A Screening of Antimalarials Extends the Range of Known Escherichia coli AcrB Efflux Substrates and Reveals Two Candidates with Antimicrobial Drug-Enhancing Activity

AcrB-dependent efflux of antimalarials

In this study, we included 24 antimalarials from various compound classes^5,15 (Table 1). First, we examined the resistance of an acrB overexpressing E. coli and its acrB deletion derivative to the antimalarials themselves. The susceptibility of the AcrB-deficient mutant to artemisinine and its derivatives dihydroartemisinine, arteether, artemether, and artesunate, to doxycycline, proguanil, pyrimethamine, tafenoquine, and to mefloquine was increased more than or equal to fourfold compared with the wild-type AcrB strain, suggesting that these compounds are efflux substrates (Table 1).

This has already been shown for doxycycline, ¹⁶ like for other tetracyclines, ¹⁷ and such indications have also been given for artesunate. ⁸ The other artemisinine derivatives as well as proguanil, pyrimethamine, tafenoquine, and mefloquine, to our best knowledge, have to be newly added to the AcrB substrate library. It should be noted that if the MICs of the antimalarials were ≥256 μg/mL for both strains (test concentrations limited due to precipitation in broth), efflux by AcrB appeared unlikely but could not be definitely excluded (e.g., if the MIC with 3-AG100 was >1,024 and that with ΔacrB 512 μg/mL).

The identification of chemically diverse AcrB substrates provides information for drug designer aiming to develop noneffluxable drugs. For example, it might be of interest that the hydrogenation of a single oxo group in artesiminine resulting in dihydroartemisinine can change the AcrB efflux ratio (quotient of the MIC of the AcrB overexpressing strain and that of its knockout mutant ΔAcrB [Table 1]).

Screening for drug-enhancing activity in E. coli

Next, we screened subinhibitory concentrations (1/4 MIC or 100 μg/mL, if MIC >256) of the antimalarials in combination with a range of antibacterial compounds in standard twofold microdilution MIC assays (Table 2). Representatives from nine antibiotic classes were used, namely of the oxazolidinones (linezolid), lincosamides (clindamycin), phenicols (chloramphenicol), β-lactams (oxacillin, cefuroxime), aminocoumarins (novobiocin), tetracyclines (tetracycline, minocycline), rifamycins (rifaximine), macrolides (clarithromycin, erythromycin, azithromycin), and FQs (levofloxacin, moxifloxacin).

With the latter (not shown in Table 2), only proguanil achieved greater than or equal to fourfold in vitro sensitizing (Table 3). The best candidates regarding broad activity were tafenoquine and proguanil (Fig. 1A), revealing ≥4-fold MIC decreases with 6 and 11 of the antibacterial test drugs (except FQs), respectively (Table 2). These compounds including amodiaquine (activity with two drugs) were additionally tested with the highly lipophilic FQ nadifloxacin.

OPEN IN VIEWER

FIG. 1.

(A) The structures of tafenoquine, proguanil, and NMP. (B) The intracellular accumulation of Hoechst 33342 added after 5 minutes of incubation with or without adjuvants. Means of normalized values shown (n ≥ 5); (C) Nile red real-time efflux in the absence and presence of adjuvants (means, n ≥ 3). The arrow indicates energizing of the deenergized Escherichia coli cells by the addition of glucose. NMP, 1-(1-naphthylmethyl)-piperazine.

The exclusive FQ sensitizing by proguanil remarkably increased with the FQ lipophilicity, a similar finding as for the EPI phenylarginine alanine β-naphthylamide (PAβN), which, however, works at lower concentrations ¹⁸ (Table 3). It should be noted that neither proguanil nor tafenoquine shows any effect on the sensitivity to the AcrB nonsubstrate gentamicin (Table 3).

Proguanil showed broader activity, even if applied at 50 μg/mL, and acts more efficiently (sensitizing up to 16-fold) than tafenoquine (Tables 2 and 3). The efficiency of proguanil appears comparable with the well-described EPI NMP ¹⁹ with somewhat differing selectivity, but worse as found with PAβN (Tables 2 and 3). However, with the latter, an additional permeabilizing activity at least partially explains increased sensitizing to high-molecular antimicrobials such as rifamycins.^18,20 Even though some permeabilization contributing to drug sensitizing cannot be totally excluded for proguanil and tafenoquine, their relatively low impact on rifamycin resistances suggests rather negligible effects on membrane integrity.

Evaluating the efflux inhibitory activity by intracellular dye accumulation

To elucidate the impact on AcrB efflux further, we monitored the kinetics of intracellular dye accumulation in E. coli cells. As shown in Fig. 1B, Hoechst 33342 accumulates in the presence of 100 μg/mL proguanil to similar extent as with 100 μg/mL NMP, meaning 1.7–1.8-fold relative to the acrB overexpressing E. coli without adjuvants, whereas it was 1.5-fold with 16 μg/mL tafenoquine and 1.9-fold due to acrB deletion (p-values ≤0.0001). In Nile red real-time efflux assays, 50 μg/mL proguanil was sufficient to achieve similar efflux inhibition as found with 100 μg/mL NMP (Fig. 1C). The impact of tafenoquine on Nile red efflux could not be examined, since its presence caused immense fluorescence decreases presumably due to quenching effects.

Taken together, among the screened antimalarials, proguanil showed broadest and most effective drug potentiating in E. coli and clearly the ability to inhibit AcrB-mediated efflux. Tafenoquine revealed less broad activity but at lower concentration. Notably, the concentrations of both compounds required are well above those reached in plasma after administration of standard doses in humans.^21,22 Being substrates of AcrB themselves, which is similar to PAβN, both antimalarials do not perfectly meet the requirements of an ideal EPI. Their AcrB substrate nature suggests a competitive rather than an allosteric mode of inhibition.