Structural Modifications of the Quinolin-4-yloxy Core to Obtain New Staphylococcus aureus NorA Inhibitors

Tackling antimicrobial resistance (AMR) represents a social responsibility aimed at renewing the antimicrobial armamentarium and identifying novel therapeutical approaches. Among the possible strategies, efflux pumps inhibition offers the advantage to contrast the resistance against all drugs which can be extruded. Efflux pump inhibitors (EPIs) are molecules devoid of any antimicrobial activity, but synergizing with pumps-substrate antibiotics. Herein, we performed an in silico scaffold hopping approach starting from quinolin-4-yloxy-based Staphylococcus aureus NorA EPIs by using previously built pharmacophore models for NorA inhibition activity. Four scaffolds were identified, synthesized, and modified with appropriate substituents to obtain new compounds, that were evaluated for their ability to inhibit NorA and synergize with the fluoroquinolone ciprofloxacin against resistant S. aureus strains. The two quinoline-4-carboxamide derivatives 3a and 3b showed the best results being synergic (4-fold MIC reduction) with ciprofloxacin at concentrations as low as 3.13 and 1.56 µg/mL, respectively, which were nontoxic for human THP-1 and A549 cells. The NorA inhibition was confirmed by SA-1199B ethidium bromide efflux and checkerboard assays against the isogenic pair SA-K2378 (norA++)/SA-K1902 (norA-). These in vitro results indicate the two compounds as valuable structures for designing novel S. aureus NorA inhibitors to be used in association with fluoroquinolones.


Introduction
Antimicrobial resistance (AMR) represents a complex global health challenge due to its natural insurgence and rapid spread caused by the use and misuse of antimicrobial agents in humans and animals [1][2][3][4]. The magnitude of the problem worldwide, the impact of AMR on human health and on costs for the health-care sector are still largely unknown [5]. Estimations report that yearly in Europe about 31100 people die as a result of multidrug-resistant (MDR) microbial infections, burdening on linked to the oxygen at C-4. The new virtual library was analysed by performing virtual screening experiments using our previously constructed pharmacophore models for NorA inhibitors to select the most interesting derivatives [35]. Only the scaffolds able to (i) correctly reproduce the mutual positions of the two above mentioned substituents and (ii) accept the two chemical functionalities in different positions were retained, thus generating about 6000 new virtual compounds. Four different scaffolds functionalized with the chemical key requirements for NorA inhibition have been selected based on the fitness values on the pharmacophore models and the chemical accessibility. In particular, the four selected scaffolds were used to synthesize eight new derivatives (3a, 3b, 4a, 4b, 5a, 5b, 6a and 6b- Figure 1) that were biologically evaluated as NorA EPIs.

In Silico Scaffold Hopping
We have recently developed two common-features pharmacophore models (hereafter called ModB and ModC) for NorA EPIs, which allowed the identification of FDA-approved drugs endowed with potent inhibitory activity [35]. Among the 3D chemical features, a positive charge appeared to be the key element in the discrimination between active and inactive compounds for both models. It should be noted that many quinoline derivatives were used to develop and validate the pharmacophore hypothesis, underlining an important role of the protonable moieties located in position 4 of the quinoline scaffold, with the ethyl-N,N-diethylamine (e.g., compound 1) and 1ethylpiperidine (e.g., compound 2) groups being generally associated with the most interesting results in terms of EPI activity and physico-chemical properties. Herein, to enrich the array of NorA inhibitors, we performed a scaffold hopping strategy of the quinolin-4-yloxy backbone, using as template the starting hit 1 (Figure 1) [33]. Combining scaffolds extracted by Food and Drug Administration (FDA) approved drugs, we built a scaffold library to replace quinolin-4-yloxy core and introduced the chemical substituents that in our quinolines gave the best NorA inhibition, i.e., the propoxyphenyl group at C-2 position and the alkylamino chain linked to the oxygen at C-4. The new virtual library was analysed by performing virtual screening experiments using our previously constructed pharmacophore models for NorA inhibitors to select the most interesting derivatives [35]. Only the scaffolds able to (i) correctly reproduce the mutual positions of the two above mentioned substituents and (ii) accept the two chemical functionalities in different positions were retained, thus generating about 6000 new virtual compounds. Four different scaffolds functionalized with the chemical key requirements for NorA inhibition have been selected based on the fitness values on the pharmacophore models and the chemical accessibility. In particular, the four selected scaffolds were used to synthesize eight new derivatives (3a, 3b, 4a, 4b, 5a, 5b, 6a and 6b- Figure 1) that were biologically evaluated as NorA EPIs.

In Silico Scaffold Hopping
We have recently developed two common-features pharmacophore models (hereafter called ModB and ModC) for NorA EPIs, which allowed the identification of FDA-approved drugs endowed with potent inhibitory activity [35]. Among the 3D chemical features, a positive charge appeared to be the key element in the discrimination between active and inactive compounds for both models. It should be noted that many quinoline derivatives were used to develop and validate the pharmacophore hypothesis, underlining an important role of the protonable moieties located in position 4 of the quinoline scaffold, with the ethyl-N,N-diethylamine (e.g., compound 1) and 1-ethylpiperidine (e.g., compound 2) groups being generally associated with the most interesting results in terms of EPI activity and physico-chemical properties.
Based on this background, compound 1 was selected as quinoline-representative leading structure for our scaffold hopping approach (Figure 2, panel A1), where new cores were selected from a library composed by 1456 small molecules (MW < 300) arising from the smart fragmentation of approved drugs [36].
highlighted the presence of a fragment derived from the approved drug cinchocaine (the fragment is highlighted in blue in Figure 2, panel B). This compound shared a high chemical similarity with 1, thus inspiring the modification of the ether linker in the quinoline compound with an amide to generate derivative 3a. The latter compound was added to the previously described database to make up the scaffold hopping set, which was filtered in several rounds according to a funnel-like approach, each time discarding the molecules that did not meet the required criteria ( Figure 2, panel C).
In the first selection procedure, in silico absorption, distribution, metabolism and excretion (ADME) filters (see Material and methods) were applied to exclusively collect drug-like molecules. Second, the output compounds (1089 derivatives) were in turn virtually analyzed in Phase [38,39] by using the two pharmacophore models (i.e., ModB and ModC) as queries. Figure 2. Schematic overview of applied in silico workflow. Panel A: two scaffold hopping runs were performed. In each run, the scaffold (highlighted in green) was virtually replaced using fragments derived by FDA-approved drugs. Panel B: drug-derived quinoline fragments were checked. The cinchocaine fragment Prestw-frag-3758 was highlighted in blue. This structure suggested the replacement of the ether linker in the quinoline family (e.g., inhibitor 1) with an amide (represented in red). Panel C: virtual screening process and selected virtual hits. Virtual hits 4a, 5a and 6a derived from the Prestwick fragments Prestw-frag-1560, Prestw-frag-2136 and Prest-frag-2303, respectively, Figure 2. Schematic overview of applied in silico workflow. Panel A: two scaffold hopping runs were performed. In each run, the scaffold (highlighted in green) was virtually replaced using fragments derived by FDA-approved drugs. Panel B: drug-derived quinoline fragments were checked. The cinchocaine fragment Prestw-frag-3758 was highlighted in blue. This structure suggested the replacement of the ether linker in the quinoline family (e.g., inhibitor 1) with an amide (represented in red). Panel C: virtual screening process and selected virtual hits. Virtual hits 4a, 5a and 6a derived from the Prestwick fragments Prestw-frag-1560, Prestw-frag-2136 and Prest-frag-2303, respectively, that are highlighted in red. The chemical features of ModB and ModC are shown in green and yellow, respectively. a Fitness values.
The first core hopping run was aimed at replacing only the quinoline scaffold, while the two moieties on position 2 and 4 of the bicyclic system were completely preserved (run1: Figure 2, panels A1 and A2). In the second scaffold hopping experiment, the core definition included the quinoline nucleus together with the oxygen atom at position 4; this approach has been pursued with the aim of increasing the chemical diversity of the explored scaffolds (run2: Figure 2, panels A1 and A2).
The two independent scaffold hopping rounds were carried out using the ligand-based core hopping utility in Schrodinger [37]. The corresponding generated libraries were merged to obtain 6393 non redundant compounds defined as scaffold hopping set ( Figure 2, panel C).
In parallel, a substructure search among the 1456 fragments using the quinoline core as query highlighted the presence of a fragment derived from the approved drug cinchocaine (the fragment is highlighted in blue in Figure 2, panel B). This compound shared a high chemical similarity with 1, thus inspiring the modification of the ether linker in the quinoline compound with an amide to generate derivative 3a. The latter compound was added to the previously described database to make up the scaffold hopping set, which was filtered in several rounds according to a funnel-like approach, each time discarding the molecules that did not meet the required criteria ( Figure 2, panel C).
In the first selection procedure, in silico absorption, distribution, metabolism and excretion (ADME) filters (see Material and methods) were applied to exclusively collect drug-like molecules. Second, the output compounds (1089 derivatives) were in turn virtually analyzed in Phase [38,39] by using the two pharmacophore models (i.e., ModB and ModC) as queries.
From each pharmacophore screening, only compounds showing a fitness score ≥1.7 were retained (723 and 292 compounds for ModB and ModC, respectively) and analyzed through a consensus approach by selecting only the compounds fitting both models. Additionally, compounds with a fitness higher than 2.0 for at least one of the two models were kept as well.
Third, the resulting compounds (167 molecules) were visually inspected looking for new compounds that had a novel scaffold and only the essential pharmacophore features.
Taking into account the in silico results and an acceptable synthetic accessibility, four virtual hits (3a-6a, Figures 1 and 2), each one having a different core, were considered worthy of experimental investigations as potential NorA EPIs. In particular, the virtual hit 6a was designed starting from the scaffold hopping derivative 7 ( Figure 2, panel C). The latter compound had a hydroxy-methyl group in position 4 of the pyridine ring. However, this substituent was not a pharmacophore element in the two models developed for NorA EPIs, and moreover its presence lowered the synthetic accessibility of the proposed compound. For this reason, the hydroxy-methyl group was removed retaining only the pyridine core to provide compound 6a.
Furthermore, an additional set of 4 new compounds was designed by replacing the ethyl-N,N-diethylamine at position 4 in the previous series (a) with the 1-ethylpiperidine (3b-6b, Figure 1), that was the other well-characterized and promising chain in the quinoline chemical family (e.g., compound 2).

Chemistry
Synthetic procedures of the planned compounds 3a, 3b, 4a, 4b, 5a, 5b, 6a and 6b have been reported in Schemes 1-4. From each pharmacophore screening, only compounds showing a fitness score ≥1.7 were retained (723 and 292 compounds for ModB and ModC, respectively) and analyzed through a consensus approach by selecting only the compounds fitting both models. Additionally, compounds with a fitness higher than 2.0 for at least one of the two models were kept as well.
Third, the resulting compounds (167 molecules) were visually inspected looking for new compounds that had a novel scaffold and only the essential pharmacophore features.
Taking into account the in silico results and an acceptable synthetic accessibility, four virtual hits (3a-6a, Figures 1 and 2), each one having a different core, were considered worthy of experimental investigations as potential NorA EPIs. In particular, the virtual hit 6a was designed starting from the scaffold hopping derivative 7 ( Figure 2, panel C). The latter compound had a hydroxy-methyl group in position 4 of the pyridine ring. However, this substituent was not a pharmacophore element in the two models developed for NorA EPIs, and moreover its presence lowered the synthetic accessibility of the proposed compound. For this reason, the hydroxy-methyl group was removed retaining only the pyridine core to provide compound 6a.
Furthermore, an additional set of 4 new compounds was designed by replacing the ethyl-N,Ndiethylamine at position 4 in the previous series (a) with the 1-ethylpiperidine (3b-6b, Figure 1), that was the other well-characterized and promising chain in the quinoline chemical family (e.g., compound 2).

Compd.
SA-1199 SA-1199B 3a 25 25 16 5 None of the investigated compounds showed antibacterial activity against both S. aureus SA-1199 and SA-1199B at concentrations ≤25 µg/mL. Moreover, no significant difference can be observed between MIC values against SA-1199 and SA-1199B, the latter characterized by a higher amount of NorA protein than SA-1199. Thus, indirectly these data suggested that all compounds were not substrate of NorA efflux pump; indeed, a significant change in MIC values between SA-1199 and SA-1199B would be expected for a NorA substrate, as observed for CPX. Since all compounds showed MIC values ≥ 25 µg/mL, we proceeded to evaluate the synergistic effect of the derivatives at 12.5 µg/mL (a concentration ≤ 1 2 MIC for all compounds) in combination with scalar concentrations of CPX against both S. aureus strains (Figure 3). MIC values of CPX when tested alone against SA-1199 and SA-1199B can be found in Table 1  We considered promising EPIs those compounds that yielded a ≥4-fold decrease in the CPX MIC against SA-1199B and no significant change (≤2 times) against SA-1199. This difference in the synergistic activity with CPX between the two S. aureus strains is essential to display that the activity of compounds is related to NorA inhibition. Since norA gene is overexpressed in SA-1199B and normally expressed in SA-1199, compounds acting on NorA should possess a significant synergistic activity with CPX only against SA-1199B. Evident synergism also against SA-1199 could be likely due to a NorA-independent effect such as disruption, permeabilization or depolarization of the bacterial membrane, promoting the CPX penetration into bacterial cells.
Focusing the attention on SAR, it was evident that the replacement of the quinolin-4-yloxy scaffold of 1 with the phthalazinone core (compounds 5a and 5b) led to a loss of the synergistic activity with the CPX against both S. aureus strains. Similarly, when the benzene moiety of the quinoline of 1 was removed to give the pyridine derivatives 6a and 6b, the synergistic activity was lost. On the other hand, quinoline-4-carboxamide derivatives (3a and 3b) and only the benzimidazole 4a displayed promising results by reducing, when tested at 12.5 µg/mL, the CPX MIC by 4-fold against SA-1199B while not producing a significant decrease in the CPX MIC against SA-1199. The lack of activity of the benzimidazole analogue 4b was of more difficult interpretation. However, based on these results, we considered 3a, 3b and 4a as interesting derivatives deserving further investigations. Thus, checkerboard assays were scheduled for all three derivatives in combination with CPX against SA-1199B (Figure 4). All three compounds showed a dose-dependent synergistic effect in combination with CPX, Figure 3. Synergistic assays of derivatives 3a, 3b, 4a, 4b, 5a, 5b, 6a and 6b and starting hit 1 at 12.5 µg/mL in combination with scalar concentrations of CPX against SA-1199B (red columns) and SA-1199 (blue columns).
We considered promising EPIs those compounds that yielded a ≥4-fold decrease in the CPX MIC against SA-1199B and no significant change (≤2 times) against SA-1199. This difference in the synergistic activity with CPX between the two S. aureus strains is essential to display that the activity of compounds is related to NorA inhibition. Since norA gene is overexpressed in SA-1199B and normally expressed in SA-1199, compounds acting on NorA should possess a significant synergistic activity with CPX only against SA-1199B. Evident synergism also against SA-1199 could be likely due to a NorA-independent effect such as disruption, permeabilization or depolarization of the bacterial membrane, promoting the CPX penetration into bacterial cells.
Focusing the attention on SAR, it was evident that the replacement of the quinolin-4-yloxy scaffold of 1 with the phthalazinone core (compounds 5a and 5b) led to a loss of the synergistic activity with the CPX against both S. aureus strains. Similarly, when the benzene moiety of the quinoline of 1 was removed to give the pyridine derivatives 6a and 6b, the synergistic activity was lost. On the other hand, quinoline-4-carboxamide derivatives (3a and 3b) and only the benzimidazole 4a displayed promising results by reducing, when tested at 12.5 µg/mL, the CPX MIC by 4-fold against SA-1199B while not producing a significant decrease in the CPX MIC against SA-1199. The lack of activity of the benzimidazole analogue 4b was of more difficult interpretation. However, based on these results, we considered 3a, 3b and 4a as interesting derivatives deserving further investigations. Thus, checkerboard assays were scheduled for all three derivatives in combination with CPX against SA-1199B ( Figure 4).
All three compounds showed a dose-dependent synergistic effect in combination with CPX, exhibiting an activity similar or greater than that of the starting hit 1. In particular, the worst of the three compounds (4a) retained a significant synergistic effect at concentrations ≥ 12.5 µg/mL, similarly to the starting hit 1. On the other hand, carboxamide derivatives 3a and 3b exhibited better results, reducing the CPX MIC by 4-fold at concentrations as low as 3.13 and 1.56 µg/mL, respectively. These results are in agreement with the data obtained from EtBr efflux inhibition assays. Indeed, the carboxamide derivatives 3a and 3b showed a stronger ability to reduce the EtBr efflux on SA-1199B than the benzimidazole analogue 4a (Table 2), with reductions up to 96-98% when used at 50 µM. 4a displayed promising results by reducing, when tested at 12.5 µg/mL, the CPX MIC by 4-fold against SA-1199B while not producing a significant decrease in the CPX MIC against SA-1199. The lack of activity of the benzimidazole analogue 4b was of more difficult interpretation. However, based on these results, we considered 3a, 3b and 4a as interesting derivatives deserving further investigations. Thus, checkerboard assays were scheduled for all three derivatives in combination with CPX against SA-1199B (Figure 4). All three compounds showed a dose-dependent synergistic effect in combination with CPX, exhibiting an activity similar or greater than that of the starting hit 1. In particular, the worst of the  Overall, the obtained results show that the quinoline-4-carboxamide scaffold of 3a and 3b can efficiently replace the quinolin-4-yloxy nucleus, while the benzimidazoles resulted less interesting. In addition, data suggested that both derivatives 3a and 3b inhibited the NorA efflux pump without nonspecific effects. Indeed, synergism against the norA-overexpressing strain (SA-1199B) was observed for both compounds at concentration as low as 3.13 and 1.56 µg/mL (Figure 4) while no effect was detected against the wild type strain (SA-1199) using concentrations up to 12.5 µg/mL (Figure 3). The further demonstration that NorA could be inhibited by carboxamide analogues 3a and 3b was confirmed by their high degree of inhibition of the EtBr efflux observed through the EtBr efflux assays on SA-1199B. However, to rule out any doubt related to the mechanism of action of both derivatives, checkerboard assays were performed even against two specific S. aureus strains, SA-K1902 (norA-) and SA-K2378 (norA++). Since these two engineered strains are different only for the presence and expression level of the norA gene [43], it is expected that compounds that show a synergism with CPX only against SA-K2378 may inhibit the NorA efflux pump. On the other hand, those compounds having a significant synergism with CPX against both engineered strains should not be considered as NorA EPIs because they exhibit a synergistic activity not dependent on the presence of the NorA pump.
Once excluded any antibacterial activity (MIC >25 µg/mL against SA-K2378 and SA-K1902), checkerboard assays against SA-K2378 (norA++) ( Figure 5A) for compounds 3a and 3b highlighted, up to the lowest concentrations used (0.39 µg/mL), a significant synergism with CPX. On the other hand, as expected for NorA inhibitors, no significant synergistic effect with CPX was observed against SA-K1902 (norA-) ( Figure 5B), thus confirming their "pure" activity against NorA. Remarkably, as a further demonstration that 3a and 3b were able to inhibit NorA, the CPX MIC against SA-K2378 in combination with EPIs never dropped below the MIC of the CPX tested alone against SA-K1902. This data is essential to prove that the synergistic activity of 3a and 3b serves to boost the CPX activity through the fully inhibition of the NorA efflux restoring the CPX MIC up to the levels observed in SA-K1902 in which norA is deleted.
hand, as expected for NorA inhibitors, no significant synergistic effect with CPX was observed against SA-K1902 (norA-) ( Figure 5B), thus confirming their "pure" activity against NorA. Remarkably, as a further demonstration that 3a and 3b were able to inhibit NorA, the CPX MIC against SA-K2378 in combination with EPIs never dropped below the MIC of the CPX tested alone against SA-K1902. This data is essential to prove that the synergistic activity of 3a and 3b serves to boost the CPX activity through the fully inhibition of the NorA efflux restoring the CPX MIC up to the levels observed in SA-K1902 in which norA is deleted. Finally, we evaluated both compounds at 3.13 µg/mL, a concentration at which they reduced CPX MIC by 4-fold against SA-1199B, for their cytotoxic activity against human THP-1 and A549 (CCL-185 TM ) cell lines. Both exhibited a vitality higher than 50%, specifically THP-1 72% (3a) and 62% (3b) and A549 about 100% in presence of both compounds. In addition, it should be considered that SA-1199B is doubly resistant to CPX both for a single mutation on its target (DNA gyrase) and the overexpression of norA gene. Therefore, considering only the effect of the NorA inhibition observable on SA-K2378, both compounds at 0.39 µg/mL reduced the CPX MIC by 4-fold, thus showing that the inhibition of NorA occurs at concentrations significantly lower than those cytotoxic for human cells.

In Silico Scaffold Hopping
The Prestwick drug-fragment library was downloaded and submitted to LigPrep [44]. The neutral form of the ligands was prepared, and the tautomeric states was generated using Epik [45,46]. Furthermore, at most 32 stereoisomers per ligand and three lowest energy conformations per ligand ring were produced. Where not defined, all the chiral form of each stereocenter was produced.
The ligand-based core hopping utility within Schrodinger was used to generate the scaffold-hopping libraries [37]. The input cores were generated starting from the prepared Prestwick drug-fragment library using the corefinder utility of Schrodinger [37], treating the entire fragment as core.
The cores were filtered using the following criteria: number of heavy atoms ≤ 15; number of hydrogen bond acceptor ≤ 8; number of hydrogen bond donor ≤ 4; number of N+O ≤ 10; number of chiral centers = 0.
The remaining compounds were submitted to a conformational search using MacroModel [48]. To enhance the conformational sampling, the maximum number of steps was set to 10,000 per molecule. Conformers in an energy window of 5 kcal/mol were saved, discarding the redundant ones on the basis of their atomic rmsd (0.5 Å cutoff). Finally, the obtained conformers were screened in Phase [38,39] using ModB and ModC as queries.

Chemistry
All starting materials, reagents and solvents were purchased from common commercial suppliers and were used as such, unless otherwise indicated. Organic solutions were dried over anhydrous Na 2 SO 4 and concentrated with a rotary evaporator at low pressure. The reactions carried out under MW irradiation were performed employing a microwave reactor BIOTAGE INITIATOR 2.0 version 2.3, build 6250. All reactions were routinely checked by thin-layer chromatography (TLC) on silica gel 60 F254 (Merck) and visualized by using UV or iodine. Flash chromatography separations were carried out on Merck silica gel 60 (mesh 230-400) or by BUCHI Reveleris ® X2 Flash Chromatography (BÜCHI Labortechnik AG, Flawil, Switzerland). Melting points were determined in capillary tubes (Stuart SNP30, Stewart Italia, Milan, Italy) and are uncorrected. Yields were of purified products and were not optimized. 1 H NMR spectra were recorded at 200 or 400 MHz (Bruker Avance DRX-200 or 400, respectively (Bruker Corporation, Massachusetts, USA)), while 13 C NMR spectra were recorded at 101 MHz (Bruker Avance DRX-400). Chemical shifts are given in ppm (δ) relative to TMS. Spectra were acquired at 298 K. Data processing was performed with standard Bruker software XwinNMR (3.0) and the spectral data are consistent with the assigned structures. The purity of the tested compounds (≥95% sample purity) was evaluated by HPLC analysis using a Jasco LC-4000 instrument equipped with a UV-Visible Diode Array Jasco MD-4015 (Jasco Corporation, Tokyo, Japan) and an XTerra MS C18 Column, 5 µm, 4.6 mm × 150 mm (Waters Corporation, Massachusetts, USA). Chromatograms were analysed by ChromNAV 2.0 Chromatography Data System software.

Antimicrobial Susceptibility Assays
The antimicrobial activity of the compounds and of their combinations with CPX was evaluated by broth microdilution MIC determination against the isogenic pair SA-1199 and SA-1199B using doubling concentrations of the tested drug, according to the CLSI guidelines [49]. Checkerboard assays were performed as previously described [28], considering synergistic the combinations leading to a ≥4-fold reduction of the CPX MIC [50]. All assays were performed in duplicate; when the results did not overlap, the higher MIC value was reported.

Cytotoxicity Assays
The cytotoxic effect of the compounds was determined by MTT assays performed on THP-1 and A549 (CCL-185 TM ) cells after 24 h exposure [51]. Cells were seeded at the density of 1 × 10 4 cells/well into 96-well flat-bottomed plates in 200 µL of RPMI or DMEM/F12 medium, respectively, and then exposed for 24 h to the compounds at the concentrations resulted synergistic in checkerboard assays. Unexposed cells were used as negative control. The colorimetric MTT assay allowed to measure the cell growth rates through the amount of the accumulated intracellular insoluble formazan crystals, subsequently dissolved using DMSO and quantified spectrophotometrically (OD 570 ), using a microplate reader (BioTeK, Winooski, VT, USA). The obtained data were analyzed by the software Gen05 v3. The percentage of viable cells was calculated as follows: % Cell viability = 100 × Experimental well absorbance/untreated control well absorbance. All assays were performed in biological and technical duplicate.

Conclusions
In this work, in silico scaffold-hopping approaches combined with 3D-pharmacophore screening followed by chemical synthesis and validated biological methods allowed for the identification of two quinoline-4-carboxamide derivatives (3a and 3b) as new NorA EPIs. Compounds 3a and 3b were able to synergize at low concentrations with CPX against the norA overexpressing S. aureus strains SA-1199B and SA-K2378. We also proved that the synergistic effect of our EPIs was due to a NorA inhibition, indirectly ruling out potential nonspecific effects by EtBr efflux inhibition on SA-1199B and checkerboard assays on the isogenic pairs SA-K2378 (norA++) and SA-K1902 (norA-). Moreover, through the synthesis and biological evaluation of different compounds, we observed some important SAR information revealing that the quinoline core is essential to retain NorA inhibition; however, functionalization at the C-4 position of this nucleus with a carboxamide moiety (instead of an ethereal function as for the starting hit 1) led to an improvement of both the synergistic activity with CPX and the cytotoxicity profile against human cell lines. Based on these promising results, virtual hits that have been discarded from this work, because they are less synthetically accessible but promising in terms of fitness scores, will be considered for future experimental validations.