Intrinsic Intermolecular DNA Ligation Activity of Eukaryotic Topoisomerase I1 POTENTIAL ROLES IN RECOMBINATION*

Drosophila melanogaster topoisomerase I1 is capable of joining 4x174 (+) strand DNA that it has cleaved to duplex oligonucleotide acceptor molecules by an intermolecular ligation reaction (Gale, K. C. and Osheroff, N. (1990) Biochemistry 29, 9538-9545). In order to investigate potential mechanisms for topoisomerase II- mediated DNA recombination, this intrinsic enzyme activity was further characterized. Intermolecular DNA ligation proceeded in a time-dependent fashion and was concentration-dependent with respect to oligonucleotide. The covalent linkage between 4x174 (+) strand DNA and acceptor molecules was confirmed by Southern analysis and alkaline gel electrophoresis. To- poisomerase 11-mediated intermolecular DNA ligation required the oligonucleotide to contain a 3’-OH terminus. Moreover, the reaction was dependent on the presence of a divalent cation, was inhibited by salt, and was not affected by the presence of ATP. The enzyme was capable of ligating 4x174 (+) strand DNA to double-stranded oligonucleotides that contained 5‘-overhang, 3’-overhang, or blunt ends. Single- stranded, nicked, or gapped oligonucleotides also could be used as acceptor molecules. These results demon-strate that the type I1 enzyme has an intrinsic ability to mediate illegitimate DNA recombination in vitro and suggests possible roles for topoisomerase I1 in nu- cleic acid recombination in vivo. The topological

The topological state of nucleic acids is regulated in uiuo by enzymes known as type I and type I1 topoisomerases (1-4). These enzymes influence virtually every aspect of DNA metabolism in eukaryotic cells (3). In addition to roles in chromosome segregation (5-lo), DNA replication (11)(12)(13), and transcription (14,15), several lines of evidence indicate that topoisomerases are involved in the process of DNA recombination (3,4,10,(16)(17)(18)(19)(20)(21)(22)(23)(24). To this point, expression of vaccinia topoisomerase I in Escherichia coli cells (25) or introduction of exogenous mammalian topoisomerase I1 into mouse cells (26) greatly increases the frequency of recombination in these systems. Topoisomerases have been ascribed a number of different roles in the process of DNA recombination. Both topoisomerases I and I1 are important for the regulation of recombination within the rDNA cluster in yeast (18,20,23).
Furthermore, topoisomerase I1 is required for the resolution of chromosomes that have undergone meiotic recombination (10). In addition, evidence suggests that topoisomerases I and I1 can mediate the DNA cleavage/ligation event that leads to some forms of illegitimate recombination in eukaryotes. For example, sites of i n vivo recombination are tightly associated with topoisomerase I or I1 recognition/cleavage sequences (16,21). Moreover, topoisomerase Ior 11-targeted antineoplastic drugs (that stabilize covalent enzyme-cleaved DNA complexes (27, 28)) promote chromosomal translocations and mutations, as well as sister chromatid exchange in treated cells and human patients (29-37). Finally, both enzymes have been shown to carry out illegitimate recombination (16,17,19,24) or intermolecular DNA ligation (38-40) i n uitro.
Proposed models for topoisomerase I-mediated illegitimate recombination all rely on the enzyme's ability to introduce transient single-stranded breaks in the DNA backbone (16,22,24). In these models, a single molecule of the type I enzyme joins cleaved nucleic acids to the free termini of separate DNA acceptor molecules by an intermolecular ligation event. In contrast, the only models for topoisomerase I1 that have been proposed require two separate enzyme homodimers to introduce double-stranded breaks in DNA. In these latter models, the recombinagenic event depends on the exchange of subunits between the two enzyme molecules, followed by ligation of the cleaved nucleic acids (17,19,22,27).
However, since the type I1 enzyme can introduce nicks into the double helix (41-44), and single-stranded breaks are a kinetic intermediate in the double-stranded DNA cleavage/religation reaction of topoisomerase I1 (4, 42, 45), alternative recombination pathways that do not require two separate enzyme homodimers or subunit exchange also should be considered.
Recent evidence indicates that topoisomerase I1 can join single-stranded DNA (that results from enzyme-mediated cleavage of 6x174 (+) strand molecules) to an oligonucleotide acceptor (40). The present study characterized this intermolecular ligation activity in order to investigate the enzyme's potential to recombine DNA molecules following the introduction of single-stranded breaks in the nucleic acid backbone. Results indicate that topoisomerase I1 can mediate illegitimate recombination events, a t least in uitro, by a mechanism that does not rely on double-stranded DNA cleavage or subunit exchange.

EXPERIMENTAL PROCEDURES
Materials DNA topoisomerase I1 was purified from the nuclei of Drosophila melanogaster Kc tissue culture cells or 6-to 12-h-old embryos, as described by Shelton et al. (46). All preparations employed were 295% homogeneous. Protein concentrations were determined by Bradford analysis using bovine serum albumin as a standard (46). Circular bacteriophage 4x174 (+) strand DNA and terminal deoxynucleotidyltransferase were from GIBCO-BRL; [a-"PIdATP, dTTP, and [Y-~'PP]ATP (-3000 Ci/ mmol) were from Amersham Corp.; sodium dodecyl sulfate and proteinase K were from E. Merck Biochemicals; polyethylene glycol 20,000 was from BDH Chemicals; Tris was from Sigma; ATP, ddATP, Sephadex G-50 (medium), and bacteriophage T4 polynucleotide kinase were from Pharmacia LKB Biotechnology, Inc.; Taq DNA polymerase, dATP, dCTP, dGTP, and dTTP were from Perkin-Elmer Cetus Instruments; restriction endonucleases EcoRI, PstI, and PuuII were from New England Biolabs; and oligonucleotides were synthesized on an Applied Biosystems DNA synthesizer. All other chemicals were analytical reagent grade.

Preparation of Double-stranded Oligonucleotide Acceptor Molecules for Intermolecular D N A Ligation Reactions
Blunt-ended 56-mer for Standard Assays-A single-stranded 56base oligonucleotide with the sequence shown in Fig. 1 (top strand) and its complementary oligonucleotide were radioactively labeled on their 5"termini in separate reaction mixtures. Reaction mixtures contained 27.5 pmol (0.5 pg) of oligonucleotide (all DNA concentrations are reported as molecules rather than total nucleotides), 15 units of polynucleotide kinase, and 37 pmol of [y-32P]ATP (-3000 Ci/mmol) in a total of 20 p1 of kinase buffer supplied by Pharmacia LKB Biotechnology Inc. Following a 2-h incubation at 37 "C, reaction mixtures were diluted to 200 p1 by the addition of 1 X T E (10 mM Tris-HC1, pH 7.9, and 1 mM EDTA) and filtered through a column of Sephadex G-50 (medium). Phosphorylated oligonucleotides were precipitated with ethanol and dried under partial vacuum as described by Sambrook et al. (47), followed by resuspension in 20 p1 of 0.5 X TE. After this procedure, oligonucleotides contained -2 X lo6 cpm/ pmol. The complementary phosphorylated oligonucleotides were annealed by mixing equimolar amounts, heating to 70 "C for 10 min, and cooling to 25 "C over a period of 2 h. Samples were diluted with nonlabeled annealed oligonucleotides to a final specific activity of -2 X lo5 cpm/pmol blunt-ended double-stranded oligonucleotide prior to intermolecular ligation reactions.
Double-stranded Oligonucleotide Lacking a 3'-OH--Radioactively labeled double-stranded 56-mer (prepared as described above) (2.8 pmol) was incubated with terminal deoxynucleotidyltransferase (30 units) and 1 mM ddATP in a total of 17 pl of 100 mM potassium cacodylate, pH 7.2, 2 mM CoC12, and 0.2 mM dithiothreitol for 2 h at 37 "C. The oligonucleotide was phenol-extracted, ethanol-precipitated, resuspended in 10 pl of 0.5 X TE, and diluted with nonradioactive 56-mer that lacked a 3"OH to a final specific activity of -2 X 10" cpm/pmol. Double-stranded Oligonucleotide Lacking a 5'-Phosphate--The single-stranded 56-base oligonucleotide shown in Fig. 1 (bottom strand) (55 pmol, and an equimolar amount of an oligonucleotide corresponding to the first 20 bases of the 56-mer (top strand) were mixed with 2.5 units of Taq DNA polymerase, 312 pmol of each dNTP, and 7 pmol of each [a-"'PIdNTP (-3000 Ci/mmol) in a total of 100 g1 of Taq DNA polymerase buffer (Perkin-Elmer Cetus Instruments). Two drops of paraffin oil were placed on top of the sample, and the 20mer primer was extended by incubating the mixture in an Ericomp Twin Block System thermocycler for 10 cycles (2 min at 95 "C, 1.5 Oligonucleotide acceptor molecule for intermolecular DNA ligation reactions. Unless otherwise stated, the bluntended 56-mer shown was utilized as the acceptor molecule for assays. Digestion with restriction endonuclease EcoRI, PstI, or PuuII (sites are denoted by the arrows) generated 45-mers with 5'-overhangs, 3'overhangs, or blunt ends, respectively, for use in some assays. min at 50 "C, and 2 min at 72 "C), followed by a final cycle of 10 min at 72 "C. (Cycles beyond the first round were employed to ensure complete extension of the 20-mer.) Reaction products were phenol/ chloroform-extracted, filtered through Sephadex G-50 (medium), ethanol-precipitated, and resuspended in 20 pI of 0.5 x TE. The resulting radioactive double-stranded 56-mer was either phosphorylated by polynucleotide kinase with nonradioactive ATP as described above (as a control) or subjected to mock phosphorylation in the absence of polynucleotide kinase. Oligonucleotides were recovered by gel filtration and ethanol precipitation as described above, resuspended in 10 p1 of 0.5 X TE, and further diluted in 0.5 X T E to a final specific activity of -2 X 10" cpm/pmol.

E c o l R1
Oligonucleotides with Different End Configurations-Nonradioactive double-stranded 56-mer (220 pmol) was digested with EcoRI (300 units), PstI (300 units), or PuuII (150 units) in a final volume of 200 pl of the appropriate restriction enzyme buffer supplied by New England Biolabs. (See Fig. 1 for the location of restriction endonuclease cleavage sites.) Mixtures were incubated for 2 h at 37 "C, followed by phenol extraction, ethanol precipitation, and resuspension in 15 pl of 0.5 x TE. Twenty microliters of loading buffer (85% formamide, 50 mM Tris borate, pH 8.3, 0.05% bromphenol blue, 0.05% xylene cyanol FF) was added, and samples were subjected to electrophoresis in a 7 M urea, 8% polyacrylamide gel in 100 mM Tris borate, pH 8.3, and 2 mM EDTA. The 45-base oligonucleotides that were generated by the restriction digests were excised from the gel, extracted overnight in 500 mM NH,OAc, 10 mM MgOAcn, and 1 mM EDTA, and purified by column chromatography on SepPak C-18 (Millipore) columns as recommended by the manufacturer. Samples were resuspended in 20 p1 of 0.5 X TE, 3' P end-labeled, and recovered by ethanol precipitation as described above. Samples were diluted with nonlabeled annealed oligonucleotides (digested with the appropriate restriction endonuclease) to a final specific activity of -2 X lo5 cpm/pmol.

Intermolecular D N A Ligation by Topoisomerase II
Assays were carried out by a modification of the procedure of Gale and Osheroff (40). Unless otherwise noted, 150 nM topoisomerase I1 and 5 nM circular 6x174 (+) strand DNA were incubated at 30 "C for 5 min in 18 p1 of 10 mM Tris-HCI, pH 7.9, 50 mM KC1, 50 mM NaC1, 0.1 mM EDTA, 2.5% glycerol (w/v), and 7.5 mM MgC1,. The ["P]phosphate end-laheled double-stranded 56-mer (25 nM) and polyethylene glycol 20,000 (5%, w/v) were added to the solution to a final volume of 24 pl. DNA ligation reaction mixtures were further incubated for 60 min at 30 "C. Assays were terminated by the addition of 1 pl of 250 mM EDTA, followed by 2 pl of 10% sodium dodecyl sulfate. Proteinase K (2 gl of a 1 mg/ml solution) was added, and topoisomerase I1 was digested at 37 "C for 45 min. Loading buffer (2 p1 of 60% sucrose, 0.05% bromphenol blue, 0.05% xylene cyanol FF, and 10 mM Tris-HCI, pH 7.9) was added, and products were subjected to electrophoresis at 10 V/cm in 1.2% agarose (MCB) gels in 100 mM Tris borate, pH 8.3, and 2 mM EDTA. Following electrophoresis, gels were stained in an aqueous solution of ethidium bromide (0.8 pg/ml). DNA bands were visualized by transillumination with ultraviolet light (300 nm) and photographed through Kodak 24A and 12 filters with Polaroid type 665 positive/negative film. Under the conditions employed, the intensity of the negative was proportional to the amount of DNA present. Levels of DNA cleavage were monitored by quantitating the percent of circular form 6x174 (+) strand. The agarose gel was dried and the incorporated radioactivity was visualized by autoradiography with Kodak XAR film and a Du Pont Lightning Plus screen. Levels of intermolecular DNA ligation were monitored by quantitating the ["'Plphosphate incorporated into the unit length linear 6x174 molecules by scanning autoradiograms with an E-C Apparatus model EC910 scanning densitometer in conjunction with Hoefer GS-370 Data System Software. Levels of intermolecular DNA ligation were normalized relative to reactions that were incubated for 60 min and contained 5 nM 6x174 DNA and 25 nM oligonucleotide.

Southern Analysis of Intermolecular D N A Ligation Products
Following intermolecular DNA ligation using nonlabeled doublestranded 56-mers, samples were subjected to electrophoresis, and the DNA bands were visualized as described above. The DNA was transferred from the agarose gel to a Genescreen Plus hybridization transfer membrane (New England Nuclear Research products) and analyzed by Southern blot hybridization according to the procedures described in Sambrook et al. (47). A radioactively labeled 20-base oligonucleotide with the sequence AGAGC/TTCTC/GAGCT/ of Topoisomerase II (;<'GCA (complementary to residues 155-174 of @X174 (+) strand DNA) was used to probe for @X174 molecules, and each strand of the M-mer was used to probe for its complementary strand.

Separation of Intermolecular D N A Ligation Reaction Products by Alkaline Gel Electrophoresis
Blunt-ended double-stranded 56-mers in which one or the other strand was ["'P]phosphate end-labeled were prepared as described above and employed as acceptor molecules in intermolecular ligation reactions. DNA products were ethanol-precipitated, resuspended in 20 pl of 100 mM NaOH and 1 mM EDTA, and subjected to electrophoresis in an alkaline gel (1.2% agarose gel formed in 100 mM Tris borate, pH 8.3, and 2 mM EDTA and soaked overnight in alkaline running buffer, 30 mM NaOH, and 1 mM EDTA). The agarose gel was dried, and the radioactivity was visualized by autoradiography as described above.

Determination of Steady-state Levels of Unit Length Linear @X174 U N A Cleavage Product Generated during Intermolecular Ligation Reaction?
Prior to assays, the linear contaminant (-15% of the total DNA) present in preparations of 6x174 (+) strand DNA substrate (6 pmol 01 total DNA) was radioactively labeled as described above on its 5' terminus with 15 units of polynucleotide kinase and 17 pmol of [y-'"PIATP (-3000 Ci/mmol) in a total of 56 pl of kinase buffer. Following ethanol precipitation, samples were resuspended in 20 p1 of 0.5 X TE and diluted with nonlabeled @X174 (+) strand DNA to a final specific activity of -3.5 X 10' cpm/pmol. Intermolecular DNA ligation assays were carried out as described above using nonlabeled double-stranded 56-mer and analyzed by agarose gel electrophoresis. The percent of DNA cleavage and the total amount of unit length linear 6x174 DNA present were determined by scanning densitometry. The amount of unit length linear contaminant in each sample was determined by autoradiography. Levels of unit length linear cleavage product generated by topoisomerase 11-mediated DNA cleavage during intermolecular ligation were calculated by subtracting the amount of unit length linear contaminant from the total amount of unit length linear @X174 molecules present a t each time point.

Intermolecular DNA Ligation Activity of Topoisomerase II-
Central to topoisomerase 11-mediated nucleic acid recombination is the enzyme's ability to carry out an intermolecular DNA ligation event. Recently, a novel assay was developed to monitor this enzyme activity (40). Although it was demonstrated that the intermolecular DNA ligation reaction is intrinsic to topoisomerase 11, the properties of this potentially important activity were not determined.
In order to investigate potential roles for topoisomerase I1 in the recombination process, the system of Gale and Osheroff (40) was employed to characterize the enzyme's intermolecular DNA ligation reaction. In this assay, topoisomerase I1 is incubated with circular bacteriophage 4x174 (+) strand molecules for 5 min to generate covalent enzyme-cleaved linear DNA intermediates. Following this initial incubation, a &fold molar excess (with respect to 4x174 DNA) of a radioactively labeled double-stranded oligonucleotide acceptor molecule is added to the reaction mixture. An autoradiogram showing a representative time course for the ligation of cleaved 4x174 DNA to a 56-mer acceptor oligonucleotide by topoisomerase I1 can be seen in Fig. 2. Intermolecular ligation results in a shift in the electrophoretic mobility of labeled oligonucleotide to the position of linear 4x174 molecules. As expected for this enzyme-mediated activity, no radioactivity was observed at the position of circular uncleaved bacteriophage DNA.
Intermolecular DNA ligation proceeded in a time-dependent fashion. At longer reaction times, some of the radioactive label was incorporated into smaller than unit length linear 4x174 (+) strand fragments. These fragments result from multiple topoisomerase I1 cleavage events on a single bacteriophage substrate (40). For the purpose of this study, levels of intermolecular DNA ligation were monitored by quantitating the label incorporated into unit length linear 4x174 molecules. To normalize data between independent assays, the amount of label incorporated under standard conditions (60 min incubation, 5 nM 4x174 (+) strand DNA, 25 nM double-stranded 56-mer oligonucleotide), as quantitated by scanning densitometry of autoradiograms, was arbitrarily assigned a value of 1.0. As determined by liquid scintillation counting of excised DNA bands (48), 2.0-2.5% of the cleaved unit length linear bacteriophage DNA was ligated to the labeled 56-mer acceptor molecule (see Fig. 1) in standard assays. Levels that exceeded 4.2% were observed under some conditions. These levels of intermolecular ligation are similar to those obtained when a 42-mer oligonucleotide acceptor was employed previously (40).
The efficiency of topoisomerase 11-mediated intermolecular ligation observed when 4x174 (+) strand DNA was used as the initial cleavage substrate is considerably higher than that obtained using double-stranded nucleic acid as substrates. For example, when double-stranded bacteriophage X was employed, levels of intermolecular ligation (determined by a genetic screen of reaction products) ranged from 0.01 to 0.15% (17, 19). In the present system, no ligation was observed (either in the absence or presence of ATP) when doublestranded pBR322 plasmid molecules were utilized as cleavage substrates (not shown). The relatively high efficiency observed with 6x174 (+) strand molecules most likely reflects the fact that the single-stranded 3'-OH DNA terminus generated upon cleavage can dissociate from the active site of topoisomerase I1 (40). This dissociation disrupts the enzyme's normal intramolecular DNA cleavage/religation equilibrium (established with double-stranded substrates) and allows a separate 3'-OH terminus to invade the topoisomerase 11-DNA cleavage complex.
The bimolecular nature of intermolecular ligation allows the affinity of topoisomerase I1 for its acceptor oligonucleotide t o be determined. This was accomplished by analyzing the concentration dependence of the 56-mer in ligation reactions.
As seen in Fig. 3, levels of intermolecular ligation increased with the oligonucleotide concentration. The apparent K,,, of topoisomerase I1 for the 56-mer was -20 nM, as assessed by Eadie-Hofstee (Fig. 3, inset) or Lineweaver-Burk (not shown) transformation (49) of the data. This value is consistent with previously derived affinity constants determined for binding interactions of the enzyme with double-stranded DNA (50).

Covalent Linkage of 4x1 74 DNA to the Acceptor Oligonucleotide Following Topoisomerase 11-mediated Intermolecular Ligation-Two independent techniques
were employed to confirm the covalent linkage between 4x174 DNA and the acceptor 56-mer following intermolecular ligation. First, reaction products from assays that employed nonradioactive oligonucleotide were examined by Southern hybridization analysis (Fig. 4, right) using DNA probes specific for either the 4x174 molecule (lanes 1-5) or the top strand (see Fig. 1) of the 56-mer (lanes [6][7][8][9][10]. In all cases, the oligonucleotide comigrated with 4x174 DNA a t positions where ligation products were observed (i.e. unit-length and sub-unit-length linear bacteriophage DNA). No oligonucleotide comigrated with uncleaved circular 4x174 molecules. Identical results were obtained when blots were examined with a probe specific for the bottom strand of the 56-mer (not shown).
Second, intermolecular DNA ligation products were analyzed by alkaline gel electrophoresis (Fig. 4, left). For these experiments, either the top strand (not shown) or the bottom strand (see Fig. 1) of the 56-mer (lanes [1][2][3] was radioactively labeled. As found with the Southern hybridization analysis, the oligonucleotide comigrated with linear, but not with uncleaved circular bacteriophage DNA. Identical results were obtained when either strand of the oligonucleotide acceptor molecule was labeled. These data provide strong evidence that topoisomerase 11-mediated intermolecular ligation results in the covalent joining of cleaved 4x174 molecules to the 56mer.  Procedures." An autoradiogram of an alkaline gel is shown a t the left. The 56-mer acceptor molecule employed for this study was radioactively labeled only on its bottom strand (see Fig. 1). Lanes 1-3, intermolecular ligation assays were carried out for 0, 30, or 60 min, respectively. An autoradiogram of a Southern blot is shown a t the right. The 56-mer acceptor molecule employed for this study was not radioactively labeled. Following the transfer of reaction products to Genescreen Plus, blots were probed with radioactive oligonucleotide complementary to either @X174 (+) strand DNA (lanes [1][2][3][4][5] or the top strand of the 56-mer (lanes 6-10). Lanes 1 and 6, assays carried out in the absence of topoisomerase 11; lanes 2-5 and 7-10, intermolecular ligation was carried out for 0, 30, 60, or 90 min, respectively. The positions of circular, linear, and multiply cleaved 6x174 (+) strand DNA, as well as the 56-mer, are as shown in Fig. 2. The abbreviations are as in Fig. 2. -I / I

Time (min)
FIG. 5. Requirement for the 3'-OH termini of the oligonucleotide acceptor for topoisomerase 11-mediated intermolecular DNA ligation. Time courses were carried out as described under "Experimental Procedures." Levels of intermolecular DNA ligation were set arbitrarily to 1.0 at the 60-min time point in assays that contained the control 56-mer acceptor molecule. Open circles, assays contained the control 56-mer; closed circles, assays contained a 56mer that lacked its 3"OH termini due to treatment with terminal deoxynucleotidyltransferase and ddATP.
for the 3'-OH termini of the 56-mer, a labeled oligonucleotide that lacked its terminal hydroxyl groups was generated. This was accomplished by incubating the 56-mer with terminal deoxynucleotidyltransferase and ddATP. When the dideoxy oligonucleotide was employed as the acceptor molecule in assays, levels of intermolecular ligation decreased by more than 85% (Fig. 5). The low level of activity remainingprobably reflects the incomplete incorporation of ddAMP into the oligonucleotide. However, to ensure that intermolecular ligation was always between the 5"phosphate terminus of cleaved 4x174 DNA and the 3'-OH termini of the 56-mer, labeled oligonucleotide acceptor molecules that lacked a 5"phosphate were generated as described under "Experimental Procedures." In two independent experiments, levels of intermolec-ular DNA ligation to the 5'-OH-containing oligonucleotide were a t least as high as those obtained in the presence of the 5'-phosphate-containing acceptor (not shown). Therefore, as expected, topoisomerase I1 covalently joins cleaved 4x174 DNA to the 3"OH termini of the oligonucleotide acceptor.
Due to interactions between topoisomerase I1 and the agarose gel matrix, the electrophoretic mobility of DNA to which it is covalently attached is severely retarded (42,54). Normal DNA electrophoretic patterns are reestablished following digestion of the bound enzyme with proteinase K (42,54). In contrast to cleaved (but unligated) 4x174 molecules (40), the mobility of ligation products did not change when the proteinase K digestion step was omitted (not shown). This indicates that the covalent phosphotyrosine bond established between topoisomerase I1 and the 5"terminus of cleaved 4x174 DNA is displaced by the 3'-OH of the acceptor oligonucleotide.
Effects of Salt, Divalent Cation, and ATP on Intermolecular DNA Ligation-Since salt, divalent cation, and ATP all affect the DNA cleavage/religation equilibrium of topoisomerase I1 (50-52, 54, 55), their effects on the enzyme's intermolecular DNA ligation reaction were characterized. However, if conclusions concerning the influence of these compounds on intermolecular ligation are to be valid, it is first necessary to control for their effects on DNA cleavage. This was accomplished by analyzing levels of unit length linear 4x174 DNA cleavage intermediates produced over the course of a ligation assay.
Unit length linear 4x174 DNA is generated by topoisomerase 11-mediated cleavage of circular substrates and is depleted when additional enzyme molecules convert unit length DNA to sub-unit length fragments (40). Since both the generation and the depletion of unit length cleavage intermediates are catalyzed by the same enzyme-mediated reaction, once a steady state is attained, levels of unit length cleavage intermediates available for ligation should remain constant. Therefore, the effects of compounds (even those that block the enzyme's cleavage reaction) on intermolecular DNA ligation can be assessed as long as they are added to assay mixtures during this steady-state window.
As seen in Fig. 6 teriophage DNA substrate becomes too low to support this steady state. Thus, all of the effectors discussed below were added to assays after 10 min of intermolecular DNA ligation had taken place.
Although noncovalent topoisomerase 11-DNA interactions are disrupted by increased salt concentrations (>95% dissociation at 0.5 M NaCI) (50), the enzyme's intramolecular double-stranded DNA religation activity is not sensitive to salt at concentrations up to 1 M (42, 52,54). As expected for a reaction that requires both topoisomerase 11-DNA binding and ligation, the intermolecular joining of cleaved 4x174 (+) strand DNA to the oligonucleotide acceptor was blocked following the addition of 0.5 M NaCl to assays (Fig. 7).
Although topoisomerase I1 does not require a divalent cation for noncovalent interactions with DNA (50, 56, 57), a divalent cation is necessary for enzyme-mediated cleavage of 4x174 (+) strand molecules (40), as well as for cleavage and intramolecular religation of double-stranded nucleic acids (51,54,55). The enzyme's requirement for a divalent cation for intermolecular ligation was determined by chelating the magnesium in assays with EDTA.
As seen in Fig. 8, EDTA completely inhibited the joining of @X174 DNA to the 56mer. This inhibition was due to the chelation of the divalent cation rather than to the presence of EDTA, per se, since the back addition of excess MgClz once again supported ligation (not shown). Therefore, the intermolecular DNA ligation reaction of topoisomerase I1 requires the presence of a divalent cation.
Neither the intramolecular religation of double-stranded DNA (51,54,55)  stranded bacteriophage X DNA by calf thymus topoisomerase I1 is stimulated -10-fold by ATP (19), the effects of the high energy cofactor on intermolecular ligation of single-stranded 4x174 DNA to the oligonucleotide acceptor by the Drosophila enzyme was determined. As shown in Fig. 9, no stimulation of intermolecular DNA ligation was observed in the presence of ATP.
Characterization of the Oligonucleotide Acceptor-As determined by analysis of ligation products on denaturing alkaline agarose gels (see Fig. 4), topoisomerase I1 joined cleaved 4x174 (+) strand DNA to either strand of the blunt-ended double-stranded 56-mer with equal efficiency. To examine the influence of the configuration around the 3'-OH termini of oligonucleotide acceptor molecules, the 56-mer was digested at the restriction endonuclease sites shown in Fig. 1. Digestion with PuuII, EcoRI, or PstI yielded double-stranded 45-mers that contained a blunt end, a 4-base 5'-overhang, or a 4-base 3'-overhang, respectively. Topoisomerase I1 utilized all three 45-mers as acceptor molecules for intermolecular DNA ligation with similar efficiencies (Table I). Although the enzyme displayed a slight preference for the 45-mer with a 5"overhang over that with a blunt end over that with a 3'-overhang, it is not clear whether the minor differences in activity are due solely to changes in the oligonucleotide end configuration or are also influenced by differences in the DNA sequence around the ends.
Three additional aspects of the oligonucleotide acceptor were examined (not shown). First, intermolecular DNA ligation was not enhanced by the inclusion of a strong topoisom-  The blunt-end, 5'-overhang, and 3"overhang oligonucleotides were prepared by digesting the 56-mer shown in Fig. 1 with restriction endonuclease PuuII, EcoRI, or PstI, respectively.
Intermolecular DNA ligation was carried out for 30 or 60 min as described under "Experimental Procedures." The level of ligation obtained a t 60 min using the blunt-ended oligonucleotide acceptor was set arbitrarily to a value of 1.00. erase I1 recognition/cleavage site (51) in the double-stranded acceptor molecule. Second, the enzyme was able to utilize single-stranded oligonucleotides as acceptors for intermolecular ligation. Finally, topoisomerase I1 was able to transfer cleaved 6x174 DNA to a 3'-OH at a nick or a 5-base gap, but with a reduced efficiency.

DISCUSSION
A number of studies implicate a role for topoisomerase I1 in mediating DNA recombination in vitro and in vivo (17,19,21,26). All of the models that have been proposed for this activity require enzyme-mediated double-stranded DNA cleavage, followed by subunit exchange between two separate homodimers of topoisomerase I1 (17,19,22,29). However, since the type I1 enzyme can introduce nicks (that are kinetic intermediates in its double-stranded DNA cleavage/religation pathway) into the double helix (41-44), alternative mechanisms for illegitimate recombination that rely on the singlestranded DNA cleavage activity of topoisomerase I1 should also be considered. The intermolecular DNA ligation mechanism that was addressed in the present study is similar to models proposed for topoisomerase I-mediated recombination (16,22,24), does not require subunit exchange, and relies on the actions of a single topoisomerase I1 homodimer.
In order to examine the potential for this alternative illegitimate recombination pathway, the properties of topoisomerase 11-mediated intermolecular DNA ligation were characterized. To this end, a model system in which the enzyme covalently joins cleaved 4x174 (+) strand DNA to a doublestranded oligonucleotide acceptor was employed. Results indicate that topoisomerase I1 ligates the cleaved bacteriophage DNA exclusively to the 3'-OH termini of acceptor molecules in a time-dependent and an oligonucleotide concentrationdependent fashion. As previously determined for the enzyme's intramolecular DNA religation activity (51,54,55), intermolecular ligation requires a divalent cation. In contrast to recombination of double-stranded DNA substrates reported for calf thymus topoisomerase I1 (19), the intermolecular ligation of cleaved single-stranded 4x174 DNA to an oligonucleotide acceptor is not stimulated by the presence of ATP. This finding demonstrates that the enzyme's DNA strand passage event is not required for recombination in the present model system. Topoisomerase I1 displayed no stringent requirements for its acceptor molecule. Single-stranded oligonucleotides, as well as double-stranded oligonucleotides that contained 3'-OH termini with blunt-ends, 5'-overhangs, or 3"overhangs all could be utilized by the enzyme. Finally, topoisomerase 11 also was capable of transferring cleaved 4x174 DNA to a 3'-OH present at either a nick or a gap.
The formation of a recombination substrate in the present system requires the 3"OH DNA terminus generated by enzyme-mediated cleavage to diffuse away from the active site of topoisomerase 11. Bacteriophage 4x174 (+) strand DNA generates such a diffusible 3'-OH terminus upon cleavage by the Drosophila type I1 enzyme (40) and, as predicted, is an excellent model substrate for illegitimate recombination. Three different nucleic acid structures have been shown to produce diffusible 3'-OH termini following topoisomerase IImediated cleavage; single-stranded DNA, duplex molecules that contain nicks adjacent to the enzyme's cleavage site, and molecules with cleavage sites at double-stranded/singlestranded junctions (53,58).' Recent evidence indicates that nucleic acid molecules with this latter structure in fact can serve as substrates for a topoisomerase 11-mediated intermolecular DNA ligation reaction (58). Although sites of 4x174 (+) strand DNA cleavage by the Drosophila enzyme have not yet been mapped, bacteriophage T4 topoisomerase I1 cleaves 4x174 DNA in single-stranded regions and in regions that contain double-stranded/single-stranded junctions (i.e. at the base of hairpin loops) (53).
The nucleic acid structures described above, as well as others that also have potential to generate diffusible 3"OH termini, all exist in vivo (Fig. 10). For example, singlestranded regions (Fig. 1OA) are generated as a result of transcription (3) and prepriming for DNA replication (59,60). Nicks (Fig. 10B) are produced in the DNA helix by the actions of damaging agents (61), repair enzymes (61, 62), and replication on the lagging strand of parental DNA (63). Double-stranded/single-stranded junctions (Fig. 10, C and D) are formed by the conversion of palindromic sequences to cruciforms (64), the production of homologous recombination intermediates (65), and gapped DNA formed during lagging strand DNA replication (63). Finally, duplex nucleic acids (undergoing cleavage by topoisomerase 11) may be converted to single-stranded molecules by the actions of helicases (66), transcription complexes (3), or replication complexes (63).
The present study describes a novel mechanism for illegitimate recombination that relies on the ability of a single topoisomerase I1 homodimer to mediate intermolecular DNA ligation. The fact that DNA structures necessary to generate appropriate recombination substrates are prevalent in uiuo suggests that this intermolecular DNA ligation model for topoisomerase 11-mediated recombination has physiological significance.