Essential Functions of C Terminus of Drosophila Topoisomerase IIIα in Double Holliday Junction Dissolution♦

Background: Topoisomerase IIIα (Top3α) and Blm dissolve Holliday junctions into non-crossover products. Results: The Top3α C terminus binds to Blm and DNA substrates and is important in vivo. Conclusion: The C-terminal domain of Top3α is required for dissolution and cellular functions. Significance: The Top3α C terminus is an essential component of the dissolvasome complex. Topoisomerase IIIα (Top3α) is an essential component of the double Holliday junction (dHJ) dissolvasome complex in metazoans, along with Blm and Rmi1/2. This important anti-recombinogenic function cannot be performed by Top3β, the other type IA topoisomerase present in metazoans. The two share a catalytic core but diverge in their tail regions. To understand this difference in function, we investigated the role of the unique C terminus of Top3α. The Drosophila C terminus contains an insert region not conserved among metazoans. This insert contributes an independent interaction with Blm, which may account for the absence of Rmi1 in Drosophila. Mutant Top3α lacking this insert maintains the ability to perform dHJ dissolution but only partially rescues a top3α null fly line, indicating an in vivo role for the insert. Truncation of the C terminus has a minimal effect on the type IA relaxation activity of Top3α; however, dHJ dissolution is greatly reduced. The Top3α C terminus was found to strongly interact with both Blm and DNA, which are critical to the dissolution reaction; these interactions are greatly reduced in the truncated enzyme. The truncation mutant also cannot rescue the viability of top3α null flies, indicating an essential in vivo role. Our data therefore suggest that the Top3α C terminus has an important role in dHJ dissolution (by providing an interaction interface for Blm and DNA) and an essential function in vivo.

Topoisomerase III␣ (Top3␣) is an essential component of the double Holliday junction (dHJ) dissolvasome complex in metazoans, along with Blm and Rmi1/2. This important anti-recombinogenic function cannot be performed by Top3␤, the other type IA topoisomerase present in metazoans. The two share a catalytic core but diverge in their tail regions. To understand this difference in function, we investigated the role of the unique C terminus of Top3␣. The Drosophila C terminus contains an insert region not conserved among metazoans. This insert contributes an independent interaction with Blm, which may account for the absence of Rmi1 in Drosophila. Mutant Top3␣ lacking this insert maintains the ability to perform dHJ dissolution but only partially rescues a top3␣ null fly line, indicating an in vivo role for the insert. Truncation of the C terminus has a minimal effect on the type IA relaxation activity of Top3␣; however, dHJ dissolution is greatly reduced. The Top3␣ C terminus was found to strongly interact with both Blm and DNA, which are critical to the dissolution reaction; these interactions are greatly reduced in the truncated enzyme. The truncation mutant also cannot rescue the viability of top3␣ null flies, indicating an essential in vivo role. Our data therefore suggest that the Top3␣ C terminus has an important role in dHJ dissolution (by providing an interaction interface for Blm and DNA) and an essential function in vivo.
Maintaining genomic stability requires efficient repair of DNA breaks with minimal genetic changes. The complex of Bloom helicase (Blm) and topoisomerase III␣ (Top3␣) 2 has been shown to convergently migrate double Holliday junctions (dHJs), an intermediate of double-strand break repair, into solely non-crossover products (1,2). The significance of this anti-recombinogenic function can be seen in the phenotype of Blm-deficient cells, which show a marked increase in sister chromatid exchanges (3) and loss of heterozygosity (4).
Blm and Top3␣ together are sufficient for dHJ dissolution (1,2,5). The two proteins are known to interact directly (6). In yeast, the ϳ100 N-terminal amino acids of Sgs1, the yeast homolog of Blm, are required for interaction with Top3 (7,8). The Top3␣-interacting domain of human Blm has also been identified, although some ambiguity remains. Wu et al. (9) showed that both the N-and C-terminal ϳ200 amino acids of human Blm can independently bind Top3␣ in a far Western analysis. However, Hu et al. (10) showed by co-immunoprecipitation that the N-terminal region is solely responsible for the interaction. Information on the region of Top3␣ that binds Blm is completely absent.
In addition to Blm and Top3␣, the small structural proteins Rmi1 and Rmi2 were found to be stabilizing components of the complex in a variety of organisms (11)(12)(13)(14). The deletion of either Rmi1 or Rmi2 recapitulates the Blm deficiency hallmark of increased sister chromatid exchanges (13,15). However, Drosophila does not appear to possess any homologs of either structural protein (13).
In metazoans, there are a pair of type IA topoisomerases: Top3␣ and Top3␤. Although both can perform relaxation on topologically constrained DNA with single-stranded regions (16 -19), Top3␤ cannot substitute for Top3␣ in dHJ dissolution (2). Their in vivo effects are similarly distinct. Whereas Top3␣ is absolutely essential, Top3␤ mutants show only a modest reduction in life span (20 -22), indicating that relaxation activity alone is insufficient for the full range of in vivo functions. Because Top3␤ has no functional interaction with Blm (23), the essential function of Top3␣ may be alongside Blm in the recombinational repair pathway.
Top3␣ and Top3␤ share the type IA catalytic domain. The ϳ600 N-terminal residues of Top3␣ and Top3␤, including the catalytic core and active site tyrosine, are very similar, with 63% similarity and 43% identity between the human enzymes (57 and 35% in Drosophila, respectively). However, the proteins greatly diverge in length and sequence composition of their remaining C termini (see Fig. 1A), suggesting a potential region of protein-protein interaction specificity.
Here, we show that the unique C terminus of Drosophila Top3␣ is necessary for the interaction between Blm and Top3␣, for double Holliday junction dissolution to occur efficiently, and to rescue the viability of top3␣ null flies. In addition to the conserved Top3␣ C terminus, the Drosophila enzyme contains an inserted region that can independently bind to Blm, perhaps compensating for the absence of the Rmi structural proteins. Top3␣ lacking this region can only partially rescue top3␣ null flies. The Top3␣ C terminus is also involved in binding to DNA substrates, substantially contributing to the binding activity. Together, these data reveal the essential role for the Top3␣ C terminus in recombinational repair.
Purification of Blm-Recombinant Blm was purified as by Weinert and Rio (24) with the following exceptions. All listed steps were performed at 4°C, and all buffers contained 5 mM 2-mercaptoethanol, 1 mM benzamidine, and 0.1 mM PMSF. The nuclear pellet was resuspended in DNA precipitation buffer (15 mM Hepes-KOH (pH 7.6), 400 mM KCl, 5 mM MgCl 2 , 1 mM EDTA, and 1 mM EGTA) and gently homogenized. This solution was allowed to incubate for 30 min before centrifugation at 20,000 ϫ g for 20 min. The supernatant was recovered, and 0.1 volume of saturated (NH 4 ) 2 SO 4 was added. Ammonium sulfate precipitation and storage were done as described above. Precipitates were resuspended in 10 ml of single-stranded DNA (ssDNA) wash buffer (20 mM Tris (pH 7.9), 200 mM NaCl, 10% glycerol, and 0.02% Triton X-100) and dialyzed twice into 2 liters of this buffer for 2 h. The dialyzed solution was then incubated with 2-3 ml of ssDNA-cellulose resin (United States Biochemical Corp.) for 30 min. The resin was washed with wash buffer for 10 min and then incubated with ssDNA elution buffer (20 mM Tris (pH 7.9), 1 M NaCl, 10% glycerol, and 0.02% Triton X-100) for 30 min. The eluate was dialyzed twice into 1 liter immunoprecipitation buffer (20 mM Tris (pH 7.5), 500 mM NaCl, 10% glycerol, and 0.02% Triton X-100) for 1 h and then incubated overnight with 100 l of anti-Glu-Glu antibody resin (Covance). The resin was washed and eluted as described above. BSA was added to the eluate to a concentration of 0.25 mg/ml; the eluate was then dialyzed into storage buffer (40 mM Tris (pH 7.5), 150 mM NaCl, and 50% glycerol) for 3 h, flashfrozen, and stored at Ϫ80°C.
Relaxation Activity Assay-Activity assays were similar to those performed in a previous study (22). Relaxation activity reactions contained 20 mM Hepes (pH 7.6), 1 mM MgCl 2 , 100 mM NaAc, 1 mM DTT, 0.1 mM EDTA, and 50 ng/l DNA. Reactions proceeded at 37°C for 30 min and were stopped by the addition of NaCl to 600 mM. After 5 min, loading dye and proteinase K (to 50 ng/l) were added. After a 30-min incubation at 37°C, samples were loaded onto a 1.5% Tris acetate/EDTAagarose gel containing 0.5 ng/l ethidium bromide and run overnight at 50 V. Gels were destained prior to imaging.
Creation of dHJ Substrate-The dHJ substrate was made as described by Plank and Hsieh (25) with the following exceptions. Cre reaction buffer contained 20 mM Tris, 300 mM LiCl, and 1 mM EDTA (pH 8.2). Instead of being electroeluted, Cre reaction products were gel-extracted; allowed to diffuse overnight into 10ϫ buffer containing 50 mM Mops (pH 7.0), 10 mM EDTA (pH 8.0), and 750 mM NaCl; and recovered through a DEAE column (Qiagen).
dHJ Substrate Dissolution Assay-Dissolution of the dHJ substrate was performed as described by Plank et al. (2) with the following exceptions. Reactions were incubated at 30°C for 30 min and contained 15 nM Top3␣, 25 nM Top3␣NT, or 20 nM Top3␣⌬CT2 and also 18 nM Blm.
Protein Binding Assay-Top3␣ and its derivatives were bound to glutathione-Sepharose resin. Equivalent amounts of bound protein were incubated with purified Blm for 30 min at 4°C in 40 mM Tris (pH 7.5), 100 mM NaCl, 4 mM MgCl 2 , and 10% glycerol. The Top3␣NT experiment included 5 g/ml ethidium bromide. After incubation, the resin was washed three times with several volumes of the same buffer. Flowthrough and resin samples were run on a 4 -20% Criterion acrylamide gel (Bio-Rad), transferred to nitrocellulose, and blotted with mouse anti-Glu-Glu antibody. The blot was incubated with HRP-conjugated goat anti-mouse antibody and substrate (Thermo Scientific) and visualized with film (Eastman Kodak).
Drosophila Crosses-The Top3␣NT and Top3␣⌬CT2-expressing transgenes were generated from the Top3␣-YFP-expressing transgene (26) by excising the appropriate tail regions and adding back the nuclear localization sequence (amino acids 845-862) between the Top3␣ and YFP sequences. The transgenes were injected into fly larvae at the Duke Model System Genomics facility. These flies were mated with the homozygous lethal top3␣ 54 line to create top3␣ 54 /CyO;NT/ϩ and top3␣ 54 / CyO;⌬CT2/ϩ. Each line was mated inter se, and the progeny were phenotyped and counted to determine rescue efficiency.
Loading buffer was added, and samples were run on an 8% Tris borate/EDTA-polyacrylamide gel overnight at 4°C. The gel was exposed to a PhosphorImager screen and scanned using a Storm scanner.

Top3␣ C Terminus Is Required for dHJ Dissolution-Because
Top3␣ and Top3␤ have distinct cellular and biochemical functions, and the major structural difference resides in their C termini (Fig. 1A), we examined the properties of the Top3␣ C terminus. We constructed a series of C-terminal truncation mutants (Fig. 1B). The N-terminal 648 amino acid construct (Top3␣NT) was purified and tested for relaxation activity. Top3␣NT was active in relaxing a hyper-negatively supercoiled plasmid, with only a slight reduction in specific activity compared with the full-length enzyme ( Fig. 2A).
We then tested Top3␣NT in the dHJ dissolution assay. The dHJ substrate contains two topologically constrained Holliday junctions separated by a homologous region of 165 bp (2). Top3␣ and Blm have previously been shown to dissolve the dHJ substrate into two separate double-stranded DNA circles (25). At comparable levels of relaxation activity, full-length Top3␣ showed clear dissolution products, whereas Top3␣NT could only minimally support dHJ substrate dissolution, with a Ͼ2-fold reduction in dissolution activity (Fig. 2B).
Previously, it has been shown that replication protein A (RPA) specifically stimulates the dHJ dissolution reaction (2). We tested Top3␣NT for stimulation of dissolution by RPA. Although stimulation of the dissolution reaction was seen for both full-length Top3␣ and Top3␣NT, the stimulation increased proportionally to RPA concentration only when the full-length enzyme was used (Fig. 2C). Stimulation using Top3␣NT was observed only at a higher level of RPA and did not continue to increase with RPA concentration.
Top3␣ C Terminus Is Involved in Binding Blm-Because dissolution requires coordination with Blm, we tested Top3␣NT for the ability to bind Blm. Top3␣ and Top3␣NT with N-terminal GST tags were bound to glutathione resin, incubated with purified Blm, and washed; bound Blm was visualized by Western blotting. Although bound Blm was clearly evident in the presence of full-length Top3␣, Top3␣NT showed only minimal binding (Fig. 3A). The clear difference between the two constructs suggests that the essential role of the Top3␣ C terminus may be to interact with Blm.
To test this hypothesis directly, Top3␣CT was constructed and purified. Top3␣CT was immobilized and incubated with purified Blm as described above. In contrast to GST alone, Top3␣CT clearly showed retained Blm (Fig. 3B). Therefore, Top3␣CT is able to bind Blm independently and presumably contributes this function to the full-length protein.
Drosophila Top3␣ C Terminus Contains Two Independent Blm-binding Domains-We made a series of C-terminal subdomain constructs to narrow down the binding region (Fig. 1B). When compared with the other metazoan enzymes, Drosophila Top3␣ contains a unique insert in its C terminus (Fig. 1A). The insert sequence is not homologous to any known protein and aligns only with the Top3␣ C termini of the genus Drosophila, among which it is conserved (supplemental Fig. S1). The first third of the insert is extremely glycine-rich (55%); the remainder is somewhat rich in lysine (8%) but contains no other outstanding features. This insert was used to define the boundaries for dividing the C terminus: the insert was designated CT2, the section prior to it was called CT1, and the section following it was termed CT3. Both individual (CT1, CT2, and CT3) and overlapping (CT1,2 and CT2,3) constructs were made (Fig. 1B). When tested individually, only the Drosophila unique region (CT2) bound Blm (Fig. 3C). This result was surprising given that Top3␣ in other species is known to interact with Blm. As expected, both CT1,2 and CT2,3 also interacted with Blm (supplemental Fig. S2A).
To determine whether CT2 was necessary or if the remaining tail was sufficient, we constructed CT1,3, which covalently connected regions 1 and 3. This CT1,3 construct was also able to bind to Blm independently (Fig. 3D). Therefore, the Drosophila Top3␣ enzyme appears to contain two regions capable of independently binding Blm: CT2, the insert present only in Drosophila, and CT1,3, which corresponds to the C terminus in other metazoans. To verify that the two domains (CT1,3 and CT2) can interact with Blm in the context of the enzyme, we added each back to the core domain, creating Top3␣⌬CT2 and Top3␣NTϩCT2. Both constructs were able to interact with Blm (supplemental Fig. S2B).
Drosophila Unique Region of Top3␣ C Terminus Is Not Required for dHJ Substrate Dissolution-To investigate the role of CT2 in the context of dHJ dissolution, we assayed the Top3␣⌬CT2 construct. This enzyme is more comparable to FIGURE 2. A, full-length Top3␣ and Top3␣NT display comparable relaxation activity. Top3␣ could relax a hyper-negatively supercoiled (HNSC) DNA to the negatively supercoiled (NSC) level but not to the fully relaxed (R) state. B, whereas full-length Top3␣ can support dissolution of the dHJ substrate, Top3␣NT shows a significant reduction in dissolution activity. NE, no enzyme. Reaction products were digested with BamHI and run on an acrylamide gel. A diagram of the dHJ substrate and reaction products is shown to the right. Reactions are quantified in the graph below based on three independent experiments. C, stimulation by RPA is assisted by the Top3␣ C terminus. Dissolution reactions were conducted with varied amounts of RPA, and the amount of product was quantified from at least three independent experiments. human Top3␣ in length and composition. Similar to Top3␣NT, Top3␣⌬CT2 was active in relaxation, with a specific activity comparable to that of the wild type (Fig. 4A). When added at equal activity, Top3␣⌬CT2 was able to support dHJ substrate dissolution similar to wild-type levels (Fig. 4B). Therefore, Drosophila Top3␣⌬CT2 is capable of recapitulating the in vitro activity of the full-length enzyme.
Top3␣CT Has Important in Vivo Role-Top3␣ is essential in flies but can be fully rescued by a transgene expressing a fulllength Top3␣-YFP construct (26). To test the in vivo role of the C terminus, a Top3␣NT-YFP-expressing transgene was constructed. The Top3␣NT-YFP construct failed to rescue the viability of the top3␣ null line (supplemental Table S1).
Although CT2 does not appear to be necessary in vitro, the region is highly conserved among Drosophila species (supplemental Fig. S1). To determine whether CT2 is necessary in vivo, a Top3␣⌬CT2-YFP-expressing transgene was constructed. A low percentage of rescue was seen with the Top3␣⌬CT2 construct (Table 1). Compared with control flies (top row), a single copy of the Top3␣⌬CT2-expressing transgene rescued only ϳ5% of the homozygous top3␣ null flies, whereas two copies rescued ϳ10%. This low level of rescue efficiency indicates that CT2 has critical in vivo functions, although the Top3␣⌬CT2 construct still retains partial functionality.
Top3␣ C Terminus Is Involved in Binding DNA-Dissolution requires a tripartite complex of Top3␣, Blm, and DNA. To determine whether Top3␣CT contributes to DNA interactions, the Top3␣ constructs were tested for DNA-binding activity using an EMSA with a series of radiolabeled DNA constructs (supplemental Table S2). Full-length Top3␣, Top3␣NT, and Top3␣CT were incubated with single-stranded, doublestranded, and static-X (a non-migratable Holliday junction) DNAs as described previously (27). The full-length enzyme showed almost complete binding of all three substrate types (Fig. 5A). In contrast, Top3␣NT showed very little binding above the background, with only the ssDNA showing a significant level of binding. This region of the protein contains type IA topoisomerase domains 1-4, which are known to interact with ssDNA (28). Top3␣CT showed significant binding on all substrates. Although not equivalent to the full-length enzyme, Top3␣CT contributed significantly to DNA binding. To test which subdomains are involved, the various Top3␣CT constructs were tested on ssDNA. CT1,3 was the only subregion able to bind strongly (Fig. 5B), indicating that CT1 and CT3 work together.
The Top3␣CT region contains four putative zinc fingers and a zinc knuckle that could contribute to DNA binding (Fig. 1). To test for the importance of the zinc fingers in DNA binding, strong binders were incubated with either 1,10-phenanthroline, which is a zinc chelator, or ethanol, as the phenanthroline is dissolved in ethanol. Full-length Top3␣, Top3␣CT, and CT1,3 all showed substantially reduced binding in the presence of the zinc chelator (Fig. 5C), indicating that the zinc ion is essential for binding. Full-length Top3␣ retained some binding, presumably contributed from the active site region (N terminus), which does not contain zinc fingers and can independently bind ssDNA (Fig. 5A). Complete binding of full-length Top3␣ to ssDNA could be restored by washing away the phenanthroline and adding back zinc (supplemental Fig. S3). Taken together, these data indicate that Top3␣CT and its zinc fingers are major contributors to DNA binding.

DISCUSSION
Despite possessing seemingly identical relaxation activity, Top3␤ cannot substitute for Top3␣ in vivo (20,23) or in dHJ dissolution (2). The main structural difference between the two isoenzymes lies in their C termini. Therefore, we investigated the necessity and function of the Drosophila Top3␣ C terminus.
The relaxation activity of Top3␣NT, which truncated the C terminus, was not significantly altered. However, dissolution of the dHJ substrate, a model recombination intermediate, was greatly reduced. Top3␣NT was unable to interact with Blm directly, whereas Top3␣CT could, indicating that Top3␣CT provides a necessary platform for protein-protein interaction. Top3␣CT is also involved in the stimulation of dissolution activity by RPA, as truncated enzyme did not display proportional stimulation. The stimulation by RPA could be related to the interaction with Blm, as Blm and RPA are known to interact directly (29), and Top3␣CT provides the interaction with Blm.
The Drosophila Top3␣ C terminus contains an insert that is not homologous to the tails of other metazoans. This insert was used to divide the Drosophila Top3␣ C terminus for further study. The insert was termed CT2, and the divided pieces were called CT1 and CT3. CT1,3, which directly connects sections CT1 and CT3, is most homologous to the Top3␣ C-terminal domain of other metazoans. CT1,3 was capable of binding Blm, which is consistent with the notion that Top3␣ from other species can interact with Blm. However, CT2 was also able to bind Blm independently. This insert may be stabilizing the Top3␣-Blm complex in place of Rmi1 and Rmi2, structural proteins found in the complex in other species that have so far not been found in Drosophila. In the species we studied, possession of CT2 consistently corresponds with a lack of Rmi proteins. Although CT1,3 and CT2 each had Blm-binding activity, CT1 and CT3 individually showed little activity in binding Blm or DNA. It is likely that CT1 and CT3 come together three-dimensionally to form a proper scaffold for binding. However, future studies are needed to determine the structural conformation of this region.
Drosophila Top3␣⌬CT2, which contains both the enzyme active site and the conserved CT1,3-binding region, was capable of performing dHJ dissolution at levels equal to the wildtype enzyme, indicating that CT2 is not required for dHJ dissolution in vitro. This mutant closely resembles the human and other metazoan Top3␣ enzymes, and therefore, the ability to perform dHJ dissolution is not unexpected. However, CT2 contains important in vivo functions, which are not apparent in our in vitro assays. When the Top3␣⌬CT2 mutant was inserted into flies, it was able to only partially rescue top3␣ null flies, and rescuing ability was correlated with the gene copy number of Top3␣⌬CT2. As Drosophila appears to lack the structural proteins Rmi1 and Rmi2, this insert may be contributing to stabilization of the Top3␣-Blm complex.
Top3␣CT also plays an extensive role in binding to DNA substrates. Although the catalytic core binds ssDNA as part of its activity, the majority of the observed ssDNA binding of the enzyme was due to Top3␣CT. The tail region also contributed the majority of the binding for double-stranded and static-X substrates. Top3␣CT is expected to bind DNA due to its putative zinc fingers, which align with those of other metazoans. When all zinc fingers were present, as in CT1,3, they were able to provide significant binding. The addition of a zinc chelator significantly reduced this binding activity, indicating that the protein is coordinating zinc ions to bind DNA substrates.
Recently, two models have been proposed for the molecular process of dHJ dissolution (30). The "unravel and unlink" model proposes a sequential reaction in which a region of ssDNA is first created between the junctions, followed by coordinated separation of the strands and renaturation. In the "HJ migration" model, the two enzymes are arranged in a complex such that Top3␣ is able to perform strand passage concomitant with the helicase migrating the junctions. Both models require concerted activity of the two enzymes. Our data reinforce the idea that the two proteins need to be coordinated for dissolution to occur. In the absence of interaction, only limited dissolution occurs. Because some limited dissolution is possible when RPA is present but Blm interaction is limited or missing (as with Top3␣NT), this may indicate that the functions of the two enzymes may be separated, as in the unravel and unlink model. In this situation, Blm may be acting first to unwind the intervening strands, which are then held in the melted state by RPA until the topoisomerase can bind and act. This would explain why dHJ dissolution with Top3␣NT is not above the background in the absence of RPA (Fig. 2C). It is also possible that the optimal situation (when full-length topoisomerase is pres-  Further study will be needed to distinguish the exact mode of action of the complex. The C terminus of Top3␣ appears to provide an important interface for protein-protein and protein-DNA interactions. These interactions are critical for the in vitro biochemical reactions, including dissolution of dHJs, as well as for the in vivo functions of the enzyme.