p-element-induced male recombination and gene conversion in

Copyright 0 1996 by the Genetics Society of America

P-Element-Induced Male Recombination and Gene Conversion in Drosophila

Christine R. Preston and William R. Engels

Laboratory of Genetics, University of Wisconsin,, Madison, Wisconsin 53706 Manuscript received June 14, 1996

Accepted for publication September 11, 1996

ABSTRACT A P-element insertion flanked by 13 restriction fragment length polymorphism (RFLP) marker sites

was used to examine male recombination and gene conversion at an autosomal site. The great majority of crossovers on chromosome arm 2R occurred within the 4kb region containing the Pelement and RFLP sites. Of the 128 recombinants analyzed, approximately two-thirds carried duplications or deletions flanking the Pelement. These rearrangements are described in more detail in the accompanying report. In a parallel experiment, we examined 91 gene conversion tracts resulting from excision of the same autosomal P element. We found the average tract length was 1463 bp, which is essentially the same as found previously at the white locus. The distribution of conversion tract endpoints was indistinguishable from the distribution of crossover points among the nonrearranged male recombinants. Most recombination events can be explained by the “hybrid element insertion” model, but, for those lacking a duplication or deletion, a second step involving double-strand gap repair must be postulated to explain the distribution of crossover points.

T HE discovery of male recombination in certain hybrids between wild-caught male and laboratory

strain female Drosophila was the first indication of the existence of the Pfamily of transposable elements (HIR- AIZUMI 1971). In more than 20 years following this ob- servation, P elements have been studied in great detail and have been used as ubiquitous tools for Drosophila molecular genetics (ENGELS 1989, 1996). Much is now known about how these elements transpose and how their activity is regulated. However, the mechanism by which they induce male recombination has never been resolved. When P elements are mobilized, recombination can be detected in the progeny of either sex. How- ever, the phenomenon is usually called “male recombination” because it is most conspicuous in males where meiotic recombination is absent.

Many experiments employing classical genetic tech- niques and strains bearing multiple P elements in scat- tered locations revealed several features about the process (HIRAIZUMI et al. 1973; KIDWELL and KIDWELL 1976; K~DWELL et nl. 1977; WOODRUFF and THOMPSON 1977; SVED 1978; WOODRUFF et al. 1978; ENGELS 1979; YANNO- POULOS 1979; ISACKSON et aZ. 1981; SINCLAIR and GRIGLI-

From these studies, the average frequency of male recombination in “dysgenic” hybrids was found to be - 1% per chromosome arm but varied depending on which strains were used. The distribution of crossover points within the genome was also a function of the P strain. Most P-induced male recombination was found

ATTI 1985; MCCARRON et al. 1989; DUTTAROY et al. 1990):

Corresponding author: William R. Engels, Laboratory of Genetics, University of Wisconsin, Madison, WI 53706. E-mail: [email protected]

Genetics 144 1611-1622 (December, 1996)

to occur in mitotic germ cells before meiosis. There was little or no interchromosomal effect or positive in- terference, as is seen with normal meiotic recombination. More recently, i t has been possible to study male recombination in better detail with the use of stocks in which Pelement transposase is supplied from a single, nonmobile, source (ROBERTSON et al. 1988), and a single mobile P element is present at a known site (SVED et al. 1990,1991,1995; MCCARRON et al. 1994; SVOBODA et al. 1995). These studies showed that the frequency of male recombination was greatly increased when a mobile Pelement was made homozygous, and that mobilizing P elements in somatic cells leads to somatic recombination.

A key question in these studies was whether the recombination events were occurring at the sites of mobile P elements, as opposed to random sites that can be attacked by P transposase. The experiments favoring random sites (MCCARRON et aZ. 1989, 1994; DUTTAROY et al. 1990) were limited by the use of highly selective screens for recombination, raising the question of whether the recombinations recovered are actually typi- cal of the majority of events. The work favoring localiza- tion to Psites (SVED et al. 1990, 1991, 1995; SVOBODA et al. 1995) relied on large genetic intervals and therefore could not place the recombination point with precision.

Another mystery was the possible relationship between male recombination and double-strand break repair leading to gene conversion. Strong evidence showed that P-element transposition leaves behind a double-strand break that is repaired by copying in a stretch of homologous template sequence (ENGELS et al. 1990; GLOOR et al. 1991; JOHNSON-SCHLITZ and ENGELS 1993; NASSIF and ENCELS 1993; ENGELS et al. 1994; NAS-

1612 C. R. Preston and W. R. Engels

SIF et al. 1994). The template is often the homologue, which would seem to provide a clear opportunity for male recombination, especially through the resolution of Holliday junction intermediates. However, several observations argued against this possibility. (1) There was only a weak correlation between conversion and recombination (ENGELS et al. 1990; JOHNSON-SCHLITZ and ENGELS 1993). (2) Many recombinant chromosomes retained a copy of the original P element (SVED et al. 1995), which would have been removed had the break been repaired off the homologue. (3) The array of conversion events observed suggested an underlying gap repair mechanism that did not involve Holliday junctions (NASSIF et al. 1994).

In the present work, we resolve these questions through the use of restriction site polymorphisms flanking an autosomal P-element site. This strategy allowed us to analyze male recombinant chromosomes with molecular precision without the necessity of using strong selective measures that tend to complicate the interpretation. In addition, the conversion tract distribution was obtained based on the same restriction site markers and mobile P element. The results show that the great majority of P-induced recombination sites do occur within 2 kb on either side of the mobile P element. We conclude that most male recombination events are not due to the random transposase cuts hy- pothesized by CHOVNICK and coworkers. Approximately two-thirds of the recombinants carried a flanking duplication or deletion, and the remaining third were simple crossovers. Among the simple recombinants, the distribution of crossover points matched the distribution of conversion endpoints associated with double-strand gap repair. The duplications and deletions are described in detail in the accompanying paper (PRESTON et al. 1996). Overall, the results suggest that most male recombination events occur by an aberrant transposition process (GRAY et al. 1996; PRESTON et al. 1996) as opposed to double-strand break repair. A minority of events might also be explained as a secondaly step following double- strand break repair initiated by unligated single-strand nicks at the conversion tract endpoints.

MATERZALS AND METHODS

Drosophila crosses: Flies were raised on standard corn- meal-molasses-agar medium and grown at room temperature (24" 5 1) unless otherwise noted. All genetic symbols not described in the text are in the Drosophila reference works (LINDSIXY and ZIMM 1992; FLYEME 1996).

DNA sequencing: All sequencing of PCR products from recombinants was done by the dye termination method (LEE et al. 1992; ROSENTHAI. and CHARNOCK~ONES 1992) using AB1 Prism model 373 and 377 automated sequencers according to the manufacturers instructions. Primers for the 50C region were based on the initial determination of portions of the sequence surrounding the insertion P(CaSpeR)(5OC), which was done by the dideoxy method (SAMBROOK et al. 1989) using plasmid clones described previously (SED et al. 1991).

Sequence information in these areas was kindly provided by J. SED, L. BIACKMAN and C. FLORES.

PCR and restriction mapping: PCR was performed in 14 pl volumes in Stratagene RoboCyclers on Drosophila genomic DNA samples extracted by the simple single-fly procedure (GLOOR and ENGELS 1992). Annealing and denaturing were at 61" and 94" for -1 min each. Extension was at 72" for between 1 and 10 min, depending on the expected length of the amplicon. The number of cycles was 29 in most cases. All amplifications used Taq DNA polymerase, but those requiring amplification of fragments longer than 3 kb also included a small amount of another thermostable DNA polymerase that has 3 exonuclease, thus enabling "long PCR" (BARNES 1994).

Amplified DNA was cut with restriction enzymes without further purification. Each PCR tube received 3-8 units of enzyme and sufficient 1OX buffer to bring the final concentration to 1 X . These components were added under the oil to the aqueous part, and the mixture was incubated for 90 min at the optimum reaction temperature for each enzyme. Samples were then electrophoresed on gels of agarose, Metaphor (FMC), or a mixture of the two, at a concentration determined by the expected size of the DNA fragments.

Most reactions utilized the primers shown in Figure lA, whose sequences are as follows: D5 CTAACGAGCAAAATC- ACTGC; G4 CTATTGCTAGACTCAGATTGC; D4 CTGACT- GTTGGCGATAAGAA, G3 TTCTTATCGCCAACAGTCAG; D3 GTAGGCCAACCAATTTGGAG GO GCCATCCTCCAA- ATTGGTTG; 2223 CGTCCGCACACAACCTTTCC; 2231 TCGCTGTCTCACTCAGACTC; G2 ATTTCGATTTCGGCG ATACG DO GCAAACGCAAACTCAATCGC; D2 ACGTCG TTGGTTCTCGTTGC; G1 TCATTTTGCTGCAAGTCTCG; Dl AATCACAGCTCCCTTGGTGC. For most amplifications, primer G4d GCAGTGATTTTGCTCGTTAG, which is located very close to G4, was used in place of G4. Some PCRs also included a second pair of primers to serve as a positive control for integrity of the genomic target DNA and PCR conditions and reagents. These primers were 4849 CTGGAAGTAGAG ATGGTCGC and 1422 GCGCAACTTCGCTCGC, which amplify a 1.0-kb fragment from the singed locus.

RESULTS

Development of the marker system: We used a P{CaSpeR) insertion at cytological position 50C to study both male recombination and conversion tracts. This insert was chosen because it has been used previously in studies of male recombination (SVED et al. 1990, 1991,1995; SVOBODA et al. 1995), and because the non- autonomous P element, PICaSpeR} carries a mini-white gene (PIRROTTA 1988).

The first step was to obtain a set of closely linked markers on both sides of P(CaSpeR) (50C). These markers would serve to localize the crossover point in male recombination events and also to delineate the conversion tracts in P-induced gene conversion. Our approach was to screen a series of distantly related Drosophila lines to find one that differed from the line carrying P(CaSpeR}(50C) at several restriction sites near the insertion point.

Primer pairs G4d-D4, G3-D3, G2-D2 and GI-Dl can be used to amplify four fragments of -1 kb each in the vicinity of the P(CaSpeR}(50C) insert, as shown in Fig- ure 1B. We amplified these four segments from the P(CaSpeR}(5OC)-bearing stock plus 21 other Drosoph-

Recombination and Conversion 1613

A 500 bp 1000 bp M

G4d G3 G1 b b b I I 1: I I I D5 D4 D3 ~ DO D2 Dl

B fragment 4 fragment 3 I fragment 2 fragment 1

I *-

r I, I I

C

FIGURE 1.-Maps of the P(CaSpeR}(50C) insertion region. All map elements are drawn to scale with the exception of the CuSpeR transposon, which is 4982 bp, and the arrowheads indicating primer sites. Orientation is according to the standard Drosophila second chromosome map, with the centromere to the left and the 2R tip to the right. The orientation of the mapped 50C region with respect to the centromere was inferred from the recombinants described below. (A) Positions of primers. See MATERIALS AND METHODS for primer sequences. (B) Amplified segments used to screen for RFLPs. Fragments 4, 3 and 1 were used to characterize conversion tracts and recombinants. In most cases, fragment 2 was not used for this purpose, since the much smaller amplicons, DO-GO or DO-2231, were used to test for the E d site. (C) RFLP differences between chromosomes A and C. Cutting sites on chromosome A are shown above the line, and those on chromosome C are below. The map does not include sites on both A and C chromosomes. For some of the tests, we used CluI (ATCGAT) to test for the TuqI (TCGA) difference. The feature indicated as a 300-bp insert on chromosome C could also be considered as a 300-bp deletion on the A chromosome.

ila stocks and cut each with various restriction enzymes to identify differences. The 21 stocks were selected on the bases of being free of Peiements (i.e., M strains) and carrying second chromosomes with no known recent common ancestor with the chromosome bearing P(CaSpeR} (50C). The restriction enzymes were mostly 4cutters to maximize the chances of finding polymorphisms. Two of the chromosomes, designated A and B, were found to be highly diverged from the CaSpeR- bearing homologue (hereafter designated C). Chromo- some A, which was used for all of the subsequent work, differed from the CaSpeR chromosome at 12 restriction fragment length polymorphism (RFLP) sites. A series of partial- and double-digests with amplified DNA yielded the map in Figure 1C. The sites included 11 changes in a recognition sequence and one length difference. In addition, subsequent sequencing showed that chromosomes A and C differed at an EarI site located -30 bp to the right of the P[CaSpeR] insertion site. In total, the 13 RFLP differences between chromosomes A and C covered a stretch of -4350 bp centered around the P[CaSpeR}(50C) insertion.

Gene conversion tracts: Previous work at the white locus suggested that in the presence of transposase, P elements excise leaving behind a double-strand break.

Gap expansion and repair result in sequence being cop- ied into the excision site from a,homologous template (ENGELS et al. 1990; GLOOR et al. 1991; JOHNSON- SCHLITZ and ENGELS 1993; NASSIF et al. 1994). The most likely templates are the sister chromatid and the homologous chromosome. Excision of P{CaSpeR] (50C) and use of the A chromosome homologue as template would result in clean loss of the CuSpeR element con- comitantly with some of the RFLP markers on the C chromosome being replaced with the corresponding A markers.

In the previous work, clean loss of the P insert in whd was phenotypically detectable. However, the 50C insertion does not mutate any gene, and its excision is therefore not phenotypically detectable. In addition, the loss of the internal w' gene contained in P(CaSpeR}(50C) is not a reliable indicator of clean loss of the element because internal deletion will yield the same result. Therefore, we used a PCR screen, as shown in Figure 2, to collect conversion events. These events are expected to occur in the premeiotic germline of the gen-0 males and recovered in the gen-1 sons. From each gen-0 cross, up to three w a1 b Dr+ sons were selected and allowed to mate individually with balancer females as shown in


CATG CATG CATG eR CATG

h. a1 b ,cn 7 bw+ gen-0 -.

a1 a Dr 82-3 h 9 a1 b I c n bw+ 9 x 4 9 al+ b+,cn+ d+ ’ D ~ +

.- 0 - CATG CATG

- Screen for loss of

site (CATG) CaSpeR and

gen-1

gen-2 gen-3

CyO, Roi en bw -. a1 b ,cn J bw

en bw 9 x 4 a1 b -cn bw+ ’ ’ Dr’ Dr+ ‘ - CATG CATG

L!

Select Cy cn bw males and females. Cross them to test for lethality and make homozygous . stocks. All lethal and recombinant chromosomes were eliminated.

mim RFLP markers for C - RFLP markers for A

FIGURE 2,”Scheme for collecting P-induced gene conversions. Excision of P{CuSpeR/(5OC) in the gen-0 male germline was recovered in the gen-1 sons selected for w a1 b Dr’ phenotype. Transposase was eliminated in the gen-1 by using only Dr+ sons. Selection for w phenotype provided a crude initial screen for full or partial loss of CaSpeR. Screening for the wild-type DO-GO amplicon provided a more precise indication of the clean loss of CuSpeR, which was eventually confirmed by loss of the NluIII site. The maternal second chromosome carries a Pfragment of -430 bp derived from an internal excision of P{CuSpeR/(5OC) in a previous experiment. This fragment retains the NluIII site (CATG) on both sides, but lacks the necessary sequences for mobility. Thus, it serves as a nonmobile marker to ensure that no wild-type-size DO-GO amplicon comes from the maternal allele. There were 391 iso-8 gen-0 crosses that were fertile, of which 191 yielded at least one son with the w a1 b Dr+ phenotype. No more than three such sons were tested by PCR from each cross. Male recombination events in the RO germline were eliminated in generation R2 if no cn bw progeny appeared. Lethal chromosomes were eliminated in R3. RFLP tests to map the conversion tracts were carried out in the homozygotes from R3 or later. The RO males were derived such that the transposase source, A2- 3, and CuSpeR came from different parents so as to avoid mobilizing CuSpeR before the experiment.

Figure 2. The same gen-1 males were then used for DNA extraction and PCR.

Those gen-1 males that yielded an approximately wild-type-size amplicon with primers DO and GO (145 bp, see Figure 1A) were candidates for conversion. To confirm that a clean loss of CuSpeR had taken place, we then cut the DO-GO amplicon with NZuIII, whose recognition sequence, CATG, begins and ends the P- element sequence. Loss of the NZuIII site is therefore a strong indication of a clean CuSpeR excision and rules out the possibility of a near precise excision in which a small part of one or both P ends is retained. Such near-precise losses are known to be the most common type of P-element excision when no wild-type template is present (TSUBOTA and SCHEDL 1986; JOHNSON- SCHLITZ and ENGELS 1993). Those lines in which the NZuIII site had been removed were retained as conversion candidates, and their gen-3 progeny were used for further tests.

We performed this screen on 369 gen-1 white-eyed a1 b sons and identified 176 with approximately the wild-type size DO-GO fragment. The rest had either a larger-than-wild-type DO-GO amplicon, indicative of an internal CuSpeR deletion, or else no amplicon at all, as would be expected if CuSpeR remained intact. All but eight of the 176 lines with a wild-type DO-GO amplicon had lost the NlaIII cut site, indicating full loss of the CuSpeR element, and presumably conversion from the homologue. Since 36.9% of the gen-1 a1 b sons were

white eyed, we estimate the overall frequency of conversion at 0.369(168/369) = 16.8%. This frequency is con- sistent with results at white when the homologue was used as the template ( JOHNSON-SCHLITZ and ENCELS 1993; NASSIF and ENGELS 1993), and it is much greater than O.1%, the frequency of npntemplated precise P losses (ENGELS et uZ. 1990). Therefore, we conclude that most or all of the 168 lines represent conversion events.

We then selected for further analysis 100 independent conversion lines by taking only one gen-1 male from each gen-0 cross. This selection ensures that no more than one representative from a premeiotic event is included in the sample. Nine of these were subse- quently eliminated because of homozygous inviability, recombination between cn and bw, or evidence of structural changes seen by PCR and restriction mapping. Each of the remaining 91 presumptive conversion chromosomes was made homozygous for DNA extraction and analysis by PCR and restriction mapping. Frag- ments 4, 3 and 1 (Figure 1B) were amplified and cut with the restriction enzymes in Figure 1C to determine whether each site was derived from the A or C chromosome. A conversion tract was defined as a run of A markers. Those in which one or both conversion tract endpoints appeared to lie in the interval between the M p I marker and the 300-bp size difference (see Figure 1) were also amplified with primers DO and GO and cut with EurI.

All but one of the conversion tracts were found to


(5OC) x = 0.99864 (white) x = 0.99855

/

-2500 -2000 -1500 -1000 -500

Position 500 1000 1500 2000

FIGURE 3.-Distribution of conversion frequencies. The proportion of the 91 conversion tracts that include each of the 13 RFLP markers is indicated. Coordinates are given as distances from the insertion site of P(CaSpeR)(50C). Five of the lines were lost before the right endpoints could be tested for E d . Since 73 of the other 82 lines had their right endpoints to the right of the E d site, we assumed the same proportion (4/5) of the untested ones would extend beyond Ea& The theoretical curves are xn, where n is the positive distance to the CaSpeR insertion point. The value of x was estimated by maximum likelihood assuming each endpoint is an independent trial, as described previously (GLOOR et al. 1991).

be continuous, meaning the run of A markers was not interrupted by C markers. In addition, all conversion tracts overlapped the P(CaSpeR](50C) insertion site. These properties are similar to those of conversion tracts obtained previously at the white locus (GLOOR et al. 1991; NASSIF and ENGELS 1993). The frequency at which each marker site was converted is shown in Figure 3.

For a quantitative comparison with previous work at the white locus, we used a model in which conversion tract endpoints are assumed to be determined by a random gap-enlargement process, as described in footnote 23 of GLOOR et al. (1991). According to this model, the distribution of conversion tracts is determined by a single parameter, x, defined as the probability of an exonuclease removing one base. Thus, the proportion of conversion tracts that include a site n nucleotides away from the P insertion site is x”. Using a maximum likelihood procedure as described (GLOOR et al. 1991), we estimate x = 0.998635 2 0.00011 from the present data. The corresponding distribution is indicated by the solid curve in Figure 3. This value corresponds to an average conversion tract length of 2x/( 1 - x) = 1463 bp. This result is in remarkably good agreement with the analogous work at the white locus, where x was estimated to be 0.99855 (GLOOR et al. 1991), and the average tract length was 1379. The dashed curve in Figure 3 shows the theoretical distribution for the white locus data.

Male recombination screen: The scheme in Figure 4A was used to generate a collection of male recombinant lines in which the A and C RFLP markers could be used to locate the crossover point. Note that recombination occurred in the germline of the gen-0 males, which are identical to the gen-0 males of Figure 2 that were used to obtain conversion events. The only pheno- typic selection was for recombination between cinnabar and brown, which are on opposite ends of chromosome arm 2R (FLYBASE 1996). The frequency of male recombination was found to be 1.05% among 23,188 flies scored. These were equally distributed among the two reciprocal classes, cn bw+ and cnf bw. The frequency of male recombination decreases when the rearing temperature of the gen-0 males is reduced to 19” (PRESTON et al. 1996).

Some of the 243 recombinant males from generation gen-1 were members of clusters of recombinants from single gena crosses. By taking no more than one of each recombinant class from each cross, we obtained 128 lines for further study. Of these, 26 were homozygous lethal and 17 were sterile. These lines were maintained as het- erozygous stocks over the balancer chromosome, CyO. Complementation tests reported elsewhere (PRESTON et al. 1996) showed that most of these were allelic to one or a combination of three essential loci, one female sterility locus and two male sterility loci in the 50CD region. PCR analysis showed that most of the lethal or sterile effects


J. Select recombinants

Select Cy cn bw+ DrC or gen-2 Cy cn+ bw Dr+ males and females.

Cross them to test for lethality and gen-3 make homozygous stocks.

B

123,188 I Total Cy Roi Dr+ sons examined from generation G1 above

c c 11271

cn + + bw recombinants recombinants

nonrecombinants (discarded)

(1 281 11151 Independent recombinants cluster members, etc. retained for funher study (discarded)

c H lssl Simple Rearranged

1 1291 1161 (811 El

FIGURE .l.--P-induced male recombination. (A) Mating scheme. Recom- bination events occur in the germline of the gen-0 males and are recovered among gen-1 progeny. All 677 fertile gen-0 crosses were iso-8. Wavy lines indicate balancer chromosomes. White and hatched boxes represent the A and C RFLP markers, as in Figure 2. Ques- tion marks on the recombinant chromosomes signify the possibility that all or part of CuSpeR remains at the recombination site. The balancer C y 0 carries amorphic al- leles of both cn and bw (CRAYMER 1980) to facili- tate recombinant screening. The gena males were derived as in Figure 2. (B) Flow chart showing the classification of gen-1 progeny and recombinant chromosomes.

Crossover Crossover Crossover point at 50C point elsewhere point at 50C point elsewhere

Crossover

can be attributed to structural changes in the vicinity of the crossover point (PRESTON et al. 1996). The 85 recombinants that were not lethal or sterile were kept as homozygous stocks. Figure 4B shows how the recombinants were classified.

Crossover points usually lie in the vicinity of CaSpeR: Crossover points were determined using the RFLP markers in Figure 1C. Amplified fragments 4, 3 and 1 (Figure 1B) were cut with the restriction enzymes of Figure 1C to determine the pattern, A or C, of each. This was done with homozygotes for the nonlethal lines. For the lethals, it was done in heterozygotes in two ways: recombinant/A and recombinant/(=. We could then deduce the pattern of the lethal by subtracting the

bands of the homologue. In addition, those whose crossover points lay between the MspI marker and the 300-bp size difference were also tested for the Earl RFLP using amplicon DO-GO or DO-2231, depending on whether CaSpeR was present.

The crossover point was considered to be within the 4kb interval surrounding P(CaSpeRJ(5OC) if the RFLP markers were C on the left side of the interval and A on the right for cn bw+ recombinant, and the reverse for cnf bw recombinants. Of the 128 recombinants, we found that 110 followed this rule (81 + 29 in Figure 4B), implying that 86% of the recombinants had crossover points within the 4kb interval. Since this interval is only 0.02% of the total distance between cn and bw,

1617

500bD 1 W o b

Recombination and Conversion

A Crossover Points of Simple Recombinants

FIGURE 5,"Distributions of crossover points and conversion tract endpoints. Mapping is to the intervals between RFLP sites ( t) , as shown in Figure 1C. Some of the chromosomes, indicated by S , were not tested with E d , and their corresponding blocks could therefore lie in either of two potential intervals. These include five conversion tract endpoints and one crossover point. These are shown in the figure distributed between the two intervals in approximately the same proportions as the remaining chromosome that were so tested. (A) Crossover points. Only the 29 simple recombinants are shown. Two of the recombinants appeared to have two crossover points within the 4kb interval and an implied crossover point outside the interval. These are represented by three blocks each. All other simple recombinants contribute a single block each. (B) Conversion tract endpoints. Each conversion tract contributes two blocks, except for a single discontinuous conversion tract, which contributes four.

these results show that P-induced male recombination occurs primarily at the site of a mobile P element.

The remaining 14% of the recombinants that occurred outside the 4kb interval can be explained as second-step events following a transposition of CaSpeR to a site elsewhere on 2 R . All but two of the 18 recombinants outside the 4kb interval were in the structurally simple category (see below), although they might have had structural changes at the crossover point outside our analyzed interval. They were approximately equally divided between the proximal and distal directions relative to 50C, with 10 being proximal and eight distal. They were also about equally divided between cn bw" and cn+ bw recombinant types (eight vs. 10).

There were 12 cases in which a single gen-0 cross produced both en bw' and cn' bw recombinants among the gen-1 sons. Only one of each type was kept for analysis, thus accounting for 24 of the 128 recombinant lines. Eight of the pairs included deletions and/or duplications, and are discussed in detail elsewhere (PRES TON et al. 1996). In the other four we found two pairs in which the crossover points were in different RFLP intervals, suggesting that they came from independent events. The remaining two pairs had crossover points

in the MspI-EarI interval, which also contains the CaSpeR insertion site, and is the most frequent site of recombination (Figure 5A). These two cases could be explained either as reciprocal products of a single event, or from two independent recombination events in the MspI-Ead interval.

Structural classification of recombinants: Each of the 128 recombinant chromosomes was classified as either "simple" or rearranged according to a series of tests including PCR, restriction mapping, and DNA sequencing. The simple classification implies that no deletion, duplication, or other rearrangement was detected. Specifically, the following tests were done in various combinations depending on the structure of each recombinant.

(1) PCR fragments 4, 3 and 1 were amplified and in some cases fragment 2 as well (see Figure IS). Classifi- cation as simple required the amplicon to be present and of the normal size for each fragment,

(2) Each RFLP site analyzed as described above must show either the A or C pattern for the recombinant to be categorized as simple. In some cases, both patterns appeared indicating a duplication. For the homozygous lethal chromosomes, where it was necessary to examine


both kinds of heterozygotes as mentioned above, there were some cases in which we observed only A bands in the recombinant/A heterozygote and only C bands in recombinant/C. In those cases we assumed that the fragment from the recombinant chromosome failed to amplify, suggesting a deletion or other rearrangement.

( 3 ) PCR using primer combinations G0/2223 and D0/2231 was done to determine whether each CaSpeR boundary was still present and flanked by the same genomic sequences. The simple classification required that both CaSpeR ends be present, or that both be absent. These two results correspond to the presence or complete loss of CaSpeR.

(4) PCR with primer combination DO/GO was done to determine whether the wild-type CaSpeR-free fragment was present and of the expected size. If the two boundary fragments described in the previous test were missing, then the simple classification required the DO/ GO fragment to be amplified and of the normal size (145 bp).

(5) DNA sequencing of the boundary fragments (G0/2223 and D0/2231) and/or the empty fragment (DO/GO) was carried out for a subset of the recombinants. Only those with the expected A or C sequence were classified as simple.

By these criteria we classified 45 of the chromosomes as simple and 83 as structurally rearranged. Twelve of the simple recombinants were confirmed by sequencing as described in step 5 above, and the rest were inferred from PCR fragment sizes. In the remainder of this report, we will focus on the simple recombinants. The rearranged cases will be described in detail in the accompanying paper (PRESTON et ul. 1996).

We found that 10 of the simple recombinants in the 4-kb interval had both Pends, as indicated by PCR amplification of both boundary fragments and lack of the expected size empty fragment (primers DO/GO). There were two additional cases in which a CaSpeR-internal deletion prevented amplification of one of the boundary fragments. In one of these, the DO/GO combination produced a larger-than-normal fragment. Further tests, especially digestion with NZaIII, showed that both ends of the P element were still present. Therefore, this recombinant was placed in the structurally simple category, and the other was assumed to be similar. The remaining 17 simple recombinants failed to amplify either boundary segment but did yield the wild-type size for the CaSpeR-empty fragment. Thus CaSpeR was cleanly lost from the 50C site of those chromosomes. For those with crossover points outside the CaSpeR RFLP interval, the presence/absence of CaSpeRwas mostly as expected. That is, recombinants in which the MspI-Ead interval carried the C markers also had the CuSpeRelement (Fig- ure 5A). The one exception to this rule can be explained as a subsequent excision of CaSpeR.

Comparison of crossover points and conversion endpoints: The crossover point for each simple recombi-

TABLE 1

Distribution of crossover and conversion points

Interval Crossover Conversion

NlaIII- ThaI 1 (3.3) 6 (3.6) ThaI- TaqI 1 (3.3) 5 (3.0)

MSPI-Ear1 13 (43.3) 70 (41.4) TaqI- MsPI 7 (23.3) 22 (13.0)

Ear1 - gap 4 (14.3) 60 (35.5) gap-Sau3AI 4 (13.3) 6 (3.6)

Data pooled from Figure 5. The number (and percentage in parentheses) of conversion points and conversion tract endpoints in each interval is indicated. These numbers were treated as a 2 X 6 contingency table in the test for independence.

nant is mapped in Figure 5A. Note that many of the crossover points lie within the 4kb region but are not within the MspI-Ead interval that includes the CaSpeR insertion site. The density of crossover points is greatest in the CaSpeRcontaining interval and decreases as one moves away in either direction. Indeed, the distribution is similar to that of the conversion tract endpoints, as shown in Figure 5B.

To test this similarity, we divided the 50C region into six intervals and totalled the number of crossover points and conversion breakpoints in each interval. Table 1 shows the counts. The chromosomes not tested with Ead (Figure 5) were excluded. The 2 X Ngeneraliza- tion of Fisher’s Exact Test for independence yielded P = 0.08. This test was performed by computing the probability and likelihood ratio statistic of all 129,051 tables with the same marginal sums as the observed. We conclude that the crossover points of the simple recombinants follow a distribution similar to that of conversion tract endpoints, with no statistically detectable difference between the two distributions.

DISCUSSION

Male recombination OCCUTS at Felement sites: Is P- induced male recombination the result of direct action of P transposase on the Drosophila genome, or is it a consequence of the transposition process itself and/or subsequent repair events? MCCARRON et aZ. (1994) ar- gue for the former model based on their results showing that recombination can occur in the absence of any mobile P elements provided a transposase source is available. Their experimental design for detecting recombination under these conditions was such that only recombinant progeny survived. Therefore, the frequency of such events could not be measured with accu- racy, but it was clearly several orders of magnitude lower than what is typically seen when even a single mobile P element is present. In addition, studies have shown that when a mobile P element is present, most of the recombination events occur in the same genetic interval


as the element (SVED et al. 1991, 1995; MCCARRON et al. 1994; SVOBODA et al. 1995), which would seem to indicate that the events are the result of transposase action on the mobile Pelement rather than on random genomic sites. An alternative interpretation advocated by CHOVNICK and coworkers (DUTTAROY et al. 1990; MCCARRON et al. 1994) is that the mobile P element merely tends to concentrate the P transposase to the general vicinity where it is more likely to interact directly with genomic sequences. As evidence for this view, they point out that the presence of a mobile P element also elevates recombination rates in neighboring genetic intervals, albeit to a lesser extent.

Use of the RFLP markers surrounding P{CaSpeR] (50C) allowed us to examine this issue with much higher resolution. We find that 86% of the recombinant chromosomes had crossover points within 2 kb on either side of a single mobile P element. This means that the concentration of crossover points in the immediate vicinity of the Pelement is -30,000-fold greater than in the rest of the chromosome arm. Such a high concentration strongly supports the idea that male recombination results from the action of transposase on mobile Pelements rather than on the genome directly.

If one assumes that male recombination occurs only at Pelement sites, the lower frequency crossover events in neighboring regions can then be explained by a two- step process in which the mobile P element transposes to a new location, followed by crossing over at the new site in a future germ cell generation. The phenomenon of “short hopping” observed for some Pelement insertions (TOWER et al. 1993; ZHANC and SPRADLINC 1993) can be invoked to account for any excess of crossover events on the same chromosome arm as the mobile P element. Finally, to explain the rarer-still events in which transposase induces crossing over in a genome lacking any mobile P elements, we postulate that the Drosophila genome contains occasional sequences with sufficient similarity to P element ends to allow transposase attack, at least with reduced affinity.

Conversion tract distributions are the same at white and 50C: A useful side result of the present work is the conversion tract distribution in Figure 3, which can be compared to analogous data generated previously at the whitelocus. Specifically, Figure 3 of the present work is comparable to Figure 7 of GLOOR et al. (1991), Figure 4 of NASSIF et al. (1993) and Figure 6 of NASSIF et al. (1994). From these comparisons, it is clear that the distribution of conversion tracts is almost identical between the white locus and 50C. The average conversion tract length at 50C was 1463 bp, essentially the same as the estimated average of 1379 bp for conversion tracts in white (GLOOR et al. 1991). For both loci, the distribution of conversion tracts was highly symmetrical with most tracts being bidirectional. This contrasts with one study in Saccharomyces cermisiae where strong polarity was observed and where most tracts were unidirectional

(SWEETSER et al. 1994). However, conversion tracts can be bidirectional in yeast when the site of the double- strand break is flanked by homeologous sequences (NELSON et al. 1996).

The mechanism of conversion tract generation is not known, but the data provide a good fit to a simple model in which a double-strand break is expanded by exonuclease activity such that the extent of the expansion determines the conversion tract endpoints. In this model, each base cleavage event occurs with some prok ability x, and the gap expansion process stops after a given base with probability 1 - x. Estimating x by maximum likelihood yields almost the same value whether the data come from 50C or the white locus (0.99864 us. 0.99855). This near identity suggests that gap expansion occurs through a cell-wide system that is independent of genomic position.

P-element-induced gap repair has been proposed for use as a gene replacement technique for target sites in the vicinity of P insertions (GLOOR et al. 1991). The distance between the target site and the P insertion is crucial for this technique, since the fraction of conversion tracts that include a target site n bases distant from the Pinsertion will be x n . Planning of such experiments will be facilitated if one can assume that x is essentially invariant between loci, as the present data suggest.

The mechanism of simple recombination events: Any proposed mechanism for the simple recombinants must explain the retention of CaSpeR in a large fraction of recombinant chromosomes, and the distribution of crossover points that resembles the distribution of conversion tract endpoints (Figure 5). The relationship between conversion tracts and crossing over has been in- vestigated for Drosophila meiosis (CURTIS et al. 1989; CURTIS and BENDER 1991), but the underlying mechanisms are likely to be different for P-induced recombination, as discussed below.

Approximately two-thirds of the male recombination events we analyzed contained structural rearrangements, usually duplications or deletions. The true proportion might be even greater, since some of those with crossover points outside the 50C region might also carry undetected rearrangements. In the accompanying paper (PRESTON et al. 1996) we conclude that duplication- and deletion-bearing recombinants are most easily explained by the hybrid element inm-tion (HEI) model (Svo- BODA et al. 1995; GRAY et al. 1996).

In the HE1 model, recombination comes about via an aberrant transposition event resulting in a recombinant chromosome carrying an intact P element and a flanking duplication or deletion (Figure 6A). Evidence for the HE1 model comes from studies in which each homologue carries only one functional P-element end (Svo- BODA et al. 1995; GRAY et al. 1996). Studies of the structure of transposase-induced recombinant chromosomes from males of this kind provided strong evidence for the HE1 model (GRAY et al. 1996). We apply the model


A B cn+ bwdeletion

c3 bw ........ ........

n HE1 crossover

one of four event yielding

products

c!!: .... -.." Al A2 A3 4

cn bwC deletion

F"..... ........ .............. cn: A1 '4 bw " .... ........ cn+ bwduplication

A1 A2 _ _ " A3 ."., c3 c4 bw F?.: .... -.....!E+ A1 A2 A3 A4 cn.' .... ........ r ............ - - s

cn bw+ duplication e2

c1 c2 F".... 2 % ; A3 A4 bw' ........ - ...........

0 homolog as Repair with

template

FIGURE 6.-Two step model for formation of simple crossover lacking CaSpeR. The WLP markers are represented by C,, Cp,

most common recombinant structures produced by a hybrid element insertion event from CuSpeR elements on the two sister chromatids to the homologue. The steps involved in this kind of HE1 event are described in detail elsewhere (PRESTON et al. 1996). For each structure, a novel junction connects the CuSpeR end with an arbitrary site in the flanking sequence of the A chromosome. The duplication or deletion is indicated by a dashed box. (B) CuSpeR excises from one of the recombinant chromosomes at some time after the HE1 event. This might occur in a subsequent mitotic germ cell division. Gap expansion and repair from the unmodified A chromosome results in a simple recombinant lacking CuSpeR, but with an apparent crossover point offset from the original CuSpeR insertion site. The process is shown starting with a cz+ bw deletion, but would be similar

. . . . or A,, A2, . . . for the C and A chromosomes. The 8-bp insertion site of P(CaSpeRJ(50C) is indicated by b. (A) The four

for the other three structures.

here to the case of a "hybrid element" formed from two fully functional Pelements, one on each sister chromatid (PRESTON et al. 1996).

In addition to the structural rearrangements, some of the 29 simple recombinants with crossover points within the 50C marked interval (Figure 5) can also be explained by the HE1 model. If the hybrid element happens to insert into the homologue at precisely the same point as the original element, the result would appear as a simple crossover in the MspI-EarI interval with retention of the CaSpeR element. From Figure 5, we see that there are nine recombinants in this category: four cn bw+ and five cn+ bw, five of which were confirmed by DNA sequencing (PRESTON et al. 1996). This interpretation requires that nine independent HE1 events, or 6% of the total, occurred at the same target site. Such an extreme hotspot would be unlikely for ordinary P-element transposition. However, insertion site selection for HE1 events is probably different from that of ordinary P-element transposition. Physical con- straints on the movement of the hybrid element are expected since it remains covalently linked to both sister strands of the homologue. Indeed, five potential

cases HE1 insertion into this site were reported in another experiment (PRESTON et al. 1996, see type V in Table 2). Therefore, an extreme excess of insertion events into the original site cannot be ruled out.

A two-step process was suggested by J. SVED and M. TANAKA (personal communication) to account for the simple recombinants in the 50C interval that have lost the CuSpeR element. The first step is an HE1 event, leaving a recombinant chromosome with an intact CaS- peR and a flanking duplication or deletion (Figure 6A). The second step is ordinary transposition of the CaSpeR element followed by double-strand gap repair (Figure 6B). The template for this repair would be the nonre- combinant A chromosome, which lacks CaSpeR Follow- ing repair, the product is a recombinant chromosome in which the CaSpeR element and the flanking duplication or deletion have been lost. In their place is sequence derived from the A chromosome. We observed 16 simple recombinants in this category. Those with crossover points within the MspI-Eurl interval are inter- preted as cases in which the second step occurred by a short conversion tract so as not to replace any of the RFLP markers, yet large enough to eliminate any dupli-


cation or deletion. Those with apparent crossover points outside the MspI-Earl interval are assumed to have longer conversion tracts. The two cases of an apparent double crossovers would, according to this interpretation, be due to discontinuous conversion events as have been reported previously (GLOOR et al. 1991; NASSIF and ENGELS 1993).

There remain four cases of simple recombinants that cannot be explained by the HE1 model, even with multiple steps. These are the chromosomes with crossover points outside the MspI-Ead interval but which retain the CuSpeR element (Figure 5). The occurrence of such chromosomes suggests that not all male recombination comes about via HE1 events. One way to account for these structures is to assume that ordinary transposition of the CaSpeR element followed by gap repair creates a recombinogenic intermediate structure that interacts with the homologue. According to various models of double-strand gap repair (SZOSTAK et al. 1983; FORMOSA and ALBERTS 1986; NASSIF et al. 1994), a likely intermediate is a repaired DNA duplex with single-stranded nicks at each end of the filled-in gap. Such nicks can be recombinogenic, invading another DNA duplex and initiating recombination (MESELSON and RADDINC

1975). If the initial repair event used the sister chromatid as the template, the copy of CuSpeR would be re- stored, but nicks would be left at some distance from the element. Subsequent recombination at these nicked sites would result in the observed structures. This kind of mechanism also offers an alternative way to explain the nine CaSpeR-bearing crossovers within the MspI- Ead interval without requiring a high frequency of insertion into the original CaSpeR site.

To summarize, the HE1 model can explain 25 of the 29 simple recombinants whose crossover points lie within the RFLP region. However, other mechanisms must be invoked to account for the remaining cases. The correspondence between the conversion tract distribution and that of the simple crossover points (Fig- ure 5) is expected from the two-step mechanism in Figure 6, since the apparent crossover point corresponds to an end of the conversion tract. In addition, the model of recombinogenic nicks at the boundaries of conversion tracts would also account for the similarity in distributions because conversion tract endpoints determine the recombination point.

Helpful comments were provided by DENA JOHNSON-SCHLITZ, CAR. LOS FLOES, KOHJI KCJSANO, JAMES CROW and JOHN(; LIM. We especially thank JOHN SVED and MARK TANAKA for the suggestion that HE1 events might he involved. This is paper 3465 from the University of Wisconsin Laboratory of Genetics (http://www.wisc.edu/genetics) supported by National Institutes of Health grant GM-30948.

LITERATURE CITED

BARNES, W. M., 1994 PCR amplification of up to 35-kb DNA with high fidelity and high yield from lambda bacteriophage templates. Proc. Natl. Acad. Sci. USA 91: 2216-2220.

CRAYMER, L., 1980 New mutants report. Dros. Inform. Sew. 5 5 197- 200.

CURTIS, D., and W. BENDER, 1991 Gene conversion in Drosophila and the effects of the meiotic mutants ~ ' - 9 a n d mei-218. Genetics

CURTIS, D., S . H. CLARK, A. CHOVNICK and W. BENDER, 1989 Molecu- lar analysis of recombination events in Drosophila. Genetics 122

DUTTAROY, A,, M. MCCARRON, K. SITARAMAN, G. DOUGHTY and A. CHOVNICK, 1990 The relationship between P elements and male recombination in Drosophila melanogaster. Genetics 124: 317-329.

ENGEIS, W. R., 1979 Germline aberrations associated with a case of hybrid dysgenesis in Drosophila melanogaster males. Genet. Res. Camb. 3 3 137-146.

ENGELS, W. R., 1989 P elements in Drosophila, pp. 437-484 in M e bile DNA, edited by D. BERG and M. HOW. American Society of Microbiology, Washington, DC.

ENGELS, W. R., 1996 P elements in Drosophila, pp. 103-123 in Transposable Elemats, edited by H. SAEDLER and A. GIERI.. Springer-Verlag, Berlin. (http://www.wisc.edu/genetics/CATG/ engels/Pelements/)

ENGELS, W. R., D. M. JOHNSON-SCHLITZ, W. B. EGGLESTON andJ. SVED, 1990 High-frequency P element loss in Drosophila is homolog- dependent. Cell 62: 515-525.

ENGEIS, W. R., C. R. PRESTON and D. M. JOHNSON-SCHLITZ, 1994 Long-range cis preference in DNA homology search extending over the length ofa Drosophila chromosome. Science 263: 1623- 1625.

RYBME, 1996 The Drosophila Genetic Database. (Available from the flybase.bio.indiana.edu network server and Gopher site.). Nucleic Acids Res. 2 4 53-56.

FORMOSA, T., and B. M. ALBERTS, 1986 DNA synthesis dependent on genetic recombination: characterization of a reaction cata- lyzed by purified bacteriophage T4 proteins. Cell. 47: 793-806.

GLOOR, G. B., and W. R. ENGEIS, 1992 Single fly DNA preps for PCR. Dros. Inform. Ser. 71: 148-149.

GLOOR, G. B., N. A. NMSIF, D. M. JOHNSON-SCHI.ITZ, C. R. PRESTON and W. R. ENGELS, 1991 Targeted gene replacement in Dro- sophila via P element-induced gap repair. Science 253 1110- 1117.

GRAY, Y., M. TANAKA and J. A. SVED, 1996 P-element-induced recombination in Drosophila melanogaster: hybrid element insertion. Ge- netics 144: 1601-1610.

HIRAIZUMI, Y., 1971 Spontaneous recombination in Drosophila melanogaster males. Proc. Natl. Acad. Sci. USA 6 8 268-270.

HIRAIZUMI, Y., B. SIATKO, C. LANGLEY and A. NILL, 1973 Recom- bination in Drosophila melanogaster male. Genetics 7 3 439-444.

ISACKSON, D. R., T. K. JOHNSON and R. E. DENELL, 1981 Hybrid dysgenesis in Drosophila: the mechanism of T-007-induced male recombination. Mol. Gen. Genet. 184 539-543.

JOHNSON-SCHI.ITZ, D. M., and W. R. ENGELS, 1993 P element-induced interallelic gene conversion of insertions and deletions in Drosophila. Mol. Cell. Biol. 13: 7006-7018.

KIDWELL, M. G., and J. F. KIDWELL, 1976 Selection for male recombination in Drosophila melanogaster. Genetics 84: 333-351.

KIDWELL, M. G., J. F. KIDWEIL and J. A. SVED, 1977 Hybrid dysgenesis in Drosophila melanogaster: a syndrome of aberrant traits including mutation, sterility, and male recombination. Genetics 86:

LEE, L. G., C. R. CONNELL, S . L. Woo, R. D. CHENG, B. F. MCARDLE et al., 1992 DNA sequencing with dye-labeled terminators and T7 DNA polymerase: effect of dyes and dNTPs on incorporation of dye-terminators and probability analysis of termination fragments. Nucleic Acids Res. 20: 2471-2483.

LINDSLEY, D. L., and G. G. ZIMM, 1992 The Genome ofDrosophila melanogaster. Academic Press, San Diego.

MCCARRON, M. Y., A. D U ~ A R O Y , G. A. DOUGHTY and A. CHOVNICK, 1989 P element transposase induces male recombination in Drosophila melanogaster. Genet. Res. 54: 137-141.

MCCARRON, M., A. DUTTAROY, G. DOUGHTY and A. CHOVNICK, 1994 Drosophila P element transposase induces male recombination additively and without a requirement for P element excision or insertion. Genetics 136: 1013-1023.

MESELSON, M. S., and C. M. RADDING, 1975 A general model for genetic recombination. Proc. Nat. Acad. Sci. USA 78: 358-361.

127: 739-746.

653-661.

813-833.


NASSIF, N. A., and W. R. ENGELS, 1993 DNA homology requirements for mitotic gap repair in Drosophila. Proc. Nat. Acad. Sci. USA

NASSIF, N. A., J. PENNLY, S . PAL, W. R ENGEIS and G. B. GLOOR, 1994 Efficient copying of nonhomologous sequences from ectopic sites via P element-induced gap repair. Mol. Cell. Biol. 14: 1613-1625.

NELSON, H. H., D. B. SWEETSER and J. A. NICKOLOFF, 1996 Effects of terminal nonhomology and homeology on double-strand-break-

2951-2957. induced gene conversion tract directionality. Mol. Cell. Biol. 16:

PIRROTTA, V., 1988 Vectors for P-mediated transformation in Dro- sophila, pp. 437-456 in Vectors, A Suruey of Molecular Cloning Vec- to rs and Their Uses, edited by R. L. RODRIGUEZ and D. T. DEN- HARUT. Butterworths, London.

PRESTON, C. R., J. A. SVED and W. R. ENGELS, 1996 Flanking duplications and deletions associated with P-induced male recombination in Drosophila. Genetics 144: 1623-1638.

ROBERTSON, H. M., C. R. PRESTON, R. W. PHILLIS, D. JOHNSON- SCHLITZ, W. K BENZ et al., 1988 A stable genomic source of P element transposase in Drosophila mehnogaster, Genetics 118: 461-470.

ROSENTHAI., A,, and D. S . CHARNOCK~ONES, 1992 New protocols for DNA sequencing with dye terminators. DNA Seq. 3: 61-64.

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Clon- ing: A Laboratmy Manual, Ed. 2. Cold Spring Harbor Press, Cold Spring Harbor, NY.

SINCIAIR, D. A. R., and T. A. GRIGLIATTI, 1985 Investigation of the nature of P-induced male recombination in Drosophila melanogaster. Genetics 110: 257-279.

SVED, J. A., 1978 Male recombination in dysgenic hybrids of Drosoph- ila melanogaster; chromosome breakage or mitotic crossing-over? Aust. J. Biol. Sci. 31: 303-309.

SVED, J. A,, W. B. EGGIXSTON and W. R. ENGEIS, 1990 Germline and somatic recombination induced by in vitro modified P elements in Drosuphila mehnogaster. Genetics 124: 331-337.

SVED, J. A,, L. M. B L A C ~ , A. S . GILCHRIST and W. R. ENGELS, 1991 High levels of recombination induced by homologous P elements in Drosophila melanogaster. Mol. Gen. Genet. 225: 443-447.

90: 1262-1266.

SVED, J. A., L. M. BLACKMAN, Y. SVOBODA and R. COLLESS, 1995 Male recombination with single and homologous P elements in Dro- sophila melanogaster. Mol. Gen. Genet. 246: 381-386.

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

SWEETSER, D. B., H. HOUGH, J. F. WHELDEN, M. ~ U T . K I . E and J. A. NICKOLOFF, 1994 Fine-resolution mapping of' spontaneous and double-strand break-induced gene conversion tracts in Saccharo- myces cerevisiae reveals reversible mitotic conversion polarity. Mol. Cell. Biol. 14: 3863-3875.

S z o s r u , J. W., T. L. ORR-WEAVER, R. J. ROTHSTEIN and F. W. STAHI., 1983 The double-strand-break repair model for recombination. Cell 33: 25-35.

TowER,J., G. H. KARPEN, N. C M c a n d A. C . SPRADLING, 1993 Prefer- ential transposition of Drosophila P element5 to nearby chromo- somal sites. Genetics. 133: 347-359.

TSUBOTA, S., and P. SCHEDI., 1986 Hybrid dysgenesis-induced re- vertants of insertions at the 5 end of the rudimentaly gene in Drosophila melanogaster: transposoninduced control mutations. Genetics 114: 165-182.

WOODRUFF, R. C., and J. N. J. THOMPSON, 1977 An analysis of spontaneous recombination in Drosophila melanogastermales: isolation and characterization of male recombination lines. Heredity 38:

WOODRI~FF, R. C., B. SIATKO and J. N. THOMPSON, 1978 Lack of an interchromosomal effect associated with spontaneous recombination in males of Drosophila melanogaster. Ohio J. Sci. 78: 310- 317.

YAVNOPOL'I.~~, G., 1979 Ability of the male recombination factor 31.1 MRF to be transposed to another chromosome in Drosophila melanagaster. Mol. Gen. Genet. 176: 247-253.

ZITANC;, P., and A. C. SPRADLING, 1993 Efficient and dispersed local P element transposition from Drosophila females. Genetics 133:

291-307.

361-373.

Communicating editor: R. S. HAWEY

p-element-induced male recombination and gene conversion in

Documents