Drosophila melanogaster topoisomerase IIIalpha preferentially relaxes a positively or negatively supercoiled bubble substrate and is essential during development.

Eukaryotic type IA topoisomerases are important for the normal function of the cell, and in some cases essential for the organism, although their role in DNA metabolism remains to be elucidated. In this study, we cloned Drosophila melanogaster topoisomerase (topo) IIIalpha from an embryonic cDNA library and expressed and purified the protein to >95% homogeneity. This enzyme partially relaxes a hypernegatively supercoiled plasmid substrate consistent with other purified topo IIIs. A novel, covalently closed bubble substrate was prepared for this study, which topo IIIalpha fully relaxed, regardless of the handedness of the supercoils. Experiments with the bubble substrate demonstrate that topo IIIalpha has much different reaction preferences from those obtained by plasmid substrate-based assays. This is presumably due to the fact that solution conditions can affect the structure of plasmid based substrates and therefore their suitability as a substrate. A mutant allele of the Top3alpha gene, Top3alpha191, was isolated through imprecise excision mutagenesis of an existing P-element inserted in the first intron of the gene. Top3alpha191 is recessive lethal, with most of the homozygous individuals surviving to pupation but never emerging to adulthood. Whereas this mutation can be rescued by a Top3alpha transgene, ubiquitous overexpression of D. melanogaster topo IIIbeta cannot rescue this allele.


From the Department of Biochemistry, Duke University Medical Center, Durham, North Carolina 27710
Eukaryotic type IA topoisomerases are important for the normal function of the cell, and in some cases essential for the organism, although their role in DNA metabolism remains to be elucidated. In this study, we cloned Drosophila melanogaster topoisomerase (topo) III␣ from an embryonic cDNA library and expressed and purified the protein to >95% homogeneity. This enzyme partially relaxes a hypernegatively supercoiled plasmid substrate consistent with other purified topo IIIs. A novel, covalently closed bubble substrate was prepared for this study, which topo III␣ fully relaxed, regardless of the handedness of the supercoils. Experiments with the bubble substrate demonstrate that topo III␣ has much different reaction preferences from those obtained by plasmid substrate-based assays. This is presumably due to the fact that solution conditions can affect the structure of plasmid based substrates and therefore their suitability as a substrate. A mutant allele of the Top3␣ gene, Top3␣ 191 , was isolated through imprecise excision mutagenesis of an existing P-element inserted in the first intron of the gene. Top3␣ 191 is recessive lethal, with most of the homozygous individuals surviving to pupation but never emerging to adulthood. Whereas this mutation can be rescued by a Top3␣ transgene, ubiquitous overexpression of D. melanogaster topo III␤ cannot rescue this allele.
DNA topoisomerases are ubiquitous enzymes found in all cells and some viruses that regulate the topology of DNA in such cellular processes as replication, transcription, and recombination (1,2). These enzymes work by making a transient covalent bond to the phosphodiester backbone of the DNA, creating a protein mediated DNA gate, which allows another strand (or strands) of DNA to pass through this reversible break. Type II topoisomerases create double-stranded breaks, and double strand passage, whereas type I topoisomerases break a single strand of DNA. The type I topoisomerases are further divided into the IA and IB subfamilies, based on structural and mechanistic differences (3,4). Topoisomerase III (topo III) 1 is a member of the type IA subfamily that is conserved from bacteria to humans.
Topo III was originally purified by following superhelical relaxation activity from extracts of Escherichia coli containing a deletion for topo I (5). This enzyme was also independently purified based on its decatenating abilities in an assay utilizing plasmid DNA replication intermediates (6). Despite the strong sequence homology that this protein shares with E. coli topo I (7), purified topo III shows a weak ability to relax negatively supercoiled DNA (5). The relaxation activity of Topo III is strongly inhibited by single-stranded DNA (ssDNA) (5), and the enzyme was subsequently shown to preferentially bind ssDNA. The decatenase activity of topo III also depends on the presence of single-stranded regions in the catenated DNA (6). Recently, Nurse et al. showed that topo III serves as a decatenase in vivo by removing precatenanes that arise during replication of circular DNAs (8). These results, combined with others, have lead to the thought that topo III is responsible for decatenating replication and possibly recombination intermediates.
In Saccharomyces cerevisiae, TOP3 was discovered in a screen for mutants that enhanced the deletion of a marker flanked by repeated DNA sequences (9). Subsequently, it was shown that TOP3 deletions result in hyperrecombination, impaired sporulation, and a slow growth phenotype (9). Some of these phenotypes can be suppressed by mutations in SGS1, a member of the RecQ helicase family (10). This interaction between topo III and RecQ helicase proteins has since been shown to be a physical interaction that is conserved through evolution into higher eukaryotes (11)(12)(13)(14)(15)(16).
In larger eukaryotes, there are two isoforms of topo III, ␣ and ␤ (17)(18)(19)(20)(21), differing primarily in their long carboxyl-terminal sequences. These two isoforms show very high homology with each other, as well as bacterial topo I, topo III, and yeast topo III, through the first 600 amino acids of the sequence, hereafter referred to as the type IA core region. This region contains the four type IA topoisomerase domains along with the active site tyrosine (4). Although both the ␣ and the ␤ carboxyl terminal regions contain several potential zinc finger motifs, there is very little homology between the two. The carboxyl terminus of E. coli topo I shows some similarity to the ␣ isozymes and has been shown to possess five zinc ribbon domains with the first three binding zinc (22)(23)(24). This region in E. coli topo I binds ssDNA (25,26) as well as interacting with RNA polymerase II (27), suggesting that these regions in the ␣ and ␤ isozymes may be multifunctional as well.
Topo III␣ has been shown to be essential for embryogenesis in the mouse (28), whereas topo III␤ deletions survive to adulthood, although they show a predisposition to tumors, a shortened life span, and reduced litter size (29,30). Biochemically, the ␣ and ␤ isozymes purified thus far have shown relaxation activity consistent with the E. coli and yeast enzymes (17,18,21,31,32). Relaxation of negatively or hypernegatively supercoiled substrates normally occurs in conditions that would favor destabilization of the DNA helix, such as elevated temperatures, high glycerol concentrations, and low monovalent and divalent salt concentrations (17,18,21,31,32). SGS1 also has homologs in mice and humans: RECQ1, RECQ4, RECQ5, WRN, and BLM (33)(34)(35)(36). Mutations in RECQ4, WRN, and BLM are associated with Rothmund-Thompson, Werner's, and Bloom's syndromes, respectively (37)(38)(39). These syndromes all display genomic instability with a predisposition to cancer, whereas Werner's and Bloom's syndromes also show severely advanced aging. Several of the RecQ helicases physically interact with topo III␣ (11,13,14), with the interaction with BLM somewhat activating the relaxation activity of topo III␣ (40). Recently it was shown that human BLM and topo III␣ can separate oligonucleotides joined by a very short (15-bp) double Holliday junction, an intermediate of homologous recombination (41).
We have identified topo III␣ in Drosophila melanogaster and report here this protein's preferences for relaxation of a hypernegatively supercoiled plasmid substrate. We also prepared a new covalently closed DNA substrate that contains a singlestranded bubble, and we compared the activity of Dmtopo III␣ on this substrate with that on the hypernegatively supercoiled plasmid substrate. The importance of topo III␣ to fruit fly development was assessed through the isolation and study of Top3␣ 191 , a recessive lethal allele.

EXPERIMENTAL PROCEDURES
Cloning the D. melanogaster Topoisomerase III␣ Gene-The Dm-Top3␣ gene was discovered in the D. melanogaster genomic DNA data base via homology searches. Oligonucleotides were designed to amplify a highly conserved region of the type IA core region of DmTop3␣. This PCR product was used as a probe to screen a D. melanogaster embryonic cDNA library, which was prepared in our laboratory. A clone with a 3.9-kb insert that corresponded to the DmTop3␣ cDNA was isolated and sequenced (deposited in GenBank TM with accession number AF255733).
Rescue of the Slow Growth Phenotype of a ⌬top3 S. cerevisiae Strain with DmTop3␣-The Top3␣ coding region was PCR-amplified from the above clone with a 5Ј primer that included a Met-His 6 coding sequence before the start codon. The product was digested and ligated into YEpG, placing it under the control of a galactose-inducible promoter, to make YEpGTop3␣. Once the sequence was confirmed, YEpG and YEpGTop3␣ were transformed into JCW253, a ⌬top3 derivative of FY251, using standard protocols. Transformants and the parental strains were grown in selective media overnight, and culture densities were adjusted to 1 ϫ 10 7 cells/ml. 10-fold serial dilutions were made, and 1 l of each was spotted onto nonselective plates containing either dextrose or galactose as the exclusive sugar source. The plates were incubated at 30°C until the colonies were readily detectable (2-3 days). Purification of topo III␣ from yeast transformed with YEpGTop3␣ was explored, but expression levels were too low to make it feasible for biochemical studies.
Preparation of an Anti-topo III␣ Antibody-Peptides TopoIII␣-1 (GI-IQSIFQCPKCNEAPLALK) and TopoIII␣-2 (YEUUDVCRSIKPNIS-VYRAT) were synthesized using the multiple antigen peptide method (42) and injected into a rabbit using standard protocols. The peptides were covalently linked to Affi-Gel 15 (Bio-Rad) amine reactive crosslinking resin using the anhydrous method provided with the resin. This column was used to affinity-purify the polyclonal antibody anti-To-poIII␣-1 from the rabbit's serum using standard methods.
D. melanogaster Topo III␣ Expression and Purification-Topo III␣ was overexpressed using the Bac-to-Bac baculovirus expression system (Invitrogen) in Sf9 cells and purified using a cleavable N-terminal GST tag and a C-terminal decahistidine tag. The GST tag was PCR-amplified from pGEX-6P-1 (Amersham Biosciences) using oligonucleotides that amplified from the start codon of GST to just after the recognition sequence for PreScission Protease (Amersham Biosciences). DmTop3␣ was PCR-amplified from its start codon to its C terminus, excluding the stop codon. The decahistidine tag followed by a stop codon was synthesized as two complementary oligonucleotides, and the three pieces were sequentially cloned into pFastBac1 and the sequence was verified to create pFBG-T3␣-H 10 . This plasmid was developed into a baculovirus (bvG-T3␣-H 10 ) using the Bac-to-Bac kit protocols. Supernatants of transfections or amplifications of the baculovirus were assayed by dot blot using an anti-gp (64) monoclonal antibody (eBioscience) to detect the presence of baculovirus. Baculovirus titer was estimated by comparison of cleared supernatants with baculovirus of known titer.
Four liters of Sf9 cells were infected at a concentration of 1 ϫ 10 6 cells/ml using an empirically determined amount of bvG-T3␣-H 10 (multiplicity of infection ϳ2). The cells were harvested at 65 h postinfection and spun down at 550 ϫ g for 10 min. All of the following steps were performed on ice or at 4°C, and all buffers were prechilled and con- The nuclei were then pelleted at 1500 ϫ g for 15 min, and the supernatant was decanted off of the nuclear pellet. The nuclear pellet was resuspended in 50 ml of nuclear resuspension buffer (20 mM Tris, pH 8.0, 1.5 mM MgCl 2 , 20 mM KCl, 25% glycerol). 50 ml of nuclear extraction buffer (80 mM Tris, pH 7.6, 2 M NaCl, 20% glycerol, 0.06% Triton X-100) was slowly added to the resuspended nuclei while stirring. 50 ml of DNA precipitation buffer (18% polyethylene glycol 8000, 1 M NaCl) was slowly added with stirring, and the solution was incubated for 30 min on ice with stirring. The insoluble material was pelleted at 20,000 ϫ g for 20 min. The cleared supernatant was added to a prechilled bottle with 2 ml of glutathione resin (Amersham Biosciences). The GST-topo III␣-H 10 was allowed to batch bind at 4°C for 1 h with gentle stirring.
The glutathione resin with bound GST-topo III␣-H 10 was gently pelleted and washed twice in batch with 20 ml of glutathione wash buffer I (50 mM Tris, pH 7.0, 1 M NaCl, 10% glycerol, 0.02% Triton X-100). The resin was then washed once in batch with 20 ml of glutathione wash buffer II (glutathione wash buffer I at 150 mM NaCl) and then resuspended in 2 ml of glutathione wash buffer II. 30 l of Pre-Scission Protease was added to the slurry, and the GST tag was digested off of the topo III␣-H 10 for 4 h at 4°C. Following digestion, the slurry was loaded into a column, and the flow-through was collected. The resin was washed once with 2 ml of glutathione wash buffer II and then with 4 ml of GST eluate dilution buffer (glutathione wash buffer II at pH 8.8). The final pH of the collected flow-through and washes is ϳ7.9, allowing it to be loaded directly onto equilibrated TALON Co 2ϩ IMAC resin (Clontech). This column was washed with six volumes of IMAC wash buffer (glutathione wash buffer II at pH 7.9) and then eluted with IMAC elution buffer (IMAC wash buffer with 400 mM imidazole). Peak fractions were pooled, and bovine serum albumin (New England Biolabs) was added to 0.25 mg/ml to help stabilize the topo III␣. A mock was prepared at this time using bovine serum albumin and IMAC elution buffer. The topo III␣ and the mock were then dialyzed against storage buffer (20 mM Tris, pH 7.6, 150 mM NaCl, 50% Glycerol, 0.5 mM dithiothreitol) and stored at Ϫ20°C. Topo III␣ concentrations were determined using densitometry of stained SDS-polyacrylamide gels using ␤-galactosidase (EIA grade; Roche Applied Science) as a reference.
Immunoprecipitation of Endogenous Topo III␣ from S2 Cells-D. melanogaster S2 cells were grown in suspension using standard procedures. 3 ϫ 10 9 S2 cells in midlog phase growth were pelleted at 550 ϫ g for 10 min, and the media were discarded. The cell pellet was resuspended in 20 ml of ice-cold hypotonic lysis buffer (10 mM HEPES, pH 8.0, 10 mM KCl, 0.5 mM EDTA, 0.5% Nonidet P-40, 5 mM 2-mercaptoethanol, and a protease inhibitor mixture consisting of 0.1 M 4-(2aminoethyl)-benzenesulfonyl fluoride, 80 nM aprotinin, 45 M leupeptin, 3.6 M bestatin, 1.5 M pepstatin A, and 1.4 M E-64) and incubated on ice for 15 min, and then the solution was homogenized gently with a Dounce homogenizer. NaCl was added to 350 mM with stirring, the solution was incubated on ice for 15 min, and the insoluble cell debris was pelleted at 20,000 ϫ g for 20 min. Topo III␣ was immunoprecipitated from 1 ml of this cleared supernatant with 10 l of protein A-agarose (Roche Applied Science) and 7 g of anti-TopoIII␣-1 using standard procedures.
Bubble substrate was prepared by annealing and linking the plusand minus-strands of pBlueScript SK with ADP and Archaeoglobus fulgidus reverse gyrase, purified according to a published procedure (43). When reverse gyrase is used with ADP as the cofactor, the enzyme relaxes a circular DNA rather than positively supercoiling it as it does with an ATP cofactor. Plasmids pBlueScript SKϩ and pBlueScript SKϪ (Stratagene) were transformed into E. coli XL-2 MRFЈ cells (Stratagene) and then infected with the helper phage M13K07 (New England Biolabs) according to the manufacturer's instructions. The phagemids were purified according to Lin et al. (44), phenol/chloroform-extracted, and dialyzed against TE (10 mM Tris, pH 7.9, 0.1 mM EDTA). Equal amounts of the plus-and minus-strands were mixed together in a reaction containing 10 mM Tris, pH 7.9, 50 mM KCl, 10 mM MgCl 2 , 0.1 mM EDTA, 0.005% gelatin, 1 mM ADP, 50 g/ml reverse gyrase, and 80 g/ml total DNA. This reaction was incubated at 80°C for 20 min, and then stopped with the addition of SDS to 1% and EDTA to 10 mM. The reaction was then extracted with phenol/chloroform and precipitated with ethanol, and the DNA was redissolved in TE. The bubble substrate was then negatively supercoiled by using the same procedure used to make hypernegatively supercoiled plasmid substrate, except with half the amount of ethidium bromide.
Positively supercoiled substrates were generated either by incubating relaxed substrates in the presence of ethidium bromide or by positively supercoiling the substrate with A. fulgidus reverse gyrase, using the reaction conditions given above except with 1 mM ATP as the cofactor.
Topo III␣ Assays-Unless otherwise indicated, topo III␣ reactions contained 40 mM HEPES, pH 7.5, 50 mM sodium acetate, 0.1 mM EDTA, 0.125 mM MgCl 2 , 1 g/l bovine serum albumin, 0.2 g/l substrate, and 10 ng/l topo III␣ for plasmid substrate reactions. Bubble substrate reactions were the same except with 2 mM MgCl 2 and 200 mM sodium acetate. The reactions were set up on ice by adding all of the components except the MgCl 2 and the substrate. The reactions were started by adding MgCl 2 and substrate and shifting the reaction to 37°C. Reactions were incubated for 15 min unless indicated otherwise and then stopped with the addition of NaCl to 500 mM. After 5 min at 37°C, DNA loading buffer (with SDS and EDTA) was added, and the reaction was diluted out 5-fold. Any reactions including ethidium bromide to positively supercoil the DNA, were stopped as described, extracted with phenol/chloroform, precipitated with ethanol, and processed for loading onto an agarose gel. After electrophoresis, gels were then stained or destained as was appropriate and imaged with a UVP Bio-Chemi System and software.
Isolation of Top3␣ 191 -Top3␣ EP2272 is an allele of Top3␣ in which a P-element is inserted 7 bases downstream of the first intron/exon junction. This allele is homozygous viable and does not have a detectable difference in the amount of topo III␣ produced. Top3␣ 191 was created from this less severe allele via imprecise excision of the P-element inserted in the first intron of Top3␣ (45). The offspring were screened for the loss of the w ϩmC marker on the P-element. These white-eyed flies were then checked for the recessive lethality that we expected from a stronger allele of Top3␣. The recessive lethal stocks were then tested for complementation with an Hsp70-promoted Top3␣ cDNA transgene. Stock 191 was rescued by the transgene, and the Top3␣ 191 allele was sequenced.
Creation of the HSP 3 Top3␣ and HSP 3 Top3␤ Transgenic Lines-The coding sequences of Top3␣ and Top3␤ were PCR-amplified from the appropriate cDNA clones. The cDNAs were then cloned into the multiple cloning site of pP{CaSpeR-hs/act}, placing them under the control of the Hsp70 heat shock-inducible promoter with the 3Ј-untranslated region of Actin5C. The P-elements were injected into w 1118 flies, and transgenic lines were isolated and mapped using standard protocols.
Time of Death for Top3a 191 -Top3␣ 191 was placed over the balancer chromosome CyO, P{w ϩmC , GAL4-twi.G}2.2, P{UAS-2xEGFP}AH2.2, which marks larvae carrying the balancer chromosome with green fluorescent protein, thus differentiating heterozygotes from Top3a 191 homozygotes. Note that the genotype of CyO/CyO dies as an embryo, so all of the fluorescent larvae are of the genotype Top3␣ 191 /CyO. This stock was crossed inter se, and the embryos were collected for 4-h intervals on grape juice plates. Approximately 36 h later, the plates were viewed with a fluorescent dissecting microscope, and the fluorescent and nonfluorescent larvae were counted and placed into separate vials with standard fly food. The vials were monitored daily and scored for pupation and eclosion of these two genotypes.

RESULTS AND DISCUSSION
Cloning of D. melanogaster Topoisomerase III␣-To isolate and study the Top3␣ gene from D. melanogaster, primers were designed from the genomic sequence to amplify a region from the highly conserved type IA core region of the gene, and this PCR product was used to screen a D. melanogaster embryonic cDNA library. A clone was isolated that contained a 3.9-kb insert. This cDNA was sequenced, and it encoded a short 5Ј untranslated region, an open reading frame, and a short 3Јuntranslated region with polyadenylation signal. The genomic structure of the gene is shown in Fig. 1A. The Top3␣ gene is located on the left arm of chromosome 2 at the cytological location 34E. The gene is very compact, containing three small introns, each of which is less than 60 bp in length, a short promoter region, and a short intergenic sequence to genes neighboring either its 5Ј-end or 3Ј-end. The largest open reading frame of the cDNA encodes a 1250-amino acid protein with high overall similarity to human and mouse topo III␣.
Like the human and mouse homologs, D. melanogaster topo III␣ has a short, unconserved region at the N terminus of the protein, which contains a putative mitochondrial localization sequence (46). Analysis using the algorithm developed by Claros and Vincens (47) gives this sequence an 88% probability of mitochondrial import, with a signal cleavage site following the import sequence. Consistent with the human and mouse homologs, a second methionine follows this region, and a nuclear localization signal lies in the remaining C-terminal portion of the protein. It is unclear at this time if this second methionine is used as an alternative start codon to create two populations of topo III␣, one localized to the mitochondria and the other to the nucleus, although human topo III␣ has been shown to be targeted to two different compartments in this manner (46).
The type IA topoisomerase core region follows the potential mitochondrial localization sequence. This region shows high overall similarity to other type IA topoisomerases and is homologous to domains I-IV of E. coli topo I and topo III (4,48), including the active site tyrosine at amino acid 356 (Fig. 1B). After the type IA core region, the topo III␣ signature C terminus begins.
The C-terminal sequence of D. melanogaster topo III␣ is conserved among C-terminal regions of various metazoans, with the exception of a large (260-amino acid) region "inserted" about one-third of the way into the typical C terminus. A detailed sequence alignment is shown in Supplemental Fig. 1. The C-terminal domain is divided into four segments (A-D) based on sequence comparison (Fig. 1B). Segment A of the D. melanogaster C-terminal region contains the first one-third of the typical topo III␣ C terminus, with two of the conserved putative zinc finger motifs, whereas segment D is similar to the latter two thirds, containing the remaining two putative zinc fingers and putative zinc knuckle. Segments B and C compose the inserted sequence that is unique to D. melanogaster topo III␣, with segment B being a glycine-rich region and segment C showing little homology to any proteins of known function.
D. melanogaster Topoisomerase III␣ Can Partially Rescue the Slow Growth Phenotype of Yeast ⌬top3 Mutant-Once the cDNA clone was obtained, we used a heterologous expression system to test the biological function of D. melanogaster Top3␣. YEpGTop3␣ was constructed from the YEpG vector, in which DmTop3␣ is placed under the control of a galactose-inducible promoter (GAL1). YEpGTop3␣ was tested for its ability to rescue the slow growth phenotype of the S. cerevisiae strain JCW253, a ⌬top3 derivative of FY251. Fig. 1C illustrates the slow growth phenotype of JCW253 when compared with FY251 grown on either dextrose or galactose. The dextrose plate shows that transforming JCW253 with YEpGTop3␣ or empty vector does not change its slow growth phenotype. When these yeast strains are grown on galactose to activate the expression of DmTop3␣, JCW253 with YEpGTop3␣ shows an accelerated growth rate when compared with either JCW253 or JCW253 with empty vector. Although the D. melanogaster topo III␣ does not bring the growth rate back to the wild-type level (FY251), it is clear that topo III␣ from the fruit fly has a biological function in yeast. The highly conserved core domain carries the same basic and conserved function across the species, since partial rescue of the slow growth phenotype of ⌬top3 yeast has also been seen for both D. melanogaster Top3␤ and human Top3␤ (17,49).
Expression and Purification of Topoismerase III␣-For further biochemical studies, topo III␣ was expressed and purified using a baculovirus expression system. The Top3␣ cDNA, with a cleavable glutathione S-transferase N-terminal tag and a decahistidine C-terminal tag, was placed under the control of the polyhedron promoter and developed into a baculovirus. This virus was used to infect early log phase Sf9 cells, which were harvested 65 h later. The GST-topo III␣-H 10 was extracted out of the nuclei of these cells and bound in batch to glutathione-Sepharose (Fig. 2A, lane 3). The GST tag was digested off of the GST-topo III␣-H 10 , and the free topo III␣-H 10 was washed off of the resin (Fig. 2A, lane 4). This material was then loaded onto a Co 2ϩ -chelating column (IMAC), the column was washed, and the topo III␣-H 10 was eluted with imidazole (see Fig. 2A, lane 5). The fraction from the IMAC column has a single polypeptide of about 140 kDa, with Ͼ95% purity as judged by SDS-PAGE.
To determine whether the purified protein was of the correct size, we performed a Western blot with an anti-topo III␣ antibody (see "Experimental Procedures") comparing the purified protein with whole cell extract from D. melanogaster S2 cells. However, the endogenous levels of topo III␣ in S2 cells are too low to be detected in this manner (data not shown). The endogenous topo III␣ was further concentrated by immunoprecipitation from S2 extracts with the anti-topo III␣ antibody. Fig. 2B shows that the purified protein does co-migrate with the major species from the immunoprecipitate. At this time, it is unknown whether the slower migrating species apparent in the immunoprecipitate is topo III␣ modified in some manner or if it is simply a cross-reacting protein.
Topoisomerase III␣ Partially Relaxes a HNSC, but Not a Positively Supercoiled (PSC), Plasmid Substrate-We assayed the purified topo III␣ in vitro for relaxation activity on a HNSC plasmid substrate. The reaction products were analyzed by gel electrophoresis in the presence of 0.5 g/ml ethidium bromide. The ethidium bromide intercalates into the DNA, causing it to become more positively supercoiled without changing the linking number of the plasmid. The HNSC plasmid substrate in this electrophoretic condition was nearly relaxed and could be resolved into topoisomers. NSC plasmid will become positively supercoiled in the presence of the ethidium bromide and migrate faster than the HNSC plasmid substrate, and relaxed plasmid will become highly positively supercoiled and migrate the fastest of all the species. Fig. 2C shows that the purified protein possesses partial relaxation activity on this substrate, relaxing it from a superhelical density of about Ϫ0.12 to about Ϫ0.07, which is slightly more negatively supercoiled than plasmid DNA isolated from bacteria. Incubating these reactions for a longer period of time will partially relax all of the substrate, but it does not noticeably change the superhelical density of the product (data not shown). These results are similar to those obtained with human topo III␣, human topo III␤, and D. melanogaster topo III␤ (17, 18, 31).  (dark gray boxes). B, the protein organization of D. melanogaster topoisomerase III␣. The unconserved N terminus with the putative mitochondrial localization sequence is indicated in white. The type IA core region is indicated in light gray, with the position of the putative active site tyrosine noted (Y). The C terminus is divided into four segments, A-D. Segments A and D (dark gray) are homologous to the typical topo III␣ C terminus and contain the five conserved putative zinc finger motifs. Segment B (diagonal stripes) is the glycine-, serine-, proline-rich region, and segment C (white) is unconserved sequence that shows little similarity to any proteins of known function. C, D. melanogaster topo III␣ can partially rescue the slow growth phenotype of a ⌬top3 yeast strain. Yeast strains were grown overnight in selective media and diluted to 1 ϫ 10 7 cells/ml. 10-fold serial dilutions were made, and 1 l of each was spotted onto nonselective plates containing either dextrose or galactose as the sole sugar source.
To test the activity of topo III␣ on a PSC plasmid substrate, relaxed plasmid was incubated with ethidium bromide to shift the substrate to a positively supercoiled state and then reacted with topo III␣. After the reactions were stopped, they were phenol-extracted to remove the ethidium bromide. If the topo III␣ is able to relax the positive supercoils, then the plasmid will become negatively supercoiled upon the removal of the ethidium bromide. This is similar to the reaction that is performed with D. melanogaster topo I to create the HNSC plasmid substrate. Fig. 2D shows that there was no difference between the topo III␣ and the mock reaction, whereas topo I relaxed the positive supercoils. This experiment was repeated using a PSC plasmid substrate that was generated by reverse gyrase with the same results (data not shown). The topo III␣ was not able to relax the PSC plasmid substrate, which is consistent with the idea that topo III␣ requires single-stranded regions upon which to function.
Preparation of a Bubble Substrate-Whereas the HNSC plasmid substrate provides a convenient assay for topo III activity, it is potentially only a topo III substrate under certain limited conditions. Topo III may function on this substrate due to the presence of unwound DNA regions induced by supercoiling. Solution conditions can help stabilize or destabilize a DNA helix, however, so the optimized reaction conditions determined for a topo III protein with this substrate are likely to be the best compromise between the preference of the enzyme and the effect of the reaction conditions on the DNA structure.
The solution to this problem would be to use a covalently closed DNA circle with single-stranded regions regardless of the solution conditions. A substrate had previously been made that was a covalently closed circle with an extraneous, singlestranded bubble (50). Whereas this substrate satisfied our requirement, it was difficult to make in large quantities, limiting the number of assays that could be performed and necessitating detection with autoradiography. To address this issue, we endeavored to make a new substrate that would be a covalently closed circular DNA with a single-stranded "bubble" region in large enough quantities to make routine analysis feasible.
We created this substrate by annealing and linking purified single-stranded circles that were created by a helper phage from pBlueScript SKϩ and pBlueScript SK- (Fig. 3A). Because these two plasmids are exactly the same except for the orientation of the f1 origin, the resulting single-stranded circles are complementary except for the f1 origin. In this region, both of the circles have the exact same sequence rather than complementary ones, so once the two circles are annealed and linked together by a topoisomerase, the f1 region will remain a singlestranded bubble. Hyperthermophilic reverse gyrase with ADP was used to anneal and link the two purified single-stranded circles in one step. When the single-stranded circles (Fig. 3B, lane 1) are incubated at 80°C in a mock reaction, the two complementary DNA sequences anneal (Fig. 3B, lane 2). Reaction of the annealed circles with reverse gyrase results in relaxed, covalently closed bubble substrate (Fig. 3B, lane 3), which is then negatively supercoiled with the same technique used to create the HNSC plasmid substrate. It is interesting to note that there is less single-stranded DNA remaining after reaction with reverse gyrase than in the mock reaction (Fig. 3B,  compare lanes 2 and 3). This is presumably due to further annealing and linking of the single-stranded DNA promoted by reverse gyrase.
Topoisomerase III␣ Fully Relaxes the Bubble Substrate, Regardless of the Handedness of the Supercoils-After reacting the NSC bubble substrate with topo III␣, almost all of the substrate ran at the position of the nicked or covalently closed open circle (Fig. 4A, lane 3). Some topoisomers are present in this lane just below the nicked/open circle position, suggesting the presence of fully relaxed DNA. To confirm this, another aliquot of the same reaction was analyzed by electrophoresis in the presence of ethidium bromide. Covalently closed substrate will become positively supercoiled with the intercalation of the ethidium bromide and migrate much faster than the nicked circle. In this electrophoretic condition, it is apparent that there is no detectable increase in the amount of nicked material when the NSC bubble substrate is reacted with topo III␣, whereas the intensity of the relaxed product band is comparable with the NSC starting material (Fig. 4B). This result confirms that a negatively supercoiled substrate with a singlestranded region can be fully relaxed by topo III␣.
Topo III␣ was also tested for its ability to relax positively supercoiled bubble substrate. Relaxed bubble substrate was reacted in the presence of ethidium bromide to generate positively supercoiled bubble substrate. The PSC bubble substrate was relaxed by topo III␣, generating negative supercoils upon the removal of the ethidium bromide (Fig. 4A, lane 6). This result was repeated, with the same outcome, using PSC bubble substrate created with reverse gyrase (data not shown). These results show that the handedness of the supercoils in the substrate is inconsequential to topo III␣ activity; only the presence of single-stranded DNA matters.
Topoisomerase III␣ Relaxes HNSC Plasmid and NSC Bubble Substrates with Similar Temperature Preferences-We tested the temperature preferences for the relaxation of the two substrates, since temperature plays a role in helix stability. While testing this variable, as well as the ones that follow, we shortened the reaction times in order to stop the reaction before it reached completion, allowing us to detect smaller differences in activity. Consistent with many other topo IIIs, topo III␣ has a temperature preference for the partial relaxation of HNSC plasmid substrate elevated above the physiological temperature of the source organism. Fig. 5A shows that the preferred temperature for this reaction is 45°C. With the NSC bubble substrate, the relaxation activity is maximal between 40 and 45°C (Fig. 5B). This result demonstrates that the temperature preferences determined for the relaxation of plasmid substrates by topo IIIs are not significantly influenced by temperature-induced destabilization of the DNA helix. Fig. 5 also illustrates the difference in the reaction rates between the HNSC plasmid substrate and the NSC bubble substrate. Densitometry analysis of the 45°C reaction products in Fig. 5A shows that about 30% of the HNSC plasmid substrate has been partially relaxed in 6 min. In the same amount of time, all of the NSC bubble substrate has been relaxed at the same incubation temperature (Fig. 5B). This demonstrates that DNA with a denatured, single-stranded region is a preferred substrate for the relaxation activity of D. melanogaster topo III␣.
NSC Bubble Substrate Relaxation Has a Higher Mg 2ϩ and Monovalent Salt Optimum than HNSC Plasmid Substrate Relaxation-We also examined the optimal buffer conditions for topo III␣ relaxation of the NSC bubble substrate and HNSC plasmid substrate. Fig. 6A shows that the relaxation of the HNSC plasmid substrate requires Mg 2ϩ but is exquisitely sensitive to its concentration in the reaction. The optimum Mg 2ϩ concentration is 0.125 mM, and there is very little relaxation of the HNSC plasmid substrate at Mg 2ϩ concentrations higher than 0.5 mM. Notice that the reaction buffer contains 0.1 mM EDTA, which attenuates the free Mg 2ϩ concentration in the reaction. In the reactions using the NSC bubble substrate, for which the assay depends on the preference of the enzyme rather than the effect of the buffer on DNA structure, optimum activity of topo III␣ begins at 0.25 mM Mg 2ϩ (Fig. 6B). Unlike the HNSC plasmid substrate reactions, the NSC bubble substrate reaction is insensitive to Mg 2ϩ concentrations from 0.25 mM to at least 8 mM. This shows that topo III␣ itself is not inhibited by high concentrations of Mg 2ϩ but that Mg 2ϩ is probably stabilizing the HNSC plasmid substrate and eliminating the denatured regions upon which topo III␣ can act.
We next investigated the monovalent salt concentration preferences for the two reactions. Topo III␣ was assayed in the presence of several different salts, and the enzyme has a preference for sodium acetate over sodium chloride, potassium acetate, and potassium chloride (data not shown). The optimum amount of sodium acetate for the relaxation of HNSC plasmid substrate is 50 mM, as judged by the disappearance of the substrate, and the reaction was completely inhibited by 200 mM sodium acetate (Fig. 6C). Also interesting to note, the more salt that is present in the reaction, the more negatively supercoiled the final product is. This is probably due to the Na ϩ stabilizing the underwound DNA, eliminating the singlestranded regions upon which the topo III␣ can work. Once the effects of counterions on DNA structure are removed by using the NSC bubble substrate, the monovalent salt optimum of topo III␣ is 250 -300 mM sodium acetate (Fig. 6D).
These results demonstrate that the ionic strength of the reaction can have a dramatic effect on the preference of topo III␣ for HNSC DNA as a substrate. The reactivity of some of the relevant topo III␣ substrates can be affected by ionic strength. Replication and recombination intermediates that topo III␣ may act upon contain single/double-stranded junctions and could have structures very sensitive to solution ionic strength. For example, synthetic Holliday junctions have also shown that the solution environment can change the shape of this structure (51).
Generation of a Recessive Lethal Top3␣ Allele-We acquired a P-element insertion mutant from the Bloomington stock center, Top3␣ EP2272 , which has an insertion just 7 base pairs after the 5Ј splice site of the first intron (see Fig. 7A). This mutation is homozygous viable, with no detectable change in the levels of topo III␣ in the adult fly or embryo (data not shown). In order to create a more severe allele of Top3␣, we mobilized the existing P-element with a genomic transposase source, and the offspring were screened for imprecise excision events. When a screen such as this is performed, there is a chance that a P-element insertion (mutation) elsewhere in the genome could create a phenotype that is not associated with the gene of interest. To minimize this possibility, only lines that had lost the w ϩmC marker and therefore do not contain a P-element in their genome were selected for further analysis. These lines were then tested for recessive lethality of the manipulated chromosome, assuming that a Top3␣ null mutation in D. melanogaster would be lethal, as it is in mice (28). The recessive lethal lines were then tested for rescue with a transgene of Top3␣ under the control of an HSP70 promoter. The rescue of a line by a Top3␣ transgene would confirm that the mutant phenotype was due to a mutation in Top3␣.
One recessive lethal line, 191, was rescued by transgenic Top3␣, and the Top3␣ 191 allele was sequenced (Fig. 7A). This allele was surprising in that the P-element still remained in the insertion site, albeit with a large internal deletion. The part of the P-element that was excised contained the w ϩmC marker (Fig. 7A), allowing it to pass the first set of criteria of our screen. The 5Ј splice site of the first intron is still intact, although it is possible that the new context of the splice site, as a result of the deletion, might affect the expression or splicing of this mRNA in ways that are different from the parental line Top3␣ EP2272 .
Top3␣ 191 Dies as a Pupa-The Top3␣ 191 allele is maintained as a stock with CyO, a dominant balancer chromosome that contains a wild-type copy of Top3␣. In order to determine when the homozygous Top3␣ 191 /Top3␣ 191 flies die, the homozygous mutant flies had to be separated from their Top3␣ 191 /CyO and CyO/CyO siblings at an early age. To do this, we used a CyO chromosome that expresses green fluorescent protein with a Twist promoter, labeling any embryo, larvae, or pupae carrying this chromosome. Embryos were collected from a Top3␣ 191 / CyO, Twist 3 GFP stock and allowed to hatch into larvae. Because the CyO/CyO embryos do not hatch, the expected ratio between fluorescent (Top3␣ 191 /CyO, Twist 3 GFP) and nonfluorescent (Top3␣ 191 /Top3␣ 191 ) larvae is 2:1. We obtained a ratio of 2.5:1 after scoring 352 larvae using a GFP dissecting microscope, indicating that 20% of Top3␣ 191 /Top3␣ 191 flies die as embryos. Of the homozygous mutant larvae, 85% pupate, which is very similar to the 89% pupation rate for the heterozygous larvae. Although the homozygous mutant larvae do pupate, they do so more slowly than their heterozygous siblings (Fig. 7B). The average time to pupate for the homozygous mutant flies was 9.8 days after hatching, whereas the average for the heterozygous control flies was 8.2 days. After pupation, the homozygous mutant flies develop melanotic bodies, probably due to necrotic tissues, and never eclose, in contrast to their heterozygous siblings (Fig. 7B).
These results indicate that Top3␣ is important for stages of development in which nuclear division is prominent. During the early embryonic phase, the nuclei are rapidly dividing, and a significant proportion of the Top3␣ 191 homozygous embryos are lost. Although topo III␣ is maternally loaded into the egg (data not shown), the amount of topo III␣ loaded from the FIG. 7. Top3␣ 191 is a recessive lethal mutation that dies during pupation. A, the generation of Top3␣ 191 from Top3␣ EP2272 . Top3␣ EP2272 has a P-element (indicated in gray) inserted within the first intron of Top3␣. An imprecise excision screen yielded Top3␣ 191 , in which the denoted portion of the P-element was deleted with the w ϩmC marker (indicated in dark gray), but no other changes were detected. B, Top3␣ 191 is delayed to pupate and never ecloses. heterozygous mothers was insufficient for embryonic development for some of the homozygous mutant embryos. There is little cell division in the larval phase, although DNA endoreplication still occurs. We do not know whether the retarded development of the larvae is due to problems with DNA replication in these polytene cells or if the small amount of DNA replication and cell division of the imaginal disks creates these delays to pupation. Melanotic bodies are visible on some late third instar larvae, indicating that some tissues have already started dying before pupation begins. Once pupated, all homozygous individuals develop melanotic bodies and never complete this phase.
Top3␤ Cannot Rescue Top3␣ 191 -Whereas the generation of a recessive lethal Top3␣ allele in a Top3␤ ϩ background shows that endogenous topo III␤ cannot fulfill an essential role of topo III␣, it is possible that this is simply due to the expression level or the expression pattern of the Top3␤ gene. To address this possibility, we used a heat shock-promoted Top3␤ transgene to attempt rescue of the Top3␣ 191 allele. This particular transgenic construct shows strong expression when checked with Western blot using an anti-topo III␤ antibody (data not shown). Table I shows that the Top3␤ transgene was unable to rescue the homozygous mutant flies, whereas two Top3␣ transgenes rescued the Top3␣ 191 allele with 40 -50% efficiency. The rescued flies were crossed inter se and were maintained as a stock for over 12 generations with no detectable phenotype.
The failure of topo III␤ to rescue the Top3␣ mutant could be attributed to several factors. The most straightforward explanation is that these two proteins may have distinct functions. The C termini of the two isoforms contain a similar structural motif, multiple zinc fingers, but their differences may make the enzymes specific for recognition of different DNA structures or may specify a different set of protein-protein interactions.
It is also possible that the two proteins are functionally redundant, at least enough so for survival, but topo III␤ is simply not localized correctly to take the place of topo III␣ in mutant cells. Topo III␣ has a potential mitochondrial import sequence at the N terminus of the protein, but when topo III␤ is analyzed by the Claros and Vincens (47) algorithm, it is only given a 22% chance of mitochondrial import with no potential signal cleavage site. If there is a role for topo III␣ in resolution of replicating mitochondrial chromosomes and this role is essential, then it would be unlikely that the ubiquitous overexpression of topo III␤ would rescue this phenotype.
Conclusions-Whereas there have been many studies on the genetics and biochemistry of topo III proteins, the important role or roles that this protein plays in DNA metabolism remain unclear. Here, we have found that D. melanogaster Top3␣ is essential during development, as in the mouse. The mouse knockout is lethal at a much earlier stage in development (28), but this organism cannot maternally load the enzyme in the egg to the same degree that D. melanogaster can, perhaps prolonging the development of mutant fruit flies. Future studies utilizing defined knockouts of D. melanogaster Top3␣ and disruption of the maternal loading of the enzyme may prove insightful for determining its importance in these early stages of development. With the genetically malleable D. melanogaster system, it would also be interesting to utilize domain deletions and chimeras of the ␣ and ␤ isozymes in null backgrounds to better define the different roles of these enzymes.
In this paper, we have also presented the expression of D. melanogaster topo III␣ in a eukaryotic system and the purification of the enzyme to near homogeneity. Whereas this enzyme has activity and reaction condition preferences similar to other topo IIIs studied on a HNSC plasmid substrate, the preparation of a covalently closed bubble substrate demonstrated that topo III␣ is more active in physiological conditions when presented with DNA with single-stranded regions. This substrate may appear artificial at first, but the action of other DNA-metabolizing enzymes, including the RecQ family of helicases, may present such DNA to topo III in vivo. It may be possible to create DNA substrates using the techniques presented here that would mimic DNA recombination and replication intermediates more closely than currently available substrates. With these biochemical and genetic tools, the biological function of topoisomerase III will be an active area for future investigation.