Structures of Kibdelomycin Bound to Staphylococcus aureus GyrB and

Jun 26, 2014 - Bacterial resistance to antibiotics continues to pose serious challenges as the discovery rate for new antibiotics fades. Kibdelomycin ...
0 downloads 0 Views 2MB Size
Download PDF
Articles pubs.acs.org/acschemicalbiology

Structures of Kibdelomycin Bound to Staphylococcus aureus GyrB and ParE Showed a Novel U‑Shaped Binding Mode Jun Lu,*,† Sangita Patel,† Nandini Sharma,† Stephen M. Soisson,† Ryuta Kishii,‡ Masaya Takei,‡ Yasumichi Fukuda,‡ Kevin J. Lumb,† and Sheo B. Singh*,§,∥ †

Merck Research Laboratories, West Point, Pennsylvania 19486, United States Kyorin Pharmaceutical Co., Ltd., 2399-1, Nogi, Nogi-machi, Shimotsuga-gun, Tochigi 329-0114, Japan § Merck Research Laboratories, Kenilworth, New Jersey 07033, United States ‡

S Supporting Information *

ABSTRACT: Bacterial resistance to antibiotics continues to pose serious challenges as the discovery rate for new antibiotics fades. Kibdelomycin is one of the rare, novel, natural product antibiotics discovered recently that inhibits the bacterial DNA synthesis enzymes gyrase and topoisomerase IV. It is a broad-spectrum, Gram-positive antibiotic without cross-resistance to known gyrase inhibitors, including clinically effective quinolones. To understand its mechanism of action, binding mode, and lack of cross-resistance, we have cocrystallized kibdelomycin and novobiocin with the N-terminal domains of Staphylococcus aureus gyrase B (24 kDa) and topo IV (ParE, 24 and 43 kDa). Kibdelomycin shows a unique “dual-arm”, U-shaped binding mode in both crystal structures. The pyrrolamide moiety in the lower part of kibdelomycin penetrates deeply into the ATP-binding site pocket, whereas the isopropyltetramic acid and sugar moiety of the upper part thoroughly engage in polar interactions with a surface patch of the protein. The isoproramic acid (1,3-dioxopyrrolidine) and a tetrahydropyran acetate group (Sugar A) make polar contact with a surface area consisting of helix α4 and the flexible loop connecting helices α3 and α4. The two arms are connected together by a rigid decalin linker that makes van del Waals contacts with the protein backbone. This “dual-arm”, U-shaped, multicontact binding mode of kibdelomycin is unique and distinctively different from binding modes of other known gyrase inhibitors (e.g., coumarins and quinolones), which explains its lack of cross-resistance and low frequency of resistance. The crystal structures reported in this paper should enable design and discovery of analogues with better properties and antibacterial spectrum.

B

modes of action that could be developed as effective treatment options against drug-resistant bacteria. We recently reported the discovery of a series of natural product antibiotics with novel modes of action by application of antisense-based screening technology yielding platensimycin,4,5 platencin,6,7 and most recently kibdelomycin8 and kibdelomycin A (Figure 1).9 The broad spectrum (Supplementary Table S1), Gram-positive, natural product antibiotic kibdelomycin was reported in 2011 and is isolated from a Kibdelosporangium sp. Kibdelomycin exerts its activity by inhibiting bacterial DNA synthesis through specific inhibition of gyrase B (GyrB) and topoisomerase IV (ParE), two highly similar but functionally different enzymes.10 Kibdelomycin exhibits a low frequency of resistance and shows no cross-resistance to Staphylococcus aureus strains resistant to other known gyrase inhibitors, such as novobiocin, coumermycin, and quinolones, suggesting a novel binding mode.8 Kibdelomycin has been shown to be a potent

acterial strains continue to select for resistance against antibiotics in clinical practice with a frightening frequency, posing a serious threat to human lives. Klevens et al.1 recently reported that methicillin-resistant Staphylococcus aureus (MRSA) infections alone are responsible for about 18,000 deaths per year in the United States. The dearth of new antibiotics due to the lack of structurally novel chemical leads that can be developed into new clinically useful antibiotics make this problem dire. The chemical structure leads amenable for delivering antibiotic drugs were all discovered over five decades ago, with the exception of linezolid and daptomycin that were approved after 2000. Most of the chemical leads originated from nature. Relatively small chemical modifications of the antibiotic lead structures discovered decades ago have led to incrementally improved antibiotics that have served well and continue to provide the current reservoir of clinical antibiotics.2,3 However, the capacities of such modifications are not limitless, and additional improvements on existing chemical scaffolds are proving increasingly challenging, particularly for improving the resistant profile within the class. The dearth of new antibiotics can be overcome only by discovery of new antibiotic scaffolds, with either known or novel © 2014 American Chemical Society

Received: February 17, 2014 Accepted: June 26, 2014 Published: June 26, 2014 2023

dx.doi.org/10.1021/cb5001197 | ACS Chem. Biol. 2014, 9, 2023−2031

ACS Chemical Biology

Articles

inhibitor of the corresponding topoisomerase IV activities (E. coli, IC50 29000 nM; S. aureus, IC50 500 nM). Kibdelomycin potently inhibited catalytic E. coli ATPase activity of gyrase B (IC50 11 nM) and topoisomerase IV (ParE) (IC50 900 nM).8 Kibdelomycin A is a less potent inhibitor of S. aureus gyrase (IC50 400 nM) and topoisomerase IV (ParE, IC50 5000 nM) but has been shown to be a potent inhibitor of E. coli gyrase B ATPase activity (IC50 9 nM) and poor inhibitor of the E. coli ParE (IC50 6400 nM).9 In order to understand the binding mode of kibdelomycin (KBD), the co-crystal structures in complex with both the 24 kDa GyrB and 43 kDa ParE proteins from S. aureus were determined. We also determined the crystal structures of novobiocin bound to the 24 kDa GyrB and 24 kDa ParE proteins from S. aureus for comparison and to understand the reason for lack of cross-resistance. These structures reveal a novel binding mode for kibdelomycin, unlike the binding mode of known gyrase inhibitors (e.g., coumarins and quinolones) and thus help explain the potent activity against resistant strains and lack of cross-resistance to key gyrase inhibitors.

■

RESULTS AND DISCUSSION GyrB- and ParE-Kibdelomycin Structural Analysis. We have crystallized and determined the structures of the N-terminal domain of S. aureus ParE (ParE 43 kDa) complexed with kibdelomycin (Figure 2, Table 1) and the N-terminal domain of Gyrase B (GyrB 24 kDa) complexed with KBD (Figure 3, Table 1) at 2.3 and 2.9 Å resolution, respectively. The KBD binding sites on GyrB and ParE are located in domain I (GyrB 2−220,

Figure 1. Chemical structures and three binding arms of kibdelomycin, kibdelomycin A, and kibdelomycin acetate.

inhibitor of Escherichia coli (IC50 60 nM, Supplementary Table S2) and S. aureus (IC50 9 nM) gyrase activities and less potent

Table 1. Crystallographic Data Collection and Refinement Statistics ParE43:KBD

GyrB24:KBD

ParE24:NOV

GyrB24:NOV

Data Collection space group unit cell parameters (Å)

unit cell parameters (deg)

molecules per asymmetric unit resolution range (Å) no. of unique observations completeness (%) multiplicity Rsymb (%) av I/σI (%) resolution range (Å) used reflections Rworkc Rfreed no. of atoms protein waters av B-factors (Å2) bond length rmsd (Å) bond angle rmsd (deg)

P21 a = 54.45 b = 136.81 c = 69.19 α = 90 β = 111.9 γ = 90 2 50.0−2.30 (2.34−2.30)a 41758 99.9(100.0) 4.0 6.4(56.8) 20.3(2.7)

C2 a = 115.68 b = 75.32 c = 76.14 α = 90 β = 92.85 γ = 90 3 50−2.30 (2.35−2.30) 28285 97.0 (98.4) 4.1 8.9 (31.8) 10.3(3.5)

P21 a = 74.28 b = 68.59 c = 85.72 α = 90 β = 91.72 γ = 90 4 74.24−2.59 (2.89−2.59) 26992 99.6 (99.9) 3.7 8.8 (35.1) 9.9(3.3)

49.7−2.29 (2.35−2.29) 41708 0.217 0.252

P21 a = 69.28 b = 75.55 c = 90.46 α = 90 β = 90.04 γ = 90 4 90.45−2.93 (3.09−2.93) 20139 99.7(100.0) 3.7 8.9 (41.9) 11.4(2.7) Refinement Statistics 30.44−2.94 (3.10−2.94) 20004 0.195 0.256

19.47−2.30 (2.39−2.30) 28173 0.222 0.267

74.24−2.59 (2.89−2.59) 26939 0.195 0.246

5514 106 52.27 0.010 1.21

5878 18 66.18 0.010 1.25

4213 50 48.81 0.010 1.23

5733 68 53.44 0.010 1.17

Numbers in parentheses are for the highest resolution shell. bRsym = Σ[|Ii − ⟨Ii⟩|]/Σ[⟨Ii⟩]. cRwork = Σ∥Fobs| − |Fcalc∥/Σ|Fobs|. dRfree is equivalent to Rwork but calculated with a randomly selected subset (5.0%) of the data not used in the refinement. a

2024

dx.doi.org/10.1021/cb5001197 | ACS Chem. Biol. 2014, 9, 2023−2031

ACS Chemical Biology

Articles

Figure 2. Co-crystal structure of the S. aureus ParE-KBD complex: (A) The overall structure, showing a surface representation of monomer A and a ribbon representation of monomer B. Domain I of monomer A is colored in green and domain II in wheat; the domain I and domain II of monomer B are colored in cyan and light blue respectively; (B) The KBD binding site with electron density map of KBD, (C) The hydrophobic interactions of KBD in the lower binding site, (D) The polar interactions of KBD in the lower binding site, (E) The upper binding site. In Figure B-E, KBD is colored in yellow and ParE in green. The red spheres in Panel B-E are the structured water molecules involved in polar interactions with KBD complexes.

ParE 2−224) that consist of an eight-stranded β-sheet and five αhelices. Domain IIs of GyrB and ParE consist of four-stranded βsheet and four α-helices. The overall binding mode of KBD to GyrB is very similar to that of KBD bound to ParE (Figure 4). KBD is located in the 24 kDa N-terminal part of ParE43 (domain I). The binding mode of KBD features an extended “dual-arm”, U-shaped conformation that clamps on domain I (Figures 2 and 3). Most of the contacts between KBD and ParE are through the two stretched arms. The lower arm of the Ushaped KBD consists of a pyrrolamide moiety and dihydroxytetrahydropyran (Sugar B, Figure 1), which penetrates into the well-known ATP-binding site. The remaining part of KBD protrudes from this lower binding pocket (Figures 2A and 2B) and wraps around domain I. The upper arm of KBD includes the isopropyl-tetramic acid (1,3-dioxopyrrolidine) and a tetrahydropyran acetate group (Sugar A, Figure 1) that makes contact with a surface area consisting of helix α4 and the flexible loop connecting helices α3 and α4 (termed the upper binding site hereafter, Figures 2B and 3A). The two arms are connected together by a rigid decalin linker that makes VDW contacts with the protein (Figures 2E and 3D). The lower binding pocket partially overlaps with the ATPbinding site, where the pyrrolamide group occupies the same pocket occupied by the adenine moiety of ATP. The pyrrole nitrogen is hydrogen-bonded to the conserved D76 (or D81 in GyrB), and the carbonyl substituent of the pyrrole moiety is within hydrogen bond distance of a conserved water molecule that is positioned as a hub of a hydrogen-bond network involving

the D76 side chain, G80 backbone amide, and T168 side chain (Figure 2C). The 3,4-dichloro substituents on the pyrrole ring provided increased hydrophobic interaction in the pocket, which are surrounded by hydrophobic residues (M81, I96, F97, V170, A122, and I46) (Figure 2C). Interestingly, the pyrrolamide has also been identified as an important scaffold through a fragmentbased, lead generation approach.11 The dihydroxy-tetrahydropyran (Sugar B) moiety extends outside of the ATP pocket and is hydrogen-bonded to the E53-R79 salt bridge, together with another conserved water molecule, which is further hydrogenbonded to the main chain carbonyls of R138 and G80 (Figure 2D). In short, the lower arm is anchored to the lower binding site by three pairs of hydrogen bonds between KBD and the floor of the lower binding site. The bis-chloropyrrole warhead has numerous hydrophobic interactions with surrounding residues. The key binding moiety of the upper arm of KBD in GyraseB structure is the isopropyl-tetramic acid group, which is clenched by Q91 (H86 for ParE) on one side (two pairs of H-bonds) and G109 (G103 for ParE) on the other side, through hydrogen bonding (Figures 2E, 3D and 4A). The highly substituted sugar (Sugar A, Figure 1) moiety sits on top of a flexible loop connecting helices α3 and α4. In both the GyrB and ParE ATPase structures with KBD, the loop (composed of residues 99−120) is highly dynamic, and its conformation is liganddependent. Part of the loop, residues 104−117 in GyrB (residue 112−123 in ParE) in our structure, is missing from the electron density map. Residue 99−102 in GyrB (105−108 in ParE) is 2025

dx.doi.org/10.1021/cb5001197 | ACS Chem. Biol. 2014, 9, 2023−2031

ACS Chemical Biology

Articles

Figure 3. Co-crystal structure of S. aureus GyrB-KBD complex: (A) The overall binding mode of KBD with electron density map of KBD, (B) The polar interaction of KBD in the lower binding site, (C) The hydrophobic interaction of KBD in the lower binding site, (D) The upper binding site. KBD is colored in yellow and GyrB in green. The red spheres in Panel A-D are the structured water molecules involved in polar interactions with KBD complexes.

Figure 4. Comparisons of S. aureus ParE-KBD and GyrB-KDB structures (colored in cyan:purple and green:yellow, respectively). (A) The overall binding mode; (B) The upper binding site; (C) The lower binding site.

stabilized through the interaction with KBD and forms a Type II β-turn. The structure of 24 kDa GyrB, in complex with KBD, was determined at 2.9 Å. The overall binding mode of KBD in the

GyrB complex is very similar to that of the KBD-43 kDa ParE complex, with a slight shift in the upper arm (Figures 3 and 4). In the lower binding pocket, all of the polar residues in contact with KBD are identical in both GyrB and ParE, such as D81, E58, and 2026

dx.doi.org/10.1021/cb5001197 | ACS Chem. Biol. 2014, 9, 2023−2031

ACS Chemical Biology

Articles

Figure 5. Co-crystal structures of (A) S. aureus gyraseB 24 kDa in complex with novobiocin and (B) S. aureus ParE 24 kDa in complex with novobiocin. (C) Six GyrB:novobiocin or ParE:novobiocin structures superimposed together. Novobiocin bound to S. aureus ParE is colored in yellow, S. aureus GyrB in gray, E. coli GyrB in cyan (PDB: 1AJ6), E. coli ParE in blue (PDB: 1S14), T. thermophiles GyrB in magnenta (PDB: 1KIJ), and Xanthomonas Oryzae pv oryza ParE in purple (PDB: 3LPS). The red spheres are the structured water molecules involved in polar interactions with novobiocin complexes.

five earlier structures. These structures show that the hydroxyl benzoate isopentyl group wraps around a proline residue and folds back away from the solvent onto the coumarin ring.12 In contrast, the same structural motif in S. aureus ParE complex exhibits an extended conformation with the benzoate moiety in a trans conformation (Figure 5B and C). The biological relevance of the binding mode remains to be elucidated, and it is possible that crystal packing might play a role in the observed binding mode. It seems less likely that novobiocin could adopt the same conformation when bound to GyrB, since D89 and Q91 of GyrB form strong hydrogen bonds with the phenolic group of the novobiocin (Figure 5A). ParE contains a G84 at the corresponding location and hence lacks this polar interaction (Figure 5B). This missing interaction might allow the benzoate group to swing out into the solvent to form the extended conformation observed in our S. aureus ParE:novobiocin structure. In the E. coli ParE:novobiocin structure (which has an asparagine at the equivalent position), the benzoate group still takes the same folded-in conformation as seen in GyrB. It is highly conceivable that in the ParE structure both conformations could be accommodated due to the lack of the anchoring interaction with residue D89.

R84. The pyrrolamide moiety of KBD, bound to 24 kDa GyrB, is also surrounded by hydrophobic residues (I86, I102, L103, I175, and I51). Most of these residues are identical except for three residues: I86, L103, and S128 in GyrB (M81, F97, and A122 in ParE). The hydrophilic residue in the pocket of GyrB is S128 (A122 in ParE). In the upper binding pocket, the prolyl-tetramic acid group is situated between the side chain of Q91 and G109. GyrB- and ParE-Novobiocin Structural Analysis. Although several crystal structures of novobiocin (novo), in complex with bacterial gyrase B and topoisomerase IV enzymes, have been previously reported, there have been no reported structures of novobiocin bound to S. aureus GyrB or ParE. In order to explain the activity of kibdelomycin against the novobiocin-resistant strain of S. aureus, we crystallized and determined the structures of novobiocin in complex with 24 kDa GyrB and ParE at 2.6 and 2.3 Å respectively. The GyrB:novobiocin structure resembles that of E. coli GyrB:novobiocin and four other published novobiocin complex structures (PDB codes: 1AJ6, 1KIJ, 1S14, and 3LPS). Interestingly, the ParE:novobiocin structure displays an unique binding mode that is different from the other novobiocin complex structures (Figure 5, Table 1). In the ParE complex, the coumarin and noviose moieties of novobiocin superimpose relatively well with the other 2027

dx.doi.org/10.1021/cb5001197 | ACS Chem. Biol. 2014, 9, 2023−2031

ACS Chemical Biology

Articles

Figure 6. (A) Comparisons of KBD with ADPNPs (PDB codes: 1EI1, 1S16) and 36 known small-molecule inhibitors including novobiocin (PDB codes: 1AJ6, 1KIJ, 1KZN, 1S14, 3FV5, 3G7B, 3G7E, 3G75, 3TTZ, 3U2D, 3U2K, 4BAE, 4B6C, 4DUH, 4EM7, 4EMV, 4GEE, 4GFN, 4GGL, 4HXW, 4HXZ, 4HYP, 4HY1, 4HZ0, 4HZ5, 4KFG, 4KQV, 4KSG, 4KSH, 4KTN, 4K4O, 4LPB, 4LP0, 4MBC, 4MB9, 4MOT). (B) Superposition of domain I of the monomer A of the S. aureus ParE:KBD complex on domain I of the T. thermophilus GyrB:novobiocin complex (in blue ribbons). The monomer A of S. aureus ParE:KBD is colored in green (domain I) and wheat (domain II). The monomer B of S. aureus ParE:KBD is colored in cyan. (C) Close-up view of Figure B in the KBD binding region in monomer A. If S. aureus ParE:KBD dimerized in the same manner as the T. thermophilus GyrB:novobiocin complex, the upper arm of KBD will collide with helix 1 of monomer B.

Upper Binding Site and Its Impact on GyrB/ParE Dimerization. This “dual-arm”, U-shaped, multicontact binding mode of kibdelomycin is unique and distinctly different from binding modes of other known GyrB/ParE inhibitors. Figure 6A shows the superpositions of S. aureus GyrB:KBD with 38 published structures of GyrB or ParE in complex with small molecule inhibitors including novobiocin and ATP substrate analogue, 5′-adenylyl-β,γ-imidodiphosphate (ADPNP). It is quite evident that none of the other small molecule inhibitors is able to reach out to the novel KBD upper binding site, including other known natural products such as novobiocin and chlorobiocin (Figure 6A). The upper arm of KBD also plays a role in destabilizing GyraseB/ParE dimerization. In the presence of ADPNP, the 43 kDa Gyrase B or ParE dimerized through the N-terminal arm

(residue 2−15 of Gyrase B), which protrudes from the surface of one monomer and wraps around domain I of the other monomer.13−15 Tyr5 on the N-terminal arm of one monomer makes contact with ADPNP of the other molecule to further stabilize the dimer.13 The 43 kDa S. aureus ParE:KDB complex crystallized as a dimer, with the major dimerization interface lying at the C-terminal helices between the two monomers. In contrast to all known 43 kDa GyrB or ParE dimer structures, complexed with either ATP analogues or novobiocin, the intermolecular contacts between domain I of the S. aureus dimer presented here are significantly reduced.15 This cannot be solely explained by the displacement of the tyrosine (Tyr5 in E. coli GyrB) of the N-terminal tail of monomer B out of the ATP pocket in monomer A, through KBD binding. The T. Thermophilus GyrB:novobiocin complex was also crystallized as 2028

dx.doi.org/10.1021/cb5001197 | ACS Chem. Biol. 2014, 9, 2023−2031

ACS Chemical Biology

Articles

inhibitory activities. The C-33 hydroxy group showed a strong hydrogen bond with E58. The acetylation of the hydroxy group to kibdelomycin acetate led to significant loss of cellular activity (MIC 16 μg/mL),9 confirming the importance of the hydrogen bond for the binding. The strong, bidentate hydrogen bond/salt bridge interactions of H86 (ParE) and Q91 (GyrB), with the tetramic acid moiety of the upper arm, give rise to greater interaction energy. The hydrogen bond with G103 (GyrB) (G109 in ParE) helps further anchor this structural motif in place with increased binding efficiency and likely improved potency. It is unclear how much binding energy is gained by stabilization of the type II β-turn from the interactions with the acetylated sugar A. Kibdelomycin exhibits a 55-fold lower ParE (IC50 500 nM) activity compared to GyrB (IC50 9 nM) from S. aureus enzymes, despite remarkably similar structures of the two enzymes. More pronounced activity differences were observed for E. coli ParE and GyrB enzymes.8 These differences were also observed for kibdelomycin A.9 Similar ParE and GyrB activity differences were also observed for novobiocin, which has been attributed to a single amino acid substitution (M74 in ParE versus I78 in GyrB) in the E. coli enzymes.8 This observation was confirmed by amino acid substitutions in each enzyme, leading to gain and/or loss of potencies.15 S. aureus ParE contains M81, and GyrB contains I86 at the corresponding location (Figure 4C). These two amino acids are located around the pyrrolamide binding site (Figure 4C) and are likely responsible for such differences in the potency for the two enzymes from both organisms. While the presence of the methyl group in kibdelomycin showed better potency versus a desmethyl analogue of kibdelomycin A against S. aureus GyrB, no activity difference was observed against E. coli GyrB. The amino acids surrounding the pyrrole moiety differ between the S. aureus and E. coli GyrB enzymes. The pyrrole methyl group penetrates into a hydrophobic pocket formed by I51, I175, and V79 in S. aureus GyrB (Figure 3C and D, also Supplementary Figure S2A and B), whereas the corresponding hydrophobic pocket contains V43, V167, and V71 in E. coli GyrB (Supplementary Figure S2A and B). The comparison of the methyl group-binding site of kibdelomycin in S. aureus GyrB with the modeled structure of KBD with E. coli GyrB shows that the ethyl group of I175 is in contact with the methyl group in S. aureus GyrB enzyme, whereas it is missing in the E. coli enzyme containing V167. The differences in the amino acids forming hydrophobic pockets surrounding the methyl group and contacts provided in one enzyme and not in the other may be sufficient to explain the identical potencies of kibdelomycin and kibdelomycin A in S. aureus and E. coli GyrB (Supplementary Figure S2A and B). Finally, the structural results presented here explain why crossresistance with other known gyrase inhibitors is not observed. The novobiocin-resistant S. aureus strain contains a mutation of D89G in GyrB. This aspartate forms a strong hydrogen bond with the phenolic group of the novobiocin in S. aureus GyrB (Figures 5A, 7A). This residue does not engage in interaction with kibdelomycin and appears to plays no role in KBD activity, and therefore it is not surprising that this mutation (D89G) has no effect on the potency of KBD and hence conferred no crossresistance (Figure 7A and Supplementary Figure S2). ParE contains a glycine, G84, at the corresponding location; hence it has no significance in the binding and activity of novobiocin. While it was relatively easy to explain the lack of cross-resistance of kibdelomycin to novoR strains using the novobiocin crystal structure bound to S. aureus GyrB (Figure 5A), explaining the

a dimer (PDB code: 1KIJ), and the presence of novobiocin prevents Tyr5 from closing the ATP-binding pocket (Figure 6C).12,16 The dimeric interface of the T. Thermophilus GyrB:novobiocin complex exhibits some degree of rearrangements but still resembles that of the E. coli GyrB:ANPPNP structure (PDB: 1EIJ), indicating that displacement of Tyr5 does not disrupt the dimer interface. However, the dimeric interface of S. aureus ParE:KBD is significantly different from those of the T. thermophilus GyrB:novobiocin complex and E. coli GyrB:ADPNP complex. Superposition of domain I of monomer A of the S. aureus ParE:KBD complex on domain I of the T. thermophilus GyrB:novobiocin complex (PDB: 1KIJ) shows a large movement of domain I of monomer B relative to that of monomer A (Figure 6B and C). It is reasonable to assume that if S. aureus ParE:KBD dimerized in the same manner as the T. thermophilus GyrB:novobiocin complex, the upper arm of KBD will collide with helix α1 of monomer B. Therefore, KBD binding not only blocks the tyrosine 5 side chain (using E. coli GyrB numbering system) from entering into the ATP pocket (similar to novobiocin) but also causes significant perturbation in the dimerization interface between the two domain I of the ParE dimer, which further inhibits the ATPase activity. KBD Binding Mode Explains Lack of Cross-Resistance. Kibdelomycin did not show cross-resistance to other known gyrase inhibitors (Supplementary Table S3), regardless of their mode of binding. This includes ATPase inhibitors, such as novobiocin and coumermycin, and DNA cleavage/resealing gyrase A/ParC inhibitors, such as levofloxacin and other quinolones.8 KBD showed no MIC change between wild type and novobiocin-resistant (harboring single D89G mutation in GyrB) S. aureus strains. It showed moderate (4-fold, Supplementary Table S3) MIC shift against coumermycin resistant S. aureus harboring three mutations in GyrB (Q136E, I175T, L455I, Supplementary Figure S1). It did not show an MIC shift against highly quinolone resistant S. aureus.8 Kibdelomycin consists of five substructural fragments with alternating hydrophobic and hydrophilic moieties. Each of these substructures are fully utilized in the unique “dual arm” binding mode that results in potent binding and inhibitory activity against GyrB and ParE. The crystal structures of kibdelomycin bound to S. aureus GyrB and ParE resolve the molecular details of the Ushaped binding mode that consists of strong polar interactions by the upper arm and both polar and hydrophobic interactions by the lower arm moieties. The hydrophobic linker connecting the upper and lower arms is stabilized by van der Waals (VDW) interactions to the protein. More specifically, the pyrrolamide moiety of the lower arm binds to the adenine-binding site of the ATP-binding pocket, which is held by polar interactions of the NH and the amide carbonyl to D76 (ParE) and D81 (GyrB). Significantly hydrophobic residues surrounding the pyrrole ring likely provide favorable binding energy. The presence of 3,4dichloro groups also provides increased hydrophobic interactions. Additionally, the chlorine atoms enhance the hydrogen bond donor properties of the pyrrole NH due to reduction of pKa by the electron-withdrawing nature of the dichloro groups. This phenomenon was confirmed from significant loss of activity exhibited by the bis-deschloro kibdelomycin analogue,9 similar to what was reported for another pyrrolamide.11 The methyl group of the pyrrole is buried in a deeper hydrophobic pocket of GyrB, formed by hydrophobic residues I51, I175, and V79, whereas the corresponding hydrophobic pocket contains I46, V170, and I74 in ParE. Substitution of the methyl group with a hydrogen leads to over 40-fold loss of GyrB activity and 10-fold loss of ParE 2029

dx.doi.org/10.1021/cb5001197 | ACS Chem. Biol. 2014, 9, 2023−2031

ACS Chemical Biology

Articles

should facilitate the design of new molecules with better druglike properties and spectrum of activity.

■

METHODS

■

ASSOCIATED CONTENT

Reagents. All reagents were obtained from Sigma-Aldrich (St. Louis, MO) unless otherwise indicated. Kibdelomycin and kibdelomycin A were isolated from Kibdelosporangium sp.8,9 GyrB and ParE Protein Expression and Purification, Crystallization, and Data Collection of 24 kDa and 43 kDa GyrBKibdelomycin. All constructs were generated using a standard PCRbased cloning strategy, and entire coding sequences were verified by sequencing. Genes encoding gyrB (1−231), parE (1−225), and parE (1−406) were cloned into the pGEX-4T3 vector and expressed in E. coli BL21DE3 as GST fusion proteins. Each protein was purified first by glutathione sepharose column followed by cleavage of GST with thrombin. 24 kDa GyrB (1−231), 24 kDa ParE (1−225), and 43 kDa ParE (1−406) were further purified using gel-filtration chromatography (Superdex S200 column). 24 kDa GyrB (20 mg mL−1) was co-crystallized with KBD (at 1:2 molar ratio) by sitting-drop vapor diffusion method using a reservoir containing 0.12 M magnesium chloride, 0.06 M Tris pH 8.5, and 15% PEG 3350. 24 kDa GyrB (20 mg mL−1) was co-crystallized with novobiocin (at 1:2 molar ratio) by sitting-drop vapor diffusion method using a reservoir containing 0.2 M magnesium chloride, 0.1 M Tris pH 8.5, and 30% PEG 3350. 43 kDa ParE (18 mg mL−1) was co-crystallized with KBD (at 1:2 molar ratio) by sitting-drop vapor diffusion method using a reservoir containing 0.2 M ammonium sulfate, 0.1 M Tris pH 8.5, and 25% PEG 3350. 24 kDa ParE (15 mg mL−1) was co-crystallized with novobiocin (at 1:2 molar ratio) by sitting-drop vapor diffusion method using a reservoir containing 0.15 M ammonium acetate, 0.075 M Tris pH 8.5, and 19% PEG 3350. Crystals were frozen in liquid nitrogen. Data for 24 kDa GyrB:KBD were collected at the SER-CAT beamline (Argonne National Laboratories, The Advanced Photon Source) using an X-ray wavelength of 1.00 Å. Data for 24 kDa GyrB:Novobiocin were collected at the ALS beamline (Lawrence Berkeley National Laboratories, Advanced Light Source) using an X-ray wavelength of 1.00 Å. Data for 43 kDa ParE:KBD and 24 kDa ParE:novobiocin were collected at the IMCA beamline (Argonne National Laboratories, The Advanced Photon Source). All data were integrated and reduced using HKL200017 or the XDS program.18 The structures for 43 kDa ParE:KBD, 24 kDa ParE:novobiocin, 24 kDa GyrB:KBD, and 24 kDa GyrB:novobiocin were determined by molecular replacement with the program Phaser19 using the previously reported structures of E. coli 43 kDa and 24 kDa ParE (PDB: 1S16 and 1S14) and 24 kDa GyrB (PDB: 1AJ6) as search models, respectively. The molecular models were built manually using Coot20 and completed using iterative rounds of refinement and rebuilding. The structures were refined using REFMAC21 as implemented in CCP422 and Buster (GlobalPhasing Ltd.).23 The final models all have favorable R/R free values, Molprobity24 scores, and excellent geometry and stereochemistry. The final structures have been deposited with the RCSB protein data bank. Structure determination statistics are provided in Table 1. Figures were prepared using the program PYMOL (The PyMol Molecular Graphics System, Schrodinger, LLC).

Figure 7. (A) Role of Asp89 in GyrB-KBD and GyrB-novobiocin structures (D89 is the residue mutated in NovoR mutant). KBD is colored in yellow, and novobiocin is colored in magenta. (B) Superposition of S. aureus GuyrB:KBD complex with E. coli GyrBclorobiocin (PDB code: 1KZN). Clorobiocin is colored in cyan, and the residue I175 is highlighted (Q136E, I175T, L455I mutated in CouR mutant).

activity against the coumermycin-resistant strain is more difficult due to the lack of a crystal structure of coumermycin complexed with these enzymes. Our efforts to crystallize coumermycin, bound to GyrB and ParE, have been unsuccessful. Fortunately, coumermycin A1 is somewhat similar to chlorobiocin, which has been co-crystallized with E. coli GyrB. The structure of E. coli GyrB:clorobiocin complex (PDB code: 1KZN) was hence superimposed with the S. aureus GyrB:KBD complex (Figure 7B). Due to the structure similarity, coumermycin A1 might bind to the ATP-binding site of S. aureus GyrB in a similar manner as chlorobiocin, with its pyrrolamide moiety positioned close to the residue I175. The coumermycin-resistant S. aureus harbors three mutations (Q136E, I175T, and L455I) in GyrB (Supplementary Figure S1). I175 is part of the hydrophobic binding pocket of the pyrrole moiety. The mutation of isoleucine 175 to threonine makes the binding pocket slightly polar and would be expected to confer unfavorable interactions, leading to a 4-fold reduction in observed activity. The other two amino acids (Q136E and L455I) do not engage in kibdelomycin binding and should not play a significant role in its activity. Quinolones bind to gyrase A/ ParC enzymes, whereas kibdelomycin binds to GyrB and ParE. The lack of binding overlap is consistent with the lack of crossresistance. The novel “dual arm”, U-shaped binding mode derived from the crystal structure of KBD, bound to S. aureus GyrB and ParE, shows no common binding interactions with novobiocin and quinolones, which explains the activity of KBD against these resistant strains. This multimodal, novel-binding mode readily explains the lack of cross-resistance to the key gyrase inhibitors and also provides understanding for the low frequency of resistance. The crystal structure also helps resolve the unknown absolute configuration of KBD, as presented here. Structural refinements with alternative stereochemistry at various chiral centers of KBD, including replacement with the opposite enantiomer of sugar A, provided a poor fit to the experimentally observed electron density. In summary, kibdelomycin is a bactericidal, Gram-positive agent with potent inhibitory activity of GyrB and ParE. It binds to the active site of enzymes in a novel, “dual-arm,” U-shaped binding mode that explains the lack of cross-resistance against known gyrase inhibitors and also is consistent with the known SAR of kibdelomycin. These crystal structures provide rationale for structure-guided chemical modifications of kibdelomycin that

S Supporting Information *

Supplementary figures showing key S. aureus GyrB mutations and close-up view of the pyrrole binding pockets of S. aureus GyrB:KBD and E. coli GyrB and tables with antimicrobial MIC, gyrase inhibitory activities, and resistance profile. This material is available free of charge via Internet at http://pubs.acs.org Accession Codes

Atomic coordinates and structure factors have been deposited in the PDB with accession codes [4url, 4urm, 4urn and 4uro] with the designation “for immediate release upon publication” (HPUB). 2030

dx.doi.org/10.1021/cb5001197 | ACS Chem. Biol. 2014, 9, 2023−2031

ACS Chemical Biology

■

Articles

(11) Eakin, A. E., Green, O., Hales, N., Walkup, G. K., Bist, S., Singh, A., Mullen, G., Bryant, J., Embrey, K., Gao, N., Breeze, A., Timms, D., Andrews, B., Uria-Nickelsen, M., Demeritt, J., Loch, J. T., 3rd, Hull, K., Blodgett, A., Illingworth, R. N., Prince, B., Boriack-Sjodin, P. A., Hauck, S., MacPherson, L. J., Ni, H., and Sherer, B. (2012) Pyrrolamide DNA gyrase inhibitors: fragment-based nuclear magnetic resonance screening to identify antibacterial agents. Antimicrob. Agents Chemother. 56, 1240− 1246. (12) Lafitte, D., Lamour, V., Tsvetkov, P. O., Makarov, A. A., Klich, M., Deprez, P., Moras, D., Briand, C., and Gilli, R. (2002) DNA gyrase interaction with coumarin-based inhibitors: the role of the hydroxybenzoate isopentenyl moiety and the 5′-methyl group of the noviose. Biochemistry 41, 7217−7223. (13) 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. (14) Brino, L., Urzhumtsev, A., Mousli, M., Bronner, C., Mitschler, A., Oudet, P., and Moras, D. (2000) Dimerization of Escherichia coli DNAgyrase B provides a structural mechanism for activating the ATPase catalytic center. J. Biol. Chem. 275, 9468−9475. (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 coli topoisomerase IV ParE subunit (24 and 43 kilodaltons): a single residue dictates differences in novobiocin potency against topoisomerase IV and DNA gyrase. Antimicrob. Agents Chemother. 48, 1856−1864. (16) 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. (17) Otwinowski, Z. A. M. (1997) Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307−326. (18) Kabsch, W. (2010) XDS. Acta Crystallogr., Sect. D: Biol. Crystallogr. D66, 125−132. (19) McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C., and Read, R. J. (2007) Phaser crystallographic software. J. Appl. Crystallogr. 40, 658−674. (20) Emsley, P., Lohkamp, B., Scott, W. G., and Cowtan, K. (2010) Features and development of Coot. Acta Crystallogr., Sect. D: Biol. Crystallogr. 66, 486−501. (21) Murshudov, G. N., Vagin, A. A., and Dodson, E. J. (1997) Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr., Sect. D: Biol. Crystallogr. 53, 240−255. (22) Winn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J., Emsley, P., Evans, P. R., Keegan, R. M., Krissinel, E. B., Leslie, A. G. W., McCoy, A., McNicholas, S. J., Murshudov, G. N., Pannu, N. S., Potterton, E. A., Powell, H. R., Read, R. J., Vaginc, A., and Wilson, K. S. (2011) Overview of the CCP4 suite and current developments. Acta Crystallogr., Sect. D: Biol. Crystallogr. D67, 235−242. (23) Bricogne, G., Blanc, E., Brandl, M., Flensburg, C., Keller, P., Paciorek, W., Roversi, P., Sharff, A., Smart, O. S., Vonrhein, C., and Womack, T. O. (2011) Buster, version 2.11.5 ed., Global Phasing Ltd, Cambridge, United Kingdom. (24) Chen, V. B., Arendall, W. B., 3rd, Headd, J. J., Keedy, D. A., Immormino, R. M., Kapral, G. J., Murray, L. W., Richardson, J. S., and Richardson, D. C. (2010) MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr., Sect. D: Biol. Crystallogr. 66, 12−21.

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Present Address ∥

SBS Pharma Consulting, Edison, NJ 08820.

Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS Authors are grateful to Dr. H. Su for his critical reading of manuscript and providing helpful comments. REFERENCES

(1) Klevens, R. M., Morrison, M. A., Nadle, J., Petit, S., Gershman, K., Ray, S., Harrison, L. H., Lynfield, R., Dumyati, G., Townes, J. M., Craig, A. S., Zell, E. R., Fosheim, G. E., McDougal, L. K., Carey, R. B., and Fridkin, S. K. (2007) Invasive methicillin-resistant Staphylococcus aureus infections in the United States. J. Am. Med. Assoc. 298, 1763− 1771. (2) Singh, S. B., and Barrett, J. F. (2006) Empirical antibacterial drug discovery-Foundation in natural products. Biochem. Pharmacol. 71, 1006−1015. (3) Walsh, C. T. (2003) Antibiotics: Actions, Origin, Resistance; ASM Press: Washington, DC. (4) Wang, J., Soisson, S. M., Young, K., Shoop, W., Kodali, S., Galgoci, A., Painter, R., Parthasarathy, G., Tang, Y., Cummings, R., Ha, S., Dorso, K., Motyl, M., Jayasuriya, H., Ondeyka, J., Herath, K., Zhang, C., Hernandez, L., Alloco, J., Basilio, Á ., Tormo, J. R., Genilloud, O., Vicente, F., Pelaez, F., Colwell, L., Lee, S. H., Michael, B., Felcetto, T., Gill, C., Silver, L. L., Hermes, J., Bartizal, K., Barrett, J., Schmatz, D., Becker, J. W., Cully, D., and Singh, S. B. (2006) Platensimycin is a selective FabF inhibitor with potent antibiotic properties. Nature 441, 358−361. (5) Singh, S. B., Jayasuriya, H., Ondeyka, J. G., Herath, K. B., Zhang, C., Zink, D. L., Tsou, N. N., Ball, R. G., Basilio, A., Genilloud, O., Diez, M. T., Vicente, F., Pelaez, F., Young, K., and Wang, J. (2006) Isolation, structure, and absolute stereochemistry of platensimycin, a broad spectrum antibiotic discovered using an antisense differential sensitivity strategy. J. Am. Chem. Soc. 128, 11916−11920 15547. (6) Wang, J., Kodali, S., Lee, S. H., Galgoci, A., Painter, R., Dorso, K., Racine, F., Motyl, M., Hernandez, L., Tinney, E., Colletti, S., Herath, K., Cummings, R., Salazar, O., Gonzalez, I., Basilio, A., Vicente, F., Genilloud, O., Pelaez, F., Jayasuriya, H., Young, K., Cully, D., and Singh, S. B. (2007) Platencin is a dual fabf and fabh inhibitor with potent in vivo antibiotic properties. Proc. Natl. Acad. Sci. U.S.A. 104, 7612−7616. (7) Jayasuriya, H., Herath, K. B., Zhang, C., Zink, D. L., Basilio, A., Genilloud, O., Diez, M. T., Vicente, F., Gonzalez, I., Salazar, O., Pelaez, F., Cummings, R., Ha, S., Wang, J., and Singh, S. B. (2007) Isolation and structure of platencin: A novel FabH and FabF dual inhibitor with potent broad spectrum antibiotic activity produced by Streptomyces platensis MA7339. Angew. Chem., Int. Ed. 46, 4684−4688. (8) Phillips, J. W., Goetz, M. A., Smith, S. K., Zink, D. L., Polishook, J., Onishi, R., Salowe, S., Wiltsie, J., Allocco, J., Sigmund, J., Dorso, K., Lee, S., Skwish, S., de la Cruz, M., Martin, J., Vicente, F., Genilloud, O., Lu, J., Painter, R. E., Young, K., Overbye, K., Donald, R. G., and Singh, S. B. (2011) Discovery of kibdelomycin, a potent new class of bacterial type II topoisomerase inhibitor by chemical-genetic profiling in Staphylococcus aureus. Chem. Biol. 18, 955−965. (9) Singh, S. B., Goetz, M. A., Smith, S. K., Zink, D. L., Polishook, J., Onishi, R., Salowe, S., Wiltsie, J., Allocco, J., Sigmund, J., Dorso, K., de la Cruz, M., Martin, J., Vicente, F., Genilloud, O., Donald, R. G., and Phillips, J. W. (2012) Kibdelomycin A, a congener of kibdelomycin, derivatives and their antibacterial activities. Bioorg. Med. Chem. Lett. 22, 7127−7130. (10) Champoux, J. J. (2001) DNA topoisomerases: structure, function, and mechanism. Annu. Rev. Biochem. 70, 369−413. 2031

dx.doi.org/10.1021/cb5001197 | ACS Chem. Biol. 2014, 9, 2023−2031