Mycobacterium tuberculosisDNA gyrase ATPase domain … · 2017. 9. 23. · DNA gyrase, a type II topoisomerase, regulates DNA topology by creating a double-stranded break in one DNA

Biochem. J. (2013) 456, 263–273 (Printed in Great Britain) doi:10.1042/BJ20130538 263

Mycobacterium tuberculosis DNA gyrase ATPase domain structures suggesta dissociative mechanism that explains how ATP hydrolysis is coupled todomain motionAlka AGRAWAL*1, Melanie ROUE†‡§1, Claus SPITZFADEN*, Stephanie PETRELLA†‡§, Alexandra AUBRY‖¶**, Michael HANN*,Benjamin BAX*2 and Claudine MAYER†‡§*Platform Technology Sciences, GlaxoSmithKline, Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire SG1 2NY, U.K., †Unite de Microbiologie Structurale,Institut Pasteur, 75015 Paris, France, ‡URA 2185, CNRS, 75015 Paris, France, §Universite Paris Diderot, Sorbonne Paris Cite, Cellule Pasteur, 75015 Paris, France, ‖UPMC UniversiteParis 06, ER5, EA 1541, Laboratoire de Bacteriologie-Hygiene, Paris, France, ¶AP-HP, Hopital Pitie-Salpetriere, Laboratoire de Bacteriologie-Hygiene, Paris, France, and **CentreNational de Reference des Mycobacteries et de la Resistance des Mycobacteries aux Antituberculeux, Paris, France

DNA gyrase, a type II topoisomerase, regulates DNA topologyby creating a double-stranded break in one DNA duplex andtransporting another DNA duplex [T-DNA (transported DNA)]through this break. The ATPase domains dimerize, in thepresence of ATP, to trap the T-DNA segment. Hydrolysis of onlyone of the two ATPs, and release of the resulting Pi, is rate-limiting in DNA strand passage. A long unresolved puzzle ishow the non-hydrolysable ATP analogue AMP-PNP (adenosine5′-[β,γ -imido]triphosphate) can catalyse one round of DNAstrand passage without Pi release. In the present paper wediscuss two crystal structures of the Mycobacterium tuberculosisDNA gyrase ATPase domain: one complexed with AMP-PCP(adenosine 5′-[β,γ -methylene]triphosphate) was unexpectedlymonomeric, the other, an AMP-PNP complex, crystallized asa dimer. In the AMP-PNP structure, the unprotonated nitrogen(P-N = P imino) accepts hydrogen bonds from a well-ordered

‘ATP lid’, which is known to be required for dimerization. Theequivalent CH2 group, in AMP-PCP, cannot accept hydrogenbonds, leaving the ‘ATP lid’ region disordered. Further analysissuggested that AMP-PNP can be converted from the imino(P-N = P) form into the imido form (P-NH-P) during the catalyticcycle. A main-chain NH is proposed to move to either protonateAMP-P-N = P to AMP-P-NH-P, or to protonate ATP to initiateATP hydrolysis. This suggests a novel dissociative mechanismfor ATP hydrolysis that could be applicable not only to GHKLphosphotransferases, but also to unrelated ATPases and GTPasessuch as Ras. On the basis of the domain orientation in ourAMP-PCP structure we propose a mechanochemical scheme toexplain how ATP hydrolysis is coupled to domain motion.

Key words: ATPase domain, ATP hydrolysis, dissociative mecha-nism, DNA gyrase.

INTRODUCTION

Type II DNA topoisomerases are essential nucleic acid-dependentnanomachines present in all organisms. They solve topologicalproblems of DNA by temporarily introducing a double-strandedbreak into one DNA segment and transporting another DNA [T-DNA (transported DNA)] duplex through this temporary break[1,2]. Most bacteria have two type II topoisomerases (DNA gyraseand Topo IV), which are are heterotetrameric enzymes composedof two subunits (A2B2). However, the Mycobacterium tuberculosisgenome posseses only one type II topoisomerase, DNA gyrase.M. tuberculosis DNA gyrase is capable of introducing negativesupercoils into DNA and has decatenation activity [3]. Thestructures of all of the M. tuberculosis GyrA domains and theC-terminal domain of M. tuberculosis GyrB have previously beendetermined [4,5]. In the present study, the focus is on the structureof the N-terminal ATPase domain of M. tuberculosis GyrB.

Type II topoisomerases function as ATP-dependent clamps(Figure 1). The ATP-gate closes to capture the T-DNA segmentand remains closed when the central G-DNA (gate DNA) opens.Capture of a T-DNA segment brings the ATPase domains closer

together [6] in a way that stimulates both ATP-gate dimerization[7,8] and ATPase activity [9,10].

The structure of the ATPase domain of the GyrB subunit ofEscherichia coli DNA gyrase [7] showed that it contained twostructural domains, an N-terminal GHKL domain [11] and a C-terminal transducer domain (Figure 1). A change in the relativepositions of the GHKL and transducer domains in responseto the hydrolysis of the first ATP [12,13] is important in guidingthe T-DNA segment through the cleaved DNA-gate during thecatalytic cycle [1]. The structures of eukaryotic [12,14], bacterial[7,8,15] and archaeal [13,16] ATPase domains in complex withthe non-hydrolysable ATP analogue, AMP-PNP (adenosine 5′-[β,γ -imido]triphosphate), are dimeric in the crystal, with theGHKL and transducer sub-domains of each monomer held in an‘ATP-restrained’ conformation [13]. In contrast, the complexeswith inhibitors or the apo forms are in an ‘open’ or ‘relaxed’conformation [13,16,18]. The GHKL domain is named after DNAgyrase, HSP90 (heat-shock protein of 90 kDa), histidine kinasesand MutL, four proteins all found to possess domains with thesame ATP-binding fold [11]. Three of these, known as the GHLATPases, transfer the terminal phosphate to a water (or OH− ion),

Abbreviations used: AMP-PCP, adenosine 5′-[β,γ-methylene]triphosphate; AMP-PNP, adenosine 5′-[β,γ-imido]triphosphate; ASEC, analytical size-exclusion chromatography; AUC, analytical ultracentrifugation; CSD, Cambridge Structural Database; HSP90, heat-shock protein of 90 kDa; P-loop,phosphate-binding loop; T-DNA, transported DNA.

1 These authors contributed equally to this work.2 To whom correspondence should be addressed (email [email protected]).

Co-ordinates and structure factor files for the M. tuberculosis GyrB ATPase domain have been deposited in the PDB under the accession codes 3ZKB,3ZKD and 3ZM7.

c© The Authors Journal compilation c© 2013 Biochemical Society

http://getutopia.com/documents

264 A. Agrawal and others

Figure 1 Domains in the GyrB subunit of M. tuberculosis DNA gyrase and activity of the ATPase domain

(A) M. tuberculosis DNA gyrase consists of two subunits GyrA and GyrB. The ATPase domain of GyrB (residues 1–427) contains two structural domains: a GHKL domain (residues 21–255) anda transducer domain (residues 256–427). (B) Simplified schematic diagram illustrating how the ATP-gate closes to capture the T-DNA segment (DNA, black cylinders) and remains closed whenthe central DNA-gate opens. GyrA subunits are white and structural domains in GyrB subunits are coloured as in (A). (C) ATPase activity of the ATPase domain (MtbGyrB47) was assessed (nM ofphosphate produced per s) at room temperature for 30 min at 15 μM protein and various ATP concentrations.

and also have the transducer domain and a residue equivalent toGlu42 in E. coli GyrB [11]. Jackson and Maxwell [19] identifiedGlu42 as a key catalytic residue in the ATPase reaction of DNAgyrase and proposed that it acts as a general base polarizing awater molecule for nucleophilic attack on the γ -phosphate. Ina comprehensive paper, Corbett and Berger [13] solved structureswith several ADP and ATP analogues to structurally dissectATP turnover in the prototypical GHL ATPase TopoVI. Thisled to the proposal of a detailed mechanism [13] for ATPhydrolysis (Supplementary Figure S1A at http://www.biochemj.org/bj/456/bj4560263add.htm), which is initiated when theconserved glutamate residue (equivalent to Glu42 in E. coliGyrB) abstracts a proton from a water (consistent with theoriginal proposal of Jackson and Maxwell [19]). In contrast, in adissociative mechanism [20], the breaking of the bond between theβ- and γ -phosphates (Supplementary Figure S1B), gives a highlyelectrophilic metaphosphate [PO3]− ion and a protonated ADP2 −

ion (Supplementary Figure S2 at http://www.biochemj.org/bj/456/bj4560263add.htm).

In the present paper, we report crystal structures of theM. tuberculosis GyrB ATPase domain (MtbGyrB47) withtwo different non-hydrolysable ATP analogues; purification,crystallization and data collection are reported elsewhere [21].A structure with AMP-PCP (adenosine 5′-[β,γ -methylene]triphosphate) is the first non-dimeric structure of a type IItopoisomerase ATPase domain with an ATP analogue, whereastwo AMP-PNP structures are dimeric and similar to previouslydetermined ‘ATP-restrained’ structures. This gives new insightsinto the catalytic cycle of M. tuberculosis DNA gyrase, andprovides a structural explanation of why ‘non-hydrolysable’AMP-PNP can drive one round of strand passage in type IItopoisomerases [22,23], whereas, with ATP, the release of Pi fromthe hydrolysed ATP is the rate-limiting step [24,25]. We notethat whereas the bridging nitrogen of AMP-PNP (SupplementaryFigure S2) is normally protonated (P-NH-P, imido), in thepresence of divalent metal ions, it is often unprotonated (P-N = P,imino) [26].

MATERIALS AND METHODS

Protein expression, purification, crystallization and data collection

Two constructs, MtbGyrB47C1 and MtbGyrB47C2, for the ATPasedomain of M. tuberculosis H37Rv GyrB (residues 1–427) were de-signed independently in two different laboratories (GlaxoSmithK-

line, U.K. and Pasteur Institute, France respectively). Both con-structs coded for residues 1–427 of M. tuberculosis GyrB, but theyhad slightly different N-terminal His6 tags. For MtbGyrB47C1,the His6 tag was systematically cleaved. For MtbGyrB47C2, theprotein was purified with the His6 tag intact. The N-terminal tagswere not seen in crystal structures and made, within experimentalerror, no significant difference in ATPase activity assays. Detailsof the expression, purification, crystallization and data collectionon crystals of MtbGyrB47C1 with AMP-PNP and of MtbGyrB47C2

with AMP-PCP are reported elsewhere [21].

ATP hydrolysis assays

ATPase activity of the M. tuberculosis DNA gyrase was assessedat the Pasteur Institute by measurement of free Pi using thepyruvate kinase/lactate dehydrogenase assay described previously[27]. The reaction mixture (100 μl) contained 50 mM Tris/HCl(pH 7.5), 50 mM KCl, 5 mM MgCl2, 0.25 mM NADH, 1 mMphosphoenolpyruvate, 2 units of pyruvate kinase and 2 unitsof lactate dehydrogenase,various amounts of ATP and M.tuberculosis DNA gyrase [an equimolar mixture of GyrA andGyrB subunits in 50 mM Tris/HCl (pH 8) and 50 mM NaCl].Reactions were performed in the absence or presence of DNA(relaxed pBR322, 5 μg/μl) at 37 ◦C and the decrease in NADHconcentration was monitored continuously as a function of timefor 90 min by measuring the absorbance at 340 nm in a UV–visiblespectrophotometer. The Km and Vmax values were determined fromthe double-reciprocal plots.

ATPase activities of the ATPase domain and GyrB subunit of theM. tuberculosis DNA gyrase were assessed at GlaxoSmithKlineby measurement of free Pi using the fluorescence methoddescribed in [28]. The proteins were first buffer-exchangedinto 20 mM Tris (pH 7.5), 100 mM NaCl and 1 mM EDTA toremove DTT, as this can produce erroneous results. Assays werecarried out in 10 μl aliquots in triplicate, and reactions werefollowed for 30 min at room temperature (22 ◦C). Fluorescencewas measured using a Gemini microplate spectrofluorometerand Softmax Pro (Molecular Devices). Initial rates were corre-lated with a phosphate calibration curve and kinetics wereanalysed with Grafit (Erithacus Software).

Structure determination and refinement

The structure of the P1(8) crystal form of MtbGyrB47C1

with AMP-PNP was determined by molecular replacement


http://www.biochemj.org/bj/456/bj4560263add.htm




ATP hydrolysis by the M. tuberculosis DNA gyrase ATPase domain 265

Table 1 Refinement parameters

Values in parentheses correspond to the highest-resolution outer shell. Data collection parameters are in [21].

Parameter AMP-PNP AMP-PNP AMP-PCP

Space group P1 P1 P21

Number of subunits in the asymmetric unit 8 16 6Resolution (A) 40–2.95 (3.0–2.95) 25–2.9 (2.95–2.9) 50–3.3 (3.5–3.3)Resolution range (A) 25–2.95 25–2.90 25–3.3Completeness (%) 84.6 98.2 99.7Number of reflections 61070 159159 40735Rwork/R free 18.2/25.7 18.2/24.0 19.6/22.3Number of atoms

Protein 22821 45997 16012Heteroatoms 371 955 283Solvent 115 439 91

Mean B value (A2) 80.3 99.1 113.9RMSDs

Bond length (A) 0.009 0.009 0.010Bond angles (◦) 1.16 1.10 1.19

Ramachandran plotFavoured (%) 87.9 90.3 90.2Allowed (%) 11.9 9.1 9.8Disallowed (%) 0.2 0.1 0.0

with PHASER [29] using the 43 kDa E. coli ATPase domain(40% amino acid identity) as the search model (PDB code 1EI1)[8] and refined. The P1(16) crystal form of MtbGyrB47C1 (withAMP-PNP) was solved by molecular replacement from the P1(8)crystal form and refined (statistics in Table 1). In the final modelsof both of the AMP-PNP crystal forms the bridging nitrogenhas been modelled in the imino form (P-N = P). Refinementof the AMP-PNP structures was carried out with Refmac [30],phenix.refine [31] and Buster [32]. Restraint dictionaries forthe imino (P-N = P) form of AMP-PNP were generated witheLBOW [33] and fit [34] into difference maps calculatedwith phenix.refine [31]. In the final 2.9 Å (1 Å = 0.1 nm) AMP-PNP structure the main-chain NH groups of Leu120 and Gly122

are 3.09 and 3.12Å from the bridging nitrogen (quoted distancesare averages of sixteen observations for each in the asymmetricunit; errors on distances estimated to be ∼0.1 Å). Analysis ofa closely related higher-resolution structure with AMP-PNP, a1.87 Å human Topo II structure (PDB code 1ZXM), confirmedthat the bridging nitrogen between the β- and γ -phosphateAMP-PNP could not be protonated (Supplementary Figure S3at http://www.biochemj.org/bj/456/bj4560263add.htm). Smallmolecule crystal structures in the CSD (Cambridge StructuralDatabase) [35], which have a bridging nitrogen between twophosphate groups, were analysed with Conquest, Mercury andMogul [36]. This analysis suggested that if two oxygens fromthe two phosphate groups co-ordinate a divalent metal ion,then the bridging nitrogen of AMP-PNP was likely to be in theimino (P-N = P or P = N-P) form (Supplementary Figure S2). Inthis unprotonated form the two P-N bond lengths are nearly equalin length (1.59 Å), suggesting delocalized π -bonding. In small-molecule crystal structures in the CSD in which the nitrogenbetween the two phosphates was protonated (P-NH-P, imido; seeSupplementary Figure S2), the P-N bond lengths were longer(1.64 Å), the P-N-P bond angle was 130 +− 1◦ (compared with123 +− 3◦ in the unprotonated structures) and co-ordination of adivalent metal ion was not observed. In small-molecule crystalstructures and in protein complexes with AMP-PNP, the presenceof a divalent metal ion tends to bring the two co-ordinatingoxygens from the two phosphates quite close together (<3 Å),so that looking along the virtual P-P bond, the two phosphate

groups are nearly eclipsed (Supplementary Figure S3). In P-NH-Pstructures looking along the virtual P-P bond the two phosphatestend to be staggered (results not shown), presumably becauseof electrostatic repulsion between negatively charged phosphateoxygens. Whereas in both the imido (P-NH-P) and imino (P-N = P) forms of AMP-PNP the nitrogen appears largely SP2 incharacter, analysis of small-molecule crystal structures suggeststhat in AMP-PCP the CH2 group is SP3 in character, with P-Cbond distances of 1.80 Å and a P-C-P bond angle of 115 +− 3◦.

The structure of MtbGyrB47C2 with AMP-PCP was determinedby molecular replacement with PHASER [29] using GHKL andtransducer domains from the refined MtbGyrB47–AMP-PNP asthe search models. Structure refinement was carried out withBUSTER [32] to 3.3 Å. Model building was performed withCoot [37]. Model refinement statistics are summarized in Table 1and structure factors and co-ordinates have been deposited in thePDB under codes 3ZKB (2.9 Å, AMP-PNP), 3ZKD (2.95 Å,AMP-PNP) and 3ZM7 (3.3 Å, AMP-PCP). Structural Figureswere drawn with PyMOL (http://www.pymol.org) unless statedotherwise.

AUC (analytical ultracentrifugation) and analytical ASEC(analytical size-exclusion chromatograohy)

Sedimentation velocity experiments were performed at 15 μMprotein (MtbGyrB47C2 or His6–MtbGyrB47C2) either in Tris buffer[50 mM Tris (pH 8) and 50 mM NaCl, +− 5 mM AMP-PCP or5 mM AMP-PNP] or in PBS (pH 7.4, +− 2 mM AMP-PNP and+− 2 mM Mg or 2 mM EDTA). The PBS samples were gentlyresuspended at the end of the AUC run and the run repeatedwith the same sample cell after 3 or 6 days of incubation at roomtemperature.

Experiments were performed in a Beckman XL-I analyticalultracentrifuge using a double sector charcoal-Epon cell at 20 ◦Cand 42000 rev./min. Interference scans were taken every 6 min.The program Sednterp 1.09 (available at http://www.jphilo/mailway.com/download.htm) was used to calculate solventdensity, solvent viscosity and partial specific volume using theamino-acid composition. The sedimentation data were analysed



http://www.pymol.org

http://www.jphilo/mailway.com/download.htm

http://www.jphilo/mailway.com/download.htm


with the program Sedfit [39] using the continuous c(s) andc(M) distributions. The theoretical sedimentation coefficient valuecalculated from the crystallographic monomer structure withoutthe ATP lid and insertion region (see below) was 3.8 S. AUCexperiments showed that the MtbGyrB47C2 behaved as a monomerin the absence or presence of AMP-PCP, irrespective of thepresence or absence of the N-terminal His6 tag.

The oligomerization state of MtbGyrB47C1 and MtbGyrB47C2

in solution was also assessed by ASEC. MtbGyrB47C1 was mixedwith 5 mM MgCl2 plus 1 mM AMP-PNP (Sigma) and applied to aTOSOH SW3000 column in 20 mM Hepes (pH 7.5) and 100 mMNa2SO4 at 0.2 ml/min. Cytochrome c (12.4 kDa), carbonicanhydrase (29 kDa), BSA (66 kDa), alcohol dehydrogenase(150 kDa), β-amylase (200 kDa), apoferritin (443 kDa) andthyroglobulin (669 kDa) from Sigma were used to calibrate thecolumn.

ASEC showed slow time-dependent dimerization on a timescale of days in the presence of AMP-PNP (results not shown),in agreement with the results from AUC. Dimerization was notobserved in the presence of novobiocin, consistent with resultsfor the E. coli GyrB ATPase domain [27].

RESULTS

ATPase activity of M. tuberculosis DNA gyrase

ATPase activities of full-length M. tuberculosis GyrB andthe ATPase domain (MtbGyrB47, residues 1–427) measuredindependently in two laboratories with different assays (see theMaterials and methods section) gave similar results.

To confirm that purified MtbGyrB47 was active, we analysedits ATPase activity using a sensitive fluorescence assay whichmeasures the production of Pi [28]. As the ATPase activity ofthe isolated ATPase domain is quite low, 15 μM protein wasused. Figure 1 shows an ATP titration for the ATPase domain.An ATP-dependent increase in activity was observed, and activitywas Mg2 + -dependent and inhibited by novobiocin (results notshown). The kcat was 0.002 s− 1. Full-length GyrB is also active,with a kcat of 0.025 s− 1 at 15-fold less protein (results not shown),suggesting that the C-terminal end of the protein (missing fromthe isolated ATPase domain) enhances ATPase activity, possiblyby making M. tuberculosis GyrB dimeric. This higher activityof the full-length protein compared with the ATPase domain hasalso been observed in Saccharomyces cerevisiae and Plasmodiumfalciparum Topo II [40,41].

The ATPase activity of M. tuberculosis DNA gyrase(GyrA and GyrB subunits at an equimolar concentration)was investigated using a PK/LDH (pyruvate kinase/lactatedehydrogenase)-linked enzyme assay as a function of enzymeconcentration at a fixed substrate concentration. The ATPaseactivity of the M. tuberculosis DNA gyrase shows a lineardependence of the rate of hydrolysis depending on the enzymeconcentration (Supplementary Figure S4 at http://www.biochemj.org/bj/456/bj4560263add.htm). Typically, 5 μM of the M.tuberculosis gyrase was found to have an activity of 30 nM/s,and activity was enhanced at least 1.5-fold (from 1.5 to 2.5) inthe presence of DNA (Supplementary Figure S4). The ATPaseactivity of the M. tuberculosis gyrase demonstrated a hyperbolicdependence on substrate concentration with a Km (app) of0.77 mM and a kcat (app), or enzyme turnover number, of0.02 s− 1 (Supplementary Figure S4). In comparison with thevalues observed for the E. coli DNA gyrase (Km = 0.83 mM andkcat = 1.2 s− 1) [10], the Km (app) is similar, but the kcat (app) wasdecreased by 60-fold, resulting in a decreased catalytic efficacyfor the M. tuberculosis gyrase. The much lower ATPase activity

of M. tuberculosis gyrase compared with E. coli gyrase has alsorecently been reported elsewhere [5].

The structures of the M. tuberculosis ATPase domain in complexwith AMP-PCP and AMP-PNP

Structures of the M. tuberculosis ATPase domain with AMP-PCPand AMP-PNP were determined by molecular replacement andrefined as described in the Materials and methods section (seeTable 1 for details). The structure of MtbGyrB47 with AMP-PCPwas, surprisingly, not dimeric (Figures 2A–2C). There was cleardensity for the AMP-PCP in all six subunits in the asymmetric unit(Figure 2C and Supplementary Figure S5 at http://www.biochemj.org/bj/456/bj4560263add.htm), but the ATP lid region (residues104–125) was largely disordered, and the GHKL and transducerdomains were differently oriented than in the AMP-PNP structure(see the next section for more detail).

The structure of MtbGyrB47 with AMP-PNP was dimeric(Figures 2D and 2E), had an ordered ATP lid region and thedimer was stabilized by the N-terminal arm crossing from onesubunit to the other. The overall dimeric structure of MtbGyrB47with AMP-PNP is similar to previously reported AMP-PNPcomplexes of ATPase domains (Supplementary Figure S6 at http://www.biochemj.org/bj/456/bj4560263add.htm) from E. coli GyrB[7,8], E. coli Topo IV [15], human Topo II [12] and other type IItopoisomerases. Two main-chain NHs were within hydrogen-bonding distance of the bridging nitrogen in the AMP-PNPstructure (Figure 2F) showing it to be in the unprotonated (P-N = P) imino form [26]. In the MtbGyrB47–AMP-PNP structuresthe ordered ATP lid region (residues 104–124) wraps around theγ -phosphate and also makes extensive interactions with the N-terminal arm from the other subunit of the dimer (Figures 2D–2F). Whereas most of the contacts to the AMP-PNP come fromthe GHKL domain (residues 21–255), Lys372 from the transducerdomain (Lys337 in E. coli) contacts the γ -phosphate of ATP, asobserved in related structures. In the AMP-PCP structure thisswitch lysine residue does not contact the γ -phosphate.

Compared with previously determined crystal structuresof bacterial type II topoisomerase ATPase domains, the M.tuberculosis GyrB ATPase domain possesses an insert of 32amino acids between the last two β-strands of the GHKLdomain (Supplementary Figure S7 at http://www.biochemj.org/bj/456/bj4560263add.htm). In our MtbGyrB47 structures thisinsert (residues 214–245) is largely disordered (Figures 2A and2D). PSI-Blast searches showed that this insert (residues 214–245) is found only in bacterial GyrB ATPase domains of theGram-positive Corynebacterineae, which include Mycobacteria,Nocardia, Rhodococcus, Gordonia and Corynebacteria [42].In the MtbGyrB47–AMP-PCP structure there were six subunits inthe asymmetric unit, all structurally similar, arranged in two verysimilar trimers. The insert region (residues 214–245) was closeto the three-fold axis of the trimers (Supplementary Figure S8at http://www.biochemj.org/bj/456/bj4560263add.htm); however,there was not clear electron density for the insert in the 3.3 Å maps,so the insert was not modelled. AUC experiments of MtbGyrB47with AMP-PCP showed monomers in solution (SupplementaryFigure S9 at http://www.biochemj.org/bj/456/bj4560263add.htm), so the trimers seen in the crystal structure are probably notbiologically relevant. The two crystal forms solved with AMP-PNP had either four dimers or eight dimers in the asymmetricunit and, in one subunit of each dimer, a short β-strand at theend of the insert region (residues 243–246) mediated a commoncrystal contact (Supplementary Figure S8C). In one of the 24AMP-PNP subunits a longer ordered region was stabilized by














Figure 2 Structures of the M. tuberculosis ATPase domain with AMP-PCP and AMP-PNP

(A and B) Two orthogonal views of the monomeric structure of MtbGyrB47 with AMP-PCP. AMP-PCP, red sticks; Mg2 + blue sphere. The ATP lid (residues 104–125, green) is largely disordered. (C)F o − F c (3 σ ) omit map (mesh) for the AMP-PCP (carbon, magenta; nitrogen, blue; oxygen, red; and phosphate, orange) and Mg2 + (small blue sphere). The protein is shown with yellow carbons,except for residues in the ATP lid which have green carbons (residues 104 and 122–125 are included in the model, residues 105–121 are disordered). (D and E) Two orthogonal views of the dimericstructure of MtbGyrB47 with AMP-PNP. AMP-PNP, red sticks; Mg2 + , blue sphere. The ATP lid (residues 104–125, green), which is shown as a solid main-chain trace and semi-transparent spheres(green) for all atoms, buries the AMP-PNP and interacts with the N-terminal arm (black) from the other subunit of the dimer. (F) F o − F c (3 σ ) omit map (mesh) for AMP-PNP and Mg2 + [colouredas in (C), except that protein atoms are with brown carbon atoms when not in the ATP lid]. The black broken lines indicate that the main-chain NH groups of Leu120 and Gly122 are 3.09 and 3.12 Afrom the bridging nitrogen. (G) Superimposition of the three conformations: RelaxT, observed in the AMP-PCP structure (yellow); ATS, ATP-restrained (AMP-PNP) conformation (orange); RelaxED,relaxed conformation (brown). For clarity, the AMP-PCP-bound structure was also used for the RelaxED structure, superimposing its GHKL and transducer domains on corresponding domains ofthe T. thermophilus gyrase (RelaxED) structure. The red square and zoom-in view highlight different orientations of the C-terminal helix. (H) Comparison of the binding modes for AMP-PCP andAMP-PNP. Note that the CH2 in AMP-PCP cannot accept hydrogen bonds from the NH groups of Leu120 and Gly122.

a unique crystal contact. A sequence search of structures inthe PDB revealed that the N-terminal domain of HSP90 fromP. falciparum also possesses a similar, but longer, insert (50%sequence similarity over the 32 residues in common), also locatedinbetween the last two strands of a GHKL domain (PDB code3PEH) (Supplementary Figure S8B). However, no function wassuggested for this insert in P. falciparum HSP90 [43], and althoughour structural studies have not identified a clear functional role forthe M. tuberculosis GyrB insert, crystal packing contacts suggestit may play a role in protein–protein interactions.

Comparison of the M. tuberculosis ATPase domain with AMP-PCPand AMP-PNP

The MtbGyrB47 structure with AMP-PCP is the first non-dimeric structure of a DNA gyrase ATPase domain in complexwith an ATP analogue. Comparing the AMP-PCP subunitstructure to that with AMP-PNP (Figure 2G), the individualGHKL and transducer subdomains are similar (Cα RMSD of0.3 Å over 152 atoms for GHKL, and Cα RMSD of 0.3 Åover 136 atoms for transducer), but the relative orientation of



Figure 3 AUC of the M. tuberculosis GyrB ATPase domain

AUC traces are shown of untagged MtbGyrB47C2 at 17 μM (bottom) and MtbGyrBC2 in thepresence of 2 mM MgCl2 and 2 mM AMP-PNP (top). Time points shown are for the freshsamples (��), after incubation at room temperature for 3 days (� � �) and6 days (��). MWapp, apparent molecullar mass.

the GHKL and transducer subdomains are different, with aswivel of approximately 17◦, moving the distal end of the C-terminal-most helix approximately 15 Å outward (Figure 2G).This movement is different from the one observed for therelaxed conformation of type II topoisomerase ATPase domains(Figure 2G). The MtbGyrB47–AMP-PCP structure seems to be ina new conformational state, which we term RelaxT (Figure 2G),intermediate between the previously observed RelaxED (or open)and ATP-restrained (ATS or closed) conformations [13,18]. Inthe AMP-PCP structure the ATP lid region (residues 104–125)is disordered, as it is in many structures with inhibitors [44] andin some structures with ADP [13]. This suggested that becausethe CH2 between the β- and γ -phosphates of AMP-PCP cannotaccept hydrogen bonds from main-chain NHs of Leu120 and Gly122

(Figures 2C, 2F and 2H), the ATP lid cannot become ordered andclose, and therefore the dimer cannot form.

AUC showed that incubation of the M. tuberculosis GyrBATPase domain with AMP-PNP initially gave monomers (asobserved with AMP-PCP and apo protein), but that dimers formedslowly (compared with the 6-10 h run time of an AUC experiment)over a period of several days (Figure 3). This suggested thatAMP-PNP was initially largely in the imido (P-NH-P) form, andthat the presence of the hydrogen on the nitrogen blocked theATP lid from becoming ordered and closing (as with AMP-PCP).However, the AMP-PNP could convert into the imino (P-N = P)form, and when in the imino (P-N = P) form the ATP lid couldclose and the dimer form; the percentage of dimer increased withtime (Figure 3). Crystals of MtbGyrB47 with AMP-PNP weregrown at pH 8.5, in the presence of between 5 and 200 mM MgCl2

[21], conditions expected to stabilize the imino (P-N = P) form.This slow interaction of AMP-PNP has also been observed withE. coli DNA gyrase [45].

Several Topo II crystal structures contain the imino (P-N = P) formof AMP-PNP

The ATP-binding pocket of type II topoisomerase ATPasedomains is highly conserved and all of the important residues

for binding and catalysis are observed in M. tuberculosis gyrase.In the AMP-PNP structures the ATP lid is ordered. Residues at theC-terminus of the ATP lid, GLHGVG (residues 119–124), forma glycine-rich P-loop (phosphate-binding loop), that has main-chain NH groups from His121, Val123 and Gly124, making hydrogenbonds to oxygens on the γ -phosphate of AMP-PNP. Gly122 has itsmain-chain NH pointing directly at the bridging nitrogen betweenthe β- and γ -phosphates, whereas the main-chain nitrogenof Leu120 is within 3.0 Å of both the bridging nitrogen and one ofthe oxygens on the γ -phosphate (Figure 4A). The binding modeobserved for AMP-PNP in MtbGyrB47 is essentially the sameas seen in the 1.8 Å yeast Topo II complex (Figure 4B) and the1.87 Å human complex (Supplementary Figure S3); we concludethat these structures also bind the imino form of AMP-PNP. Amore distantly related Topo VIB structure with AMP-PNP [16] isalso in the imino (P-N = P) form [26].

Two E. coli Topo II structures appear to contain the imido (P-NH-P)form of AMP-PNP

The position of the nitrogen between the β- and γ -phosphates ina 2.3 Å E. coli gyrase and a 2.1 Å E. coli Topo IV structurewith AMP-PNP are quite similar to each other (Figures 4Cand 4D), but this nitrogen is in a different position from otherstructures (Figure 4). Moreover, in these two structures there isnot a Mg2 + adjacent to the β- or γ -phosphates (although the2.1 Å E. coli Topo IV structure has a nearby Mg2 + ; Figure 4D).The different position of the bridging nitrogen (and the absenceof the adjacent Mg2 + ion) in the E. coli gyrase and Topo IVstructures (Figures 4C–4E) suggests that the nitrogen is in theimido form (Supplementary Figure S2). In the 2.1 Å E. coli TopoIV structure (Figure 4D), the closest main-chain NH from theATP lid is some 3.9 Å away; the bridging nitrogen is too far awayto accept hydrogen bonds from the ATP lid. However, the ATP lidis ordered and in the same conformation as observed in the otherAMP-PNP structures, moreover these two E. coli structures aredimeric (Supplementary Figure S6). This observation was initiallypuzzling because the MtbGyrB47 AMP-PCP structure suggestedthat to initially become ordered and close, the ATP lid needed tomake contact with the atom between the β- and γ -phosphates.Only once the ATP lid has closed can the dimer form.

One possibility is that dimers of the two E. coli proteins formedwhile the AMP-PNP was in the imino form (P-N = P), but atsome point during crystal formation the NH group of the adjacentglycine residue protonated the P-N = P group to be P-NH-P.In the scheme shown in Figure 5 (Figures 5A–5C), movementof the P-loop and the AMP-PNP would then allow a hydrogenfrom the γ -phosphate to reprotonate the glycine residue. Theextensive interactions with the N-terminal arm of the other subunit(Figures 2D and 2E) would keep the ATP lid ordered after theAMP-PNP had adopted the imido form.

DISCUSSION

A fully dissociative mechanism for ATP hydrolysis

The proposed scheme for the catalytic conversion of the imino (P-N = P) into the imido (P-NH-P) form of AMP-PNP (Figures 5A–5C) suggests a simple dissociative mechanism for ATP hydrolysisby M. tuberculosis GyrB and related ATPases (Figures 5G–5I).In this proposed mechanism an initial movement of Gly122 causesits main-chain NH to protonate the bridging oxygen of the ATP.Stabilization of the resulting negative charge on the protein isproposed to be, in M. tuberculosis gyrase, by transfer to Tyr33



Figure 4 In structures of AMP-PNP with type IIA topoisomerases, the nitrogen between the β- and γ -phosphates is found in two different positions

(A) The 2.9 A structure of M. tuberculosis GyrB (the present study). (B) The 1.8 A structure of S. cerevisiae (PDB code 1PVG). (C) The 2.3 A structure of E. coli GyrB (PDB code 1EI1). (D) The 2.1 Astructure of E. coli Topo IV (PDB code 1S16). (E) Structures in (A–D) are superimposed. The two orientations for the bridging nitrogen are consistent with imino (P-N = P) or imido (P-NH-P) formsof AMP-PNP.

via His104 (Figure 5D). In human Topo II a similar charge-relaynetwork (Figures 5E and 5F) involves His42 (equivalent to Ala25

in M. tuberculosis), although direct transfer of the charge to Tyr151

might also be possible. The protonation of ATP gives rise to anADP2 − and a free metaphosphate ion (Figure 5H). The activatedwater immediately attacks the highly reactive metaphosphate ion(Figure 5H) to give a phosphate ion. The ADP2 − (SupplementaryFigure S2) then reprotonates the main-chain nitrogen of Gly122

(Figure 5I), becoming an ADP3 − ion in the process. If the activatedwater is not correctly positioned to attack the metaphosphate ion,once the ADP3 − ion is formed it would be correctly positioned tore-attack the metaphosphate ion to reform ATP [20].

A scheme for the reverse reaction, the synthesis of ATP bytype II topoisomerases [25], is shown in Supplementary FigureS10 (at http://www.biochemj.org/bj/456/bj4560263add.htm). InGHKL domain histidine kinases, the activated water is not present,and the phosphate group is transferred from ATP to a histidine sidechain [46]. In two-component signal transduction the phosphategroup is then transferred from the histidine side chain to anaspartic acid side chain (Supplementary Figure S10).

The proposed dissociative mechanism seems chemicallyreasonable for GHKL domain histidine kinases. The previouslyproposed mechanisms for ATP hydrolysis [13,19] by type IItopoisomerases show the reaction being initiated when theactivated water (or OH− ion) makes a nucleophilic attack on thehighly negatively charged γ -phosphate (Supplementary FigureS1). This type of mechanism, initiated by a nucleophilic attack ona negatively charged tetrahedral γ -phosphate, has been proposedfor many other ATPases and GTPases.

Could other ATPase/GTPase families also have a fully dissociatedmechanism?

The glycine-rich P-loop in M. tuberculosis GyrB and relatedGHKL phosphotransferases does not correspond to the WalkerA motif (GxxxxGK[T/S]) found in many ATP- and GTP-bindingproteins [47]. However, because of structural similarities in theway the atom corresponding to the bridging oxygen between theβ- and γ -phosphates is co-ordinated in transition state complexesof Ras and F1-ATPase and M. tuberculosis GyrB–AMP-PNP (Supplementary Figure S11 at http://www.biochemj.org/bj/456/bj4560263add.htm), it is tempting to speculate that WalkerA motif-containing proteins may have, by convergent evolution,arrived at a similar mechanism for ATP/GTP hydrolysis. Main-

chain NHs from glycine residues point at the bridging oxygenbetween the β- and γ -phosphates (Supplementary Figure S11) intransition state analogue complexes of the small G-protein Ras[48] (bold Gly13 in GaggvGKS) and in F1-ATPase [49] (boldGly159 in GgagvGKT). We propose that movement of this glycineresidue past the bridging oxygen between the β- and γ -phosphatesforces it to leave its main-chain NH proton behind on the bridgingoxygen promoting GTP or ATP hydrolysis. The bridging oxygenbetween the β- and γ -phosphates of GTP is negatively charged[50]. A fully dissociative mechanism for GTP hydrolysis byRas (Supplementary Figure S12 at http://www.biochemj.org/bj/456/bj4560263add.htm) can explain why point mutations at Gly12,Gly13 or Gln61 impair intrinsic rates of GTP hydrolysis byRas, while increasing intrinsic rates of hydrolysis of some GTPanalogues [51].

How is ATP hydrolysis coupled to domain motion in M. tuberculosisDNA gyrase?

Analysis of previously available type II topoismerase ATPasestructures (GHKL + transducer domains) supported the existenceof ‘ATP-restrained’ and ‘relaxED’ conformational states [13,18].In the M. tuberculosis GyrB complex with AMP-PCP, a secondrelaxed conformation (RelaxT) was observed, with the transducerand GHKL domains in different relative positions; we term thisconformation ‘RelaxT’ and suggest it may correspond to themonomeric ATP-bound form. In ‘ATP-restrained’ conformationsthe size of the ‘hole’ between the two transducer domains is oftentoo small to accommodate the T-segment DNA [12,14].

In the present study we compare schemas showing how thecatalytic cycle of DNA gyrase could be governed by (i) ATPhydrolysis [24,25] or (ii) protonation of imino AMP-P-N = P toimido AMP-P-NH-P. The catalytic cycle in the presence of ATPis described first (Figure 6A). A closed ATP lid is indicated bya green square, an open ATP lid by a green line. When the ATPlid is open, nucleotide exchange can take place. (i) At the start ofthe catalytic cycle both GHKL domains are shown occupied byATP (represented by the letter T), but the two GHKL domains aretoo far apart to dimerize. (ii) When the T-DNA comes betweenthe transducer domains, they are attracted inwards towards it[6,52]. Once the two GHKL domains are close enough togethera closed ATP lid in one subunit will bind the terminal N-armfrom the other subunit, closing the ATP-gate dimer interface. (iii)In forming this dimer the first ATPase domain adopts the ATS








Figure 5 Main-chain NH protonation of AMP-P-N = P to AMP-P-NH-P suggests a fully dissociative mechanism for ATP hydrolysis

(A–C) A scheme for enzymatic conversion of imino (P-N = P) into imido (P-NH-P) forms of AMP-PNP. (D) The MtbGyrB47–AMP-PNP structure showing residues that could stabilize the negativecharge after Gly122 protonates ATP. (E) Human Topo II structure (PDB code 1ZXM) with AMP-PNP (equivalent view). (F) Superimposition of structures in (D) and (E). (G–I) A dissociative mechanismfor ATP hydrolysis by M. tuberculosis DNA gyrase.

conformation (orange) and the ATP lid hydrolyses the first ATP.This first GHKL domain now contains ADP and Pi. (iv) When theATP lid in the second GHKL domain (which still contains ATP)becomes ordered, the domain tries to adopt the ATP-restrainedconformation, but it cannot adopt this ATS conformation untilthe T-DNA segment has been squeezed out from between thetwo transducer domains. During this conformational change thePi is released from the first ATPase domain, allowing the firstATPase domain, now containing ADP (D) to move to a relaxedconformation. (v) The second ATPase domain can adopt the ATP-restrained conformation (orange) once the T-segment has passedthrough the cleaved DNA, helping to close the central G-gate.(vi) After the T-DNA segment has passed through the exit gate,

closure of the exit gate may signal for the second ATP to behydrolysed via interactions between DNA-gate domains and thetransducer domain [53]. In the presence of two ADPs (D), theATP lid is no longer restrained so tightly and the ATP-gate canre-open, returning the enzyme to step (i) to rerun the catalyticcycle.

The proposed schema when the catalytic cycle is regulated byAMP-PNP (Figure 6B) is very similar to that for ATP. Note thatthe ATP lid can only initially close when the AMP-PNP is in theimino (P-N = P) form (shown by letter N in Figure 6B), but oncethe dimer is formed and the ATP lid is held in place by extrainteractions with the N-arm of the other subunit, the ATP lid willremain closed with the imido (P-NH-P) form (shown by letters NH



Figure 6 Schemes illustrating coupling of ATP hydrolysis (or AMP-PNP protonation) with domain motion in M. tuberculosis gyrase

(A) ATP hydrolysis. T, ATP; D, ADP. (B) AMP-PNP protonation. NH, AMP-P-NH-P; N, AMP-P-N = P. Different conformations of the ATPase domains are indicated by different colours: yellow, RelaxTconformation; orange, ATS conformation; brown, RelaxED conformation. A green central region in a GHKL domain indicates the ATP lid is closed. An open (disordered) ATP lid is indicated by a greenline. Disordered N-terminal arms (broken blue or black lines) become ordered (continuous blue or black lines) when they interact with an ordered ATP lid across the dimer interface. Gate DNA isshown in black and T-DNA is in purple. (The C-terminal DNA-wrapping domain of GyrA is omitted for clarity). (C) Models of the ATPase domain of M. tuberculosis showing how the space betweenthe C-terminal helices increases between states (iii) and (iv) (as in A), allowing passage of the T (transport) segment DNA (purple). The RelaxED conformation (brown) was modelled on PDB code1KIJ by superimposing the GHKL and transducer domains.

in Figure 6B). The protonation of the imino to the imido formsis hypothesized to be accompanied by the expulsion of Mg2 +

normally co-ordinated by oxygens from β- and γ -phosphates. Itis suggested that the different conformation of the γ -phosphateand the absence of Mg2 + allows the AMP-P-NH-P-bound ATPasedomain to adopt relaxed conformations.

Conclusions

The crystal structure of the M. tuberculosis DNA gyrase ATPasedomain with AMP-PCP was in a novel monomeric form, whereascrystal structures with AMP-PNP were dimeric and similar tomany related structures in the literature. By including hydrogensin the refinement and comparing our structures with others fromthe literature we were able to come up with an explanation for theability of AMP-PNP to catalyse one round of the reaction cycle. Inthis explanation the enzyme protonates the imino (P-N = P) formof AMP-PNP to give the more common imido form (P-NH-P).This insight allowed a novel mechanism for ATP hydrolysis to beproposed. In this fully dissociative mechanism for ATP hydrolysisa main-chain NH moves to protonate ATP, causing it to dissociateinto ADP and Pi. To the best of our knowledge this is the firsttime a main-chain amide has been proposed to play a direct rolein catalysis. Moreover, this mechanochemical mechanism mayhave a broad applicability to proteins such as F1-ATPase and G-proteins such as Ras, as well as proteins involved more directlyin movement. As well as providing insights into the coupling of

ATP hydrolysis and domain movement in M. tuberculosis ATPasegyrase, these studies open up new possibilities for the rationaldesign of covalent inhibitors against DNA gyrase, HSP90 andpossibly other challenging, but important, drug targets.

AUTHOR CONTRIBUTION

In GlaxoSmithKline, Alka Agrawal carried out cloning, protein expression, purification,crystallization experiments, refinement and activity assays. Claus Spitzfaden performedAUC and ASEC. Benjamin Bax solved structures, completed refinements and deviseddissociative mechanism. In Paris, Claudine Mayer conceived and supervised the work.Melanie Roue carried out cloning, protein expression, purification, activity assays,crystallization experiments, AUC experiments, crystallographic data collection, andrefinement with the help of Claudine Mayer. Alka Agrawal, Melanie Roue, Benjamin Baxand Claudine Mayer wrote the paper with assistance from Stephanie Petrella, AlexandraAubry and Michael Hann.

FUNDING

The work carried out in Paris was supported by the Pasteur Institute [ProgrammesTransversaux de Recherche number 367]. A. Agrawal was funded by a National ScienceFoundation International Research Fellowship [grant number INT-0202606].

ACKNOWLEDGEMENTS

For experimental assistance we thank members of the Crystallisation and Crystallography,and the Biophysics of Macromolecules and their Interactions Platforms at the PasteurInstitute and members of Biological Reagents and Assay Development department withinGlaxoSmithKline. We also thank Helene Munier-Lehman for help with the ATPase assays,



Robert Nolte and Nigel Moriarty for computational help, and Oliver Smart, Jeremie Pitonand Olivier Poch for helpful discussions.

REFERENCES

1 Schoeffler, A. J. and Berger, J. M. (2008) DNA topoisomerases: harnessing andconstraining energy to govern chromosome topology. Q. Rev. Biophys. 41, 41–101

2 Wang, J. C. (2009) A journey in the world of DNA rings and beyond. Annu. Rev. Biochem.78, 31–54

3 Aubry, A., Fisher, L. M., Jarlier, V. and Cambau, E. (2006) First functional characterizationof a singly expressed bacterial type II topoisomerase: the enzyme from Mycobacteriumtuberculosis. Biochem. Biophys. Res. Commun. 348, 158–165

4 Piton, J., Petrella, S., Delarue, M., Andre-Leroux, G., Jarlier, V., Aubry, A. and Mayer, C.(2010) Structural insights into the quinolone resistance mechanism of Mycobacteriumtuberculosis DNA gyrase. PLoS ONE 5, e12245

5 Tretter, E. M. and Berger, J. M. (2012) Mechanisms for defining supercoiling set point ofDNA gyrase orthologs: II. the shape of the GyrA subunit C-terminal domain (CTD) is not asole determinant for controlling supercoiling efficiency. J. Biol. Chem. 287,18645–18654

6 Gubaev, A. and Klostermeier, D. (2011) DNA-induced narrowing of the gyrase N-gatecoordinates T-segment capture and strand passage. Proc. Natl. Acad. Sci. U.S.A. 108,14085–14090

7 Wigley, D. B., Davies, G. J., Dodson, E. J., Maxwell, A. and Dodson, G. (1991) Crystalstructure of an N-terminal fragment of the DNA gyrase B protein. Nature 351, 624–629

8 Brino, L., Urzhumtsev, A., Mousli, M., Bronner, C., Mitschler, A., Oudet, P. and Moras, D.(2000) Dimerization of Escherichia coli DNA-gyrase B provides a structural mechanismfor activating the ATPase catalytic center. J. Biol. Chem. 275, 9468–9475

9 Maxwell, A. and Gellert, M. (1984) The DNA dependence of the ATPase activity of DNAgyrase. J. Biol. Chem. 259, 14472–14480

10 Gross, C. H., Parsons, J. D., Grossman, T. H., Charifson, P. S., Bellon, S., Jernee, J.,Dwyer, M., Chambers, S. P., Markland, W., Botfield, M. and Raybuck, S. A. (2003)Active-site residues of Escherichia coli DNA gyrase required in coupling ATP hydrolysisto DNA supercoiling and amino acid substitutions leading to novobiocin resistance.Antimicrob. Agents Chemother. 47, 1037–1046

11 Dutta, R. and Inouye, M. (2000) GHKL, an emergent ATPase/kinase superfamily. TrendsBiochem. Sci. 25, 24–28

12 Wei, H., Ruthenburg, A. J., Bechis, S. K. and Verdine, G. L. (2005) Nucleotide-dependentdomain movement in the ATPase domain of a human type IIA DNA topoisomerase. J. Biol.Chem. 280, 37041–37047

13 Corbett, K. D. and Berger, J. M. (2005) Structural dissection of ATP turnover in theprototypical GHL ATPase TopoVI. Structure 13, 873–882

14 Classen, S., Olland, S. and Berger, J. M. (2003) Structure of the topoisomerase II ATPaseregion and its mechanism of inhibition by the chemotherapeutic agent ICRF-187. Proc.Natl. Acad. Sci. U.S.A. 100, 10629–10634

15 Bellon, S., Parsons, J. D., Wei, Y., Hayakawa, K., Swenson, L. L., Charifson, P. S., Lippke,J. A., Aldape, R. and Gross, C. H. (2004) Crystal structures of Escherichia colitopoisomerase IV ParE subunit (24 and 43 kilodaltons): a single residue dictatesdifferences in novobiocin potency against topoisomerase IV and DNA gyrase. Antimicrob.Agents Chemother. 48, 1856–1864

16 Corbett, K. D. and Berger, J. M. (2003) Structure of the topoisomerase VI-B subunit:implications for type II topoisomerase mechanism and evolution. EMBO J. 22, 151–163

17 Reference deleted18 Lamour, V., Hoermann, L., Jeltsch, J. M., Oudet, P. and Moras, D. (2002) An open

conformation of the Thermus thermophilus gyrase B ATP-binding domain. J. Biol. Chem.277, 18947–18953

19 Jackson, A. P. and Maxwell, A. (1993) Identifying the catalytic residue of the ATPasereaction of DNA gyrase. Proc. Natl. Acad. Sci. U.S.A. 90, 11232–11236

20 Frey, P. A. and Hegeman, A. D. (2007) Enzymatic Reaction Mechanisms. OxfordUniversity Press, New York

21 Roue, M., Agrawal, A., Volker, C., Mossakowska, D., Mayer, C. and Bax, B. (2013)Purification, crystallization and preliminary X-ray crystallographic studies of theMycobacterium tuberculosis DNA gyrase ATPase Domain. Acta Crystallogr., Sect. F:Struct. Biol. Crystal. Commun. 68, 178–180

22 Osheroff, N., Shelton, E. R. and Brutlag, D. L. (1983) DNA topoisomerase II fromDrosophila melanogaster. Relaxation of supercoiled DNA. J. Biol. Chem. 258,9536–9543

23 Sugino, A. and Cozzarelli, N. R. (1980) The intrinsic ATPase of DNA gyrase. J. Biol.Chem. 255, 6299–6306

24 Baird, C. L., Harkins, T. T., Morris, S. K. and Lindsley, J. E. (1999) Topoisomerase II drivesDNA transport by hydrolyzing one ATP. Proc. Natl. Acad. Sci. U.S.A. 96, 13685–13690

25 Baird, C. L., Gordon, M. S., Andrenyak, D. M., Marecek, J. F. and Lindsley, J. E. (2001)The ATPase reaction cycle of yeast DNA topoisomerase II. Slow rates of ATP resynthesisand Pi release. J. Biol. Chem. 276, 27893–27898

26 Dauter, M. and Dauter, Z. (2011) Deprotonated imidodiphosphate in AMPPNP-containingprotein structures. Acta Crystallogr., Sect. D: Biol. Crystallogr. 67, 1073–1075

27 Ali, J. A., Jackson, A. P., Howells, A. J. and Maxwell, A. (1993) The 43-kilodaltonN-terminal fragment of the DNA gyrase B protein hydrolyzes ATP and binds coumarindrugs. Biochemistry 32, 2717–2724

28 Vazquez, M. J., Rodriguez, B., Zapatero, C. and Tew, D. G. (2003) Determination ofphosphate in nanomolar range by an enzyme-coupling fluorescent method. Anal.Biochem. 320, 292–298

29 McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. andRead, R. J. (2007) Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674

30 Murshudov, G. N., Skubak, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A.,Winn, M. D., Long, F. and Vagin, A. A. (2011) REFMAC5 for the refinement ofmacromolecular crystal structures. Acta Crystallogr., Sect. D: Biol. Crystallogr. 67,355–367

31 Adams, P. D., Afonine, P. V., Bunkoczi, G., Chen, V. B., Davis, I. W., Echols, N., Headd,J. J., Hung, L. W., Kapral, G. J., Grosse-Kunstleve, R. W. et al. (2010) PHENIX: acomprehensive Python-based system for macromolecular structure solution. ActaCrystallogr., Sect. D: Biol. Crystallogr. 66, 213–221

32 Smart, O. S., Womack, T. O., Flensburg, C., Keller, P., Paciorek, W., Sharff, A., Vonrhein,C. and Bricogne, G. (2012) Exploiting structure similarity in refinement: automated NCSand target-structure restraints in BUSTER. Acta Crystallogr., Sect. D: Biol. Crystallogr. 68,368–380

33 Moriarty, N. W., Grosse-Kunstleve, R. W. and Adams, P. D. (2009) electronic LigandBuilder and Optimization Workbench (eLBOW): a tool for ligand coordinate and restraintgeneration. Acta Crystallogr., Sect. D: Biol. Crystallogr. 65, 1074–1080

34 Terwilliger, T. C., Klei, H., Adams, P. D., Moriarty, N. W. and Cohn, J. D. (2006) Automatedligand fitting by core-fragment fitting and extension into density. Acta Crystallogr., Sect.D: Biol. Crystallogr. 62, 915–922

35 Allen, F. H. (2002) The Cambridge Structural Database: a quarter of a million crystalstructures and rising. Acta Crystallogr., Sect. B: Struct. Sci. 58, 380–388

36 Bruno, I. J., Cole, J. C., Kessler, M., Luo, J., Motherwell, W. D., Purkis, L. H., Smith, B.R., Taylor, R., Cooper, R. I., Harris, S. E. and Orpen, A. G. (2004) Retrieval ofcrystallographically-derived molecular geometry information. J. Chem. Inf. Comput. Sci.44, 2133–2144

37 Emsley, P., Lohkamp, B., Scott, W. G. and Cowtan, K. (2010) Features and development ofCoot. Acta Crystallogr., Sect. D: Biol. Crystallogr. 66, 486–501

38 Reference deleted39 Brown, P. H. and Schuck, P. (2006) Macromolecular size-and-shape distributions by

sedimentation velocity analytical ultracentrifugation. Biophys. J. 90, 4651–466140 Olland, S. and Wang, J. C. (1999) Catalysis of ATP hydrolysis by two NH2-terminal

fragments of yeast DNA topoisomerase II. J. Biol. Chem. 274, 21688–2169441 Dar, M. A., Sharma, A., Mondal, N. and Dhar, S. K. (2007) Molecular cloning of

apicoplast-targeted Plasmodium falciparum DNA gyrase genes: unique intrinsic ATPaseactivity and ATP-independent dimerization of PfGyrB subunit. Eukaryotic Cell 6, 398–412

42 de Sousa-d’Auria, C., Kacem, R., Puech, V., Tropis, M., Leblon, G., Houssin, C. and Daffe,M. (2003) New insights into the biogenesis of the cell envelope of corynebacteria:identification and functional characterization of five new mycoloyltransferase genes inCorynebacterium glutamicum. FEMS Microbiol. Lett. 224, 35–44

43 Corbett, K. D. and Berger, J. M. (2010) Structure of the ATP-binding domain ofPlasmodium falciparum Hsp90. Proteins 78, 2738–2744

44 Chan, P. F., Huang, J., Bax, B. D. and Gwynn, M. N. (2013) Recent developments ininhbitors of bacterial type IIA topoisomerases. In Antibiotics: Targets, Mechanisms andResistance (Gualerzi, C. O., Brandi, L., Fabbretti, A. and Pon, C. L., eds), pp. 263–297,Wiley, Germany

45 Tamura, J. K., Bates, A. D. and Gellert, M. (1992) Slow interaction of5′-adenylyl-β ,γ -imidodiphosphate with Escherichia coli DNA gyrase. Evidence forcooperativity in nucleotide binding. J. Biol. Chem. 267, 9214–9222

46 Casino, P., Rubio, V. and Marina, A. (2010) The mechanism of signal transduction bytwo-component systems. Curr. Opin. Struct. Biol. 20, 763–771

47 Saraste, M., Sibbald, P. R. and Wittinghofer, A. (1990) The P-loop: a common motif inATP- and GTP-binding proteins. Trends Biochem. Sci. 15, 430–434

48 Scheffzek, K., Ahmadian, M. R., Kabsch, W., Wiesmuller, L., Lautwein, A., Schmitz, F. andWittinghofer, A. (1997) The Ras-RasGAP complex: structural basis for GTPase activationand its loss in oncogenic Ras mutants. Science 277, 333–338

49 Kagawa, R., Montgomery, M. G., Braig, K., Leslie, A. G. and Walker, J. E. (2004) Thestructure of bovine F1-ATPase inhibited by ADP and beryllium fluoride. EMBO J. 23,2734–2744



50 Li, G. and Zhang, X. C. (2004) GTP hydrolysis mechanism of Ras-like GTPases. J. Mol.Biol. 340, 921–932

51 Ahmadian, M. R., Zor, T., Vogt, D., Kabsch, W., Selinger, Z., Wittinghofer, A. andScheffzek, K. (1999) Guanosine triphosphatase stimulation of oncogenic Ras mutants.Proc. Natl. Acad. Sci. U.S.A. 96, 7065–7070

52 Tingey, A. P. and Maxwell, A. (1996) Probing the role of the ATP-operated clamp in thestrand-passage reaction of DNA gyrase. Nucleic Acids Res. 24, 4868–4873

53 Schmidt, B. H., Osheroff, N. and Berger, J. M. (2012) Structure of a topoisomeraseII-DNA-nucleotide complex reveals a new control mechanism for ATPase activity. Nat.Struct. Mol. Biol. 19, 1147–1154

Received 15 April 2013/4 September 2013; accepted 9 September 2013Published as BJ Immediate Publication 9 September 2013, doi:10.1042/BJ20130538


Biochem. J. (2013) 456, 263–273 (Printed in Great Britain) doi:10.1042/BJ20130538

SUPPLEMENTARY ONLINE DATAMycobacterium tuberculosis DNA gyrase ATPase domain structures suggesta dissociative mechanism that explains how ATP hydrolysis is coupled todomain motionAlka AGRAWAL*1, Melanie ROUE†‡§1, Claus SPITZFADEN*, Stephanie PETRELLA†‡§, Alexandra AUBRY‖¶**, Michael HANN*,Benjamin BAX*2 and Claudine MAYER†‡§*Platform Technology Sciences, GlaxoSmithKline, Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire SG1 2NY, U.K., †Unite de Microbiologie Structurale,Institut Pasteur, 75015 Paris, France, ‡URA 2185, CNRS, 75015 Paris, France, §Universite Paris Diderot, Sorbonne Paris Cite, Cellule Pasteur, 75015 Paris, France, ‖UPMC UniversiteParis 06, ER5, EA 1541, Laboratoire de Bacteriologie-Hygiene, Paris, France, ¶AP-HP, Hopital Pitie-Salpetriere, Laboratoire de Bacteriologie-Hygiene, Paris, France, and **CentreNational de Reference des Mycobacteries et de la Resistance des Mycobacteries aux Antituberculeux, Paris, France

Figure S1 Comparison of two possible mechanisms for ATP hydrolysis by M. tuberculosis DNA gyrase

(A) A mechanism in which the ATP hydrolysis is initiated when a water (or OH− ion) makes a nucleophilic attack on the phosphate (associative). This scheme is essentially that proposed in [1], anexcellent paper dissecting the ATP turnover of a GHL ATPase. (B) A mechanism in which the hydrolysis is initiated when the oxygen between the β- and γ -phosphates is protonated. This causesATP to dissociate into ADP and a metaphosphate ion [PO3] − .

1 These authors contributed equally to this work.2 To whom correspondence should be addressed (email [email protected]).

Co-ordinates and structure factor files for the M. tuberculosis GyrB ATPase domain have been deposited in the PDB under the accession codes 3ZKB,3ZKD and 3ZM7


A. Agrawal and others

Figure S2 Common protonation states for Pi, ADP and ATP with pK values, and protonation and tautomeric states of AMP-PCP and AMP-PNP

A is adenosine. Adenosine is uncharged between pH 4.8 and 12. The common protonation states of Pi, ADP and ATP are shown [3,4]. In the presence of bound Mg2 + ions the more negativelycharged forms can be stabilized [5]. Solution NMR studies suggest that AMP-PNP probably has the hydrogen on the bridging nitrogen [6]. Although E. coli alkaline phosphatase can remove theterminal phosphate of AMP-PNP [7], most enzymes do not efficiently hydrolyse AMP-PNP.


ATP hydrolysis by the M. tuberculosis DNA gyrase ATPase domain

Figure S3 The 1.87 A human Topo IIA structure with AMP-PNP is consistentwith a bridging imino (P-N = P), but not a bridging imido (P-NH-P)

(A) The 1.87 A human Topo IIA structure with AMP-PNP (PDB code 1ZXM). Distances frommain-chain nitrogens of Gly164 and Arg162 are indicated by broken lines (estimated error ∼0.1 Abased on two molecules in asymmetic unit). (B) The original AMP-PNP was removed, hydrogenswere added to the protein and a difference map was calculated (insert 5 σ F o − F c). The iminoP-N = P form of AMP-PNP was modelled [8] into the F o − F c map using PHENIX [9] anddisplayed in Coot [10]. Distances from hydrogens on main-chain nitrogens of Gly164 and Arg162

are indicated by broken lines. (The hydrogen on the main-chain nitrogen of Asn163 is some2.9 A from the bridging nitrogen). (C) Superimposition of (A) and (B).

Figure S4 ATP activity assays on M. tuberculosis DNA gyrase

(A) ATPase activity of the M. tuberculosis DNA gyrase as a function of protein concentration.Rates are initial velocities. The substrate (ATP) concentration was 1 mM, and the subunits GyrAand GyrB were mixed in equimolar concentrations. (B) ATPase activity of the M. tuberculosisDNA gyrase in the presence and absence of DNA (relaxed pBR322 at 5 μg/ml). Rates areinitial velocities. The substrate (ATP) concentration was 1 mM, and the subunits GyrA and GyrBwere mixed in equimolar concentrations. Increased ATPase activity in the presence of DNA haspreviously been shown for E. coli DNA gyrase, yeast, human and protozoan topoisomerasesII, and the P. falciparum DNA gyrase [11–17]. (C) ATPase activity of the M. tuberculosis DNAgyrase as a function of substrate (ATP) concentration at a constant enzyme concentration. Ratesare initial velocities. The subunits GyrA and GyrB both at 5 μM. The K m was calculated by theLineweaver–Burk plot. Initial rates were correlated with a phosphate calibration curve.



Figure S5 Fo − Fc ( + 2.5 σ ) maps showing difference density for the six AMP-PCPs in the asymmetric unit

The final structure is shown with a map phased on the original molecular replacement solution from which the ATP lid, N-arm and, in molB, residues from the transducer domain have been deleted.The map shows the density for all AMP-PCPs, but no density for the ATP lid. For molA the AMP-PNP structure is shown superimposed.



Figure S6 Comparison of structures of type IIA topoisomerases with AMP-PNP

Three orthogonal views of (A) M.tuberculosis GyrB (the present study), (B) 2.3 A E.coli GyrB (PDB code 1EI1) and (C) 2.1 A structure E. coli Topo IV (PDB code 1S16). (D) Superimposition of(A–C) (the structures are superimposed using one GHKL domain, indicated by an arrow in the centre of A). (E) The 1.87 A human Topo IIA (PDB code 1ZXM). Because the superimpositions in thisFigure were done on a single GHKL domain (indicated by an arrow in A), this Figure emphasizes differences between the structures.



Figure S7 Structure-corrected sequence alignment of the ATPase domain from type IIA topoisomerases

The sequence names are as follows: M.tuberculosis_GyrB, M. tuberculosis DNA gyrase; M.leprae_GyrB, M. leprae DNA gyrase; 1EI1_E.coli_Gyr, E. coli DNA gyrase (PDB code 1EI1);1S16_E.coli_TopoIV , E. coli Topo IV (PDB code 1S16); 1PVG_S.cerev_TopoII, S. cerevisiae Topo II (PDB code 1PVG); and 1ZXM_Human_TopoII, human Topo II (PDB code 1ZXM). α-Helices andβ-strands are shown above the sequences as red and blue cylinders respectively, and are defined from the E. coli DNA gyrase ATPase domain (PDB code 1EI1) from PDBSum. The ATP lid is indicatedby a green arrow. The vertical green bar delimits the end of the GHKL and the beginning of the transducer sub-domains. The insert regions (32 or 34 residues) of M. tuberculosis and M. leprae areindicated in brown. The two constructs used in the present paper both had the M. tuberculosis GyrB sequence shown in alignment, but they had different N-terminal tags: MtbGyrB47C1 had a 19residue tag MGHHHHHHLEVLFQ/GPLGS and MtbGyrB47C2 had a 15 residue N-terminal tag with MAHHHHHHVDDDDK/ with the cleavage site just prior to Val1 (where / represents a cleavage site).



Figure S8 The insert region of the M. tuberculosis ATPase domain is involved in crystal contacts

(A) View of the Mtb GyrB47–AMP-PNP dimer showing positions of the largely disordered insert regions (framed in red). (B) Superimposition of the GHKL domain of P. falciparum HSP90 (PDBcode 3PEH) showing an insert at the same position. (C) Dimer–dimer interactions observed in all MtbGyrB47–AMP-PNP dimers [shown between the AB and EF chains in the P1(8) form]. Residuesfrom the insert are in brown. (D) The insert region is also close to crystal contacts in the MtbGyrB47–AMP-PCP structure. The first residue of β10 and last residue of β11 are represented in brownCPK (Corey–Pauling–Koltun).

Figure S9 AUC of the M. tuberculosis GyrB ATPase domain

AUC traces are shown of untagged MtbGyrB47C2 at 17 μM (bottom) and MtbGyrBC2 in thepresence of 2 mM MgCl2 and 2 mM AMP-PCP (top).



Figure S10 Dissociative reaction schemes for the (A) synthesis of ATP by the GyrB ATPase domain and (B) for the transfer of a phosphate from a histidineto an aspartic acid residue in two-component signal transduction

Figure S11 Comparison of transition state complexes of Ras and F1-ATPase with the AMP-PNP structure of M. tuberculosis GyrB47

(A) Ras (cyan carbons)/RasGAP(dark green carbons) complexed with ADP and AlF3 (PDB code 1WQ1). The Mg2 + is shown as a small green sphere. Arg789 from RasGAP accelerates ATP hydrolysis.(B) F1-ATPase with ADP and BeF3. The Mg2 + is shown as a small green sphere. Two F1-ATPase subunits are indicated by carbons in different shades of green. (C) MtbGyrB47–AMP-PNP. (D)Superimposition of (A–C).



Figure S12 A simplified dissociative mechanism for the hydrolysis of GTP by Ras

(A) In the presence of wild-type Ras, Gln61 helps to position a water, so that it can attack the metaphosphate ion [PO3] − , once it is formed. Movement of the NH of Gly13 past the bridging oxygen isresponsible for the initial protonation, and by the time the glycine has moved again (to a position where it can take back the proton from GDP) the phosphate ion has formed. (B) In the presence ofan Ala61 mutant the water is not optimally positioned, so the metaphosphate ion is likely to be re-attacked by GDP before the water can attack.

REFERENCES

1 Corbett, K. D. and Berger, J. M. (2005) Structural dissection of ATP turnover in theprototypical GHL ATPase TopoVI. Structure 13, 873–882

2 Frey, P. A. and Hegeman, A. D. (2007) Enzymatic Reaction Mechanisms. OxfordUniversity Press, New York

3 Saenger, W. (1983) Principles of Nucleic Acid Structure. Springer-Verlag, New York4 Alberty, R. A. and Goldberg, R. N. (1992) Standard thermodynamic formation properties

for the adenosine 5′-triphosphate series. Biochemistry 31, 10610–106155 Storer, A. C. and Cornish-Bowden, A. (1976) Concentration of MgATP2 − and other ions

in solution. Calculation of the true concentrations of species present in mixtures ofassociating ions. Biochem. J. 159, 1–5

6 Reynolds, M. A., Gerlt, J. A., Demou, P. C., Oppenheimer, N. J. and Kenyon, G. L. (1983)15N and 17O NMR studies of the proton binding sites in imidodiphosphate, tetraethylimidodiphosphate, and adenylyl imidodiphosphate. J. Am. Chem. Soc. 105, 6475–6481

7 Yount, R. G., Babcock, D., Ballantyne, W. and Ojala, D. (1971) Adenylylimidodiphosphate, an adenosine triphosphate analog containing a P-N-P linkage.Biochemistry 10, 2484–2489

8 Moriarty, N. W., Grosse-Kunstleve, R. W. and Adams, P. D. (2009) electronic LigandBuilder and Optimization Workbench (eLBOW): a tool for ligand coordinate and restraintgeneration. Acta Crystallogr., Sect. D: Biol. Crystallogr. 65, 1074–1080

9 Adams, P. D., Afonine, P. V., Bunkoczi, G., Chen, V. B., Davis, I. W., Echols, N., Headd,J. J., Hung, L. W., Kapral, G. J., Grosse-Kunstleve, R. W. et al. (2010) PHENIX: acomprehensive Python-based system for macromolecular structure solution. ActaCrystallogr., Sect. D: Biol. Crystallogr. 66, 213–221

10 Emsley, P., Lohkamp, B., Scott, W. G. and Cowtan, K. (2010) Features and development ofCoot. Acta Crystallogr., Sect. D: Biol. Crystallogr. 66, 486–501

11 Maxwell, A. and Gellert, M. (1984) The DNA dependence of the ATPase activity of DNAgyrase. J. Biol. Chem. 259, 14472–14480

12 Lindsley, J. E. and Wang, J. C. (1993) On the coupling between ATP usage and DNAtransport by yeast DNA topoisomerase II. J. Biol. Chem. 268, 8096–8104

13 Hammonds, T. R. and Maxwell, A. (1997) The DNA dependence of the ATPase activity ofhuman DNA topoisomerase IIα. J. Biol. Chem. 272, 32696–32703



14 Gardiner, L. P., Roper, D. I., Hammonds, T. R. and Maxwell, A. (1998) The N-terminaldomain of human topoisomerase IIα is a DNA-dependent ATPase. Biochemistry 37,16997–17004

15 Sengupta, T., Mukherjee, M., Das, A., Mandal, C., Das, R., Mukherjee, T. and Majumder,H. K. (2005) Characterization of the ATPase activity of topoisomerase II from Leishmaniadonovani and identification of residues conferring resistance to etoposide. Biochem. J.390, 419–426

16 Dar, M. A., Sharma, A., Mondal, N. and Dhar, S. K. (2007) Molecular cloning ofapicoplast-targeted Plasmodium falciparum DNA gyrase genes: unique intrinsic ATPaseactivity and ATP-independent dimerization of PfGyrB subunit. Eukaryotic Cell 6,398–412

17 Raghu Ram, E. V., Kumar, A., Biswas, S., Kumar, A., Chaubey, S., Siddiqi, M. I. and Habib,S. (2007) Nuclear gyrB encodes a functional subunit of the Plasmodium falciparum gyrasethat is involved in apicoplast DNA replication. Mol. Biochem. Parasitol. 154, 30–39

Received 15 April 2013/4 September 2013; accepted 9 September 2013Published as BJ Immediate Publication 9 September 2013, doi:10.1042/BJ20130538


Documents

Mycobacterium tuberculosisDNA gyrase ATPase domain … · 2017. 9. 23. · DNA gyrase, a type II topoisomerase, regulates DNA topology by creating a double-stranded break in one DNA