Structural basis of DNA gyrase inhibition by antibacterial QPT-1, anticancer drug etoposide and moxifloxacin

New antibacterials are needed to tackle antibiotic-resistant bacteria. Type IIA topoisomerases (topo2As), the targets of fluoroquinolones, regulate DNA topology by creating transient double-strand DNA breaks. Here we report the first co-crystal structures of the antibacterial QPT-1 and the anticancer drug etoposide with Staphylococcus aureus DNA gyrase, showing binding at the same sites in the cleaved DNA as the fluoroquinolone moxifloxacin. Unlike moxifloxacin, QPT-1 and etoposide interact with conserved GyrB TOPRIM residues rationalizing why QPT-1 can overcome fluoroquinolone resistance. Our data show etoposide's antibacterial activity is due to DNA gyrase inhibition and suggests other anticancer agents act similarly. Analysis of multiple DNA gyrase co-crystal structures, including asymmetric cleavage complexes, led to a ‘pair of swing-doors' hypothesis in which the movement of one DNA segment regulates cleavage and religation of the second DNA duplex. This mechanism can explain QPT-1's bacterial specificity. Structure-based strategies for developing topo2A antibacterials are suggested.

The contacts mapped onto the five bacterial sequences are from S.aureus DNA gyrase complexes with four different classes of inhibitors: QPT-1 = contacts from the ba subunit in the 2.5Å ba_ba' 2-QPT structure (pdb code 5CDM). Note that two key interactions with QPT-1 come from two highly conserved sequence motifs in the TOPRIM domain of GyrB, EGD SA and PLR/KGK (R/K is a conservative Arginine or Lysine change). Etop. = contacts from the BA subunit in the 2.8Å BA_BA' 2-etop complex (first, BA, subunit). Etop.' = contacts from the BA' subunit in the 2.8Å BA_BA' 2-etop complex (second, BA', subunit) (pdb code 5CDN). NBTI = contacts from the BA subunit in the 3.5Å BA_BA' 1-NBTI complex (pdb code 2XCR). Moxi. = contacts from the BA subunit in the 2.95Å BA_BA' 2-moxi complex (note for moxifloxacin the Mg 2+ ion and waters of the water ion bridge were counted as part of the compound (pdb code 5CDQ). The contacts in the 3.25Å A. Baumannii TopoIV mo xifloxacin complex structure, 2XKK, are not shown but are nearly identical to those from the 2.95Å S.aureus moxifloxacin complex -(here mapped onto the A. baumanii TopoIV sequences in the alignment). Note, because the BA_BA' 2-etop complex is not C2 symmetric the contacts from the BA and BA' subunits are not the same.
The contacts mapped onto the three eucaryotic structures are from two binary complexes with DNA (pdb codes: 3LJ4 and 4FM9) and the 2.16Å etoposide complex with human top2β (pdb code: 3QX3).
The secondary structural elements in S. aureus gyrase complex with DNA and GSK299423 are shown above the sequence alignment (Bax et al., 2010), * or * above the sequence alignment indicate the positions of the catalytic residues (Glu B435, Asp B508, Asp B510, Arg A122, Tyr A123) in S. aureus GyrB or GyrA. Residues underlined on the SaGyrBA sequence (in the WHD and TOPRIM domains) were used in superpositions (Supplementary Table 8). When the DNA has been cleaved the catalytic tyrosine (Tyr 123 in S.aureus Gyrase A) is covalently attached to the scissile phosphate, and the residue was considered as a phosphotyrosine when calculating contacts.   Comparisons of 'catalytically-competent' (CRsym type) conformations of a procaryotic and a eucaryotic topo2A. Superposition of S.aureus DNA gyrase (2XCSblack/cyan) and yeast topo II structure (4L4Kmagenta). A metal ion is present at the catalytic metal binding site (the A or 3' site) in both structures. In the S.aureus structure the catalytic tyrosine has been mutated to a phenylalanine (cyan), and the DNA is uncleaved (path of DNA backbone indicated by cyan and orange line passing through scissile phosphate). In the yeast structure the 3' hydroxyl (3' OH/SH) has been replaced with a sulfur to inhibit religation, and the DNA has been cleaved by the catalytic tyrosine (Tyr P ). In our study, a CRsym cluster of five structures were all similar in conformation to 2XCS, a structure proposed to be in a cleavage competent conformation prior to the first cleavage step. b, and c are views at lower resolution down different axis of the same comparison. (d) Comparison of an asymmetric (Casym) S. aureus etoposide cleavage-complex (black/blue) with S. aureus 2XCS (black/cyan). Note that because the S.aureus etoposide complex is asymmetric, superposing the BA subunit, or the BA' subunit gives different resultsboth are shown. Note that one of the etoposide catalytic Tyr P s is closer to the catalytic (cyan) position than the other. In the etoposide cleavage complex a single metal ion is seen at the non-catalytic B position (where it interacts with the DNA backbone phosphate on the 3' side of the cleavage site, via a water). Note that although the protein-protein interactions between subunits at the DNA-gate are almost identical in the three GSK299423 complexes ( Supplementary Fig. 7         Note: The nitrogen (arrowed in tautomer number 6) shared between the central and the dimethyl morpholine rings was modelled as sp 3 in only one of of six QPT-1 binding sites. Note that the carbon shared between the central ring and the barbituric acid ring becomes chiral if the barbituric acid tautomer is not symmetric. Conservation of amino-acids at the DNA gate between procayrotic and eucaryotic topo2As. Absolutely conserved amino-acid residues involved in (a) protein:DNA interactions and (b) protein:protein interactions at the DNA gate, between the five procaryotic and three eucaryotic sequences shown in Supplementary Fig. 5.

Supplementary
In the table residues are counted as contacting the DNA (or protein) if they are contact residues in at least one of the three eucaryotic structures and at least one of the five S. aureus structures, whose contacts are mapped onto five different bacterial sequences in Supplementary Fig. 5  Catalytic RY residues are not included in the contact sets.

A B C
Note the differences in coordinates between different tautomers are small (e.g. refined coordinates of (A) tautomer 3 and (B) tautomer 4 as fit by Afitt with 3.5 sigma Fo-Fc density from diffence map shown. In (C) the structures are superposed). Note also in Afitt docked structures the pucker on the dimethylmorpholino ring was not always correct.

SUPPLEMENTARY DISCUSSION
A 'swing-doors mechanism' for DNA gyrasein which movement of the transport-DNA regulates formation and religation of the double-stranded DNA break in the gate-DNA.
Type IIA topoisomerases (topo2As) are essential enzymes that regulate DNA topology by: creating a double-stranded break in one DNA segment (the gate or G-DNA), then passing a DNA duplex (the transport or T-segment) through this break, before religating the break (Fig. 1). In bacteria, the creation of a temporary double-stranded DNA break is essential for the function of DNA gyrase but it also poses a risk to the cell. If the two 'halves' of the enzyme become separated while the DNA is doubly cleaved, the genomic integrity and viability of the cell may be lost. The 'pair of swing-doors mechanism' that we propose for DNA gyrase ( Supplementary Fig. 11) suggests when doubly cleaved DNA is bound across the DNA gate, the gate will tend to automatically swing closed and will tend to remain closed until the T-DNA is pushed through 6 11 and with other X-ray crystal structures 12,13 . Our 'swing-doors mechanism' is distinct from a previously proposed 'mechanism for coordinating inter-subunit interactions with DNA cleavage' 14 .
Two alternative, but quite similar, mechanisms for metal catalysed DNA-cleavage and religation by topo2As have been proposed: a single moving metal mechanism 1, 15 and a two metal mechanism 16 .
In the two metal mechanism it is proposed that both metal binding sites (sites A and B) on the TOPRIM domain can be occupied at the same time, while in the single moving metal mechanism it is proposed that the metal ion moves between the B (also called Y) site and the catalytic A (also called 3') site and cannot occupy both sites at the same time. In both the two metal and single moving mechanisms a metal (Mg 2+ ) ion is required at the A site for metal catalysed DNA-cleavage or religation (see supplementary   Fig. 8). The 'swing-doors mechanism' described here requires a metal at the A site for metal catalysed DNA-cleavage or religation ( Supplementary Fig. 11), but is compatible with either the two metal or single moving metal mechanisms. A fuller description of the 'swing-doors mechanism' is given below.
In our scheme in Supplementary Fig. 11, prior to the initial cleavage step, the catalytic TOPRIM and WHD domains are 'shown schematically' in a configuration similar to that observed in the apo S.aureus DNA gyrase structure 1 . The binding of the DNA changes the relative orientations of the TOPRIM and WHD domains within a gyrase CORE subunit, but once the DNA is bound the relative positions of the TOPRIM and WHD domains in a S.aureus gyrase subunit remain fixed (to a good approximation) through the rest of the catalytic cycle. In the first DNA cleavage step ( Supplementary   Fig. 11) the catalytic domains move through the CRsym conformation, to cleave the first DNA strand.
DNA cleavage can only occur when the two 'half active sites', from opposite subunits, are aligned across the dimer interface and a catalytic metal (Mg 2+ ) ion is bound at the A site.
Once one DNA strand has been cleaved the stretched gate-DNA 1 is proposed (in the absence of a captured T-DNA segment) to relax, moving the enzyme 17 Fig. 11). Once the transport DNA has passed through the DNA gate, the 'swing doors' will swing close and adopt the CRsym conformation to religate the first DNA strand. In the mechanism shown in Supplementary Fig. 11, it is proposed that when the transport T-DNA is sitting between the DNA-gate and the exit gate, the Greek key domain moves the YKGLG motif to 'switch off' the DNA cleavage mechanism, by preventing the catalytic metal moving to the A site.
However, the mechanism proposes that at this stage of the catalytic cycle the CRsym conformation religates the doubly cleaved-DNA using a catalytic lysine residue; as discussed below.
Consistent with our proposed mechanism, a recent paper suggests that DNA gyrase can accomplish the first DNA religation step in the absence of metal ions 18 -since incubating quinolone cleavage complexes with EDTA produced singly cleaved DNA. In addition type 1A topoisomerases, which cleave a single DNA strand to modify DNA topology and have a TOPRIM domain and a catalytic tyrosine on a WHD at an active site that resembles that of type 2A topoisomerases 1 , can use a lysine residue in catalysis 19 . While the type IA catalytic lysine is not present in topo2As, in our 3.5 Å GSK299423 structure 1 a lysine residue was observed pointing at the scissile phosphate ( Supplementary   Fig. 12). This lysine residue (residue 581 in S. aureus DNA gyrase) is from a topo2A conserved sequence motif, YKGLG ( Supplementary Fig. 5) that is just C-terminal to the small mobile Greek key domain. We suggest that S. aureus DNA gyrase may be able to use this lysine residue to catalyse the first DNA religation step. Mutation of the equivalent lysine in yeast Topo II (Lys 603 - Supplementary   Fig. 13) gives a mutant which is capable of cleaving DNA but is defective in DNA religation 20 . The position of the Greek key domain (which is deleted in many of our structures) suggests that it may function to help prevent the exit gate from opening while both strands of the gate DNA are cleaved. If lysine 581 can indeed catalyse the first DNA religation step (but not DNA cleavage), positioning this lysine and reordering the YKGLG motif to prevent a divalent metal from occupying the catalytic A site could ensure that the gate DNA cannot be re-cleaved while the exit gate is open.
The YKGLG motif also adopts different conformations in different crystal structures of eukaryotic topo2As (Supplementary Fig. 13), suggesting that this motif may play a similar role in eukaryotic topo2As. However, the Casym conformation has not been observed in eucaryotic topo2A crystal structures, and the dropped trap-door WHD conformation 16
The solid was collected by filtration to give 3 (0.78 g, 1.979 mmol, 82 % yield) as yellow powder.
LCMS and NMR indicated a single diastereomer and 1 H NMR indicated some methanol was trapped in the solid. The material was used without further purification. LCMS: M+1=375.  11.59 (br. s., 1 H) 11.89 (br. s., 1 H). (ent-QPT-1). 230 mg of 3 was put on the column with 10 injections (23mg/5mL) (column: chiralpak IA, 5 micron, 21x250mm; mobile phase: heptane/EtOH, 50%B isocratic; flow rate: 20mL/min). The major peaks were collected and coded E1 and E2 respectively (retention times: 9 min & 19 min). It should be noted that there were solubility issues with the sample in EtOH/heptane. 4mL of ETOH and 1 mL heptane was used to dissolve the sample with warming. The sample was then filtered through a 0.25 micron acrodisc and quickly injected onto the column. The sample was warm but not hot upon injection. Each sample had to be prepared just before injection. The final enantiomers were collected and the solvent was evaporated. The yield was lower than expected likely because of the solubility issues. Much remained on the filter and sides of the vials which was recovered as possible (total of 67 mg recovered). The fractions enantiomer 1 and enantiomer 2 were combined appropriately and the solvent was removed under vacuo to give QPT-1 (enantiomer 1: 42 mg, 17% yield, bright yellow solid) and ent-QPT-1 (enantiomer 2: 39 mg, 16% yield, bright yellow solid). A reverse phase system was used to evaluate chemical purity and the chiral normal phase system was used to evaluate the enantiomeric purity.