DNA-induced narrowing of the gyrase N-gatecoordinates T-segment capture and strand passageAirat Gubaeva,b and Dagmar Klostermeiera,b,1

aInstitute for Physical Chemistry, University of Muenster, Corrensstrasse 30, D-48149 Muenster, Germany; and bBiozentrum, Department of BiophysicalChemistry, University of Basel, Klingelbergstrasse 70, CH-4056 Basel, Switzerland

Edited by Martin Gellert, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD, and approved June 30,2011 (received for review February 8, 2011)

DNA gyrase introduces negative supercoils into DNA in an ATP-dependent reaction. DNA supercoiling is catalyzed by a strand-passage mechanism, in which a T-segment of DNA is passedthrough the gap in a transiently cleaved G-segment. Strand pas-sage requires the coordinated closing and opening of three proteininterfaces in gyrase, the N-gate, DNA-gate, and C-gate. We showhere that DNA binding to the DNA-gate of gyrase and wrappingof DNA around the C-terminal domains of GyrA induces a nar-rowing of the N-gate. This half-closed state prepares capture ofa T-segment in the upper cavity of gyrase. Subsequent N-gateclosure upon binding of ATP then poises the reaction toward strandpassage. The N-gate reopens after ATP hydrolysis, allowing forfurther catalytic cycles. DNA binding, cleavage, and wrappingand N-gate narrowing are intimately linked events that coordinateconformational changes at the DNA and the N-gate.

gyrase dynamics ∣ nucleotide-induced conformational changes ∣topoisomerase mechanism ∣ DNA wrapping ∣ negative supercoiling

DNA topoisomerases catalyze the interconversion of DNAtopoisomers and impact key cellular events such as replica-

tion, recombination, and transcription (1, 2). Bacterial gyrase isthe only type II topoisomerase that introduces negative supercoilsinto DNA in an ATP-dependent reaction (3). The active form ofgyrase is a heterotetramer of two GyrA and two GyrB subunits(Fig. 1). Structures of fragments of Escherichia coli GyrB (4) andGyrA (5, 6), and of a fusion of parts of GyrB and GyrA (7) havebeen determined, but the overall architecture of gyrase remainsunknown. DNA supercoiling by gyrase occurs via a strand-passage mechanism (8–11). A double-stranded DNA segmentbinds to the DNA-gate (5, 12), close to the catalytic tyrosines thatperform the cleavage reaction. Cleavage of this gate DNA(G-segment) generates a gap through which a second DNA seg-ment (T-segment) is transported. We have previously shown thatDNA bound to the DNA-gate is distorted in a process coupled tocleavage (13). The N-gate formed by the GyrB subunits acts as anATP-dependent clamp that closes due to nucleotide-induceddimerization, leading to the trapping of a T-segment (4, 14) in theupper cavity delimited by the N and the DNA-gate. The T-seg-ment can pass the gap in the G-segment created by double-strandcleavage. Importantly, the DNA-gate is predominantly closedduring negative supercoiling (13), indicating that opening of thegate occurs only transiently to allow for strand passage. An activecontribution of the T-segment to the opening of the DNA-gatehas been proposed (14–16). Closure of the DNA-gate subsequentto strand passage may then lead to expulsion of the transportedsegment from the lower cavity through the C-gate (17, 18). A“double lock” rule according to which each gate can only adoptan open conformation if the two others are closed, has beensuggested as a simple principle to ensure unidirectional strandpassage in type II topoisomerases (19, 20). Directed strandpassage thus requires the regulated opening and closing of theN-gate, the DNA-gate, and the C-gate, and may be guided bythe GyrA C-terminal domains (CTDs; Fig. 1).

The high level of flexibility required for gyrase function, incombination with its appreciable size, has rendered structuralstudies of the complete enzyme difficult. Whereas biochemicaland structural data have provided support for the proposedextensive conformational changes, opening and closing of indivi-dual gates has not been observed directly, and the coordination ofthe conformational changes of gyrase is unclear. Single moleculefluorescence resonance energy transfer (smFRET) is ideallysuited to investigate the dynamics underlying DNA supercoilingby gyrase. In smFRETexperiments, we monitored the conforma-tional state of the N-gate in the catalytic cycle of Bacillus subtilisgyrase. The N-gate closes when ATP binds to the GyrB subunits,and reopening occurs after ATP hydrolysis. Interestingly, bindingof DNA already leads to a narrowing of the N-gate before nucleo-tide is bound. Using linear DNA substrates of different lengths,we demonstrate that wrapping of the DNA around the CTDstriggers narrowing of the N-gate. Our results point toward abidirectional communication between DNA-gate and N-gate thatensures the coordination of gate movements during DNA super-coiling by gyrase.

ResultsA GyrB–GyrA Fusion to Investigate the Conformation of the N-Gate inB. subtilis DNA Gyrase. The active form of DNA gyrase is a hetero-tetramer of two GyrA and two GyrB subunits. B. subtilis GyrA isa stable dimer, even at picomolar concentrations (13), whereasisolated GyrB is monomeric (21). To prevent formation of incom-plete complexes (22, 23) or complex dissociation during smFRETexperiments that address conformational changes of the N-gateformed by the GyrB subunits, we fused the coding regions forGyrB and GyrA, such that the two subunits are connected viaa linker of the sequence G-A-P (Fig. 1A). The resulting fusionprotein GyrBA (164 kDa) eluted as an apparent dimer from acalibrated size-exclusion column, and retained wild-type-likeDNA relaxation and ATP-dependent negative supercoiling activ-ities (Fig. S1).

The N-Gate Closes Upon ATP Binding, and Reopens when ATP Is Hydro-lyzed. Cysteines were introduced at the N-gate of GyrBA at re-sidues S7, E17, and E131. The cysteine mutants were labeled withthe donor fluorophore AlexaFluor 488 (A488), and the acceptorfluorophore AlexaFluor 546 (A546). Topoisomerase activities ofthe fluorescently labeled GyrBA constructs were only slightlyreduced compared to wild-type gyrase (Fig. S1). smFRET histo-grams (Fig. 1C) show unimodal distributions of FRETefficienciesaround 0.1, in agreement with an open N-gate in the absence of

Author contributions: A.G. and D.K. designed research; A.G. performed research; A.G.and D.K. analyzed data; and A.G. and D.K. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: dagmar.klostermeier@uni-muenster.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1102100108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1102100108 PNAS ∣ August 23, 2011 ∣ vol. 108 ∣ no. 34 ∣ 14085–14090

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

1

DNA or nucleotide. Binding of the nonhydrolyzable ATP analogADPNP induces dimerization of GyrB (4, 21). Upon additionof ADPNP, the FRET efficiencies of donor/acceptor-labeledGyrBA increased to 0.34 (S7C), 0.97 (E17C), and 0.52 (E131C),consistent with a closure of the N-gate due to nucleotide binding(Fig. 1C). In the structure of the dimeric GyrB ATPase domainfrom E. coli (4), the corresponding residues for S7 and E131 arelocated further away from the GyrB dimer interface than E17C.S7 is located on the intertwined arms of the GyrB dimer whoseconformation in the monomeric protein is unknown, whereasE131 is on the GyrB body. The FRET efficiencies observed,and the derived interdye distances (5.8 nm for S7C, >3.1 nmfor E17C, 5.6 nm for E131C) are similar or slightly larger com-pared to the Cβ–Cβ-distances between these residues in the crys-tal structure of the GyrB dimer (4), in agreement with similarconformations of GyrB in the crystal structure of the isolatedATPase domain and in the closed conformation of the N-gatein gyrase.

To identify at which state in the nucleotide cycle of gyrasethe N-gate reopens, we analyzed the N-gate conformation in thepresence of ADP · BeFx, ADP · MgFx, and ADP. ADP · BeFxis a prehydrolysis ATP analog (24–26), and ADP · MgFx acts asa posthydrolysis analog mimicking a product state prior to phos-phate release (27–30). For GyrBA_S7C and _E17C, observed

FRETefficiencies in the ADP · BeFx-bound state were indicativeof a closed N-gate. In contrast, FRET efficiencies in the ADP ·MgFx- and ADP-bound states corresponded to an open N-gate(Fig. S2). These data are consistent with reopening of the N-gateafter hydrolysis of both ATP molecules.

DNA Binding to Gyrase Induces a Narrowing of the N-Gate. Next, westudied the effect of DNA on N-gate conformation. Addition ofrelaxed plasmid DNA caused an increase in FRETefficiency forall three constructs (Figs. 1D and 3A), demonstrating that DNAbinding affects the conformation of the N-gate. Interestingly, theincrease in FRETefficiency was also observed for the complex ofgyrase with negatively supercoiled DNA (Fig. 2A and Fig. S3),indicating that the movement of the N-gate is independent ofthe supercoiling state of the DNA substrate. We have previouslyestablished that a linear DNA of 60 bp (Fig. S3) is a suitable mod-el for a gate DNA (13). Interestingly, binding of this DNA had noeffect on N-gate conformation (Fig. S3). Thus, binding of plasmidDNA, but not of short linear DNA, can trigger a conformationalchange of the N-gate already in the absence of nucleotides.

The FRET efficiency for the GyrBA_E17C/DNA complex issimilar to the value for the closed N-gate in the ADPNP complex.In contrast, the FRET efficiencies for the GyrBA_S7C and_E131C/DNA complexes (0.72 for S7C, 0.39 for E131C, Fig. 1D)

Fig. 1. Constructs and conformation of the N-gate in the nucleotide cycle. (A) Scheme of the gyrase fusion protein GyrBA used in this study. The cartoon showsthe domain arrangement in the GyrA2GyrB2 heterotetramer, the position of the labels (D, A), and N-gate closure upon ADPNP binding. (B) Cysteines labeledwith donor and acceptor fluorophores for smFRETexperiments are highlighted in the dimer of the GyrB ATPase domain in the presence of ADPNP [PDB ID code1EI1(4)]. One monomer is depicted in blue, the second in white. S7 and E17 are located on the arms interchanged between the two monomers, and E131 islocated on the body of the ATPase domain. (C) The smFRET histograms for GyrBA_E17C, GyrBA_S7C, and GyrBA_E131C in the absence (color code as in B, openN-gate indicated by o) and presence of ADPNP (black, closed N-gate, indicated by c). (D) The smFRET histograms for GyrBA_E17C, GyrBA_S7C, and GyrBA_E131Cin the absence (color code as in B) and presence of relaxed plasmid (black) show a DNA-induced conformational change of the N-gate. The N-gate conformationin the DNA-bound state (i, intermediate) is different from the ADPNP-bound state (c).

14086 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1102100108 Gubaev and Klostermeier

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

1

are clearly different from the values in the ADPNP-bound, closedN-gate (0.34 for S7C and 0.52 for E131C, Fig. 1C). Thus, the con-formations of the N-gate in the ADPNP- and DNA-bound statesof gyrase are different. S7C is located on the N-terminal arms ofGyrB, but FRET efficiencies between dyes attached to E131Cdirectly report on movements of the GyrB body. The FRETeffi-ciency of GyrBA_E131C for the plasmid-bound state is higherthan for the open N-gate, and lower than for the closed N-gate.Overall, these data thus point to the formation of an intermediatestate in the presence of DNA, in which the GyrB subunits havemoved closer together and the N-gate has narrowed.

DNA-Induced N-Gate Narrowing Is Hampered in Cleavage-DeficientGyrase. To investigate the role of DNA cleavage for the DNA-induced conformational change of the N-gate, we followed the

N-gate conformation in the cleavage-deficient mutants GyrBA_E17C_Y123Fand GyrBA_S7C_Y123F. For both constructs, simi-lar FRET efficiencies to the cleavage-competent counterpartswere observed in the absence and in the presence of ADPNP,confirming that neither the open N-gate conformation nor theADPNP-induced N-gate closure are affected by the mutation(Fig. S4). However, upon addition of plasmid DNA to cleavage-deficient gyrase, a significant fraction of complexes with an openN-gate remained, indicating that plasmid-induced narrowing ofthe N-gate was hampered (Fig. 2C and Fig. S4). These data sug-gest that DNA cleavage may facilitate efficient DNA-inducednarrowing of the N-gate, and may precede N-gate narrowing.

Different Requirements of CTDs for Nucleotide- and DNA-MediatedN-Gate Closure. To address the role of the CTDs in the DNA-induced conformational change of the N-gate, we created donor/acceptor-labeled GyrBA lacking the CTDs. GyrBA_ΔCTD cat-alyzed the relaxation of positively and negatively supercoiledDNA (Fig. S5), but did not supercoil DNA. Whereas GyrBA_ΔCTD still bound DNA (Fig. S6), it did not show DNA-stimu-lated ATPase activity (Fig. 4C). FRET histograms for GyrBA_E17C_ΔCTDwere similar to GyrBA_E17C (Fig. 1C and Fig. S7).Despite the lack of DNA-stimulated ATPase activity, the N-gateof GyrBA_E17C_ΔCTD closed upon binding of ADPNP(Fig. S7), leading to the population of a high FRET species(EFRET ¼ 1). The similar response to ADPNP compared to wild-type is in line with the CTDs not being required for N-gateclosure per se. In contrast to wild-type GyrBA, however, Gyr-BA_E17C_ΔCTD did not show N-gate narrowing upon plasmidDNA binding (Fig. 3A). These results illustrate that the CTDsplay a critical role in DNA-induced N-gate narrowing, but aredispensable for N-gate closure due to GyrB dimerization uponnucleotide binding.

Narrowing of the N-Gate Requires DNA Wrapping Around the CTDs.A60-bp gate DNA is not able to trigger a conformational change inthe gyrase N-gate as observed with plasmid DNA, suggesting thateither wrapping of the DNA around the CTDs or the presence ofa T-segment is required as a trigger for this conformationalchange of the N-gate. To distinguish between these alternatives,we prepared linear dsDNA fragments with varying lengths from90, 110, 119, and 139 bp (Fig. S3) that contain a preferred clea-vage site for B. subtilis gyrase (31) in the center. Binding of theseDNAs to gyrase and cleavage at the preferred site were con-firmed (13) (Fig. 4 A and B). Kd values were determined fromdisplacement titrations of a gyrase/DNA complex containingfluorescently labeled DNA (Fig. 4B and Fig. S6). Quantivativeanalysis yielded Kd values of 220 nM for the 60-bp DNA(Fig. S6). For the pUC18 plasmid, the number of gyrase bindingsites is not known, precluding a quantitative analysis. However,50% of fluorescently labeled DNA bound to gyrase are displacedby approximately 150 nM 60-bp DNA, approximately 80 nM90-bp DNA, or approximately 1 nM plasmid, corresponding toconcentrations of 9 μM, 7.2 μM, and 2.7 μM in terms of basepairs. Thus, the titration curves in comparison clearly demon-strate that the Kd value for the 60-bp DNA sets as a lower limitfor DNA affinity. Based on this limiting value for DNA affinity,the effect of saturating concentrations of the DNAs on N-gateconformation was tested in smFRET experiments with GyrBA_E17C and _S7C that display the largest changes in EFRET uponN-gate narrowing (Fig. 4D and Fig. S6). The FRET efficienciesremained low in the presence of the 60- and 90-bp DNA, demon-strating that binding of these short DNAs does not affect the con-formation of the N-gate. In contrast, the FRET efficienciesincreased when the 110-, 119-, and 139-bp DNAs were bound,indicating that these DNAs triggered narrowing of the N-gatein both constructs. The cutoff at approximately 110 bp is in agree-ment with the previously determined length of DNA protected

Fig. 3. Role of the CTDs in N-gate narrowing. (A) The smFRET histogramof GyrBA_E17C_ΔCTD in the presence of pUC18rel (black). The histogramfor GyrBA_E17C (gray) is shown for comparison. (B) The smFRET histogramof a phosphorothiolate-modified donor/acceptor-labeled 60-bp DNA in com-plex with GyrBA_E17C_ΔCTD. The mutant lacking the CTDs does not severelydistort the DNA. Percentages refer to fractions of low-FRET (gray) and high-FRET species (black). See also Fig. S7.

Fig. 2. N-gate conformation of cleavage-deficient gyrase. (A) The smFREThistograms for GyrBA_S7C in the presence of pUC18rel (gray) and pUC18−sc

(black). (B) The smFRET histograms for GyrBA_S7C in the presence ofpUC18rel (gray) and pUC18−sc (black) and ADPNP, added to the DNA complex.(C) The smFRET histograms for GyrBA_S7C_Y123F (black) in the absence andpresence of pUC18rel. The histogram for cleavage-competent GyrBA_S7C inthe presence of pUC18rel (gray) is shown for comparison. (D) The smFRET his-tograms for GyrBA_S7C_Y123F (black) in the presence of pUC18rel andADPNP (added to the DNA complex). The histogram for cleavage-competentGyrBA_S7C in the presence of pUC18rel and ADPNP (gray) is shown for com-parison. (E) The smFRET histograms for GyrBA_S7C with pUC18rel and ATP.

Gubaev and Klostermeier PNAS ∣ August 23, 2011 ∣ vol. 108 ∣ no. 34 ∣ 14087

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

1

upon wrapping (12). Although at this point it cannot be excludedthat the ends of the 110-, 119-, and 139-bp DNAs may form firstcontacts with GyrB, geometric considerations suggest that theseDNAs are too short to provide a bona fide T-segment (SI Text),suggesting that a T-segment is not required for N-gate narrowing.The increase in affinity with DNA length predicts that theseDNAs are wrapped around the CTDs to increasing extents. Over-all, our data would thus be consistent with complete wrapping ofDNA around the CTDs as a trigger for N-gate narrowing.

Interestingly, the GyrBA ATPase activity was stimulated verylittle by the 60- and 90-bp DNAs, but significantly increased inthe presence of the 110-, 119-, and 139-bp DNAs and pUC18(Fig. 4C), similar to E. coli DNA gyrase (32). The relative DNA-stimulated ATPase activities paralleled the fraction of gyrase with anarrowed N-gate determined from smFRETexperiments, pointingtoward a role of N-gate narrowing in ATPase stimulation.

N-Gate Narrowing and its Relation to N-Gate Closure and T-SegmentCapture. DNA-induced narrowing of the N-gate occurred withlinear and circular DNAs, independent of the supercoiling state.Once DNA has been bound and the N-gate has narrowed, nu-cleotide binding will lead to complete N-gate closure and conco-mitant capture of a T-segment in the upper cavity. To follow theconformation of the N-gate in this subsequent step of the catalyticcycle, we added ADPNP to gyrase/plasmid complexes. Interest-ingly, the effect of ADPNP on the conformation of the N-gate inthese complexes differed depending on the supercoiling state ofthe bound plasmid: For gyrase bound to relaxed plasmid, additionof ADPNP converted only a fraction of gyrase into an intermedi-ate-FRET species (FRET efficiency 0.3) indicative of a closedN-gate (Fig. 2B). In contrast, addition of ADPNP to the complexof GyrBA with negatively supercoiled DNA led to a higherfraction of gyrase molecules with a closed N-gate (Fig. 2B). In allexperiments, gyrase is saturated with DNA and nucleotide, ex-cluding that the different species reflect bound and unboundforms. The experimentally observed differences in the responseof the two DNA complexes to ADPNP may be a consequence ofdifferent propensities for wrapping: Gyrase wraps DNA arounditself with a positive handedness (12, 33), leading to the presenta-tion of a T-segment with the correct geometry for capture andstrand passage. A positive node can be achieved more easilywith relaxed DNA than with negatively supercoiled DNA. Henceone would predict that with relaxed DNA a larger fraction of the

gyrase complexes has wrapped the DNA such that a T-segment isplaced between the GyrB arms than in complexes of gyrase withnegatively supercoiled DNA. When comparing this predictionto the experimental observations, a higher fraction of gyrase com-plexes with a T-segment present correlates with a higher fractionof complexes that retain a narrowed N-gate when ADPNP isadded, and thus with a reduced propensity to close the N-gate.The correlation suggests that the complexes without T-segmentin the upper cavity close the N-gate in response to ADPNP addi-tion, whereas the fraction that is not able to close the N-gate uponADPNP addition represents gyrase that contains a T-segmentbetween the GyrB arms. While this interpretation remainstentative in the absence of a direct measure for the presence ofa T-segment, our data would thus point toward an interferenceof a T-segment in the upper cavity with complete N-gate closureby ADPNP. The supercoiling activity of the GyrBA fusion proteinexcludes that this observation is an experimental artifact causedby the construct. Due to the reduced coupling of nucleotide bind-ing and strand passage with ADPNP compared to ATP (34) (seeDiscussion), only a small fraction of complexes with a T-segmentwill undergo strand passage, and will then become trapped with aclosed N-gate.

When ATP is added to the gyrase/plasmid complex, under con-ditions where strand passage and negative supercoiling occur, theN-gate is either in the open state, or in the intermediate state witha narrowed N-gate, indicating that the N-gate closes only transi-ently (Fig. 2E). Analysis of the time-dependent FRETefficienciesin individual fluorescent bursts also shows that the N-gate in in-dividual gyrase molecules spends most of the time in the open andnarrowed states during the supercoiling reaction, and only fewbrief excursions to a closed N-gate are observed (Fig. S8).

DNA Distortion, N-Gate Narrowing, and T-Segment Capture.We haveshown previously that DNA bound to gyrase is distorted (13).Cleavage-deficient gyrase neither shows DNA distortion (13)nor narrowing of the N-gate (this work), indicating that bothprocesses may be linked. To test this hypothesis, we directly testedif GyrBA_ΔCTD can distort DNA in smFRET experimentswith phosphorothiolate-modified donor/acceptor-labeled DNA.FRET histograms showed bimodal distributions, correspondingto similar fractions of slightly and severely distorted DNA aswe previously observed for wild-type gyrase (Fig. S7 and ref. 13).When ADPNP was added to the GyrBA/DNA complex, the low

Fig. 4. N-gate closure and the dependence on DNA length. (A) DNA cleavage assay for 90-, 119-, and 139-bp DNA in the presence of ciprofloxacin. (B) Aniso-tropy titration of GyrBA in complex with an A546-labeled 60-bp DNAwith unlabeled 60-, 90-bp, and plasmid DNA. (C) Steady state ATPase activity of gyrase inthe presence of linear DNAs and pUC18. Squares, GyrBA; open circles, GyrBA_ΔCTD. The stimulation of the ATPase activity increases with the DNA length. In thepresence of plasmid DNA, the ATPase activity is reduced compared of the 139-bp DNA. Here, strand passage occurs while the plasmid is negatively supercoiled,possibly leading to a change in the rate-limiting step. The stars denote the fraction of gyrase with intermediate N-gate conformation, as determined fromquantitative analysis of smFRET histograms (see D). The fractions of gyrase with a narrowed N-gate and the ATPase activities are correlated. (D) The smFREThistograms of GyrBA_S7C with 60-, 90-, 110-, 119-, and 139-bp DNA (from left to right). N-gate narrowing requires binding of DNA >110 bp.

14088 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1102100108 Gubaev and Klostermeier

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

1

FRET species corresponding to the severely distorted DNA wasreduced (Fig. S7), again similar to wild-type gyrase. These dataconfirm that the DNA-gate and its communication with theN-gate are functional in the fusion protein. In contrast, the FRETdistribution for GyrBA_E17C_ΔCTD was broad, but unimodalwhen the DNA was bound, and did not change upon ADPNPaddition (Fig. 3B and Fig. S7), indicating that here the DNAis not distorted. These results demonstrate that the CTDs arerequired for distortion of the DNA at the DNA-gate, and linkDNA distortion with N-gate narrowing.

Finally, we tested whether DNA distortion is a prerequisite forpresenting a T-segment to the upper cavity. Cleavage-deficientgyrase cannot distort or cleave the DNA substrate, and its Ngate does not respond to DNA binding. Addition of ADPNPto a GyrBA_S7C_Y123F/plasmid complex (Fig. 2D) led to anincrease in FRETefficiency, indicative of N-gate closure. In linewith our (tentative) interpretation that a T-segment between theGyrB arms interferes with ADPNP-induced N-gate closure(see above), this behavior suggests that no T-segment had beencaptured, and is consistent with DNA distortion and cleavagecontributing to T-segment capture. Altogether, our results pro-vide substantial evidence that DNA distortion, cleavage, andwrapping, N-gate narrowing, and T-segment capture are separate,yet intimately linked events that are highly coordinated in thecatalytic cycle of DNA gyrase.

DiscussionUsing smFRET, we have monitored the conformation of theN-gate in the catalytic cycle of gyrase. The N-gate closes uponATP binding, and reopens after ATP hydrolysis. Strikingly, DNAbinding affects the conformation of the N-gate, and leads to anintermediate state with a narrowed N-gate that has not been ob-served before. DNA cleavage, distortion, and wrapping, and N-gatenarrowing appear to be separate events that occur in a highly con-certed manner to achieve T-segment capture and strand passage.

The intermediate state with a narrowed N-gate is a well-defined, yet flexible conformation that constitutes a hithertounidentified link in coordinating events at N-gate and DNA-gate.In this state, the ATP-binding domains have come closer, but notyet close enough to engage in dimer contacts. Narrowing of theN-gate may favor productive encounters of the GyrB subunits atthe N-gate, and hence the transition of gyrase to the state with aclosed N-gate. Dimerization of GyrB leads to formation of theactive site for ATP hydrolysis (4), and complete N-gate closureshould be required for ATPase stimulation. Yet, DNA stimulationof ATP hydrolysis is correlated with the fraction of gyrase with anarrowed N-gate. Cleavage-deficient gyrase and gyrase lackingthe CTDs do not show N-gate narrowing in response to DNA,and show no DNA-stimulated ATP hydrolysis (this work andrefs. 9 and 10), further supporting that ATPase stimulation andN-gate narrowing are linked. Possibly, gyrase spends less timein slowly-hydrolyzing states in the presence of DNA, leading tostimulation of the ATPase.

Although at this point we cannot completely exclude that DNAextending from the CTDs interacts with GyrB and contributes toN-gate narrowing, the presence of a T-segment in the upper cavitydoes not seem to be strictly required for this conformationalchange to occur. This interpretation is supported by the observa-tion that N-gate narrowing does not depend on the supercoilingstate of the bound DNA. Instead, the probability to capture aT-segment between the GyrB arms is related to the supercoilingstate of the bound DNA, and is higher for relaxed than for nega-tively supercoiled DNA. Notably, gyrase relaxes positively super-coiled DNA more efficiently than negatively supercoiled DNAwhen the CTDs are deleted, in support of the idea that theintrinsic activity of gyrase is the relaxation of positive supercoilsin DNA that are present in the DNA, or are generated by DNAwrapping. The formation of gyrase with a narrowed N-gate may

help to tightly couple G-segment binding and T-segment capture,thereby favoring capture of a T-segment adjacent to the G-seg-ment, and thus intramolecular strand passage. Possibly, such astate may not be found in topoisomerases that catalyze decatena-tion by intermolecular strand transfer.

The default state of the N-gate is the open conformation, andonly brief excursions to the closed state occur during negativesupercoiling, in line with transient closing. N-gate narrowing islinked to gate-DNA distortion and cleavage, processes that havebeen suggested to contribute to unlocking of the DNA-gate,which can then open transiently (13). Our smFRET data are inagreement with a bidirectional communication, from the DNA-gate to the N-gate during DNA-gate unlocking and N-gate nar-rowing, and in the reverse direction during N-gate closure andDNA-gate opening. Concerted, anticorrelated movement of theN-gate and DNA-gate in the presence of a T-segment (flip-flopmechanism) would explain the observed interdependence ofnucleotide and DNA binding (34), and the suggested active roleof a T-segment for DNA-gate opening (4, 14, 19, 20, 35), and is inagreement with the widely accepted double lock rule for type IItopoisomerases (19, 20). The state of gyrase with a narrow N-gatewould constitute a spring-loaded state, waiting for ATP bindingto close the N-gate completely, and concomitantly open theDNA-gate if a T-segment is present. The observed interferenceof a T-segment with N-gate closure upon ADPNP binding mayindicate that the smaller binding energy from ADPNP comparedto ATP does not suffice to efficiently open the DNA-gate, in linewith the reported reduced coupling of ADPNP binding to strandpassage (34). According to the suggested contribution of a T-seg-ment to DNA-gate opening, N-gate closure will not open theDNA-gate in the absence of a T-segment. In this case, ATPhydrolysis liberates gyrase for further (productive) cycles. Ithas been shown before that gyrase does not show 100% couplingof ATP hydrolysis to strand passage, and keeps hydrolyzingATP even when no more supercoils can be introduced (34, 36).

Once a T-segment has been captured in the upper cavity, it canequilibrate on both sides of the open DNA-gate (10). However,the upper cavity is too small to accommodate a T-segment whenboth N-gate and DNA-gate are closed (4, 14). Therefore, as longas the N-gate remains closed, the DNA-gate can only return to itsdefault closed state when the T-segment is in the (larger) bottomcavity. As a consequence, N-gate closure will enforce strand pas-sage through the open DNA-gate, and push the T-segment intothe (larger) bottom cavity. Return of the DNA-gate to the closedstate may then drive expulsion of the T-segment through the

Fig. 5. Model of conformational changes in the catalytic cycle of gyrase andtheir relation to strand passage. DNA gyrase (1) binds DNA. DNA distortion,cleavage, and wrapping around the CTDs lead to a change in N-gate confor-mation toward an intermediate, narrow state (2, highlighted). ATP bindingcloses the N-gate. A T-segment captured in the upper cavity contributes toa concerted opening of the DNA-gate (3), resulting in strand passage (4). TheN-gate reopens upon ATP hydrolysis after strand passage has been completed(5), and ADP and phosphate are released. DNA can be released (1) or theG-segment remains bound, and rewrapping will start a new catalytic cycle(2). Yellow ellipse: ATP, split ellipse: ADP∕Pi .

Gubaev and Klostermeier PNAS ∣ August 23, 2011 ∣ vol. 108 ∣ no. 34 ∣ 14089

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

1

C-gate (15, 37). In this scenario, strand passage is a mere conse-quence of the coordination of gate movements. Once the N-gatehas closed, strand passage is inevitable, and one supercoilingcycle will be completed, nicely illustrating how ATP binding pro-vides the driving force for strand passage (38). ATP hydrolysis,the rate-limiting in the catalytic cycle of B. subtilis gyrase in theabsence of DNA (21), leads to reopening of the N-gate, and in-itiates a new catalytic cycle, thereby providing a timing function.It is currently unknown at which stage of the catalytic cycle ATPis hydrolyzed, and thus when the conformational change of theN-gate happens. Tight coupling of ATP hydrolysis and negativesupercoiling would predict ATP hydrolysis and reopening of theN-gate to occur after strand passage, when the DNA-gate hasclosed. The coupling of the concerted conformational changesduring strand passage to ATP hydrolysis imposes irreversibility onthe supercoiling reaction. Fig. 5 summarizes our current knowl-edge of the conformational states of gyrase during the supercoil-ing reaction.

Overall, the formation of the intermediate state with a narrowN-gate upon DNA binding at the beginning of the catalytic cycle

initiates a multilayered process that couples T-segment capture toG-segment binding, and coordinates N-gate and DNA-gate con-formational changes. Strand passage and negative supercoilingare a direct consequence of the subsequent concerted openingand closing of the three gyrase gates.

Materials and MethodsThe B. subtilis GyrBA fusion was constructed by overlap-extension PCR suchthat the coding region for GyrB was fused to the coding region for GyrA viaa linker coding for a peptide linker of the sequence GAP. The fusion proteinwas purified via ion exchange and size-exclusion chromatography (see SIText). Kd values of gyrase/DNA complexes, rates of ATP hydrolysis, and DNAtopoisomerase activity were determined as described (13). The smFRETexperiments were performed and analyzed as described (13). See SI Textfor detailed descriptions of experimental procedures.

ACKNOWLEDGMENTS. We thank Martin Lanz and Markus Rudolph forcomments on the manuscript, and Diana Blank, Ines Hertel, and AndreasSchmidt for excellent technical assistance. This work was supported by theVolkswagenStiftung, the Swiss National Science Foundation, and theNational Center of Competence in Research (NCCR) Nanoscale Sciences.

1. Corbett KD, Schoeffler AJ, Thomsen ND, Berger JM (2005) The structural basis forsubstrate specificity in DNA topoisomerase IV. J Mol Biol 351:545–561.

2. Schoeffler AJ, Berger JM (2008) DNA topoisomerases: Harnessing and constrainingenergy to govern chromosome topology. Q Rev Biophys 41:41–101.

3. Gellert M, Mizuuchi K, O’Dea MH, Nash HA (1976) DNA gyrase: An enzyme thatintroduces superhelical turns into DNA. Proc Natl Acad Sci USA 73:3872–3876.

4. Wigley DB, Davies GJ, Dodson EJ, Maxwell A, Dodson G (1991) Crystal structure of anN-terminal fragment of the DNA gyrase B protein. Nature 351:624–629.

5. Morais Cabral JH, et al. (1997) Crystal structure of the breakage-reunion domain ofDNA gyrase. Nature 388:903–906.

6. Corbett KD, Shultzaberger RK, Berger JM (2004) The C-terminal domain of DNAgyrase A adopts a DNA-bending beta-pinwheel fold. Proc Natl Acad Sci USA 101:7293–7298.

7. Schoeffler AJ, May AP, Berger JM (2010) A domain insertion in Escherichia coliGyrB adopts a novel fold that plays a critical role in gyrase function. Nucleic AcidsRes 38:7830–7844.

8. Williams NL, Maxwell A (1999) Probing the two-gate mechanism of DNA gyrase usingcysteine cross-linking. Biochemistry 38:13502–13511.

9. Williams NL, Maxwell A (1999) Locking the DNA-gate of DNA gyrase: Investigating theeffects on DNA cleavage and ATP hydrolysis. Biochemistry 38:14157–14164.

10. Kampranis SC, Bates AD, Maxwell A (1999) A model for the mechanism of strandpassage by DNA gyrase. Proc Natl Acad Sci USA 96:8414–8419.

11. Williams NL, Howells AJ, Maxwell A (2001) Locking the ATP-operated clamp of DNAgyrase: Probing the mechanism of strand passage. J Mol Biol 306:969–984.

12. Liu LF, Wang JC (1978) DNA-DNA gyrase complex: The wrapping of the DNA duplexoutside the enzyme. Cell 15:979–984.

13. Gubaev A, Hilbert M, Klostermeier D (2009) The DNA-Gate of Bacillus subtilis gyraseis predominantly in the closed conformation during the DNA supercoiling reaction.Proc Natl Acad Sci USA 106:13278–13283.

14. Tingey AP, Maxwell A (1996) Probing the role of the ATP-operated clamp in the strand-passage reaction of DNA gyrase. Nucleic Acids Res 24:4868–4873.

15. Wang JC (1998) Moving one DNA double helix through another by a type II DNAtopoisomerase: The story of a simple molecular machine. Q Rev Biophys 31:107–144.

16. Schoeffler AJ, Berger JM (2005) Recent advances in understanding structure-function relationships in the type II topoisomerase mechanism. Biochem Soc Trans33:1465–1470.

17. Roca J, Wang JC (1994) DNA transport by a type II DNA topoisomerase: Evidence infavor of a two-gate mechanism. Cell 77:609–616.

18. Roca J, Berger JM, Harrison SC, Wang JC (1996) DNA transport by a type IItopoisomerase: Direct evidence for a two-gate mechanism. Proc Natl Acad Sci USA93:4057–4062.

19. Roca J (2009) Topoisomerase II: A fitted mechanism for the chromatin landscape.Nucleic Acids Res 37:721–730.

20. Roca J (2004) The path of the DNA along the dimer interface of topoisomerase II. J BiolChem 279:25783–25788.

21. Gottler T, Klostermeier D (2007) Dissection of the nucleotide cycle of B. subtilis DNAgyrase and its modulation by DNA. J Mol Biol 367:1392–1404.

22. O’Dea MH, Tamura JK, Gellert M (1996) Mutations in the B subunit of Escherichia coliDNA gyrase that affect ATP-dependent reactions. J Biol Chem 271:9723–9729.

23. Kampranis SC, Maxwell A (1998) Hydrolysis of ATP at only one GyrB subunit issufficient to promote supercoiling by DNA gyrase. J Biol Chem 273:26305–26309.

24. Chabre M (1990) Aluminofluoride and beryllofluoride complexes: a new phosphateanalogs in enzymology. Trends Biochem Sci 15:6–10.

25. Bigay J, Deterre P, Pfister C, Chabre M (1987) Fluoride complexes of aluminum orberyllium act on G-proteins as reversibly bound analogues of the gamma phosphateof GTP. EMBO J 6:2907–2913.

26. Fisher AJ, et al. (1995) X-ray structures of the myosin motor domain of Dictyosteliumdiscoideum complexed withMgADP.BeFx andMgADP.AlF. Biochemistry 34:8960–8972.

27. Antonny B, Bigay J, Chabre M (1990) A novel magnesium-dependent mechanism forthe activation of transducin by fluoride. FEBS Lett 268:277–280.

28. Park S, Ajtai K, Burghardt TP (1999) Inhibition of myosin ATPase by metal fluoridecomplexes. Biochim Biophys Acta 1430:127–140.

29. Graham DL, et al. (2002) MgF3- as a transition state analog of phosphoryl transfer.Chem Biol 9:375–381.

30. Ahmad Z, Senior AE (2006) Inhibition of the ATPase activity of Escherichia coli ATPsynthase by magnesium fluoride. FEBS Lett 580:517–520.

31. Bashkirov VI, Zvingila DJ (1991) Sequence specificity of Bacillus subtilis DNA gyrase invivo. Genetica 85:3–12.

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

33. Orphanides G, Maxwell A (1994) Evidence for a conformational change in the DNAgyrase–DNA complex from hydroxyl radical footprinting. Nucleic Acids Res 22:1567–1575.

34. Bates AD, O’Dea MH, Gellert M (1996) Energy coupling in Escherichia coli DNA gyrase:The relationship between nucleotide binding, strand passage, and DNA supercoiling.Biochemistry 35:1408–1416.

35. Roca J, Wang JC (1992) The capture of a DNA double helix by an ATP-dependentprotein clamp: A key step in DNA transport by type II DNA topoisomerases. Cell71:833–840.

36. Sugino A, Cozzarelli NR (1980) The intrinsic ATPase of DNA gyrase. J Biol Chem255:6299–6306.

37. Berger JM, Gamblin SJ, Harrison SC, Wang JC (1996) Structure and mechanism ofDNA topoisomerase II. Nature 379:225–232.

38. Sugino A, Higgins NP, Brown PO, Peebles CL, Cozzarelli NR (1978) Energy couplingin DNA gyrase and the mechanism of action of novobiocin. Proc Natl Acad Sci USA75:4838–4342.

14090 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1102100108 Gubaev and Klostermeier

Dow

nloa

ded

by g

uest

on

Aug

ust 9

, 202

1