Kinetic Analysis of DNA and RNA Strand Transfer Reactions Catalyzed by Vaccinia Topoisomerase*

Vaccinia topoisomerase binds duplex DNA and forms a covalent DNA-(3′-phosphotyrosyl) protein adduct at the sequence 5′-CCCTT↓. The enzyme reacts readily with a 36-mer CCCTT strand (DNA-p-RNA) composed of DNA 5′ and RNA 3′ of the scissile bond. However, a 36-mer composed of RNA 5′ and DNA 3′ of the scissile phosphate (RNA-p-DNA) is a poor substrate for covalent adduct formation. Vaccinia topoisomerase efficiently transfers covalently held CCCTT-containing DNA to 5′-OH-terminated RNA acceptors; the topoisomerase can therefore be used to tag the 5′ end of RNA in vitro. Religation of the covalently bound CCCTT-containing DNA strand to a 5′-OH-terminated DNA acceptor is efficient and rapid (k rel > 0.5 s−1), provided that the acceptor DNA is capable of base pairing to the noncleaved DNA strand of the topoisomerase-DNA donor complex. The rate of strand transfer to DNA is not detectably affected by base mismatches at the 5′ nucleotide of the acceptor strand. Nucleotide deletions and insertions at the 5′ end of the acceptor slow the rate of religation; the observed hierarchy of reaction rates is as follows: +1 insertion > −1 deletion > +2 insertion ≫ −2 deletion. These findings underscore the importance of a properly positioned 5′-OH terminus in transesterification reaction chemistry, but they also raise the possibility that topoisomerase may generate mutations by sealing DNA molecules with mispaired or unpaired ends.

Vaccinia topoisomerase, a 314-amino acid eukaryotic type I enzyme, binds and cleaves duplex DNA at a specific target sequence, 5Ј-(T/C)CCTT2 (1)(2)(3). Cleavage is a transesterification reaction in which the Tp2N phosphodiester is attacked by Tyr-274 of the enzyme, resulting in the formation of a DNA-(3Ј-phosphotyrosyl) protein adduct (4). The covalently bound topoisomerase catalyzes a variety of DNA strand transfer reactions. It can religate the CCCTT-containing strand across the same bond that was originally cleaved (as occurs during the relaxation of supercoiled DNA) or it can ligate the strand to a heterologous acceptor DNA 5Ј end, thereby creating a recombinant molecule (5)(6)(7).
Duplex DNA substrates containing a single CCCTT target site have been used to dissect the cleavage and strand transfer steps. A cleavage-religation equilibrium is established when topoisomerase transesterifies to DNA ligands containing Ն18 bp 1 of duplex DNA 3Ј of the cleavage site (8 -11). The reaction is in equilibrium because the 5Ј-OH-terminated distal segment of the scissile strand remains poised near the active site by virtue of the fact that it is stably base paired with the nonscissile strand. About 20% of the CCCTT-containing strand is covalently bound at equilibrium (11). "Suicide" cleavage occurs when the CCCTT-containing substrate contains six or fewer base pairs 3Ј of the scissile bond, because the short leaving strand dissociates from the protein-DNA complex. In enzyme excess, Ͼ90% of the suicide substrate is cleaved (11).
The suicide intermediate can transfer the incised CCCTT strand to a DNA acceptor. Intramolecular strand transfer occurs when the 5Ј-OH end of the noncleaved strand of the suicide intermediate attacks the 3Ј phosphotyrosyl bond and expels Tyr-274 as the leaving group. This results in formation of a hairpin DNA loop (5). Intermolecular religation occurs when the suicide intermediate is provided with an exogenous 5Ј-OH-terminated acceptor strand, the sequence of which is complementary to the single strand tail of the noncleaved strand in the immediate vicinity of the scissile phosphate (5). In the absence of an acceptor strand, the topoisomerase can transfer the CCCTT strand to water, releasing a 3Ј-phosphateterminated hydrolysis product, or to glycerol, releasing a 3Јphosphoglycerol derivative (12). Although the hydrolysis and glycerololysis reactions are much slower than religation to a DNA acceptor strand, the extent of strand transfer to non-DNA nucleophiles can be as high as 15-40%.
The specificity of vaccinia topoisomerase in DNA cleavage and its versatility in strand transfer have inspired topoisomerase-based strategies for polynucleotide synthesis in which DNA oligonucleotides containing CCCTT cleavage sites serve as activated linkers for the joining of other DNA molecules with compatible termini (13). In the present study, we examined the ability of the vaccinia topoisomerase to cleave and rejoin RNAcontaining polynucleotides. It was shown previously that the enzyme did not bind covalently to CCCTT-containing molecules in which either the scissile strand or the complementary strand was composed entirely of RNA (9). To further explore the pentose sugar specificity of the enzyme, we have prepared synthetic CCCTT-containing substrates in which the scissile strand is composed of DNA-and RNA-containing halves. In this way, we show that the enzyme is indifferent to RNA downstream of the scissile phosphate, but it does not form the covalent complex when the region 5Ј of the scissile phosphate is in RNA form. Also, we assess the contribution of base pairing by the 5Ј end of the acceptor strand to the rate of the DNA strand transfer reaction.

MATERIALS AND METHODS
Preparation of Tandem RNA-p-DNA and DNA-p-RNA Oligonucleotides-CCCTT-containing 36-mer oligonucleotides containing a single internal 32 P-label at the scissile phosphate were prepared by ligating two 18-mer strands (synthetic RNA or DNA oligonucleotides) that had been hybridized to a complementary 36-mer DNA strand. The sequence of the proximal CCCTT-containing 18-mer strand was 5Ј-CATATCCGT-GTCGCCCTT as DNA or 5Ј-CAUAUCCGUGUCCCUU as RNA. The sequence of the distal 18-mer strand was 5Ј-ATTCCGATAGTGAC-TACA as DNA or 5Ј-AUUCCGAUAGUGACUACA as RNA. The distal 18-mer strand was 5Ј-labeled in the presence of [␥-32 P]ATP and T4 polynucleotide kinase and then gel-purified. The sequence of the 36-mer strand was 5Ј-TGTAGTCACTATCGGAATAAGGGCGACACGGATA-TG. The strands were annealed in 0.2 M NaCl by heating at 65°C for 2 min, followed by slow cooling to room temperature. The molar ratio of the 5Ј-labeled distal 18-mer to the proximal 18-mer and the 36-mer strand in the hybridization mixture was 1:4:4. The singly nicked product of the annealing reaction was sealed in vitro with purified recombinant vaccinia virus DNA ligase (14,15). The ligation reaction mixtures (400 l) contained 50 mM Tris-HCl (pH 8.0), 5 mM dithiothreitol, 10 mM MnCl 2 , 1 mM ATP, 10 pmol of 5Ј 32 P-labeled nicked substrate, and 160 pmol of ligase. After incubation for 4 h at 22°C, the reactions were halted by the addition of EDTA to a final concentration of 25 mM. The samples were extracted with phenol-chloroform, and the labeled nucleic acid was recovered from the aqueous phase by ethanol precipitation. The 36-mer duplex products were dissolved in TE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA). Ligation of the labeled 18-mer distal strand to the unlabeled CCCTT-containing 18-mer strand to form an internally labeled 36-mer product was confirmed by electrophoresis of the reaction products through a 17% denaturing polyacrylamide gel. The extents of ligation [36-mer/(36-mer ϩ 18-mer)] were as follows: DNA-p-DNA, 88%; DNA-p-RNA, 67%; and RNA-p-DNA, 66%.
Covalent Binding of Topoisomerase to Internally Labeled 36-mer Duplexes-Recombinant vaccinia topoisomerase was expressed in bacteria and purified via phosphocellulose and SP5PW column chromatography as described (16,17). Reaction mixtures for assay of covalent adduct formation contained (per 20 l) 50 mM Tris-HCl (pH 8.0), 0.2 pmol of 36-mer duplex, and 1 pmol of topoisomerase. The reactions were initiated by adding topoisomerase and halted by adding SDS to 1% final concentration. The samples were analyzed by SDS-polyacrylamide gel electrophoresis. Covalent complex formation was revealed by the transfer of radiolabeled polynucleotide to the topoisomerase polypeptide (3). The extent of adduct formation was quantitated by scanning the gel using a FUJIX BAS1000 phosphorimager and was expressed as the percentage of the input 5Ј 32 P-labeled 36-mer substrate that was covalently transferred to protein.
DNA Strand Transfer to an RNA Acceptor-An 18-mer CCCTTcontaining DNA oligonucleotide (5Ј-CGTGTCGCCCTTATTCCC) was 5Ј end-labeled in the presence of [␥-32 P]ATP and T4 polynucleotide kinase, then gel-purified and hybridized to a complementary 30-mer strand to form the 18-mer/30-mer suicide cleavage substrate. Covalent topoisomerase-DNA complexes were formed in a reaction mixture containing (per 20 l) 50 mM Tris-HCl (pH 8.0), 0.5 pmol of 18-mer/30-mer DNA, and 2.5 pmol of topoisomerase. The mixture was incubated for 5 min at 37°C. The strand transfer reaction was initiated by the addition of an 18-mer acceptor strand, 5Ј-ATTCCGATAGTGACTACA (either DNA or RNA), to a concentration of 25 pmol/20 l (i.e. a 50-fold molar excess over the input DNA substrate), while simultaneously the reaction mixtures were adjusted to 0.3 M NaCl. The reactions were halted by addition of SDS and formamide to 0.2 and 50%, respectively. The samples were heat-denatured and then electrophoresed through a 17% polyacrylamide gel containing 7 M urea in TBE (90 mM Tris borate, 2.5 mM EDTA). The extent of strand transfer (expressed as the percentage of input labeled DNA converted to a 30-mer strand transfer product) was quantitated by scanning the wet gel with a phosphorimager.
Preparation of 32 P-labeled 36-mer RNA-A 36-nucleotide run-off transcript was synthesized in vitro by T3 RNA polymerase from a pBluescript II-SK(Ϫ) plasmid template that had been linearized by digestion with endonuclease EagI. A transcription reaction mixture (100 l) containing 40 mM Tris-HCl (pH 8.0), 6 mM MgCl 2 , 2 mM spermidine, 10 mM NaCl, 10 mM dithiothreitol, 0.5 mM ATP, 0.5 mM CTP, 0.5 mM UTP, 6.25 M [␣ 32 P]GTP, 5 g of template DNA, and 100 units of T3 RNA polymerase (Promega) was incubated for 90 min at 37°C. The reaction was halted by adjusting the mixture to 0.1% SDS, 10 mM EDTA, and 0.5 M ammonium acetate. The samples were extracted with phenol-chloroform and ethanol-precipitated. The pellet was resuspended in formamide and electrophoresed through a 12% polyacrylamide gel containing 7 M urea in TBE. The radiolabeled 36mer RNA was localized by autoradiography of the wet gel and eluted from an excised gel slice by soaking for 16 h at 4°C in 0.4 ml of buffer containing 1 M ammonium acetate, 0.2% SDS, and 20 mM EDTA. The eluate was phenol-extracted and ethanol-precipitated. The RNA was resuspended in TE. Dephosphorylation of the RNA 5Ј terminus was carried out in a reaction mixture (30 l) containing 10 mM Tris-HCl (pH 7.9), 50 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol, 10 pmol of 36-mer RNA, and 30 units of calf intestine alkaline phosphatase (New England Biolabs). After a 1-h incubation at 37°C, the mixture was phenol-extracted and ethanol-precipitated. The phosphatase-treated 36-mer transcript was repurified electrophoretically as described above.

Covalent Binding of Topoisomerase to a Duplex Substrate
Containing RNA 3Ј of the Scissile Phosphate-Vaccinia topoisomerase does not bind covalently to CCCTT-containing RNA duplexes, nor does it form a covalent complex on RNA-DNA hybrid duplexes in which one of the two strands is RNA (9). Control experiments showed that the failure to form a covalent adduct on a CCCUU-containing RNA strand was not caused by uracil substitution for the thymine bases in the CCCTT sequence (9). To better understand why vaccinia topoisomerase does not form a covalent complex with all-RNA strands, we prepared 36-bp duplex substrates in which the scissile strand was a tandem RNA-DNA or DNA-RNA copolymer and the noncleaved strand was all DNA (Fig. 1). These duplexes were uniquely labeled with 32 P at the scissile phosphodiester. The substrate molecules were constructed by annealing two 18-mer oligonucleotides (one of which had been 5Ј 32 P-labeled) to a complementary 36-mer DNA strand to form a singly nicked duplex. The 5Ј-labeled 18-mer strand was then joined to the unlabeled CCCTT-strand (or CCCUU strand) in a reaction catalyzed by vaccinia virus DNA ligase. (The properties of vaccinia ligase in joining RNA and DNA strands will be described elsewhere.) The 36-mer duplex products were isolated and then used as substrates for vaccinia DNA topoisomerase. We will refer to these substrates as DNA-p-DNA, DNA-p-RNA, and RNA-p-DNA, with the labeled phosphate being denoted by p.
Transesterification by topoisomerase at the CCCTT site will result in covalent binding of a 3Ј 32 P-labeled 18-mer oligonucleotide to the enzyme. The extent of covalent complex formation on the DNA-p-RNA substrate in 10 min was proportional to input topoisomerase; 80 -85% of the 36-mer strand was transferred to the topoisomerase at saturating enzyme (Fig. 1). The same level of topoisomerase covalently bound less than 1% of the RNA-p-DNA 36-mer strand. Hence, the topoisomerase tolerated RNA substitution downstream of the scissile phos- phate but was impeded from forming the covalent adduct when the CCCTT sequence was in RNA form.
We assessed the kinetics of the covalent binding reaction at a saturating level of topoisomerase (Fig. 2). An all-DNA 36-mer (DNA-p-DNA) was bound to an end point of 21% in 2 min ( Fig.  2A). The apparent cleavage-religation equilibrium constant (K cl ϭ covalent complex/noncovalent complex) was 0.26, which agrees with the values of 0.2-0.25 that were reported previously for equilibrium cleavage of a 5Ј end-labeled CCCTTcontaining DNA substrate (10,11). The DNA-p-RNA 36-mer was bound covalently to an end point of 80% in 5 min ( Fig. 2A and other data not shown). The apparent equilibrium constant for DNA-p-RNA (K cl ϭ 4) was significantly higher than that observed for the all-DNA ligand.
The RNA-p-DNA 36-mer was transferred to the topoisomerase, albeit very slowly. After 4 h, 4% of the CCCUU-containing RNA strand was bound covalently to the enzyme (Fig. 2B). An end point was not established in this experiment. However, by comparing the initial rate of covalent adduct formation on RNA-p-DNA (0.04% of input substrate cleaved per min) to the amount adduct formed on DNA-p-DNA at the earliest time point (12% in 10 s), we estimate that RNA substitution of the CCCTT-portion of the substrate slowed the rate of covalent complex formation by about 3 orders of magnitude.
DNA Strand Transfer to an RNA Acceptor-Rejoining of the cleaved strand occurs by attack of a 5Ј-hydroxyl-terminated polynucleotide on the 3Ј-phosphodiester bond between Tyr-274 and the CCCTT site. This transesterification step can be studied independent of strand cleavage by assaying the ability of a preformed topoisomerase-DNA complex to religate the covalently held strand to a heterologous acceptor strand (5, 11). To form the covalent topoisomerase-DNA donor complex, the enzyme was initially incubated with a suicide substrate consisting of a 5Ј 32 P-labeled 18-mer scissile strand (CGTGTCGC-CCTTATTCCC) hybridized to a 30-mer strand. Cleavage of this DNA by topoisomerase is accompanied by dissociation of the 6-nucleotide leaving group, ATTCCC. With no readily available acceptor for religation, the enzyme is essentially trapped on the DNA as a suicide intermediate (Fig. 3). In a 5-min reaction in enzyme excess, Ͼ90% of the 5Ј 32 P-labeled strand becomes covalently bound to protein. The strand transfer reaction was initiated by adding a 50-fold molar excess of an 18-mer acceptor strand (either DNA or RNA) complementary to the 5Ј singlestrand tail of the covalent donor complex (Fig. 3) while simultaneously increasing the ionic strength to 0.3 M NaCl. Addition of NaCl during the religation phase promotes dissociation of the topoisomerase after strand closure and prevents recleavage of the strand transfer product. Ligation of the covalently held 12-mer CGTGTCGCCCTT to the 18-mer yields a 32 P-labeled 30-mer (Fig. 4, lane 1). The suicide intermediate transferred 94% of the input CCCTT-containing strand to the 18-mer DNA strand (Fig. 3). The extent of religation at the earliest time point (5 s) was 90% of the end point value. We calculated from this datum a religation rate constant (k rel ) of Ͼ0.5 s Ϫ1 . We had determined previously (from experimental values for k cl and K eq at 37°C) a k rel value of ϳ1.3 s Ϫ1 (18).
Topoisomerase readily ligated the covalently held 12-mer DNA to an 18-mer RNA acceptor to form a 30-mer product (Fig.  4, lane 5). 89% of the input CCCTT-strand was transferred to RNA, with 40% of the end point value attained in 5 s. We used this datum to estimate a rate constant of 0.1 s Ϫ1 for singleturnover strand transfer to RNA. Thus, religation to DNA was about 10 times faster than religation to RNA. The slowed rate of RNA religation is likely to account for the observed increase in the cleavage-religation equilibrium constant (K eq ϭ k cl /k rel ) on the DNA-p-RNA 36-mer.
Analysis of the Strand Transfer Reaction Product-The predicted product of strand transfer to RNA is a 30-mer tandem DNA-RNA strand (5Ј-CGTGTCGCCCTTAUUCCGAUAGU-GACUACA) uniquely 32 P-labeled at the 5Ј end. The structure of this molecule was confirmed by analysis of the susceptibility of this product to treatment with NaOH. The labeled 30-mer RNA ligation product was converted nearly quantitatively into a discrete species that migrated more rapidly than the input 18-mer CCCTT-containing DNA strand (Fig. 4, lane 6). The mobility of this product was consistent with a chain length of 13 nucleotides. The expected 32 P-labeled alkaline hydrolysis product of the RNA strand transfer product is a 13-mer (5Ј-CGTGTCGCCCTTAp). Control reactions showed that neither the 32 P-labeled 18-mer scissile strand of the suicide substrate nor the 30-mer product of strand transfer to DNA was susceptible to alkali (Fig. 4, lanes 4 and 2). We conclude that topoisomerase can be used to ligate RNA to DNA in vitro.
DNA Ligand Tagging of an RNA Transcript Synthesized in Vitro by T3 RNA Polymerase-Practical applications of topoisomerase-mediated strand transfer to RNA include the 5Ј tagging of RNA transcripts. Bacteriophage RNA polymerases have been used widely to synthesize RNA in vitro from plasmid DNA templates containing phage promoters. To test whether such transcripts were substrates for topoisomerase-catalyzed ligation, we constructed a CCCTT-containing suicide cleavage substrate that, when cleaved by topoisomerase, would contain a 5Ј single-strand tail complementary to the predicted 5Ј sequence of any RNA transcribed by T3 RNA polymerase from a pBluescript vector (Fig. 5). A 36-nucleotide T3 transcript was synthesized in a transcription reaction containing [␣ 32 P]GTP. The RNA was treated with alkaline phosphatase to dephosphorylate the 5Ј terminus. The topoisomerase-DNA covalent intermediate was formed on an unlabeled suicide substrate. Incubation of the radiolabeled T3 transcript with the suicide intermediate resulted in the conversion of the 36-mer RNA into a novel species that migrated more slowly during polyacrylamide gel electrophoresis (not shown). The apparent size of this product (48 nucleotides) was indicative of ligation to the 12-mer CCCTT DNA strand. The kinetics of DNA ligation to the T3 transcript are shown in Fig. 5. The reaction was virtually complete within 1 min; at its end point, 29% of the input RNA had been joined to DNA. No DNA-RNA ligation product was formed in reaction containing a T3 transcript that had not been treated with alkaline phosphatase (not shown).

Formation of Insertions and Deletions: A Kinetic Analysis-
The acceptor polynucleotides used in the experiments described above were capable of hybridizing perfectly with the 5Ј single-strand tail of the topoisomerase-DNA donor complex. It had been shown previously that the vaccinia virus topoisomerase is capable of joining the CCCTT-strand to an acceptor oligonucleotide that hybridizes so as to leave a single nucleotide gap between the covalently bound donor 3Ј end and the 5Ј terminus of the acceptor. Religation across this gap generated a 1-base deletion in the product compared with the input scissile strand (5). The enzyme also catalyzes strand transfer to an acceptor oligonucleotide that, when hybridized, introduces an extra nucleotide between the donor 3Ј end and the penultimate The samples were extracted with phenol-chloroform and ethanol-precipitated. The pellets were resuspended in either 12 l of 0.1 M NaOH, 1 mM EDTA (NaOH ϩ ) or 12 l of 10 mM Tris-HCl (pH 8.0), 1 mM EDTA (NaOH Ϫ ). These samples were incubated at 37°C for 16 h. Control samples containing the input 18-mer DNA substrate that had not been exposed to topoisomerase were treated in parallel (lanes 3 and 4). The alkali-treated samples were neutralized by adding 1.2 l of 1 M HCl. All samples were then ethanol-precipitated, resuspended in formamide, heated for 5 min at 95°C, and electrophoresed through a 17% polyacrylamide gel containing 7 M urea in TBE. An autoradiograph of the gel is shown. The positions of the 30-mer religation product and the 18-mer input strand are indicated at the left. Alkaline hydrolysis of the RNA strand transfer reaction product (lane 6) yielded a discrete species denoted by the asterisk. Incubation was at 37°C. Aliquots (15 l) were removed at the times indicated and quenched immediately by adding SDS and EDTA. The samples were adjusted to 50% formamide, heated for 5 min at 95°C, and electrophoresed through a 12% polyacrylamide gel containing 7 M urea in TBE. Transfer of the 12-nucleotide DNA donor strand to the 5Ј end of the labeled 36-mer T3 transcript yielded a labeled 48-mer product. Conversion of input 36-mer to 48-mer was quantitated by scanning the gel with a phosphorimager. base paired nucleotide of the acceptor. Religation in this case will produce a 1-base insertion (5). Deletion and insertion formation in vitro have also been documented for mammalian type I topoisomerase (19). However, there has been no report of the effects of acceptor strand gaps and insertions on the rate of strand joining by these enzymes.
We assessed the kinetics of strand transfer by the vaccinia topoisomerase covalent intermediate to acceptor oligonucleotides that base pair with the donor complex to form either a fully base-paired 3Ј duplex segment or 3Ј duplexes with a 1-or 2-nucleotide gap. 84% of the input DNA substrate was ligated to the fully paired acceptor in 10 s, the shortest time analyzed (Fig. 6A). The size of the strand transfer product was 30 nucleotides, as expected (Fig. 7, lane 3). No 30-mer product was formed in the absence of the added acceptor strand (Fig. 7, lane  2).
Religation across a 1-nucleotide gap was highly efficient, albeit slow. 85% of the input substrate was joined across a 1-nucleotide gap to yield the expected 29-nucleotide product ( Fig. 6A and Fig. 7, lane 4). The kinetic data in Fig. 6 fit well to a single exponential with an apparent rate constant of 0.005 s Ϫ1 . Thus, single-turnover strand closure by topoisomerase across a 1-nucleotide gap was 2 orders of magnitude slower than the rate of joining across a fully paired nick. Vaccinia topoisomerase catalyzed strand transfer across a 2-nucleotide gap to form the anticipated 28-nucleotide product (Fig. 7, lane 5), but this reaction was feeble (Fig. 6A). We observed linear accumulation of the 2-nucleotide gap product over a 2-h incubation, at the end of which only 10% of the input DNA had been joined. We estimated, based on the initial rate, that religation across the 2-nucleotide gap was 2 orders of magnitude slower than joining across a 1-nucleotide gap (and hence 4 orders of magnitude slower than the rate of joining across a nick).
Similar experiments were performed using DNA acceptors that contained either 1 or 2 extra nucleotides at their 5Ј ends (Fig. 6B). Religation to these acceptors yielded labeled strand transfer products of 31 and 32 nucleotides, respectively (Fig. 7,  lanes 6 and 7). 90% of the input DNA was religated to form the 1-nucleotide insertion product (Fig. 6B). We calculated a rate constant of 0.04 s Ϫ1 for religation with 1-nucleotide insertion. A similar end point was achieved in the formation of a 2-nucleotide insertion product, but the strand transfer rate was considerably slower (Fig. 6B). The observed rate constant for 2-nucleotide insertion was 0.001 s Ϫ1 , i.e. 3 orders of magnitude lower than k rel at a nick.
Effect of 5Ј Acceptor Base Mismatch on Strand Transfer-We examined strand transfer by topoisomerase to a set of 18-mer acceptors that were capable of forming base pairs with the 5Ј tail of the donor complex from positions Ϫ2 to Ϫ18 (relative to the scissile ϩ1 T:A base pair of the CCCTT element) but that have a base mismatch at the Ϫ1 position immediately 3Ј of the scissile bond. The control acceptor, which has a normal Ϫ1 A:T base pair, reacted to completion in 10 s; 89% of the end point was achieved in 5 s (Fig. 8). DNAs containing T:T, C:T, or G:T mispairs at the Ϫ1 position supported the same extent of strand transfer; 77% of the end point was attained in 5 s in each case (Fig. 8). Thus, within the limits of detection of this experiment, mismatch at the Ϫ1 position had little effect on the strand transfer reaction. There are clear and instructive differences between the effects of base mismatches versus a single nucleotide deletion on the rate of the strand joining step.
Kinetics of Intramolecular Hairpin Formation-In the absence of an exogenous acceptor oligonucleotide, the 5Ј-OH terminus of the nonscissile strand of the 12-mer/30-mer covalent complex can flip back and act as the nucleophile in attacking the DNA-(3-phosphotyrosyl) bond (5). The reaction product is a hairpin molecule containing a 12-bp stem and an 18-nucleotide loop. The kinetics of this reaction were examined under single turnover conditions. In the experiment shown in Fig. 9A, 65% of the input CCCTT strand was converted to hairpin product in 3 h. The observed rate constant was 5.7 ϫ 10 Ϫ4 sec Ϫ1 . In parallel, we analyzed the rate of hairpin formation by the covalent complex formed on an 18-bp cleavage substrate (Fig. 9A). In this case, an attack by the 5Ј-OH of the nonscissile strand yielded a hairpin molecule containing a 12-bp stem and a 6-nucleotide loop. 69% of the input CCCTT strand was converted to hairpin product in 10 h. The observed rate constant was 8.2 ϫ 10 Ϫ5 sec Ϫ1 . Thus, the 18-nucleotide 5Ј tail was ϳ7 times more effective than the 6-mer 5Ј tail as the attacking nucleophile for strand transfer in cis. Note that hairpin formation by these covalent complexes occurs without any potential for base pairing by the single-strand tails.
To examine the contribution of base pairing to the rate of religation, we altered the 5Ј terminal and penultimate bases of bottom strand of the 18-mer/30-mer substrate to 5Ј-AT (Fig.  9B). Now, the 5Ј-terminal three bases of the bottom strand (5Ј-ATT) are identical to the 5Ј-terminal bases of the leaving strand (5Ј-ATTCCC); hence, the single-strand tail is self-complementary and capable of forming three base pairs adjacent to the scissile phosphate. Intramolecular hairpin formation on this DNA was extremely fast; the reaction was complete in 10 -20 s (Fig. 9B). The observed religation rate constant was 0.2 s Ϫ1 . By comparing this value to the religation rate constant on the noncomplementary 18-mer/30-mer substrate (Fig. 9A), we surmise that three base pairs accelerated the reaction ϳ350-fold.
Kinetics of Single-turnover Cleavage of a CCCTT-containing Hairpin Molecule-The 42-nucleotide 5Ј 32 P-labeled hairpin product was gel-purified and tested as a substrate for covalent adduct formation by the vaccinia topoisomerase. 55% of the input radioactivity was transferred to the topoisomerase polypeptide in 15 s at 37°C; an end point of 90% transfer was attained in 60 s (data not shown). The apparent rate constant for cleavage of the hairpin was 0.06 s Ϫ1 . Thus, the topoisomer-ase rapidly and efficiently cleaved a CCCTT-containing molecule in which there were no standard paired bases downstream of the scissile phosphate. The hairpin cleavage rate constant is about one-fifth of k cl on the 18-mer/30-mer suicide substrate, which contains five paired bases of duplex DNA 3Ј of the CCCTT site.

DISCUSSION
Vaccinia topoisomerase catalyzes a diverse repertoire of strand transfer reactions. Religation of the covalently bound DNA to a perfectly base-paired acceptor DNA oligonucleotide provides a model for the strand closure step of the DNA relaxation reaction. Here, we have analyzed the kinetics of strand transfer to alternative nucleic acid acceptors. Our findings provide new insights into the parameters that affect transesterification rate, illuminate the potential for topoisomerase to generate mutations in vivo, and suggest practical applications of vaccinia topoisomerase as an RNA modifying enzyme.
Sugar Specificity for Covalent Adduct Formation Resides within the CCCTT Element-Vaccinia topoisomerase is apparently incapable of binding covalently to CCCUU-containing RNA strands. This is the case whether the CCCUU strand is part of an RNA-RNA or an RNA-DNA duplex (9). We have now shown that the sugar specificity of the enzyme is attributable to a stringent requirement for DNA on the 5Ј side of the scissile phosphate, i.e. the CCCTT site must be DNA. Moreover, the CCCTT element must be a DNA-DNA duplex, because earlier experiments showed that a CCCTT strand is not cleaved when it is annealed to a complementary RNA strand (9). The RNA-DNA hybrid results are informative, because they suggest that the CCCTT site must adopt a B-form helical conformation to be cleaved. RNA and DNA polynucleotide chains adopt different conformations within an RNA-DNA hybrid, with the RNA strand retaining the A-form helical conformation (as found in double-stranded RNA), whereas the DNA strand adopts a conformation that is neither strictly A nor B, but is instead intermediate in character between these two forms (20,21). Vaccinia topoisomerase makes contacts with the nucleotide bases of the CCCTT site in the major groove (9,22). It also makes contacts with specific phosphates of the CCCTT site (23). Adoption by the CCCTT site of a non-B conformation may weaken or preclude these contacts.
Our finding that vaccinia topoisomerase is relatively insensitive to the nucleotide sugar composition downstream of the scissile phosphate implies that the conformation of the helix in this portion of the ligand is not important for site recognition or reaction chemistry. Topoisomerase cleaves DNA-p-RNA strands in which the leaving strand is RNA. Indeed, the extent of cleavage at equilibrium is significantly higher than that achieved on a DNA-p-DNA strand.
Strand Transfer to RNA-The increase in the cleavage-religation equilibrium constant K eq (ϭ k cl /k rel ) on the DNA-p-RNA substrate can be explained by our finding that the rate of single-turnover RNA religation k rel(RNA) is about one-tenth that of k rel(DNA) . Nonetheless, the extent of religation to RNA is quite high, i.e. ϳ90% of the input CCCTT strand is religated to an 18-mer RNA acceptor strand in a 2-min reaction. We have shown that a CCCTT-containing DNA strand can be rapidly joined by topoisomerase to a transcript synthesized in vitro by bacteriophage RNA polymerase; ϳ30% of the RNA was transferred to the DNA strand in a 2-5 min reaction. This property can be exploited to 5Ј tag any RNA for which the 5Ј terminal RNA sequence is known, i.e. by designing a suicide DNA cleavage substrate for vaccinia topoisomerase in which the nonscissile strand is complementary to the 5Ј sequence of the intended RNA acceptor. Some practical applications include: (i) 32 Plabeling of the 5Ј end of RNA; and (ii) affinity labeling the 5Ј end of RNA, e.g. by using a biotinylated topoisomerase cleavage substrate. A potential advantage of topoisomerase-mediated RNA strand joining (compared with the standard T4 RNA ligase reaction) is that ligation by topoisomerase can be targeted by the investigator to RNAs of interest within a complex mixture of RNA molecules.
Frameshift and Missense Mutagenesis-It was reported earlier that vaccinia topoisomerase can religate to complementary DNA acceptors containing recessed ends or extra nucleotides, thereby generating the equivalent of frameshift mutations (5). Similar reactions have been described by Henningfeld and Hecht (19) for the cellular type I topoisomerase. A key question is whether these aberrant religation reactions are robust enough to implicate topoisomerase as a potential mutagen in vivo. Our kinetic analysis suggests that they are and provides the first clue as to what spectrum of frameshift reactions is most likely to occur (taking into account only the intrinsic properties of the topoisomerase). For the vaccinia enzyme, the hierarchy of frameshift generating religation reactions is as follows: ϩ1 insertion Ͼ Ϫ1 deletion Ͼ ϩ2 insertion Ͼ Ͼ Ϫ2 deletion.
The slowest of these topoisomerase-catalyzed reactions is strand closure across a 2-nucleotide gap (initial rate ϭ 0.002% of input DNA religated/sec). In this situation, the attacking nucleophile is held in place at some distance from the DNAprotein phosphodiester by base pairing with the nonscissile strand. Moving the 5Ј-hydroxyl 1 base pair closer to the phosphodiester enhances reaction rate by a factor of 100. Extra nonpaired nucleotides appear to pose much less of an impediment to strand joining to form 1-or 2-nucleotide insertions. The active site of the topoisomerase may be able to accommodate extrahelical nucleotides; alternatively, these nucleotides may intercalate into the DNA helix at the topoisomerase-induced nick.
There are two potential pathways for topoisomerase to form minus frameshifts in vivo, which differ as to how the acceptor strand is generated: (i) the 5Ј end of the leaving strand can be trimmed by a nuclease, after which ligation could occur across the resulting gap; or (ii) a homologous DNA single strand could attack the covalent intermediate. The second pathway presumably requires a helicase to form the invading strand (and perhaps also to displace the leaving strand). In the case of plus frameshifts, only the latter pathway would be available to the topoisomerase, i.e. because no mechanism exists to add nucleotides to the 5Ј terminus of the original leaving strand. No matter which pathway is taken, it is reasonable to assume that the most rapidly catalyzed mutagenic strand-joining reactions are the ones most likely to make their mark in vivo. If the religation reaction is slow, as for Ϫ2 frameshifting, then the cell has greater opportunity to repair the mutagenic lesion, e.g. by removing the covalently bound topoisomerase. This could entail: (i) excision of a patch of the DNA strand to which the topoisomerase is bound; or (ii) hydrolysis of the topoisomerase-DNA adduct. An enzyme that catalyzes the latter reaction was discovered recently by Yang et al. (24).
Introducing a base mismatch at the Ϫ1 position immediately flanking the scissile phosphate has almost no effect on the rate of religation. This result is in stark contrast to the 10 Ϫ2 rate effect of a 1-nucleotide gap. We infer that the Ϫ1 base mismatches do not significantly alter the proximity of the 5Јhydroxyl nucleophile of the terminal nucleotide to the scissile phosphate at enzyme's active site. Our results indicate clearly that topoisomerase has the capacity to generate missense mutations in vitro. The single-strand invasion pathway invoked above for frameshift mutagenesis could, in principle, provide the opportunity for topoisomerase to create missense mutations in vivo. The kinetics of ligation in vitro suggest that topoisomerase-generated missense mutations would predominate over frameshifts.
Kinetic Contribution of Base Complementarity-Kinetic analysis of intramolecular hairpin formation by the vaccinia topoisomerase provides the first quantitative assessment of the role of base complementarity in strand closure. The rate constant for attack on the DNA-(3Ј-phosphotyrosyl) bond by a nonpairing 18-nucleotide single strand linked in cis to the covalent complex was 5.7 ϫ 10 Ϫ4 sec Ϫ1 . Altering only the terminal bases of the single-strand tail to allow the formation of base pairs at the Ϫ1, Ϫ2, and Ϫ3 positions increased the rate constant for hairpin formation by 350-fold. The rate of religation in cis with 3 potential base pairs was nearly the same as the rate of religation to a non-covalently linked acceptor strand that forms 18 base pairs 3Ј of the scissile bond. The ability of the covalently bound enzyme to take up and rapidly rejoin DNA strands with only three complementary nucleotides lends credence to the suggestion that vaccinia topoisomerase catalyzes the formation of recombination intermediates in vivo (25), either via strand invasion or by reciprocal strand transfer between two topoisomerase-DNA complexes. Efforts to model these reactions in vitro are under way.