Mutations of vaccinia virus DNA topoisomerase I that stabilize the cleavage complex.

Two mutations in vaccinia virus topoisomerase I, K167D and G226N, have been characterized. SOS induction was observed in Escherichia coli expressing vaccinia topoisomerase I with either one of these mutations. The mutant enzymes were purified to homogeneity and compared with the wild type enzyme for relaxation activity and the partial activities of substrate binding, site-specific DNA cleavage and DNA religation to determine the mechanism of SOS induction. The K167D mutant enzyme had reduced binding affinity for the DNA substrate with a Kapp that was 10-fold higher than wild type. Nevertheless, in reactions with high enzyme concentration, its substrate cleavage activity was 90% that of wild type. The G226N mutant enzyme had virtually wild type binding and cleavage activities. However, intermolecular religation by these two mutants were observed to be significantly reduced. The cleavage complexes formed with the K167D and G226N mutants were more stable to high salt than the wild type cleavable complex. We propose that these mutants in vivo induce the SOS response in E. coli due to the shift of topoisomerase cleavage-religation equilibrium towards cleavage and increased stability of the cleavage complex. The mutation thus has a similar effect as the topoisomerase-targeting inhibitors that turn topoisomerases into DNA damaging agents.

DNA topoisomerases are enzymes that catalyze the interconversion of DNA topological forms by cleavage of DNA followed by DNA strand passage, and then rejoining of the cleaved DNA (reviewed in Refs. 14). DNA cleavage is due to the nucleophilic attack of DNA phosphodiester bond by the hydroxyl group of a tyrosine residue in the topoisomerase active site. A phosphotyrosine linkage is formed in the resulting covalent complex. After strand passage or swivel action at the cleavage site leading to topological change, nucleophilic attack on the phosphotyrosine linkage by the DNA hydroxyl group displaced previously rejoins the DNA phosphodiester backbone. DNA cleavage-rejoining is highly concerted. To isolate the covalent cleavage complex with duplex DNA, denaturation of enzyme or unusual substrates that abolish certain protein-DNA interactions are required (2, 5-81. A large number of clinically important antimicrobial and anti-cancer drugs have been shown to target against topoisomerases (9)(10)(11)(12)(13). Many of these drugs act by enhancing the formation of the covalent topoisomerase-DNA cleavage complexes either by increasing the DNA cleavage rate or inhibition of DNA religation rate. For example, the anti-* This work was supported by Grant CH478 from the American Cancer Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed. Tel.: 914-993-4061; Fax: 914-993-4058. cancer drug camptothecin has been shown to inhibit the relaxation activity and enhance the cleavage complex of eukaryotic type I DNA topoisomerases from human, yeast, Drosophila, wheat germ, Zbtrahyrnena, and calf thymus (14)(15)(16)(17)(18). Unlike these eukaryotic type I topoisomerases, vaccinia virus topoisomerase I is resistant to camptothecin (19). We have demonstrated previously that the single mutation D221V can convert the vaccinia topoisomerase I to camptothecin sensitive (20). It therefore has a potential interaction site for camptothecin despite its small size (32,000 M,) compared to the other eukaryotic type I topoisomerases (90,000-100,000 Mr).
The relatively small size of vaccinia topoisomerase I facilitates structure and function analysis. We have been attempting to isolate and characterize topoisomerase mutations that mimic the action of the topoisomerase poisons in inhibiting the religation of topoisomerase-mediated DNA cleavage. Cells expressing such mutants are expected to have increased accumulation of the topoisomerase-DNA covalent cleavage complex. In yeast, topoisomerase inhibitors targeting against both topoisomerase I and topoisomerase I1 have been shown to induce P-galactosidase production from chromosomal DIN3::lacZ fusion (18). The DIN3 gene is a DNA damage-inducible gene that is probably involved in repair of topoisomerase-mediated lesion. In bacteria, gyrase inhibitors are also known to induce the SOS response (21). The r e d and dinDl genes of Escherichia coli are two of the genes induced in the SOS repair pathway in response to DNA damage (22)(23)(24). E. coli strains with chromosomal dinD1::lacZ fusion have been utilized to characterize cleavage of DNA by restriction enzymes i n vivo (25,26). A ApL::lacZ fusion strain has also been used to isolate IS10 transposase mutants that produce a transposition intermediate and induce the SOS response but are unable to complete transposition (27). Our screening for i n vivo SOS induction by vaccinia topoisomerase I mutants using E. coli with dinD1::lacZ or red::lacZ (28) fusions has identified two mutants, K167D and G226N. The lac2 fusion strains expressing these mutant enzymes formed smaller and blue colonies on indicator plates containing X-gal,' a chromogenic substrate of P-galactosidase. Cells expressing the wild type vaccinia topoisomerase I formed larger and white colonies. I n vitro characterization of DNA relaxation and partial enzyme activities confirmed that these mutations have affected the DNA cleavage-religation equilibrium and stabilized the enzyme cleavage complex.

EXPERIMENTAL PROCEDURES
Materials-DNA oligonucleotides used as primers for site-directed mutagenesis were synthesized in an Applied Biosystems synthesizer and used without purification. Oligonucleotide substrates for the enzyme were purified by high performance liquid chromatography. Plasmid pKS-top was constructed by cloning the BumHI-PstI fragment containing the entire vaccinia topoisomerase I coding sequence from p1940 (19) into the corresponding sites in phagemid pKS+ from Stratagene.

573
Negatively supercoiled plasmid DNA used in relaxation assay was purified by alkaline lysis (29) and CsCl centrifugation. E. coli strains GE94 (28) and JH140 (26) were kindly provided by Dr. G. M. Weinstock of the University of Texas Medical School, Houston, and Dr. J. Heitman of Duke University, respectively.
Site-directed Mutagenesis-The Muta-Gene oligonucleotide-directed in vitro mutagenesis kit from Bio-Rad was used. Single-stranded phagemid pKVT DNA containing part of the vaccinia topoisomerase coding sequence (20) was used as template. The G226N mutagenesis primer has the sequence 5'-CGTATAATTGACATTATACGTTCGGAGA- 3'. The K167D mutagenesis primer has the sequence 5"AC'MTGT-CATCTCCTACAA-3'. Mutants were identified by direct DNA sequencing of the double-stranded phagemid DNA isolated from transformants of E. coli strain MV1190 included in the kit.
Expression of Mutant Topoisomerases-Mutant vaccinia topoisomerase I sequence was excised in a NsiI-BamHI restriction fragment from pKVT (20) and ligated to the largest fragment from NsiI-BarnHI digestion of the T7 promoter expression plasmid pA9topo (19). The noncoding region between the BamHI restriction sites of pA9topo is absent in the resulting clone. That has no apparent effect on enzyme expression. The induction of expression and enzyme purification were carried out as described previously (19). Wild type and mutant vaccinia topoisomerase I preparations appeared to be homogeneous in SDS-polyacrylamide gel electrophoresis analysis. Protein concentrations were determined by the Bio-Rad protein assay.
Relaxation Assay-0. Oligonucleotide Cleavage Assay-The cleavage reaction (20 pl) contained 50 m~ Tris-HCI, pH 7.5, 12.5 m~ NaCl, 200 fmol of JP6/JP12 oligonucleotide hybrid, and various amount of wild type and mutant vaccinia topoisomerase I. Following incubation for 5 min a t 37 "C, the reactions were stopped with 2 p1 of 10% SDS, 18 p1 of water, and 1 pl of 10 mg/ml protease K to proteolyze the covalently bound protein for 60 min at 37 "C. After phenol and chloroform extractions, formamide was added to 50% and the reaction products were heated a t 95 "C for 5 min before electrophoresis in a 20% acrylamide gel containing 7.5 M urea. The DNA substrate and cleavage product bands in the gel were visualized by autoradiography. Using the autoradiogram as template, bands corresponding to uncleaved substrate and cleavage product were excised from the gel, mixed with 10 ml of EcoScint scintillation mixture, and counted for radioactivity.

TATTCAACA'MTCCGTGTCGCCCTTAT-3') labeled with 32P a t the 5'-
Intermolecular Religation Assay-Five pmol of the 27-mer (5'end was hybridized to 5.5 pmol of an unlabeled 50-mer (3"ATAAGT-5'). Two-hundred fifty fmol of the 27-med50-rner hybrid was incubated with topoisomerase in 20 pl of 50 m~ Tris-HC1, pH 7.5, for 5 min a t 37 "C to allow cleavage of the 27-mer a t two bases from the 3'-end. After termination of the cleavage reaction by addition of NaCl to 200 m~, 35 pmol of an unlabeled 25-mer acceptor (5'-ATTCCCTTMTTGCGGC-A'Il'lTGCC-3') was added and further incubated for 5 min. A 50-mer religation product formed from intermolecular religation of the covalently bound DNA sequence to the 5'-hydroxyl end of the acceptor. Reactions were stopped by addition of 2 pl of 10% SDS. After proteinase K treatment for 60 min, samples were phenol-chloroform extracted and adjusted to 50% formamide. Electrophoresis and quantitation of substrate and religation product were carried out as described for the cleavage assay. coli GE94 transformed with plasmid was prepared as described previously (32). 0.5 pg of supercoiled plasmid was incubated with 1 pl of the extract as described under "Experimental Procedures." Lane 1, control with no extract. Lane 2, extract of GE94 transformed with pKS+, grown a t 30 "C for 8 h. Lane 3, extract of GE94 transformed with pKS+, grown a t 30 "C for 3 h, then shifted to 42 "C for 5 h. Lane 4, extract of GE94 transformed with pKS-top, grown a t 30 "C for 8 h. Lane 5, extract of GE94 transformed with pKS-top, grown a t 30 "C for 3 h, then shifted to 42 "C for 5 h. Lanes 6 and 7 correspond to extract used for lane 5, diluted 3-and 10-fold, respectively. MI, supercoiled monomer; MIo, relaxed monomer; DI, supercoiled dimer; DIo, relaxed dimer.

Use of pKS-top for Characterization of Vaccinia Topoisomeruse I Mutants in E. coli--To screen for vaccinia topoisomerase
I mutants that induced the SOS response in vivo in E. coli, we needed an expression system that would produce a significant amount of vaccinia enzyme, but overexpression of vaccinia topoisomerase I had to be avoided because that would limit the viability of the host (31,32). We also wanted to avoid SOS induction from overexpression of the wild type enzyme. It was shown that induction of expression of vaccinia topoisomerase I from the very strong T7 promoter caused arrest of growth and recA-dependent lysogenic induction in a A-lysogen (31). That level of overexpression was therefore not suitable for the SOSinduction assay we have planned to use for identification of mutants deficient in DNA religation. We demonstrated previously (32) that in plasmid p1940 (20), a low level of vaccinia topoisomerase I was expressed at 30 "C, with no apparent ill effect on cell growth. Topoisomerase expression could be induced by shift to 42 "C, and cells would have decreased p1940 copy number and lose viability (32). We constructed a pKS-top phagemid as an expression plasmid for vaccinia topoisomerase I that could be obtained in the single-stranded form. The 5'flanking sequence for the vaccinia topoisomerase I gene was identical to that found in p1940. We confirmed the expression of vaccinia topoisomerase I DNA by assaying for relaxation activity in cell lysate of E. coli GE94 (Fig. 1) and JH140 (data not shown) transformed with pKS-top. The reaction mixture contained 2.5 mM EDTA and no added divalent ions so the bacterial topoisomerase activities were not active under these conditions and only eukaryotic type I topoisomerase activity could convert the negatively supercoiled DNA to more relaxed forms. We found that these transformants expressed a low level of vaccinia topoisomerase I when grown at 30 "C. A higher level was expressed at 42 "C as found in E. coli expressing p1940 (32). SOS induction due to the presence of the wild type vaccinia topoisomerase I was not detectable in GE94 (red-la&) or JH140 (dinD1::lac.Z) colonies after overnight incubation on Xgal indicator plate at 30,37, or 42 "C. These strains were there- fore suitable for screening for expression of mutant vaccinia topoisomerase I with stabilized covalent cleavage complex and resulting SOS induction due to DNA damage response.
Identification of the K167D and G226N Site-directed Mutant as SOS-inducing-We had initially hoped to construct mutants of vaccinia topoisomerase I using single-stranded pKS-top as template. Unfortunately the yield of single-stranded pKS-top was low, possibly due to expression of vaccinia topoisomerase I. A number of site-directed mutants in the conserved region of vaccinia topoisomerase I postulated to be important for DNA cleavage-religation were therefore first generated using singlestranded pKVT phagemid (20) as template. This phagemid lacked the coding sequence for the first 68 amino acids of the enzyme. The mutants were first cloned in the T7 expression plasmid pA9topo for enzyme purification purpose. To determine if these site-directed mutants would induce the SOS response in E. coli, they were cloned via the NsiI-BglII fragment from pA9topo into pKS-top. Ligation using NsiI-BglII fragment wild type pA9topo and other site-directed mutan+-resulted in white transformants of JH140 and GE94 on X-gal indicator plates. In comparison, the ligation using the NsiI-BglII fragment carrying the K167D and G226N mutations yielded a large number of transformants at 30 "C that were blue and had poor growth on X-gal indicator plates. The blue color of the colonies were more intense at 37 and 42 "C, when enzyme expression level was expected to be higher.
Plasmid Copy Number of pKS-top Expressing and K167D a n d G226N Mutant Was Lowered-When 30 "C overnight cultures of GE94 colonies expressing the G226N mutant were used for plasmid preparation by rapid lysis, we found that the yield of the mutant pKS-top plasmid was drastically reduced compared to the yield of wild type pKS-top plasmid from GE94 (Fig. 2). The expression of the G226N mutant enzyme might have resulted in selection of cells with lowered plasmid copy number, probably due to some damaging effect of the mutant enzyme. Just like plasmid p1940 that also expresses wild type vaccinia topoisomerase I(32), the plasmid pKS-top was isolated as a dimer. The mutant plasmid G226N appeared mostly to be monomeric, as seen previously for insertionally mutated forms of p1940 (32). Transformants of pKS-top with the K167D mutation had even lower copy number when compared to the G226N mutant plasmid (data not shown).
Mutant Topoisomerases Had Reduced Relaration A c t i v i t p G226N and K167D mutant enzymes were expressed in E. coli strain BL21 from the T7 promoter in pA9topo upon supply of T7 polymerase by phage A CE6 infection. The yields of mutant vaccinia topoisomerase I in the lysate and after purification to apparent homogeneity were about the same as wild type (data not shown). When assayed for relaxation activity using negatively supercoiled plasmid DNA as substrate (Fig. 3), both mutants were found to be less active than wild type. Quantitation of the results from relaxation assay by densitometry scanning showed that the G226N mutant was -5-10-fold less active than wild type while the K167D mutant was -30-50-fold less active. At least one of the individual steps involved in relaxation of DNA by enzyme was thus being affected by the mutation.
Noncovalent DNA Binding by Mutant lbpoisomerases-Vaccinia topoisomerase I specifically binds to duplex DNA sites with the conserved sequence 5'-(c/T)cc'IT (35). w e use a labeled oligonucleotide duplex containing the CCC'IT sequence to assay this step of enzyme action by measuring the percentage of substrate DNA that was retained on nitrocellulose filters after incubation with the enzyme (35). There was no contribution of covalent intermediate to overall binding measured in this assay as demonstrated by previous results from the mutant with the active site tyrosine 274 converted to a phenylalanine, eliminating the covalent binding activity (36). The results of our experiments (Fig. 4) showed that the noncovalent binding afinity (determined by concentration needed for half-maximal binding) of the G226N mutant was nearly identical to that of the wild type. "he K167D mutant had a binding affinity that was about 10-fold lower than wild type. The amount of maximal binding compared to wild type was 83% for the G226N mutant and 72% for the K167D mutant.
Cleavage of DNA by Mutant Topoisomerases-The ability of the mutant topoisomerases to carry out DNA strand cleavage was determined by quantitation of the percentage of substrate that was cleaved using the labeled oligonucleotide duplex substrate with a specific cleavage site. "he results (Fig. 5) for the K167D mutant showed that due to weaker binding, about 10fold higher K167D mutant enzyme was required for half-maximal cleavage compared to the wild type enzyme and the G226N mutant. The maximal percentage of DNA cleavage for the G226N mutant was 99% of that of the wild type. For the K167D mutant, it was 92% of the maximal percentage of DNAcleavage by wild type. These ratios of mutant to wild type activity were significantly higher than those obtained for noncovalent binding. Since a noncovalent complex had to form first before cleavage can occur, these results suggested that the cleavage-religation equilibrium was shifted towards cleavage for the two mutant enzymes. The cleavage specificity of the mutant enzymes was compared to the wild type enzyme by using 3"end-labeled pUC19 DNA(1inearized with BsrF I) as substrate. The pattern of cleavage products generated by the mutant enzymes were analyzed by electrophoresis in the sequencing gel (35). The sites of cleavage were identical to those of the wild type enzyme (data not shown), indicating that these two mutations did not alter the DNA sequence specificity in binding by the enzyme.

Religation by the G226N and K167D Mutants Was Reduced-
To test further if the DNA cleavage-religation equlilibrium was altered for the two mutants, we carried out an intermolecular religation assay. A 5"end-labeled 27-mer (5'-TATTCAACATll" CCGTGTCGCCCTTAT-3') was hybridized to a n unlabeled 50mer (3'-ATAAG'ITGTAAAGGCACAGCGGGAATAAGGGAAA- . The vaccinia topoisomerase I cleaved the 27-mer at two bases from the 3'-end. A 25-mer was then added as acceptor so that religation of the vaccinia topoi-  somerase covalent intermediate to the 5"OH of this acceptor generated a new labeled 50-mer product. Quantitation of the percentage of input substrate that formed the intermolecular religation product (Fig. 6) showed that the maximal amount of ligated product formed by both the G226N and K167D mutant was only 65% of that of the wild type enzyme even though the ratio of maximal cleavage product for both mutants was >90% of that formed by wild type. This provided further support that the equilibrium between cleavable and ligated DNA-protein complexes was skewed towards the cleavable complex for the two mutants. We also followed the time course of religation by the wild type and mutant enzymes. Fig. 7 showed that both mutants had a slower rate of religation compared to the wild type, again supporting our conclusion that the mutant enzymes had impaired DNA religation.

NaCl Promoted Closure of the Wild IJpe Cleavable Complex More Readily Than the G226N and K167D Mutant Cleavable
Complex-It has been shown previously (35) that addition of NaCl to cleavable complex of vaccinia topoisomerase I formed at a low salt concentration resulted in closure of the cleavable complex from religation of the nicked DNA strand. We compared the salt stability of the cleavable complexes formed by the G226N and K167D mutants to that of the wild type by first forming the cleavable complex in the presence of 50 mM NaCl. After 5 min, NaCl was added to the reactions to achieve different final concentrations. After 5 min of further incubation, the reactions were terminated and analyzed for residual cleavage product as a percentage of the amount of cleavage product obtained at 50 mM NaCl. The results shown in Fig. 8 demonstrated that the cleavable complex formed by the G226N and K167D mutants were both more salt stable than that of the wild type. Closure of the nicked strand upon addition of NaCl occurred less readily in these mutant's enzyme. DISCUSSION We have characterized the effect of two single mutations, K167D and G226N, on the activities of vaccinia DNA topoisomerase I . We are interested in these mutants because when these mutant topoisomerases were expressed in E. coli, the SOS response was induced. They were much less active (<20%) than the wild type enzyme when assayed for relaxation of negatively supercoiled plasmid DNA. When expressed from the same plasmid vector, the wild type enzyme did not induce the SOS response. Therefore it was unlikely that the SOS response was induced by the effect of relaxing activity of the mutant enzymes. We therefore characterized their partial activities that represent the different steps of catalysis. Our initial hypothesis in the screening of these SOS-inducing mutants was that the mutations we would obtain through this screening process might be mimicking the action of the drugs that stabilize the cleavable complex of topoisomerases, causing increased cleavage of DNA in viuo. The results from the enzymatic assays we carried out on the two SOS-inducing mutant enzymes, G226N and K167D, have confirmed this hypothesis. After measurement of DNA binding, cleavage, and religation by the G226N mutant enzyme, the only partial activity that was reduced compared to the wild type enzyme mutant was in DNA religation. This mutation is thus unlikely to affect folding of the enzyme to a large extent. We do not have an assay that can measure rate of strand passage or strand swiveling after initial DNA nicking by enzyme. We therefore do not know if that plays a part in the overall 5-fold reduction of relaxing activity in the G226N mutant. Besides affecting DNA religation, the K167D mutation also reduced significantly the binding affinity of the enzyme for the DNA substrate. We postulate that the lower DNA binding affinity of the K167D mutant may be due to loss of a direct interaction which is required for substrate affinity but not for DNA sequence recognition. This mutant still maintained the same maximal DNA cleavage activity and sequence specificity so it is unlikely to have a drastic change in protein folding. FIG. 6. Religation by wild type and mutant topohmerases. The religation assay was carried out as described under "Experimental Procedures." The percentage of input DNA that formed the intermolecular religated product (average of data from three separate experiments) was plotted here. have about 26% homology to the corresponding region of Saccharomyces cerevisiae, s. pombe, and human enzymes (33,371. Many lines of evidence have linked this domain to DNA cleavage-religation activity of the enzyme even though it does not contain the active site Ty?74 (34,38). Conversion of Aspzz1 to Val reduced the cleavage activity of the enzyme, with the religation activity in this mutant becoming sensitive t~ inhibition by camptothecin even though the wild type enzyme was resistant (20). Mutations of Glyx3' to Asp and to Gln rendered the enzyme inert in formation of covalent complex (36). Mutations of Thr147 to Ile and Gly'32 to Ser also caused severe defects in covalent complex formation (36). None of these previously described mutations affected the noncovalent DNA binding significantly. The two mutations we characterized here belonged to a different class even though they are found in the same conserved region. DNA cleavage activity was not affected in these two mutants but DNA religation activity was reduced and stability of the cleavable complex increased. It is possible to suggest different possible mechanisms for these mutations to affect the cleavage-religation activity of vac- , 50 nm NaCl. 6.7 pmol of K167D mutant enzyme and 0.65 pmol of wild type or G226N mutant enzyme were used in the individual reaction so that approximately the same amount of cleavable complex would be formed at 50 nm NaCl. After incubation at 37 "C for 5 min, different amounts of 5 M NaCl solution were added to the reaction mixture to achieve the final NaCl concentration indicated. After incubation for an additional 5 min, the reactions were stopped and the amount of cleavage product and substrate was determined as described under "Experimental Procedures." The cleavage ratio was first calculated as the percentage of input substrate that then divided by the percentage of cleavage seen at 50 n m NaCl to obtain remained as cleavage product at each final NaCl concentration and the percentage of residual cleavage shown here. cinia topoisomerase I. The tyrosine hydroxyl acts as a nucleophile in phosphodiester bond cleavage during the cleaved complex formation. During the religation step, the 5"hydroxyl on the cleaved DNA is now the nucleophile that acts to displace the phosphotyrosine linkage. Basic residues on the enzyme might assist in these catalytic steps by withdrawing a proton from the hydroxyl groups to increase their strength as nucleophiles, similar to the general acid-base catalysis mechanism proposed for DNase I (39). These basic residues if mutated, would affect the corresponding cleavage or religation activity of the enzyme severely. Hydrogen bonding or non-ionic interactions between the enzyme and DNA are likely to differ to a certain degree when the noncovalent enzyme-DNA complex undergoes the transition to cleaved complex. Basic residues such as Lys167 can also participate directly in binding to DNA by charge-charge interaction with the DNA backbone phosphate. The energy of binding for the two kinds of complexes relative to each other will affect the equilibrium between noncovalent and cleaved complexes. A mutation can affect the energy of protein-DNA complexes by either directly being involved in noncovalent interaction with DNA, especially at the region downstream from the cleavage site where the DNA strand containing the 5'-hydroxyl group is not covalently bound to the enzyme, or through steric and protein folding effects. If a mutation has a relative to the noncovalent complex, it will result in stabilization of the cleaved complex and shift the equilibrium towards DNA cleavage. Alternatively, if the mutation promotes dissociation of the noncovalently bound DNA from the protein after the cleavage complex is formed, the 5'-hydroxyl will then be out of position for interactions required for reforming the phosphodiester linkage. That would also decrease DNA religation and stabilize the cleavage complex. The latter mechanism is more likely for the K167D mutant which has decreased noncovalent binding affinity for DNA. Further results, preferably structural information on the enzyme, are needed to determine how the G226N and K167D mutations reduce DNA religation and stabilize the cleavage complex with DNA.