Efflux Pump Overexpression Profiling in Acinetobacter baumannii and Study of New 1-(1-Naphthylmethyl)-Piperazine Analogs as Potential Efflux Inhibitors

ABSTRACT

Overexpression of efflux pumps extruding antibiotics currently used for the treatment of Acinetobacter baumannii infections has been described as an important mechanism causing antibiotic resistance. The first aim of this work was to phenotypically evaluate the overexpression of efflux pumps on a collection of 124 ciprofloxacin-resistant A. baumannii strains. An overexpression of genes encoding one or more efflux pumps was obtained for 19 out of the 34 strains with a positive phenotypic efflux (56%). The most frequent genes overexpressed were those belonging to the RND family, with adeJ being the most prevalent (50%). Interestingly, efflux pump genes coding for MATE and MFS families were also overexpressed quite frequently: abeM (32%) and abaQ (26%). The second aim was to synthesize 1-(1-naphthylmethyl)-piperazine analogs as potential new efflux pump inhibitors and biologically evaluate them against strains with a positive phenotypic efflux. Quinoline and pyridine analogs were found to be more effective than their parent compound, 1-(1-naphthyl methyl)-piperazine. Stereochemistry also played an important part in the inhibitory activity, as quinoline derivative (R)-3a was identified as being the most effective and less cytotoxic. Its inhibitory activity was also correlated with the number of efflux pumps expressed by a strain. The results obtained in this work suggest that quinoline analogs of 1-(1-naphthylmethyl)-piperazine are promising leads in the development of new anti-Acinetobacter baumannii therapeutic alternatives in combination with antibiotics for which an efflux-mediated resistance is suspected.

INTRODUCTION

Over the past few decades, the widespread and inappropriate use of broad-spectrum antibiotics in both human and veterinary medicines strongly contributed to the emergence of highly resistant bacterial strains. The most common multidrug-resistant (MDR) pathogens have been grouped under the acronym ESKAPEE, which stands for Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter spp., and Escherichia coli (1–3). These pathogens have been identified as the leading cause of MDR nosocomial infections. Among them, A. baumannii has come out as one of the main causes of hospital-acquired infections, especially in intensive care units and in immunocompromised patients (4, 5). A. baumannii is especially involved in respiratory tract infections such as severe ventilator-associated pneumonia, bacteremia, and skin/wound infections, while rare cases of nosocomial meningitis (mostly postoperative meningitis) have been documented (6). In 2017, the World Health Organization (WHO) ranked A. baumannii as the most critical antibiotic-resistant bacteria, against which it is vital to develop new and effective antibiotic therapies (7). Indeed, treatment of Acinetobacter infections is increasingly difficult, since this bacterium survives extremely well in the environment and has a remarkable ability to disseminate. Moreover, A. baumannii has a plastic genome (8), which significantly contributes to the acquisition and dissemination of multiple resistance mechanisms in response to antibiotic pressure (9, 10). This dissemination occurs through horizontal gene transfers via mobile elements such as plasmids, integrons, transposons, and resistance islands (11) and is the main factor involved in the emergence of MDR strains (12). Among the other main mechanisms of antibiotic resistance described, the overexpression of efflux pumps (EPs) also plays a significant role in multidrug resistance (4, 13).

Active EPs are widely distributed, especially in non-glucose-fermenting Gram-negative rods (14). They also have a wide variety of substrates (antibiotics, dyes, biocides, detergents, antiseptics, etc.) (15). In A. baumannii, six main EP superfamilies coexist. Most of them are chromosomally encoded, except for some EPs belonging to the major facilitator superfamily (MFS) or the small multidrug resistance (SMR) family. According to their protein sequences, EPs are classified into the ATP-binding cassette (ABC) transporters, the SMR family, the MFS, the multidrug and toxic compound extrusion (MATE) transporters, the resistance-nodulation-division (RND) superfamily, or the PACE (proteobacterial antimicrobial compound efflux) family (16). While most EPs are constituted by a single protein encased in the inner membrane, the members of the RND family are composed of three parts. Moreover, these active transporters use different energy sources. The ABC family relies on ATP to provide energy, while the others act as antiports, coupling the drug efflux to the downhill transport of sodium ions (MATE family) or protons (SMR, MFS, and RND families) along a concentration gradient (17). From the five superfamilies of pumps, RND systems have been described as the most important ones in multiresistant A. baumannii strains (12). Indeed, most papers dealing with the study of antibiotic resistance mediated by efflux overexpression are reporting on RND efflux pumps (18–20). The contribution to antibiotic resistance of EPs belonging to other families is seldom described for A. baumannii. Therefore, the first aim of this work was to study the overexpression of EPs belonging to both the RND family and the MFS, MATE, and SMR ones in a collection of old and recent ciprofloxacin-resistant A. baumannii strains isolated in Amiens University Hospital (Amiens, France). This overexpression is evaluated by reverse transcription-quantitative PCR (RT-qPCR) on strains displaying a phenotypic efflux characterized by using two well-known efflux pump inhibitors (EPIs), 1-(1-naphthylmethyl)-piperazine (NMP) and phenylalanine-arginine-β-naphthylamide (PAβN) (Fig. 1) (21–23).

FIG 1.

NMP, PAβN, mefloquine, enpiroline, and newly designed QPE (S)- or (R)-3a and PPE (S)- or (R)-3b.

As the development of EPIs to restore antibiotic activity on resistant strains could be the key to new helpful treatments (24), the second aim of this study was to synthetize new NMP analogs possessing a quinoline or a pyridine core as potential inhibitors. These new NMP analogs are afterwards evaluated on a panel of A. baumannii strains found to display a phenotypic efflux. The assessment of their cytotoxicity is also carried out to check their usefulness in the treatment of human A. baumannii infections.

RESULTS

Bacterial strains and antibiotic susceptibility testing.

A total of 130 nonredundant strains were included in this study. The majority (n = 100) were isolated from screening (n = 51, 39%) and respiratory (n = 49, 38%) samples. Other origins were less frequent: urine (n = 9), wound (n = 8), blood (n = 7), intravenous line (n = 2), feces (n = 2), and body fluids (n = 2) (see Table S1 in the supplemental material).

Among this collection, six isolates were pan-susceptible, 2 were classified as non-multidrug resistant (non-MDR), 25 as MDR, and 97 as extensively drug resistant (XDR) based on the criteria developed by Magiorakos et al. (25) (Table S1).

High resistance rates were observed in the three main families of anti-Gram-negative antibiotics. Most strains (73%) were resistant to both imipenem and meropenem, 95% were resistant to both ciprofloxacin and levofloxacin (i.e., all strains except the 6 pan-susceptible ones), and 78% were resistant to amikacin, tobramycin, and gentamicin (Table S1).

Phenotypic efflux determination.

On the 130 tested strains, 34 (26%) gave a positive result (efflux factor, F [MIC without EPI/MIC with EPI], of 4 or above) for at least one EPI (NMP and/or PAβN). Ciprofloxacin MICs with and without EPI (NMP or PaβN, used at a fixed final concentration of 50 μg/ml) and the resulting calculated Fs for the 34 efflux-positive strains are summarized in Table 1. MICs for NMP and PAβN were superior to 200 μg/ml for all strains except for ABAM 128 (100 μg/ml for both NMP and PAβN) and ABAM 144 (100 μg/ml for NMP). Results obtained with strain ABAM 118 were also included in this table, as this strain was previously characterized by the French National Reference Centre for Antimicrobial Resistance (NRC-AR) as overexpressing the AdeABC efflux system. Among the 34 isolates, 11 gave a positive result for both NMP and PAβN, 11 only for NMP, and 12 only for PAβN.

TABLE 1.

Phenotypic efflux factors and EP gene overexpression of the 34 efflux positive strains, ABAM 118, and reference strain ABAM 113^c

Strain	Ciprofloxacin MIC (μg·ml−1)			Phenotypic Fa		Pump gene overexpressed
	Without EPI	+ NMPb	+ PAβNb	NMP	PAβN
ABAM 7	256	64	128	4	2	None
ABAM 9	512	128	128	4	4	adeB, adeJ
ABAM 12	512	128	128	4	4	adeB, adeJ
ABAM 14	128	32	16	4	8	adeB, adeG, adeJ, adeR, abeM, abaQ, craA
ABAM 15	256	128	64	2	4	adeJ, adeR
ABAM 16	128	32	64	4	2	adeJ
ABAM 25	512	256	128	2	4	adeB, adeJ
ABAM 26	128	16	64	8	2	adeG, adeJ, adeR, abeM, abaQ, craA
ABAM 27	256	64	128	4	2	adeJ
ABAM 28	128	16	64	8	4	adeG, adeJ, abeM, craA
ABAM 30	512	64	128	8	4	None
ABAM 31	512	128	128	4	4	adeG, adeJ, abeM
ABAM 32	128	64	32	2	4	adeG, adeJ, abeM
ABAM 35	512	128	128	4	4	None
ABAM 48	256	128	32	2	8	adeJ, abeM, abaQ, craA
ABAM 52	32	8	32	4	0	adeB, adeG, adeJ, adeR, abeM, abaQ, craA
ABAM 54	256	128	64	2	4	None
ABAM 57	256	64	64	4	4	adeB
ABAM 65	256	64	256	4	0	None
ABAM 67	128	32	64	4	2	adeB, adeG, adeJ, abeM, abaQ, craA
ABAM 70	128	32	128	4	0	adeB, adeG, adeJ, adeR, abeM, abaQ, craA
ABAM 73	512	256	128	2	4	None
ABAM 74	512	128	128	4	4	None
ABAM 77	128	16	32	8	4	adeB, adeG, adeJ, abeM, abaQ, amvA, craA
ABAM 78	64	16	32	4	2	adeG, adeJ, adeR, abeM, abaQ, amvA
ABAM 79	256	128	64	2	4	None
ABAM 80	128	64	32	2	4	adeG, abaQ
ABAM 95	256	64	128	4	2	None
ABAM 97	128	32	128	4	0	None
ABAM 108	256	128	64	2	4	None
ABAM 113	0.25	0.25	0.25	0	0	Reference strain
ABAM 118	512	256	256	2	2	None
ABAM 123	64	32	16	2	4	None
ABAM 129	128	128	32	0	4	None
ABAM 132	128	64	32	2	4	None
ABAM 133	64	8	16	8	4	None

^aF, MIC without EPI/MIC with EPI, significant when ≥4 (3–5).

^bNMP and PAβN were used at a fixed final concentration of 50 μg/ml.

^cMICs for NMP and PAβN were superior to 200 μg/ml for all strains except for ABAM 128 (100 μg/ml for both NMP and PAβN) and ABAM 144 (100 μg/ml for NMP).

EP gene overexpression.

The most stable reference gene was identified as rpoB, which consequently served for the normalization of qPCR results. Individual normalized calibrated ratios (NCRs) results were calculated with the pan-susceptible strain ABAM 113 as the calibrator and reported as means ± standard errors of the means (SEM) (Table S2).

At least one EP was overexpressed in 19 (56%) out of the 34 strains with a positive phenotypic efflux inhibition test (Table 1). A high prevalence of RND EPs overexpression was observed, as 17 (50%), 11 (32%), and 9 (26%) strains overexpressed adeJ, adeG, and adeB, respectively. Four isolates overexpressed only one EP (namely, adeJ for 3 strains and adeB for the last one) and 16 at least 2 EPs (Table 1). Among the four strains overexpressing two EPs, the combinations adeB-adeJ and adeG-abaQ were found in 3 and 1 of them, respectively. Joint overexpression of adeG-adeJ-abeM was witnessed in 2 strains (ABAM 31 and ABAM 32). Overexpression profiles including adeG-adeJ-abeM-craA and adeJ-abeM-abaQ-craA were each found in a single strain. Combinations of 5 and more overexpressed EPs included adeG-adeJ-abeM-abaQ-craA (1 strain), adeG-adeJ-abeM-abaQ-amvA (1 strain), and adeB-adeG-adeJ-abeM-abaQ-craA (5 strains). Only two strains overexpressed amvA (ABAM 77 and ABAM 78). Finally, looking at regulatory gene expression, none of the strains overexpressed adeS, while 6 overexpressed adeR.

Correlation between the expression of EP genes, CIP MICs, and the phenotypic efflux factor F.

The NMP F was loosely correlated with the expression of RND efflux pumps in the tested strains (Spearman’s r_s = 0.342, 0.345, and 0.365 for adeB, adeG, and adeJ, respectively; P < 0.05), while no correlation could be highlighted with the PAβN F (Fig. 2). A negative correlation was found between CIP MICs in the presence of NMP and expression of various EPs (Spearman’s r_s = −0.389, −0.401, −0.381, −0.395, and −0.339 for adeG, adeJ, abaQ, amvA and craA, respectively; P < 0.05).

FIG 2.

Matrix of correlation coefficients between expression of efflux pump genes, MICs with and without efflux pump inhibitors, and efflux factors (F).

FTIR typing.

Based on Fourier transform infrared (FTIR) spectra, a dendrogram was built that included all 34 efflux-positive isolates, ABAM 118, and ABAM 66, an efflux-negative strain (Fig. 3). The cutoff was manually set at 0.199 and allowed for the constitution of 10 clusters. Strains grouped in clusters I to III as well as in cluster V did not overexpress any EP. Furthermore, all strains from clusters I to III were 2017 isolates. All strains in clusters VI and X overexpressed at least one EP, while cluster IV was also mainly constituted of strains overexpressing at least one EP. Additionally, the latter cluster also encompassed all strains isolated in 2014 as well as 5 strains out of 14 from 2016. Finally, all strains isolated from 1995 to 1998 were grouped in cluster IX.

FIG 3.

Dendrogram obtained by clustering FTIR spectra for the 34 efflux-positive A. baumannii, ABAM 118, and ABAM 66, an efflux negative strain. The vertical blue line represents the cutoff for the Euclidian distance, and Roman numbers are arbitrarily attributed to individual clusters generated by the software. The date of isolation is given beside each strain.

Synthesis of (S)- or (R)-QPEs and PPEs 3a-b.

As shown in Fig. 4, newly designed (S)- or (R)-quinolinylpiperazinylethanols (QPEs) and pyridinylpiperazinylethanols (PPEs) 3a-b were prepared in two steps, starting from the corresponding (S)- or (R)-quinoline- or pyridine‐based epoxides 1a-b. These constitute key intermediates in the enantioselective synthesis of antimicrobial drugs previously described (26, 27). In the present synthetic pathway, a microwave-promoted regioselective S_N2 ring-opening reaction was carried out to give the N-Boc-piperazinylethanol derivatives (S)- or (R)-2a-b in 73 to 98% yields. N-Boc removal by treatment with trifluoroacetic acid afforded the desired final products in good yields (Fig. 3). Nuclear magnetic resonance (NMR) spectra are provided as figures in the supplemental material.

FIG 4.

Synthesis of new QPEs and PPEs (26, 27).

Evaluation of QPE and PPE inhibition of EP activity.

The screening for a potential inhibition of EPs by the newly synthesized NMP analogs was carried out on a selection of 15 strains with a positive phenotypic efflux test. Overall, phenotypic efflux factors (Fs) obtained for QPEs and PPEs ranged from 0 to >512 (Table 2). In the quinoline series, an F of at least 4 was found for 53% (8 strains) and 80% (12 strains) of the tested strains for enantiomers (S) and (R), respectively. The highest Fs were displayed by ABAM 14, ABAM 26, and ABAM 28, with values reaching >256, >512, and >512, whatever the enantiomer. In the pyridine series, an F of at least 4 was found for 53% (8 strains) and 67% (10 strains) of the tested strains for enantiomers (S) and (R), respectively. F was superior to 256 or 512 for five strains with the enantiomer (R), while none reached this level with the enantiomer (S). None of the N-Boc-protected precursors (S)- or (R)-2a-b provided a positive phenotypic efflux factor in the quinoline or in the pyridine series.

TABLE 2.

Activity of newly synthetized QPEs and PPEs (S)- or (R)-3a-b on 15 efflux-positive strains

Strain	CIP mIC (μg·ml−1)	Quinoline series						Pyridine series
		(S)-3a MIC	CIP + (S)-3a	Fa	(R)-3a MIC	CIP + (R)-3a	Fa	(S)-3b MIC	CIP + (S)-3b	Fa	(R)-3b MIC	CIP + (R)-3b	Fa
ABAM 7	128	100	8	16	100	16	8	100	16	8	100	<0.25	>512
ABAM 9	128	50	128	0	50	128	0	50	128	0	25	128	0
ABAM 14	64	100	<0.25	>256	100	<0.25	>256	100	1	64	100	<0.25	>256
ABAM 16	64	25	64	0	50	16	4	25	64	0	25	32	2
ABAM 26	128	100	<0.25	>512	100	<0.25	>512	100	2	64	100	<0.25	>512
ABAM 28	128	100	<0.25	>512	100	<0.25	>512	100	2	64	100	<0.25	>512
ABAM 30	128	50	64	2	100	32	4	50	64	2	25	32	4
ABAM 35	128	50	128	0	100	128	0	50	64	2	25	64	2
ABAM 48	64	50	16	4	50	4	16	50	8	8	25	<0.25	>256
ABAM 57	128	50	64	2	25	32	4	25	64	2	25	32	4
ABAM 65	256	50	64	4	25	32	8	50	64	4	25	32	8
ABAM 77	64	50	16	4	50	8	8	50	16	4	25	2	32
ABAM 97	128	50	64	2	50	64	2	50	64	2	25	64	2
ABAM 118	128	50	128	0	50	32	4	50	64	2	50	64	2
ABAM 132	256	50	64	4	100	32	8	50	32	8	25	32	8

^aF, phenotypic efflux factor, i.e., MIC without EPI/MIC with EPI, significant when ≥4.

A positive correlation was found between adeG expression and Fs for all the new EPIs [Spearman’s r_s = 0.630, 0.621, 0.589, and 0.565 for (S)-3a, (R)-3a, (S)-3b and (R)-3b, respectively; P < 0.05]. Similarly, craA expression and Fs were positively correlated [Spearman’s r_s = 0.524, 0.577, 0.581, and 0.565 for (S)-3a, (R)-3a, and (R)-3b, respectively; P < 0.05]. For compound (S)-3b, the correlation failed to reach statistical significance by a narrow margin (Spearman’s r_s = 0.518, P = 0.051). For the quinoline series, adeJ expression was additionally positively correlated to F [Spearman’s r_s = 0.544 and 0.563 for (S)-3a and (R)-3a, respectively; P < 0.05]. For compound (R)-3b, a further positive correlation was uncovered between abeM and F (Spearman’s r_s = 0.528, P = 0.046). Finally, a positive correlation between the overall number of overexpressed pumps and F was found for compound (R)-3a (Spearman’s r_s = 0.556, P = 0.034).

Cytotoxicity evaluation.

The cytotoxicity of the newly synthesized compounds was evaluated on Hep-G2 cells, a human liver cancer cell line. Compound (S)-3b cytotoxicity could not be assessed because of an insufficient amount of product. Fifty percent inhibitory concentrations (IC₅₀s) for cell viability of the NMP analogs (S)- or (R)-3a-b appeared to be lower than those of their N-Boc-protected precursors, (S)- or (R)-2a-b. An important decrease could be noted for (S) compounds 2a-b compared to their (R) enantiomers. However, quinoline derivative (R)-3a exhibited a higher value than its 6-phenylpyridine analog, (R)-3b (Fig. 5).

FIG 5.

Cytotoxicity evaluation of QPEs and PPEs (S)- or (R)-3a-b and their N-Boc-protected precursor, (S)- or (R)-2a-b, on Hep-G2 cell line. Bars represent the means ± SD from triplicates.

DISCUSSION

This work first aims to describe the prevalence of efflux overexpression in a collection of old and recent clinical isolates with various antibiotic resistance profiles. Nevertheless, all strains included, except for the 6 pan-susceptible ones included to serve as calibrators in the qPCR assays, are resistant to ciprofloxacin. This molecule is used as a marker to detect efflux-mediated resistance to antibiotics, as it has been described as a substrate for various A. baumannii EPs (28–30). Combining the results obtained with PAβN and NMP efflux inhibitors, a phenotypic efflux was identified in 26% of the strains included in this study. This prevalence is lower than the one previously published (48%) for the sole NMP on fluoroquinolone-resistant A. baumannii strains (21), but this discrepancy could arise from sampling differences between the 130 strains included in this work and the 21 studied previously. The strain ABAM 118 gave a negative result for the phenotypic evaluation with both NMP and PAβN (F-value of 2) but was previously characterized by the French National Reference Centre for Antimicrobial Resistance (NRC-AR) as overexpressing the AdeABC efflux system by RT-qPCR. This strain was therefore included in the genotypic tests as well as in the evaluation of NMP analogs.

To further characterize efflux in this collection of strains, the overexpression of EP genes belonging to the RND and other families was assessed by RT-qPCR. In order to standardize culture conditions and to ensure the relevance of the correlations drawn, all A. baumannii strains were grown up to the late exponential phase prior to RNA extraction to match culture conditions used for MIC determination. Indeed, the growth stage has been shown to influence the expression levels of EP genes (31). One could argue that the mRNA expression level for a given gene is not directly related to the actual amount of protein produced by the bacteria. However, Yoneda et al. (32) previously showed a good correlation between the amounts of MexY and MexB proteins and their respective mRNA expression in P. aeruginosa. Under these standardized experimental conditions, at least one EP was overexpressed in 56% of the strains identified as displaying a phenotypic efflux. The highest prevalence of overexpression was observed for EPs of the RND family. These results are in agreement with those previously described in A. baumannii (33–35). Within the RND family, the highest prevalence of overexpression was registered for adeJ. A high prevalence (48%) of adeJ overexpression was similarly reported by Rumbo et al. (35) on a panel of 62 Spanish strains, even though they were not selected on fluoroquinolone resistance. Zhang et al. (8) also described overexpression of adeB, adeG, and adeJ by 6.9-, 4.0-, and 2.1-fold in A. baumannii imipenem-resistant selected mutants compared to the parent imipenem-susceptible strain. Kim et al. recently observed a positive correlation between adeJ overexpression and the MDR phenotype of bacteria (18). This observation, along with the high conservation of adeJ sequence through evolution (36), point to adeJ as a better target candidate for developing a more efficient EPI than adeB, which was found to be less consistently conserved.

A rather high (32%) prevalence of overexpression coupled with a positive correlation with the F of one of the newly synthesized NMP analogs was witnessed for abeM, a member of the MATE family that is also highly conserved (36). This finding was somewhat unexpected, as abeM was previously not found to be overexpressed in clinical strains (35). However, only the relative expression was taken into account in this paper to assess overexpression, and no calibrator was used. On the other hand, the overexpression of abeM so far has not been reported to be correlated with antibiotic resistance (37, 38). Nevertheless, a greater relative expression of abeM was reported in 5 isolates of imipenem-resistant A. baumannii than in their susceptible counterparts (39).

AbaQ, a putative MFS transporter, was first described in 2018 by Pérez-Varela et al. (40). This EP is mainly involved in the extrusion of quinolone-type drugs as well as in surface-associated motility and in virulence of A. baumannii. Only two studies published so far investigated the presence of abaQ in clinical isolates (28, 40). Attia et al. reported that every one of the 21 MDR, XDR, or pandrug-resistant A. baumannii clinical isolates studied carried abaQ. In the present study, abaQ was also detected in all of the 35 strains tested by RT-qPCR. To our knowledge, this is the first study assessing the overexpression of abaQ in a collection of clinical strains, yielding an overexpression prevalence of 26%. A negative correlation between CIP MICs in the presence of NMP and abaQ expression was further highlighted in this work. Therefore, this EP would warrant further investigations as a good target for EPIs. A comparable overexpression prevalence (24%) was observed for craA, another MFS EP investigated in this work. This was not the case for amvA, the third gene encoding an MFS EP included in this work, as only 2 strains were found to overexpress it. The amvA gene was previously reported to be present in all A. baumannii strains studied but was only overexpressed in isolates exhibiting higher drug resistance (34). While amvA was also found in all the qPCR-tested strains (MDR or XDR) of this study, it was not found to be overexpressed in the most resistant ones.

No EP genes were found to be overexpressed in 15 of the 35 strains tested in the qPCR experiments. The strain ABAM 118 was found among them, which is consistent with the lack of phenotypic efflux obtained with PAβN and NMP. However, these results are at odds with those obtained from the French NRC-AR. Although the same technique (RT-qPCR) was employed to evaluate adeB overexpression, this discrepancy might be explained by a different strain used to calibrate results and/or the possibility of genetically discrete subpopulations. As for the 14 remaining strains, which displayed F values of 4 and above, several hypotheses could be put forward to account for these discrepancies: (i) an EP not screened by the qPCR analysis performed in this study could be responsible for the significant phenotypic efflux factor, and/or (ii) the differences observed between ciprofloxacin MICs with and without the EPIs were not due to an efflux mechanism. In agreement with the latter hypothesis, an F value of 4 and above for PAβN may not be the most relevant parameter to coin an efflux in a given bacterial strain, as two mechanisms of action have been described for this molecule. The first one is, indeed, a direct inhibition of RND EPs through competition with their substrates and was first reported by Lomovskaya et al. (41). However, another mechanism involving membrane permeabilization was subsequently highlighted by Lamers et al. (42). This could explain why no overexpression of EP genes could be uncovered for 7 strains with an F value of 4 and above only obtained with PAβN. However, the 7 remaining strains with no EP gene overexpressed were considered to display a phenotypic efflux because of NMP alone or both NMP and PAβN. For the latter strains, the possibility of efflux driven by an EP not investigated by the qPCR analysis cannot be ruled out and is the most likely explanation for the observed discrepancies.

Another fact pointing at a lesser suitability of PAβN as an EPI for A. baumannii strains is the absence of correlation between calculated PAβN efflux factors and gene expression of EPs selected in this study, while NMP efflux factors were found to be correlated with the expression of RND efflux pumps. These results suggest that NMP is more active than PAβN on A. baumannii EPs. They are also consistent with those reported by Pannek et al. (21). These authors demonstrated that NMP was more efficient in reverting MDR in a panel of 50 clinical A. baumannii strains than PAβN. This efficiency was even better in clinical strains with reduced susceptibility to fluoroquinolones. Along with the results obtained in this study, this observation strengthened our choice to synthetize NMP analogs as potential new inhibitors of A. baumannii EPs.

Fourier transform infrared (FTIR) spectroscopy is a phenotypic typing method with a good correlation with MLST for various Gram-negative rods, such as E. cloacae, K. pneumoniae, P. aeruginosa (43), E. coli (44), and Klebsiella oxytoca (45). It was applied to the 35 phenotypic efflux-positive strains as well as to an efflux-negative one used as an outlier. This step was undertaken to assess whether the clustering of these strains would depend on the EP overexpression profile. The cutoff for the Euclidian distance between strains was manually set at 0.199. This value has been described as giving a Rand index around 0.55 for A. baumannii compared to MLST, which is slightly lower than that for other species tested (43). Even though not optimal, this choice was made as a compromise between a realistic Euclidian distance value and a good concordance with MLST typing. The resulting dendrogram showed that the outlier strain clustered with a single strain that was not found to overexpress any EP by RT-qPCR. Most clusters were defined by the isolation date rather than efflux pump overexpression profiles. Nevertheless, the main cluster (cluster IV) harbored 12 strains, most of which overexpressed at least one EP.

To meet the urgent need to develop new therapeutic alternatives against MDR A. baumannii, the second aim of this work was to synthesize and evaluate new NMP analogs as potential EPIs. In addition to the elements stated above that drove us to choose NMP as a starting point, QPEs and PPEs were also designed as analogs of the antimalarial drugs mefloquine (MQ) and enpiroline (EN) (Fig. 1). MQ, which bears a quinoline structure, was previously described as an interesting inhibitor of efflux in both P. aeruginosa and E. coli, possibly targeting RND EPs (46). Several MQ analogs were also previously synthesized, with one of them displaying interesting results on Enterobacter (Klebsiella) aerogenes and Klebsiella pneumoniae strains (47). Likewise, EN, which bears a 6-phenylpyridine core, was identified as a hit in the screening for potential EPIs and found to bind the AcrA protein within AcrAB-TolC EP of E. coli (48).

The efflux inhibition activity of the compounds synthesized in this study was assessed on 15 efflux-positive strains harboring various inhibition profiles of phenotypic efflux: (i) 6 strains positive only for NMP inhibition, (ii) 6 strains both positive for NMP and PAβN inhibition, (iii) 2 strains positive only for PAβN inhibition, and (iv) ABAM 118, the strain previously characterized by the French NRC-AR as overexpressing the AdeABC efflux system.

The maximum observed F value for NMP and PAβN was 8, while in the quinoline series as well as in the pyridine series, F values reached 512. More specifically, on strains ABAM 14, ABAM 26, and ABAM 28, which harbored various overexpressed pumps, NMP provided an F of 4, 8, and 8, respectively, while compounds (S)-3a, (R)-3a, and (R)-3b displayed F values superior to 256. Additionally, a positive correlation between the number of overexpressed pumps and F values for (R)-3a was uncovered (P = 0.035). The replacement of the naphthalene core of NMP by a quinoline or a 6-phenylpyridine bioisosteric one in the newly synthesized analogs greatly improved their efflux pump inhibitory properties.

In the quinoline series, both enantiomers displayed similar inhibitory activities, while in the pyridine series the (R) enantiomer was more active than its (S) counterpart. The (R) enantiomer of the quinoline series was also less cytotoxic. These results once again back up the importance of stereochemistry in the activity and cytotoxicity of MQ analogs such as QPEs. Indeed, (+)-MQ (11S, 12R) was reported as slightly more active as an antimicrobial drug and less cytotoxic than (−)-MQ (11R, 12S) on some Plasmodium falciparum strains (49), while it was less active on Mycobacterium avium (50). Furthermore, the N-Boc-protected precursors (S)- or (R)-2a-b of QPEs and PPEs were revealed to be inactive as EPIs. From this comparison, it is apparent that for QPEs and PPEs, (S)- or (R)-3a-b, the free amine function of piperazine group plays a crucial role in the inhibition of EP activity. However, the low IC₅₀s registered for unprotected compounds preclude their use alone to reach a relevant antibiotic activity. Without further pharmacomodulation to reduce their cytotoxicity, these molecules could only be useful in combination, provided that a synergistic effect with known antibiotics, such as ciprofloxacin, could lower their effective concentrations below IC₅₀s.

To conclude, this work showed that the most overexpressed EP gene in this collection of strains was adeJ. The overexpression prevalence of the other two RND EPs was similar to those of abeM, abaQ, and craA. Among the 4 NMP analogs screened in this study as potential new EPIs, the QPE derivative (R)-3a was identified as a hit molecule. Further investigations are warranted to (i) implement pharmacomodulations on this promising structure to try and lower cytotoxicity, (ii) perform checkerboard assays to better characterize a potential synergy between CIP and these new molecules, and (iii) ascertain that this synergy is indeed mediated through efflux inhibition using growth-independent efflux assays and efflux-deficient strains. Selective efflux-deficient strains would also permit us to better identify which efflux pump is the target of these molecules.

MATERIALS AND METHODS

Bacterial strains and antibiotic susceptibility testing.

This study was held on 124 ciprofloxacin-resistant and 6 pan-susceptible A. baumannii strains isolated from clinical or screening samples in Amiens University Hospital (Amiens, France) from 1995 to 2018. The research project was exempted from the requirement for ethical approval because no health-related personal data were used. All isolates were identified by using matrix-assisted laser desorption ionization-time of flight (MALDI‐TOF) mass spectrometry (MALDI Biotyper; Bruker Daltonics, France). Susceptibility testing was performed using the standard disk agar diffusion method on Mueller-Hinton (MH) agar medium (Bio‐Rad, Marne‐la‐Coquette, France) according to the CA-SFM/EUCAST guidelines (51). Ciprofloxacin resistance was confirmed through liquid MIC determination as described below. All strains were kept frozen at −80°C on microbeads (Cryobank, MAST Diagnostics, Amiens, France) until use.

Liquid ciprofloxacin MIC and phenotypic efflux determinations.

To determine the presence of efflux systems in the 130 A. baumannii strains, efflux pump inhibition tests were carried out by measuring the ciprofloxacin MIC of each strain by the microdilution method in the presence or the absence of NMP or PAβN used as EPIs (Sigma-Aldrich Co., St. Louis, MO). MICs were determined over a range of 0.25 to 512 μg/ml on 96-well microtiter plates (VWR International, PA) using the broth microdilution technique in cation-adjusted MH according to the guidelines provided by the Clinical and Laboratory Standards Institute (CLSI) (52). Either NMP or PAβN was added to the standard MIC determination procedure at a fixed final concentration of 50 μg/ml (53). MIC determinations for NMP and PAβN were also carried out for each strain over concentrations ranging from 0.1 to 200 μg/ml to ascertain that the 50 μg/ml concentration used for EP inhibition was not inhibitory in itself. All bacterial isolates were cultured overnight on Columbia agar plates with 5% sheep blood (Becton, Dickinson, NJ) at 35°C. Inocula were then prepared according to the recommendations of CLSI (52). After inoculation, all 96-well plates were incubated at 35°C under aerobic atmosphere for 18 to 24 h. MICs were determined as the lowest concentration at which wells remained clear. A phenotypic efflux was deemed significant when the ciprofloxacin MIC determined in the presence of an EPI was at least 4-fold lower than in its absence (53–55), leading to a calculated phenotypic efflux factor F (MIC without EPI/MIC with EPI) of 4 and above.

In vitro EPI activity and MICs of new NMP analogs.

The EPI activity and MICs of 8 NMP analogs synthetized in this study were determined by the microdilution method with the same protocol as that for phenotypic efflux determination. This evaluation was performed using 15 of the efflux-positive isolates selected on the basis of their NMP and PAβN efflux inhibition profiles.

MIC determinations for NMP analogs were carried out for each strain over concentrations ranging from 0.1 to 200 μg/ml. EPIs were then used in association with ciprofloxacin for the EP inhibition assays at a final fixed concentration lower than their MIC, i.e., 50 μg/ml or 10 μg/ml.

RNA extraction and reverse transcription.

Efflux-positive isolates were assessed for the overexpression of several genes encoding EPs. Strains were cultured in cation-adjusted Mueller-Hinton broth for 18 h at 35°C up to the late exponential phase prior to EP expression determination. Total RNA extraction was performed using the Qiagen RNeasy minikit with the addition of RNAprotect bacterial reagent (Qiagen, Courtaboeuf, France). Briefly, 1 ml of bacterial culture was centrifuged (8,000 × g, 5 min), and the bacterial pellet was resuspended in 1 ml of RNAprotect reagent, mixed by vortexing for 5 s, and incubated at room temperature for 5 min. After the decantation of the RNAprotect reagent (8,000 × g, 10 min), 200 μl of 30 mM Tris–1 mM ethylenediaminetetraacetic acid (pH 8.0) buffer containing lysozyme (15 mg/ml) and proteinase K (0.56 mg per strain) were added and left to incubate for 10 min at room temperature to ensure a complete lysis of the bacterial cell wall. Further precipitation and purification of nucleic acids were carried out according to the manufacturer’s recommendations, and the final elution step was performed using 50 μl of RNase-free water. The concentration of RNA in each sample was quantified with a spectrophotometer (NanoDrop; ThermoFisher Scientific, France) and RNA extracts stored at −20°C until further use.

Prior to cDNA synthesis, genomic DNA (gDNA) was removed from 1 μg of total RNA using a gDNA wipeout buffer (Qiagen). The reverse transcription was performed using the Transcriptor high-fidelity cDNA synthesis kit (Roche Diagnostics, Bâle, Sweden) in a volume of 20 μl, including 9.4 μl of template RNA (extract concentrations were adjusted to contain 1 μg of RNA), 1.1 μl of reverse transcriptase, 4 μl of RT reaction buffer 5×, 0.5 μl of Protector RNase inhibitor, 2 μl of deoxynucleotide mix, 2 μl of random hexamer primers (600 pmol/μl), and 1 μl of dithiothreitol (DTT). Reverse transcription was performed in a Veriti PCR thermal cycler (Applied Biosystems, France) in three steps: 10 min at 65°C to ensure denaturation of RNA secondary structures, followed by 30 min at 42°C and 5 min of incubation at 85°C to inactivate the reverse transcriptase. All reactions, including RNA handling, were carried out on ice.

Quantitative PCR.

Oligonucleotide primers used in this study are shown in Table 3. Primers used for the MFS efflux pump abaQ were designed using Primer3 software, available online (https://bioinfo.ut.ee/primer3/). The stability of expression for the 4 candidate reference genes 16S rRNA, rpoB, gyrB, and recA was evaluated through geNorm, BestKeeper, NormFinder, and comparative ΔC_T, and, finally, overall stabilities were ranked using the comprehensive RefFinder tool (https://www.heartcure.com.au/reffinder/?type=reference%23). Based on the geometric mean value of each gene, comprehensive stability was evaluated through the RefFinder tool, and candidate reference genes were ranked from 1 (most stable) to 4 (least stable). The gene ranked 1 was used to normalize expression results.

TABLE 3.

Oligonucleotide primers used in this study

Target gene	Sequence (5′–3′)	Amplicon size (bp)	Reference or source	Efficiency
recA	FWD, TGAAGGCACATGTACCACCAG	109	58	2.280
	REV, ACCAAAAGGCCGTATTATCG
gyrB	FWD, TTCACAAACAACATTCCACAAAAAG	139	59	1.871
	REV, GCATCATCACCAGTCACATTCA
rpoB	FWD, GAGTCTAATGGCGGTGGTTC	109	60	2.129
	REV, ATTGCTTCATCTGCTGGTTG
16S	FWD, CGTAAGGGCCATGATGACTT	150	61	2.107
	REV, CAGCTCGTGTCGTGAGATGG
adeB	FWD, AACGGACGACCATCTTTGAGTATT	83	60	2.136
	REV, CAGTTGTTCCATTTCACGCATT
adeG	FWD, ATCGCGTAGTCACCAGAACC	92	61	2.054
	REV, CGTAACTATGCGGTGCTCAA
adeJ	FWD, TGCGTATCTGGCTTGATCCA	110	59	1.979
	REV, CACCTAACTGACCTACGGCAACT
adeR	FWD, TGGGTTAAAAGGCTTCACCA	114	62	2.212
	REV, ACGCCAAAAAGCTCAGACTC
adeS	FWD, GCATTTTTGACGGAAACCTC	120	62	1.827
	REV, TTAGTCACGGCGACCTCTCT
abeM	FWD, TGCCAATTGGTTTAGCTGTG	100	63	2.003
	REV, TACTTGGTGTGCGGCAATAA
abaQ	FWD, ATCCCAAATGGACCGACATA	148	This study	1.911
	REV, TTGGCTGTAGTTGCGTTCTG
amvA	FWD, ACGATTGATGCAACGGTAATGC	82	8	2.013
	REV, TCCATAAAAGCTGATTGGCAGT
craA	FWD, TGTGCAACTCTTTCCTGCATT	140	8	2.121
	REV, GCAATGATTGAGCTTGTACGCTAT

A LightCycler 480 (Roche Diagnostics) apparatus was used for all quantitative PCRs. One strain was considered for the role of reference (calibrator) in RT-qPCR assays: ABAM 113, a wild A. baumannii strain, susceptible to ciprofloxacin. All PCR amplification reactions were performed in 384-well plates in a 10-μl final volume containing 2.5 μl of diluted (1:20) template cDNA, 1 μl of each primer (corresponding to a final concentration of 0.5 μM), 5 μl of PowerUp SYBR green master mix (Applied Biosystems), and 0.5 μl RNase/DNase-free water (Qiagen). The cycling program was set as (i) activation with 1 cycle at 95°C for 15 min, (ii) amplification of 45 cycles, including a 15-s denaturation at 95°C, a 25-s annealing at 60°C, and a 15-s elongation at 72°C, and (iii) melting curve of 1 cycle, including 5 s at 95°C, 1 min at 65°C, and a final increase at 97°C, with a transition rate of 0.11°C/s. Each reaction was carried out in duplicate, and the experiment was repeated on two different sets of RNA extracts (biological replicate).

Evaluation of real-time PCR results.

Relative standard curves describing the PCR efficiency (E) for each primer pair were generated for each amplicon (56) (Table 3). Prior to any other analysis, melting curves were checked for the presence of primer dimers and other artifacts. The relative expression between the target and reference genes was calculated using the formula [E_goi^{(1Ct, goi)}]/[E_ref^(1Ct,ref)], where E stands for the PCR efficiency factor, goi for gene of interest, and ref for reference. Results were subsequently expressed as NCRs.

Statistical analysis.

The statistical analysis was performed using XLstat 2021 software (Addinsoft, France). The normal distribution of the NCR values was first evaluated using Shapiro-Wilk test. Differences in the mean expression level of each gene of interest were tested using the Wilcoxon rank sum test versus a cutoff of 2 for overexpression (57). The relationships between ciprofloxacin MIC values with and without EPIs, the MIC reduction factor (F), and gene expression were assessed by calculating Spearman’s correlation coefficients. Statistical significance was inferred for P values below 0.05.

FTIR spectroscopy analysis.

Efflux-positive isolates were typed with an IR-Biotyper (Bruker Daltonics). Isolates were cultured twice on Mueller-Hinton agar medium for 24 h at 35 ± 2°C. Samples were prepared under room temperature (23.5°C) and hygrometry (46%). An overloaded 1-μl inoculation loop was taken from the confluent part of the bacterial culture and suspended into 50 μl of 70% ethanol in a 1.5-ml Bruker suspension vial with inert metal cylinders. Suspensions were vortexed before addition of 50 μl of deionized water. Fifteen microliters of each suspension were spotted in triplicate on a 96-spot silicon microtiter plate (Bruker Daltonics) and dried at 35°C ± 2°C for 30 min. Fifteen microliters of each of the two controls used (Bruker infrared test standards [IRTS] 1 and 2) was spotted twice per target. Spectra were acquired and analyzed using OPUS 7.5 software (Bruker GmbH, Bremen, Germany). Dendrograms were built with IR Biotyper client software v1.5 (Bruker GmbH) using Euclidian distance and the average linkage clustering method.

Chemistry.

All commercially available products were used without further purification. All solvents were dried using the PureSolv MD 5 solvent purification system. Routine monitoring of reactions was carried out using Merck thin-layer chromatography (TLC) Silica gel 60 F254 plates, visualized under UV light (254 nm) and revealed using a 95% phosphomolybdic acid solution in ethanol. Column chromatography was performed over 60-Å Merck Silica gel (40 to 63 μm). Microwave reactions were performed using a Discover SP apparatus (CEM Corporation). Melting points (mp) were determined on Stuart SMP3 and SMP50 devices and reported uncorrected. Infrared (IR) measurements were performed on a Jasco FT/IR-4200 system fitted with an ATR-golden gate and were reported using the frequency of absorption (per centimeter). Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker 400 Cryosonde NMR instrument. Chemical shifts (δ) are expressed in parts per million (ppm) downfield from tetramethylsilane as an internal standard and were referenced to the deuterated solvent. ¹H and ¹³C data are reported in the order of chemical shift, multiplicity (s, singlet; br s, broad singlet; d, doublet; t, triplet and m, multiplet), integration, and coupling constants, J, in Hertz (Hz). The optical rotations were measured on a Jasco P-2000 polarimeter with a 100-mm cell at a temperature of 22°C. Specific rotations [α_D^T] are reported in degree·cm³·g⁻¹·dm⁻¹ and concentrations in g per 100 ml. High-resolution mass spectra (HRMS) were obtained from a Micromass-Waters Q-TOF Ultima spectrometer in electrospray ionization (ESI) mode (positive or negative).

(i) Synthesis of N-Boc protected precursors (S) or (R)-2a-b.

To a solution of the (S)- or (R)-quinoline or pyridine‐based epoxide 1a-b (0.050 g,0.150 to 0.163 mmol) in ethanol (1 ml) was added N-Boc-piperazine (1.5 equiv). The reaction mixture was heated in a microwave at 130°C for 1 h. The solvent was evaporated under reduced pressure, and the residue was purified by column chromatography on silica gel (CH₂Cl₂–MeOH, 99:1) to give the desired product (S)- or (R)-2a-b.

(a) tert-butyl (S)-4-(2-(2,8-bis(trifluoromethyl)quinolin-4-yl)-2-hydroxyethyl)piperazine-1-carboxylate [(S)-2a]. Compound (S)-2a was obtained from compound (S)-1a (0.163 mmol) according to the general procedure as a white solid (59 mg, 73%). mp 134°C. IR ν_max: 3,332, 1,696, 1,309, 1,162, 1,126, 1,101, 772 cm⁻¹. ¹H NMR (400 MHz, chloroform‐d) δ 1.47 (s, 9H), 2.49 to 2.55 (m, 3H), 2.77 to 2.79 (m, 2H), 2.84 (dd, 1H, ³J= 19.2 Hz, ⁴J= 3.2 Hz), 3.51 to 3.53 (m, 4H), 4.24 (s, 1H), 5.57 (dd, 1H, ³J= 10.5 Hz, ⁴J= 3.2 Hz), 7.72 (t, 1H, ³J= 7.9 Hz), 8.15 to 8.17 (m, 3H). ¹³C NMR (101 MHz, chloroform‐d) δ 28.5 (3C), 43.4, 44.0, 53.0 (2C), 64.8, 65.3, 80.2, 114.7 (q, J= 2.9 Hz), 121.1 (d, J= 225.0 Hz), 123.8 (q, J= 224.4 Hz), 126.6, 126.8, 127.3, 128.8 (q, J= 5.4 Hz), 129.8 (q, J= 30.2 Hz), 143.8, 148.8 (q, J= 35.3 Hz), 150.9, 154.7. HRMS (ESI): m/z calculated for C₂₂H₂₅F₆N₃O₃, [M+H]⁺ 494.1878, found 494.1887; [α_D²²]: +41 (c 0.1, MeOH).

(b) tert-butyl (R)-4-(2-(2,8-bis(trifluoromethyl)quinolin-4-yl)-2-hydroxyethyl)piperazine-1-carboxylate [(R)-2a]. Compound (R)-2a was obtained from compound (R)-1a (0.163 mmol) according to the general procedure as a colorless oil (73.5 mg, 92%). IR ν_max: 1,697, 1,308, 1,161, 1,127, 1,103, 767 cm⁻¹. ¹H NMR (400 MHz, chloroform‐d) δ 1.47 (s, 9H), 2.47 to 2.55 (m, 3H), 2.78 to 2.80 (m, 2H), 2.84 (dd, 1H, ³J= 12.8 Hz, ⁴J= 3.3 Hz), 3.51 to 3.53 (m, 4H), 4.31 (s, 1H), 5.58 (dd, 1H, ³J= 10.6 Hz, ⁴J= 3.2 Hz), 7.72 (t, 1H, ³J= 7.9 Hz), 8.14 to 8.17 (m, 3H). ¹³C NMR (101 MHz, chloroform‐d) δ 28.5 (3C), 43.4, 44.0, 53.0 (2C), 64.9, 65.3, 80.2, 114.7 (q, J= 2.20 Hz), 121.1 (q, J= 227.1 Hz), 123.9 (q, J= 225.8 Hz), 126.7, 126.8, 127.3, 128.9 (q, J= 5.5 Hz), 129.8 (q, J= 30.2 Hz), 143.8, 148.9 (q, J= 35.3 Hz), 150.9, 154.7. HRMS (ESI): m/z calculated for C₂₂H₂₆F₆N₃O₃, [M+H]⁺ 494.1878, found 494.18871; [α_D²²]: −38 (c 0.1, MeOH).

(c) tert-butyl (S)-4-(2-hydroxy-2-(2-(trifluoromethyl)-6-(4-(trifluoromethyl)phenyl)pyridin-4-yl)ethyl)piperazine-1-carboxylate [(S)-2b]. Compound (S)-2b was obtained from compound (S)-1b (0.150 mmol) according to the general procedure as a white oil (58.3 mg, 76%). IR ν_max: 2,932, 1,682, 1,323, 1,165, 1,120, 844 cm⁻¹. ¹H NMR (400 MHz, chloroform‐d) δ 1.46 (s, 9H), 2.51 to 2.57 (m, 3H),2.72 to 2.80 (m, 3H), 3.52 to 3.54 (m, 4H), 4.97 (dd, 1H, ³J= 10.6 Hz, ⁴J= 3.3 Hz), 7.67 (s, 1H), 7.72 (d, 2H, ³J= 8.2 Hz), 8.00 (s, 1H), 8.17 (d, 2H, ³J= 8.1 Hz). ¹³C NMR (101 MHz, chloroform‐d) δ 28.5 (3C), 43.7 (2C), 53.0 (2C), 65.3, 67.6, 80.3, 116.7 (q, J= 2.7 Hz), 120.2, 121.5 (q, J= 260.1 Hz), 124.2 (q, J= 258.2 Hz), 125.9 (2C, q, J= 3.8 Hz), 127.6 (2C), 131.7 (q, J= 32.6 Hz), 141.1, 148.7 (q, J= 34.6 Hz), 154.4, 154.7, 156.6. HRMS (ESI): m/z calculated for C₂₄H₂₈F₆N₃O₃, [M+H]⁺ 520.2035, found 520.2036; [α_D²²]: +13 (c 0.1, MeOH).

(d) tert-butyl (R)-4-(2-hydroxy-2-(2-(trifluoromethyl)-6-(4-(trifluoromethyl)phenyl)pyridin-4-yl)ethyl)piperazine-1-carboxylate [(R)-2b]. Compound (R)-2b was obtained from compound (R)-1b (0.150 mmol) according to the general procedure as a white oil (68.3 mg, 88%). IR ν_max: 2,974, 1,686, 1,323, 1,164, 1,120, 845 cm⁻¹. ¹H NMR (400 MHz, chloroform‐d) δ 1.47 (s, 9H), 2.42 to 2.50 (m, 3H),2.67 to 2.77 (m, 3H), 3.45 to 3.54 (m, 4H), 4.88 (dd, 1H, ³J= 10.7 Hz, ⁴J= 3.6 Hz), 7.67 (s, 1H), 7.74 (d, 2H, ³J= 8.2 Hz), 8.00 (s, 1H), 8.19 (d, 2H, ³J= 8.2 Hz). ¹³C NMR (101 MHz, chloroform‐d) δ 28.7 (3C), 44.0 (2C), 53.1 (2C), 65.5, 67.7, 80.3, 116.9 (q, J= 2.3 Hz), 120.3, 121.7 (q, J= 260.8 Hz), 124.4 (q, J= 257.2 Hz), 126.1 (2C, q, J= 3.7 Hz), 127.8 (2C), 131.9 (q, J= 32.5 Hz), 141.3, 148.9 (q, J= 34.5 Hz), 154.7, 154.9, 156.8. HRMS (ESI): m/z calculated for C₂₄H₂₈F₆N₃O₃, [M+H]⁺ 520.2035, found 520.2050; [α_D²²]: −11 (c 0.1, MeOH).

(ii) Synthesis of QPEs and PPEs (S)- or (R)-3a-b.

To a stirred solution at 0°C of N-Boc-protected precursors (S)- or (R)-2a-b (0.045 g,0.087 to 0.091 mmol) in CH₂Cl₂ (3 ml) was added trifluoroacetic acid (20 equiv). The mixture was stirred at room temperature for 22 h. After evaporating under reduced pressure, the residue was diluted in CH₂Cl₂. The organic phase was washed with a 2 N NaOH solution and brine, dried over Na₂SO₄, and concentrated in vacuo to give the desired product (S)- or (R)-3a-b.

(a) (S)-1-(2,8-bis(trifluoromethyl)quinolin-4-yl)-2-(piperazin-1-yl)ethan-1-ol [(S)-3a]. Compound (S)-3a was obtained from compound (S)-2a (0.091 mmol) according to the general procedure as brown crystals (17.8 mg, 50%). IR ν_max: 2,920, 1,308, 1,103, 771 cm⁻¹. mp 131 to 133°C. ¹H NMR (400 MHz, chloroform‐d) δ 2.46 (dd, 1H, ²J= 12.8 Hz, ³J= 10.5 Hz), 2.51 to 2.56 (m, 2H), 2.81 to 2.87 (m, 3H),2.95 to 3.06 (m, 4H), 5.56 (dd, 1H, ³J= 10.5 Hz, ⁴J= 3.1 Hz), 7.72 (t, 1H, ³J= 7.7 Hz),8.14 to 8.19 (m, 3H). ¹³C NMR (101 MHz, chloroform‐d) δ 46.2 (2C), 54.2 (2C), 65.3, 65.4, 114.8 (q, J= 2.1 Hz), 121.3 (q, J= 227.6 Hz), 124.0 (q, J= 222.6 Hz), 126.9, 127.0, 127.4, 129.0 (q, J= 5.5 Hz), 129.9 (q, J= 30.3 Hz), 143.9, 149.0 (q, J= 35.1 Hz), 151.4. HRMS (ESI): m/z calculated for C₁₇H₁₈N₃OF₆, [M+H]⁺ 394.1354, found 394.1352; [α_D²²]: +60 (c 0.1, MeOH).

(b) (R)-1-(2,8-bis(trifluoromethyl)quinolin-4-yl)-2-(piperazin-1-yl)ethan-1-ol [(R)-3a]. Compound (R)-3a was obtained from compound (R)-2a (0.091 mmol) according to the general procedure as a yellow powder (22.3 mg, 62%). IR ν_max: 2,920, 1,306, 1,131, 1,103, 766 cm⁻¹. mp 59 to 61°C. ¹H NMR (400 MHz, chloroform‐d) δ 2.45 (dd, 1H, ²J= 12.8 Hz, ³J= 10.5 Hz), 2.51 to 2.54 (m, 2H), 2.79 to 2.87 (m, 3H), 2.94 to 3.01 (m, 4H), 5.56 (dd, 1H, ³J= 10.5 Hz, ⁴J= 3.3 Hz), 7.72 (t, 1H, ³J= 7.9 Hz),8.14 to 8.19 (m, 3H). ¹³C NMR (101 MHz, chloroform‐d) δ 46.3 (2C), 54.5 (2C), 65.2, 65.5, 114.8 (q, J= 1.9 Hz), 121.3 (q, J= 227.2 Hz), 124.0 (q, J= 222.5 Hz), 126.8, 127.0, 127.3, 128.9 (q, J= 5.5 Hz), 129.9 (q, J= 30.3 Hz), 143.9, 148.9 (q, J= 36.6 Hz), 151.4. HRMS (ESI): m/z calculated for C₁₇H₁₈N₃OF₆, [M+H]⁺ 394.1354, found 394.1350; [α_D²²]: −42 (c 0.1, MeOH).

(c) (S)-2-(piperazin-1-yl)-1-(2-(trifluoromethyl)-6-(4-(trifluoromethyl)phenyl)pyridin-4-yl)ethan-1-ol [(S)-3b]. Compound (S)-3b was obtained from compound (S)-2b (0.087 mmol) according to the general procedure as a yellow powder (10.9 mg, 30%). ¹H NMR (400 MHz, chloroform‐d) δ 2.42 (dd, 1H, ²J= 12.5 Hz, ³J= 10.7 Hz), 2.46 to 2.51 (m, 2H), 2.69 (dd, 1H, ²J= 12.5 Hz, ³J= 3.7 Hz), 2.75 to 2.80 (m, 2H), 2.92 to 3.02 (m, 4H), 4.87 (dd, 1H, ³J= 10.7 Hz, ⁴J= 3.6 Hz), 7.67 (s, 1H), 7.74 (d, 2H, ³J= 8.2 Hz), 8.00 (s, 1H), 8.20 (d, 2H, ³J= 8.1 Hz). ¹³C NMR (101 MHz, chloroform‐d) δ 46.0 (2C), 54.0 (2C), 65.7, 67.3, 116.6 (q, J= 3.1 Hz), 120.0, 121.4 (q, J= 257.7 Hz), 124.6 (q, J= 256.3 Hz), 125.8 (2C, q, J= 3.8 Hz), 127.5 (2C), 131.6 (q, J= 32.8 Hz), 141.1, 148.6 (q, J= 34.5 Hz), 154.7, 156.5. HRMS (ESI): m/z calculated for C₁₉H₂₀N₃OF₆, [M+H]⁺ 420.1511, found 420.1501.

(d) (R)-2-(piperazin-1-yl)-1-(2-(trifluoromethyl)-6-(4-(trifluoromethyl)phenyl)pyridin-4-yl)ethan-1-ol [(R)-3b]. Compound (R)-3b was obtained from compound (R)-2b (0.087 mmol) according to the general procedure as a yellow powder (27.3 mg, 75%). IR ν_max: 3240, 2919, 1323, 1118, 1061, 844 cm⁻¹. mp 53.5°C. 1H NMR (400 MHz, chloroform‐d) δ 2.42 (dd, 1H, ³J= 12.5 Hz, ⁴J= 10.7 Hz),2.49 to 2.51 (m, 2H), 2.69 (dd, 1H, ³J= 12.5 Hz, ⁴J= 3.7 Hz), 2.75 to 2.80 (m, 2H),2.92 to 3.03 (m, 4H), 4.87 (dd, 1H, ³J= 10.7 Hz, ⁴J= 3.6 Hz), 7.67 (s, 1H), 7.74 (d, 2H, ³J= 8.2 Hz), 7.99 (s, 1H), 8.19 (d, 2H, ³J= 8.2 Hz). ¹³C NMR (101 MHz, chloroform‐d) δ 46.2 (2C), 54.1 (2C), 65.9, 67.4, 116.7 (q, J= 2.4 Hz), 120.1, 121.5 (q, J= 257.0 Hz), 124.3 (q, J= 252.7 Hz), 125.9 (2C, q, J= 3.8 Hz), 127.7 (2C), 131.7 (q, J= 32.7 Hz), 141.2, 148.7 (q, J= 34.7 Hz), 154.9, 156.6. HRMS (ESI): m/z calculated for C₁₉H₂₀N₃OF₆, [M+H]⁺ 420.1511, found 420.1506; [α_D²²]: −17 (c 0.1, MeOH).

Cytotoxicity.

The cytotoxicity of newly synthesized compounds was evaluated on a human hepatoma cell line (HepG2 from European Collection of Authenticated Cell Cultures [ECACC], Merck, Darmstadt, Germany) cultured in 75-cm² sterile flasks in modified Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (Gibco) in a humidified atmosphere of 5% carbon dioxide at 37°C. Tested compounds in different concentrations (from 0 to 100 μM) were added to cells. The vehicle (dimethyl sulfoxide) was used as a control. After 48 h of incubation, cell viability was determined using the CellTiter-Glo luminescent cell viability assay (Promega, Madison WI, USA), according to the manufacturer’s protocol. Cell viability was expressed as the percentage of relative light units (RLU). IC₅₀s were calculated using an online tool available at https://www.aatbio.com/tools/ic50-calculator.