Exploring Alternative Pathways to Target Bacterial Type II Topoisomerases Using NBTI Antibacterials: Beyond Halogen-Bonding Interactions

Novel bacterial topoisomerase inhibitors (NBTIs) are a new class of antibacterial agents that target bacterial type II topoisomerases (DNA gyrase and topoisomerase IV). Our recently disclosed crystal structure of an NBTI ligand in complex with DNA gyrase and DNA revealed that the halogen atom in the para position of the phenyl right hand side (RHS) moiety is able to establish strong symmetrical bifurcated halogen bonds with the enzyme; these are responsible for the excellent enzyme inhibitory potency and antibacterial activity of these NBTIs. To further assess the possibility of any alternative interactions (e.g., hydrogen-bonding and/or hydrophobic interactions), we introduced various non-halogen groups at the p-position of the phenyl RHS moiety. Considering the hydrophobic nature of amino acid residues delineating the NBTI’s binding pocket in bacterial topoisomerases, we demonstrated that designed NBTIs cannot establish any hydrogen-bonding interactions with the enzyme; hydrophobic interactions are feasible in all respects, while halogen-bonding interactions are apparently the most preferred.


Introduction
Bacterial type II topoisomerases (DNA gyrase and topoisomerase IV) are well-validated targets for antibacterial chemotherapy. They are vital bacterial enzymes involved in the maintenance of the correct spatial topological state(s) of DNA molecule, which is of fundamental importance for the major biological processes in bacteria such as DNA replication, transcription, and control of gene expression [1]. Despite the high level of structural resemblance that DNA gyrase and topoisomerase IV (topoIV) share, they are substantially involved in distinct intracellular functions in bacteria; DNA gyrase is responsible for the introduction of negative supercoils into the DNA molecule, while topoIV is responsible for DNA's decatenation activity [2]. Consequently, inhibition of the function of either or both enzymes leads to distortion in the correct spatial DNA topology, resulting in death of the bacterial cell [2,3]. These heterotetrameric enzymes are composed of two GyrA and two GyrB subunits (A 2 B 2 in DNA gyrase), i.e., two ParC and two ParE subunits (C 2 E 2 in topoIV) [2,4]. A widely known class of antimicrobial agents targeting these bacterial enzymes are fluoroquinolones [5,6]. Some representatives of this class of antibacterials, including ciprofloxacin, levofloxacin, and moxifloxacin, are still successfully used in the clinical practice, but unfortunately their extensive and/or inappropriate application has led Our recently disclosed crystal structure of Staphylococcus aureus DNA gyrase in complex with a potent NBTI ligand (AMK-12, IC 50 = 0.035 nM) revealed that the halogen atom at the para position of the phenyl RHS moiety is able to establish strong symmetrical bifurcated halogen bonds with the backbone carbonyl oxygens of Ala68 residues of both GyrA subunits (PDB ID: 6Z1A, Figure 2b) [21]. To evaluate and re-confirm the significance of the halogen bonds that p-halogenated phenyl RHSs form with the enzymes, making them accountable for the excellent enzyme inhibitory potencies of NBTIs against DNA gyrase and topoIV [22], we opted to further probe the possibility and potential impact of some alternative interactions (e.g., hydrogen-bonding and/or hydrophobic interactions) on the NBTIs' inhibitory potencies and antibacterial activities. For this purpose, a small, targeted library of new NBTI analogues was designed by substituting the halogen atom at the para position of NBTIs RHSs with non-halogen groups (e.g., p-methyl and p-amino fragments, Scheme 1).  [17], comprising a p-chloro phenyl RHS moiety that establishes strong symmetrical bifurcated halogen bonds with the backbone carbonyl oxygens of both GyrA Ala68 residues. The DNA gyrase is represented as ribbons (GyrA subunits in blue and GyrB subunits in salmon red, respectively), GyrA's Ala68 residues and AMK-12 ligand are represented by balls and sticks and coloured by element, while bacterial DNA is represented in solid orange. The halogen-bonding interactions distances and angles are depicted as yellow dots. Scheme 1. Synthetic scheme for synthesis of final compounds. (a) TFA, DCM, rt; (b) p-substituted benzaldehyde, AcOH, DKM, rt; then NaCNBH 3, 0 • C to rt.

Results and Discussion
With the intention of investigating whether the newly designed NBTI analogues comprising non-halogenated RHS moieties (Table 1) are able to establish any hydrogenbonding or hydrophobic interaction with the amino acid residues delineating an NBTI's binding site, a series of flexible molecular docking calculation trials were performed within the NBTI binding site of S. aureus DNA gyrase enzyme (PDB ID: 6Z1A) [17]. The molecular docking calculations of NBTI analogues comprising a p-amino phenyl RHS moiety (e.g., Table 1, 9-11) revealed that the amino group could not establish any reasonable hydrogenbonding interactions with the backbone carbonyl oxygens of Ala68 residues, primarily due to its specific spatial geometry. This is clearly demonstrated in Figure 3a, depicting the hydrogen-bonding distances and angles obtained post-docking (d 1 = 2.75 Å and θ 2 = 107 • , i.e., d 2 = 3.82 Å and θ 2 = 118 • ), whose results are simply not in line with the crystallographic evidence of these types of hydrogen bonds in biological systems (e.g., d < 2.72 Å and θ = 150 • -110 • ) [23]. In contrast to this, NBTI analogues comprising a methyl group at the para position on the phenyl RHS moiety (e.g., Table 1, 5-8) are able to establish some hydrophobic interactions with amino acid residues delineating the NBTI's binding site (e.g., Ala68, Val71, Met75, and Met121 in S. aureus DNA gyrase), which are mainly hydrophobic in nature (Figure 3b).  [17], depicting the improper angles and spatial orientation of the amino hydrogens with respect to the A68 backbone carbonyl oxygens. GyrA subunits and DNA are represented as ribbon (GyrA in blue and DNA in yellow, respectively), while A68 residues and p-amino phenyl RHS as balls and sticks, coloured by element. The hydrogen-bonding distances and angles are shown as green dots. Compound 9 was selected as a representative of p-amino RHS analogues due to its highest inhibitory potency against S. aureus DNA gyrase, among other p-amino phenyl analogues. (b) 2D ligand interaction diagrams of the NBTI analogues comprising a methyl group at the para position on the phenyl RHS moiety (5)(6)(7)(8), forming hydrophobic interactions with the surrounding amino acid residues in S. aureus DNA gyrase (e.g., Ala68, Val71, Met75, and Met121).
To check the validity of the molecular docking predictions, our focused library of newly designed NBTIs was synthetized and subsequently biologically evaluated in vitro. All details related to the various synthetic procedures employed for obtaining LHS-linker constructs are thoroughly described in our previous publication [24], while the final compounds (Table 1) were synthesized utilizing a general reductive amination procedure described in detail in the experimental section.
The in vitro evaluated inhibitory potencies of the synthesized NBTIs on bacterial type II topoisomerase enzymes (DNA gyrase and topoIV) originating from S. aureus and E. coli, respectively, and their selectivity against human topoisomerase IIα enzyme are listed in Table 1, while their antibacterial activities are summarized in Table 2.
As shown in Table 1, one can observe that the experimentally determined in vitro on-target potencies of this series of NBTI analogues are in line with the predictions derived from molecular docking calculations. Both the in vitro enzyme inhibitory (Table 1) and antibacterial activity data ( Table 2) of NBTI analogues comprising a p-amino phenyl RHS moiety (9)(10)(11) clearly demonstrate that these compounds completely abolished the enzyme inhibitory (IC 50 > 100 µM) and antibacterial activity in both Gram-positive and Gram-negative bacteria (MIC > 128 µg/mL). This clearly pinpoints that p-amino phenyl NBTI analogues (9)(10)(11) are not able to establish any hydrogen-bonding interaction with GyrA A68 residues due to the inappropriate spatial conformation of the p-amino group (i.e., unsuitable bond angles and geometry), as previously predicted by the molecular docking ( Figure 3a). Moreover, considering that the amino group itself is hydrophilic, and the NBTI's binding pocket in bacterial type II topoisomerases is strictly hydrophobic, there is also a low probability even of a hydrophobic interaction due to the desolvation penalties that negatively affect hydrophobic interactions. Contrary to the NBTI analogues containing p-amino phenyl RHS fragments (9-11), NBTIs comprising p-methyl phenyl RHSs (5-8) establish some hydrophobic contact with the hydrophobic amino acid residues delineating the NBTI's binding site in bacterial type II topoisomerases (e.g., Ala68, Val71, Met75, and Met121 in S. aureus DNA gyrase), as depicted in Figure 3b. These compounds show strong antibacterial activity against Gram-positive bacteria, and potent enzyme inhibition of S. aureus DNA gyrase, (Tables 1 and 2), while in Gram-negative bacteria, they exhibit weaker enzyme inhibitory and antibacterial activity. This implies that hydrophobic interactions apparently play an important role in DNA gyrase inhibition in Gram-positive bacteria. Furthermore, comparison of compounds containing p-methyl phenyl RHSs (5)(6)(7)(8) and p-bromo phenyl RHSs (1-4) clearly indicates that bromo analogues exhibit stronger enzyme inhibitory (1-to 12-fold) and antibacterial activity (1-to 64-fold), a result that unequivocally implies the pivotal role and importance of halogen bonds for the excellent antibacterial potencies, as previously confirmed [17,18,21]. It can also be noted that topoIV is probably the primary target in E. coli, as it is observed that compounds with IC 50 s < 100 nM on E. coli topoIV exhibit strong antibacterial activities (lower MICs) against E. coli (Tables 1 and 2) as well. Moreover, it seems that hydrophobic interactions are also favourable for inhibiting topoisomerase enzyme(s) in the hard-to-treat Gram-negative bacillus Acinetobacter baumannii, as particularly exemplified by the stronger antibacterial activities of compounds 5 and 8, relative to those determined for our previously published compounds (1)(2)(3)(4). Although compounds 5, 6 and 7 show similar inhibition of E. coli DNA gyrase, only compound 5 has potent antibacterial activity against E. coli. Compounds with cyclohexane (6) and tetrahydropyran (7) linkers may have lower permeability and are more easily pumped out by the efflux pumps. Compared with wild-type E. coli, the activity was enhanced in the AcrA knockout E. coli N43 strain and in E. coli D22 strain, with a mutation in the lpxC gene that increases membrane permeability. This suggests that the membrane permeability of the compounds is suboptimal for crossing the membrane of Gram-negative pathogens and, more importantly, that they are efficiently pumped out by the E. coli efflux pumps. Moreover, the comparison of compounds 1 and 12 shows that the CF 3 group on the meta-position of p-bromo phenyl RHS leads to stronger inhibition of topoIV, but a weaker inhibition of DNA gyrase. As expected, gepotidacin shows stronger inhibition of both enzymes (DNA gyrase and topoIV) in Gram-negative bacteria such as E. coli, relative to the newly synthetized NBTIs, but also a weaker inhibition in Gram-positive bacteria such as S. aureus. According to cytotoxicity assessments on human umbilical vein endothelial cells (HUVECs) and HepG2 liver cancer cell lines (Table 3), compounds 5, 8 and 12 showed weaker effects by one to three orders of magnitude on human cells compared to antibacterial activity against Gram-positive bacteria (Table S3), and the same order of magnitude compared to Gram-negative bacteria (Table S3). Due to the weak antibacterial activity of other compounds (6, 7, 9-11) against Gram-positive and Gram-negative bacteria, MICs were approximately of the same order of magnitude as their cytotoxic IC 50 values.
Since NBTIs stop the enzymatic reaction at the level of single-strand DNA breaks, there is a correlation between IC 50 supercoiling and single-strand breaks. They disrupt the spatial symmetry and topology of nascent DNA due to the asymmetry of the 6-methoxy-1,5-naphthyridine LHS intercalated between DNA base pairs. Consequently, they stabilize single-strand DNA breaks in DNA gyrase in a similar manner to the majority of NBTIs, as we confirmed for structurally similar NBTIs in a DNA gyrase cleavage assay in our previous publication [21].   The SWISSADME tool [26] was used to calculate the physicochemical properties of the newly synthesized NBTIs (Table 4). With the exception of compound 12, they exhibit similar physicochemical properties. Despite the higher molecular weight, larger number of heavy atoms, rotatable bonds, and H-bond acceptors, it shows strong biological activity. The major difference between compounds with p-methyl phenyl RHSs (5-8) and p-amino phenyl RHSs (9-11) is in their topological polar surface area (TPSA) parameter, suggesting that a larger polar surface area leads to lower biological activity. The physicochemical properties were calculated with the SWISSADME tool [26].
Different RHS moieties form different interactions with the amino acid residues of the binding pocket, resulting in a different structure-activity relationship (SAR). As shown in Figure 4, NBTIs with p-bromo phenyl RHSs show inhibition for three out of four enzymes, due to hydrophobic interactions and additional strong halogen bonds in the binding pocket. Compounds with p-methyl phenyl RHSs that form a hydrophobic interaction show inhibition, but it is most likely weaker, as can be deduced from the enzyme inhibition data. Since no inhibition was observed for p-amino phenyl RHS NBTIs, we presume that they fail to form expected hydrogen bonds with backbone carbonyl oxygens of Ala68 in the enzyme-binding pocket.

Molecular Docking Calculations
Molecular docking calculations of newly designed NBTIs within the binding site of S. aureus DNA gyrase were performed by employing the GOLD docking suite in a flexible manner [27]. Our recently disclosed crystal structure of S. aureus DNA gyrase in complex with a potent NBTI ligand (AMK-12; PDB ID: 6Z1A) [21] was used for the calculations. The experimental coordinates of the co-crystallized ligand (AMK-12) were used for defining the NBTI's binding site-cavity radius of 15.5 Å for S. aureus DNA gyrase. Each of the NBTI compounds was docked into the defined binding site up to 10 times using the GOLD genetic algorithm (the settings and parameters of which are shown in Table S1). During the calculations, the amino acid residues, Met75, Asp83, and Met121, in S. aureus GyrA were flexible. The protocol for molecular docking was verified by conducting three redocking validation trials of the co-crystallized AMK-12 ligand within the defined binding site ( Figure S1 and Table S1) [28]. As a criterion for well-tuned structure-based settings, the heavy atoms' root mean square deviation values (RMSD ≤ 2.0 Å) between each calculated AMK-12 docking solution and the co-crystallized AMK-12 conformation were calculated (Table S2) [28]. To check the correct spatial orientation and conformation of the docked NBTIs in the binding site relative to co-crystallized AMK-12 ligand conformation, they were also visually examined.

General Chemical Methods
The general chemical methods (chemicals, thin layer chromatography (TLC) analysis, NMR spectroscopy, high resolution mass spectra HPLC analysis) were the same as in our previous publication [25].

General Procedure for Reductive Amination
Acetic acid (2 drops) was added to a solution of Figure 5A-D (1 eq) and suitable p-substituted benzaldehyde (1.3 eq) dissolved in methanol (~5 mL). The mixture was stirred for 1 h at room temperature. Then, Na(CH 3 COO) 3 BH (4 eq) was added to the reaction mixture and was stirred for 16 h at room temperature. The solvent was evaporated and the crude product was extracted with in ethyl acetate (20 mL) and 3 × 10 mL of 0.1 M NaOH. The organic phase was dried over Na 2 SO 4 , filtered and evaporated. The product was purified by flash chromatography on silica gel to afford the title compounds (5-12).

In Vitro DNA Gyrase and topoIV Inhibition
To determine the IC 50 values of the compounds for S. aureus and E. coli, a DNA gyrase supercoiling high-throughput plate assay [29,30] and a topoIV relaxation high-throughput plate assay [31,32] were performed. Assay kits and black streptavidin-coated 96-well microtiter plates are available from Inspiralis (Norwich, UK). For pre-screening, an inhibitor concentration of 100 µM was used. For compounds showing residual activity less than 50% at 100 µM, IC 50 values were calculated in GraphPad Prism 6.0 software. A nonlinear regression-based fit of the inhibition curves was used, with the log [inhibitor] versus slope of the reaction variables (four parameters) and a symmetric equation. Compounds that exhibited residual activity of more than 50% were labelled inactive. To determine IC 50 values, two independent repetitions were performed and the average was calculated. The positive control was gepotidacin, which showed 0.374 µM and 0.244 µM for S. aureus and E. coli DNA gyrase, and 8.30 µM and 0.049 µM for S. aureus and E. coli TopoIV, as determined by our procedure.

Human TopoIIα Selectivity Determination
For selectivity determination, a human topoisomerase II alpha relaxation high-throughput assay was used. The assay kits and black streptavidin-coated 96-well microtiter plates are available from Inspiralis (Norwich, UK) [33]. Two independent measurements were performed with an inhibitory concentration of 100 µM. Raw data were calculated as the mean ± SD percentage of the residual enzyme activity.

Antimicrobial Activity Assay
For the determination of minimum inhibitory concentrations (MIC), the broth microdilution method was used in 96-well plate format, according to the guidelines of the Clinical and Laboratory Standards Institute [34] and the recommendations of the European Committee on Antimicrobial Susceptibility Testing (EUCAST) [35]. The dilutions of the bacterial suspension of the specific bacterial strain, corresponding to the 0.5 McFarland turbidity standard, were prepared with cation-matched Mueller-Hinton broth with TES (Thermo Fisher Scientific) to obtain an end inoculum of 10 5 CFU/mL. The compounds were dissolved in DMSO and mixed with the inoculum. The mixture was incubated at 35 • C for 20 h, and MIC values were determined visually. The MIC were the lowest dilutions of compounds that did not exhibit turbidity. The MIC was determined as the lowest concentration of antibiotic that prevented visible bacterial growth. Tetracycline was used as a positive control on each test plate.

Metabolic Activity Assessment
HUVEC (ATCC ® CRL-1730™) and HepG2 (ATCC ® HB-8065™) cell lines were cultured in Dulbecco's Modified Eagle Medium (DMEM; Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% heat-deactivated FBS (Gibco, Grand Island, NE, USA), 2 mM L-glutamine, 100 U/mL penicillin, 100 µg/mL streptomycin (all from Sigma-Aldrich, St. Louis, MO, USA) in a humidified chamber at 37 • C and 5% CO 2 . The experiments were carried out on passages 4-7 for both cell lines. The tested compounds were dissolved in DMSO and further diluted in culture medium to a desired final concentration. The cells were then seeded into 96-well plates at 5 × 10 4 cells/mL (100 µL/well) or 8 × 10 4 cells/mL (100 µL/well) for HUVEC and HepG2, respectively. The cells were then incubated for 24 h to allow attachment onto the wells. The cells were treated with 1 µM and 50 µM of each compound of interest, or the corresponding vehicle as control. For IC 50 determination, the cells were seeded into 96-well plates, and after attachment, treated with nine different concentrations (the final concentrations were in the range of 0.39-100 µM) of the selected compounds, or the corresponding vehicle as control. Their metabolic activity was assessed after 72 h treatment using a CellTiter 96 Aqueous One Solution Cell Proliferation Assay (Promega, Madison, WI, USA), in accordance with the manufacturer's instructions. The absorbance was measured at 492 nm on an automated microplate reader, Tecan Spark (Tecan Group Ltd., Männedorf, CH, Switzerland). Data are presented as the percentage of metabolic activity of control cells stimulated with vehicle, or as an IC 50 value (mean ± SD) from three independent experiments, each conducted in triplicate (for screening) or duplicate (for IC 50 determination). IC 50 values were calculated using a non-linear regression model and with Graph Pad Prism 9 software.

Conclusions
In summary, with the newly designed series of NBTIs presented in this study, we have probed whether alternative interactions may offer improvements to the well-established halogen bond-forming moieties exemplified in our previous work. Molecular docking calculations predicted that the geometry of the amino group of the p-amino phenyl RHS moiety (in particular, the interatomic distances and angles) is simply not suitable for establishing hydrogen-bonding interactions with the backbone carbonyl oxygens of the S. aureus GyrA A68 residues. In vitro biological evaluation of the NBTIs containing p-amino phenyl RHS showed no measurable antibacterial activity (MIC > 128 µg/mL) and enzyme inhibition potency (IC 50 > 100 µM) in both Gram-positive and Gram-negative bacteria, and confirmed the prediction of the molecular docking calculations. In contrast, hydrophobic interactions can occur mainly in Gram-positive topoisomerases, and are apparently favourable for NBTI's antibacterial activity against A. baumannii. This clearly pinpoints that van der Waals/hydrophobic interactions offer a valid alternative to the halogen bonds at NBTI's binding site. However, these interactions do not match the halogen-bonding interactions in terms of the NBTI's enzyme inhibitory potency, and consequently, they induce less potent antibacterial activity, as demonstrated in our previously published series of halogenated NBTI analogues.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/antibiotics12050930/s1, Figure S1. Structural comparison between co-crystallized AMK-12 conformation and its calculated docked poses in the crystal structure of S. aureus DNA gyrase (PDB ID: 6Z1A) The co-crystallized ligand AMK-12 is represented in grey, while its re-docked poses in violet. Figure S2. NMR spectra, Figure S3. HPLC traces, Figure S4. HRMS spectra. Table S1. Settings and parameters of the GOLD genetic algorithm. Table S2. Re-docking validation data. Table S3. Antimicrobial activity of the NBTIs against different Gram-positive and Gram-negative bacteria.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding authors.