Antibacterial activity of synthetic 1,3‐bis(aryloxy)propan‐2‐amines against Gram‐positive bacteria

Abstract Synthetic 1,3‐bis(aryloxy)propan‐2‐amines have been shown in previous studies to possess several biological activities, such as antifungal and antiprotozoal. In the present study, we describe the antibacterial activity of new synthetic 1,3‐bis(aryloxy)propan‐2‐amines against Gram‐positive pathogens (Streptococcus pyogenes, Enterococcus faecalis and Staphylococcus aureus) including Methicillin–resistant S. aureus strains. Our compounds showed minimal inhibitory concentrations (MIC) in the range of 2.5–10 μg/ml (5.99–28.58 μM), against different bacterial strains. The minimal bactericidal concentrations found were similar to MIC, suggesting a bactericidal mechanism of action of these compounds. Furthermore, possible molecular targets were suggested by chemical similarity search followed by docking approaches. Our compounds are similar to known ligands targeting the cell division protein FtsZ, Quinolone resistance protein norA and the Enoyl‐[acyl‐carrier‐protein] reductase FabI. Taken together, our data show that synthetic 1,3‐bis(aryloxy)propan‐2‐amines are active against Gram‐positive bacteria, including multidrug–resistant strains and can be a promising lead in the development of new antibacterial compounds for the treatment of these infections.

worse if we consider that in the past 40 years only two classes of narrow-spectrum antibiotics (daptomicin and linezolid) were developed (Clatworthy, Pierson, & Hung, 2007). The scarcity of new therapeutic options against antibiotic-resistant strains has led to the return of older drugs previously disregarded due to its significant toxicity, such as colistin (Li et al., 2006). However, resistance mechanisms continue to emerge even for these drugs leading to the appearance of virtually untreatable infections (Malhotra-Kumar et al., 2016).
Among the infections with resistant bacteria, one can high light the group of pathogens known as ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species). These infections are associated with longer periods of hospitalization, increases in hospital costs, higher use of antimicrobial drugs and higher mortality rates. The number of deaths caused by infection with methicillin-resistant S. aureus (MRSA) strains, for instance, surpassed the number of deaths from HIV/AIDS and tuberculosis combined in the US (Boucher et al., 2009).
The main strategy to overcome the problem of bacterial resistance is the development of new antibacterial agents. Regarding this strategy, the synthesis of new compounds and modification of the existing ones is promising and can extend the options of new drugs with a broader spectrum of activity, lower toxicity and/or reduced sensitivity to resistance mechanisms (Silver, 2011). This approach has resulted in the introduction of some new antibacterial agents for clinical use, such as retapamulin, a compound derived from pleuromutilin, and some of the classical modifications of penicillins, the aminopenicillins (Gao et al., 2017;Lobanovska & Pilla, 2017).
In the present work we have evaluated the antibacterial activity of a series of 1,3-bis(aryloxy)propan-2-amines, several synthetic intermediates and N-substituted amines (Figure 1).
F I G U R E 1 Compounds screened for antibacterial activity in the present study  (Gomes et al., 2015).

| Minimal inhibitory concentration determination
The antibacterial activity of the compounds was evaluated using the broth microdilution method in 96-well microplates according to the Clinical and Laboratory Standards Institute protocol (CLSI, 2017). First, synthetic compounds were diluted in Mueller Hinton broth (MHB; Oxoid, Thermo Scientific, UK) to concentrations ranging from 20 to 2.5 μg/ml. The same volume of a bacterial suspension containing 10 5 CFU/ml was added to each of the previous solutions, resulting in final compound concentrations from 10 to 1.25 μg/ml. After incubation at 35°C for 24 hr, the plates were inspected visually for inhibition of bacterial growth. In each plate was included a viability control (bacterial suspension only), an inhibitory control (MHB containing five times the minimal inhibitory concentration (MIC) of penicillin G for Gram-positive and Gentamicin for Gram-negative bacteria, or a serial dilution of vancomycin ranging from 16 to 2 μg/ml or 5.52 to 0.69 μM for MRSA strains) and a sterility control (medium only). All conditions were tested in triplicate and the results shown are representative of three independent assays.

| Minimal bactericidal concentration determination
To evaluate the minimal bactericidal concentration (MBC) of tested compounds, the content of wells that showed no visual growth in the previous experiments, plus the well containing the viability control were plated in Mueller Hinton agar plates. After incubation at 35°C for 24 hr, the colonies were counted and the percentage of inhibition was calculated. MBC is defined as the lowest compound's concentration that inhibits at least 99.9% of the bacterial cell count compared to nontreated viability control (Clinical and Laboratory Standards Institute, 1999).

| Cytotoxicity to mammalian cells
The cytotoxicity of active compounds to mammalian cells was assessed using the MTT reduction assay (Mosmann, 1983). Vero and BSC-40 cells were seeded in 96-well plates (8 × 10 4 cells per well) and incubated at 37°C and 5% CO 2 atmosphere. After 24 hr of incubation, 200 μL of fresh medium containing a serial dilution of compounds (10-1.25 μg/ml) were added to the plates. After 48 hr of incubation in the same conditions, 100 μL of MTT solution in MEM or DMEM (5 mg/ml) was added to each well and incubated for 3 hr at 37°C and 5% CO 2 atmosphere. The medium was removed and 100 μL of DMSO was used to solubilize formazan crystals. Absorbance at 570 nm of each well was read using a spectrophotometer (VersaMax, Molecular Devices). The cytotoxic concentration of 50% (CC 50 ) is defined as the lowest concentration of a specific compound that reduces by 50% the viability of cultured cells.

| Putative molecular target identification by 3D chemical similarity and interaction profiling by molecular docking
First, the lowest energy conformations of tested compounds showing antibacterial activity were obtained by conformational analysis performed on OMEGA 2.5.1.4 software (Hawkins, Skillman, Warren, Ellingson, & Stahl, 2010). Then, the database of compounds with known effects over S. aureus, S. pyogenes and E. faecalis proliferation was retrieved from ChEMBL v23 (Bento et al., 2014). The three obtained databases were filtered to remove entries without experimental activity determined, inactive compounds and mixtures of compounds.
For all compounds, the structures had their protonation states calculated according to pH = 7.4 using fixpka software implemented on QUACPAC 1.7.0.2 (OpenEye Scientific Software, 2016) and, then, the lowest energy conformers were generated using OMEGA.
Chemical similarity queries were created for each active compound by considering common chemical features (rings, H-bond donors and acceptors, ions and hydrophobes) and the overall compound shape using the program ROCS 3.2.1.4 (Hawkins, Skillman, & Nicholls, 2007). ROCS software was used to identify the most similar compounds from the database against our queries. ROCS can overlay the library of conformers against a query composed of the shape and colors (representing chemical properties) derived from a compound.
The output conformers were ranked according to their similarity with the query using a Tanimoto-combo coefficient (TC, a linear sum of Tanimoto coefficient for molecular shape and colors) and the compounds were considered for further analysis when TC > 1, representing at least 50% of chemical similarity (Rush, Grant, Mosyak, & Nicholls, 2005). Within this chemically similar dataset, compounds with experimental activity against molecular targets were identified and used in docking studies. Those targets were retrieved from the Protein Data Bank (PDB) or constructed using homology modeling.
Identified proteins were prepared by adding the adjusting protonation states of amino acids and fixing missing side-chain atoms (PrepWiz, Maestro v2017.4). Molecular docking was performed around the cocrystallized ligand of the different protein using the default settings of the Glide program (Glide v7.7, Maestro v2017.4) in extraprecision mode, with at least five poses selected for visual inspection (Friesner et al., 2006). The amino acid residues were considered rigid and structural water molecules were kept during calculation. The employed docking protocol was evaluated with redocking experiments. Our target prediction protocol was based on the previously published methodology (Vallone et al., 2018).

| Homology modeling
Homology model of the S. aureus NorA (uniport accession number P0A0J7) was inferred using the E. coli homolog (PDB code: 4ZP0, resolution: 2.0 Å, sequence similarity: 77.3%) as a template. 3D model of the SaNorA domain was generated using the online server HHPred (Söding, Biegert, & Lupas, 2005) for template identification and alignment followed by Modeller 9v19 (Eswar et al., 2006) for the model construction. The quality of the final structure was accessed by MolProbity (Davis, Murray, Richardson, & Richardson, 2004) showing three residues out of the Ramachandran allowed region, which was then fixed by the protein preparation step prior to docking.

| RE SULTS
To investigate the antibacterial potential of 1,3-bis(aryloxy)propan-2-amines, 22 compounds of this class, variations in the nature and position of the substituents on the aromatic ring, were evaluated against Gram-positive and Gram-negative bacteria. These compounds, named as CPD1-CPD22, were synthesized in four steps (Figure 2), as previously described by Lavorato et al. (2017).

| Initial screening for antibacterial activity
Among the compounds initially tested, four-CPD18, CPD20, CPD21 and CPD22-presented antibacterial activity at the concentration of 10 μg/ml. Among the six bacterial species tested (Escherichia coli, K. pneumoniae, P. aeruginosa, E. faecalis, S. aureus and S. pyogenes), the activity was observed only against Gram-positive bacteria. CPD20 and CPD22 inhibited the growth of all Gram-positive bacteria tested (E. faecalis, S. aureus and S. pyogenes), while CPD18 and CPD21 showed activity against S. aureus and S. pyogenes.

| Minimal inhibitory and MBC determination
Compounds that showed antibacterial activity in the initial screening were submitted to MIC determination by broth microdilution method.
Among the four active compounds in the initial screening, CPD20 showed the best results, with MIC values of 2.5 μg/ml (6.58 μM) against S. pyogenes and S. aureus and 5 μg/ml (13.16 μM) against E. faecalis.

| Antibacterial activity against MRSA strains
In order to evaluate the efficacy of compounds against antibiotic resistant strains, we performed a broth microdilution method using MRSA.
Corroborating the findings above, compound CPD20 showed MIC values of 2.5 μg/ml (6.58 μM) against all MRSA strains, being the most promising among all compounds tested. Values of MIC ranged from 2.5 to 5 μg/ml (5.99-11.97 μM) for CPD22 and from 5 to 10 μg/ml (

| Changes in chemical group in R position abolish the antibacterial activity of tested compounds
To investigate the importance of the amino group to antibacterial activity, compounds CPD23-CPD31, synthetic intermediates of the most active amines CPD20, CPD21 and CPD22, were selected for biological testing. As shown in Figure 3, they were obtained in one, two or three steps according to the substituent in C-2.
In this second screening, we also prepared a series of secondary and tertiary amines for evaluation. As shown in Figure 4, compounds CPD38 and CPD39 were used as precursors to synthesize N-substituted amines CPD32-CPD36 and both compounds were obtained from alcohol CPD37. CPD37 was obtained as previously described by Lavorato et al. (2017). The ketone CPD38 was obtained from CPD37 by Albright-Goldman oxidation using DMSO and acetic anhydride (Fritsche, Elfringhoff, Fabian, & Lehr, 2008), while the tosylate CPD39 was prepared by reacting CPD37 with p-toluenesulfonyl chloride in dry pyridine (King & Bigelow, 1952). The secondary amines CPD32 and CPD33 were obtained by reductive amination reaction of CPD38 with benzylamine or butylamine, respectively, in the presence of NaCNBH 3 as reducing agent (Borch, Bernstein, & Durst, 1971). The nucleophilic substitution reaction between CPD39 and the heterocyclic amines morpholine, N-methylpiperazine and piperidine under heating at 100°C resulted in the tertiary amines CPD34, CPD35 and CPD36, respectively (Yuxiu, Guiqin, & Guangren, 2000).
None of these compounds presented antibacterial activity, with no complete inhibition of bacterial growth in all concentrations tested (up to 10 μg/ml, data not shown).

| Cytotoxicity concentration (CC 50 ) of active compounds in mammalian cells
We also evaluated the cytotoxicity in Vero and BSC-40 cell lines of the active compounds using the colorimetric MTT assay.

| The putative molecular targets of CDP20-22 and binding mode proposal
In order to identify the putative molecular target for the active compounds, we apply a ligand-based similarity approach combined with inverse docking using compounds CPD20, CPD21 and CPD22 as templates, since they presented stronger antibacterial activity in previous assays. Ligand-based similarity searches for each active compound were performed against a database of  Figure 5a). CPD21 and CPD22 proposed interaction mode within the SaNorA active site (Figure 5b,c) shares hydrophobic interactions mainly with Leu62 and Leu236, but not limited to, with also a large number of hydrophobic side-chains surrounding both ring systems. Thai and collaborators by a comprehensive computational workflow have shown that SaNorA has a conserved large binding site within the channel offering more opportunities for binding sites than the one exploited here in this study (Thai et al., 2015).  (Kronenberger et al., 2017;Mistry et al., 2016), but have a chlorine atom oriented at H-bond region.
Lastly, the cysteine transpeptidase Sortase has been proposed as a putative molecular target for the CPD20. Sortase commonly binds to flexible ligands such as signaling peptides but can also be covalently inhibited by small compound fragments. Redocking in the PDB structure 1QWZ revealed moderate capacity of prediction for this target (Table A1 and Figure 7a), 1QWZ has a large binding site when compared to other sortase structures (Jacobitz et al., 2014). The two double-ring systems of CPD20 were positioned by docking near the aromatic residues Phe114 and Tyr181, however, no pi-pi interactions could be established (Figure 7b).
Structural studies of the SaSortase B complexed with the substrate have shown a substrate-stabilized oxyanion hole involving Arg233 and Glu224 residues, which could accommodate the substrate (Jacobitz et al., 2014). Additionally, they also reported the close proximity of the ligands towards Tyr181, which could have a role in stabilizing the active conformation.

| D ISCUSS I ON
The compounds tested in the present study belong to the chemical class of 1,3-bisaryloxypropan-2-amines, which have shown several biological activities in the literature and have easy access by synthesis (Heerding et al., 2003;Yuxiu et al., 2000).
Our results showed that four out of 36 compounds presented Other studies in the literature regarding the antibacterial activity of synthetic compounds have shown similar results, for example, Heerding et al. (2003) reported that an asymmetric diaryloxipropan-  The synthetic intermediates of active compounds CPD20, CPD21 and CPD22 were also evaluated in order to get some insights about the role of an amino group in the aliphatic chain to the antibacterial activity of these compounds. The substitution of the amino group by a hydroxyl, a mesyl or an azide group leads to loss of activity (CPD23-31), indicating that this group is essential for the antibacterial activity. Since we recognize the importance of an amino group placed in the aliphatic chain, we also verified if the substitution of the amino group would interfere with their activity, synthetizing and evaluating secondary and tertiary amines derived from CPD18. The results indicated a loss of activity when the amino group is substituted (CPD32-36), suggesting that these compounds need to be a primary amine to promote antibacterial effects.
Among the evaluated primary amines, CPD18, CPD20, CPD21 and CPD22 are the ones with the highest calculated partition coefficient (ClogP) values (Table A2) The cytotoxic concentration of 50% (CC 50 ) of active compounds in mammalian cells was generally also in the low micromolar range, corroborating the results obtained by Lavorato et al. (2017). Hence, the selectivity index (SI) values obtained ranged from 0.25 to 2.19. These low SI values can be improved by changing critical chemical groups in the molecule, which can reduce its toxicity or enhance its activity.
Here, experimental validation demonstrated the ability of CPD20, CPD21 and CPD22 to interfere with the growth of S. aureus, S. pyogenes and E. faecalis. However, the molecular target of these drug candidates remains undetermined. In order to identify putative protein targets, we have employed a combination of chemical similarity search with inverse docking approaches. The three-dimensional chemical similarity between our hits and compounds with known activity against the organisms of interest was used to select a set of compounds with known biological targets. The prediction of the binding mode suggests that the compounds can interact with same pockets/regions of known cocrystallized inhibitors, which indicate the possibility of CPD22 to be a multi-target antibacterial. Furthermore, for the suggested targets, both 3D chemical similarity and parallels in terms the protein-ligand interactions between our compounds with known inhibitors supports this binding mode and encourages further in vitro testing. For instance, the dichloro-benzene groups of CPD22 interacting with the FabI hydrophobic pocket (Figure 6d).
Taken together, our data show that some compounds belong to the class of symmetric 1,3-bis(aryloxy)propan-2-amine tested in the present study showed a relevant antibacterial activity against important Gram-positive pathogens (including antibiotic-resistant strains), with minimal inhibitory and bactericidal concentration in the low micromolar range. As a perspective, we intend to investigate the activity of this class of amines against other clinically relevant resistant bacteria, such as E. faecium, vancomycin-resistant enterococci and glycopeptide-intermediate S. aureus. Through an in silico approach, we identified three putative molecular targets for these compounds and we hope that these data may contribute, in the long-term, to lead these compounds for further optimization towards selectivity, aiming to treat bacterial infections, including those caused by resistant Gram-positive bacteria.

ACK N OWLED G EM ENT
This work was supported by the CNPq, CAPES, FAPEMIG and UFMG intramural funds. Authors would like to thank OpenEye Scientific Software for OMEGA, ROCS and QUACPAC academic licenses.
[Correction added on 22 October 2019 after first online publication: Acknowledgement has been updated].

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

E TH I C S S TATEM ENT
None required.

DATA ACCE SS I B I LIT Y
The data that support the findings of this study are available from the corresponding author upon request. All reagents were obtained from commercial suppliers and used without further purification, unless stated otherwise. IR spectra were obtained using a Spectrum One, Perkin-

, 3 -B I S (3 ,4 -D I CH LO RO PH EN OX Y ) PRO PA N -2-Y L A ZI D E (CPD31)
Sodium azide (10.9 mmol) was added to a stirred solution of CPD30 (1.09 mmol) in DMF (3 ml) at 80°C. Crushed ice was added to the flask after 24 hr of reaction. The product was extracted with dichloromethane (3 × 30 ml) and the organic layer was washed with water (5 × 50 ml). The organic layers were combined, dried over anhydrous

, 3-B I S (3 ,4 -D I CH LO RO PH EN OX Y ) PRO PA N -2-A M I N -I U M CH LO R I D E (CPD2 2)
10% Palladium on activated carbon (20 mg) was added to a solution of CPD31 (0.47 mmol) in Tetrahydrofuran (THF) (10 ml). The reaction was kept under stirring and hydrogen atmosphere for 4 hr. Then, the catalyst was removed using filtration and the filtrate was concentrated. The residue was reconstituted in methanol and concentrated hydrochloric acid was added dropwise until a slight precipitate is formed. The solvent was evaporated to give a white solid in 41%

S TITUTE D A M I N E S
To a solution of the appropriate primary amine in THF and absolute ethyl alcohol (30 ml, 1:1) was added a methanolic solution of hydrochloric acid 5 M until pH 6. Then, CPD38 and NaCNBH 3 were added. The reaction was stirred under room temperature for 72 hr. The reaction was quenched with water (1 ml). The solvent was then removed under reduced pressure, and the residue was reconstituted in dichloromethane (30 ml) and washed with 0.5 M aqueous hydrochloric acid and water (3 × 30 ml). The organic layers were combined, dried over anhydrous Na 2 SO 4 , filtrated and concentrated under reduced pressure.

1-(1 , 3 -B I S (4 -CH LO RO PH EN OX Y ) PRO PA N -2-Y L ) PI PER-I D I N E (CPD36)
From piperidine (3.9 mmol) and CPD38 (0.39 mmol) and using the general procedure 1, a brown oil was isolated using silica gel column chromatography (dichloromethane 100%) in 74% yield; IR (ATR) v max 2,933, 2,852, 2,808, 1,595, 1,580, 1,489, 1,467, 1,235, 1,031, 1,017, TA B L E A 1 Redocking validation protocol (and cross-docking for NorA) and docking pose scores (kcal/mol). RMSD values expressed in Ångström (Å) was calculated from the comparison between redocking results with the cocrystallized conformation and are shown between parenthesis after the calculated energy. Poses are ranked by their GlideScore XP with more negative values representing more energetically stable interactions.