5-Arylideneimidazolones with Amine at Position 3 as Potential Antibiotic Adjuvants against Multidrug Resistant Bacteria

Searching for new chemosensitizers of bacterial multidrug resistance (MDR), chemical modifications of (Z)-5-(4-chlorobenzylidene)-2-(4-methylpiperazin-1-yl)-3H-imidazol-4(5H)-one (6) were performed. New compounds (7–17), with fused aromatic rings at position 5, were designed and synthesized. Crystallographic X-ray analysis proved that the final compounds (7–17) were substituted with tertiary amine-propyl moiety at position 3 and primary amine group at 2 due to intramolecular Dimroth rearrangement. New compounds were evaluated on their antibiotic adjuvant properties in either Gram-positive or Gram-negative bacteria. Efflux pump inhibitor (EPI) properties towards the AcrAB-TolC pump in Enterobacter aerogenes (EA289) were investigated in the real-time efflux (RTE) assay. Docking and molecular dynamics were applied to estimate an interaction of compounds 6–17 with penicillin binding protein (PBP2a). In vitro ADME-Tox properties were evaluated for compound 9. Most of the tested compounds reduced significantly (4-32-fold) oxacillin MIC in highly resistant MRSA HEMSA 5 strain. The anthracene-morpholine derivative (16) was the most potent (32-fold reduction). The tested compounds displayed significant EPI properties during RTE assay (37–97%). The naphthyl-methylpiperazine derivative 9 showed the most potent “dual action” of both oxacillin adjuvant (MRSA) and EPI (E. aerogenes). Molecular modeling results suggested the allosteric mechanism of action of the imidazolones, which improved binding of oxacillin in the PBP2a active site in MRSA.

In our previous research on 5-arylideneimidazolone derivatives, unsubstituted piperazine at position 2, displayed high potency to restore the efficacy of β-lactams and fluoroquinolone antibiotic ciprofloxacin in MDR strains of S. aureus (4a and 4b, Figure 2) [25]. Furthermore, the compounds with arylidene moiety at position 5 and amine at position 2 (4a, 4b, 5) were able to re-sensitize MDR strains of Gram-negative bacteria for selected antibiotics and displayed strong efflux pump inhibitory (EPI)  On the grounds of similarity of hydantoin core to a part of PAβN, which is a well-known efflux pump inhibitor, modifications in this group were carried out to obtain potential adjuvant activity. Further studies indicate imidazolone derivatives with amphiphilic properties as the most promising [10].
In our previous research on 5-arylideneimidazolone derivatives, unsubstituted piperazine at position 2, displayed high potency to restore the efficacy of β-lactams and fluoroquinolone antibiotic ciprofloxacin in MDR strains of S. aureus (4a and 4b, Figure 2) [25]. Furthermore, the compounds with arylidene moiety at position 5 and amine at position 2 (4a, 4b, 5) were able to re-sensitize MDR strains of Gram-negative bacteria for selected antibiotics and displayed strong efflux pump inhibitory (EPI) properties towards AcrAB-TolC. Compound 5 caused the highest reduction (32-fold) of rifampin MIC (minimal inhibitory concentration) and an 8-fold reduction of a few antibiotics MIC values: oxacillin, chloramphenicol, linezolid and clarytromycin [15]. In our previous research on 5-arylideneimidazolone derivatives, unsubstituted piperazine at position 2, displayed high potency to restore the efficacy of β-lactams and fluoroquinolone antibiotic ciprofloxacin in MDR strains of S. aureus (4a and 4b, Figure 2) [25]. Furthermore, the compounds with arylidene moiety at position 5 and amine at position 2 (4a, 4b, 5) were able to re-sensitize MDR strains of Gram-negative bacteria for selected antibiotics and displayed strong efflux pump inhibitory (EPI) properties towards AcrAB-TolC. Compound 5 caused the highest reduction (32-fold) of rifampin MIC (minimal inhibitory concentration) and an 8-fold reduction of a few antibiotics MIC values: oxacillin, chloramphenicol, linezolid and clarytromycin [15]. On the basis of those interesting results obtained for 5-arylideneimidazolones with an unsubstituted piperazine moiety, we decided to explore their methylpiperazine analogues starting from p-chlorobenzylideneimidazolone 6 ( Table 1). The compound 6 gave yellow crystals and was used for X-ray crystallographic analysis, as described earlier [26].
Taking into consideration comprehensive structural data for the compound 6, which also initiated a new group of methylpiperazine derived 5-arylideneimidazolones, the compound was selected as a lead structure for further modifications in search for new antibiotic adjuvants performed within this study. The modifications aimed to introduce a spacer between imidazolone and the methylpiperazine at position 2 as well as an extension of aromatic area at position 5 of imidazolone (7′-17′, part A, Table 1), in respect to the chemical structure of favorable compounds (4a, 4b and 5) found previously [15,25]. However, the structures of the new desired compounds turned out to be unexpectedly different (7-17, part B, Table 1). In this context, the lead structure and its derivatives 7-17 were the main subject of this study. Thus, synthesis of the new series and X-ray structural consideration was performed. Final products were tested in microbiological assays for their antibiotic adjuvant potency as well as in molecular modelling to determine the potential mechanism of action towards PBP2a. For selected compounds, EPI properties were examined. One of the most active compounds (9) was selected for ADME-Tox assays in vitro. Based on the results obtained, structureactivity relationship (SAR) was discussed. Table 1. Structure of investigated compounds: lead 6 and its modifications (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17). On the basis of those interesting results obtained for 5-arylideneimidazolones with an unsubstituted piperazine moiety, we decided to explore their methylpiperazine analogues starting from p-chlorobenzylideneimidazolone 6 ( Table 1). The compound 6 gave yellow crystals and was used for X-ray crystallographic analysis, as described earlier [26]. Table 1. Structure of investigated compounds: lead 6 and its modifications (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17).
ciprofloxacin in MDR strains of S. aureus (4a and 4b, Figure 2) [25]. Furthermore, the compounds with arylidene moiety at position 5 and amine at position 2 (4a, 4b, 5) were able to re-sensitize MDR strains of Gram-negative bacteria for selected antibiotics and displayed strong efflux pump inhibitory (EPI) properties towards AcrAB-TolC. Compound 5 caused the highest reduction (32-fold) of rifampin MIC (minimal inhibitory concentration) and an 8-fold reduction of a few antibiotics MIC values: oxacillin, chloramphenicol, linezolid and clarytromycin [15]. On the basis of those interesting results obtained for 5-arylideneimidazolones with an unsubstituted piperazine moiety, we decided to explore their methylpiperazine analogues starting from p-chlorobenzylideneimidazolone 6 ( Table 1). The compound 6 gave yellow crystals and was used for X-ray crystallographic analysis, as described earlier [26].
Taking into consideration comprehensive structural data for the compound 6, which also initiated a new group of methylpiperazine derived 5-arylideneimidazolones, the compound was selected as a lead structure for further modifications in search for new antibiotic adjuvants performed within this study. The modifications aimed to introduce a spacer between imidazolone and the methylpiperazine at position 2 as well as an extension of aromatic area at position 5 of imidazolone (7′-17′, part A, Table 1), in respect to the chemical structure of favorable compounds (4a, 4b and 5) found previously [15,25]. However, the structures of the new desired compounds turned out to be unexpectedly different (7-17, part B, Table 1). In this context, the lead structure and its derivatives 7-17 were the main subject of this study. Thus, synthesis of the new series and X-ray structural consideration was performed. Final products were tested in microbiological assays for their antibiotic adjuvant potency as well as in molecular modelling to determine the potential mechanism of action towards PBP2a. For selected compounds, EPI properties were examined. One of the most active compounds (9) was selected for ADME-Tox assays in vitro. Based on the results obtained, structureactivity relationship (SAR) was discussed.

Chemical Synthesis
Synthesis of compounds 6-17 was performed according to the Scheme 1. Details of synthesis for 6 were described elsewhere [26]. Compounds 7-17 were synthesized within three stages, starting from corresponding initial steps to obtain intermediates 18-27 and 28-37, respectively (Scheme 1). In

Chemical Synthesis
Synthesis of compounds 6-17 was performed according to the Scheme 1. Details of synthesis for 6 were described elsewhere [26]. Compounds 7-17 were synthesized within three stages, starting from corresponding initial steps to obtain intermediates 18-27 and 28-37, respectively (Scheme 1). In

N-CH 3 13
Molecules 2018, x, x FOR PEER REVIEW 4 of 27

Chemical Synthesis
Synthesis of compounds 6-17 was performed according to the Scheme 1. Details of synthesis for 6 were described elsewhere [26]. Compounds 7-17 were synthesized within three stages, starting from corresponding initial steps to obtain intermediates 18-27 and 28-37, respectively (Scheme 1). In

Chemical Synthesis
Synthesis of compounds 6-17 was performed according to the Scheme 1. Details of synthesis for 6 were described elsewhere [26]. Compounds 7-17 were synthesized within three stages, starting from corresponding initial steps to obtain intermediates 18-27 and 28-37, respectively (Scheme 1). In

Chemical Synthesis
Synthesis of compounds 6-17 was performed according to the Scheme 1. Details of synthesis for 6 were described elsewhere [26]. Compounds 7-17 were synthesized within three stages, starting from corresponding initial steps to obtain intermediates 18-27 and 28-37, respectively (Scheme 1). In

Chemical Synthesis
Synthesis of compounds 6-17 was performed according to the Scheme 1. Details of synthesis for 6 were described elsewhere [26]. Compounds 7-17 were synthesized within three stages, starting from corresponding initial steps to obtain intermediates 18-27 and 28-37, respectively (Scheme 1). In

Chemical Synthesis
Synthesis of compounds 6-17 was performed according to the Scheme 1. Details of synthesis for 6 were described elsewhere [26]. Compounds 7-17 were synthesized within three stages, starting from corresponding initial steps to obtain intermediates 18-27 and 28-37, respectively (Scheme 1). In

Chemical Synthesis
Synthesis of compounds 6-17 was performed according to the Scheme 1. Details of synthesis for 6 were described elsewhere [26]. Compounds 7-17 were synthesized within three stages, starting

Chemical Synthesis
Synthesis of compounds 6-17 was performed according to the Scheme 1. Details of synthesis for 6 were described elsewhere [26]. Compounds 7-17 were synthesized within three stages, starting Taking into consideration comprehensive structural data for the compound 6, which also initiated a new group of methylpiperazine derived 5-arylideneimidazolones, the compound was selected as a lead structure for further modifications in search for new antibiotic adjuvants performed within this study. The modifications aimed to introduce a spacer between imidazolone and the methylpiperazine at position 2 as well as an extension of aromatic area at position 5 of imidazolone (7 -17 , part A, Table 1), in respect to the chemical structure of favorable compounds (4a, 4b and 5) found previously [15,25]. However, the structures of the new desired compounds turned out to be unexpectedly different (7-17, part B, Table 1). In this context, the lead structure and its derivatives 7-17 were the main subject of this study. Thus, synthesis of the new series and X-ray structural consideration was performed. Final products were tested in microbiological assays for their antibiotic adjuvant potency as well as in molecular modelling to determine the potential mechanism of action towards PBP2a. For selected compounds, EPI properties were examined. One of the most active compounds (9) was selected for ADME-Tox assays in vitro. Based on the results obtained, structure-activity relationship (SAR) was discussed.

Crystallographic Studies
The overall shape of molecule 7 with atom numbering is presented in Figure 3. The 5-(pchlorobenzylidene)-imidazolone fragment is almost planar. The angle between the planes of these two rings is 12.36(5)°. The molecule possesses a double bond C5=C14 and can form two geometric isomers (Z or E). In the crystal structure of 7 the Z isomer is observed. This isomer also occurred in other crystal structures, determined earlier, containing 5-(p-chlorobenzylidene)-imidazolone moiety [26,28]. We have conducted a search of the Cambridge Structural Database [29] and in all crystal structures containing 5-arylideneimidazolone fragments only the Z isomer was noted. The molecule of the investigated compound adopts a bent conformation, with the torsion angles of linker: N3-C6-C7-C8 = −60.3(2)°, C6-C7-C8-N2 = −51.7(2)°. This conformation is imposed by the intramolecular hydrogen bond N5-H5B···N2 ( Table 2). The piperazine ring has chair conformation with equatorial positions of substituents at nitrogen atoms.

Crystallographic Studies
The overall shape of molecule 7 with atom numbering is presented in Figure 3. The 5-(pchlorobenzylidene)-imidazolone fragment is almost planar. The angle between the planes of these two rings is 12.36(5) • . The molecule possesses a double bond C5=C14 and can form two geometric isomers (Z or E). In the crystal structure of 7 the Z isomer is observed. This isomer also occurred in other crystal structures, determined earlier, containing 5-(p-chlorobenzylidene)-imidazolone moiety [26,28]. We have conducted a search of the Cambridge Structural Database [29] and in all crystal structures containing 5-arylideneimidazolone fragments only the Z isomer was noted. The molecule of the investigated compound adopts a bent conformation, with the torsion angles of linker: N3-C6-C7-C8 = −60.3(2) • , C6-C7-C8-N2 = −51.7 (2) • . This conformation is imposed by the intramolecular hydrogen bond N5-H5B···N2 ( Table 2). The piperazine ring has chair conformation with equatorial positions of substituents at nitrogen atoms.

Crystallographic Studies
The overall shape of molecule 7 with atom numbering is presented in Figure 3. The 5-(pchlorobenzylidene)-imidazolone fragment is almost planar. The angle between the planes of these two rings is 12.36(5)°. The molecule possesses a double bond C5=C14 and can form two geometric isomers (Z or E). In the crystal structure of 7 the Z isomer is observed. This isomer also occurred in other crystal structures, determined earlier, containing 5-(p-chlorobenzylidene)-imidazolone moiety [26,28]. We have conducted a search of the Cambridge Structural Database [29] and in all crystal structures containing 5-arylideneimidazolone fragments only the Z isomer was noted. The molecule of the investigated compound adopts a bent conformation, with the torsion angles of linker: N3-C6-C7-C8 = −60.3(2)°, C6-C7-C8-N2 = −51.7(2)°. This conformation is imposed by the intramolecular hydrogen bond N5-H5B···N2 ( Table 2). The piperazine ring has chair conformation with equatorial positions of substituents at nitrogen atoms.   The crystal network in the studied structure can be characterized by N5-H5A···O1 intermolecular hydrogen bonds, creating a chain C(6) along [001] direction [30]. Furthermore, the C13-H13B···Cl1 interactions between the chains, which lead to the formation of layers (Figure 4), are observed ( Table 2).   The crystal network in the studied structure can be characterized by N5-H5A···O1 intermolecular hydrogen bonds, creating a chain C(6) along [001] direction [30]. Furthermore, the C13-H13B···Cl1 interactions between the chains, which lead to the formation of layers (Figure 4), are observed ( Table 2). Results of the crystallographic studies for compound 7 allowed us to identify the Dimroth rearrangement and real structure of the series of compounds (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17). It was difficult to recognize this structure only on the basis of the results of both, NMR and LC-MS analyses, since the molecular mass and number of protons is the same in the expected compound 7′ and the real Dimroth product (7). Before the crystallography results, we had identified the bright singlet (7.80 ppm) occurring in 1 H-NMR spectrum of basic form of 7 (see Supplementary) as that coming from 2H of guanidine-like fragment formed by N1 (in position 1), C2-NH-(position 2) and N3-H (position 3) of the imidazolone ring. These two acid protons could give one bright singlet due tautomerism. The X-ray results have undermined that wrong hypothesis, showing that this peak of two protons comes from the NH2 group placed at position 2. This peak occurred in the basic form, while it was exchanged into a very bright (almost plane) singlet at 9-10 ppm (typical for N1-H protons of 5-arylidene-2-thiohydantoin), which indicated a probable transformation of NH2 (at C2) into the tautomeric form of N1-H and C2 = NH (see Supplementary) in the case of the hydrochloric form of 7. Based on the results of X-ray analysis for 7, it is also distinctly seen that the triplet at ~3.50 ppm comes from N3-CH2 protons because similar shifts were observed for N3-CH2 protons of N3-propyl-substituted hydantoin derivatives [25], whereas this shift is too high for any propyl-CH2 protons in the case of an unrearranged compound (7′). Thus, the crystallography studies allowed us to right assign 1 H-NMR peaks to the suitable protons of the Dimroth rearrangement product. Similar trends that occurred in the 1 H-NMR spectra of the rest of the series confirm the Dimroth rearrangement structure for the rest Results of the crystallographic studies for compound 7 allowed us to identify the Dimroth rearrangement and real structure of the series of compounds (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17). It was difficult to recognize this structure only on the basis of the results of both, NMR and LC-MS analyses, since the molecular mass and number of protons is the same in the expected compound 7 and the real Dimroth product (7). Before the crystallography results, we had identified the bright singlet (7.80 ppm) occurring in 1 H-NMR spectrum of basic form of 7 (see Supplementary) as that coming from 2H of guanidine-like fragment formed by N1 (in position 1), C2-NH-(position 2) and N3-H (position 3) of the imidazolone ring. These two acid protons could give one bright singlet due tautomerism. The X-ray results have undermined that wrong hypothesis, showing that this peak of two protons comes from the NH 2 group placed at position 2. This peak occurred in the basic form, while it was exchanged into a very bright (almost plane) singlet at 9-10 ppm (typical for N1-H protons of 5-arylidene-2-thiohydantoin), which indicated a probable transformation of NH 2 (at C2) into the tautomeric form of N1-H and C2 = NH (see Supplementary) in the case of the hydrochloric form of 7. Based on the results of X-ray analysis for 7, it is also distinctly seen that the triplet at~3.50 ppm comes from N3-CH 2 protons because similar shifts were observed for N3-CH 2 protons of N3-propyl-substituted hydantoin derivatives [25], whereas this shift is too high for any propyl-CH 2 protons in the case of an unrearranged compound (7 ). Thus, the crystallography studies allowed us to right assign 1 H-NMR peaks to the suitable protons of the Dimroth rearrangement product. Similar trends that occurred in the 1 H-NMR spectra of the rest of the series confirm the Dimroth rearrangement structure for the rest of the compounds (8)(9)(10)(11)(12)(13)(14)(15)(16)(17), although the hydrochloric form and more extended aromatic areas disturb the detection of such clear and regular peaks as those for the phenyl symmetric-substituted 7 in its basic form.

Biological Assays
All compounds were investigated on their adjuvant action in the reference and MDR Gram-positive Staphylococcus aureus strains. Selected compounds (9-13, 16, 17) were also tested in Gram-negative E. aerogenes strains, by employing both the microdilution and the real-time efflux (RTE) assays. In the first step of the study, direct antibacterial activity of compounds against aforementioned bacteria was evaluated. Next, the influence of compounds (at the concentrations corresponding to 25% of their intrinsic MICs) on MICs of antibiotics was investigated. Finally, real-time efflux (RTE) assays were performed in order to determine efflux pump inhibitory properties of compounds towards AcrAB-TolC in E. aerogenes.

Direct Antibacterial Activity
Gram-positive S. aureus Initially, 12 arylideneimidazolone derivatives were evaluated in vitro for their intrinsic antibacterial effect against two strains of S. aureus: the reference strain S. aureus ATCC 25923 and extremely-drug resistant (XDR) MRSA HEMSA 5 clinical isolate. This step of the study was necessary for: (i) elucidation whether molecules tested are devoid of antistaphylococcal activity and thus cannot become antimicrobial agents by themselves; and (ii) determination of the concentrations of compounds suitable for the further assay on their antibiotic adjuvant potency. In addition, the antistaphylococcal efficacy of oxacillin, which was paired with compounds tested, in the following assays was also assessed (Table 3). Table 3. Intrinsic antibacterial activity of compounds tested against S. aureus strains used in the study. The results obtained for arylideneimidazolone derivatives 6-17 have demonstrated that the compounds did not exhibit any notable antibacterial activity against S. aureus strains used in the study. Among all compounds tested, the lowest MIC value was determined for the compound 13, which inhibited the growth of the reference strain ATCC 25923 and drug-resistant strain MRSA HEMSA 5 at the concentrations of 0.0625-0.125 mM (28.25-56.5 µg/mL). The MICs of remaining compounds were in the range of 0.25 mM to more than 2 mM. Gram-Negative E. aerogenes Then, MIC values of the compounds were evaluated with Gram-negative E. aerogenes. Three following strains of this bacterial species were employed in the studies: (i) the clinical isolate Ea-289 overexpressing the AcrAB-TolC efflux pump and exhibiting a porin-deficient phenotype; (ii) the CM-64 strain which also overexpresses the AcrAB-TolC efflux pump but has no changes in porin content; (iii) the Ea-294 and Ea-298 that are Ea-289 strain derivatives which are devoid of AcrAB and TolC, respectively (Table 4). For most of the compounds, precipitation was observed after the addition of bacterial suspension in MH II broth (10,11,16,17). For these molecules, the MICs could not be determined precisely.  (2) * Compounds, which precipitated after addition of bacterial suspension in MH II broth. Values in brackets represent concentrations of antibiotics in µg/mL. Table 4, none of the compounds tested exhibited remarkable antibacterial effects against the bacterial strains used in the study (MIC from 0.25 mM to >2 mM). Since imidazolone derivatives 10, 12 and 13 did not show any significant differences when assayed in efflux pump overexpressing strains (CM-64 and Ea-289) and efflux pump deficient strains (Ea-294 and Ea-308), these compounds were not considered as substrates for the AcrAB-TolC transporter. Furthermore, the absence of porins in Ea-289, Ea-298 and Ea-294 has no significant effect on the activity. Compounds 11 and 17 influenced the strains growth in corresponding manner to that of known EPI, PAβN, increasing MICs at least in 4-fold for Ea-289 and CM-64 in comparison to Ea-298 and Ea-294. Similar trends can be expected for 9 and 16, but it was impossible to determine it exactly due lack of growth inhibition for Ea-289 and CM-64 caused by both compounds at their highest tested concentration.

Influence on Antibiotic Efficacy
Influence on Antibiotic Susceptibility in S. aureus Strains The lack of a direct antibacterial activity of arylideneimidazolone derivatives tested allowed analyzing their anti-resistance potency in combination with the β-lactam antibiotic, oxacillin. Compounds studied were examined at the concentrations corresponding to 25% of their respective MIC values or at the highest concentrations, at which they did not precipitate. The potential of compounds to become antibiotic adjuvants was evaluated by comparing the drug effectiveness in the presence and absence of compounds tested, thereby, by determining activity gain parameter A ( Table 5). Most of the imidazolone derivatives (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16) were able to significantly improve the susceptibility of MRSA to oxacillin. At the same time, compounds did not exert any activity against the reference S. aureus strain (A < 4). The strongest chemosensitizing effect was demonstrated by compounds 16 and 9, which, at a concentration of 0.125 mM, reduced the MIC of oxacillin in 16-to 32-fold against MRSA HEMSA 5. Slightly lower adjuvant-like activity was determined for the imidazolone derivatives 7 and 10 (A = [8][9][10][11][12][13][14][15][16]. The efficacy of the remaining compounds was slightly less marked, however, also significant. Although compounds tested improved noticeably the antibacterial activity of oxacillin against MRSA clinical isolate, none of them restored the activity of the antibiotic (MIC < 2 µg/mL) against this highly resistant strain. Influence on Antibiotic Susceptibility in E. aerogenes Since the compounds were found to be deprived of an intrinsic antibacterial effect in E. aerogenes strains, their ability to enhance the activity of several antibiotics was assessed. The following four antibiotics were selected for the assays: chloramphenicol, erythromycin, doxycycline, and norfloxacin. The compounds were tested at the concentration corresponding to 1 4 of their MIC. For compounds 9, 11, 16 and 17, for which determination of the exact intrinsic MIC value was not possible (MIC > 2 mM, Table 4), the concentration of 0.5 mM was chosen. As in previous studies [10], PAβN was used as reference molecule in the assays. Results of the assays indicated that none of the tested compounds significantly increased the effectiveness of chloramphenicol, erythromycin, doxycycline and norfloxacin against both AcrAB-TolC-overexpressing strains and AcrAB-TolC-deficient strains (A ≤ 2). The A value obtained for PAβN combined with the aforementioned antibiotics coincided with data found in the literature [31].

Efflux Pump Inhibitory Properties
Taking into account the structural similarity of the explored group (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17) to previously found potent efflux pump inhibitors for AcrAB-TolC (4a, 4b and 5), the representative structures (9-13, 16 and 17) were also evaluated for their ability to inhibit the activity of AcrAB-TolC pump in E. aerogenes by measuring the efflux of a dye-substrate in the real-time efflux assay (RTE). The RTE assay is a dynamic tool by which the ability of a potential EPI to inhibit substrate transport in a drug efflux pump can be measured as a function of time and energy. Moreover, the test cannot discriminate for a possible outer membrane-destabilizing action of compounds. The dye, 1,2 -dinaphthylamine (1,2 -dNA), a substrate of the AcrAB-TolC efflux pump, is a sensitive to highly lipophilic environment where it is fluorescent, but it is almost non-fluorescent in an aqueous solution [32]. The experiment was performed in the E. aerogenes strain overexpressing AcrAB-TolC transporter (Ea-289).
Results have demonstrated that majority of the tested compounds caused a decrease in the fluorescence of 1,2 -dNA in the treated bacterial cell. The effect observed was most probably due to molecular interactions between fluorophore and compounds tested that result in increasing fluorescence quenching [33,34]. In order to analyze and quantitatively compare the ability of tested compounds to inhibit the efflux of 1,2 -dNA in the RTE assay, the inhibition efficiency (IE) of each compound was calculated. The IE parameters have indicated that all the tested compounds promoted the accumulation of 1,2 -dNA through the inhibition of its efflux outside bacterial cells overexpressing AcrAB-TolC ( Figure 5).
Molecules 2018, x, x FOR PEER REVIEW 10 of 27 compound was calculated. The IE parameters have indicated that all the tested compounds promoted the accumulation of 1,2′-dNA through the inhibition of its efflux outside bacterial cells overexpressing AcrAB-TolC ( Figure 5). The highest EPI property was detected for the compound 13, which at the concentration of 100 µM almost completely blocked 1,2′-dNA efflux from the bacteria overexpressing the AcrAB-TolC transporter. Slightly lower effectiveness was found for the compound 9 (80% of efflux inhibition). By contrast, the weakest EPI activity among all compounds tested was determined for the compound 11 which blocked the efflux of the dye in 37% in the Ea-289 strain.

Docking Studies
In order to estimate potential molecular mechanism of the significant oxacillin adjuvant action of the tested imidazolones in MRSA strains, molecular modeling was applied. The attempts to explain the mechanism were based on the verification if compounds tested had the capacity to interact with the main proteins conferring β-lactam resistance to MRSA strains. For this purpose, all the compounds tested (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17) were docked to the crystal structure of PBP2a protein. Moreover, the arylidenemidazolones were subjected to molecular dynamics simulation analysis, which offers an important insight into the molecular behavior of a molecule in its immediate microenvironment.
The results of docking studies are presented in the form of two groups of docking poses, in respect to the active (   The highest EPI property was detected for the compound 13, which at the concentration of 100 µM almost completely blocked 1,2 -dNA efflux from the bacteria overexpressing the AcrAB-TolC transporter. Slightly lower effectiveness was found for the compound 9 (80% of efflux inhibition). By contrast, the weakest EPI activity among all compounds tested was determined for the compound 11 which blocked the efflux of the dye in 37% in the Ea-289 strain.

Docking Studies
In order to estimate potential molecular mechanism of the significant oxacillin adjuvant action of the tested imidazolones in MRSA strains, molecular modeling was applied. The attempts to explain the mechanism were based on the verification if compounds tested had the capacity to interact with the main proteins conferring β-lactam resistance to MRSA strains. For this purpose, all the compounds tested (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17) were docked to the crystal structure of PBP2a protein. Moreover, the arylidenemidazolones were subjected to molecular dynamics simulation analysis, which offers an important insight into the molecular behavior of a molecule in its immediate microenvironment.
The results of docking studies are presented in the form of two groups of docking poses, in respect to the active ( Figure 6) and allosteric (Figure 7) sites of PBP2a. Detailed schemes with particular amino acid residues interacting with compounds 6-17 are presented in ligand interaction diagrams (Supporting Information). Due to the relatively high number of considered compounds, they are grouped in pictures according to their structural similarity, therefore compounds 6-8, 9-11, 13-15 are presented together, and the docking pose of compound 12 is presented together with compounds 16 and 17. Despite small structural differences occurring within the analyzed groups of compounds, there are quite significant variations in their docking poses ( Figure 6). When compounds 6-8 are considered (among which compound 7 displayed the highest activity, reducing MIC of oxacillin by 16-fold in comparison to 8-fold reduction observed for compounds 6 and 8), the most pronounced difference is connected with the orientation of the phenyl ring, which for compound 7 is located the closest to the serine at position 403 of PBP2a. Moreover, for compound 7 and 8, a set of hydrogen bonds is observed (charged assisted hydrogen bond of piperazine ring with glutamic acid at position 447 and 460 and hydrogen bond of amine group with serine at position 461), which do not occur for compound 6 (see ligand interaction diagrams in the Supporting Information). Compounds 9-11, which are analyzed together, also displayed relatively high activity in terms of restoring oxacillin activity, with the most active compound 9 (16-fold reduction of MIC of oxacillin). All those compounds are differently oriented in the active site of PBP2a in comparison to compounds 6-8, and all of them form hydrogen bond with oxygen from imidazolone ring; however, different residues from the side of protein are involved in this type of interaction: N464 for compound 9, T600 for compound 10, and S403 for compound 11. Compounds 13-15, for which the docking poses to the active site of PBP2a were analyzed together, displayed slightly lower activity. Compound 14, reducing the MIC of oxacillin by 8-fold, oriented its piperazine part rather outside of the protein and oxygen from imidazolone ring formed hydrogen bond with E602 and Q521. Compounds 13 and 15 were located in different part of the PBP2a active site than compound 14, with aromatic rings forming pi-pi interactions with Y446 and H583, respectively, and hydrogen bonds of oxygen in imidazolone moiety with N464 and S462, respectively. The last group of analyzed compounds was the most diversified in terms of activity; there was the most active compound 16 (16-32 fold reduction of MIC of oxacillin), compound 12, which displayed an 8-fold MIC reduction, and compound 17, which was not effective in restoring oxacillin activity (2-fold MIC reduction). Compound 16 is differently oriented in comparison to compounds 12 and 17. It seems that the clue for activity of this compound When compounds 6-8 are considered (among which compound 7 displayed the highest activity, reducing MIC of oxacillin by 16-fold in comparison to 8-fold reduction observed for compounds 6 and 8), the most pronounced difference is connected with the orientation of the phenyl ring, which for compound 7 is located the closest to the serine at position 403 of PBP2a. Moreover, for compound 7 and 8, a set of hydrogen bonds is observed (charged assisted hydrogen bond of piperazine ring with glutamic acid at position 447 and 460 and hydrogen bond of amine group with serine at position 461), which do not occur for compound 6 (see ligand interaction diagrams in the Supporting Information). Compounds 9-11, which are analyzed together, also displayed relatively high activity in terms of restoring oxacillin activity, with the most active compound 9 (16-fold reduction of MIC of oxacillin). All those compounds are differently oriented in the active site of PBP2a in comparison to compounds 6-8, and all of them form hydrogen bond with oxygen from imidazolone ring; however, different residues from the side of protein are involved in this type of interaction: N464 for compound 9, T600 for compound 10, and S403 for compound 11. Compounds 13-15, for which the docking poses to the active site of PBP2a were analyzed together, displayed slightly lower activity. Compound 14, reducing the MIC of oxacillin by 8-fold, oriented its piperazine part rather outside of the protein and oxygen from imidazolone ring formed hydrogen bond with E602 and Q521. Compounds 13 and 15 were located in different part of the PBP2a active site than compound 14, with aromatic rings forming pi-pi interactions with Y446 and H583, respectively, and hydrogen bonds of oxygen in imidazolone moiety with N464 and S462, respectively. The last group of analyzed compounds was the most diversified in terms of activity; there was the most active compound 16 (16-32 fold reduction of MIC of oxacillin), compound 12, which displayed an 8-fold MIC reduction, and compound 17, which was not effective in restoring oxacillin activity (2-fold MIC reduction). Compound 16 is differently oriented in comparison to compounds 12 and 17. It seems that the clue for activity of this compound lies in the position of aromatic rings of anthracene, located closely to S403 residue. The network of hydrogen bonds is similar for all of these compounds, as they all form this type of interaction with S461 and S463. Figure 7 presents docking results to the allosteric pocket of PBP2a with the compounds grouped in an analogous way as previously.
Molecules 2018, x, x FOR PEER REVIEW 12 of 27 lies in the position of aromatic rings of anthracene, located closely to S403 residue. The network of hydrogen bonds is similar for all of these compounds, as they all form this type of interaction with S461 and S463. Figure 7 presents docking results to the allosteric pocket of PBP2a with the compounds grouped in an analogous way as previously.

Molecular Dynamic Studies
As docking captures the compound orientation only in one moment, molecular dynamic simulation studies were carried out in order to explain more comprehensively the observed activity relationships. Changes in the ligand-protein interactions in time that were observed during simulations are presented in Figures 8 and 9 for dockings to the active and allosteric site of PBP2a, respectively. For simplicity, we have presented the results for only selected compounds; all data is available in the Supporting Information.

Molecular Dynamic Studies
As docking captures the compound orientation only in one moment, molecular dynamic simulation studies were carried out in order to explain more comprehensively the observed activity relationships. Changes in the ligand-protein interactions in time that were observed during simulations are presented in Figures 8 and 9 for dockings to the active and allosteric site of PBP2a, respectively. For simplicity, we have presented the results for only selected compounds; all data is available in the Supporting Information. When simulations of compounds in the PBP2a active site are analyzed (Figure 8), the most striking observation is the change in orientation of the most active compound (16), approximately after 50 ns of simulation, which resulted in the loss of interaction with Q637 and M641 and the formation of interaction with S643, Y644, K647, I648 and K651. On the other hand, a slightly less active compound (10) preserved interaction with Y446 (which was lost by compound 16 at the beginning of the simulation), Q637 and M641, although the last two interactions are not very frequent despite being present from time to time during the whole simulation. Similar changes in compound orientation occurred for compound 15 (4-fold reduction of MIC of oxacillin), which was observed after approximately 40 ns of simulation and was connected with the loss of a set of interactions (e.g., T444, Y446, N464, Q521, and E602), the formation of interactions with S437, N510 and I512, and later also with K506, N507, and L513. Additionally, inactive compound 17 moved from its initial docking orientation and it happened the fastest out of all tested compounds (after approximately 20 ns). It led When simulations of compounds in the PBP2a active site are analyzed (Figure 8), the most striking observation is the change in orientation of the most active compound (16), approximately after 50 ns of simulation, which resulted in the loss of interaction with Q637 and M641 and the formation of interaction with S643, Y644, K647, I648 and K651. On the other hand, a slightly less active compound (10) preserved interaction with Y446 (which was lost by compound 16 at the beginning of the simulation), Q637 and M641, although the last two interactions are not very frequent despite being present from time to time during the whole simulation. Similar changes in compound orientation occurred for compound 15 (4-fold reduction of MIC of oxacillin), which was observed after approximately 40 ns of simulation and was connected with the loss of a set of interactions (e.g., T444, Y446, N464, Q521, and E602), the formation of interactions with S437, N510 and I512, and later also with K506, N507, and L513. Additionally, inactive compound 17 moved from its initial docking orientation and it happened the fastest out of all tested compounds (after approximately 20 ns). It led to the loss of interactions e.g., with E447, S461, N464, and K597 and formation of interactions with S403, Q521, L525, T600, E602, and Q613.
On the other hand, when the outcome of molecular dynamic simulations within the allosteric site of PBP2a is analyzed (Figure 9), the most visible observation from the ligand interaction diagrams is the significantly greater perseverance for all of the considered compounds. For the most active compound (16), the very strong interaction with R241 occurs for almost 100% of the simulation time, and although interactions with other protein residues are less frequent, they are also present during the whole simulation. The slightly different pose of compound 10 led to a different set of present interactions, the most frequent of which was with E239. Compound 15 interacted more frequently with K215 and T216, although the latter interaction appears after about 20 ns of simulation. For the inactive compound 17, there is no such interaction that occurs for almost 100% of simulation time, although the compound pose in the allosteric site of PBP2a is also rather preserved during the whole molecular dynamic studies. to the loss of interactions e.g., with E447, S461, N464, and K597 and formation of interactions with S403, Q521, L525, T600, E602, and Q613. On the other hand, when the outcome of molecular dynamic simulations within the allosteric site of PBP2a is analyzed (Figure 9), the most visible observation from the ligand interaction diagrams is the significantly greater perseverance for all of the considered compounds. For the most active compound (16), the very strong interaction with R241 occurs for almost 100% of the simulation time, and although interactions with other protein residues are less frequent, they are also present during the whole simulation. The slightly different pose of compound 10 led to a different set of present interactions, the most frequent of which was with E239. Compound 15 interacted more frequently with K215 and T216, although the latter interaction appears after about 20 ns of simulation. For the inactive compound 17, there is no such interaction that occurs for almost 100% of simulation time, although the compound pose in the allosteric site of PBP2a is also rather preserved during the whole molecular dynamic studies. The observed dependencies in docking poses and changes in ligand-protein interactions during molecular dynamic simulation studies suggest the allosteric mechanism of action of the studied compounds that is an interaction with PBP2a in the allosteric site, improving binding of oxacillin in the PBP2a active site.

ADMET Studies
Compound 9, as the most active agent considering both oxacillin adjuvant-and AcrAB-TolC inhibitor properties, was chosen to be tested on ADME-Tox properties in vitro. The observed dependencies in docking poses and changes in ligand-protein interactions during molecular dynamic simulation studies suggest the allosteric mechanism of action of the studied compounds that is an interaction with PBP2a in the allosteric site, improving binding of oxacillin in the PBP2a active site.

ADMET Studies
Compound 9, as the most active agent considering both oxacillin adjuvant-and AcrAB-TolC inhibitor properties, was chosen to be tested on ADME-Tox properties in vitro.

Membrane Permeability
The permeability of compound 9 was estimated by the commercially available Pre-coated PAMPA (Parallel Artificial Membrane Permeability Assay) Plate System Gentest™ according to previously described protocols and formulas [35,36]. The calculated permeability coefficient for 9 (Pe = 0.72 ± 0.27 × 10 −6 cm/s) was comparable to the Pe value obtained for the low permeable reference norfloxacin and much lower than that for the highly permeable caffeine (Table 6). Thus, the passive movement of compound 9 across the cell membranes was determined as low. Table 6. PAMPA results for compound 9 and the references: well-permeable caffeine (CFN) and low-permeable norfloxacin (NFX).

Safety Assays In Vitro
The bacterial and eukaryotic in vitro cell growths were used to get results of potential mutagenicity and hepatotoxicity of compound 9.
The mutagenicity was evaluated using Ames microplate fluctuation protocol (MPF) performed with Salmonella typhimurium TA100 strain, which is able to detect base pair substitutions [37]. Calculation of medium control baseline (MCB) was performed basing on mean number of revertants in standard medium control in addition with one standard deviation. According to the manufacturer's protocol, the mutagen alert is determined as 2-fold of medium control baseline (2 x MCB), which in this study was calculated as 24 revertants. For the control mutagen nonyl-4-hydroxyquinoline-N-oxide (NQNO, 0.5 µM) more than 40 revertants was observed, whereas in the presence of tested compound 9 maximum 11 revertants occurred ( Table 7). The obtained data indicates no mutagenicity potential of the tested compound. However, the Binomial B-value = 0.0093, calculated for 9 at the 10 µM concentration, showed possible cytotoxic effect of 9 against S. typhimurium TA100 (Table 7). Table 7. The results of mutagenicity assay for 9 and the reference mutagen nonyl-4-hydroxyquinoline-N-oxide (NQNO). Determination of potential hepatotoxicity was performed with hepatoma HepG2 cells by standard colorimetric MTS procedure. Compound 9 was diluted into the cell culture media in the following concentrations: 0.1, 1, 10 and 100 µM, added to the cells and incubated for 72 h at 37 ºC in the presence of 5% CO 2 . The cytostatic drug doxorubicin (DX) was used at 1 µM, as the reference. The obtained results showed that compound 9 caused statistically significant decrease of the HepG2 cells viability at the concentrations 10 and 100 µM ( Figure 10). However, in comparison to the result obtained for the reference DX the hepatotoxicity potential of tested compound is moderate, as its decreased cells' viability to~74% of control at 10 µM, whereas DX to~15% of control at 1 µM ( Figure 10).

SAR Discussion
All synthesized compounds (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17) belong to the group of 5-arylideneimidazolones with amine moiety at position 2 (6) and after Dimroth rearrangement (Scheme 1) also at position 3 (7-17). A variety of arylidene moieties at position 5 and distinct position and kind of amines give an opportunity to evaluate impact of these fragments for the antibiotic adjuvant properties considered. Most of the tested compounds displayed significant (at least four) reduction of oxacillin MIC in MRSA HEMSA 5 strain (Table 5). In previous tests of 2-amine-5-arylideneimidazolones, only two compounds displayed a significant reduction of oxacillin MIC in the same bacterial strain. However, the decrease of antibiotic MIC was higher (64-to 128-fold) [25]. Additionally, a synergistic effect with oxacillin was not observed in the case of the susceptible strain (ATCC), what indicated an ability to selectively block mechanisms of resistance occurring in the MDR pathogen. The outcome seems to be more profitable for imidazolones with condensed aromatic rings at the position 5 (8-32-fold activity gain for 9-11, 13, 15 and 16). The same conclusion was made for previously tested 5arylideneimidazolones with unsubstituted piperazine at position 2 [25]. Comparing the 2-naphthyl derivative (9) and 1-naphthyl (15), the first one (9) was more potent. There was no difference between the adjuvant potency of biphenyl (14) and 4-phenoxyphenyl (12) derivatives. In the case of monoaromatic 5-benzylideneimidazolones, chlorine (7) in para position was more profitable than MeO. Compound 16 with an anthracene moiety and the morpholine-terminated fragment at position 3 showed the highest ability to restore oxacillin activity, and also demonstrated more potent ligandprotein interactions in the allosteric site in molecular modelling studies. The similar anthracene derivative of methylpiperazine (11) was less active, thus indicating a predominant role for morpholine. Additionally, the 5-anthracylmethylideneimidazolone derivative with a 2-piperazine moiety was active in the same MRSA strain [25]. On this account, this relationship seems to be aryldependent. In the case of phenoxybenzylidene derived imidazolones, morpholine derivative (17) were less active than the methylpiperazine analogue (12). Considering the topology of an amine moiety, the presence of aminoalkyl fragment at position 3 of imidazolone (7) seems to be significantly more favourable than the presence of the tertiary amine substituted directly at imidazolone position 2 (6).
Compounds 9-13, 16 and 17 were additionally tested for their activity in E. aerogenes strains. Although 5-arylideneimidazolone derivatives did not display antibiotic adjuvant activity in assays on their synergistic effects with chloramphenicol, erythromycin, doxycycline and norfloxacin, these compounds were able to block AcrAB-TolC efflux activity in RTE assay. The highest inhibitory efficiency (97%) demonstrated the fluorene derivative (13), while imidazolones with three fused Control-1% DMSO in the cell culture media. The statistical significance was evaluated by a one-way ANOVA, followed by Bonferroni's Comparison Test (*** p < 0.001 compared with control).

SAR Discussion
All synthesized compounds (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17) belong to the group of 5-arylideneimidazolones with amine moiety at position 2 (6) and after Dimroth rearrangement (Scheme 1) also at position 3 (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17). A variety of arylidene moieties at position 5 and distinct position and kind of amines give an opportunity to evaluate impact of these fragments for the antibiotic adjuvant properties considered. Most of the tested compounds displayed significant (at least four) reduction of oxacillin MIC in MRSA HEMSA 5 strain (Table 5). In previous tests of 2-amine-5-arylideneimidazolones, only two compounds displayed a significant reduction of oxacillin MIC in the same bacterial strain. However, the decrease of antibiotic MIC was higher (64-to 128-fold) [25]. Additionally, a synergistic effect with oxacillin was not observed in the case of the susceptible strain (ATCC), what indicated an ability to selectively block mechanisms of resistance occurring in the MDR pathogen. The outcome seems to be more profitable for imidazolones with condensed aromatic rings at the position 5 (8-32-fold activity gain for 9-11, 13, 15 and 16). The same conclusion was made for previously tested 5-arylideneimidazolones with unsubstituted piperazine at position 2 [25]. Comparing the 2-naphthyl derivative (9) and 1-naphthyl (15), the first one (9) was more potent. There was no difference between the adjuvant potency of biphenyl (14) and 4-phenoxyphenyl (12) derivatives. In the case of mono-aromatic 5-benzylideneimidazolones, chlorine (7) in para position was more profitable than MeO. Compound 16 with an anthracene moiety and the morpholine-terminated fragment at position 3 showed the highest ability to restore oxacillin activity, and also demonstrated more potent ligand-protein interactions in the allosteric site in molecular modelling studies. The similar anthracene derivative of methylpiperazine (11) was less active, thus indicating a predominant role for morpholine. Additionally, the 5-anthracylmethylideneimidazolone derivative with a 2-piperazine moiety was active in the same MRSA strain [25]. On this account, this relationship seems to be aryl-dependent. In the case of phenoxybenzylidene derived imidazolones, morpholine derivative (17) were less active than the methylpiperazine analogue (12). Considering the topology of an amine moiety, the presence of aminoalkyl fragment at position 3 of imidazolone (7) seems to be significantly more favourable than the presence of the tertiary amine substituted directly at imidazolone position 2 (6).
These results confirmed the previous findings about EPI's action in AcrAB-TolC pump in different E. coli strains in the group of imidazolones [15]. In turn, the presence of condensed aromatic rings at position 5 was beneficial either for MRSA or for E. aerogenes MDR reversal action. In previous studies, the 2-piperazine derivative of 5-(2-naphthyl)imidazolone was the most potent (128-fold reduction of oxacillin MIC) in the MRSA HEMSA 5 strain [25]. Although, taking into account results of both, the adjuvant potency in MRSA and efflux pump inhibitory action in E. aerogenes, the 2-naphthyl and methylpiperazine derivative (9) was found as the most potent "dual agent". Hence, the compound was selected as representative structure for primary ADMET assays in vitro. The obtained results allowed us to classify this compound as lowly permeable (Table 6), non-mutagenic but cytotoxic for prokaryotic cells (Table 7), and displaying slight hepatotoxic effects ( Figure 10). This rather moderate "drugability" found did not exclude a better drug-like property for other active members of the investigated imidazolones, but indicated a strong need of further studies within this interesting chemical group. Thus, new chemical modifications as well as an extension of biological-and "drugability" screening will be intentional for this new family of 3-substituted 5-arylideneimidazolones in the near future.

Real-Time Efflux Assay
In the first step, bacteria were loaded with 1,2 -dNA, a fluorescent membrane probe which is a substrate of the AcrAB-TolC efflux pump, in the presence of a well-known EPI carbonyl cyanide m-chlorophenylhydrazone (CCCP) that inactivates the pump in the pump energy-collapsing mode (H + -consuming). After the addition of compounds, at the final concentration of 100 µM, an efflux was initiated by automated injection of glucose that provides pump energy (H + -donating compound) to the final concentration of 50 µM. The fluorescence intensity was measured using a microplate reader (Tecan) with an excitation wavelength of λ ex = 370 nm and an emission wavelength of λ em = 420 nm. In order to analyze and quantitatively compare the EPI activity of compounds tested, the pre-energization fluorescence intensity was adjusted to 100 relative fluorescence units and the inhibition efficiency (IE) of each compound was calculated according to Equation (2): where ∆i 1 corresponds to the difference between the fluorescence of the dye in the presence and absence of compound tested after the addition of a source of energy for the pump (glucose) and ∆i 2 is the difference between the fluorescence of the dye in the presence and absence of compound tested before the addition of a source of energy for the pump.

In Silico Studies
The three-dimensional conformations of compounds and respective protonation states (for pH 7.0 +/− 2.0) were generated with the use of LigPrep [49]. At first, all the compounds were docked to the crystal structure of PBP2a protein (PDB code: 3ZFZ [50]). Following the suggested mechanism of interaction of compounds via the allosteric modulation of this target [50,51], the studied compounds were docked to the active and allosteric sites of the protein (with grids were centered at S403 and S240, respectively; the docking was performed in Glide [52,53], and the compounds were docked in extra precision).
The poses with the best Glide docking score were use as starting points for molecular dynamic (MD) simulations. MD simulations were performed in Desmond [54,55], using TIP3P solvent model [56] and lasted 100 ns.
The interactions between ligands and PBP2a protein were analyzed manually and with the use of the Simulation Interaction Diagram from the Schrodinger Suite.

Membrane Permeability
Pre-coated PAMPA Plate System Gentest™ was purchased from Corning (Tewksbury, MA, USA). The tested compound 9 and the references (200 µM) were prepared first in PBS buffer (pH = 7.4) and added to the donor wells of PAMPA Plate System. PBS was added to the acceptor wells. The plate was incubated for 5 h at room temperature. The compounds' as well as references' concentrations in acceptor and donor wells were estimated by the UPLC-MS analyses with use of internal standard. The permeability coefficients (Pe, cm/s) were calculated according to formulas provided by the manufacturer.

Safety
Ames microplate fluctuation protocol (MPF) assay was obtained from Xenometrix AG (Allschwil, Switzerland). The occurrence of mutagen-induced or spontaneous reversion events to histidine prototrophy was determined as a growth of S. typhimurium TA100 in the indicator medium without histidine after 72 h incubation at 37 • C temperature. NQNO (0.5 µM) was used as a positive control. Compound 9 was tested in triplicate at the final concentrations 1 and 10 µM. The revertants' growth induced the colour change of medium which was analysed next colourimetrically with a microplate reader (EnSpire, PerkinElmer, Waltham, MA, USA) at 420 nm.
The HepG2 cells were seeded in 96-well plates at a concentration of 1 × 10 4 cells/well and incubated for 24 h at 37 • C in 5% CO 2 atmosphere to reach 50% of confluence. Compound 9 was diluted into fresh growth medium and added to the cells at the final concentrations 0.1 µM-100 µM. The positive controls DX and CCCP were added at 1 µM and 10 µM, respectively, and the cells were incubated for 72 h. The MTS reagent (CellTiter 96 ® AQueous One Solution Cell Proliferation Assay, Promega, Madison, WI, USA) was added next to the each well and incubated for 4 h. The absorbance was measured next using a microplate reader (EnSpire, PerkinElmer, Waltham, MA USA) at 490 nm to determine cells' viability.

Conclusions
In order to search for successful antibiotic adjuvants, useful against both Gram-positive and Gram-negative MDR pathogens, the performed chemical modifications of the lead structure (Z)-5-(4-chlorobenzylidene)-2-(4-methylpiperazin-1-yl)-3H-imidazol-4(5H)-one (6) have provided a series of new, and even unexpected, chemical structures due to Dimroth rearrangements within the imidazolone ring. Most of those original compounds significantly improved oxacillin activity in MRSA, while they did not influence the activity of oxacillin in the reference S. aureus strain, suggesting an impact on MDR mechanisms associated with PBP2a. Docking and molecular dynamic simulations have confirmed this hypothesis, indicating the ability of the most active compounds to interact with the allosteric site of PBP2a, and to enhance the binding of oxacillin to the active site. On the other hand, results of the RTE assay for the new 5-arylideneimidazolones towards AcrAB-TolC in E. aerogenes (EA289) have confirmed their potent capacity to inhibit this important MDR efflux system. The performed SAR analysis indicated the anthracene-morpholine derivative (16) as the most active oxacillin adjuvant in the MRSA, and the naphthalene-methylpiperazine imidazolone (9) as the most potent "dual-action" compound displaying both oxacillin potentiating action in MRSA in the range of compound 16 and efflux pump inhibitory properties in E. aerogenes. The primary ADMET screening in vitro for 9 showed a rather moderate "drugability" profile for this active compound. In the context of the obtained results, further comprehensive studies for this interesting chemical family are needed. Overall, compound 9, and then 16, seem to be new lead structures for further modifications in order to discover therapeutically useful adjuvants, able to restore the effectiveness of common antibiotics against MDR pathogens.