still exhibited somatic mosaicism in the presence of transposase, and selected a secondary mutation in which the mosaicism was reduced still further. In this case, most of the 3' deletions were typical in size and location to the 5' deletions (average size 9.5 bp; see Table 2), showing that, in fact, small 3' deletions can be detected when they occur.

The asymmetry of the 5' and 3' deletions therefore suggests a polarity in recognition of the ends of peach by the Mos1 transposase, whereby the 5' end is recognized and cleaved first and usually somewhat protected during the repair process to minimize long deletions. When the 5' end is not recognized efficiently due to a small deletion in the 5' end, the 3' end is chosen occasionally as the substrate, and then typical small deletions are formed at the 3' end. The ITRs of peach are 28 bp in length and differ at positions 1, 16, 18, and 26; but perhaps internal recognition sites rather than (or in addition to) the differences in the ITRs account for the polarity. For the six alleles in which secondary 3' deletions were recovered (alleles 1R, 18R, 22R, 48R, 86R, and 100R in Table 1), the primary 5' deletion had an altered TA dinucleotide and one to three changes in the first three nucleotides of the 5' ITR. Certainly, secondary mutations at the 5' end resulting in even less mosaicism would have been detected had they occurred, but no mutations of this type were found. The absence of secondary 5' mutations suggests that the lesion in the 5' end prevented efficient recognition or strand scission by the transposase, and that the intact 3' ITR was targeted instead. For example, the allele 18R (Table 1) has a 5' deletion that resulted in only two changes in peach: the TA dinucleotide is altered to TG, and the C at nucleotide 1 is changed to G. The location of the secondary deletion at the 3' end in allele 18RS (Table 2) suggests that, compared with an intact 5' end, the mutated 5' end in 18R is a poorer substrate for the transposase than the intact 3' end, which in this case has only one change in the corresponding region (nucleotide 1286). These results further suggest that the region including the TA dinucleotide and nucleotides 1–3 of the 5' ITR are critical to enable the transposase to distinguish between the 5' and 3' ITRs.

The mutations with alterations in the 5' ITR often resulted in changes to the flanking TA dinucleotide and/or the extreme 5' end of the ITR. The altered response of these alleles to transposase provides information about the sequences required for transposase recognition. The TA dinucleotide adjacent to the 5' ITR is important, but not essential, for peach excision. This finding contrasts with those obtained for Tc1, in which the 5' TA appears to be essential for excision (VOS et al. 1996 ), and for Tc3, in which mutation of the 5' TA has no effect on excision (VAN LUENEN et al. 1994 ). In the mariner system, one change in the 5' dinucleotide from TA to TC (allele 96R) reduces somatic mosaicism significantly, and two changes (to GG in allele 41R or to GT in allele 54R) also reduce it (Table 1). However, in none of these cases was somatic mosaicism abolished. In the mutant 54R allele, the footprint of germline revertants after peach excision was GTXXXTA, showing that transposase was able to recognize the 5' GT dinucleotide, despite the presence of TA dinucleotides in close proximity. Similarly, if the 5' TA dinucleotide is intact, but nucleotides 1, 2, or 3 of the ITR are deleted (or if there is a C to A transversion at nucleotide 2) somatic mosaicism is reduced but not abolished (Table 1). These results demonstrate that the TA dinucleotide and the first three nucleotides of peach are required for efficient excision. They also reveal the robustness of the Mos1 transposase, because most mutant peach elements with 5' lesions can be attacked by the transposase, although with reduced efficiency.

The mutations observed in this screen differ significantly from those found with the P element (STAVELEY et al. 1995 ). These authors studied a P-element insertion in the 5' end of the vestigial gene in the allele vg²¹. Selection for phenotypic revertants, for a stronger vestigial phenotype, or for altered PCR fragment length from the P element yielded internal deletions in P in which the last 15–17 bp of the termini were retained. The deletions recovered from the phenotypic screens were smaller than those recovered from the PCR screen and were biased by the selection for stronger or weaker vestigial expression, but the PCR screen yielded almost complete deletions that retained only the terminal 15–17 bp on each end (occasionally as small as 10 bp). The terminal 15–17 bp bind with the inverted-repeat binding protein (IRBP), and the authors proposed that the IRBP acts to protect the P-element ends when transposition creates a double-strand gap.

In the case of mariner, the termini of the inverted repeats are not as strongly protected. Among the 5' lesions, we find deletions extending through the terminal 18 bp of the inverted repeat of 28 bp (allele 65R in Table 1), and the 3' deletions often include the entire 3' inverted repeat (Table 2). In contrast, the very different lengths of the 5' deletions and the 3' deletions (average 6–8 vs. 669 bp) suggest that there may be some protection of the 5' end, perhaps by the transposase itself. In proposing such a model, one must be careful about possible biases introduced by phenotypic selection. Our selection was based on reduced somatic mosaicism, and it is conceivable that any peach element that retains the terminal few base pairs of the inverted repeats remains an efficient target for excision; therefore, only those lesions affecting these base pairs would be recovered. However, this possibility can be excluded based on two lines of evidence. The first is that, in three mutant alleles with 5' lesions (41R, 54R, and 96R in Table 1), the repair process by chance restored the exact sequence of the 5' inverted repeat, but the adjacent TA dinucleotide was altered to GG (allele 41R), GT (54R), or TC (96R). Yet all three of these alleles show at least a 10-fold decrease in the number of mosaic spots per eye (Table 1). The second line of evidence is that, among the 20 mutant peach elements not discussed in detail in this article, 6 were found to have deletions completely internal to the inverted repeats, but they nevertheless reduced somatic excision significantly (data not shown).

Template-directed gap repair:
The allele 41R with its characteristic phenotype and 5' end afforded a unique opportunity to test for the presence of TDGR in mariner excision. This process is well established for the P element (GLOOR et al. 1991 ), but has not been demonstrated previously for the mariner/Tc1 superfamily. Transposable elements that move via a cut-and-paste mechanism cannot multiply easily in a population. Since the chromosome from which the element is "cut" loses a copy, whereas the chromosome with the "paste" gains a copy, there is no net gain in copy number due to transposition. TDGR using the element present in the sister chromatid, or in the homologous chromosome, as a template tips the balance, because each transposition accompanied by TDGR results in a net gain of one copy in the genome.

To estimate the rate of homolog-dependent gap repair, we tested the heteroallelic combination sn⁺ wpch lz⁺/sn 41R lz in the presence of transposase, where the 41R allele was flanked by the closely linked markers sn and lz to rule out double recombination. Excision accompanied by 41R-directed gap repair would lead to sn⁺ lz⁺ progeny with the characteristic rosy-like eye color of 41R and the diagnostic 5-bp insertion at the end of the 5' ITR. Excision not accompanied by 41R-directed gap repair would lead to sn⁺ lz⁺ progeny with the characteristic wild-type eye color. The data indicate that homolog-dependent gap repair of mariner is very efficient. For wpch excision, the ratio of peach deletions to gap repair using a 41R allele present in the homologous chromosome as a template is ~2.4:1. This value does not include use of the sister chromatid as a template, which might be expected to be at least equally frequent, yielding an overall frequency of TDGR of perhaps 50%. This fraction could be even higher to the extent that the sister chromatid is favored over the homolog as the repair template. The high frequency of TDGR explains why the maximum frequency of wpch excision observed as revertants to a wild-type eye-color phenotype is only ~15% (LOHE et al. 1997 ), since these are the (perhaps minority of) excision events not accompanied by TDGR. Under similar experimental conditions, a rate of ~15% is also observed for P-element precise excision, probably for the same reason (JOHNSON-SCHLITZ and ENGELS 1993 ).

Hotspot of recombination:
Effects of transposable element mobilization on recombination have been examined closely in the case of the P element in Drosophila (SVOBODA et al. 1995 ; GRAY et al. 1996 ; TANAKA et al. 1997 ). Effects of mariner mobilization have not been examined previously. The studies reported here were motivated by the finding that two mariner-like elements in human chromosome region 17p11.2-p12 are hotspots of unequal recombination (KIYOSAWA and CHANCE 1996 ; REITER et al. 1996 ; ROBERTSON et al. 1996 ). These elements are separated by about 1.5 Mb and flank the gene for peripheral myelin protein 22. Mispairing and unequal crossing over between the mariner-like elements result in reciprocal duplication and deletion products, the former associated with Charcot-Marie-Tooth disease type 1A (CMT1A) and the latter with hereditary neuropathy with liability to pressure palsies (HNPP).

In the case of Drosophila, we found that presence of a homozygous wpch element increased recombination by ~1 map unit in the sn-lz region in the presence of active transposase, but that neither wpch alone nor the transposase alone had any statistically significant effect (Table 3). The rate of recombination in the sn-lz region is ~1 map unit/250 kb (HEINO et al. 1994 ). Hence, recombination is stimulated ~200-fold by the 1.3-kb peach element in the presence of active transposase. The increase is even greater to the extent that the target of the transposase is smaller than 1.3 kb. If the target consisted only of the ITRs, for example, the increase in recombination would be by a factor of ~4500, which makes it a very hot hotspot indeed.

	ACKNOWLEDGMENTS

We are grateful to the suggestion from John Sved that the demonstration of template-directed gap repair would be important in understanding the multiplication and persistence of mariner elements in natural populations. This caused us to be on the lookout for any wpch mutants with a structure that would allow this question to be addressed. We are also grateful to Nathan Blow for contributing some of the DNA sequencing. This work was supported by a grant from the Australian Federal Government (A.R.L.) and by National Institutes of Health grant GM-33741 (D.L.H.).

Manuscript received July 29, 1999; Accepted for publication October 4, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AUGEGOUILLOU, C., Y. BIGOT, N. POLLET, M. H. HAMELIN, and M. MEUNIERROTIVAL et al., 1995  Human and other mammalian genomes contain transposons of the mariner family. FEBS Lett. 368:541-546[Medline].

BLACKMAN, R. K., and W. M. GELBART, 1989 The transposable element hobo of Drosophila melanogaster, pp. 523–529 in Mobile DNA, edited by D. E. BERG and M. HOWE. American Society for Microbiology, Washington, DC.

BRYAN, G. J., J. W. JACOBSON, and D. L. HARTL, 1987  Heritable somatic excision of a Drosophila transposon. Science 235:1636-1638[Abstract/Free Full Text].

BRYAN, G., D. GARZA, and D. L. HARTL, 1990  Insertion and excision of the transposable element mariner in Drosophila. Genetics 125:103-114[Abstract].

ENGELS, W. R., 1989 P elements in Drosophila melanogaster, pp. 437–484 in Mobile DNA, edited by D. E. BERG and M. HOWE. American Society for Microbiology, Washington, DC.

GARCIA-FERNÁNDEZ, J., J. R. BAYASCAS-RAMÍREZ, G. MARFANY, A. M. MUÑOZ-MÁRMOL, and A. CASALI et al., 1995  High copy number of highly similar mariner-like transposons in planarian (Platyhelminthe): evidence for a trans-phyla horizontal transfer. Mol. Biol. Evol. 12:421-431[Abstract].

GARZA, D., M. MEDHORA, A. KOGA, and D. L. HARTL, 1991  Introduction of the transposable element mariner into the germline of Drosophila melanogaster.. Genetics 128:303-310[Abstract].

GLOOR, G. B., N. A. NASSIF, D. M. JOHNSON-SCHLITZ, C. R. PRESTON, and W. R. ENGELS, 1991  Targeted gene replacement in Drosophila via P element-induced gap repair. Science 253:1110-1117[Abstract/Free Full Text].

GRAY, Y. H. M., M. M. TANAKA, and J. A. SVED, 1996  P-element-induced recombination in Drosophila melanogaster: hybrid element insertion. Genetics 144:1601-1610[Abstract].

HARTL, D. L., 1996  The most unkindest cut of all. Nat. Genet. 12:227-229[Medline].

HARTL, D. L., A. R. LOHE, and E. R. LOZOVSKAYA, 1997a  Modern thoughts on an ancyent marinere: function, evolution, regulation. Annu. Rev. Genet. 31:337-358[Medline].

HARTL, D. L., E. R. LOZOVSKAYA, D. I. NURMINSKY, and A. R. LOHE, 1997b  What restricts the activity of mariner-like transposable elements? Trends Genet. 13:197-201[Medline].

HEINO, T. I., A. O. SAURA, and V. SORSA, 1994  Maps of the salivary gland chromosomes of Drosophila melanogaster.. Dros. Inf. Serv. 73:619-738.

JACOBSON, J. W., M. M. MEDHORA, and D. L. HARTL, 1986  Molecular structure of a somatically unstable transposable element in Drosophila.. Proc. Natl. Acad. Sci. USA 83:8684-8688[Abstract/Free Full Text].

JARVIK, T. and K. G. LARK, 1998  Characterization of Soymar1, a mariner element in soybean. Genetics 149:1569-1574[Abstract/Free Full Text].

JOHNSON-SCHLITZ, D. M. and W. R. ENGELS, 1993  P-element-induced interallelic gene conversion of insertions and deletions in Drosophila melanogaster.. Mol. Cell. Biol. 13:7006-7018[Abstract/Free Full Text].

KIYOSAWA, H. and P. F. CHANCE, 1996  Primate origin of the CMT1A-REP repeat and analysis of a putative transposon-associated recombinational hotspot. Hum. Mol. Genet. 5:745-753[Abstract/Free Full Text].

KURKULOS, M., J. M. WEINBERG, D. ROY, and S. M. MOUNT, 1994  P element-mediated in vivo deletion analysis of white-apricot: deletions between direct repeats are strongly favored. Genetics 136:1001-1011[Abstract].

LEVIS, R., P. M. BINGHAM, and G. M. RUBIN, 1982  Physical map of the white locus of Drosophila melanogaster.. Proc. Natl. Acad. Sci. USA 79:564-568[Abstract/Free Full Text].

LOHE, A. R. and D. L. HARTL, 1996  Autoregulation of mariner transposase activity by overproduction and dominant-negative complementation. Mol. Biol. Evol. 13:549-555[Abstract].

LOHE, A. R., D.-A. LIDHOLM, and D. L. HARTL, 1995  Genotypic effects, maternal effects and grand-maternal effects of immobilized derivatives of the transposable element mariner.. Genetics 140:183-192[Abstract].

LOHE, A. R., D. T. SULLIVAN, and D. L. HARTL, 1996  Genetic evidence for subunit interactions in the transposase of the transposable element mariner.. Genetics 144:1087-1095[Abstract].

LOHE, A. R., D. DE AGUIAR, and D. L. HARTL, 1997  Mutations in the mariner transposase: the "D,D(35)E" consensus sequence is nonfunctional. Proc. Natl. Acad. Sci. USA 94:1293-1297[Abstract/Free Full Text].

MORGAN, G. T., 1995  Identification in the human genome of mobile elements spread by DNA-mediated transposition. J. Mol. Biol. 254:1-5[Medline].

NIKITIN, A. G. and R. C. WOODRUFF, 1995  Somatic movement of the mariner transposable element and lifespan of Drosophila species. Mutat. Res. 338:43-49[Medline].

O'HARE, K. and G. M. RUBIN, 1983  Structure of P transposable elements and their sites of insertion and excision in the Drosophila melanogaster genome. Cell 34:25-35[Medline].

OOSUMI, T., W. R. BELKNAP, and B. GARLICK, 1995  Mariner transposons in humans. Nature 378:672[Medline].

REITER, L. T., T. MARUKAMI, T. KOEUTH, L. PENTAO, and D. M. MUZNY et al., 1996  A recombination hotspot responsible for two inherited peripheral neuropathies is located near a mariner transposon-like element. Nat. Genet. 12:288-297[Medline].

ROBERTSON, H. M., 1993  The mariner transposable element is widespread in insects. Nature 362:241-245[Medline].

ROBERTSON, H. M. and D. J. LAMPE, 1995  Recent horizontal transfer of a mariner transposable element among and between Diptera and Neuroptera. Mol. Biol. Evol. 12:850-862[Abstract].

ROBERTSON, H. M. and E. G. MACLEOD, 1993  Five major subfamilies of mariner transposable elements in insects, including the Mediterranean fruit fly, and related arthropods. Insect Mol. Biol. 2:125-139[Medline].

ROBERTSON, H. M., K. L. ZUMPANO, A. R. LOHE, and D. L. HARTL, 1996  Reconstructing the ancient mariners of humans. Nat. Genet. 12:360-361[Medline].

STAVELEY, B. E., T. R. HESLIP, R. B. HODGETTS, and J. B. BELL, 1995  Protected P-element termini suggest a role for inverted-repeat-binding protein in transposase-induced gap repair in Drosophila melanogaster.. Genetics 139:1321-1329[Abstract].

SVOBODA, Y. H. M., M. K. ROBSON, and J. A. SVED, 1995  P-element-induced male recombination can be produced in Drosophila melanogaster by combining end-deficient elements in trans.. Genetics 139:1601-1610[Abstract].

TANAKA, M. M., X. M. LIANG, Y. H. M. GRAY, and J. A. SVED, 1997  The accumulation of P-element-induced recombinants in the germline of male Drosophila melanogaster.. Genetics 147:1769-1782[Abstract].

VAN LUENEN, H. G. A. M., S. D. COLLOMS, and R. H. A. PLASTERK, 1994  The mechanism of transposition of Tc3 in C. elegans. Cell 79:293-301[Medline].

VOS, J. C., I. DE BAERE, and R. H. A. PLASTERK, 1996  Transposase is the only nematode protein required for in vitro transposition of Tc1.. Genes Dev. 10:755-761[Abstract/Free Full Text].

This article has been cited by other articles:

				J. M. Marshall The Impact of Dissociation on Transposon-Mediated Disease Control Strategies Genetics, March 1, 2008; 178(3): 1673 - 1682. [Abstract] [Full Text] [PDF]

				C. Miskey, B. Papp, L. Mates, L. Sinzelle, H. Keller, Z. Izsvak, and Z. Ivics The Ancient mariner Sails Again: Transposition of the Human Hsmar1 Element by a Reconstructed Transposase and Activities of the SETMAR Protein on Transposon Ends Mol. Cell. Biol., June 15, 2007; 27(12): 4589 - 4600. [Abstract] [Full Text] [PDF]

				K. Lipkow, N. Buisine, D. J. Lampe, and R. Chalmers Early Intermediates of mariner Transposition: Catalysis without Synapsis of the Transposon Ends Suggests a Novel Architecture of the Synaptic Complex Mol. Cell. Biol., September 15, 2004; 24(18): 8301 - 8311. [Abstract] [Full Text] [PDF]

				S. R. Yant and M. A. Kay Nonhomologous-End-Joining Factors Regulate DNA Repair Fidelity during Sleeping Beauty Element Transposition in Mammalian Cells Mol. Cell. Biol., December 1, 2003; 23(23): 8505 - 8518. [Abstract] [Full Text] [PDF]

				D. A. O'Brochta, N. Sethuraman, R. Wilson, R. H. Hice, A. C. Pinkerton, C. S. Levesque, D. K. Bideshi, N. Jasinskiene, C. J. Coates, A. A. James, et al. Gene vector and transposable element behavior in mosquitoes J. Exp. Biol., November 1, 2003; 206(21): 3823 - 3834. [Abstract] [Full Text] [PDF]

				S. E. J. Fischer, E. Wienholds, and R. H. A. Plasterk Continuous Exchange of Sequence Information Between Dispersed Tc1 Transposons in the Caenorhabditis elegans Genome Genetics, May 1, 2003; 164(1): 127 - 134. [Abstract] [Full Text] [PDF]

				A. R. Lohe and D. L. Hartl Efficient Mobilization of mariner in Vivo Requires Multiple Internal Sequences Genetics, February 1, 2002; 160(2): 519 - 526. [Abstract] [Full Text] [PDF]

				E. R. Lozovsky, D. Nurminsky, E. A. Wimmer, and D. L. Hartl Unexpected Stability of mariner Transgenes in Drosophila Genetics, February 1, 2002; 160(2): 527 - 535. [Abstract] [Full Text] [PDF]

				D. L. Hartl Discovery of the Transposable Element Mariner Genetics, February 1, 2001; 157(2): 471 - 476. [Full Text]

				L.-L. Han, M. P. Keller, W. Navidi, P. F. Chance, and N. Arnheim Unequal exchange at the Charcot-Marie-Tooth disease type 1A recombination hot-spot is not elevated above the genome average rate Hum. Mol. Genet., July 22, 2000; 9(12): 1881 - 1889. [Abstract] [Full Text] [PDF]

				M. Caceres, M. Puig, and A. Ruiz Molecular Characterization of Two Natural Hotspots in the Drosophila buzzatii Genome Induced by Transposon Insertions Genome Res., August 1, 2001; 11(8): 1353 - 1364. [Abstract] [Full Text] [PDF]