The role of the carboxyl-terminal amino acid residues in Escherichia coli DNA topoisomerase III-mediated catalysis.

The role that the carboxyl-terminal amino acids of Escherichia coli DNA topoisomerase I (Topo I) and III (Topo III) play in catalysis was examined by comparing the properties of Topo III with those of a truncated enzyme lacking the generalized DNA binding domain of Topo III, Topo I, and a hybrid topoisomerase polypeptide containing the amino-terminal 605 amino acids of Topo III and the putative generalized DNA binding domain of Topo I. The deletion of the carboxyl-terminal 49 amino acids of Topo III decreases the affinity of the enzyme for its substrate, single-stranded DNA, by approximately 2 orders of magnitude and reduces Topo III-catalyzed relaxation of supercoiled DNA and Topo III-catalyzed resolution of DNA replication intermediates to a similar extent. Fusion of the carboxyl-terminal 312 amino acid residues of Topo I onto the truncated molecule stimulates topoisomerase-catalyzed relaxation 15-20-fold, to a level comparable with that of full-length Topo III. However, topoisomerase-catalyzed resolution of DNA replication intermediates was only stimulated 2-3-fold. Therefore, the carboxyl-terminal amino acids of these topoisomerases constitute a distinct and separable domain, and this domain is intimately involved in determining the catalytic properties of these polypeptides.

The role that the carboxyl-terminal amino acids of Escherichia coli DNA topoisomerase I (Topo I) and III (Topo III) play in catalysis was examined by comparing the properties of Topo III with those of a truncated enzyme lacking the generalized DNA binding domain of Topo III, Topo I, and a hybrid topoisomerase polypeptide containing the amino-terminal 605 amino acids of Topo III and the putative generalized DNA binding domain of Topo I. The deletion of the carboxyl-terminal 49 amino acids of Topo III decreases the affinity of the enzyme for its substrate, single-stranded DNA, by approximately 2 orders of magnitude and reduces Topo III-catalyzed relaxation of supercoiled DNA and Topo III-catalyzed resolution of DNA replication intermediates to a similar extent. Fusion of the carboxyl-terminal 312 amino acid residues of Topo I onto the truncated molecule stimulates topoisomerase-catalyzed relaxation 15-20-fold, to a level comparable with that of full-length Topo III. However, topoisomerase-catalyzed resolution of DNA replication intermediates was only stimulated 2-3-fold. Therefore, the carboxyl-terminal amino acids of these topoisomerases constitute a distinct and separable domain, and this domain is intimately involved in determining the catalytic properties of these polypeptides.
Escherichia coli contains two type I DNA topoisomerases, DNA topoisomerase I (1) and III (2,3). DNA Topoisomerase I (Topo I), 1 a 865-amino acid polypeptide (4), has been proposed to be involved with the maintenance of the superhelical density of the bacterial chromosome by acting as an antagonist to DNA gyrase (reviewed by Wang (5)). It has been shown to possess an ATP-independent relaxation activity (1) and is capable of catalyzing the catenation of nicked, duplex DNA circles (6 -8) and the decatenation of these singly linked catenanes (6,9).
DNA topoisomerase III (Topo III), a 653-amino acid polypeptide (10), was originally purified as a superhelical DNA relaxation activity from cells containing a deletion of the gene encoding Topo I (topA) (2,3). It was subsequently purified as a potent decatenating activity based on its ability to resolve plasmid DNA replication intermediates in vitro (11). Unlike Topo I, Topo III-catalyzed relaxation of negatively supercoiled DNA was virtually undetectable under standard assay conditions (10 mM Mg 2ϩ , 50 mM Na ϩ or K ϩ , 37°C), requiring high temperature (52°C) and low magnesium (1 mM) and monovalent salt concentrations (Ͻ20 mM) to exhibit maximal activity (2,3,11). The decatenation of multiply interlinked plasmid DNA dimers and resolution of DNA replication intermediates catalyzed by Topo III, however, does not require these extreme conditions and proceeds quite efficiently under the standard reaction conditions stated above (11). Interestingly, Topo I does not catalyze the resolution of DNA replication intermediates in vitro (12). This has led to the suggestion that Topo III may play a role in the decatenation of newly replicated DNA, whereas Topo I is involved in the maintenance of the superhelical density of the chromosome (11).
Analysis of the gene encoding Topo III (topB) indicated that this topoisomerase exhibited striking protein sequence homology to Topo I (10). Topo I and Topo III, therefore, are an example of two proteins, sharing extensive protein sequence homology, that catalyze distinct reactions. The homology between Topo I and Topo III extends only through the first 600 amino acids of the two polypeptides. The carboxyl-terminal amino acid residues of Topo I and Topo III show no homology, but each has been shown to be involved in substrate binding (13,14). The carboxyl terminus of Topo I contains three zinc finger motifs and a high density of arginine and lysine residues (15). The carboxyl terminus of Topo III does not contain any known motif but, similar to Topo I, contains a high density of clustered, positively charged amino acid residues (14). Since the carboxyl-terminal residues differ in the two polypeptides, the role that this region plays in topoisomerase-catalyzed relaxation and decatenation was examined. The nature of the differences in the two polypeptides was addressed by comparing the properties of Topo III with a truncation of the enzyme (lacking the putative carboxyl-terminal 49-amino acid residue substrate binding domain (14)), Topo I, and a hybrid molecule that contained the amino-terminal 605 amino acids of Topo III fused to the carboxyl-terminal 312-amino acid residue substrate binding domain of Topo I.

MATERIALS AND METHODS
DNA and Nucleotides-X174 RF I DNA was purchased from Life Technologies Inc. DNA oligonucleotides were prepared by the University of Maryland Biopolymer Laboratory. Radiolabeled nucleoside triphosphate was purchased from Amersham Corp.
Enzymes and Reagents-Acrylamide and agarose were from Life Technologies, Inc. Bacteriophage T4 polynucleotide kinase was from New England Biolabs Inc. Nuclease P1 was purchased from Boehringer Mannheim.
Protein Determination-Protein concentration was determined by the method of Bradford (16) using a Bio-Rad protein assay kit.
Radiolabeling of Oligonucleotides-Oligonucleotides were 5Ј end-labeled using bacteriophage T4 polynucleotide kinase and [␥-32 P] ATP as per the manufacturer's recommendations. The labeled oligonucleotides were fractionated through a polyacrylamide gel. The region containing the labeled oligonucleotide was excised, and the DNA was isolated by direct elution of the fragment into 10 mM Tris-HCl (pH 7.5 at 22°C), 1 mM EDTA. The radiolabeled oligonucleotides were diluted to a specific activity of 2000 cpm/pmol by the addition of excess unlabeled oligonucleotide.
Construction of the Chimeric Topo I-Topo III Gene-The gene encoding the Topo III-Topo I fusion protein was constructed by first introducing a PvuII restriction endonuclease site, using oligonucleotidedirected site-specific mutagenesis (17), into Mp19 DNA that contained topB (10) (the gene encoding Topo III) and topA (12) (the gene encoding Topo I) that had been engineered to be subcloned into the T7 transient expression vector pET-3c (18). The PvuII endonuclease site was positioned to cleave the topB sequence after the codon specifying amino acid Gln 605 and to cleave the topA sequence after the codon specifying amino acid Gln 552 . The Mp19 DNA (RF I) that contained the altered topB gene was then cleaved with NdeI and PvuII restriction endonucleases to liberate a fragment encoding the first 605 amino acids of Topo III. The NdeI-PvuII fragment was separated by agarose gel electrophoresis, the band was excised from the gel, and the fragment was purified using an IBI electroelution apparatus. The Mp19 DNA (RF I) that contained the altered topA gene was cleaved with PvuII and BglII restriction endonucleases to liberate a fragment encoding the carboxyl-terminal amino acids (amino acids 553-865) of Topo I (a BglII site was previously placed downstream of topA in order to allow ligation into the BamHI site of pET-3c (12)). The PvuII-BglII fragment was separated by agarose gel electrophoresis, the band was excised from the gel, and it was purified with an IBI electroelution apparatus.
The two fragments were used in a ligation containing NdeI-BamHI cut pET-3c in order to generate the hybrid gene (pT31Z). A plasmid DNA containing the correct insert was isolated and was then transformed in E. coli BL21.
Purification of Topo 31Z, Topo III, Topo I, and Topo III 604 -The induction of the chimeric polypeptide (Topo 31Z) was initiated by infection of the expression strain, harboring the pT31Z plasmid DNA, with bacteriophage CE6 (18). Induction of the hybrid protein was performed at 30°C in order to maximize the topoisomerase activity present in the crude lysate. Chromatography through trypsin inhibitor agarose was included in the purification of each polypeptide to reduce proteolysis (12). In order to prevent any contamination of the hybrid polypeptide with endogenous Topo III, it was purified from E. coli strain BL21 in which the gene encoding Topo III (topB) had been disrupted (12). Both polypeptides were purified by a modification of a previously described protocol that included DE52, Biogel HT, single-stranded DNA cellulose, and Sephacryl S-200 chromatography (11). The purification of Topo III, Topo I, and Topo III 604 has been described previously (12,14).
Superhelical DNA Relaxation Assays-Superhelical DNA relaxation reaction mixtures (25 l) contained 40 mM Hepes-KOH buffer (pH 8.0 at 22°C), 1 mM magnesium acetate (pH 7.0), 0.1 mg/ml bovine serum albumin, 40% (v/v) glycerol, 200 ng of X174 form I DNA, and the indicated amount of topoisomerase (14). Reactions were incubated at 52°C for 10 min, and the reaction products were separated through an agarose gel and visualized by staining with ethidium bromide as described previously (19).
oriC DNA Replication Assay-The replication of oriC-containing DNA, in vitro, was performed as described previously (12). The replication products were separated by agarose gel electrophoresis and visualized by autoradiography (19). The percentage of replication products existing as Form II molecules was quantified using a Fuji BAS 1000 phosphor imager.
Oligonucleotide Gel Mobility Shift Assays-Reaction mixtures (10 l) contained 40 mM Hepes-KOH buffer (pH 8.0 at 22°C), 0.1 mg/ml bovine serum albumin, 1 mM magnesium acetate (pH 7.0), 12% glycerol, 5 pmol of radiolabeled oligonucleotide, and the indicated amount of topoisomerase. The reactions were incubated for 5 min at 37°C, and the products were separated through a polyacrylamide gel (30:0.8) using 0.5 ϫ TBE as the running buffer. The gels were electrophoresed at 15 mA for 1.5 h, dried, and autoradiographed. The 45-base oligonucleotide used in the assay was 5Ј-CAGAATCAGAATGAGCCGC2-AAC2T1TC2GGG2 ATGAAAATGCTCACAAT-3Ј (oligonucleotide 45C), where 1 indicates the site of Topo III cleavage and 2 indicates the site of Topo I cleavage. The autoradiographs were quantified using a Pharmacia Biotech Inc. Ultrascan laser densitometer. In addition, bands from the gels, representing the indicated topoisomerase-oligonucleotide complex, were also excised, and the amount of radiolabeled oligonucleotide was determined using a Beckman LS 5801 liquid scintillation counter.
Topoisomerase The indicated amount of Topo III, Topo 31Z, or Topo I, was incubated for 3 min at 37°C, and the reaction was stopped by the addition of SDS to 2%. The reactions were adjusted to 45% formamide, 10 mM EDTA, 0.025% bromphenol blue, 0.025% xylene cyanol, and heat-denatured for 5 min at 90°C. The reaction products were separated by electrophoresis through a polyacrylamide gel (19:1) containing 50% (w/v) urea. The gels were then dried and autoradiographed. Nuclease P1 Protection Assay-Reaction mixtures (5 l) contained 40 mM Hepes-KOH buffer (pH 8.0 at 22°C), 0.1 mg/ml bovine serum albumin, 1 mM magnesium acetate (pH 7.0), and 200 fmol of the 45-base radiolabeled oligonucleotide. Reactions containing Topo I and Topo 31Z were incubated for 3 min at 37°C followed by the addition of 3 ϫ 10 Ϫ3 units of P1 nuclease as described previously (14). The reactions were incubated an additional 10 min at 37°C and terminated by the addition of EDTA to 10 mM. The reactions were adjusted to 45% formamide, 0.025% bromphenol blue, 0.025% xylene cyanol and heat denatured for 5 min at 90°C. The reaction products were separated by electrophoresis through a polyacrylamide gel (19:1) containing 50% (w/v) urea. The gels were then dried and subjected to autoradiography.

RESULTS
Overproduction and Purification of Topo 31Z, a Topoisomerase III-Topoisomerase I Chimera-In order to examine what was responsible for the different catalytic properties of Topo I and Topo III, a chimeric molecule, Topo 31Z, was synthesized ( Fig. 1A) that combined the first 605 amino acid residues of Topo III (which are homologous to those of Topo I) with the carboxyl-terminal amino acid residues of Topo I (which show no homology with Topo III). The chimeric gene was engineered in such a way that the only significant change between the two enzymes was the replacement of valine 553 of Topo I with a leucine residue (Fig. 1B). This Topo III-Topo I hybrid gene was  (15), whereas the carboxyl-terminal amino acids of Topo III contain clustered lysine and arginine residues (14). Panel B, the amino acid sequences of Topo I, Topo III, and Topo 31Z are presented. The synthesis of Topo 31Z resulted in a single change in the carboxyl-terminal residues of Topo I: Val 553 was changed to a leucine residue (filled circle) as a result of the cloning procedure. cloned into the bacteriophage T7 transient expression vector pET-3c (18), and the gene product was overproduced and purified to apparent homogeneity as described previously (12). The purification procedure yielded the expected 103-kDa polypeptide (Topo 31Z) and a 98-kDa proteolysis product (Fig. 2). The amount of full-length Topo 31Z (70% of the preparation) is similar to the amount of full-length Topo I present when purified by a similar procedure (12). The pattern of proteolysis is also consistent between Topo 31Z and Topo I, suggesting that the proteolysis occurs within the identical carboxyl-terminal amino acid residues of the two enzymes.
The Presence of Topo I Carboxyl-terminal Amino Acid Residues Stimulates Topo 31Z-catalyzed Relaxation of Negatively Supercoiled DNA-Previous studies have demonstrated that the truncation of Topo III from 653 amino acids (full-length) to 604 amino acids (Topo III 604 ) had a drastic effect on the catalytic activity of the enzyme. This reduction in activity correlated with the loss of enzyme's ability to bind to DNA and a change in mechanism from a processive to a distributive enzyme (14). The data in Fig. 3 demonstrates that the defect in catalyzing the relaxation of negatively supercoiled DNA by Topo III 604 is relieved by the incorporation of the carboxylterminal residues of Topo I. The specific activity of Topo 31Z is approximately 15-20-fold higher than that of Topo III 604 (Fig.  3, upper panel; compare lanes 2-5 with lanes 6 -9). The specific activity of Topo 31Z is approximately 50% of both intact Topo III and Topo I (Fig. 3, lower panel). The relaxation activity measurements were performed under conditions where Topo III activity is maximal (high temperature (52°C) and 1 mM Mg 2ϩ (2, 3, 11)); however, when the enzymes were assayed under more physiological conditions (37°C, 10 mM Mg 2ϩ ) Topo III and Topo 31Z relaxation activity was barely detectable. Topo I activity was only reduced slightly (data not shown). Therefore, the relaxation properties of Topo 31Z mimic those of Topo III rather than those of Topo I.
The Addition of the Carboxyl-terminal Amino Acid Residues of Topo I to Topo III 604 Stimulates the Ability of the Topoisomerase to Bind to Its Substrate-Since the carboxyl-terminal residues of both Topo I and Topo III appear to be involved in substrate binding, the dramatic increase in Topo 31Z-catalyzed DNA relaxation activity could be the result of the restoration of the enzyme's ability to bind its substrate. In order to evaluate this hypothesis, oligonucleotide mobility shift experiments were performed using a 45-base oligonucleotide (14), containing both Topo I and Topo III cleavage sites, as a substrate (Fig.  4). Topo III, Topo III 604 , Topo 31Z, and Topo I were evaluated for their ability to form a stable complex with the 45-base oligonucleotide (oligonucleotide 45C). In accord with previous results (14), the truncation of Topo III to 604 amino acid residues dramatically decreases equilibrium binding of the enzyme to its substrate. In fact, no oligonucleotide-topoisomerase complex is observed with the levels of the truncated enzyme used in this experiment (lanes 5-7). The addition of the carboxyl-terminal residues of Topo I to the truncated enzyme, however, stimulates equilibrium binding of the enzyme to its substrate. Topo 31Z (lanes 8 -10) is approximately 11% as efficient as Topo III (lanes 2-5) and Topo I (lanes [11][12][13] in binding to oligonucleotide 45C. Previous studies have indicated that the equilibrium binding of Topo III 604 is approximately 0.5% of the fulllength polypeptide (14); therefore, the addition of the carboxylterminal amino acids of Topo I stimulates equilibrium binding of Topo III 604 by 20-fold. This is in excellent agreement with stimulation observed for Topo 31Z in the relaxation of negatively supercoiled DNA substrates.
The cleavage site specificity of Topo 31Z was also compared with Topo III and Topo I (Fig. 5). A 22-base oligonucleotide containing a subsequence of the 45-base oligonucleotide was used as a substrate in a topoisomerase-catalyzed DNA cleavage assay. This oligonucleotide contains distinct Topo III and Topo I cleavage sites. Topo 31Z-induced cleavage of this substrate occurs at the strong Topo III site (compare lanes 1 and 2 to lane 4) rather than the Topo I cleavage sites (lane 3). This is consistent with the observation that the requirements for the relaxation of supercoiled DNA by the hybrid molecule closely  2, 6, and 10), 100 fmol (lanes 3, 7, and 11), 200 fmol (lanes 4, 8, and 12), and 500 fmol (lanes 5, 9, and 13) of either Topo 31Z (lanes 2-5), Topo I (lanes 6 -9), or Topo III (lanes 10 -13). OC, open circle (nicked or gapped circular DNA). SC, supercoiled (negatively supercoiled circular DNA). resemble those of Topo III and is also consistent with the previous observation that Topo III 604 exhibits the same cleavage site specificity as Topo III, in spite of the removal of its generalized DNA binding domain (14).

The Presence of Topo I Carboxyl-terminal Amino Acid Residues Does Not Significantly Stimulate Topo 31Z-catalyzed Resolution of DNA Replication Intermediates-Topo III has been
shown to be a potent decatenase, in vitro, using both pBR322 (11) and oriC-containing plasmids (12) as templates. The ability of Topo 31Z to resolve DNA replication intermediates was assessed by titrating the enzyme into an in vitro replication reaction (Fig. 6). As previously shown (12), Topo I was incapable of resolving DNA replication intermediates into plasmid monomers (Fig. 6A, lanes 2-5). The linking number of the interlinked plasmid dimers decreases with increasing amounts of Topo I, but fully decatenated monomeric plasmid molecules were never observed. The lack of fully decatenated plasmid monomers may be a consequence of the inhibition of the replication reaction by increasing amounts of Topo I (Fig. 6A, lanes  1-5). It has been shown that the replication reaction can be inhibited by relaxation of the input template by Topo I (20,21). It is not clear whether Topo I is incapable of fully decatenating the replication intermediates or simply if the amount of Topo I required to generate fully decatenated products results in the inhibition of the replication reaction. Topo III, however, was a potent decatenase and readily resolved the plasmid DNA replication intermediates (Fig. 6A, lanes 7-10).
Interestingly, in contrast to the 15-20-fold stimulation of DNA relaxation activity compared with that of the truncated enzyme, the addition of the carboxyl-terminal amino acid residues of Topo I to the truncated enzyme (Topo 31Z) had only a minimal effect on the resolution of DNA replication intermediates (Fig. 6B, lanes 2-5). The hybrid enzyme showed, at most, a 2-3-fold stimulation of decatenation activity when compared with the 604-amino acid truncation of Topo III (Fig. 6B, lanes  7-10). Topo III 604 is 1-2% as active as the full-length polypeptide, consistent with the DNA relaxation activity exhibited by the enzyme.
In contrast to Topo I, however, Topo 31Z was capable of completely resolving plasmid DNA replication intermediates (albeit at a reduced efficiency). This may represent an intrinsic and unique property of the first 605 amino acids of Topo III.

The Presence of the Carboxyl-terminal Residues of Topo I in Topo 31Z Alters the Substrate Binding Properties of Topo III-
Although the restoration of substrate binding efficiency by the carboxyl-terminal residues of Topo I could account for the stimulation of Topo 31Z-catalyzed DNA relaxation activity (Fig. 3), it could not account for the significantly reduced stimulation of the resolution of DNA replication intermediates (Fig. 6). Therefore, the binding properties of the hybrid enzyme were further characterized by a nuclease protection experiment using both Topo I and Topo 31Z (Fig. 7). Topo III has been shown to be a site-specific binding protein that protects a 14-base region surrounding its cleavage site (22). In contrast to Topo III (Fig. 7A,  lane 1), a titration of Topo 31Z (Fig. 7A, lanes 2-5) and Topo I (Fig. 7B, lanes 1-4) revealed no distinct nuclease protection pattern. This observation suggests either that a single molecule of Topo I or Topo 31Z protects the entire oligonucleotide or that the topoisomerase molecules bind along the length of the entire oligonucleotide. It is clear, however, that the binding properties of the enzyme have been altered, and the protection pattern generated by Topo 31Z is similar to that generated by Topo I. This change in the binding properties of the hybrid molecule may account for the altered properties of the enzyme.

DISCUSSION
The carboxyl-terminal residues of Topo I and Topo III are required for the formation of a stable enzyme-substrate complex, and it has been postulated that this region may constitute a generalized DNA binding domain (13,14). Since this domain lies outside the region of homology between the two polypeptides, the possibility that the distinct reactions catalyzed by the two enzymes were the result of the properties of their heterologous carboxyl-terminal domains was examined. This was accomplished by determining the biochemical properties of a chimeric enzyme in which the carboxyl-terminal residues of Topo I were substituted for those of Topo III.
A 604-amino acid truncation of Topo III (Topo III 604 ) had been shown to possess a very low affinity for single-stranded DNA when compared with the full-length molecule (14). The protein sequence homology between Topo I and Topo III breaks around this point in the amino acid sequence comparison of the two enzymes; therefore, this region was chosen as the site for splicing the carboxyl-terminal amino acid residues of Topo I to Topo III. The hybrid enzyme was overexpressed, purified, and assayed for supercoiled DNA relaxation activity and its ability to resolve DNA replication intermediates in vitro.
The addition of the carboxyl-terminal amino acids of Topo I to Topo III 604 was able to restore the ability of the enzyme to bind to single-stranded DNA substrates as well as dramatically stimulate topoisomerase-catalyzed relaxation of negatively supercoiled DNA. The hybrid enzyme was approximately 50% as active as either intact Topo I or Topo III. The biochemical properties of relaxation by the hybrid enzyme (Topo 31Z) as well as cleavage site specificity were identical to those of Topo III, suggesting that the first 605 amino acid residues of Topo III also contribute to the unique characteristics of Topo III.
Since the carboxyl-terminal amino acid residues of Topo I could be fused with another topoisomerase molecule and reconstitute functional topoisomerase activity, this region must constitute a distinct and separable domain of the enzyme. In addition, the properties of this domain appear to be critical in determining the biochemical characteristics of both Topo I and Topo III.
Although the biochemical characteristics of the relaxation activity of Topo 31Z were very similar to Topo III, Topo 31Z was very inefficient in catalyzing the resolution of DNA replication intermediates in vitro. The addition of the amino acid residues of Topo I stimulated the relaxation activity of Topo 31Z 15-20fold relative to Topo III 604 and stimulated the binding of the enzyme to single-stranded DNA to a similar extent (the equilibrium binding of Topo III 604 to the identical oligonucleotide has been shown to be 0.5-1% of that of the full-length polypeptide (14)), but the decatenation properties of Topo 31Z were enhanced only 2-3-fold relative to the truncated enzyme.
An analysis of binding specificity of both Topo I and Topo 31Z indicated that these enzymes bind and protect the entire 45base oligonucleotide substrate. This result is distinctly different from the 14-base protection pattern exhibited by Topo III (22). Therefore, it is clear that the substitution of the generalized binding domain of Topo I for that of Topo III alters the manner in which Topo III binds to its substrate and alters the catalytic properties of the enzyme. A striking difference between Topo I and Topo III is that in addition to a preference for single-stranded substrates, Topo I also has a considerable affinity for double-stranded DNA 2 (23). In contrast, only singlestranded DNA appears to be an effective substrate for Topo III binding (11,14).
A model for the mechanism of both decatenation and relaxation, based on the properties of the generalized binding domains of Topo I and Topo III, data from other laboratories, and from the known three-dimensional structure of Topo I is presented in Fig. 8. The mechanism of topoisomerase-induced decatenation is based on the model by Mondragon and colleagues (24). In this model (Fig. 8A), one helix (represented by the circle) is located in the cavity of the inverted "U" structure of the topoisomerase. The generalized DNA binding domain (represented by the rectangle) is positioned asymmetrically across the body of the topoisomerase. This domain is responsible for noncovalent interactions with the substrate 5Ј to the topoisomerase cleavage site (22). Decatenation occurs when a tyrosine residue in the active site of the enzyme (triangle) transiently nicks the single-stranded DNA, creating a "gate" that allows the helix located in the cavity of the enzyme to pass 2 S. Malpure and R. J. DiGate, unpublished results. through the nick. Since the generalized DNA binding domain of Topo III binds single-stranded DNA, the enzyme is always capable of passing a helix through the gate and, hence, always capable of decatenating the two molecules. The generalized DNA binding domain of Topo I, however, has a significant affinity for double-stranded DNA in addition to its affinity for single-stranded DNA (23). 2 If double-stranded DNA were present in the enzyme active site, the gate would be blocked by the uncleaved strand of DNA since a type I topoisomerase can only cleave one strand of the substrate. The same result would occur if the enzyme bound to a region of a single-stranded substrate that did not contain a Topo I cleavage site. These binding properties would result in an enzyme that is inefficient at decatenation. This would also explain the inefficiency observed in Topo 31Z-catalyzed decatenation since, the majority of the time, the enzyme would be presented with a substrate that is refractory to decatenation.
Relaxation of negatively supercoiled DNA (Fig. 8B) is accomplished by the binding of double-stranded DNA to the generalized DNA binding domain of Topo I in a mechanism analogous to the bridging model originally proposed by Cozzarelli and colleagues (7,26). In this case, the single-stranded region required for cleavage need only be present locally around the active site tyrosine (triangle) of the enzyme. The three-dimensional structure of Topo I indicates that there is a cleft capable of fitting single strand DNA in the vicinity of the active site of the enzyme (24). The generalized binding domain may simply feed the double-stranded molecule into this channel resulting in the local denaturation of the molecule into two single strands (negative superhelicity is also required at this step). After strand scission, the torque present in the supercoiled molecule would provide the driving force of strand passage through the gate.
The relaxation properties of the chimeric enzyme (Topo 31Z) mimic those of Topo III. This may be explained by the fact that the active site of Topo III lacks the ability to locally unwind the substrate (this would be unnecessary since the generalized DNA binding domain of Topo III only binds and feeds singlestranded DNA into the active site); therefore, relaxation by Topo 31Z (or by Topo III) requires conditions that produce stable single-stranded regions within the predominantly double-stranded substrate. This is consistent with the observation that both Topo 31Z and Topo III require high temperature and low Mg 2ϩ concentration to efficiently relax supercoiled substrates. These conditions both favor the stabilization of singlestranded DNA within a negatively supercoiled molecule.
Ultimately, the reason for the different catalytic properties of Topo I and Topo III will be resolved by a comparison of the three-dimensional structures of the two enzymes. For example, decatenation of two interlinked molecules requires that a type I topoisomerase interact at a node between a single strand gap or nick and the helix of DNA that must be passed through the transient gate in the single strand. A potent decatenating enzyme, such as Topo III, may contain a structure(s) that pro-  The generalized binding domain (rectangle) of Topo III has been positioned upon the known threedimensional structure of Topo I (inverted "U") (24). Since the two enzymes share extensive protein sequence homology within their first 600 amino acids (10), it is assumed that they will have a similar three-dimensional structure. Decatenation by Topo III requires that the generalized binding domain hold the substrate to be cleaved across the body of the enzyme (the substrate shown is a gapped molecule, where the DNA in the active site region is single-stranded (single wavy line) and the region outside of the active site is double-stranded). The dotted line indicates where the substrate is held within the binding domain. The double-stranded helix to be unlinked (circle) is held in the cavity present within the body of enzyme. The single-stranded DNA is then transiently cleaved by a tyrosine residue (triangle) present in the active site of the enzyme, with the DNA fragment 3Ј of the cleavage site held covalently through the tyrosine residue and the DNA fragment 5Ј of the cleavage site held by the generalized DNA binding domain. The helix is then passed through this transient gate, and the nick is resealed. B, model for Topo I-catalyzed relaxation of negatively supercoiled DNA. Relaxation of negatively supercoiled DNA by Topo I requires that the generalized DNA binding domain of Topo I hold a double-stranded DNA across the body of the enzyme. The generalized binding domain simply feeds the double-stranded molecule into this channel resulting in the local denaturation of the molecule into two single strands (this denaturation step is greatly stimulated by negative superhelicity). Relaxation occurs by the transient cleavage of one of the strands (as described above) followed by the strand passage of the opposite strand (shown in boldface) through the gate. This energy is provided by potential energy store stored in the form of superhelicity within the molecule. Decatenation could not be accomplished even if a helix were present in the cavity of the enzyme since the uncleaved strand of DNA would serve to prevent strand passage. A more detailed description is given under "Discussion." motes the creation of a node and/or stabilizes this intermediate.
The crystal structure of the amino-terminal 596 residues of Topo I has been determined to 2.2-Å resolution (24). Unfortunately, this structure does not contain the carboxyl-terminal substrate binding domain. The structure of carboxyl-terminal domain of Topo I has been determined using multidimensional NMR methods (25); however, since this structure was obtained from a purified carboxyl-terminal peptide, it is unclear how this structure relates to the known crystal structure of the enzyme. However, crystals have been obtained of the full-length Topo III polypeptide. 3 In addition, the availability of a catalytically inactive Topo III polypeptide that has the same binding specificity as the active enzyme (22) should allow the structural determination of a Topo III-substrate complex.