Mutational Analysis of Escherichia coli Topoisomerase IV III. IDENTIFICATION OF A REGION OF ParE INVOLVED IN COVALENT CATALYSIS*

The products of three dominant-negative alleles of parE, encoding the ATP-binding subunit of topoisomerase IV (Topo IV), were purified and their activities characterized when reconstituted with ParC to form Topo IV. The ability of the ParE E418K, ParE G419D, and ParE G442D mutant Topo IVs to bind DNA, hydrolyze ATP, and close their ATP-dependent clamp was relatively unaffected. However, their ability to relax negatively supercoiled DNA was compromised significantly. This could be attributed to severe defects in covalent complex formation between ParC and DNA. Thus, these residues, which are far from the active site Tyr of ParC, contribute to covalent catalysis. This indicates that a dramatic conformational rearrangement of the protein likely occurs subsequent to the binding of the G segment at the DNA gate and prior to its opening.

Type II topoisomerases are capable of passing segments of duplex DNA through transient double-strand breaks in a reaction that is coupled to ATP-binding and hydrolysis. When the passed segment (the T segment) and the transient doublestrand break (the DNA gate) are on the same circular DNA molecule, a change in the linking number of the DNA results. When the two segments are on two different DNA molecules, catenation or decatenation occurs. In prokaryotes, the ATP binding and hydrolysis and the DNA cleavage and religation activities reside on different subunits that associate to form a heterotetramer (1)(2)(3).
Even though ParC, the DNA cleavage and religation subunit of Escherichia coli topoisomerase IV (Topo IV) 1 is a stable dimer in solution (4), it cannot cleave DNA by itself, it must be associated with the ATP-binding subunit, ParE (4,5). Because ParC binds DNA nearly as well as the intact enzyme, 2 this implies that ParE either contributes amino acid residues to the active site that are necessary for catalysis or that signals from ParE, in the form of conformational changes presumably driven by either the binding or hydrolysis of ATP, are required for DNA cleavage to proceed. As described in an accompanying article (6), Topo IV proteins reconstituted from ParC and mutant ParE proteins that are defective in either ATP binding or hydrolysis are still capable of covalent complex formation and DNA cleavage. This suggests that there must be other amino acid residues of ParE that are not involved in either ATP binding or hydrolysis, but are required for covalent catalysis. We have identified several such residues in this report.
We isolated six independent alleles in a screen for dominantnegative mutations in parC (7). The corresponding mutant ParE proteins all formed catalytically inactive Topo IV when reconstituted with ParC. In an accompanying article (6), we describe the characterization of three of the mutant Topo IV proteins that had defects in ATP binding and hydrolysis. Here, we report our characterization of the ParE E418K, ParE G419D, and ParE G442D mutant Topo IV proteins. While these proteins were relatively unaffected in their ATP-directed functions, they were severely reduced in their ability to form covalent complex or cleave DNA: a reaction mediated by ParC. Given the current understanding of the structure of type II topoisomerases (8), this indicates that a large conformational rearrangement is required to bring these amino acid residues in ParE in the vicinity of the active site Tyr in ParC in order to open the DNA gate.

MATERIALS AND METHODS
Reagents, Enzymes, and DNAs-Reagents, enzymes, and DNAs are all as described in the accompanying articles (6,7).
Topo IV Binding to DNA-Binding to linear pBR322 DNA was as described by Peng and Marians (9). K D values were calculated using the Hill equation as described (10). Binding to a duplex 24-nucleotide-long oligonucleotide composed of a defined Topo IV binding site was as described by Marians and Hiasa (11).
DNA Cleavage Assay-DNA Cleavage Assay was as described by Peng and Marians (12).
Other Assays-Assays for ATP hydrolysis, ParE dimer formation, decatenation of Crithidia kinetoplast DNA (kDNA), ParC covalent complex formation, and fast protein liquid chromatography gel filtration of Topo IV were as described in the accompanying articles (6,7).

The Mutant Topo IVs Can Hydrolyze ATP and Close Their
ATP-dependent Clamps-ParE is the ATP binding and hydrolysis subunit of Topo IV (4,5). In the first article in this series, we described six mutant ParE proteins that, when reconstituted with ParC, gave Topo IV proteins that were unable to relax negatively supercoiled DNA (7). Three of these, the ParE G110S, ParE S123L, and ParE T201A Topo IV, proteins also exhibited hyper-DNA cleavage. These enzymes were shown to be defective in ATP hydrolysis and DNA religation (6). Here, we investigate the underlying defects in the other three mutant Topo IVs.
All three of the mutations arose in amino acid residues that are highly conserved between type II topoisomerases. Glu 418 and Gly 419 are part of the EGDSA motif that is conserved in all type II topoisomerases (12). Gly 442 is part of the PLRGKILN motif and is conserved in all type II topoisomerases except Caenorhabditis elegans 2C, where the corresponding residue is an Arg (12). None of these residues are present in the structure of AMP-P(NH)P bound to the N-terminal fragment of GyrB (13). Thus, it is unlikely that they are involved in either ATP binding or hydrolysis directly. However, these amino acid substitutions could still affect these activities; Ser 127 and Thr 201 are also distant from the ATP, but they affect binding and hydrolysis (7).
The ability of the ParE E418K, ParE G419D, and ParE G442D Topo IV proteins to hydrolyze ATP was therefore examined ( Fig. 1). ATP hydrolysis by the ParE E418K and ParE G442D enzymes was essentially indistinguishable from that of the wild type protein, whereas the specific activity of the ParE G419D Topo IV was about one-third that of the wild type. This was a modest decrease in activity, particularly when compared with the defect in ATPase activity manifested by the ParE G110S, ParE S123L, and ParE T201A Topo IVs, where activity was decreased to between one-fifteenth and one-fortieth that of the wild type (7).
Because the ParE E418K, ParE G419D, and ParE G442D Topo IVs could all hydrolyze ATP, they should also be able to dimerize in the presence of AMP-P(NH)P. This is measured by protein-protein cross-linking in the presence of dimethyl suberimidate. All three of the mutant ParE proteins could be crosslinked to give a dimer of ParE in the presence of AMP-P(NH)P (Fig. 2). Thus, none of these three mutants were defective in either of the signature activities of ParE: ATP hydrolysis or the ability to close their ATP-dependent clamp.
Although neither the ParE E418K, ParE G419D, nor ParE G442D Topo IVs could relax DNA, at very high concentrations the ParE G442D enzyme appeared to be able to catenate DNA (7). We therefore examined the ability of the mutant enzymes to decatenate kDNA ( Fig. 3). At 40 M ATP, a concentration sufficient to support maximal rates of decatenation by the wild type enzyme (Fig. 3A), none of the mutant proteins could decatenate DNA when present at equivalent concentrations as the wild type (Fig. 3, B-D). At both higher protein and ATP concentrations, however, the ParE G442D Topo IV could decatenate the kDNA (Fig. 3E). A roughly 100-fold greater concentration of the mutant protein than the wild type effected about the same level of decatenation. As argued in the accompanying article (6), it is unlikely that this activity is a result of wild type ParE contaminating the preparation of ParE G442D.
The Mutant Topo IVs Are Defective in Covalent Catalysis-The mutant Topo IVs were not significantly defective in ATP hydrolysis and could close their ATP-dependent clamp. This suggested that they were defective either in binding DNA or in covalent catalysis. A significant defect in DNA binding seemed unlikely, because the ATPase activity of all the mutant proteins was stimulated by DNA. 2 This was checked directly by measuring the ability of the mutant Topo IVs to bind to linear pBR322 DNA using a nitrocellulose filter binding assay (Fig. 4).
The ParE E418K Topo IV bound DNA as well as the wild type, with K D values of 16.5 and 12.5 nM, respectively. Both the ParE G419D and ParE G442D Topo IVs showed modest defects in DNA binding, giving K D values of 51.2 and 66.3 nM, respectively. This is unlikely to account for their inability to relax negatively supercoiled DNA, because no activity was manifest even at concentrations of protein as high as 1 M. In addition, the K D for binding to supercoiled DNA is expected to be onetenth that for binding to linear DNA (9).
These data pointed to a defect for the mutant enzymes in covalent catalysis. There are three aspects of this that can be examined: (i) the ability to mediate SDS-dependent DNA cleavage, (ii) the ability to form a covalent complex between ParC and DNA, and (iii) the ability to form a ternary complex between the topoisomerase, DNA, and a quinolone antibacterial drug such as norfloxacin. All three aspects of covalent catalysis were examined for the mutant enzymes.
Once bound to DNA, topoisomerases engage in a cleavagereligation equilibrium. In the open form of the topoisomerase-DNA complex, the DNA is cleaved and covalently bound to the enzyme. This represents the opening of the DNA gate. In the case of Topo IV, the DNA is bound to ParC. Normally, free cleaved DNA is not obvious in a reaction where a topoisomerase is allowed to bind to DNA. To observe DNA cleavage, the topoisomerase must be denatured while it is bound to DNA in the open complex. Treatment with SDS of Topo IV bound to DNA that has a 5Ј-32 P label on one end produces a characteristic set of fragments representing the distance between the Topo IV cleavage site and the labeled end of the DNA. This pattern is evident for the wild type enzyme in Fig. 5. None of the mutant Topo IV proteins showed any DNA cleavage activity in this assay (Fig. 5).
In general, the cleavage-religation equilibrium is far to the side of religation. This acts to protect the cell from the acciden- tal generation of a lethal double-strand break. Thus, it was possible that the lack of cleavage observed for the mutant Topo IV proteins derived from a shift in the cleavage-religation equilibrium even further toward religation. The quinolone family of antibacterials affect bacterial type II topoisomerase activity by shifting the cleavage-religation equilibrium toward cleavage. Thus, they would be expected to increase DNA cleavage in the assay shown in Fig. 5. That experiment was performed in the absence of such drugs.
We have developed a very sensitive assay for formation of the Topo IV-DNA-quinolone ternary complex. It makes use of the fact that Topo IV cannot bind in a stable fashion to a duplex 24-nucleotide-long oligonucleotide composed of a defined Topo IV binding site sequence in the absence of quinolone (11). Binding is measured using nitrocellulose filters. The ability of the wild type and mutant enzymes to bind to the 24-mer in the presence of 0.5 mM norfloxacin was therefore determined (Fig.  6). Only binding by the wild type enzyme could be detected; no binding for any of the mutant Topo IVs was evident. This suggested that the observed DNA cleavage defect derived from an inability to form the covalent complex. This was tested directly.
If Topo IV is allowed to come to equilibrium bound to uni- formly 32 P-labeled DNA followed by digesting the DNA with nuclease, a fraction of the label will be transferred to the ParC subunit as a result of the Topo IV-DNA complex becoming destabilized and falling apart. Analysis by SDS-PAGE can be used to reveal the label transfer. When treated in this fashion, covalent complex formation by the mutant Topo IVs could not be detected, although it was clearly evident with the wild type enzyme (Fig. 7).
ParE G419D Cannot Form a Stable Heterotetramer with ParC-As discussed above, even though ParC can bind DNA, it cannot form a covalent complex with DNA in the absence of heterotetramer formation with ParE. Thus, the failure of the mutant Topo IVs to form a covalent complex could be because of the lack of a stable association between the mutant ParE proteins and ParC. This was examined by fast protein liquid chromatography gel filtration chromatography. Both ParE E418K and ParE G442D formed heterotetramers with ParC that could be isolated by gel filtration (data not shown). On the other hand, the combination of ParE G419D and ParC clearly did not elute during gel filtration as a heterotetramer (Fig. 8). However, ParE G419D did elute slightly ahead of the position of free ParE. This suggested that it did interact with ParC, but that the interaction was weakened compared with the wild type. Opening of the DNA gate clearly requires a large conformational rearrangement of the enzyme. Two crystal structures give us an idea of the movement required. The structure of the central core of the yeast topoisomerase II (14) presumably represents the open conformation. Here, the dimer-related ac-tive site tyrosines are over 27 Å apart, and it was estimated that they would have to move 35-40 Å toward and past each other to assume an appropriately staggered position to catalyze DNA cleavage. The structure of a 59-kDa N-terminal fragment of GyrA (equivalent to the AЈ region of the yeast topoisomerase II structure) presumably represents the closed conformation (15). Here the active site tyrosines are appropriately staggered, but the inter-tyrosyl distance (30 Å) is larger than the width of a double helix of DNA, suggesting that some additional rearrangement of either the DNA or the enzyme must occur to bring the tyrosines close enough to the scissile bonds to effect cleavage.
The chemical mechanism of DNA cleavage and religation remains obscure. Based on the GyrA59 structure, Cabral et al. (15) proposed that the active site of the breakage-reunion activity is formed from Tyr 122 , which forms the covalent bond with DNA, and Arg 121 from one monomer and His 80 , Arg 32 , and Lys 42 from the other monomer. Fast protein liquid chromatography gel filtration through a Superose 6 column was performed with the indicated subunits, and the wild type (WT) and mutant Topo IV proteins (with ParE in 10% excess). Aliquots of the indicated fractions were analyzed by SDS-PAGE through 10% gels. Glu 9 , Asp 111 , Asp 113 , and Glu 115 in the bacterial enzyme and Glu 449 , Asp 526 , Asp 528 , and Asp 530 in the yeast enzyme.
Glu 449 is part of the conserved EGDSA motif (12). These residues are also part of the toprim motif identified by Aravind et al. (18) on the basis of amino acid sequence comparisons. Toprim is a proposed conserved catalytic domain in type 1A and type II topoisomerases, DnaG-type primases, OLD family nucleases, and RecR proteins. It has been proposed that these acidic residues coordinate Mg 2ϩ in bacterial topoisomerase I (17). In the existing structure of the yeast enzyme (14), these residues project away from the active-site Tyr (Fig. 9). Thus, Berger et al. (16) proposed that a rotation of the BЈ domain about the linker to the AЈ domain to bring these residues in close proximity to the active-site Tyr is effected in the conformation of the enzyme where DNA is bound across the DNA gate and is about to be cleaved. Our finding of amino acid substitutions in the BЈ region of ParE that affect covalent complex formation by Topo IV is consistent with this proposal.
In our analysis of the biochemical properties of ParE proteins encoded by dominant-negative alleles of parE, we found two groups of mutations (7). When reconstituted with ParC, the ParE G110S, ParE S123L, and ParE T201A Topo IVs were defective in ATP hydrolysis and exhibited hyper-DNA cleavage. The characterization of these proteins is described in the accompanying report (6). The ParE E418K, ParE G419D, and ParE G442D Topo IVs were unaffected in their ability to either hydrolyze ATP or to close their ATP-dependent clamp; however, none of them could relax negatively supercoiled DNA. This inactivity could be attributed to a defect in the ability of these mutant enzymes to form the covalent intermediate.
Although two of these proteins, the ParE G419D and ParE G442D Topo IVs, did show a modest decrease in DNA binding, this was unlikely to account for the defect in covalent complex formation. These experiments were performed at Topo IV concentrations of 60 nM. Thus, even in the case of the ParE G442D Topo IV, which binds linear DNA with a K D of 66 nM, about 5-fold that of the wild type, 50% of the DNA in the experiment will be bound at any time. Yet, covalent complex formation was undetectable. We therefore conclude that these amino acid substitutions interfere with covalent catalysis.
Glu 418 is part of the toprim (18) and EGDSA (12) motifs and was predicted to be directly involved in covalent catalysis by coordinating Mg 2ϩ (17). If this is in fact the case, the reversal of charge effected by the E418K substitution would clearly have a profound effect on covalent catalysis. An equivalent substitution, E449A, inactivates the DNA relaxation and covalent complex-forming activities of yeast topoisomerase II (19).
Gly 419 is also part of the EGDSA motif (12). However, the lack of a functional group makes it unlikely that this residue participates directly in covalent catalysis. It seems likely that the G419D substitution imposes an effect on catalysis by interfering with the hydrogen-bonding network that would be required to coordinate Mg 2ϩ . In the yeast topoisomerase II structure (14) the equivalent residue, Gly 450 , lies very close to Glu 449 (Glu 418 in ParE), within 2-4 Å, is fairly close to Asp 526 , within 5-7 Å, and is farther from Asp 528 and Asp 530 , from 7 to Ͼ10 Å (Fig. 9). These latter three Asp residues are part of the putative acidic tetrad necessary for Mg 2ϩ coordination. Intrusion of an additional negative charge into this region could easily shift the pattern of hydrogen bonding between the Mg ion and the acidic tetrad. Consistent with this is that in topoisomerase II from Candida albicans, the Asp in the EGDSA motif is a Leu (12) and that substitution of the Ser with Ala in the yeast topoisomerase II does not inactivate the enzyme (19). Neither of these two amino acid replacements would be expected to interfere with a hydrogen-bonding network.
Gly 442 is part of the conserved PLRGKILN motif. In the yeast enzyme, substitutions of the Arg and Lys in this motif with Ala has no effect on the catalytic activities of the topoisomerase (20), whereas Ala substitution of the Asn does ablate the superhelical DNA relaxation activity and reduce covalent complex formation to about one-tenth the level of the wild type (19). This suggests that residues in this motif may not be directly involved in catalysis. Consistent with this is the observation that the ParE G442D Topo IV retained some catalytic activity. This enzyme could decatenate kDNA at concentrations 100-fold greater than that required for the wild type enzyme. Clearly, this implies that this mutant enzyme retains some low level ability to form the covalent complex. Our inability to detect this presumably relates to an inability to approach the protein concentrations that would be required in the assay for covalent complex formation.
Gly 442 is considerably farther from the acidic tetrad than Gly 419 . In the yeast topoisomerase II structure (14), the equivalent residue, Gly 476 , is 8 -9 Å from Glu 449 (Glu 418 in ParE), 7-9 Å from Asp 526 and Asp 528 , and greater than 12 Å from Asp 530 . Thus, whereas the G442D substitution could interfere with the hydrogen-bonding network involving the acidic tetrad, it is less likely.
By making mixed heterodimers of yeast topoisomerase II where one protomer carried a mutation in this region of the BЈ domain and the other protomer carried the Y782F mutation that inactivates covalent complex formation, Liu and Wang (19) have shown that amino acid residues in the BЈ domain, such as the acidic tetrad, cooperate in trans with the AЈ region of the other protomer in the dimer. The structural organization of this conformation is unknown. Thus, an explanation for the disruptive effect of the ParE G442D substitution may only become apparent when additional crystal structures are available.
It is clear, however, that our data and that from other laboratories discussed above, point up another of the large conformational changes necessary for activity of the type II topoisomerases. The cooperation in trans between the BЈ and AЈ regions of the two halves of the enzyme presumably contributes to the stabilization of the protein as the DNA gate opens. This view is supported by the apparent instability of the ParE G419D Topo IV tetramer.