Crystal Structure of a Bacterial Type IB DNA Topoisomerase Reveals a Preassembled Active Site in the Absence of DNA*

Type IB DNA topoisomerases are found in all eukarya, two families of eukaryotic viruses (poxviruses and mimivirus), and many genera of bacteria. They alter DNA topology by cleaving and resealing one strand of duplex DNA via a covalent DNA-(3-phosphotyrosyl)-enzyme intermediate. Bacterial type IB enzymes were discovered recently and are described as poxvirus-like with respect to their small size, primary structures, and bipartite domain organization. Here we report the 1.75-Å crystal structure of Deinococcus radiodurans topoisomerase IB (DraTopIB), a prototype of the bacterial clade. DraTopIB consists of an amino-terminal (N) β-sheet domain (amino acids 1–90) and a predominantly α-helical carboxyl-terminal (C) domain (amino acids 91–346) that closely resemble the corresponding domains of vaccinia virus topoisomerase IB. The five amino acids of DraTopIB that comprise the catalytic pentad (Arg-137, Lys-174, Arg-239, Asn-280, and Tyr-289) are preassembled into the active site in the absence of DNA in a manner nearly identical to the pentad configuration in human topoisomerase I bound to DNA. This contrasts with the apoenzyme of vaccinia topoisomerase, in which three of the active site constituents are either displaced or disordered. The N and C domains of DraTopIB are splayed apart in an “open” conformation, in which the surface of the catalytic domain containing the active site is exposed for DNA binding. A comparison with the human topoisomerase I-DNA cocrystal structure suggests how viral and bacterial topoisomerase IB enzymes might bind DNA circumferentially via movement of the N domain into the major groove and clamping of a disordered loop of the C domain around the helix.

Topoisomerases I and II are involved in virtually all DNA transactions and are the targets of clinically effective anti-cancer and anti-infective drugs (1,2). Topoisomerases exploit a tyrosine nucleophile to attack the phosphodiester backbone, yielding a covalent enzyme-DNA adduct on one side of the resulting break that permits the passage of strand(s) through the break. Type I enzymes operate by cleaving one DNA strand and passing another strand through the nick; type II enzymes cleave both DNA strands and allow passage of a duplex segment through the double strand break. Closure of the break by reversal of the cleavage step results in a change to the topology of DNA, with no net effect on its chemical structure.
Vaccinia TopIB is composed of two domains separated by a flexible protease-sensitive linker (22,23). The 234-aa 3 carboxyl-terminal (C) domain contains the active site and catalyzes DNA transesterification and supercoil relaxation but has reduced affinity for DNA compared with the full-length enzyme (24). The 80-aa amino-terminal (N) domain interacts with DNA in the major groove (22,25,26). The atomic structures of the individual domains of vaccinia TopIB have been determined by x-ray crystallography (7,22). Whereas the fold and active site of the predominantly ␣-helical catalytic domain are conserved in nuclear type IB topoisomerases (nuclear TopIB) and the tyrosine recombinases (7), the amino-terminal domain, which consists primarily of ␤-strands (22), has no counterpart in tyrosine recombinases, although it is structurally homologous to part of the much larger amino-terminal domain of nuclear TopIB (27).
Bacterial type IB topoisomerases were discovered recently (5) and are similar to poxvirus type IB topoisomerases with respect to their size, primary structures, and domain organization (Fig. 1). Mutational analyses indicate that the transesterification mechanism and active site constituents of bacterial TopIB adhere closely to those of vaccinia TopIB (5). A major difference between the viral and bacterial enzymes is their cleavage site specificity. Whereas poxvirus and mimivirus TopIB transesterify at a pentapyrimidine cleavage site 5Ј-(C/T)CCTT2 in duplex DNA (6,28), the same element cannot be cleaved by Deinococcus radiodurans topoisomerase IB (DraTopIB), the only member of the bacterial TopIB clade that has been characterized to date (5). Nuclear TopIB is also unable to transesterify at the poxvirus cleavage site (29). It has been proposed that contacts of the poxvirus TopIB with DNA trigger assembly of a competent active site by recruitment of several of the catalytic residues that are either disordered or out of position in the free enzyme (7, 30 -32). Structural studies of tyrosine recombinases underscore the theme that the active site might not be preassembled in the free enzyme prior to productive DNA binding (11,12).
A fuller understanding of target site recognition by type IB enzymes and the conformational changes that accompany DNA binding will depend on capturing the structures of a single enzyme in free and DNAbound states or on structural comparisons of enzymes with different site specificity in the same functional states along the reaction pathway. These goals have been elusive. The structures of a free full-length poxvirus or nuclear TopIB remain unsolved, perhaps because domain motions about the linker are an impediment to crystallization. Whereas the human TopIB has been crystallized bound to DNA (8), no DNA cocrystal has been obtained for the vaccinia enzyme. Here we report the structure of the intact D. radiodurans topoisomerase IB, a prototypal bacterial type IB topoisomerase. The structure verifies the predicted similarity between viral, nuclear, and bacterial type IB enzymes. Important findings are: (i) the active site of DraTopIB is substantially preassembled and seemingly poised for transesterification and (ii) the relative orientation of the catalytic and amino-terminal domains in the apoenzyme clearly needs to change to engage duplex DNA in the circumferential binding mode used by viral and nuclear enzymes (7,8,33). A comparison with the human TopIB-DNA cocrystal structure suggests how viral and bacterial TopIB enzymes might bind circumferentially to DNA.

EXPERIMENTAL PROCEDURES
DraTopIB Purification and Crystallization-DraTopIB was produced and purified as described previously (5). Selenomethionine-substituted protein was produced using the methionine pathway inhibition method (34). For crystallization, the protein was further purified by hydrophobic (Tosoh Biosep; TSK-Gel Phenyl-5PW) and gel filtration chromatography (Pharmacia Corporation; Sephacryl S-100). Purified protein was dialyzed into 50 mM Tris-HCl (pH 7.5), 0.5 M NaCl, 1 mM EDTA, and 1 mM dithiothreitol, and concentrated to 5 mg/ml. Initial crystallization trials, using 96-well crystallization plates set up with a Hydra-II crystallization robot, gave small plate-like crystals in a variety of polyethylene glycol conditions at 10°C, which were refined further. Refined conditions yielded plate-like crystals in 12% polyethylene glycol 3350 (w/v), 0.1 M MES (pH 6.5), 0.2 M ammonium chloride using the hanging drop vapor diffusion method. Typical crystals were .4ϫ.4ϫ.05 mm. Prior to data collection, DraTopIB crystals were transferred to a solution containing 25% glycerol in addition to the mother liquor in several steps (5% glycerol increments/step and soaking for 2 min/step), harvested using a rayon crystal-mounting loop, and flash-cooled in liquid nitrogen. Throughout the transfer, the crystals were kept at 4°C.
Structure Determination and Refinement-All data were collected from a single crystal at 100 K using synchrotron radiation at the DuPont-Northwestern-Dow Collaborative Access Team Synchrotron Research Center at the Advanced Photon Source. Diffraction data were measured at a wavelength corresponding to an energy value of ϳ35 eV above the theoretical absorption edge of selenium (E ϭ 12,693 eV, ϭ 0.9768 Å). To minimize radiation damage, the crystal was translated halfway through data collection. In addition, a low resolution data set was collected from the same crystal after the high resolution data had been measured. All data were processed using XDS (35) and scaled using SCALA software programs (36). DraTopIB crystals belong to space group P2 1 with unit cell dimensions a ϭ 38.09, b ϭ 64.97, and c ϭ 76.62 Å, ␤ ϭ 91.78°and have one molecule in the asymmetric unit. Data collection statistics are listed in Table 1.
Three of five possible selenium sites were identified using the Shake and Bake program (37,38). An additional site was identified after refinement of the heavy atom model using the SHARP program (39). The missing fifth site corresponds to the disordered amino-terminal methionine. The four selenium sites were refined and phases to 1.75 Å were calculated using SHARP. Phases were improved by density modification using Solomon (40) and DM programs (41) as implemented in SHARP. The resulting electron density map clearly showed elements of protein secondary structure. Experimental density for the active site region is shown in Fig. 2. The high quality of the experimental map enabled the use of ARP/warp programs (42) to trace automatically 253 of 346 amino acids in the polypeptide chain mainly corresponding to the C domain, which is better ordered. Additional model building was carried out in the program O (43). The model was refined with the REFMAC5 program (44) using data to 1.75 Å resolution. The program ARP was used to place water molecules into peaks Ͼ1.5 in a 2F o Ϫ F c difference Fourier map and within hydrogen bonding distances. No electron density was seen for several loops in the N domain and one loop in the C domain. As part of the refinement, the translation/liberation/screw parameters were refined for each domain. The final model contained residues 2-16, 20 -24, 39 -138, and 149 -337 plus 321 water molecules and 1 unidentified atom. The model also included 12 amino acids showing alternative side chain conformations. All residues were found within the most favored or allowed regions in the Ramachandran plot. Refinement statistics are listed in Table 1

RESULTS AND DISCUSSION
Overview of the DraTopIB Structure-The structure of DraTopIB was determined by Single Anomalous Dispersion phasing as described under "Experimental Procedures." The final model, containing one Dra-TopIB protomer in the asymmetric unit, was refined at 1.75 Å resolution (R free ϭ 23.2% and r ϭ 19.7%; Table 1). DraTopIB is composed of an N domain spanning aa 1-90 and a C domain from aa 91-346. Most of the amino acids in the C domain were clearly visible in the electron density map (e.g. see Fig. 2), except for a disordered loop from residues 139 -148. The density in the N domain was of poorer quality, with several disordered loops and missing side chains, especially near the amino terminus.
The N domain (Fig. 1, red) comprises three ␣ helices plus a five strand antiparallel ␤-sheet in a tertiary structure similar to that of the N domain of vaccinia TopIB (22). The first helix, ␣0 N , which has no equivalent in vaccinia TopIB, packs against the C domain. The structure of the C domain, which contains the active site, is similar to the catalytic domain of vaccinia TopIB (7). It can be described as a two-lobed fold. tude and E ϭ residual lack of closure error. e R cullis ϭ ͚ʈFh(obs)͉ Ϫ ͉Fh(calc)ʈ/͚͉Fh(obs)͉, where ͉Fh(obs)͉ ϭ observed heavy atom structure factor amplitude and ͉Fh(calc)͉ ϭ calculated heavy atom structure factor amplitude. f R-factor ϭ ͚ʈFo͉ Ϫ ͉Fcʈ/͚͉Fo͉, where ͉Fo͉ ϭ observed structure factor amplitude and ͉Fc͉ ϭ calculated structure factor amplitude. g R free , R-factor based on 5% of the data excluded from refinement. h Numbers in parentheses represent r.m.s.d.
Lobe 1 (Fig. 1, cyan) comprises residues 115-233 and contains four helices plus two ␤-sheets; one sheet is formed by three anti-parallel strands (␤1-␤3) and the second by two anti-parallel strands (␤4-␤5). Lobe 2 (Fig. 1, blue) comprises eight ␣ helices (␣1 and ␣6 -10) formed by residues 91-113 and 234 -346. The last nine amino acids of the protein are part of lobe 2 but are not visible in the crystal structure. The two lobes make extensive interactions, and the active site residues are located at the interface between the lobes.
Similarity to Viral TopIB-A search of the Protein Data Bank using the DALI server (48) and the Structural Classification of Proteins data base (49) using secondary structure matching (50) confirmed that the C domain of DraTopIB is structurally similar to the catalytic domains of vaccinia TopIB, human TopIB, and several tyrosine recombinases, thereby supporting the proposed common ancestry of the TopIB and tyrosine recombinase enzymes (5,7). The search also highlighted the similarity of the N domains of DraTopIB and vaccinia TopIB; otherwise, no significant similarities to any other protein or class of proteins were uncovered.
The N domains of DraTopIB and vaccinia TopIB are superimposable with an r.m.s.d. of 1.05 Å for 42 C␣ atoms; the C domains superposed with an r.m.s.d. of 3.0 Å for 132 C␣ atoms (Fig. 3A). Considered individually, lobes 1 and 2 of DraTopIB superposed on lobes 1 and 2 of vaccinia TopIB with r.m.s.d. values of 2.05 Å and 2.35 Å for 82 and 52 C␣ atoms, respectively. The secondary structure elements of lobe 1 aligned well, except for the presence of a surface ␤-sheet inserted between helices ␣4 and ␣5 of DraTopIB (aa 204 -216) (Fig. 1). A disordered surface loop in DraTopIB (aa 139 -148) corresponds closely to a disordered segment in vaccinia TopIB (aa 129 -136), with an important difference being that there is well defined electron density for the first essential arginine of the catalytic pentad (Arg-137) in DraTopIB immediately preceding the start of the disordered loop, whereas the equivalent Arg-130 side chain of vaccinia TopIB is part of the missing segment (Fig. 1).
Lobe 2 of vaccinia and DraTopIB differ significantly with respect to their secondary structures flanking the tyrosine nucleophile of the active site. Whereas Tyr-274 of vaccinia TopIB is located within a long continuous helix (helix ␣8 from aa 269 -283), the corresponding region of DraTopIB is broken into two shorter helices (␣8Ј and ␣8Љ), such that the Tyr-289 nucleophile is now situated in the loop connecting the short helices ( Figs. 1 and 3C). This has the effect of changing dramatically the position of the catalytic tyrosine, as discussed under "A Preassembled Active Site in DraTopIB." The structures of vaccinia and DraTopIB also differ in the positioning of the ␣9 helices distal to the catalytic tyrosine ( Fig. 1), although the final ␣10 helix does occupy an equivalent position in both proteins.
Similarity to Nuclear TopIB and Insights into Circumferential Binding- Fig. 3B shows a superposition of the N and C domains of DraTopIB on the structure of human nuclear TopIB in its complex with DNA (8). Nuclear TopIB is a much larger protein than vaccinia or Dra-TopIB and contains multiple structural components that have no counterpart in the viral and bacterial proteins. Nuclear TopIB forms a C-shaped clamp around duplex DNA. The two ends of the C-clamp consist of loops (the so-called "lips") that meet in a noncovalent "kiss" to envelop the target site (Fig. 3B). A subset of the ␤-strands of the N domain of DraTopIB that comprises a DNA-binding surface in vaccinia TopIB (24,25,51) is superimposable on a homologous segment of the N domain of human TopIB (Fig. 3B). Although DraTopIB has no structural equivalent of the N domain lip (lip 1) of human TopIB, the ␤4 N -␤5 N loop of DraTopIB would occupy a position on the same face of the DNA as lip 1 of the human enzyme and thus might engage in lip-like contacts to the C domain when DraTopIB is bound to DNA.
The C domain of DraTopIB superposes on the catalytic domain of human TopIB with an r.m.s.d. of 2.46 Å for 173 C␣ atoms. The similarity improves when component lobes 1 and 2 are considered separately (r.m.s.d. of 1.78 Å for lobe 1 for 108 C␣ atoms and 1.44 Å for lobe 2 for 65 C␣ atoms). The DraTopIB and human TopIB C domain structures diverge downstream of the helices that flank the tyrosine nucleophile. The environment of the catalytic tyrosine is similar in human and Dra-TopIB, where it is situated within a loop connecting two conserved helices. In this sense, the free DraTopIB and DNA-bound human TopIB structures have more in common with each other than with the vaccinia TopIB apoenzyme. On the other hand, the two strand ␤-sheet on the surface of DraTopIB (aa 204 -216), which is missing in vaccinia TopIB, is also not present in the human enzyme.
A noteworthy finding from the superposition is that the disordered loop in the C domain of DraTopIB (aa 139 -148), which is similarly disordered in vaccinia TopIB, corresponds to lip 2 of the human TopIB structure (aa 490 -503). The structural elements immediately preceding and following the disordered loop of DraTopIB superpose well on the human enzyme, suggesting that the disordered loop: (i) becomes ordered only when the enzyme binds to DNA and (ii) is likely to serve a function analogous to lip 2 of the human TopIB in stabilizing a circumferential protein-DNA clamp. In human TopIB, the interaction between the two lips entails a salt bridge between a lysine in lip 1 and an aspartate in lip 2. The cross-domain contact, which is necessarily transient to allow ingress and egress of DNA from the clamp, can be covalently stabilized by replacing opposing pairs of N and C domain side chains with cysteines that form a disulfide bond when human TopIB is bound to DNA (52,53).
Although footprinting and cross-linking studies of the vaccinia enzyme provide the first evidence that TopIB binds circumferentially to DNA (33), it remains unclear how the clamp assembles for members of the viral/bacterial TopIB enzyme subfamily. An unresolved question is  Fig. 1) on the two domains of vaccinia TopIB (yellow). B, the DraTopIB N and C domains were superposed separately on the human TopIB structure in its noncovalent complex with DNA (8). The DraTopIB N domain is superposed on the core subdomain I structure of the human protein, whereas the C domain of DraTopIB is superposed on the core subdomain III and the carboxyl-terminal domain of the human TopIB. Core subdomain II in human TopIB has no equivalent in the bacterial or viral TopIB proteins. Human TopIB is shown in magenta, except for the core subdomain II, which is colored gray. C, close-up view of the active site regions of Dra, human, and vaccinia TopIB. The stereo diagram shows the region around the active site. In particular, it shows that, in vaccinia TopIB, the active site tyrosine (Tyr-274) resides in a long helix and is located far away from the other residues in the pentad. In DraTopIB and human TopIB, the helix melts at the center and brings the catalytic tyrosine (which was mutated to Phe in the structure of human TopIB) closer to the other catalytic residues. The color coding is identical to the one in A and B. and DraTopIB ( 143 YARQ 146 ) have a tyrosine and a glutamate/glutamine as putative equivalents of the glutamates of the human enzyme. Mutation of Tyr-136 to alanine in vaccinia TopIB was reported to suppress DNA supercoil relaxation by a factor or 100 (54) and reduce the rate of DNA cleavage by more than two orders of magnitude while exerting a more modest effect on the rate of religation (15). Based on these data, it has been proposed that Tyr-136 facilitates a precleavage activation step (15). In light of the present structural considerations, we speculate that this step corresponds to participation of the loop in clamp closure.
A Preassembled Active Site in DraTopIB-The five amino acids of DraTopIB that comprise the catalytic pentad (Arg-137, Lys-174, Arg-239, Asn-280, and Tyr-289) are well defined in the electron density map (Fig. 2) and are located at the interface between the two lobes of the C domain (Fig. 1). Tyr-289 is pointing toward the protein surface, as befits its role as the nucleophile in the DNA cleavage reaction. Tyr-289 interacts through a water molecule with Arg-239 (Fig. 4). Asn-280 accepts hydrogen bonds from both Arg-239 and Arg-137 (Fig. 4).
A comparison of the active sites of DNA-bound human TopIB and DraTopIB shows that the catalytic residues are in nearly identical positions (Fig. 3C). Thus, the active site of DraTopIB is effectively preassembled in the apoenzyme crystal into a conformation that is ready (or nearly ready) to engage in transesterification. This contrasts sharply with the vaccinia apoenzyme crystal structure, in which only two cata-lytic residues (Arg-223 and His-265) occupy positions analogous to their counterparts in human TopIB (Arg-580 and His-632) or Dra-TopIB (Arg-239 and Asn-280). In vaccinia TopIB, the Tyr-274 nucleophile is grossly displaced from the active site (Fig. 3C), as is the catalytic residue Lys-167 (equivalent to human/Dra Lys-532/174); the essential vaccinia Arg-130 side chain is not seen at all in the structure, because it resides within the disordered loop.
The functions of the four residues that catalyze the attack of the tyrosine nucleophile on DNA have been elucidated by transient state kinetic analysis of vaccinia TopIB, entailing dissection of the effects of subtle chemical and stereochemical modifications of the scissile phosphodiester, alone and in tandem with mutations in the enzyme active site (16, 19 -21, 55, 56). In brief, transesterification entails an S N 2-type displacement of the 5Ј-O-leaving strand by the attacking tyrosine nucleophile through a pentacoordinate phosphorane transition state. The enzyme exploits both transition state stabilization and general acid catalysis to achieve an estimated 10 9 -10 12 chemical rate enhancement. The individual catalytic residues Arg-130, Lys-167, Arg-223, and His-265 accelerate the chemical step by factors of 10 5 , 10 4 , 10 5 , and 10 2 , respectively (13)(14)(15). Lys-167 and Arg-130 comprise a proton relay that catalyzes the expulsion of the 5Ј-O of the leaving DNA strand (19,21). His-265 donates a hydrogen bond to the nonbridging pro-Sp oxygen of the scissile phosphodiester to stabilize the transition state (16,56). The role of Arg-233 is to counter the extra negative charge developed in the transition state (20).
The present structure of DraTopIB, together with initial mutational analysis (5), is entirely consistent with similar roles for Arg-137, Lys-174, Arg-239, and Asn-280. A signature feature of bacterial TopIB enzymes (5), as well as mimivirus TopIB (6), is that their catalytic pentads have an asparagine in lieu of the histidine found in poxvirus and nuclear TopIB. This difference is functionally benign insofar as replacement of His-265 of vaccinia TopIB with Asn has only a 3-fold effect on DNA relaxation and single turnover DNA cleavage (14).
Structural Comparisons Highlight Two Types of Precleavage Conformation Transitions-The addition of a full-length DraTopIB apoenzyme structure to the ensemble that includes free vaccinia TopIB domains and multiple structures of human TopIB bound to DNA (8,(57)(58)(59) fills a gap in understanding the protein conformational steps that accompany DNA binding. The earlier comparison of the free vaccinia and DNA-bound human catalytic domains (plus tyrosine recombinase catalytic domains) emphasized the incompleteness of the active site in the free state and the need to trigger its full assembly upon DNA binding. It was also evident that the relative positions of the N and C Comparison of the C domain structures provides a roadmap for active site assembly by the vaccinia TopIB that minimally entails recruitment of three of its essential residues, Tyr-274, Arg-130, and Lys-167. Bringing Tyr-274 to the active site apparently occurs by melting a long helix to form two smaller helices and a loop. This melting represents a structurally significant conformational switch, conceptually similar to the ϳ20-Å movement of the tyrosine nucleophile of bacteriophage integrase (a prototypal tyrosine recombinase) during its transition from the apoenzyme to the DNA-bound state (60). The melted helix-(Tyr-274)-helix conformation of DraTopIB is apparently stabilized by a bidentate hydrogen bond from Thr-244 O-␥ to the backbone carbonyl of Tyr-274 and the backbone amide of Cys-291. Previous mutational analysis of the equivalent Asn-228 residue of vaccinia TopIB has shown that the hydrogen bonding potential of this side chain is important for DNA cleavage (but not religation) and that Asn can be functionally replaced by serine, which is the natural occupant of this position in human TopIB (30). Indeed, it was explicitly proposed that the Asn side chain of vaccinia TopIB recruits the tyrosine nucleophile to the active site by rearranging the tyrosine-containing helix (30). The structure of DraTopIB supports the proposal.
Arg-130 of vaccinia TopIB, which is initially disordered, apparently becomes ordered as the lip-like loop of the C domain docks around the DNA duplex. Lys-167 sits atop the conformationally flexible loop between the second and third strands of a ␤-sheet found in all TopIB and tyrosine recombinase enzymes. Modeling shows that the "general acid loop" (␤2-␤3 loop) of the vaccinia TopIB apoenzyme lies outside of the DNA helix and must traverse the circumference of the helix so that the lysine can engage the substrate in the minor groove (7,19). The Lys-167 side chain is disordered in the crystal structure of free vaccinia TopIB. The corresponding lysine side chain is similarly disordered in three of the four protomers of HP1 integrase in its free crystal structure (Lys-230) (10). The conformation of the loop in HP1 integrase also varies in each of the four protomers with respect to whether it points toward or away from the body of the catalytic domain. The ␤2-␤3 loop of Cre recombinase, which is well ordered in the covalent Cre-loxA complex, is poorly ordered and not visualized in the complex of Cre bound noncovalently to Holliday junction HJ2 (61) and the "nonreactive" protomer of CreY324F bound to loxS (62). For HP1 integrase and Cre recombinase, movement of the lysine in and out of position may coordinate the action of individual protomers within a tetrameric synaptic complex. DraTopIB and integrase (10) are seemingly exceptional among TopIB/recombinases in that their lysine general acid is positioned at the active site prior to productive DNA binding.
The question arises: why is the active site of the poxvirus TopIB apoenzyme so incomplete when that of DraTopIB is preassembled without DNA? One explanation is that the specificity of vaccinia TopIB for transesterification at the pentapyrimidine sequence 5Ј-(T/ C)CCTT2 is enforced by a requirement for target site recognition prior to assembly of the active site. This requirement is sensible given that noncovalent binding of vaccinia TopIB to duplex DNA is relatively nonspecific (63). Site recognition to trigger transesterification entails contacts to specific DNA bases and backbone phosphate oxygens that flank the scissile phosphodiester (31,32). The sequence preferences (if any) for cleavage by DraTopIB are not yet known. It is conceivable that Dra-TopIB, similar to nuclear TopIB (64,65), is fairly promiscuous in DNA cleavage and might prefer to have a cleavage-ready active site in the free enzyme.
Even with a preassembled active site, DraTopIB must undergo a major rearrangement of the N and C domains when it binds to DNA. The apoenzyme crystal captures an open conformation in which the surface of the C domain containing the active site is exposed and the N domain is located far away from the active site. This could represent one of the conformations prior to the encounter of DraTopIB with duplex DNA. When DNA is contacted, the N domain could swivel about the ␣3 N -␣1 loop that connects it to the C domain creating a continuous helix-spanning ␣3 N and ␣1 and allowing the N domain to dock into the major groove on the face opposite the scissile phosphodiester, as modeled in Fig. 5. The loop between the domains is ordered in the apoenzyme but does not adopt a defined secondary structure. Thus, the linker is likely to be flexible (as it is in vaccinia TopIB) and serve as a fulcrum for the proposed domain movements. The model of DraTopIB in complex with DNA, which is guided by the structural alignment of the N and C domains to the DNA-bound human TopIB structure (8), recapitulates features of a vaccinia TopIB-DNA model (7). It incorporates insights gained presently on the likely existence of a lip 2-like loop in the C domain that forms part of the DNA clamp and the possibility that the loop might fully encircle the duplex via contacts with the N domain. The N domain is modeled with strand ␤5 N in the major groove. The equivalent ␤-strand in vaccinia TopIB contains two amino acids, Tyr-70 and Tyr-72, that cross-link to cytosine bases in the major groove of the CCCTT target site (25). The equivalent residues are Tyr-77 and Tyr-79 in DraTopIB and Tyr-426 and Met-428 in human TopIB. As expected, the model shows a good fit of the catalytic pentad residues of DraTopIB around the scissile phosphodiester, with Tyr-289 oriented appropriately for attack on the phosphorus. Although the model does not provide accurate atomic details, it represents a useful guide to further experiments while efforts are underway to attain crystal structures of the viral and bacterial TopIB enzymes bound to DNA.