Module 4.4.9

DNA Topoisomerases

KATHERINE EVANS-ROBERTS† AND ANTHONY MAXWELL*

[SECTION EDITOR: SUE LOVETT]

Posted July 29, 2009

Department of Biological Chemistry, John Innes Centre, Norwich NR4 7UH, United Kingdom

*Corresponding author. Mailing address: Department of Biological Chemistry, John Innes Centre, Norwich Research Park, Colney, Norwich NR4 7UH, United Kingdom. Phone: 44 (0) 1603 450771, Fax: 44 (0) 1603 450018, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

†Present address: Health Enterprise East Ltd., Papworth Hospital, Papworth Everard, Cambridge CB23 3RE, United Kingdom.

“Since the two chains in our model are intertwined, it is essential for them to untwist if they are to separate…. Although it is difficult at the moment to see how these processes occur without everything getting tangled, we do not feel that this objection will be insuperable.” —J. D. Watson and F. H. C. Crick (217).

This quotation, taken from the second paper detailing the structure of DNA, predicts a potential problem inherent in the double-helical structure of DNA. The processes that utilize DNA, such as transcription, replication, and recombination, require either the temporary or permanent separation of the complementary strands of the double helix. The structure of duplex DNA inevitably leads to changes in DNA topology, such as the introduction of supercoils, during these processes. These changes in topology are resolved by members of a ubiquitous family of enzymes known as DNA topoisomerases (25, 39, 68, 212). Topoisomerases alter DNA topology by binding to the DNA, cleaving either one or both strands of the double helix, then passing either the other strand of the same helix or another double strand through the break, and finally, resealing the DNA backbone. DNA cleavage always involves the formation of a transient phosphodiester bond between one end of the broken strand and a tyrosine in the active site of the topoisomerase. Some topoisomerases require divalent metal ions as cofactors in the DNA cleavage-religation reaction. The reactions performed by DNA topoisomerases are depicted in Fig. 1 and Fig. 2. It should be pointed out that many enzymes (e.g., ligases and recombinases) can alter DNA topology but are not referred to as topoisomerases, which is a term reserved for enzymes that have this as their specific role. Similarly, there may be enzymes currently classified as topoisomerases (e.g., topoisomerase III [topo III] and reverse gyrase) whose principal cellular function may be another activity.

Fig. 1Reactions performed by type I topoisomerases.

Source: Reprinted from Advances in Protein Chemistry (133) with the permission of the publisher.

Fig. 2Reactions performed by type II DNA topoisomerases.

Source: Reprinted from Advances in Protein Chemistry (133) with the permission of the publisher.

DNA supercoiling can be either positive (right handed) or negative (left handed). The binding of proteins to DNA is often dependent on the DNA being negatively supercoiled; initiation of the replication of bacterial plasmids requires negative supercoiling to facilitate the unwinding of the origin sequence (178). As DNA replication proceeds, positive supercoils are generated ahead of the replication fork and precatenanes may build up behind it (Fig. 3) (161). The supercoils are removed by topoisomerases to prevent excess supercoiling and the breakdown of the replication machinery.

Fig. 3Model of the topology of a replicating chromosome. The chromosome is separated into domains, with the boundaries represented as orange boxes; the replication fork is in the center. Positive supercoiling occurs ahead of the replication fork, and precatenanes may form behind it.

Source: Reprinted from the Proceedings of the National Academy of Sciences of the United States of America (161) with the permission of the publisher.

When two replication forks converge at the end of DNA replication, catenated DNA rings can be formed if the daughter molecules are interwound (Fig. 4) (205, 213). These rings can be separated by decatenation, in which one DNA ring is cleaved and the other ring is passed through the double-strand break. As discussed below, this situation can be resolved by the activities of various topoisomerases.

Fig. 4Formation of catenated DNA at the termination of replication. (a and b) Converging replication forks (a) lead to the interwinding of daughter molecules and the formation of precatenanes (b). (c) Upon the completion of replication, the products are catenated DNA circles.

Source: Reprinted from DNA Topology (11) with the permission of the publisher.

Transcription may also result in changes to DNA topology, for example, if the DNA is anchored to a fixed point in the cell. It has been proposed that during transcription DNA rotates on its axis to allow the RNA polymerase to follow the helical path of the DNA strands (125). This rotation leads to the buildup of positive supercoils ahead of the transcription complex and negative supercoils behind it, and these supercoils can be removed by the enzymes DNA gyrase and topo I, respectively (126, 162). The transcription of many genes has been shown to be influenced by the level of supercoiling in the cell (11). This is because supercoiling affects the DNA binding of RNA polymerase and other proteins that repress or activate transcription (120). In fact, the levels of transcription of the genes encoding topo I (topA) and DNA gyrase (gyrA and gyrB) are both affected by the degree of supercoiling, in what is thought to be a homeostatic mechanism to control the amount of supercoiling in the cell (137). Increased negative supercoiling increases the transcription of topA and decreases the expression of gyrA and gyrB. More recently, topo IV has been shown to also participate in supercoiling control by contributing to DNA relaxation, in addition to playing a role in decatenation (231).

In this chapter, we discuss the structures, roles, and mechanisms of the different topoisomerases. We also describe compounds that inhibit topoisomerases. Although this review focuses mainly on prokaryotic topoisomerases, some details of eukaryotic topoisomerases are also provided for comparison.

CLASSIFICATION OF TOPOISOMERASES

DNA topoisomerases can be divided into two classes: type I topoisomerases introduce transient breaks into DNA single strands, whereas type II enzymes introduce transient double-strand breaks (25). The two types of topoisomerases can be further subdivided into type IA, IB, IIA, and IIB enzymes according to structural, mechanistic, and evolutionary considerations. The properties of the different groups of topoisomerases are listed in Table 1, which summarizes the in vitro activities of the different enzymes. Alignments of the domains of the type I and type II topoisomerases are shown in Fig. 5 and Fig. 6, respectively.

Fig. 5Alignment of the domains of type IA topoisomerases, based on their homology to the cleavage-strand passage domain of E. coli topo I. S. acidocaldarius, Sulfolobus acidocaldarius.

Source: Reprinted from the Annual Review of Biochemistry (25) with the permission of the publisher.

Fig. 6Alignment of the domains of the type II topoisomerases based on amino acid sequence homologies to E. coli gyrase. (a) Type IIA topoisomerases. (b) Topo VI, the only topoisomerase in the type IIB class.

Source: Reprinted from the Annual Review of Biochemistry (25) with the permission of the publisher.

Type I Topoisomerases

Topo I.

Topo I was the first topoisomerase discovered and was originally named ω protein (214). It is found in both prokaryotes and eukaryotes and can relax negative supercoils and catenate and decatenate nicked DNA (202). Bacterial type I enzymes, because they relax only negatively supercoiled DNA, fall into the category of type IA (Table 1); eukaryotic topo I enzymes are in the category of type IB, because they relax both positively and negatively supercoiled DNA and are evolutionarily and mechanistically distinct from the bacterial enzymes. Escherichia coli topo I is a 97-kDa protein consisting of three domains (121). The first domain, which consists of 582 amino acids at the N terminus, is responsible for cleavage and strand passage and contains the active-site tyrosine at position 319. The next 162 amino acids make up a Zn(II)-binding domain that contains three tetracysteine motifs. The C-terminal third domain, which contains 121 amino acids, is rich in basic amino acids and contributes to substrate binding. The structure of the N-terminal 67-kDa fragment of E. coli topo I has been resolved (121) (Fig. 7). The structure forms a “base” and a “lid” around a cavity with a diameter of 28 Å, which could accommodate double-stranded DNA. The active-site tyrosines are positioned at the entrance to this cavity.

Fig. 7N-terminal 67-kDa fragment of E. coli topo I. Domains in the ribbon diagram are indicated.

Source: Reprinted from Nature (121) with the permission of the publisher.

In the proposed “enzyme-bridging” model of DNA relaxation by topo I, the enzyme cleaves a single strand of DNA and bridges the gap through which the intact strand is passed (121). The clamp then closes around the intact strand, and the cleaved strand is religated. The protein then reopens to release the passed strand and closes again to complete the cycle (Fig. 8).

Fig. 8Proposed mechanism for E. coli topo I. The enzyme binds DNA and cleaves one strand, forming a 5′-phosphodiester linkage (black circle). The complementary strand is passed through the gap and into the central cavity of the enzyme. The nick is resealed, and the strand is released.

Source: Reprinted from DNA Topology (11) with the permission of the publisher.

Mutations in Salmonella enterica serovar Typhimurium and E. coli topo I enzymes (topA mutations) are generally nonlethal but lead to the acquisition of compensatory mutations in the DNA gyrase genes, gyrA and gyrB (53, 163, 167, 189); however, topA mutations have been shown to lead to cold sensitivity (189). Despite the apparently nonessential nature of topo I, the stabilization of the complex of topo I and cleaved DNA induces the SOS response, leading to cell death (31). There is, therefore, potential to develop antibacterial compounds targeted to topo I (30, 31); indeed, compounds leading to the stabilization of the topo I cleavage complex have been isolated recently (30).

Eukaryotic topo I is capable of relaxing both positively and negatively supercoiled DNA (26). In the proposed mechanism for eukaryotic topo I (originally proposed for vaccinia virus topo I), a single strand of the DNA is broken following DNA binding (187). The 5′-OH of the broken strand can then rotate around the other strand before the break is resealed, in a process known as controlled rotation (186).

Topo III.

Topo III is a type IA topoisomerase that relaxes and decatenates DNA but also has the ability to cleave and decatenate RNA molecules (51, 52). Topo III is well conserved across evolutionary lineages and is found in prokaryotes, eukaryotes, and archaea, but here, we will describe the prokaryotic enzyme. Topo III has significant homology to E. coli topo I, and its crystal structure is also very similar to that of topo I (Fig. 7), with four distinct domains (52, 141). Topo III deletion mutants are viable, so it is thought that the enzyme shares activities with other topoisomerases (52). A topo III (topB) deletion mutation in a topo IV temperature-sensitive background is lethal (127), which may point to the ability of topo III to decatenate. One difference between the structure of E. coli topo III and that of topo I is an additional loop in topo III known as the decatenation loop. Topo I is able to decatenate only singly catenated molecules (202), whereas topo III can unlink multiply catenated dimers, so it is possible that the decatenation loop provides topo III with the ability to carry out multiple decatenation reactions (119). Indeed, the deletion of this loop reduces the decatenation ability of topo III (119). Topo III does not appear to play a role in regulating supercoiling in E. coli but has been suggested to be involved in disentangling recombination intermediates; i.e., it does not seem to be a conventional topoisomerase (127). RecQ helicases are often linked to topo III, and the two types of enzymes may function in cooperation to unlink DNA (72, 219). More recently, it has been proposed that these enzymes have a role in maintaining genome stability by resolving the situation of stalled replication forks (193), such as in the process illustrated in Fig. 4.

Topo V.

Topo V has been described as a type IB topoisomerase (180). It shows similarities to eukaryotic topo I and can relax negatively and positively supercoiled DNA. It also has a role in DNA repair (14). It was discovered in the hyperthermophile Methanopyrus kandleri (180), and this is the only species it has been found in so far (39). A structure of the N-terminal region of topo V has recently been resolved and has been shown to exhibit a novel fold (195). This structure may reveal a new family of topoisomerases, which has been described as class IC.

Reverse Gyrase.

Reverse gyrase is a type IA enzyme that is found in hyperthermophilic archaea and eubacteria (111). The enzyme can relax negatively supercoiled DNA, but interestingly, it can also introduce positive supercoils into relaxed DNA in an ATP-dependent manner (69, 179). The crystal structure of reverse gyrase has been determined (173). The C-terminal domain of the structure is similar to topo I. Indeed, if reverse gyrase is truncated at the N terminus, then the relaxation of negative supercoils can take place in the absence of ATP, as is the case with topo I (47). Reverse gyrase has a helicase-like N-terminal domain, and it has been suggested that this region is where DNA is bound (172). It has been proposed that the controlled unwinding of DNA by the helicase domain ensures that strand passage occurs in the direction of positive supercoiling (171). As positively supercoiled DNA is more likely than negatively supercoiled DNA to be resistant to the harmful effects of high temperature, it is likely that the action of reverse gyrase is an adaptation to the extreme habitat occupied by the hyperthermophilic archaea.

Type II Topoisomerases

Type II topoisomerases occur in both prokaryotes and eukaryotes, and these enzymes show a number of similarities (Fig. 6). Although the main topic of this review is prokaryotic enzymes, it is useful to summarize what we know about the eukaryotic enzymes first so that comparisons can be made.

Topo II.

Eukaryotic topo II is a type IIA topoisomerase that can relax both positively and negatively supercoiled DNA in an ATP- and Mg²⁺-dependent manner. It can also decatenate DNA and has been found in many eukaryotes, including humans (8), Drosophila melanogaster (102), and yeast (123). Most higher eukaryotes contain two isoforms, termed topo IIα and topo IIβ (58), which appear to be expressed at different times in the cell cycle and in different cell types (22). Topo IIα is found in proliferating cell types, and expression peaks during the G₂ and M phases of the cell cycle. Topo IIβ is found in all cell types, and its expression is constant throughout the cell cycle. Topo II is required for the condensation and segregation of daughter chromosomes following DNA replication (54). Topo II in mammals has also been linked to chromosome condensation during apoptosis. Many studies have shown topo II in yeast to be cell cycle regulated.

Topo II has homology to DNA gyrase and topo IV (see below). The N-terminal domain of topo II aligns with GyrB and the topo IV subunit ParE (Fig. 6) (23); the C-terminal domain aligns with GyrA and ParC. It is thought that topo II may have evolved following the fusion of the genes encoding the A and B subunits of gyrase (128). One area where topo II, topo IV, and DNA gyrase differ is the C terminus. In DNA gyrase and topo IV, this domain is important mechanistically, whereas in topo II, it is thought to have a regulatory role and include nuclear localization signals (218). The crystal structure of a 92-kDa fragment of Saccharomyces cerevisiae topo II has been resolved (Fig. 9b) (15, 63). This portion of the enzyme can bind and cleave DNA but does not include the ATPase domain, so it cannot relax supercoils. The B′ region of the topo II structure is homologous to the 47-kDa C-terminal domain of GyrB (Fig. 6). More recently, the structure of the same region of yeast topo II in a complex with a DNA oligonucleotide has been resolved (57). This structure revealed that topo II introduces a 150° bend into the bound G segment of DNA. However, it should be noted that the extent of the bend in the co-crystal structure may be influenced by crystal packing and the use of a doubly nicked DNA substrate. The structure of the ATPase domain of yeast topo II, in a complex with ADPNP (5′-adenylyl β,γ-imidodiphosphate, a nonhydrolyzable ATP analog) and the chemotherapeutic agent ICRF-187, has also been published (Fig. 9a) (33). This structure is similar to that of the ATPase domain of GyrB but does have some differences, such as a smaller central cavity (6 Å) that would be unable to accommodate a DNA duplex.

Fig. 9Structures of regions of topo II from S. cerevisiae. (a) ATPase domain with ADPNP (in blue) (33). (b) The 92-kDa fragment (amino acids 410 to 1202) (15).

Source: Panel a is reprinted from the Proceedings of the National Academy of Sciences of the United States of America (33); panel b is reprinted from Nature with the permission of the publisher.

A two-gate mechanism for topo II action, similar to that for DNA gyrase, has been proposed (16, 169). The gate segment of DNA is bound across the A′ region of topo II. The binding of ATP to the ATPase region results in the dimerization of the topo II monomers and the capture of the T segment. Topo II cleaves the double-stranded G-segment DNA, with a 4-bp stagger between the cuts in the two strands. The T segment passes through the gap in the G segment and into the cavity formed by the two A′ regions. This G segment is religated, the T segment passes out of the enzyme through the bottom gate, and ATP hydrolysis allows the enzyme to return to its original conformation. As the ATPase region has a relatively small cavity, it has been proposed that the closure of the ATP-operated clamp stimulates strand passage to allow the clamp to fully shut.

Topo VI.

Topo VI is an archaeal type IIB topoisomerase that has recently also been found in plants (18, 82); it is found in all known archaea. It is able to decatenate circular DNA and relax both positive and negative supercoils, and it acts as an A₂B₂ heterotetramer (18). Apart from three motifs in the ATPase domain and the topoisomerase-primase (TOPRIM) fold, topo VI shows no obvious sequence homology to other type II topoisomerases and, therefore, is in its own subfamily (type IIB) (6, 17). The A subunit of topo VI is homologous to a protein called Spo11, which is ubiquitous in eukaryotes and is involved in initiating homologous recombination during meiosis by cleaving DNA. In this sense, Spo11 is similar to a topoisomerase that does not religate the DNA after cleavage (36). The B subunit of topo VI appears to bind and hydrolyze ATP. The structures of the topo VI A subunit (topo VIA) from Methanococcus jannaschii and topo VIB from Sulfolobus shibatae have been resolved (38, 149). A major difference between the A subunit structure of topo VI and those of other type II topoisomerases is the lack of a post-strand passage cavity (149). The ATPase region of topo VIB shows structural similarity to the ATPase regions of the type IIA topoisomerases, despite limited sequence homology (38). The structures of topo VIB in a variety of conformations involving a range of nucleotides have been resolved (37, 38); this work has revealed a detailed outline of the nucleotide hydrolysis events and associated protein conformational changes. It is likely that other topo II enzymes go through a similar series of events. Recently, the structure of intact topo VI has been determined using a combination of X-ray crystallography and X-ray scattering to analyze enzymes from Methanosarcina mazei and Sulfolobus shibatae (35, 80). These structures give valuable insights into the mechanism of strand passage by topo VI and other topoisomerases.

Topo IV.

Topo IV is a bacterial type IIA enzyme that is important in decatenating replication products (116). In addition to having decatenation activity, topo IV can relax positive supercoils and, less efficiently, negative supercoils (44). It requires ATP for enzyme turnover (155). E. coli topo IV consists of two subunits, encoded by the parC and parE genes, which form a heterotetramer (108); in other organisms (e.g., Staphylococcus aureus), the corresponding genes are termed grlA and grlB (64). The ParC subunit (84 kDa in E. coli) and the ParE subunit (70 kDa) are homologous to GyrA and GyrB, respectively (Fig. 6), although topo IV is unable to introduce negative supercoils into DNA (156). Also, topo IV is ~100 times more active at decatenation in vivo in E. coli cells than DNA gyrase (230). Topo IV appears to have a role in chromosome segregation after DNA replication. The activity of topo IV in E. coli is stimulated by positively supercoiled DNA (178). Although topo IV, and not gyrase, is responsible for decatenation in vivo, gyrase mutants have problems decatenating their chromosomes. This finding implies that DNA compaction by gyrase is necessary for the action of topo IV (204). DNA replication is stopped more quickly as a result of mutations in both topo IV and gyrase than as a result of a mutation in gyrase alone (161). It seems therefore that, despite their sequence similarities, gyrase and topo IV have quite distinct cellular roles. Topo IV has the predominant role in decatenation (and unknotting), whereas gyrase is the only supercoiling enzyme (48, 204, 230, 232). Indeed, one of the roles for gyrase (see below) can be viewed as supercoiling DNA catenanes to make them better substrates for topo IV.

The structure of a 43-kDa N-terminal fragment of ParE, in a complex with an ATP analog, has been resolved (13). This structure (Fig. 10) shows significant similarity to that of the corresponding region of GyrB (see Fig. 13). The structure of this domain gives important insight into the mechanism of ATP hydrolysis and the actions of the aminocoumarin antibiotics, which also bind to this region of the protein (see below).

Fig. 10The 43-kDa N-terminal domain of E. coli ParE, with one monomer shown in blue and the other in yellow. ADPNP is depicted in stick form near the topoisomerase of the molecule

Source: . Reprinted from Antimicrobial Agents and Chemotherapy (13).

Fig. 13Structure of the 43-kDa N-terminal domain of E. coli GyrB with ADPNP (221). The protein structure is shown as ribbons, and ADPNP is shown as a space-filling representation.

Although no structures of full-length ParE (or GrlB) have been reported, structures of full-length E. coli ParC and the ParC C-terminal domain (from Bacillus stearothermophilus) have been published (40, 103). The full-length ParC structure resembles those of fragments of GyrA and yeast topo II (15, 63, 142) but has a markedly different conformation, with the DNA gate wide open (Fig. 11a). One of the main structural differences between gyrase and topo IV is in the C-terminal domains of GyrA and ParC. The GyrA C-terminal domain forms a six-bladed β-pinwheel (see Fig. 15) (41); the structure of the C-terminal domain of ParE consists of a broken five-bladed β-pinwheel (Fig. 11b) (40). The C-terminal domain of topo IV is anchored to the N-terminal domain, which would appear to allow only minimal movement of the domain. In contrast, the C-terminal domain of DNA gyrase is connected to the N-terminal domain by a flexible linker, allowing movement (41, 43). This distinction means that topo IV cannot wrap DNA in the same way as DNA gyrase, providing an explanation for the inability of topo IV to negatively supercoil DNA. Indeed, the deletion of the wrapping domain of gyrase converts it into a topo IV-like enzyme (107). Topo IV efficiently decatenates DNA and relaxes positively supercoiled DNA more efficiently than negatively supercoiled DNA (44). The enzyme's ability to sense DNA chirality has been confirmed using single-molecule technology (188).

Fig. 11aStructure of full-length ParC (40).

Fig. 11bhe ParC C-terminal domain forms a broken five-bladed β-pinwheel (40).

Fig. 15Structure of the 35-kDa C-terminal domain of GyrA from B. burgdorferi. (a) The protein structure is shown as ribbons, with each of the six blades being a different color. (b) Representation of the β-pinwheel. One strand is highlighted in red.

Source: Adapted from the Proceedings of the National Academy of Sciences of the United States of America (41) with the permission of the publisher.

DNA Gyrase.

DNA gyrase is a type IIA topoisomerase that is unique in its ability to introduce negative supercoils into relaxed DNA in the presence of ATP (75). It can also perform relaxation (ATP-independent), catenation-decatenation, and knotting-unknotting reactions. In the presence of ATP, gyrase can relax positively supercoiled DNA in a reaction equivalent to the introduction of negative supercoils; in the presence of the nonhydrolyzable analog ADPNP, only limited turnovers occur (190). DNA gyrases are ubiquitous in bacteria, and mainly that from E. coli has been studied. DNA gyrases have also been discovered in plants (32, 211) and in apicomplexan parasites (24, 46) but do not appear to be present in other eukaryotes. E. coli DNA gyrase is an A₂B₂ heterotetramer made up of two 97-kDa GyrA subunits and two 90-kDa GyrB subunits, encoded by the gyrA and gyrB genes, respectively (1, 114, 191, 194). DNA gyrase is an essential bacterial enzyme and, as such, has been the target for several antibacterial agents.

(i) Structure of DNA Gyrase.

Each of the GyrA and GyrB subunits consists of two domains, as revealed by limited proteolysis (Fig. 12) (106, 165). GyrB consists of a 43-kDa N-terminal domain responsible for ATP binding and hydrolysis (21, 221) and a 47-kDa C-terminal domain, which is thought to interact with GyrA and DNA (21). The 47-kDa domain may be further split into two subdomains, the TOPRIM domain and the tail (42). GyrA consists of a 59-kDa N-terminal domain responsible for DNA breakage (101) and a 35-kDa C-terminal domain that wraps DNA and is essential for the ability of DNA gyrase to negatively supercoil DNA (107, 164); the deletion of the C-terminal wrapping domain of GyrA converts gyrase into a conventional (DNA-relaxing) enzyme, like topo IV (107).

Fig. 12Domain structure of the DNA gyrase subunits. GyrA consists of an N-terminal 59-kDa domain and a C-terminal 35-kDa domain. The QRDR and active-site Tyr¹²² are marked. The GyrB subunit consists of an N-terminal 43-kDa domain and a C-terminal 47-kDa domain.

The protein structures of all the gyrase domains, with the exception of that of the 47-kDa C-terminal domain of GyrB, have been resolved. Crystals of the full-length GyrB protein from B. stearothermophilus have been described (200), but the structure has not been resolved. The first gyrase domain structure to be determined was that of the 43-kDa domain of GyrB in a complex with ADPNP (221). The structure is a dimer (Fig. 13). Each monomer consists of an N-terminal ATP-binding site (amino acids 2 to 220) and a C-terminal portion that forms the walls of a central 20-Å cavity, potentially large enough to hold a DNA duplex. The N-terminal portion contains the residues involved in dimer contacts (amino acids 2 to 15) and also four motifs conserved in members of the GHKL ATPase/kinase superfamily of proteins (61). Two residues in the C-terminal portion (Gln³³⁵ and Lys³³⁷) have been shown to interact with ATP (182). The central cavity formed by the C-terminal portion is lined with positively charged arginine residues. Mutagenesis studies revealed that these residues are important for DNA binding and strand passage (196).

The structure of the 59-kDa N-terminal domain of GyrA was resolved in 1997 (142). This structure is also a dimer and contains a 30-Å central cavity (Fig. 14). There are two dimer interfaces, at the top and bottom of the structure. The interface at the top of the dimer contains the active-site tyrosines, which form phosphotyrosine bonds with the 5′ ends of the broken DNA. The region across the dimer interface at the top of the domain provides a positively charged saddle, which is proposed to allow DNA binding.

Fig. 14Structure of the 59-kDa N-terminal domain of E. coli GyrA (142). The protein structure is shown as ribbons, with the active-site tyrosines shown as space-filling representations.

The structure of the 35-kDa C-terminal domain of DNA gyrase from Borrelia burgdorferi has been determined (41). This domain forms a six-bladed β-pinwheel (Fig. 15) in which each β-strand backs against and completes the previous blade of the structure. The outer two-thirds of the surface of the structure is basic in charge, suggesting that this region may be involved in the binding and bending of DNA. The crystal structure of the E. coli GyrA C-terminal domain has also been resolved (174). This domain also forms a circular β-pinwheel fold but takes the shape of a spiral, unlike the B. burgdorferi structure, which is flat.

No high-resolution structures of the whole DNA gyrase enzyme have been presented. A low-resolution structure of the entire GyrA protein has been determined using small-angle X-ray scattering (SAXS) (43). Ab initio modeling shows the 59-kDa N-terminal domain forming a dimeric core, with a pear-shaped density pattern on either side (Fig. 16a). These densities may accommodate the crystal structure of the GyrA C-terminal domain (41) attached to the N-terminal domain by a flexible linker. SAXS has also been used to suggest a structure for the full-length GyrB protein (42). Investigation by analytical ultracentrifugation revealed that GyrB, unlike GyrA, is predominantly a monomer in solution; the molecular envelope of GyrB has a tadpole shape (Fig. 16b). The ATPase domain structure of GyrB (221) fits into the head of this envelope, with the remainder being made up of the TOPRIM fold (15) and the tail subdomain, which can be divided further into tail 1 and tail 2 (42). The structural information from topo II structures (15) implies that GyrB sits above GyrA in the complex (Fig. 9). The SAXS data imply that the GyrB ATPase domains are above the DNA cleavage active site and may be perpendicular to the main plane of the GyrA dimer (42).

Fig. 16Solution structures of GyrA (a) and GyrB (b) as determined by SAXS and ab initio modeling (42, 43). (a) The GyrA59 structure (142) is shown in blue, with the active-site tyrosines shown in yellow as space-filling representations. The GyrA59 carboxy termini are colored green. A model of the C-terminal domain (41) is shown as orange ribbons and is attached to GyrA59 by a linker. (b) The model of GyrB is shown in grey. The GyrB43 structure (221) is shown as blue ribbons. The TOPRIM region (15), tail 1, and tail 2 are shown as red, purple, and green ribbons, respectively.

(ii) Mechanism of Action of DNA Gyrase.

Biochemical characterization of the roles of GyrA and GyrB has revealed details of the mechanism of supercoiling by gyrase. Negative supercoiling occurs via a two-gate mechanism (139) (Fig. 17). A section of DNA termed the gate, or G segment, is proposed to be bound across the top dimer interface of the GyrA N-terminal domains (142). An adjacent section of DNA is wrapped around the C-terminal domain of GyrA, such that it positions a further DNA section, termed the transport, or T segment, across the G segment (89). This wrapping by the GyrA C-terminal domain provides gyrase with its unique ability to supercoil DNA (41, 164, 174). The total length of DNA bound by gyrase is estimated to be between 120 and 150 bp (65, 113, 124, 143, 152). ATP is bound to the N-terminal domains of the two GyrB subunits, resulting in their dimerization and the closure of the clamp. This closure traps the T segment in the complex (105); ADPNP is sufficient for this step to occur (2, 3, 122, 170). The active-site tyrosines form phosphotyrosine bonds with the G segment, generating a double strand break with 4-bp overhangs (101, 143). Two Mg²⁺ ions bound within the TOPRIM fold of GyrB are required for the cleavage of the DNA strands (150). The top dimer interface is pulled apart, and with it the G segment, allowing the T segment to pass through into the cavity formed by the GyrA N-terminal domains. The GyrA cavity carries a positive charge and so provides a favorable environment for DNA (142). The G segment is religated, and the T segment is released through the bottom gate of the GyrA N-terminal domains. It is not currently clear what drives the movement of the T segment at this stage. It has been proposed that it may be the closure of the top gate, which makes the GyrA cavity smaller (215). ATP hydrolysis allows the resetting of the enzyme to perform the reaction again (190). ADPNP is sufficient for a single strand-passage reaction to occur, but the enzyme is then trapped in an inactive state (169). The ATPase activity is stimulated by cleaved DNA in the presence of GyrA (222). It has been proposed that the rate-limiting aspect of the DNA supercoiling reaction is the rate of ADP and phosphate release (7).

Fig. 17Model for negative supercoiling by DNA gyrase. The domains are colored as follows: GyrB N-terminal domain, yellow; GyrB C-terminal domain, orange; GyrA N-terminal domain, dark blue; and GyrA C-terminal domain, light blue. The G and T DNA segments are colored green and red, respectively. (1 and 2) The G segment binds across GyrA, and the GyrA C-terminal domain wraps the DNA to present the T segment. (2 and 3) ATP is bound, which closes the GyrB clamp. (3 and 4) The G segment is cleaved, and the T segment passes through. (4 and 5) The T segment exits through the bottom gate, and the G segment is religated. (5 and 2) The hydrolysis of ATP resets the enzyme.

Source: Reprinted from Biochemistry (12) with the permission of the publisher.

DNA gyrase can also relax negatively supercoiled DNA, and this effect occurs as the reverse of the reaction described previously, with the T segment passing through the enzyme in the opposite direction (223). This reaction occurs in the absence of ATP, as it is energetically favorable. This relaxation activity of DNA gyrase is far less efficient than the supercoiling reaction (73, 95). DNA gyrase can also relax positively supercoiled DNA. This reaction occurs in the same way as negative supercoiling and requires ATP, even though it is energetically favorable (21). The catenation-decatenation and knotting-unknotting reactions performed by gyrase are also ATP dependent (115, 123). The mechanisms of gyrase-catalyzed supercoiling and relaxation and the kinetic steps in these processes have been analyzed in single-molecule experiments (78, 151).

DRUGS AND TOXINS THAT TARGET DNA TOPOISOMERASES

DNA topoisomerases are important targets for antimicrobial drugs. DNA gyrase is essential for the survival of bacteria but is largely absent in eukaryotes and is therefore a good drug target. Although bacterial topo I appears to be nonessential, the discovery of compounds that stabilize the cleavage complex and show antibacterial activity raises the possibility of new antibacterials targeted to topo I in the future (30). At the moment, the only bacterial topoisomerase target is DNA gyrase.

There are two well-known classes of drugs that target gyrase: the quinolones and aminocoumarins (132). The quinolones are synthetic, whereas the aminocoumarins are products of Streptomyces species (Table 2).

Aminocoumarins

The aminocoumarins are more potent inhibitors of gyrase than the quinolones in vitro, but their low solubility and toxicity in eukaryotes make them less useful clinically (131). These compounds are produced by Streptomyces species and include novobiocin, clorobiocin, and coumermycin A₁ (Fig. 18) (96, 97, 175, 184).

Fig. 18Structures of aminocoumarins.

Aminocoumarins have been shown to inhibit supercoiling, leading to the identification of DNA gyrase as the target (76). However, the aminocoumarins do not inhibit ATP-independent relaxation (74, 192), consistent with their being competitive inhibitors of ATP hydrolysis. This conclusion is supported by work showing that novobiocin strongly inhibits the gyrase ATPase reaction, which is relatively unaffected by the quinolone oxolinic acid (140). Aminocoumarin-resistant strains of E. coli frequently contain a mutation of Arg¹³⁶ (34, 50), a residue not directly implicated in ATP binding (221). This discrepancy was explained with the resolution of crystal structures of the N-terminal 24-kDa subdomain of GyrB in complexes with novobiocin and clorobiocin (98, 117, 201); the structure of the corresponding region of E. coli ParE (topo IV) in a complex with novobiocin has also been determined (13). Indeed, topo IV has been shown to be a secondary target of novobiocin (81). There is only a partial overlap between the aminocoumarin drugs and ATP-binding sites, with the novobiose sugar of novobiocin overlapping with the adenine ring of ATP (Fig. 19). Novobiocin forms a hydrogen bond with Arg¹³⁶ and has been shown to prevent the dimerization of the 43-kDa GyrB N-terminal domains (2, 117).

Fig. 19The binding sites of novobiocin and ADPNP in GyrB partially overlap. Part of the N-terminal GyrB structure is shown, with ADPNP in red and novobiocin in blue (117).

The biosynthetic gene clusters for novobiocin, coumermycin A₁, and clorobiocin have been cloned and sequenced (118, 160, 185, 216). All three clusters contain a gene encoding an aminocoumarin-resistant GyrB subunit, and the clorobiocin and coumermycin A₁ gene clusters also encode a ParY subunit, encoding a drug-resistant subunit (ParE) of topo IV (118). Differences among the gene clusters for these compounds correspond to differences in the compound structures; for example, novobiocin contains a methyl group at position 8 on the aminocoumarin ring, and its biosynthetic gene cluster contains novO, a C-methyltransferase gene (118). Clorobiocin contains a chlorine atom at position 8 on the ring, and the corresponding gene (clo-hal) in the clorobiocin gene cluster has similarity to a reduced flavin adenine dinucleotide-dependent halogenase gene. The replacement of the clo-hal gene in the clorobiocin gene cluster with novO results in an 8′-methylated derivative (62). The modification of the aminocoumarin gene clusters by genetic engineering, therefore, has the potential to generate novel antibiotics. For example, analogs of novobiocin and clorobiocin, termed novclobiocins, have been generated (4, 66). Although most of these analogs are less potent than the parent molecules, novobiocin and clorobiocin, some show equal or even greater potency. Moreover, many of the analogs have improved physicochemical properties and present the possibility of developing superior chemotherapeutic agents.

Simocyclinones

Simocyclinones D4 and D8 (see Fig. 22b) are gyrase-inhibiting antibiotics from Streptomyces antibioticus (99, 177). It was noted that some of the genes in the simocyclinone cluster have similarities to those in the aminocoumarin gene cluster, and the simocyclinone structure includes an aminocoumarin moiety (71, 199). Simocyclinone D8 has been shown to be a potent inhibitor of both the supercoiling and relaxation activities of gyrase in vitro (67). However, despite the similarities between simocyclinone D8 and the classical aminocoumarins, they show differences in their modes of action. Simocyclinone D8 does not competitively inhibit the ATPase reaction, nor does it stabilize the gyrase-DNA cleavage complex like the quinolones (67). Simocyclinone D8 appears to act at an early stage in the gyrase catalytic cycle to prevent DNA binding. Simocyclinone D8 is not particularly promising as a drug candidate as a result of its low solubility and toxicity in eukaryotes, but it is hoped that modification of the structure may circumvent these problems. In particular, its novel mode of action suggests that there are further unexploited strategies for inhibiting bacterial gyrase.

Fig. 22bStructure of the DNA gyrase inhibitor simocyclinone D8.

Quinolones

The quinolones are the most therapeutically important class of DNA gyrase inhibitors, and they have been used to treat a wide range of infections (59, 206). The quinolones traditionally have been divided into two categories: the older acidic quinolones, such as nalidixic acid, which act against gram-negative bacteria, and the amphoteric fluoroquinolones, such as ciprofloxacin (Fig. 20). More recently, the quinolones have been classified in terms of the evolution of their structures and clinical indications: narrow-spectrum drugs include nalidixic acid; expanded-spectrum drugs include norfloxacin and ciprofloxacin; broad-spectrum quinolones include levofloxacin and sparfloxacin; and “fourth-generation” drugs include trovafloxacin (112).

Fig. 20Chemical structures of a selection of quinolones. Nalidixic acid and oxolinic acid are examples of older acidic quinolones. Ciprofloxacin and norfloxacin are examples of the amphoteric fluoroquinolones (expanded-spectrum compounds).

The quinolones have been shown to inhibit DNA supercoiling and relaxation by binding to both gyrase and DNA and stabilizing the formation of the gyrase-DNA cleavage complex (74, 192). However, the specific details of their mechanism of action are not yet clear. The inhibition of DNA synthesis is due not to the inhibition of gyrase activity per se but to the quinolone-gyrase-DNA complex's blocking of transcription and the DNA replication machinery and, therefore, the blocking of cell growth (94, 220, 224). This effect is likely to result in the bacteriostatic action of quinolones (60). Lethality is likely to be due to DNA breaks, which are a second step in the process and can occur by both protein synthesis-dependent and -independent routes (60, 129). DNA replication is stopped rapidly when DNA gyrase is targeted with quinolone drugs, apparently due to the collision of replication forks with cleaved complexes (183). For example, norfloxacin has been shown to cause stalled replication forks in vivo; however, this inhibition cannot be the immediate cause of cell death, as it is reversible (79, 159). Also, bacteria in which DNA replication has been inhibited can subsequently be killed by treatment with nalidixic acid or ciprofloxacin (233). The inhibition of DNA replication by quinolones results in the induction of the SOS regulon in a RecBC-dependent manner (136). One of the genes induced is an inhibitor of cell division, so the SOS response results in cell filamentation, which may lead to the slow death of the cell (55). The release of double-strand breaks from several cleavage complexes may cause chromosome fragmentation and leads to rapid cell death (29). Chromosome fragmentation may be dependent on or independent of protein synthesis. Protein synthesis-dependent chromosome fragmentation is inhibited by chloramphenicol, and it has been proposed that a suicide factor is involved (129, 130). There is evidence to support the idea that the RecB and RecC proteins are involved in nalidixic acid-induced breaks (147). The MICs of quinolones are 10- to 100-fold greater in vivo than in vitro (56, 74). This property can be explained by a small concentration of the inhibitor in vivo triggering the downstream responses that lead to cell death. In bacteria, susceptibility to quinolones is always dominant over resistance, as the presence of double-strand breaks is lethal.

Quinolone-resistant strains of bacteria are being increasingly reported (100). Some of these strains are deficient in the uptake of the drug, but many have mutations in gyrase itself. In addition to mutations in gyrase, some strains also have mutations in topo IV, a secondary target for quinolones in some bacteria (110) and the primary target in others, such as S. aureus (64, 148, 227). The mutations in gyrase that confer resistance to quinolones are often found between amino acids 67 and 106 of GyrA (E. coli numbering). This region of the subunit has been termed the quinolone resistance-determining region (QRDR) (Fig. 21). The mutations cluster at the head-dimer interface, near the active-site tyrosines (142); mutations conferring quinolone resistance have also been found in the corresponding region of topo IV (209). Although most quinolone resistance mutations in gyrase occur in GyrA, mutations in GyrB have also been found, in both E. coli and S. Typhimurium (77, 91, 226).

Fig. 21Structure of the E. coli GyrA N-terminal domain dimer (142) showing the position of Tyr¹²² in red and, in yellow, four residues that are commonly mutated in ciprofloxacin-resistant strains of bacteria.

The residues most commonly mutated in ciprofloxacin-resistant strains are Ser⁸³ and Asp⁸⁷ in GyrA (228); corresponding mutations in S. Typhimurium have also been found (92, 166). Both of these residues are in close proximity to the active-site tyrosine, and mutations of these residues are thought to reduce quinolone binding (225). Resistance-conferring mutations outside the traditional QRDR have also been identified. For example, an Ala⁵¹-to-Val mutation results in a sixfold increase in ciprofloxacin resistance (70). Mutations of Asp⁴²⁶ or Lys⁴⁴⁶ of GyrB have also been shown to result in quinolone resistance (226, 229). It appears that these residues may be close to the active-site tyrosine and thus may form part of the quinolone-binding pocket (86). In addition to the quinolone resistance mutations found in gyrase, some clinical strains have recently been shown to have plasmid-borne resistance to quinolones (104). This resistance turns out to be due to pentapeptide repeat proteins, Qnr in E. coli and MfpA in Mycobacterium tuberculosis (208), which can bind gyrase and inhibit its activity (see below).

Although we currently do not have a high-resolution structure of the gyrase-quinolone-DNA complex, there is evidence that quinolones interact with GyrA, GyrB, and DNA (87), leading to a model in which the drugs bind in a pocket at the GyrA-GyrB interface, also interacting with the bases in DNA at the cleavage site. It is thought that this binding pocket is revealed following protein conformational changes that precede the DNA cleavage reaction (86, 144).

Albicidins

Albicidins are antibiotics and phytotoxins produced by the bacterium Xanthomonas albilineans, a plant pathogen responsible for sugarcane leaf scald disease, which have been shown to block prokaryotic DNA replication (20). The target for albicidin is DNA gyrase (84). Albicidin stabilizes the gyrase cleavage complex in an ATP-dependent manner, and cross resistance is seen with mutations of Ser⁸³ in GyrA that result in quinolone resistance. Albicidin does not inhibit ATPase activity. It has therefore been proposed that albicidin acts in a fashion similar to that of the quinolones but requires the presence of ATP (84). This requirement may be because albicidin binds when the enzyme is in a particular conformation following an ATP-mediated step in the supercoiling cycle. DNA gyrase from X. albilineans is distinct from that of E. coli in terms of its enzymatic properties and also because the GyrA subunit confers albicidin resistance (83). In addition, X. albilineans encodes a pentapeptide repeat protein, AlbG, within the albicidin biosynthesis region that also confers albicidin resistance (84); this protein is analogous to Qnr of E. coli and MfpA of M. tuberculosis, which confer quinolone resistance (see below).

Mcc B17

Aside from the range of drugs and small molecules that target gyrase, there are a number of bacterial toxins that can also inhibit gyrase activity. One of these is microcin B17 (MccB17) (Fig. 22a), which is a glycine-rich peptide found in strains of E. coli containing the plasmid-borne mccB17 operon. It induces double-strand cleavage of DNA and inhibits DNA synthesis, and its action results in cell death (93, 210). MccB17 acts by stabilizing the gyrase-DNA cleavage complex in a way similar to the actions of the quinolones, and like the quinolones, it does not require ATP (88, 157). However, the presence of ATP greatly enhances the effect of MccB17 (157). MccB17 requires at least 150 bp of DNA to induce the formation of cleavage complexes; however, the DNA-wrapping domain of GyrA is not required, nor is the ATPase domain of GyrB (158). MccB17 is therefore likely to interact with the N-terminal domain of GyrA, the C-terminal domain of GyrB, and/or DNA. Mutation of the amino acid Trp⁷⁵¹ to Arg in the C-terminal domain of GyrB in E. coli results in MccB17 resistance (49, 88). Several quinolone resistance mutations also result in cross resistance to MccB17, implying that the binding sites of quinolones and MccB17 may overlap (88). Modification of MccB17 has revealed amino acids in the toxin (Asn⁵³ and Asn⁵⁹) that are required for activity (153). Although the structure of MccB17 is not known, a model has been suggested (153), and it is likely that the folded structure is required for biological activity. The novel mode of action of MccB17, distinct from that of quinolone and coumarin drugs, presents scope for the development of small-molecule analogs that could be potential lead molecules for drug design. A pentapeptide protein, McbG, is involved in defense against MccB17 in vivo (176).

Fig. 22aStructure of the DNA gyrase inhibitor MccB17.

CcdB

CcdB is a bacterial toxin which with CcdA makes up the ccd toxin-antitoxin system. Toxin-antitoxin systems are involved in the postsegregational killing of plasmid-cured cells (85). CcdA prevents the action of CcdB by binding tightly to it (19). Cells bearing the F plasmid express both ccdA and ccdB at low levels, but if the F plasmid is lost, CcdA is degraded, leaving no defense against the CcdB toxin (207). CcdB has been shown to target DNA gyrase (19, 138), and a crystal structure of CcdB bound to a section of GyrA has indicated that it acts as a wedge which stabilizes an open conformation of the enzyme with bound DNA (45). The data support a model in which CcdB binds to a conformation of gyrase that is revealed during the strand passage reaction (181).

Pentapeptide Repeat Proteins (Qnr and MfpA)

The pentapeptide repeat proteins are a large family (>500 members) found in prokaryotes and eukaryotes that contain tandemly repeated amino acids (208). Qnr and MfpA were discovered through their abilities to protect bacteria from quinolones (see above), but they have subsequently been shown to also be inhibitors of gyrase; QnrA has also been shown to interact with topo IV (198) in addition to gyrase (197). The plasmid-borne gene, qnr, encodes a 218-amino-acid protein (10). The structure of a protein homologous to Qnr, MfpA from M. tuberculosis, has recently been resolved (90); this work reveals a protein fold, the right-handed quadrilateral β-helix, that resembles B-form DNA. This structure derives from the pentapeptide repeat motifs and suggests that other family members will exhibit this structural feature. MfpA was shown to be a DNA mimic capable of inhibiting gyrase activity by competing with DNA for the G segment-binding site (90). Although these proteins inhibit gyrase, they also confer protection by preventing cleavage complex formation by other agents. Other relevant pentapeptide proteins are McbG, which is involved in defense against MccB17 (176), and AlbG, which confers albicidin resistance (84); it is likely that these proteins also act by binding to the G segment-binding site in gyrase.

Other Inhibitors

Aside from the drugs and toxins mentioned above, there are other compounds that target bacterial gyrases about which we know rather less (132). These include cyclothialidines, clerodicin, and GyrI. Cyclothialidines are natural products from Streptomyces species that inhibit gyrase by binding to the ATP-binding site of GyrB in a manner similar to that of aminocoumarins (134). Although these compounds are potent inhibitors of gyrase, they are not very effective as antibacterial agents. Crystal structures of cyclothialidines bound to the N-terminal subdomain of GyrB reveal the details of drug-protein interactions (117, 154). As with aminocoumarins, knowledge of the structure of the drug-protein complex and the availability of pathway engineering in Streptomyces raise the possibility of making more effective analogs of cyclothialidines. However, it has also been shown that this goal can be achieved with synthetic chemistry (5).

Clerocidin is a diterpenoid natural product that can inhibit bacterial gyrase and eukaryotic topo II (109, 135). It is a cleavage complex-stabilizing drug that differs from quinolones in its action. It has recently been shown to also stabilize cleavage complexes with Streptococcus pneumoniae topo IV (168). GyrI is a small (~18-kDa) protein encoded by E. coli that binds to and inhibits DNA gyrase (145, 146); it was also identified independently as SbmC, a protein imparting resistance to MccB17 (9). Like Qnr and MfpA, GyrI can protect DNA from damage from other agents, such as MccB17 and quinolones (27, 28), but appears to be able to fulfill a broader role, negating the effects of alkylating agents such as mitomycin C and N-methyl- N-nitro- N-nitroguanidine (28).

FUTURE DIRECTIONS

It is clear that DNA topoisomerases have essential roles in DNA metabolism, and new knowledge of their structures and mechanisms will be of fundamental importance. The recently determined structures of a large fragment of yeast topo II in a complex with DNA (57) and of intact topo VI complexes (35, 80) point the way to more ambitious structural studies and a deeper understanding of topoisomerase mechanisms. Key issues include the coupling of ATP hydrolysis to strand passage in type II topoisomerases, the ability of these enzymes to simplify DNA topology (12), and the protein conformational changes that must accompany the topoisomerase reaction. More dynamic information about topoisomerase mechanism would be an important contribution to complement the available crystal structures.

Coupled with new knowledge of topoisomerase structures and mechanisms will be a better understanding of how existing drugs (such as quinolones) work and new strategies for targeting topoisomerases. The key step in the topoisomerase reaction, the stabilization of a break in DNA, is also the Achilles heel of these enzymes that makes them such good drug targets. In the case of the type II enzymes, gyrase and topo IV, there are already a variety of natural products that exploit this step. Better understanding of these drugs and toxins should potentiate the development of new chemotherapeutics. In the case of the type I enzymes, topos I and III, there are no good therapeutic compounds at this point, but recent work points the way to possible drugs in the future (30, 31, 203).

ACKNOWLEDGMENTS

A.M.'s lab is funded by BBSRC; K.E.-R. was supported by a studentship from BBSRC and Syngenta.