Cloning and Characterization of DrosophilaTopoisomerase IIIβ

We cloned cDNA encodingDrosophila DNA topoisomerase III. The top3cDNA encodes an 875-amino acid protein, which is nearly 60% identical to mammalian topoisomerase IIIβ enzymes. Similarity between the Drosophila protein and the topoisomerase IIIβs is particularly striking in the carboxyl-terminal region, where all contain eight highly conserved CXXC motifs not found in other topoisomerase III enzymes. We therefore propose theDrosophila protein is a member of the β-subfamily of topoisomerase III enzymes. The top3β gene is a single-copy gene located at 5 E-F on the X chromosome. P-element insertion into the 5′-untranslated region of this gene affects topoisomerase IIIβ protein levels, but not the overall fertility and viability of the fly. We purified topoisomerase IIIβ to near homogeneity and observed relaxation activity only with a hypernegatively supercoiled substrate, but not with plasmid DNA directly isolated from bacterial cells. Despite this difference in substrate preference, the degree of relaxation of the hypernegatively supercoiled substrate is comparable to relaxation of plasmid DNA by other type I enzymes. Drosophila topoisomerase IIIβ forms a covalent linkage to 5′ DNA phosphoryl groups, and the DNA cleavage reaction prefers single-stranded substrate over double-stranded, suggesting an affinity of this enzyme for DNA with non-double-helical structure.

Topoisomerases play important roles in many biological processes, such as DNA replication, transcription, and chromosome condensation (see Ref. 1 for review). They act by cleaving either one (type I enzymes) or both (type II enzymes) strands of DNA in order to allow strand-passage events to occur before rejoining the broken DNA ends. Type I topoisomerases are divided into two families, IA and IB, based on structural and mechanistic differences (2,3). The type IA family is composed of archeabacterial reverse gyrase, bacterial topoisomerase (topo) 1 I and topo III, and the eucaryotic topo III enzymes. All form covalent 5Ј-phosphotyrosine linkages with cleaved DNA. This is in contrast with type IB enzymes, which link to 3Ј phosphoryl groups. For several decades, a paradigm existed suggesting all topoisomerases are directly involved in regulat-ing the intracellular levels of DNA supercoiling. This paradigm has been challenged by the study of the topo III enzymes. On the amino acid sequence level, the topo III enzymes are similar to Escherichia coli topo I (4). On an enzymatic level, however, they possess a rather weak activity in relaxing negative supercoils. In this respect, E. coli topo III is only onequarter as active as E. coli topo I (5). Decatenation appears to be the preferred cellular activity for E. coli topo III (6), although this enzyme can also knot and unknot single-stranded RNA circles in vitro (7). The supercoil relaxation activity of bacterial topo III is increased at higher temperatures, suggesting a preference for partially denatured substrates (5). Such an affinity has been further demonstrated with yeast topo III, which more readily relaxes a plasmid substrate when it contains a 29-nucleotide single-stranded loop (8). Furthermore, experiments monitoring cellular DNA supercoiling in a yeast strain under conditions where neither topo I nor topo II was active demonstrated that topo III does not play a major role in regulating DNA supercoiling (9).
Apart from these biochemical observations, one key toward understanding the biological function(s) of the topo III enzymes has come through the study of mutants. Neither bacteria nor yeast topo III is essential, although mutation in yeast results in a pleiotropy of phenotypes: slow growth, hyper-recombination between repeated sequences, increased mitotic and meiotic chromosome nondisjunction, and failure to sporulate (10,11). The slow growth defect is most likely due to an extended S/G 2 transition (12). In addition, the chromosome nondisjunction and sporulation defects may be the result of an essential role for yeast topo III in meiotic recombination. This is supported by the fact that deletion of the SPO11 gene bypasses recombination and allows the top3⌬ mutants to sporulate (13).
An extragenic mutation, sgs1, was found to suppress the slow growth and hyper-recombination phenotypes of the top3 yeast mutant (12). Sgs1 is a member of a family of 3Ј-5Ј helicases, which includes bacterial RecQ and five homologs in humans, including the Blm and Wrn helicases (14 -19). Studies with bacteria have shown that RecQ helicase can both initiate homologous recombination and disrupt illegitimate recombination intermediates (20,21). Mutations in the BLM and WRN genes result in Bloom's syndrome and Werner's syndrome, respectively (22,23). These disorders are characterized by genomic instability and elevated rates of cancer, as well as the early onset of aging-related phenotypes in Werner's patients. Furthermore, these two genes, BLM and WRN, are capable of suppressing hyper-recombination in the yeast sgs1 mutant (24). Sgs1 protein physically interacts with topo III (12), and possibly topo II as well (25). Studies in yeast revealed that sgs1 mutation alone elevates recombination levels and results in decreased lifespan by the early onset of aging-related phenotypes (26,27). This premature aging is thought to be due to an accumulation of extrachromosomal rDNA circles (up to levels equivalent to the genomic DNA), which then results in nucle-olar fragmentation (28). The rDNA circle accumulation may be the result of increased recombination events between the tandem repeats within the rDNA gene cluster.
Recent interest in this field is underscored by the discovery of two topoisomerase III isozymes in mammals, topo III␣ and topo III␤ (29 -32). Human topo III␤ encodes three alternatively spliced transcripts, and the largest of these gene products can interact with yeast Sgs1 protein (33). In mice, topo III␣ is essential, for knockouts die in utero (34). Therefore, the study of mutants provides a valuable tool for understanding the biological function(s) of the topo III enzymes. Given the power of Drosophila genetics and cell biology, we set out to identify topoisomerase III in this organism and report our initial findings in this paper.

EXPERIMENTAL PROCEDURES
Cloning the Drosophila Topoisomerase III Gene-We synthesized several degenerate oligonucleotides based on the regions of high homology among type IA topoisomerase sequences and used them as primers for PCR amplification of Drosophila genomic DNA and cDNA. Two of these primers with the sequences of TA ( Fig. 1A. The PCR fragment was purified and sequenced to confirm that the cloned sequence is from the topoisomerase III gene. PCR was also used to prepare labeled DNA as a probe to screen a Drosophila embryonic cDNA library, which was prepared in our laboratory (35). A cDNA clone with a 3-kb insert was isolated, and sequences of both strands were determined.
Expression Constructs and Generation of Topo III Antibody-After removing 270 bp from the 5Ј-untranslated region, the top3 cDNA was ligated into a pET3a vector (Novagen) for isopropyl-1-thio-␤-D-galactopyranoside-inducible expression in BL21(DE3)pLysS bacterial cells (36). A protein of 97 kDa was overproduced after induction, and it was purified by SDS-polyacrylamide gel electrophoresis. The gel-purified protein was used both as an antigen to immunize a rabbit and as a ligand for affinity purification of the rabbit antibody.
For yeast expression, 270 bp were first removed from the 5Ј-untranslated region of top3. Two oligonucleotides designed to introduce a new start codon and a 6-histidine tag were annealed and ligated immediately upstream of, and in frame with, the top3 start codon (5Ј-GGCCA-GATCTAATGTCTCACCATCATCATCACCA and 5Ј-TATGGTGATGA-TGATGGTGAGACATTAGATCT). The 6-His-top3 insert was ligated into YEpG to create the pTWtop3 expression vector.
Protein Purification-Cells of the S. cerevisiae strain JEL1(⌬top1) were transformed to URA ϩ with pTWtop3. A single colony was used to inoculate synthetic media lacking uracil, supplemented with 2% dextrose (SDϪU). The SDϪU starter culture was diluted into 8 liters of rich medium containing 2% galactose to induce plasmid expression. The 8-liter culture was harvested at mid-log phase (approximately 48 h after induction), and the cell pellets were washed once with cold, sterile water, combined, and stored at Ϫ80°C.
The pellet was thawed at room temperature and resuspended in Buffer S (1 M D-sorbitol, 25 mM NaPO 4 , pH 7.4, and 10 mM MgCl 2 ) containing 10 mM DTT. The cells were centrifuged at 2,600 ϫ g for 10 min and resuspended in buffer S with 10 mM DTT. The cells were treated with yeast lytic enzyme (ICN Biomedicals Inc.), then pelleted at 2,600 ϫ g for 15 min. The pellet was twice resuspended in a buffer of 40 mM MES, pH 6.4, 10 mM MgCl 2 , 0.2% Triton X-100, 0.1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, and 1 g/ml each leupeptin and pepstatin A. With each resuspension, pellets were Dounce homogenized, then centrifuged at 2,600 ϫ g for 10 min. The supernatants were discarded, and the pellet was resuspended in a buffer of 10% glycerol, 15 mM NaPO 4 , pH 7.4, 1 M NaCl, 0.1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, and 1 g/ml each leupeptin and pepstatin A. The resuspended pellet was cleared by centrifugation at 10,200 ϫ g for 15 min.
The supernatants (nuclear extract) were made 50 mM for imidazole, pH 7, and filtered before loading onto a nickel-NTA-agarose column (Qiagen). The protein was eluted in 500 mM imidazole, 50 mM NaCl. Peak fractions off the nickel column were diluted with five volumes of buffer P (15 mM NaPO 4 , pH 7.4, and 10% glycerol), and passed over single-stranded DNA-agarose (Life Technologies, Inc.). The protein was eluted using a 0.2-2.0 M NaCl step gradient in buffer P. Peak fractions from the DNA column were concentrated over hydroxylapatite (Bio-Rad), eluting with 0.5 M NaPO 4 , pH 7.4. The peak fraction was dialyzed into a buffer of 15 mM NaPO 4 , pH 7.4, 50% glycerol, 5 mM NaCl, 0.1 mM phenylmethylsulfonyl fluoride, and 2 g/ml each leupeptin and pepstatin A, then stored at Ϫ20°C.
Plasmid Relaxation Assays-The hypernegatively supercoiled substrates were generated by adding excess ethidium bromide to negatively supercoiled plasmid DNA at a DNA bp to ethidium ratio of 2:1. The reactions were incubated with Drosophila topo I-ND423 (37) for 1 h at 30°C to relax the positive supercoils. The reactions were stopped by adding 10 mM EDTA, then phenol-chloroform-extracted to remove the ethidium. The DNA was then ethanol-precipitated and resuspended in TE buffer. The topoisomer ladder was generated similarly, using DNA bp to ethidium ratios ranging from 2:1 to 80:1. After phenol-chloroform extraction, DNA loading dye was added directly to the aqueous phase.
Standard topo III relaxation conditions contain 0.3 g (0.15 pmol) of DNA, 50 -100 ng (0.5 pmol) of purified topo III, 40 mM Hepes-KOH, pH 7.5, 1 mM MgCl 2 , and 0.05 mg/ml bovine serum albumin in a final volume of 20 l. Reactions are incubated at 37°C for 30 min, then stopped by adding DNA loading dye containing 1 mg/ml proteinase K and electrophoresed on a 1.2% agarose gel in TPE buffer. Reactions were preincubated at either 30°C (topo I and II) or 37°C (topo III) for 5 min before adding topoisomerase to initiate cleavage. After mixing, the reactions were stopped immediately by adding an equal volume of 2ϫ Laemmli sample buffer. The samples were boiled and electrophoresed on an 8% polyacrylamide gel. The gel was either dried onto Whatman paper or transferred to nitrocellulose before being subjected to autoradiography and Western blot analysis.

RESULTS
Cloning and Sequencing the Drosophila top3 Gene-Using regions of sequence similarity between the bacterial, yeast, and human type IA enzymes, we designed two degenerate oligonucleotides and used them as primers for PCR amplification of Drosophila genomic DNA (see "Experimental Procedures"). The PCR products were first cloned and sequenced to confirm their homologies with topoisomerase III sequences. The PCR reaction was then used to generate a radiolabeled probe for hybridization screening of a Drosophila embryo cDNA library. One of the positive clones was isolated, and both strands of the 2979-base pair insert were sequenced.
The largest open reading frame of the obtained clone is predicted to encode a protein of 875 amino acids (calculated molecular mass 97.0 kDa), the sequence of which is shown in Fig. 1A. This protein is within the same size range as the E. coli topo I and human and murine topo III enzymes. Residue 332 is the predicted active-site tyrosine, as it is contained within a highly conserved GYISYPRTET sequence. One potential bipartite nuclear localization signal is found in the amino terminus (38). In addition, four potential zinc fingers of the tetracysteine motif are located in the carboxyl terminus of the protein. While the smaller bacterial and yeast topo III enzymes appear to lack zinc fingers, E. coli topo I has been shown to coordinate three zinc(II) atoms, and the COOH-terminal domain containing these tetracysteines may have an important role in DNA binding (39,40). In addition to the tetracysteine motifs, the COOHterminal portion of the Drosophila open reading frame is also characterized by clusters of glycine and arginine residues (Fig.  1A). In the COOH-terminal 62 amino acids of this protein, there are 23 glycines and 6 arginines, which accounts for 47% of the residues.
Drosophila Topo III Is Closely Related to the Topo III ␤-Subfamily-Sequence comparison using the BLAST algorithm identifies the cloned Drosophila sequence as being a member of the type IA family of topoisomerases (Table I). Homology comparisons with other eucaryotic topo IIIs show similarity ranging from 49% to 71% and identity from 33% to 59%. Mammals contain two topo III isozymes, designated ␣ and ␤, and the Drosophila protein is obviously more closely related to the ␤-subfamily (58% versus 38% identity). The alignment of the Drosophila sequence with those of the topo III␤s is especially striking in the COOH-terminal domain, which is characterized by 4 tetracysteines; both the eight CXXC sequences and the intervening spacers are highly conserved (Fig. 1B). This is in contrast with the comparison between Drosophila topo III and mammalian topo III␣ or between mammalian topo III␤ and topo III␣, where most of the homologies lie in the NH 2 -terminal half (a PILEUP comparison of type IA topoisomerases is contained in the online version of this article). Another hallmark for the topo III␤ sequences is the presence of GR-rich clusters in their COOH termini. We therefore propose that the cloned Drosophila top3 sequence belongs to the ␤-subfamily.
Drosophila top3␤ Is a Single-copy Gene-A biotin-labeled top3␤ DNA probe was used for in situ hybridization with polytene chromosomes from the salivary glands of 3rd instar larvae (41). The probe hybridizes between regions 5E and 5F on the X chromosome (data not shown). Hybridization is observed only at this locus, suggesting top3␤ exists as a single-copy gene on the X chromosome. Furthermore, genomic Southern hybridization also suggests that top3␤ is a single-copy gene (data not shown). These experiments were carried out under stringent hybridization conditions, which cannot exclude the possibility that Drosophila may possess another topo III isoform. Indeed, submission of human topoisomerase III␣ cDNA for BLAST search against the Berkeley Drosophila Genome Project data base results in several matches with an 87.8-kb P1 genomic clone located at 37E1-37E2 on the left arm of the 2nd chromosome (GenBank accession no. AC005428). Translation of these sequences identifies a putative new protein, Drosophila topoisomerase III␣. Experiments are currently under way to isolate and analyze this new gene.
Expression of Topoisomerase III␤ during Drosophila Development-To probe its biological function(s), we investigated the expression pattern of topo III␤ protein throughout Drosophila development. Extracts made from Drosophila at various stages of development were analyzed by Western blot using affinitypurified topo III␤ antibody ( Fig. 2; see "Experimental Procedures" for source of this antibody). The antibody recognizes a protein of approximately 97 kDa. In contrast to topo I and topo II proteins, which peak during the 6 -12-h period of development (42,43), topo III␤ protein levels peak during the first 6 h of embryogenesis. Throughout this time, the topo III␤ levels are fairly constant (data not shown). The protein levels decline during the later stages of embryogenesis, the larval stages, and the pupal stage, but increase again during adulthood. It is interesting to note that top3␣-knockout mice die during embryogenesis (34). Our developmental Western blot suggests that Drosophila topo III␤ may also play an important role during the first few hours of the fruit fly life.
Topoisomerase III␤ Levels Are Greatly Reduced in an EP(X)1432 Mutant Fly-There are no known mutations located at the mapped cytological position of top3␤. However, when we submitted our sequence for BLAST search against the Berkeley Drosophila genome project data base, one match to top3␤ was obtained with a sequence of 183 bp surrounding the insertion site of EP(X)1432. The EP(X) flies are a series of transgenic lines with P-element insertions on the X chromosome (44, 45). Based on sequence comparison, we can map that EP(X)1432 is inserted in the 5Ј-untranslated region of the top3␤ transcribed Residue 332 is the predicted active-site tyrosine (bold). The amino terminus contains one potential bipartite nuclear localization signal (double underline). The carboxyl terminus contains eight CXXC motifs (bold underline), which may encode as many as four zinc fingers. The COOHterminal tail is also rich in glycine and arginine residues (wavy underline). B, COOH-terminal sequence alignment of four proposed topo III␤ enzymes, with the CXXC motifs underlined. sequence and is located 29 bp upstream of the translation initiation codon. Therefore, while the P-element insertion is expected to affect the expression of top3␤, it would not necessarily result in a null mutation. We obtained this fly stock and used PCR to confirm the presence and location of the P-element insertion. We then used Western blot analysis to compare topo III␤ levels in male and female transgenic flies to their wild type counterparts. While topo I and topo II levels are approximately the same in the wild type and mutant flies, the topo III␤ protein level is greatly diminished in the EP(X)1432 mutant (Fig. 3). These mutant flies are both viable and fertile. However, we did not examine whether they have reduced fertility/viability, or whether they have an altered recombination frequency. This result suggests that the overall viability and fertility of the fruit fly are not sensitive to the levels of topo III␤.
Drosophila top3␤ Suppresses the Yeast top3⌬ Slow Growth Phenotype-Mutation of the Saccharomyces cerevisiae TOP3 gene is known to result in several phenotypes, including a growth rate which is only 50% that of wild type (11). In order to assess whether Drosophila top3␤ possesses functional similarity to the yeast TOP3 gene, we cloned the top3␤ cDNA into a YEpG vector to generate pTWtop3. In this construct, 270 bp have been removed from the 5Ј-untranslated region to facilitate heterologous expression. In addition, a new initiation codon, followed by a 6-histidine tag, has been inserted just upstream of the original top3␤ start codon. Expression of this construct is under control of the galactose-inducible GAL1 promoter. This expression construct was transformed into JCW253, a yeast strain deleted for TOP3. The Drosophila top3␤ cDNA can rescue the slow growth of the top3⌬ mutant when grown in media containing galactose (Fig. 4). This improved growth rate is not observed when YEpG vector lacking the top3␤ insert is transformed into the yeast mutant, or when the strains are grown on media containing glucose (data not shown). Therefore, Drosophila top3␤ can be functionally expressed in yeast, and it shares functional similarity with the yeast TOP3 gene. Recent results have shown that human top3␤ can also rescue the top3⌬ growth defect in yeast (33).
Relaxation of Hypernegatively Supercoiled, but Not Negatively Supercoiled, DNA by Topoisomerase III␤-The functional expression of Drosophila topo III␤ in yeast allowed us to purify this protein for biochemical studies. To eliminate any potential contamination of major type I topoisomerase activity, namely that from yeast topo I, we expressed Drosophila topo III␤ in top1 Ϫ yeast (46). JEL1(⌬top1) yeast transformed with the pT-Wtop3 expression vector were grown in 8 liters of rich medium containing 2% galactose. The cells were harvested and lysed, and nuclear extract was prepared with an extraction buffer containing 1 M NaCl. The nuclear extract was fractionated over a nickel-NTA column, followed by a single-stranded DNA agarose column and hydroxylapatite (Fig. 5). Most of the purification was achieved at the step of the nickel-NTA affinity chromatography, and the 97-kDa protein is the predominant species in the purified fraction.
We tested the purified topo III␤ protein for relaxation activity toward plasmid DNA isolated directly from bacterial cells. The protein showed no activity toward this negatively supercoiled substrate in the range of temperature tested, from 30 to 65°C (Fig. 6A, lanes 1-6). The observation that type IA enzymes have an affinity for single-stranded DNA (8,47) led us to test the relaxation activity of topo III␤ toward highly unwound, or hypernegatively supercoiled, plasmids. These hypernegatively supercoiled substrates were generated by incubating the plasmid DNA with an excess of ethidium bromide, followed by relaxation with Drosophila topo I. Upon phenol extraction to remove the ethidium, the topo I-relaxed DNA becomes hypernegatively supercoiled. When we assay for relaxation with this substrate, a slight but definite reduction in mobility is observed, with relaxation at an optimal between 37 and 45°C (Fig. 6A, lanes 7-12). Relaxation of this highly underwound substrate by topo III␤ is only partial. The observed shift in  mobility appears to terminate at a definite point, with the bulk of the topoisomers migrating with approximately the same mobility as the negatively supercoiled plasmid marker. In addition, topo III␤ relaxation does not appear to be sequencespecific, since we have observed this same phenomenon with four different plasmids ranging in size from 4 to 13 kb.
This partial relaxation of hypernegatively supercoiled DNA by topo III␤ is not due to an insufficient amount of enzyme in the reaction. The relaxation reaction is essentially complete within 1 h; either prolonged incubation for another 12 h or addition of a second aliquot of enzyme does not result in further shift in mobility (Fig. 6B, lanes 1-4). Topo III␤ remains active in the reaction mixture since addition of another hypernegatively supercoiled substrate to the reaction results in similar relaxation of the larger substrate (Fig. 6B, lane 5). This suggests that the topo III␤ enzyme is specific for highly underwound substrates. Consistent with this idea is the observation that Drosophila topo III␤ does not relax positively supercoiled DNA (data not shown).
We further investigated the effect of divalent cations and monovalent salt on the activity of Drosophila topo III␤. Relaxation activity can be observed in the absence of added divalent cation, but addition of EDTA abolishes this activity, demonstrating a requirement for divalent cations (Fig. 7A, lanes 1 and  5). Activity can be restored when a molar excess of Mg ϩ2 , Mn ϩ2 , or Ca ϩ2 , but not Co ϩ2 , Cu ϩ2 , or Zn ϩ2 , are included in addition to EDTA (Fig. 7A, lanes 6 -11). The optimal Mg ϩ2 concentration is about 1 mM and at higher concentrations the reaction is inhibited (Fig. 7A, lanes 2-4). The preference of topo III␤ for low ionic strength can also be seen in the observation that the optimal monovalent salt concentration is below 50 mM (Fig.  7B). It is possible that an extended region of single strand DNA in the circular DNA substrate is requisite for topo III␤ reaction and the presence of a higher concentration of divalent cations reduces this single-strandedness in the DNA.
While the agarose gel electrophoresis employed here provides a convenient method to monitor the relaxation of hypernegatively supercoiled DNA by topo III␤, its limitation in the resolution of the negatively supercoiled DNA precludes us from determining the change in linking number that occurred in this reaction. To this end, we have used a series of DNAs with different linking numbers as a reference for gel electrophoresis in the presence of ethidium bromide (Fig. 8). This series of topoisomers was generated to cover the range of supercoiling from the hypernegatively supercoiled substrate to the fully relaxed species. The pieces of the ladder were made by incubating plasmid DNA with varying amounts of ethidium bromide (DNA bp:ethidium ratios of 2:1 to 80:1), followed by relaxation with Drosophila topo I, and phenol extraction. The pieces of the topoisomer ladder, along with the hypernegatively supercoiled substrate (H), topo III reaction product (T), negatively supercoiled plasmid (N), and the fully relaxed species (R), were arranged in the order of linking number changes and the trends in mobility shifts. Identical sets of DNA samples were resolved by electrophoresis through agarose gels either in the presence of 0.75 or 0.05 g/ml ethidium (Fig. 8, A and B). It is interesting to notice that, while there is only a small shift in the mobilities between the hypernegatively supercoiled substrate and its topo III␤ reaction product when analyzed by gel electrophoresis in the absence of ethidium (Figs. 6 and 7) or in the presence of a low concentration of ethidium (Fig. 8B), there is a much greater shift in the mobilities at a higher ethidium concentration (Fig. 8A). Under such an electrophoretic condition, the hypernegatively supercoiled DNA remains slightly negatively supercoiled and the topo III␤ product is converted into a highly positively supercoiled species. Based on analysis from these data and from other gel electrophoretic conditions, we can estimate that the linking difference between H and T DNA is between 30 and 35, T and N between 5 and 10, and N and R around 30. Therefore, the linking change from H to T is nearly equivalent to that from N to R, which occurs in the reaction catalyzed by most of the DNA topoisomerases. Drosophila topo III␤ is as proficient in reducing the linking number deficit as other type I enzymes, except it seems to operate in a different supercoiling range.
Topoisomerase III␤ Binds Covalently to DNA 5Ј Phosphoryl Ends-To further confirm we had identified a type IA enzyme, we investigated whether topo III␤ could link covalently to DNA 5Ј phosphoryl groups. In this experiment, the topoisomerase can be trapped in a covalent complex with labeled DNA when the reaction is stopped by a strong denaturant like SDS. In effect, the label on the DNA is transferred to the protein, and the labeled protein can be detected by autoradiography. Two 16-mers, each with a 4-nucleotide 5Ј-protruding end, were annealed prior to radiolabeling at either their 5Ј ends with T4 polynucleotide kinase, or their 3Ј ends with T4 DNA polymerase. As controls for the experiment, we incubated our labeled oligonucleotides with Drosophila topo II (which is known to bind to 5Ј phosphoryl groups) and an amino-terminally truncated form of Drosophila topo I (which is known to bind to 3Ј phosphoryl groups). An autoradiograph signal is seen for topo III␤ only when it is incubated with the 3Ј end-labeled sub-strate, indicating that it binds to 5Ј phosphoryl groups (Fig.  9A). This result is expected for a type IA topoisomerase. Western blot on the same samples shows that the autoradiograph signal due the label transfer coincides with the protein signal for the three topoisomerases. It also demonstrates that the same amount of protein was incubated with both the 5Ј and 3Ј end-labeled substrates. When we used a larger oligomer, however, an upward shift in mobility was observed, as would be expected for a protein bound to a larger DNA fragment (data not shown).
The label-transfer experiment provides another assay for the interactions between topo III␤ and its DNA substrates. We examined this reaction under different divalent cation conditions and with both single-and double-strand DNA substrates (Fig. 9B). The single-stranded substrates were generated by heat denaturation of the double-stranded oligonucleotides used above. Under all conditions tested, the single-stranded DNA is cleaved to a greater extent than the double-stranded DNA. In some cases, a doublet is observed, indicating that the oligonucleotide sequence contains at least two cleavage sites for topo III␤. Cleavage is observed in the absence of added divalent cation. However, it is stimulated when a divalent cation like Mg ϩ2 or Mn ϩ2 is added. The presence of Mn ϩ2 seems to support the cleavage reaction at least as well as Mg ϩ2 , which is also the case for the relaxation of hypernegatively supercoiled DNA. It will be interesting to examine if other eucaryotic topo III also have a preference for Mn ϩ2 , just like the Drosophila enzyme. DISCUSSION BLAST sequence alignments suggest Drosophila topo III is a member of the ␤-subfamily of type IA topoisomerases, being nearly 60% identical to mammalian topo III␤ enzymes. The homology among the topo III␤ enzymes is particularly striking in the COOH-terminal region where all contain eight highly conserved CXXC motifs, with the spacing between these motifs being highly conserved as well. These CXXC motifs suggest the topo III ␤s may possess as many as four zinc fingers. These motifs do not conform to the proposed zinc finger motif for E. coli topoisomerase I (Cys-X 2 -Cys-Gly-X 2 -Met-X 12-13 -Cys-X 4 -10 -Cys) (39). There are three such zinc finger motifs in the COOH terminus of E. coli topo I, which has been shown to bind three zinc(II) ions and require zinc coordination for cleavable complex formation with DNA (40). Interestingly, the mammalian topo III␣s contain zinc finger motifs similar to E. coli topo I (excluding the consensus for a glycine at the 5th position), but lack the eight CXXC motifs found in the topo III␤s. Based on these sequence observations, we suggest the type IA enzymes may fall into three subfamilies: 1) enzymes that lack zinc finger motifs, such as reverse gyrase, bacterial topo III and yeast topo III; 2) enzymes containing the motif found in E. coli topo I, which includes the mammalian topo III␣s; 3) enzymes containing eight CXXC motifs, as found in the topo III␤s.
Immediately following the CXXC motifs in the topo III␤s are clusters of glycine and arginine residues. While all of the type IA enzymes contain glycines and arginines in their COOH termini, only the topo III␤ enzymes have them arranged in clusters. We first noticed this phenomenon in the Arabidopsis topo III␤ sequence (GenBank accession AAD15404), which has a stretch of 20 consecutive Gs and Rs near the COOH-terminal end of the protein. The significance of these GR clusters is not known. It is possible they may mediate nucleic acid binding through the positive charge of the arginines. Alternatively, these GR clusters may specify a protein-protein interaction domain. Similar GR clusters have also been identified in sev-eral RNA-binding and nucleolar-localizing proteins, such as nucleolin (48), mammalian protein C23 (49), and nucleolar scleroderma antigen (50).
Like other topo III enzymes, Drosophila topo III␤ appears to possess a weak activity in relaxing negatively supercoiled plasmid DNA isolated directly from bacterial cells. In fact, we were only able to observe relaxation activity with a highly underwound, or hypernegatively supercoiled, substrate. This suggests our assay may be useful for the identification of new members of the topo III family. In addition, it affects the idea that the topo III enzymes are not inefficient in supercoil relaxation, but rather that they work within an atypical linking number range. To this end, we were able to quantitate the level of hypernegative supercoil relaxation induced by Drosophila topo III␤ and found it to be comparable to the degree of relaxation of negatively supercoiled plasmid DNA by E. coli topo I. Our results also support the idea that the topo III enzymes possess a unique substrate requirement, namely one that has exposed single-stranded regions. These single-stranded regions can be generated either by heating the reaction to high temperature, as is the case for E. coli topo III (5), or through a hypernegative supercoiling of the substrate. It is interesting to speculate on which biological processes may create such highly underwound DNA species. Strand separation events, such as DNA replication and transcription, could provide structures that are hot spots for topo III activity. Alternatively, topo III may act on the DNA intermediates created during the process of recombination.
Genetic experiments in yeast have demonstrated that TOP3 plays a role in suppressing mitotic recombination and in resolving recombined homologous chromosomes during meiosis I (11,13). Furthermore, the combined action of either yeast or bacterial topo III and the DNA helicase RecQ can promote the formation of DNA catenanes (51). Similar strand passage reactions may be involved in the initiation and resolution steps during recombination. The unwinding action of a RecQ-type helicase appears to generate a DNA structure that can be recognized by a topo III enzyme (51). It will be interesting to determine whether our topo III␤ interacts with the newly identified Drosophila RecQ homolog, Dmblm (52), and what role, if any, these enzymes play in the above mentioned processes.
The developmental Western blot demonstrates that the greatest expression of topo III␤ occurs during the first 6 h of embryogenesis. This suggests an important function for topo III␤ during the first few hours of the fly lifecycle. The topo III␤ protein levels decrease during later stages of embryogenesis, the instar larval stages, and the pupal stage, then increase again during adulthood. This expression pattern suggests topo III␤ may be important for some unique aspect of the the DNA replication and chromosome segregation process, given embryogenesis is characterized by rapid cycles of DNA replication and chromosome segregation in the absence of intervening G phases, while endoreplication is prolific in the larval stages (53).
While we do not know whether the EP(X)1432 mutant is hypomorph or null for topo III␤ activity, our results indicate that the mutant flies are fertile and viable under greatly reduced levels of topo III␤. It is still possible that the top3␤ mutants may contain defects in aspects yet to be tested, including recombination and repair. However, it appears the biological function of topo III␤ may not be required for the viability and fertility of Drosophila. Through a search of the published data base for the Drosophila genome, we have identified DNA sequences that may encode topo III␣. Drosophila therefore likely contains both forms of the topo III isozyme, just like the mammalian cells. The experiments using knock-out mice have FIG. 9. Drosophila topo III␤ binds covalently to the 5 end of cleaved DNA. A, 5Ј and 3Ј designate the 5Ј end-labeled and 3Ј endlabeled oligonucleotides used in the reaction, while I, II, and III indicate reactions with Drosophila topo I-ND423, topo II, and topo III␤, respectively. The reactions were stopped with 2ϫ Laemmli sample buffer then electrophoresed on an 8% SDS-polyacrylamide gel. The samples were subjected to autoradiography (left panel), followed by Western blot (right panel) using topo I, topo II, and topo III␤ antibodies. B, The 3Ј end-labeled oligonucleotides, both double-stranded (DS) and singlestranded (SS), were incubated with topo III␤ under various divalent cation conditions. Reactions were performed in the presence of 40 mM Hepes-KOH (lanes 1 and 2) plus 1 mM CaCl 2 (lanes 3 and 4), 0.5 mM MgCl 2 (lanes 5 and 6), 1 mM MgCl 2 (lanes 7 and 8), 0.5 mM MnCl 2 (lanes 9 and 10), or 1 mM MnCl 2 (lanes 11 and 12). demonstrated that the function of topo III␣ is essential (34). It will be interesting to determine whether topo III␤ is essential as well. The specific and overlapping functions of these two isozymes in Drosophila and mammalian cells will clearly be an important issue to address in the future.