Potent DNA gyrase inhibitors bind asymmetrically to their target using symmetrical bifurcated halogen bonds

Novel bacterial type II topoisomerase inhibitors (NBTIs) stabilize single-strand DNA cleavage breaks by DNA gyrase but their exact mechanism of action has remained hypothetical until now. We have designed a small library of NBTIs with an improved DNA gyrase-binding moiety resulting in low nanomolar inhibition and very potent antibacterial activity. They stabilize single-stranded cleavage complexes and, importantly, we have obtained the crystal structure where an NBTI binds gyrase–DNA in a single conformation lacking apparent static disorder. This directly proves the previously postulated NBTI mechanism of action and shows that they stabilize single-strand cleavage through asymmetric intercalation with a shift of the scissile phosphate. This crystal stucture shows that the chlorine forms a halogen bond with the backbone carbonyls of the two symmetry-related Ala68 residues. To the best of our knowledge, such a so-called symmetrical bifurcated halogen bond has not been identified in a biological system until now.


Supplementary
Values in parentheses are for the outer resolution shell. a merge = ∑ℎ ∑ | (ℎ )   (ℎ )|/ ∑ℎ ∑ (ℎ ). b meas = ∑ℎ [ /(  1)] 1/2 × ∑ | (ℎ )   (ℎ )|/ ∑ℎ ∑ (ℎ ) , where Ii(hkl) is the ith observation of reflection hkl, I(hkl) is the weighted average intensity for all observations i of reflection hkl and N is the number of observations of reflection hkl. c CC½ is the correlation coefficient between symmetry equivalent intensities from random halves of the dataset. d The data set was split into "working" and "free" sets consisting of 95 and 5% of the data respectively. The free set was not used for refinement. e The R-factors Rwork and Rfree are calculated as follows: = (| obs − calc |)/| obs |, where Fobs and Fcalc are the observed and calculated structure factor amplitudes, respectively. f From MolProbity. 27 g From the PDB validation server.

Molecular Docking
methoxy-naphthiridine LHS between central DNA base pairs) relative to the apo form (a significant fluctuation of ~2.00-3.24 Å). Among studied NBTIs ( Supplementary Fig. 2c), p-bromophenyl derivative (5) can be assigned as the most stable one (~1.3-1.   Omit mFobs-DFcalc difference electron density for the bound ligand calculated at 2.3-Å resolution and contoured at 1.6 σ (magenta mesh) and at 6.0 σ (cyan semi-transparent surface). Note that the latter highlights the confidence in the placement of the chlorine atom. The three orientations are chosen to emphasize the how the density associated with each ring is flattened, which aids in the assignment of their orientations. b The refined ligand has a overall B-factor that is somewhat elevated relative to the surrounding protein, but is comparable to that of the DNA. A breakdown shows there is a noticeably higher value for the aminopiperidine linker, which is in line with the lack of interactions with protein or DNA for this region of the ligand. For clarity, the ligand is shown in isolation (left) and in the context of the complex (right). In both panels, atoms are coloured according to their B-factors by interpolation between the colours shown in the key. So, for example an atom with a B-factor of 55 Å 2 would have a colour intermediate between green and yellow. Figure 7: Predicted binding modes of compound 9. Molecular docking results representing compound 9 (p-dimethylamino phenyl; sticks representation colour-coded by element) within the S. aureus DNA gyrase NBTI-binding site of our crystal structure (PDB ID: 6Z1A), revealing key intermolecular interactions. Hydrogen bonds are represented as blue dots, whereby, each methyl group forms pseudohydrogen bond with Ala68. GyrA, GyrB, and DNA are represented as cartoons in blue, yellow, and red, respectively. The amino acid residues are represented as sticks colour-coded by element.

Supplementary Figure 8 A truncated system representing the complex between compound 4 and gyrase used for QM calculations of the interaction energies.
Compound 4 (taken from PDB ID: 6Z1A) was truncated solely on the RHS part (chlorophenyl moiety; sticks representation colour-coded by element) that establishes direct, bifurcated halogen-bonding interactions (yellow dots) with the backbone carbonyl oxygens of Ala68 residues (sticks representation colour-coded by element) of each GyrA subunit (truncated and patched with methyl groups).

1-(2-(6-methoxy-1,5-naphthyridin-4-yl)ethyl)piperidin-4-amine (10)
Tert-butyl 1-(2-(6-methoxy-1,5-naphthyridin-4-yl)ethyl)piperidin-4-ylcarbamate (0.63 g, 1.63 mmol, 1 equiv) was dissolved in 12.5 ml of 1 M HCl in acetic acid and stirred for 30 min at room temperature. Solvent was evaporated, the residue dissolved in 30 mL of saturated NaHCO 3(aq) and pH adjusted to 12 with 1 M NaOH. Water layer was washed with DCM (3 × 30 mL), combined organic layers dried over Na 2 SO 4 and concentrated in vacuum to afford compound 10 as dark-brown viscous liquid (0.33 g, 71% General procedure for reductive amination: To a solution of amine 10 (1 equiv) in dry methanol the appropriate aldehyde (1.05-1.3 equiv) was added, together with catalytic amount of acetic acid. The reaction mixture was stirred for 2 h at room temperature under inert atmosphere. Next, the reaction mixture was cooled to 0 °C and NaCNBH 3 (3-5 equiv) dissolved in small amount of dry methanol was added drop-wise. After stirring overnight at room temperature under inert atmosphere, the solvent was evaporated in vacuum and the residue diluted in ethyl acetate. Organic phase was washed with saturated Na 2 CO 3 three times, dried over Na 2 SO 4 , filtered and the solvent evaporated under reduced pressure. The residue was purified by flash column chromatography to afford appropriate substituted product.

Determination of IC 50 values of compounds 5 and 6 by using S. aureus gyrase supercoiling assays Assay
In all experiments, the activity of the enzyme was determined prior to the testing of the compounds and 1 U defined as the amount of enzyme required to fully supercoil the substrate. This amount of enzyme was used in all subsequent assays. The final DMSO concentration in all the reactions was 1% (v/v). Compounds were serially diluted in DMSO and added to the reaction before the addition of the enzyme. Control compound for all assays was ciprofloxacin.

Data acquisition and analysis
Bands were visualised by ethidium staining for 10 min, destained for 10 min in water and bands were analysed by gel documentation equipment (Syngene, Cambridge, UK) and quantitated using Syngene Gene Tools software. Raw gel data (fluorescent band volumes) collected from Syngene, GeneTools gel analysis software were converted to a % of the 100% control (the fully supercoiled or relaxed DNA band) and plotted against the inhibitor concentration using SigmaPlot Version 13 (2015). The global curve fit non-linear regression tool was used to calculate IC 50 data using the following equation: Equation: Exponential Decay, Single, 2 Parameter = −  Fig. 10. Gel Image. Inhibition of S. aureus DNA gyrase supercoiling by Ciprofloxacin. Supplementary Fig. 11. DNA gyrase supercoiling inhibition plot. The plot shows the results of the supercoiling inhibition with Ciprofloxacin. IC 50 calculated from assay 1 is 27.49 M and 25.18 M from assay 2.

Supplementary
Supplementary Fig. 12. Gel Image. Inhibition of S. aureus DNA gyrase supercoiling by compound 5. Ciprofloxacin (CFX) was used for comparison. Supplementary Fig. 13. DNA gyrase supercoiling inhibition plot. The plot shows the results of the supercoiling inhibition with compound 5. IC 50 calculated from assay 1 is 0.007 M and 0.007 M from assay 2. Data were plotted up to 0.5 M to improve the appearance of the plot.
Supplementary Fig. 14. Gel Image. Inhibition of S. aureus DNA gyrase supercoiling by compound 6. Ciprofloxacin (CFX) was used for comparison. Supplementary Fig. 15. DNA gyrase supercoiling inhibition plot. The plot shows the results of the supercoiling inhibition with compound 6. IC 50 calculated from assay 1 is 0.013 M and 0.009 M from assay 2. Data were plotted up to 0.5 M to improve the appearance of the plot.

Cleavage assay with E. coli DNA gyrase
The gels relate to the Fig. 3 in the main text.

Cardiovascular hERG inhibition
Cardiovascular safety profile was entirely outsourced. hERG screening was performed at TCG Lifesciences Pvt. Ltd. India with FP-based hERG binding assay (Invitrogen kit).  8 . Our search was performed with advanced search of the Structure title, Primary Citation title, and PubMed abstract fields, utilizing keywords "bifurcated halogen bonds", "symmetrical halogen bonds", and "two halogen bonds". Two complexes (5YC6 and 5YC7) arose from this search. Hits for all complexes are exemplified in Tables S7 and Figure S11, along with information about X···O distances,  1 , and  2 angles.

Results of small molecule database (Cambridge Structural Database (CSD)) search
The search of small molecules in Cambridge Structure Database 9 (CSD version 2020.1), able to form bifurcated halogen bonds was performed using ConQuest (version 2020.1.1) program with the following criteria: halogen atom (Cl, Br or I) is bound to phenyl and interacts with two carbonyl oxygen atoms. All the obtained structures were visually inspected using Mercury (version 2020.1). Those structures with the distances between the halogen atom and both oxygen atoms lower than the sum of van der Waals radii were selected as hits. As reported by Shinada et al. a majority of chlorine atoms forms halogen bonds at a distance 0.3 Å longer than the sum of van der Waals radii. 8 Therefore we applied this criteria for another selection of structures with a chloro atom.
Although the search was based on halogen bonding with carbonyl oxygen atoms, structures in which halogen forms halogen bond with two non-carbonyl oxygens or one carbonyl and one non-carbonyl were also considered as hits. Hits for all halogen atoms are exemplified in Tables S8-11 along with information about X···O distances,  1 , and  2 angles. Statistics on their geometric parameters are shown on Fig. S12-15.