FEBS Letters 589 (2015) 2683–2689

journal homepage: www.FEBSLetters .org

NMR structural characterization of the N-terminal active domain of thegyrase B subunit from Pseudomonas aeruginosa and its complex with aninhibitor

http://dx.doi.org/10.1016/j.febslet.2015.07.0440014-5793/� 2015 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

Author contributions: C.K., J.C., J.H. and T.H.K designed the experiments. Y.L, Y.X.W.,Z.Y.P., Y.L.W., M.Y.L., H.Q.N., B.L, A.W.H., J.C., J.H., T.H.K. and C.K. conducted andanalyzed the experiments. C.K., Y.L., J.C., J.H. and T.H.K. finalized the manuscript.⇑ Corresponding author.

E-mail address: cbkang@etc.a-star.edu.sg (C. Kang).

Yan Li, Yun Xuan Wong, Zhi Ying Poh, Ying Lei Wong, Michelle Yueqi Lee, Hui Qi Ng, Boping Liu,Alvin W. Hung, Joseph Cherian, Jeffrey Hill, Thomas H. Keller, CongBao Kang ⇑Experimental Therapeutics Centre, Agency for Science, Technology and Research (A*STAR), 31 Biopolis Way, Nanos, #03-01, Singapore 138669, Singapore

a r t i c l e i n f o

Article history:Received 29 April 2015Revised 8 July 2015Accepted 26 July 2015Available online 10 August 2015

Edited by Christian Griesinger

Keywords:NMRDrug discoveryTopoisomeraseGyrase BStructure-based drug design

a b s t r a c t

The N-terminal ATP binding domain of the DNA gyrase B subunit is a validated drug target forantibacterial drug discovery. Structural information for this domain (pGyrB) from Pseudomonasaeruginosa is still missing. In this study, the interaction between pGyrB and a bis-pyridylurea inhi-bitor was characterized using several biophysical methods. We further carried out structural anal-ysis of pGyrB using NMR spectroscopy. The secondary structures of free and inhibitor bound pGyrBwere obtained based on backbone chemical shift assignment. Chemical shift perturbation and NOEexperiments demonstrated that the inhibitor binds to the ATP binding pocket. The results of thisstudy will be helpful for drug development targeting P. aeruginosa.� 2015 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction

The bacterial genome encodes two types of topoisomerases, Iand II which differ in the mechanism of DNA strand breakage [1].The type II topoisomerases consist of two types, DNA gyrase andtopoisomerase IV (TopoIV). These enzymes play essential roles inDNA replication by managing the topological states of DNA inthe cell [2,3]. In prokaryotes, type II topoisomerases consist oftwo subunits and are functional in tetrameric form, which is differ-ent from the eukaryotic type II topoisomerases that exist ashomodimers [3]. Prokaryotic DNA gyrase contains two gyrase A(GyrA) and gyrase B (GyrB) subunits respectively to form theheterotetramer. For Escherichia coli (E. coli), GyrA is a 97kDa proteinthat is involved in DNA binding and GyrB, it is a 90kDa protein withan N-terminal ATP binding domain [4].

Interfering with bacterial DNA replication by targeting type IItopoisomerases has been shown to be an efficient strategy todevelop antibacterial agents [5]. Successful examples include the

fluoroquinolone class of antibiotics [6]. Many other novel andpotent inhibitors have been developed in recent years [7]. TheN-terminal domain of the type II topoisomerases contains the ATPbinding pocket and has been of great interest in drug developmentbecause this domain exhibited high sequence homology amongpathogenic bacteria and low homology with eukaryotes [5,7].Structure-based drug design has been demonstrated to be a pow-erful tool in developing inhibitors targeting both GyrB andTopolV ATP binding domains. Several classes of inhibitors havebeen discovered using this approach [8–11].

The structures of the N-terminal ATP binding domains of bothGyrB and E subunit of TopoIV (ParE) from E. coli have been reported[12,13]. The structures of GyrB/ParE and inhibitor complexesdemonstrated that most of the inhibitors are binding with theATP binding pocket [9]. Although the structure of this domain issimilar among the type II topoisomerases, a single residue differ-ence among different topoisomerases can result in different inhibi-tor potency [2]. Understanding protein-inhibitor interactions willprovide useful information in the drug development process.NMR spectroscopy has been proven to be a useful tool in drugdevelopment [14]. Despite the extensive X-ray studies of GyrBsand ParEs, few NMR studies have been conducted for theN-terminal domain of GyrB and ParE from bacteria except for the

Fig. 1. NMR spectra of pGyrB. (A) 1H–15N-TROSY spectra of pGyrB. The NMR spectra of pGyrB in the absence (black) and presence (red) of the inhibitor were collected andsuperimposed, and two peaks that undergo significant shifts upon complex formation are highlighted. Inside spectra are selected regions of spectra with inhibitor/proteinratios of 0 (black), 0.5 (green) and 1 (red), respectively. The interaction is undergoing slow exchange. (B) 3D-HNCACB of pGyrB in the absence and presence of the inhibitor.Select strips of HNCACB spectrum for several residues are shown. Upper and lower panels are free and inhibitor-bound pGyrB, respectively. The peaks are labeled with residuenumber and atom types.

2684 Y. Li et al. / FEBS Letters 589 (2015) 2683–2689

assignments of the P24 fragment of Staphylococcus aureus and theN-terminal 24kDa fragment of GyrB from E. coli [15,16].

In this study, we obtained the N-terminal 24kDa domain of theGyrB from Pseudomonas aeruginosa (P. aeruginosa) (referred aspGyrB) for NMR studies. As the structure of this domain is notavailable, structural information for this domain will be usefulfor structure-based drug design because P. aeruginosa is an impor-tant pathogenic species. We managed to obtain the backboneassignments for both free and inhibitor-bound forms of pGyrB.The secondary structure and dynamic property of pGyrB in solu-tion were analyzed and the bis-pyridylurea inhibitor was shownto bind to the ATP binding pocket.

2. Materials and methods

2.1. Sample preparation

The cDNA encoding the pGyrB was amplified by polymerasechain reaction using genomic DNA of P. aeruginosa as a templateand cloned into NdeI and XhoI sites of pET29b. The resulting plas-mid can express residues 1–222 of GyrB and extra 8 residues(LEHHHHHH) at the C-terminus. To express pGyrB from E. coli,the plasmid was transformed in E. coli (BL21DE3) competent cells.The recombinant protein was expressed and purified using affinitypurification and gel filtration chromatography [17,18]. Briefly, sev-eral colonies were picked up from the plate and inoculated in20 mL of M9 medium. The overnight culture at 37 �C was thentransferred into 1 L of M9 medium. The recombinant protein wasinduced for 18 h at 18 �C by adding b-D-1-thiogalactopyranoside(IPTG) to 1 mM. The E. coli cells were harvested by centrifugationand the recombinant protein was purified in a buffer that con-tained 20 mM sodium phosphate, pH 6.5, 80 mM KCl, 2 mM DTTand 0.5 mM EDTA. A triple-labeled sample (13C, 15N and 2H) wasprepared by growing E. coli in a M9 medium that contained 1 g/L15NH4Cl, 2 g/L 2H-13C-glucose and D2O (99.9%). Purified proteinwas concentrated to 0.5–0.8 mM for further studies.

2.2. Backbone resonance assignment

Uniformly 15N- and 13C/15N/2H-labeled proteins were used inNMR data acquisition. Two- (2D) and three-dimensional (3D)experiments and transverse relaxation-optimized spectroscopy(TROSY) [19,20]-based experiments including HSQC, HNCACB,HNCOCACB, HNCOCA, HNCA, HNCACO and HNCO were collectedand processed. For pGyrB and inhibitor complex, protein was firstpurified and inhibitor was then added into the solution to a pro-tein: inhibitor molar ratio of 1:1.2. Inhibitor was synthesized andpurified as described [21]. All the experiments were conducted at25 �C on a Bruker Avance 700 spectrometer equipped with a cry-oprobe. All the spectra were processed using NMRPipe [22] orTopspin 2.1 and analyzed using NMRView [23] and CARA (http://www.mol.biol.ethz.ch/groups/wuthrich_group). The secondarystructure was analyzed using TALOS+ based on the backbonechemical shifts [24].

2.3. Protein-inhibitor interactions

1H–15N-HSQC spectra of pGyrB in the absence and presence ofthe inhibitor were compared and chemical shift perturbations(CSP) were monitored [25]. The combined chemical shift changes(Dd) were calculated using the following equation. Dd =((DdHN)2 + (DdN/5)2)0.5, where DdHN is the chemical shift changesupon inhibitor binding in the amide proton dimension and DdN isthe chemical shift changes in the amide dimension [25]. To obtainprotein-inhibitor inter-molecular NOEs, a NOESY-TROSY experi-ment with a mixing time of 100 ms was recorded using a samplethat contained 0.5 mM of 13C/15N/2H-labeled pGyrB and 1 mM ofinhibitor.

2.4. Effect of inhibitor on protein thermal stability

Thermal shift experiment was carried out on a Roche LC480 PCRmachine. Each assay well contained 10 lM pGyrB, 20� spyro

Y. Li et al. / FEBS Letters 589 (2015) 2683–2689 2685

orange. The assay buffer contained 50 mM HEPES, pH 7.2, 250 mMNaCl and 5 mM MgCl2.

2.5. Isothermal Titration Calorimetry (ITC) experiment

ITC experiment was performed on an Auto-iTC200 instrument(Microcal Inc.). The experiment was carried out at 25 �C. Proteinwas prepared in a buffer that contained 50 mM HEPES, pH 7.5,250 mM NaCl and 5 mM MgCl2 at concentrations of 100 lM.Inhibitor was prepared in the same buffer and loaded into the syr-inge automatically. Titration was carried out with 18 injectionsover a period of 40 min with stirring at 1000 rpm.

2.6. Protein relaxation analysis

The 15N longitudinal T1, and transverse T2 relaxation rates andbackbone 1H–15N-heteronuclear NOE (hetNOE) experiments [26]were collected at 298 K using a purified pGyrB sample in theabsence and presence of the inhibitor at a Bruker Avance700 MHz magnet. For T1 measurements, the relaxation delays of100, 300, 500, 1000, 1400, 1800, 2000, 2500 and 3000 ms wererecorded. For T2 measurements, the data were acquired withdelays of 16.9, 34, 51, 68, 85, 102, 119, 136 and 153 ms. ThehetNOE was obtained using two datasets that were collected withand without initial proton saturation for a period of 3 s. The col-lected spectra were then processed with NMRPipe [22] and ana-lyzed with NMRView [23].

Fig. 2. TSA and ITC analysis. (A) TSA of pGyrB in the absence and presence of theinhibitor. Effect of DMSO, different concentrations of inhibitor on the thermalstability of pGyrB is plotted. (B) ITC of the inhibitor against pGyrB. The bindingconstant was 54.6 nM.

3. Results

3.1. NMR spectra of pGyrB

Structural studies of GyrB from E. coli revealed that theN-terminal ATP binding domain contains eight b-strands backedon the side with several helices [27]. Free pGyrB exhibited well dis-persed cross peaks in 1H–15N-TROSY spectrum (Fig. 1A), which alsosuggested that it constrains b-strands. The OD280/OD260 of purifiedprotein was approximately 0.6, suggesting that pGyrB sample doesnot contain any nucleotides. Protein aggregation was observedwhen the sample was kept at room temperature for more than2 days. Although 3D NMR experiment data were collected for back-bone assignment, the data quality was not good enough to com-plete the assignment using the conventional strategy due to theweak signal in the HNCACB experiment (Fig. 1B). In the presenceof bis-pyridylurea, a potent inhibitor of both ParE and GyrB ofE. coli [21], chemical shift perturbation was observed, suggestingthat pGyrB binds to the inhibitor in solution (Fig. 1A). Theprotein-inhibitor complex was stable for more than 7 days andthe quality of the 3D spectra was improved (Fig. 1B), which madethe backbone assignment possible. Both thermal shift assay (TSA)and ITC were carried out to understand protein-inhibitor interac-tions. TSA showed that thermal shift (DTm) caused by inhibitorbinding was more than 9 �C (Fig. 2A). ITC result suggested thatthe binding affinity (KD) was 54.6 nM and reaction stoichiometry(n) was approximately 1 (Fig. 2B), which explained the interactionwas undergoing slow exchange observed in the NMR study (Fig. 1).

3.2. Backbone assignment of free pGyrB and complex

Backbone resonance assignment for the pGyrB-inhibitor com-plex was obtained using conventional 3D experiments. The assign-ments of the 1H–15N-TROSY spectra of pGyrB in the absence andpresence of the inhibitor are shown in Fig. 3A. Most of the back-bone amides and amide protons were assigned except M1, L100and V120. Other backbone resonance assignments including Ca

(218 of 222), Cb (190 of 198) and C0 (217 of 222) have beenobtained. The assignment of free pGyrB was achieved by referringto the assignment of the complex. The assignments of free andinhibitor-bound pGyrB have been deposited in the BiologicalMagnetic Resonance Bank (BMRB) with accession numbers 26597and 26598, respectively. Compared with the 1H–15N-TROSY spec-trum of free pGyrB, no extra peaks appeared and no line-broadening was observed in the 1H–15N-TROSY spectrum of thepGyrB-inhibitor complex because the interaction was undergoingslow exchange (Figs. 1A and 2).

3.3. Secondary structural analysis of pGyrB and the complex

The secondary structure analysis for pGyrB in the absence andpresence of the inhibitor was conducted using TALOS+ [24]. Bothforms showed similar secondary structural elements to E. coli

Fig. 3. Assignments and secondary structural analysis for pGyrB and its complex. (A and B) Assignment of the 1H–15N-TROSY spectra of pGyrB in the absence (B) and presenceof the inhibitor (A). Right panel is the enlarged region of the box in the left panel. Residue specific assignment of backbone 1H and 15N frequencies is shown with residue nameand sequence number. (C) Secondary structure analysis of pGyrB. Box indicates helical structures. Arrow indicates strands and line indicates loops. Secondary structuralelements for free pGyrB derived from NMR study and eGyrB derived from X-ray structures (PDB ids 1EI1 and 4PRX) are shown in black and red, respectively. The inhibitor-bound pGyrB has the same secondary structural elements as free pGyrB. The sequence alignment was conducted using ClustalW (http://www.ebi.ac.uk/Tools/msa/clustalw2/).The different residues between eGyrB and pGyrB are highlighted in red. (D) Homology model of pGyrB. A model was built using structure of GyrB of E. coli as a template. Leftpanel is the crystal structure of eGyrB (PDB id 1EI1). The ADPNP is show in sticks. Middle panel is surface representation of the eGyrB-ADP complex. The right panel is themodel of pGyrB.

2686 Y. Li et al. / FEBS Letters 589 (2015) 2683–2689

Fig. 4. Residues involved in protein and inhibitor interactions. (A) Plot of CSP as a function of residue number. (B) Difference of Ca chemical shifts in the absence and presenceof the inhibitor. The chemical shift dereference DCa was plotted against residue number. DCa = DCa (free) – DCa (inhibitor). (C) Mapping of affected residues on the pGyrBmodel. Residues with CSP more than 0.3 ppm, between 0.2 and 0.3 ppm and between 0.1 and 0.2 ppm are labeled in red, brown and yellow, respectively. The loop between a3and a4 is shown in blue. (D) Model of pGyrB and inhibitor complex. Left panel is the model of the pGyrB and inhibitor complex, which was based on the structure of ParE-inhibitor complex (PDB id 4LP0). Dashed lines indicate protein-inhibitor NOEs observed in the NOE experiment. The color code is similar to Fig. 4C. Residues withunambiguous NOEs with the inhibitor are highlighted with boxes. Middle panel is the structure of the inhibitor used in this study. Carbons with unambiguous assignmentsare labeled. Right panel contains select strips of the NOESY-TROSY spectrum. The resonances that may arise from incomplete deuteration or ambiguous assignments of theinhibitor are labeled with question marks. The NOESY-TROSY spectrum was collected using a 13C/15N/2H-labeled pGyrB and inhibitor at a molar ratio of 1:2.

Y. Li et al. / FEBS Letters 589 (2015) 2683–2689 2687

GyrB (Fig. 3C). There are eight b strands including b1 (residuesC58–I65), b2 (residues S70–N76), b3 (residues E131–R138), b4(residues K141–H148), b5 (residues L156–T162), b6 (residuesS165–F171), b7 (residues V202–D208) and b8 (residues K213–E220)and five a helices including a1 (residues L18–M27), a2 (residuesT36–A55) a3 (residues A92–T98), a4 (residues V122–L128) anda5 (residues W185–L198) present in the pGyrB. The pGyrB sharesvery high sequence homology (more than 75% sequence identity)with GyrB of the E. coli (eGyrB). The secondary structures ofpGyrB derived from NMR data are similar to X-ray structureof eGyrB, except that there is a short helix present at theN-terminus of eGyrB, and the lengths of a1, a2, b1, a4, b3, b5,b6, and b8 are slightly different (Fig. 3C). There are several residuesthat are different between these two proteins (Fig. 3C). Althoughthe difference did not alter the structure of pGyrB, it may affectinhibitor binding because some of different residues are at theinhibitor binding regions (Fig. 3C). A homology model of pGyrBwas built using the SWISS-MODEL server (Fig. 3D) using theX-ray structure of eGyrB bound with adenylyl-imidodiphosphate(ADPNP) as a template [2,28]. Long-range NOEs of residues in theb-strands of pGyrB were observed, which supports the homologymodel (data not shown).

3.4. The bis-pyridylurea inhibitor binds to the ATP binding pocket

To determine which residues were affected by inhibitor bindingto pGyrB, CSP caused by inhibitor binding was plotted against resi-due number (Fig. 4A). As the chemical shifts of amide and amideprotons are sensitive to the environment, residues showing signif-icant CSP might be involved in inhibitor binding. It was clear thatthose residues from the a2, b2, b6, the loop between b2 and a3,a3 and a4 were important for inhibitor interaction (Fig. 4A).Whether the inhibitor could cause structural changes on pGyrBwas investigated by analyzing the changes of the Ca chemicalshifts that are sensitive to the secondary structures. Although sev-eral residues showed changes in the Ca chemical shifts (Fig. 4B),the overall structure of pGyrB was not altered as analyzed byTALOS+ (Fig. 3). The residues from a2 including 44–52 are affectedsignificantly in the presence of the inhibitor, suggesting that theyare critical for inhibitor binding. This result may also explain thehigh quality HNCACB experiment obtained for the complexbecause the inhibitor can affect the chemical environments of Caand Cb carbons. Residues showing CSPs were mapped to thehomology model of pGyrB (Fig. 4C). Compared with the X-raystructure of the ParE-inhibitor complex, the inhibitor also binds

2688 Y. Li et al. / FEBS Letters 589 (2015) 2683–2689

to the ATP binding pocket of pGyrB (Fig. 4D). NOEs between resi-dues and inhibitor were also observed and the orientation of theinhibitor in pGyrB is similar to the one in ParE of Streptococcuspneumonia (S. pn).

3.5. Backbone relaxation analysis of pGyrB and its complex

Backbone relaxation data T1, T2 and hetNOE revealed thedynamic properties of both free and inhibitor bound pGyrB(Fig. 5). Compared with complex, fewer residues of free pGyrBwere used for analysis because the peaks are too weak to be accu-rately analyzed. Surprisingly, there is no significant changeobserved for these relaxation parameters in the absence and pres-ence of the inhibitor. The N-terminal 25 residues and theC-terminal 5 residues are flexible in the absence and presencethe inhibitor, which is characterized with low T1 and hetNOE valuesand high T2 values (Fig. 5). The average T1 values excluding bothN- and C-terminal residues and the loop between a3 and a4 forfree pGyrB and complex are 1.43 s and 1.48 s, respectively. Theaverage T2 values are 37.7 and 37.2 ms, respectively. The loop a3and a4 compassing residues 100–120 is flexible for both freepGyrB and complex, suggesting that the loop is not involved inthe molecular interaction with this inhibitor. The hetNOE valuesof other residues are higher than 0.82 that is expected for NHgroups in a grid globular protein [26], indicating that these resi-dues are rigid in solution. Further dynamic study in other timescales or protein side chain relaxation study will be helpful forunderstanding the effect of inhibitor on GyrB dynamics.

4. Discussion

Due to the bacterial resistance to antibiotics, there is a greatneed to develop novel antibacterial agents. The rate-limiting steps

Fig. 5. 15N relaxation parameters for pGyrB in free (s) and inhibitor bound forms (N). Unbe quantified. The relaxation experiments were collected using a sample that contabis-pyridylurea inhibitor.

in the antibacterial discovery process are twofold [29]. First, it isimportant to select a target that is not prone to resistance develop-ment and second, it is important to increase chemistry diversity toovercome the barriers to bacterial entry [29]. Bacterial type IItopoisomerases have been proven to be a good target for antibac-terial development due to their high sequence homology amongthe pathogenic bacteria and low homology with eukaryotes [5,7].Structure-based drug design has been an important tool in thedevelopment of these novel inhibitors such as tricyclic GyrB/ParEinhibitors and azaindole class of antibacterial agents [7,10,30].Understanding protein and inhibitor interaction is important indrug development. In this study, we carried out NMR studies onthe pGyrB. It is interesting that the assignment for the free pGyrBwas challenging due to the protein stability and low signal sensi-tivity (Fig. 1B), which might be the reason that there is no struc-tural information available for both ParE and GyrB fromP. aeruginosa, an extremely important Gram-negative strain withhigh pathogenicity. In the presence of the inhibitor that binds topGyrB with a KD of 54.6 nM, the protein stability was improved(Fig. 2A) and we obtained backbone resonance assignments forboth free and inhibitor bound forms of pGyrB (Fig. 3). The sec-ondary structural elements of pGyrB in solution were determinedbased on the assignment (Fig. 3). Although the inhibitor binds topGyrB with an affinity in nanomolar range, there was no significantsecondary structural change observed for pGyrB upon inhibitorbinding. Further relaxation results also demonstrated that thebackbone dynamic of pGyrB was not changed dramatically(Fig. 5). It has been noted that our 3D spectra and relaxation datasuggest that the side chain dynamics of residues in the ATP bindingpocket might be important for ligand binding. In the presence ofthe inhibitor, the backbone amide dynamics of pGyrB were notaffected significantly, while the side chain dynamics was influ-enced (Figs. 1B and 2A). These results imply that careful protein

analyzed residues include prolines and the ones that are overlapped or too weak toined 0.6 mM 13C/15N/2H-labeled pGyrB in the absence and presence of 1.2 mM

Y. Li et al. / FEBS Letters 589 (2015) 2683–2689 2689

and ligand interaction need to be studied in drug developmentbecause the binding affinities between the inhibitor and eGyrB/ParE might be different from pGyrB due to the difference of resi-dues in the ATP binding pocket. Our study on the ParE of S. pnshowed that swapping a single residue with a corresponding resi-due in P. aeruginosa in the ATP binding pocket can affected inhibi-tor binding affinity (Kang et al., unpublished data). Our study alsoconfirmed that the loop between a3 and a4 of pGyrB was notinvolved in inhibitor binding because there were no CSP observedand dynamic changes were minor in the presence of the inhibitor(Fig. 5).

The inhibitor used in this study is a pyridylurea scaffold derivedfrom fragment-based drug design and is ATP competitive [21]. Itsactivity against gram-negative strains such as E. coli and gram-positive strains was investigated in detail [21]. X-ray crystal struc-ture revealed that D78 of b2 and T172 of b6 from ParE of S. pnformed hydrogen bonds with the inhibitor. R81 and M83 fromS. pn ParE from the loop between b2 and a3 were shown to interactwith the inhibitor [21]. We carried out biophysical characterizationfor the molecular interaction between pGyrB and the inhibitor(Figs. 1 and 2). Based on the backbone resonance assignment, theCSP caused by inhibitor binding was investigated in this study.Residues from b2 (D75), b6 (E168 and V169), loop between b2and a3, and a3 were shown to be important for inhibitor binding(Fig. 4). Residues exhibited CSP with more than 0.2 ppm upon inhi-bitor binding were localized at the ATP binding pocket (Fig. 4).NOEs between pGyrB and the inhibitor were observed (Fig. 4),which further confirmed the residues that are important for inhibi-tor binding. Our results provide direct evidence to show that thebis-pyridylurea inhibitor binds to the ATP binding pocket ofpGyrB, which is similar to the ParE of S. pn. These results will beuseful to understand protein and inhibitor interactions, which willbe helpful in antibacterial drug development. Although it will beuseful to carry out further studies to understand protein dynamicin other time scales, this study provides an example to show thatinhibitors or ligands can facilitate structural studies of proteinsby improving spectral quality.

In summary, we purified pGyrB and conducted structural stud-ies and its interaction with a bis-pyridylurea inhibitor. This inhibi-tor binds to pGyrB with a KD of 54.6 nM and could improve itsthermal stability. Secondary structures of pGyrB were definedand inhibitor did not cause significant conformational changes.CSP and NOE analysis demonstrate that the inhibitor binds to theATP binding pocket.

Acknowledgments

We appreciate support from A*STAR JCO Grants (1331A028,1231B015). We also thank Prof Ho Sup Yoon and Dr. Hong Ye fromNanyang Technological University for the NMR experiments. Theauthors appreciate valuable discussion from members of the drugdiscovery team in Experimental Therapeutics Centre andAstraZeneca.

References

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

[2] Bellon, S. et al. (2004) Crystal structures of Escherichia coli topoisomerase IVParE subunit (24 and 43 kilodaltons): a single residue dictates differences innovobiocin potency against topoisomerase IV and DNA gyrase. Antimicrob.Agents Chemother. 48, 1856–1864.

[3] Champoux, J.J. (2001) DNA topoisomerases: structure, function, andmechanism. Annu. Rev. Biochem. 70, 369–413.

[4] Reece, R.J. and Maxwell, A. (1991) DNA gyrase: structure and function. Crit.Rev. Biochem. Mol. Biol. 26, 335–375.

[5] Collin, F., Karkare, S. and Maxwell, A. (2011) Exploiting bacterial DNA gyrase asa drug target: current state and perspectives. Appl. Microbiol. Biotechnol. 92,479–497.

[6] Mitscher, L.A. (2005) Bacterial topoisomerase inhibitors: quinolone andpyridone antibacterial agents. Chem. Rev. 105, 559–592.

[7] Manchester, J.I. et al. (2012) Discovery of a novel azaindole class ofantibacterial agents targeting the ATPase domains of DNA gyrase andTopoisomerase IV. Bioorg. Med. Chem. Lett. 22, 5150–5156.

[8] Brvar, M., Perdih, A., Oblak, M., Masic, L.P. and Solmajer, T. (2010) In silicodiscovery of 2-amino-4-(2,4-dihydroxyphenyl)thiazoles as novel inhibitors ofDNA gyrase B. Bioorg. Med. Chem. Lett. 20, 958–962.

[9] Brvar, M., Perdih, A., Renko, M., Anderluh, G., Turk, D. and Solmajer, T. (2012)Structure-based discovery of substituted 4,50-bithiazoles as novel DNA gyraseinhibitors. J. Med. Chem. 55, 6413–6426.

[10] Tari, L.W. et al. (2013) Tricyclic GyrB/ParE (TriBE) inhibitors: a new class ofbroad-spectrum dual-targeting antibacterial agents. PLoS ONE 8, e84409.

[11] Tari, L.W. et al. (2013) Pyrrolopyrimidine inhibitors of DNA gyrase B (GyrB)and topoisomerase IV (ParE). Part I: Structure guided discovery andoptimization of dual targeting agents with potent, broad-spectrumenzymatic activity. Bioorg. Med. Chem. Lett. 23, 1529–1536.

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

[13] Lafitte, D. et al. (2002) DNA gyrase interaction with coumarin-basedinhibitors: the role of the hydroxybenzoate isopentenyl moiety and the 50-methyl group of the noviose. Biochemistry 41, 7217–7223.

[14] Shuker, S.B., Hajduk, P.J., Meadows, R.P. and Fesik, S.W. (1996) Discoveringhigh-affinity ligands for proteins: SAR by NMR. Science 274, 1531–1534.

[15] Klaus, W., Ross, A., Gsell, B. and Senn, H. (2000) Backbone resonanceassignment of the N-terminal 24 kDa fragment of the gyrase B subunit fromS. aureus complexed with novobiocin. J. Biomol. NMR 16, 357–358.

[16] Bellanda, M., Peggion, E., Otting, G., Weigelt, J., Perdona, E., Domenici, E.,Marchioro, C. and Mammi, S. (2002) Backbone 1H, 13C and 15N resonanceassignment of the N-terminal 24 kDa fragment of the gyrase B subunit fromE. coli. J. Biomol. NMR 22, 369–370.

[17] Kim, Y.M. et al. (2013) NMR analysis of a novel enzymatically active unlinkeddengue NS2B–NS3 protease complex. J. Biol. Chem. 288, 12891–12900.

[18] Li, Q., Ng, H.Q., Yoon, H.S. and Kang, C. (2014) Solution structure of the cyclic-nucleotide binding homology domain of a KCNH channel. J. Struct. Biol. 186,68–74.

[19] Pervushin, K., Ono, A., Fernandez, C., Szyperski, T., Kainosho, M. and Wuthrich,K. (1998) NMR scalar couplings across Watson-Crick base pair hydrogenbonds in DNA observed by transverse relaxation-optimized spectroscopy.Proc. Natl. Acad. Sci. U.S.A. 95, 14147–14151.

[20] Salzmann, M., Pervushin, K., Wider, G., Senn, H. and Wuthrich, K. (1998)TROSY in triple-resonance experiments: new perspectives for sequential NMRassignment of large proteins. Proc. Natl. Acad. Sci. U.S.A. 95, 13585–13590.

[21] Basarab, G.S. et al. (2013) Fragment-to-hit-to-lead discovery of a novelpyridylurea scaffold of ATP competitive dual targeting type II topoisomeraseinhibiting antibacterial agents. J. Med. Chem. 56, 8712–8735.

[22] Delaglio, F., Grzesiek, S., Vuister, G.W., Zhu, G., Pfeifer, J. and Bax, A. (1995)NMRPipe: a multidimensional spectral processing system based on UNIXpipes. J. Biomol. NMR 6, 277–293.

[23] Johnson, B.A. (2004) Using NMRView to visualize and analyze the NMR spectraof macromolecules. Methods Mol. Biol. 278, 313–352.

[24] Shen, Y., Delaglio, F., Cornilescu, G. and Bax, A. (2009) TALOS+: a hybridmethod for predicting protein backbone torsion angles from NMR chemicalshifts. J. Biomol. NMR 44, 213–223.

[25] Williamson, M.P. (2013) Using chemical shift perturbation to characteriseligand binding. Prog. Nucl. Magn. Reson. Spectrosc. 73, 1–16.

[26] Kay, L.E., Torchia, D.A. and Bax, A. (1989) Backbone dynamics of proteins asstudied by 15N inverse detected heteronuclear NMR spectroscopy:application to staphylococcal nuclease. Biochemistry 28, 8972–8979.

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

[28] Biasini, M. et al. (2014) SWISS-MODEL: modelling protein tertiary andquaternary structure using evolutionary information. Nucleic Acids Res. 42,W252–8.

[29] Silver, L.L. (2011) Challenges of antibacterial discovery. Clin. Microbiol. Rev.24, 71–109.

[30] Mitton-Fry, M.J. et al. (2013) Novel quinoline derivatives as inhibitors ofbacterial DNA gyrase and topoisomerase IV. Bioorg. Med. Chem. Lett. 23,2955–2961.