Using a Biochemical Approach to Identify the Primary Dimerization Regions in Human DNA Topoisomerase IIα*

Eukaryotic topoisomerase II is a nuclear enzyme essential for DNA metabolism and chromosome dynamics. The enzyme has a dimeric structure, and subunit dimerization is vital to the cellular functions and activities of the enzyme. Two biochemical approaches based on metal ion affinity chromatography and immunoprecipitation have been carried out to map the dimerization region(s) in human topoisomerase IIα. The results demonstrate that two regions spanning amino acids 1053–1069 and 1124–1143 are both essential for dimerization. The regions correspond to the interaction domains revealed in yeast topoisomerase II after crystallization of a central fragment of this enzyme, indicating that the overall C-terminal dimerization structure of eukaryotic topoisomerase II is conserved from yeast to human. Furthermore, linker insertion analysis has demonstrated that the two dimerization regions are located in a highly flexible part of the enzyme. Topoisomerase IIα mutant enzymes unable to dimerize via the C-terminal primary dimerization regions due to lack of one of the defined dimerization regions can still be forced to dimerize if DNA and an ATP analog are added to the reaction mixture. The result indicates that secondary interactions occur by ATP analog-mediated clamp closing when the subunits are brought together on DNA.

Eukaryotic topoisomerase II is a nuclear enzyme essential for DNA metabolism and chromosome dynamics. The enzyme has a dimeric structure, and subunit dimerization is vital to the cellular functions and activities of the enzyme. Two biochemical approaches based on metal ion affinity chromatography and immunoprecipitation have been carried out to map the dimerization region(s) in human topoisomerase II␣. The results demonstrate that two regions spanning amino acids 1053-1069 and 1124 -1143 are both essential for dimerization. The regions correspond to the interaction domains revealed in yeast topoisomerase II after crystallization of a central fragment of this enzyme, indicating that the overall C-terminal dimerization structure of eukaryotic topoisomerase II is conserved from yeast to human. Furthermore, linker insertion analysis has demonstrated that the two dimerization regions are located in a highly flexible part of the enzyme. Topoisomerase II␣ mutant enzymes unable to dimerize via the C-terminal primary dimerization regions due to lack of one of the defined dimerization regions can still be forced to dimerize if DNA and an ATP analog are added to the reaction mixture. The result indicates that secondary interactions occur by ATP analog-mediated clamp closing when the subunits are brought together on DNA.
Eukaryotic topoisomerase II is a nuclear enzyme involved in regulating the topological conformation of DNA (1,2). It fulfills essential functions during DNA replication (3,4) and chromosome segregation in both mitosis (5,6) and meiosis (7), and it is thought to play a key role in certain types of DNA recombination events (8 -12). The enzyme has furthermore been suggested to constitute a component of the nuclear scaffold (13,14), where it is involved in chromosome condensation (15) and decondensation (16).
Mammalian cells contain two structurally similar but biochemically distinct topoisomerase II isoforms, topoisomerase II␣ and II␤, existing in either a homo-or heterodimeric configuration (17)(18)(19). The dimeric nature of topoisomerase II is essential to enzyme function. Thus, during the catalytic cycle of the enzyme, it introduces a transient double-strand break into the DNA backbone, where the subunits become covalently linked to the 5Ј-ends of the respective DNA strands through O 4 -phosphotyrosine bonds. A second DNA duplex is then transported through the cleaved DNA before it is finally religated (1,20,21).
The functional implications of a dimeric nature of eukaryotic topoisomerase II have been supported by several observations. Hydrodynamic studies indicate that the enzyme exists in solution as a dimer (18,19,22), and this has further been strengthened by studies of topoisomerase II heterodimers, which have been found to be highly stable in vitro without the ability to dissociate and produce enzymes with a homodimeric conformation (19,23). Furthermore, structural data obtained either from crystallization of a 92-kDa fragment of yeast topoisomerase II or from electron microscopy studies of human topoisomerase II␣ have demonstrated a dimeric nature of these enzymes (24,25). Despite the strong evidence in favor of a dimeric structure of eukaryotic topoisomerase II, biochemical data on a mapping of the exact regions involved in subunit interaction are still lacking.
Early analysis of the amino acid sequence of human topoisomerase II␣ revealed the presence of a leucine zipper (26), a motif that has been shown to selectively direct homo-and heterodimerization of proteins by forming a coiled coil structure, and it was suggested that topoisomerase II dimerization could be mediated through this motif. The importance of the leucine zipper in topoisomerase II dimerization has, however, been questioned by the absence of a similar motif in human topoisomerase II␤ (27), as well as by results obtained by Kroll et al. (28), showing that disruption of the leucine zipper did not have any effect on topoisomerase II dimerization. The same group detected by use of topoisomerase II␣ fragments in far Western blotting a region covering amino acids 951-1042 to constitute a minimal subunit association region, whereas maximal homodimerization required sequences C-terminal to position 1042 (29). Similar, another study, using a small human topoisomerase II␣ peptide with the ability to adopt a coiled coil structure, has demonstrated that this peptide, covering amino acids 1013-1056, forms a stable homodimer in solution (30).
The most compelling data on eukaryotic topoisomerase II dimerization so far have derived from resolution of the crystal structure of a 92-kDa fragment of yeast topoisomerase II (31). This structure shows a primary interaction domain in the C-terminal part of the fragment involving two minor regions covering amino acids 1031-1046 and 1114 -1130 in the yeast topoisomerase II enzyme.
In the present study, we have mapped the regions involved in primary dimerization of human topoisomerase II␣ using two biochemical approaches based on metal ion affinity chromatography and immunoprecipitation. The defined sequences are located in the C-terminal region of the enzyme and correspond to the interaction domains discovered in yeast topoisomerase II from crystallization of the central fragment of the yeast en-zyme. A further investigation of human topoisomerase II␣ mutant enzymes lacking the primary C-terminal dimerization region has shown that such enzymes are still capable of dimerization via secondary dimerization regions if DNA and an ATP analog are present.
Yeast Transformation and Complementation-Yeast cells were transformed by using a modified version of the LiAc method of Ito et al. (34). To test the ability of the mutated topoisomerase II constructs to complement the lack of endogenous topoisomerase II in BJ201 the LEU2-based constructs were transformed into BJ201, and cells were transferred to medium plates containing 5Ј fluoro-orotic acid (1 mg/ml) to select against the URA3 plasmid carrying the Schizosaccharomyces pombe TOP2 gene (33).
Metal Ion Affinity Chromatography-Single-and double-transformed yeast cells were grown in 50 ml of selection medium to late log phase and then transferred to 1 liter of YPD medium and grown for approximately 12 h before harvesting by centrifugation. Yeast cells were extracted with 2 volumes of extraction buffer (50 mM Tris-HCl, pH 7.8, 1 M NaCl, 1 mM phenylmethylsulfonyl fluoride) and 1 volume of acid-washed glass beads (425-600 m, Sigma) by vortexing at 4°C for 30 min. Glass beads and cell debris were removed by centrifugation at 3000 rpm for 10 min, and the supernatant was further centrifuged for 20 min at 12000 rpm. After filtration (0.65 m pore size filters), the crude extract was loaded on a 2-ml Ni 2ϩ -nitrilotriacetic acid agarose column (matrix purchased from Qiagen) equilibrated with equilibration buffer (20 mM imidazole, pH 8, 1 M NaCl, 10 mM KPi, pH 8, 10% glycerol). The column was subsequently washed with equilibration buffer, and protein was eluted with elution buffer (1 M imidazole, pH 8, 1 M NaCl, 10 mM KPi, pH 8, 10% glycerol) in a gradient of imidazole ranging from 20 mM to 1 M. All chromatographic steps were performed using an Amersham Pharmacia Biotech FPLC system under the control of an LCC-500 plus. When metal ion affinity chromatography was used for enzyme purification, yeast cell extraction and the following Ni 2ϩ column chromatography were performed as described above, except that all buffers contained 500 mM NaCl rather than 1 M. Partly purified enzymes were stored in elution buffer supplemented with 50% glycerol.
Antibodies and Immunostaining-The C-terminally tagged human topoisomerase II␣ mutants were detected by the mouse monoclonal antibody MYC1-9E10.2 (diluted 1:500), purchased from Genosys. Untagged human topoisomerase II versions were recognized by the antitopoisomerase II␣ antibody CRB (Genosys) raised against the C-terminal peptide (RAKKPIKYLEESDEDLF) of the wild-type human topoisomerase II␣ enzyme (diluted 1:5000). For immunostaining, samples collected from the Ni 2ϩ column were subjected to SDS-PAGE 1 on 8.5% polyacrylamide gels. Proteins were transferred to nitrocellulose filters (Schleicher & Schü ll, 0.45 m), and immunostaining was carried out according to the ECL protocol (Amersham Pharmacia Biotech). Horseradish peroxidase-conjugated antibodies were used as secondary antibodies (Jackson Immunochemicals). For studies of secondary dimerization, immunostaining of immunoprecipitated material was performed using the penta-His antibody (Qiagen).
Immunoprecipitation-Immunoprecipitation of human topoisomerase II␣ was carried out using Dynabeads-280 sheep anti-rabbit IgG (Dynal) according to the manufacturer's instructions. Briefly, 50 l of beads were incubated with the anti-topoisomerase II␣ antibody CRB (2 g) overnight followed by extensive washing with phosphate-buffered saline buffer (1 mM NaH 2 PO 4 , 5 mM Na 2 HPO 4 , 140 mM NaCl, 0.1% bovine serum albumin). The beads were then incubated with 500 l of yeast extract containing the different mutated topoisomerase II enzymes for 2-4 h at 4°C with gentle shaking. Following extensive wash, 60 l of Laemmli loading buffer was added, and the samples were boiled for 2-5 min before being loaded onto 8% SDS protein gels. Immunostaining was carried out with the MYC1-9E10.2 antibody. Immunoprecipitation using partly purified topoisomerase II␣ mutant enzymes for studies of secondary dimerization was performed using protein A-Sepharose. 250 mg of protein A-Sepharose was extensively washed in distilled water followed by three washes in immunoprecipitation buffer (50 mM Tris-HCl, pH 8, 0.35% Triton X-100, 1 mM EDTA, 100 mM NaCl). The Sepharose was finally dissolved in 5 ml of immunoprecipitation buffer. For each immunoprecipitation reaction, 100 l of Sepharose was coated with 2 g of anti-topoisomerase II␣ antibody (CRB) by incubation overnight at 4°C with gentle shaking. Coated Sepharose was next washed in immunoprecipitation buffer and used for immunoprecipitation by adding the h⌬1053-1069 enzyme. Following incubation for 1-2 h at 4°C with gently shaking, the h⌬C1121 mutant enzyme was added either alone or together with DNA (28-mer duplex containing a strong topoisomerase II recognition sequence (8)) and/or an ATP analog (AMP-PNP, Roche Molecular Biochemicals; final concentration, 500 M), and incubation was continued for 1-2 h before precipitation by centrifugation at 5000 rpm. The NaCl concentration was throughout the immunoprecipitation reaction kept at 100 mM. To avoid unspecific interaction, the precipitated material was washed four times in immunoprecipitation buffer supplemented with 1 M NaCl and 0.1% SDS and one time in 10 mM Tris-HCl, pH 7.5. The pellet was further prepared for SDS-PAGE as described above.
Topoisomerase II-mediated DNA Decatenation-The activity of topoisomerase II␣ mutant enzymes was tested by in vitro decatenation of kinetoplast DNA (kDNA) using yeast extract overexpressing the topoisomerase II enzyme of interest. Extract was incubated with 300 ng of kDNA in 1 mM ATP, 5 mM MgCl 2 , 1 mM dithiothreitol, 0.5 mM EDTA, 50 mM bis-Tris-propane (pH 8), 120 mM potassium glutamate in a total volume of 20 l. The samples were incubated for 30 min at 37°C and subsequently subjected to electrophoresis on a 0.7% agarose gel. The contribution of the endogenous yeast topoisomerase II enzyme to decatenation is negligible, as no decatenation activity was seen in yeast extracts when overexpression of the human topoisomerase II␣ enzymes was omitted.

Metal Ion Affinity Chromatography as a Biochemical Method to Identify Interactions between Human DNA Topoisomerase
II␣ Subunits-In eukaryotic topoisomerase II, subunit dimerization is a key process that is essential for the function of this highly complex enzyme. To map the regions that participate in dimerization in human topoisomerase II␣, we have taken a biochemical approach employing a previously developed protein-protein interaction assay, which takes advantage of the affinity of metal ions for histidine-containing proteins (19). The assay involves coexpression in yeast of two versions of topoisomerase II␣, one of which is His-tagged, followed by analysis of the extract by metal ion affinity chromatography. The strategy behind the assay is that an untagged topoisomerase II enzyme will be retained on a Ni 2ϩ column only if it has interacted with a His-tagged version. A positive interaction can afterward be visualized if the Ni 2ϩ column fractions are examined by immunostaining.
To first verify the specificity of the assay, a human topoisomerase II␣ mutant enzyme truncated at residue 1233 (h⌬C1233) and fused at its C-terminal end to a bicomposite tag consisting of a hexahistidine tail, and a c-Myc epitope was 1 The abbreviation used is: PAGE, polyacrylamide gel electrophoresis. expressed in the yeast strain BJ201. Crude extract was subjected to metal ion affinity chromatography under highly stringent conditions, and samples of the eluted material were analyzed by SDS-PAGE. Immunostaining with anti-c-Myc antibody demonstrated the presence of the histidine-tagged h⌬C1233 topoisomerase II␣ mutant in the elution profile ( Fig.  1, top panel). In contrast, when an untagged version of wildtype human topoisomerase II␣ was analyzed in the dimerization assay, it exhibited no detectable affinity for binding to the nickel column, as evident from the immunostain shown in Fig.  1 (bottom panel), in which the enzyme is visualized by the anti-topoisomerase II␣ antibody. The assay is thus highly specific for topoisomerase II enzymes containing a hexahistidine tag, and the method therefore provides a powerful tool to investigate possible interactions between tagged and untagged topoisomerase II subunits.
Importance of the C-terminal Region of Human Topoisomerase II␣ in Subunit Dimerization-Several biochemical studies have implied an important role for the C-terminal region of eukaryotic topoisomerase II in mediating subunit interaction (30,33,35,36), although direct biochemical evidence is still lacking. As an initial step to map the dimerization region of human topoisomerase II␣, extracts from yeast cells coexpressing untagged wild-type human topoisomerase II␣ and a Hisand c-Myc-tagged C-terminal truncation mutant of the enzyme deleted at either amino acid 1233 or 1121 were analyzed in the described dimerization assay ( Fig. 2 and Table I). When the wild-type enzyme was expressed together with h⌬C1233, the anti-c-Myc antibody identified h⌬C1233 as a constituent of the column eluate, as expected. The wild-type form of human topoisomerase II␣ was retained on the column as well and eluted in the same fractions as the truncated form, as demonstrated by staining with the anti-topoisomerase II␣ antibody (Fig. 2, top panel). It has previously been shown that under the conditions employed, unspecific interaction, as well as multimerization, does not occur at least to a detectable level between topoisomerase II dimers (19). The coexistence of the truncated and wild-type forms in the elution profile therefore can only be a result of an interaction between the two different subunits, demonstrating that h⌬C1233 still carries the region essential for dimerization. When extract from yeast cells coexpressing wild-type human topoisomerase II␣ and the C-terminal truncation mutant deleted at residue 1121 (h⌬C1121) was analyzed in the dimerization assay, only the tagged h⌬C1121 mutant was eluted (Fig. 2, bottom panel). The h⌬C1121 enzyme is therefore unable to participate in a heterodimeric configuration, suggesting that the region spanning residues 1121-1233 in human topoisomerase II␣ has an important role in directing subunit interaction. However, the obtained result does not rule out the possibility that h⌬C1121 is able to form homodimers. Furthermore, the possibility cannot be excluded that the observed lack of heterodimerization is caused by an altered or destroyed folding of the h⌬C1121 enzyme.
Two Minor Regions Located in the C-terminal Part of Topoisomerase II␣ Are Essential for Subunit Interaction-The interpretation of the results obtained with the C-terminal truncated topoisomerase II enzymes is that a region residing between amino acids 1121 and 1233 is required for subunit interaction in human topoisomerase II␣. A further dissection of the dimerization region(s) was performed based on information derived from the crystallization of a 92-kDa central fragment of yeast topoisomerase II (31). As evident from the crystal structure a major dimerization interface is present in the C-terminal region of the yeast enzyme involving residues 1031-1046 and 1114 -1130. Because dimerization is a highly conserved feature of eukaryotic topoisomerase II, it is likely that subunit contacts will be mediated by corresponding regions in topoisomerase II enzymes of different eukaryotic origin. Therefore, a human topoisomerase II␣ mutant was first constructed carrying a deletion of amino acids 1124 -1143 (h⌬1124 -1143) equivalent to the most C-terminal dimerization region observed in the crystal structure of yeast topoisomerase II (residues 1114 -1130). This mutant was coexpressed in yeast together with the histidine-tagged, truncated form of human topoisomerase II␣, h⌬C1233, which still contains the potential to dimerize. In the experiment, h⌬C1233 was preferred to the wild-type enzyme, as this mutant can easily be distinguished from the deletion mutant upon SDS-PAGE due to its smaller size. Analysis of the crude yeast extract in the dimerization assay revealed that no heterodimer formation takes place between h⌬C1233 and the deletion mutant h⌬1124 -1143 (Fig. 3, top panel), because only the h⌬C1233 enzyme was retained on the column. The result strongly suggests that the deletion mutant no longer contains the region required for subunit interaction. This is further supported by a failure of h⌬1124 -1143 to sustain mitotic growth of the top2 deletion strain BJ201, as well as an inability of the mutant enzyme to decatenate kinetoplast DNA, demonstrating that h⌬1124 -1143 lacks a feature necessary for enzy- matic activity (Table I).
The observation underscores the result obtained from the experiment employing the h⌬C1121 mutant (Fig. 2). In contrast to this severely C-terminal truncated mutant, the dele-tion mutant only lacks 20 amino acids, and it is highly unlikely that this deletion would obstruct the overall tertiary conformation of the enzyme. Therefore, the dimerization inability of h⌬C1121 is most probably due to lack of a region essential for FIG. 2. Regions involved in subunit interaction are located in the C-terminal domain of human topoisomerase II␣. Extract prepared from yeast cells co-expressing a His-tagged C-terminal truncated version and an untagged wild-type form of human topoisomerase II␣ was analyzed by immunostaining following metal ion affinity chromatography. Top panel, extract contains the His-tagged h⌬C1233 truncation mutant in addition to the wild-type enzyme. Bottom panel, extract contains the His-tagged h⌬C1121 truncation mutant and the wild-type enzyme. In both cases, immunostaining was performed with a combination of the anti-c-Myc and anti-topoisomerase II␣ antibodies. Column fractions are indicated above the immunostainings. Although h⌬C1121 gives a strong signal in the eluted material, the enzyme gives only a weak signal in the extract due to a relatively low expression of this truncated enzyme.

TABLE I Analysis of C-terminal mutants of human topoisomerase II␣
Schematic representation of the human topoisomerase II␣ constructs employed. The names of the constructs are indicated to the left, where the numbers refer to amino acid residues in topoisomerase II. wt-yTOPII and wt-hTOPII illustrate wild-type yeast and human topoisomerase II␣, respectively. C-terminal truncation mutants of human topoisomerase II␣ are denoted by h⌬C followed by a number reflecting the truncation position. Mutant h1121-12 contains a 12-amino acid linker, indicated as a solid bar, inserted at amino acid position 1121. Constructs with internal deletions are illustrated as h⌬ followed by numbers indicating the deleted amino acids. The enzymes are divided into three major domains, where the grey box represents the N-terminal ATPase domain, the open box depicts the central cleavage/religation domain, and the hatched box illustrates the divergent C-terminal region. The crystallized fragment of yeast topoisomerase II is indicated by the black line below the schematic representation of S. cerevisiae topoisomerase II. The two minor dimerization regions in yeast topoisomerase II and human topoisomerase II␣ are shown as black vertical lines. Deletion of one or both of these primary dimerization regions in human topoisomerase II␣ is indicated by ⌬ and a replacement of the black line with a grey line. Constructs fused to a bicomposite tag consisting of the c-Myc epitope, and a hexahistidine tail is marked with Ϫc-Myc-His 6 . The abilities of all constructs to mediate dimerization, to complement mitotic growth of the yeast strain BJ201, and to decatenate kinetoplast DNA are indicated by ϩ and Ϫ signs. nd, not determined. subunit interaction rather than incorrect folding.
To establish whether the region corresponding to amino acids 1031-1046 in the yeast enzyme also influences dimerization in human topoisomerase II␣, a second deletion mutant, deleted from amino acid 1053 to 1069, was constructed and tested in the dimerization assay together with h⌬C1233. From the immunostain shown in Fig. 3 (middle panel), it is seen that only the h⌬C1233 enzyme was retained on the column, revealing that the h⌬1053-1069 deletion mutant had also lost its dimerization ability. In correlation with these data, deletion of residues 1053-1069 abolished the ability of the deletion mutant to complement the top2 null strain BJ201 (Table I). From these observations, we conclude that the region from amino acid 1053 to 1069 plays an essential role in dimerization. Thus, like in the yeast enzyme, two minor dimerization regions exist in the C-terminal end of human topoisomerase II␣, where lack of either one of these regions or both ( Fig. 3 and Table I) is detrimental for dimerization. Our biochemical data therefore leads support to the structural information derived from the crystal structure. Moreover, the overall organization of subunit contacts seems to be conserved from yeast to human.
To address whether an exact spacing between the two primary dimerization regions in human topoisomerase II␣ is a prerequisite for dimer formation, we have investigated the dimerization potential of a mutant containing a linker inserted at position 1121. In this mutant, a 12-amino acid linker including two proline residues further separates the two subdomains involved in dimerization (33). The linker mutant h1121-12 was tested for its ability to interact with h⌬C1233 in the dimerization assay (Fig. 3, bottom panel; Table I). Immunostaining with the anti-topoisomerase II␣ antibody together with the anti-c-Myc antibody identifies two distinct bands of the appropriate sizes in the elution profile (Fig. 3, bottom panel). Because this result demonstrates that insertion of an extensive linker between the two dimerization regions is tolerated without functional consequences on the dimerization ability of the enzyme, a correct positioning of the two minor dimerization domains is not required for proper subunit interaction. It cannot be ruled out that the introduced peptide in the linker insertion mutant loops out in a folded conformation due to the presence of two proline residues, thereby conserving the spacing between the two dimerization regions. However, insertion of either two or seven amino acids, designed not to permit outlooping, at position 1121 in human topoisomerase II␣, does not affect the ability of the enzyme to complement the top2 deletion strain BJ201 (33), demonstrating that these mutant enzymes are still in a dimeric configuration (data not shown). Thus, dimerization of human topoisomerase II␣ is resistant to alterations in the length separating the two dimerization regions, suggesting a more flexible structure of the C-terminal part of the enzyme.
Recently, a fragment spanning residues 1109 -1163 in S. cerevisiae topoisomerase II was found to interact with the enzymes Sgs1 and Pat1 known to be involved in DNA metabolism (37,38). It is a somewhat surprising observation that more or less the same regions in S. cerevisiae topoisomerase II are involved in subunit contact as well as in directing other protein-protein interactions. This lends further support to a more accessible and flexible structure of the C-terminal domain in eukaryotic topoisomerase II.
In human cells two topoisomerase II isoforms are present, which can exist in either a homo-or heterodimeric configuration. To investigate whether the regions responsible for homodimerization of human topoisomerase II␣ are identical to the regions through which this isoform interacts with the ␤ isoform, the untagged human topoisomerase II␣ mutant h⌬1053-1069 was coexpressed in yeast together with a histidine-tagged wild-type version of human topoisomerase II␤. Analysis of the extract in the dimerization assay demonstrates that the tagged topoisomerase II␤ enzyme is unable to retain the mutated human topoisomerase II␣ enzyme on the column (data not shown). Thus, a topoisomerase II␣ mutant deleted in one of the primary dimerization regions is deficient in both homo-and heterodimerization, suggesting that the ␣ isoform can interact with either ␣ or ␤ through the same region(s).
Numerous studies have addressed the mechanism and nature of subunit interaction in eukaryotic topoisomerase II. As predicted from the human topoisomerase II␣ cDNA, it has been postulated that a potential leucine zipper motif spanning residues 994 -1021 plays a key role in mediating subunit interaction (26). However, point mutations disrupting the leucine re- peat sequence in the enzyme does not influence dimerization (28), which is in accordance with our demonstration of a more C-terminal localization of the dimerization regions. Also, a leucine zipper motif is lacking in the human topoisomerase II␤ isoform (27), which together with the natural occurrence of ␣/␤ heterodimers in human cells further rules out the importance of the leucine zipper in human topoisomerase II dimerization (19).
Recently, an in vitro study employing human topoisomerase II␣ fragments defined a minimal region between residues 951 and 1042 to be required for homodimerization, whereas amino acids C-terminal to position 1042 were found to be necessary for maximal subunit interaction (29). In addition, a peptide fragment covering amino acids 1013-1056 in the human ␣ enzyme was found to form stable dimeric coiled-coil structures, indicating that this small fragment possesses the structural requirements needed for dimerization (30). These observations, however, are not completely compatible with the structural data derived from the crystallization of yeast topoisomerase II and with our biochemical dimerization data on human topoisomerase II␣, where the identified interaction regions are found more C-terminally. The discrepancy in these results might be due to use of only small peptide fragments of human topoisomerase II␣ in the two former in vitro studies. It is possible that such fragments may not be folded in the same way and thereby not exposed when present in the full-length enzyme. Furthermore, in the study by Kroll (29), the methods employed to detect dimerization between topoisomerase II fragments involved conditions of low stringency such as 50 mM NaCl. In contrast, the in vivo dimerization assay applied in our study is carried out in the presence of 1 M NaCl to avoid unspecific interactions as well as topoisomerase II multimerization. It cannot be excluded that our assay might be too stringent to detect weak interactions between subunits of topoisomerase II, which could be identified in the above described in vitro studies. Given that the regions observed from the in vitro experiments locate in the vicinity of the dimerization domains mapped in our study and to those in the yeast crystal, it is likely that they play an important role in stabilizing subunit interaction in human topoisomerase II␣.
Immunoprecipitation Verifies the Requirement of Two Minor Regions for Subunit Dimerization in Human Topoisomerase II␣-To confirm the above data and to provide an alternative line of evidence for the localization of the two dimerization domains in human topoisomerase II␣, immunoprecipitation experiments were carried out on topoisomerase II from crude yeast extract. For this purpose, extract was prepared from cells expressing the combination of two topoisomerase II enzymes previously tested in the dimerization assay. One of the components is either one of the His-c-Myc-tagged C-terminally truncated enzymes, h⌬C1233 or h⌬C1121, and the other component is the wild-type enzyme, one of the deletion mutants, h⌬1053-1069 and h⌬1124 -1143, or the linker insertion mutant h1121-12. In all cases, the extract was incubated with Dynabeads previously coated with the anti-topoisomerase II␣ antibody recognizing the 15 outermost C-terminal residues in human topoisomerase II␣. If both topoisomerase II enzymes present in the extract contain the regions involved in subunit interaction, the enzymes recognized by the antibody coupled to the beads will allow co-immunoprecipitation of the truncated topoisomerase II enzyme. The presence of the truncation mutant in the immunoprecipitated material can subsequently be visualized by immunostaining using the c-Myc antibody. As shown in Fig.  4, the truncated h⌬C1233 mutant enzyme co-precipitates with wild-type human topoisomerase II␣ (lane 1). When further amino acids are deleted from the C-terminal region to position 1121, the truncated form is no longer precipitated by the wildtype enzyme, indicating a loss of dimerization (lane 2). However, the linker insertion mutant h1121-12 is able to precipitate the h⌬C1233 mutant (lane 3), verifying that dimerization is not abolished by insertion of 12 amino acids between the two proposed minor dimerization regions. Finally, neither the h⌬1053-1069 nor the h⌬1124 -1143 deletion mutant contains the capacity to bring down the truncated h⌬C1233 mutant (lanes 4 and 5), so these deletion mutants are deficient in mediating subunit interaction. The immunoprecipitation experiments were carried out in the presence of 1 M NaCl, providing high stringency and thereby eliminating unspecific interaction or multimerization between the different topoisomerase II enzymes. Moreover, the anti-topoisomerase II␣ antibody used for immunoprecipitation is highly specific (19), and immunoprecipitation performed with rabbit preimmune serum provides an additional control (lane 6). Thus, the immunoprecipitation data are in correlation with the results obtained from the dimerization assay.
Secondary Dimerization in Human Topoisomerase II␣ Mutant Enzymes Lacking the Primary Dimerization Region-Biochemical studies by Wang and co-workers (39 -41) have demonstrated that topoisomerase II operates as an ATP modulated protein clamp. Based on this work as well as information obtained from studies of the homologous gyrase enzyme, a model has been presented in which an ATP-dependent dimerization region exists in the far N-terminal part of eukaryotic topoisomerase II besides the primary dimerization regions in the C-terminal end. Permanent clamp closure was only detected in the presence of an ATP analog, indicating that ATP binding results in clamp closing, whereas ATP hydrolysis causes enzyme turnover and reopening of the clamp. Another N-terminal interaction face has been suggested from elucidation of the crystal structure of the yeast enzyme, where also this interaction is believed to be ATP-dependent, as well as DNA-dependent (31). Electron microscopy studies have demonstrated that a similar overall conformation, including several dimerization faces, is present in the human topoisomerase II␣ enzyme (25).
To test whether mutant topoisomerase II␣ enzymes lacking the primary C-terminal dimerization regions are able to dimerize in the presence of an ATP analog due to the existence of a secondary dimerization region(s), an immunoprecipitation experiment was performed using h⌬1053-1069 and h⌬C1121, each lacking one of the C-terminal dimerization regions (Fig.  5). In the immunoprecipitation reactions the h⌬1053-1069 enzyme was incubated with protein A-Sepharose coated with the anti-topoisomerase II␣ antibody recognizing the outermost C- terminal region of topoisomerase II␣ before h⌬C1121 was added. Immunoprecipitation was performed in the absence or in the presence of an ATP analog and/or DNA to study the effect of these molecules on the ability of h⌬C1121 to dimerize with h⌬1053-1069 through secondary dimerization regions. As seen from the immunostaining in Fig. 5, secondary dimerization between the two mutant enzymes only takes place if both DNA and an ATP analog is present (lane 1). Precipitation in the presence of ATP analog (lane 2) or DNA (lane 3) alone or in the absence of these cofactors (lane 4) can only be revealed after strong overexposure of the film presented in Fig. 5 and probably relates to unspecific protein-protein interactions. The fact that neither DNA nor the ATP analog by itself is able to promote dimerization suggests a model in which DNA is required to bring the two different enzyme subunits together, after which, the ATP analog, by its ability to stably close the N-terminal clamp between the two protein partners, ensures continuous dimerization. The absence of a considerable precipitation in the presence of an ATP analog alone suggests that the two mutant enzymes are not directly in contact with each other in the absence of DNA, because the ATP analog otherwise would be expected to make this interaction stable. In the case in which only DNA is present, the protein partners should be in contact according to the described model, but as the precipitated material is washed in 1 M NaCl prior to SDS-PAGE, the complex will dissociate from the DNA, and apparently, the secondary dimerization is too weak to withstand this high salt concentration in the absence of an ATP analog. Thus, our results are consistent with a DNA requirement for secondary dimerization of human topoisomerase II␣ enzymes lacking the primary C-terminal dimerization region, in which it is so far unknown whether DNA binding per se is sufficient or if DNA cleavage is required. Alternatively, our results can be ex-plained by DNA-independent dimerization followed by subunit exchange. In this case, the two C-terminal topoisomerase II␣ mutant enzymes exist as homodimers already at the time of purification, and heterodimerization is only occurring as a result of subunit exchange. If this holds true, our results indicate that a high level of DNA-dependent subunit exchange occurs when h⌬C1121 is added to h⌬1153-1169 coupled to protein A-Sepharose. However, we have earlier demonstrated that subunit exchange does not occur to a detectable level with human topoisomerase II␣ (19), which makes this explanation less favorable.
Recently, Maxwell and co-workers (42) presented results obtained with an N-terminal human topoisomerase II␣ fragment covering amino acids 1-435. By cross-linking experiments, the fragment was shown to be in a dimer configuration in the absence of DNA, no matter whether an ATP analog or ATP was present or not. The result suggests that secondary dimerization regions present in the N-terminal fragment allow stable interaction under the low salt conditions used in the experiment. The discrepancy between the results obtained by Maxwell and co-workers (42) and our results, in which DNA is fully required to obtain secondary dimerization, might be caused by the use of very different enzyme fragments in the two experiments. Thus, with the almost full-length enzymes used in our experiments, other regions in the enzymes that are absent in the N-terminal peptide might influence the secondary dimerization of the N-terminal regions. Furthermore, our precipitated material was washed in 1 M NaCl before analysis, which will disrupt weak protein interactions, but as discussed above, if the mutant enzymes dimerize in the absence of DNA, we would have expected to see the same amount of precipitation of h⌬C1121 whenever an ATP analog was added, no matter whether DNA is present or not.
Taken together, our study has demonstrated the presence of two small regions in the C-terminal domain of human topoisomerase II␣ spanning residues 1053-1069 and 1124 -1143, each of which is essential for homodimerization of the enzyme. The overall C-terminal dimerization structure is conserved from yeast to human, although the regions involved are located in a highly flexible domain of the enzyme. In our study, deleting part of the C-terminal interaction region abolished the dimerization ability of the enzyme, indicating that if secondary interfaces exist in the protein in the absence of ATP analogs, they are not sufficient to keep the enzyme in a dimeric form under the conditions employed here or, alternatively, that the Nterminal interactions are dependent on a functional C-terminal interaction. However, enzymes lacking the C-terminal dimerization regions can still be brought together if DNA is present. Such dimers are highly unstable in the absence of ATP analogs that ensure clamp closing of the N-terminal domain of the enzyme. Utilization of the biochemical dimerization assay described in this study may allow a dissection of the potential ATP-dependent secondary dimerization regions in human topoisomerase II␣. A complete knowledge on the dimerization faces will aid the further elucidation of the structural organization of topoisomerase II␣ and thereby provide valuable information on the catalytic mechanism of this complex enzyme.