De novo design of type II topoisomerase inhibitors as potential antimicrobial agents targeting a novel binding region

By 2050, it is predicted that antimicrobial resistance will be responsible for 10 million global deaths annually, more deaths than cancer, costing the world economy $100 trillion. Clearly, strategies to address this problem are essential as bacterial evolution is rendering our current antibiotics ineffective. The discovery of an allosteric binding site on the established antibacterial target DNA gyrase offers a new medicinal chemistry strategy. As this site is distinct from the fluoroquinolone binding site, resistance is not yet documented. Using in silico molecular design methods, we have designed and synthesised a novel series of biphenyl-based inhibitors inspired by a published thiophene-based allosteric inhibitor. This series was evaluated in vitro against Escherichia coli DNA gyrase and E. coli topoisomerase IV with the most potent compounds exhibiting IC50 values towards the low micromolar range for DNA gyrase and only ∼2-fold less active against topoisomerase IV. The structure–activity relationships reported herein suggest insights to further exploit this allosteric site, offering a pathway to overcome developing fluoroquinolone resistance.


General Information and Instrumentation
All solvents and reagents were obtained from commercial suppliers and used without further purification. Solvents used were HPLC or analytical grade. Thin layer chromatography was performed on aluminium backed silica gel supplied by Merck, visualised using an ultraviolet lamp. Flash column chromatography was performed using silica gel 60 (40-63 µm particles). Automated flash column chromatography was performed on a Biotage® Isolera™ One machine using Biotage® Sfär columns of varying sizes between 5 g and 100 g. Automated reverse phase flash column chromatography was performed using C18 silica columns. Microwave syntheses were performed using an Anton Parr Monowave 50 reactor. Hydrogen and carbon NMR data were collected on a Bruker Avance III 500. All shifts were recorded against an internal standard of tetramethyl silane. Solvents used for NMR (chloroform-d, methanol-d 4 and DMSO-d 6 ) were obtained from Sigma-Aldrich. 1 H NMR data is reported in the following format: ppm (splitting pattern, coupling constant (Hz), number of protons, proton assignment). Signal assignments were deduced with the aid of TopSpin, MestReNova, DEPT 135, COSY, HSQC and HMBC. LC-MS (liquid chromatography-mass spectrometry) data were recorded on a Donex Ultimate 3000 LC system with a MeCN/H 2 O +0.1% formic acid gradient. HR-MS data were recorded using a Bruker MaXis impact spectrometer using electron spray ionisation. Infrared spectra were recorded on a Perkin-Elmer one FTIR spectrometer. Melting points were recorded on Griffin Education MELTP melting point apparatus.

Docking Protocol with Schrödinger
The protein structure (5NPP) was imported into Maestro 3 and prepared for molecular modelling using the default protein preparation wizard settings. No side chains were fixed/altered as any errors in the structure were distant to the allosteric site. Water molecules >5 Å from a heteroatom were removed and the protein minimised using the default settings. Following protein preparation, grid generation was undertaken using the 5NPP thiophene structure as the centre of the grid.
Putative inhibitors were built using the 2D sketcher or modified from existing SDF.file structures. All small molecule structures were prepared for docking using the LigPrep tool with default settings to generate appropriate 3D conformations.
The inhibitors were docked within the prepared grid using SP mode and default setting, only changing the output to sdf file. Docking results were visualised in Maestro using the in-built protein-ligand interactions toggle. Triaging of molecules for chemical synthesis was based upon visual inspection of the protein-ligand interactions, clashes with the protein and the docking score.

Supplementary Data
SPROUT de novo design run SPROUT is a de novo design programme which allows the user to design small molecules from fragments using first principles. It is licenced via the University of Leeds. 4 To create new ideas for the hydrophobic area of the allosteric pocket on gyrase, the following protocol was used. 5NPP was split into a receptor file (protein) and cavity file (thiophene 1). Two target sites were created: a hydrophobic site 1 which overlays with the 2-chlorophenyl region of 1 (pink, below) and a spheric hydrophobic site 2 which overlays with the 2-methyl group on the thiophene ring ( Figure S1). To site 1 was assigned 5-and 6-membered aromatic fragments, to site 2 a 5-membered aromatic fragment and a single sp 3 carbon atom. Methylene and benzene were selected as spacer templates, linking fragments between sites 1 and 2 that would generate whole molecular 'skeletons'. A total of 327 solutions were produced, which were scored via the SPROUT scoring function. This process consists of scoring the skeletons for predicted binding affinity to the selected target sites and steric constraints of the allosteric site. The skeletons were then re-ranked for molecular complexity and triaged by the authors (MJM & KMO).

Figure S1
De novo design process in SPROUT showing the two targets sites 1 and 2, the selected fragment templates for each and the spacer templates used to generate whole skeletons.
Skeletons were selected if they satisfied the following criteria: 1) Covered both target sites 1 and 2 2) Contained a simple disconnection between the groups at sites 1 and 2 -9-3) Contained a vector suitable for connecting to the chiral amine right hand portion of thiophene 1.
Examples from the SPROUT design process can be seen below (Table 1S) where the SPROUT skeleton is shown in orange within the allosteric site overlaid with thiophene 1 (pink). All three skeletons satisfy criteria 1 and 2 but the bottom skeleton does not contain a suitable growth vector to satisfy criteria 3.

Docking Scores of biphenyl analogues
The Glide SP docking scores (DS) for compounds 1-33 are included in the Table 2S.
The more negative the value, the better the predicted binding affinity. For reference, thiophene 1 was re-docked to the 5NPP allosteric site using the protocol above.

Figure S2
Gel electrophoresis data for compounds 21-23. While IC 50 values could not be determined for these compounds, visually compound 23 showed some inhibition at 100 µM which led to the synthesis of compounds 24-26. -12-

Reported Mass Spectrometry Data
The HR-MS data reported constitutes the largest peak that was observed in the spectrum irrespective of the counterion. This is reported to four decimal places. The mass observed is rationalised for each compound and includes the counterion in the mass calculation. In general, the accompanying counterion is H + , Na + or K + and these are reported.
LC-MS reports data to two decimal places due to the significantly lower resolution capacity of this instrumentation. In some cases, there are counterions. In other cases, molecular units that are renowned for possessing weak stability, such as a Boc group, may have been stripped from the compound by ESI. This is a common phenomenon in mass spectrometry. For LC-MS data, neither counterions nor removed molecular groups are denoted, but an example below illustrates how the numbers are calculated.
For compound 18, the following is reported:

Compound Numbering for NMR Assignment
This section contains information regarding the procedure and methods adopted for numbering and assigning the NMR spectra of novel compounds synthesised throughout this project. Numbering of atoms follows a logical ascension primarily based as similarly as possible to the IUPAC name provided. Second to this, it is based on the addition of supplementary atomic chains, commonly in the form of aromatic rings (S3).

IUPAC Names:
Scheme S3. Numbering of atomic chains within novel compounds in this manuscript Following the first example 17, C1 and C2 form the initial chain based on the IUPAC name specifying the order of said carbon atoms, i.e. with the hydroxy and phenylethyl motifs being at the 2-position. The phenyl ring constitutes a separate atomic chain, and therefore restarts numbering at C1' to allow for distinction. Following syntheses, the introduction of another atomic chain (in the case of a phenyl ring) across the carbonyl moiety of the amide restarts numbering for a second time at C1'', again following the IUPAC name of 20. Following further syntheses, the introduction of a final atomic chain restarts numbering for a third time at C1''' for this chain in 6. In this example, it is important to note the switch in nomenclature priority of C1 and C2 following the removal of a Boc-protecting group.
Degassed propanol (2-5 mL) and 2 M aqueous Na 2 CO 3 (3 eq.) were added. The resulting solution was refluxed for 4-20 hours before being cooled to RT. The mixture was filtered through a Celite pad and washed thoroughly with methanol before the solvent was stripped in vacuo. The resulting precipitate was then dissolved in ethyl acetate (20-50 mL) and washed with water (3 x 20-50 mL). The organics were dried (MgSO 4 ) before the solvent was again removed in vacuo. The crude product was purified using manual or automated column chromatography on silica gel to yield the title compounds. See individual compounds for detailed purification methods.

Method B: TIPS-Deprotection
To a solution of the chosen triisopropylsilane (1 eq.) in THF (10 mL), 1 M TBAF in THF (2.5 eq.) was added. The mixture was stirred at RT for 1-4 hours. The solvent was then stripped in vacuo and the crude residue purified using manual or automated flash column chromatography on silica gel to yield the title compounds. See individual compounds for detailed purification methods.

Method C: Boc-Deprotection
To a flask charged with the chosen Boc-protected amine (1 eq.), 4 N HCl in dioxane (1-5 mL) was added and the solution stirred at RT for 1-4 hours. The solvent was then stripped in vacuo. Hexane (20 mL) was added to the crude residue and the mixture sonicated for 5 minutes. A precipitate was released, filtered, washed with hexane, and the title compound obtained. carbamate (17) This procedure is modified from a literature source. 5 To a solution of (S)-2-amino-1-phenylethanol (2.02 g, 14.6 mmol) in THF (10 mL) at 0°C, di-tert-butyl dicarbonate (3.60 mL, 15.5 mmol) was added. The solution was stirred at RT for 1 hour and then the solvent was removed in vacuo. Hexane (40 mL) was added and the suspension sonicated for 10 minutes. The solution was filtered to yield the title compound as a colourless solid (3.10 g, 13.0 mmol, 89%).  carbamate (18) This procedure is modified from a literature source. 5 17 (5.60 g, 23.5 mmol), phthalimide (3.48 g, 23.7 mmol) and triphenyl phosphine (7.40 g, 28.2 mmol) in anhydrous THF (40 mL) was placed under an atmosphere of N 2 and cooled to 0°C. DEAD (4.44 mL, 28.2 mmol) was added dropwise over 10 minutes. The reaction mixture was then stirred for a further 15 minutes at 0 °C and then at RT for 18 hours. The solvent was then removed in vacuo and the yellow solid recrystallised from methanol to yield the title compound as a colourless solid (5.43 g, 14.8     Tert-butyl N- [2-hydroxy-2-(3-hydroxyphenyl)