Efflux-Mediated Resistance to New Oxazolidinones and Pleuromutilin Derivatives in Escherichia coli with Class Specificities in the Resistance-Nodulation-Cell Division-Type Drug Transport Pathways

A major contribution of the resistance-nodulation-cell division (RND)-transporter AcrB to resistance to oxazolidinones and pleuromutilin derivatives in Escherichia coli was confirmed. However, we discovered significant differences in efflux inhibitor activities, specificities of the homologous pump YhiV (MdtF), and the impact of AcrB pathway mutations.

ABSTRACT

A major contribution of the resistance-nodulation-cell division (RND)-transporter AcrB to resistance to oxazolidinones and pleuromutilin derivatives in Escherichia coli was confirmed. However, we discovered significant differences in efflux inhibitor activities, specificities of the homologous pump YhiV (MdtF), and the impact of AcrB pathway mutations. Particularly, entrance channel double-mutation I38F I671T and distal binding pocket mutation F615A revealed class-specific transport routes of oxazolidinones and pleuromutilin derivatives. The findings could contribute to the understanding of the RND-type multidrug transport pathways.

INTRODUCTION

The emergence of multidrug resistance (MDR) in Gram-negative pathogens has surpassed the supply of new antibiotics by far. Resistance-nodulation-cell division (RND)-type efflux transporters, such as AcrB from Escherichia coli, play a crucial role in MDR development (1). Their function is proton-motive force (PMF) dependent, and many of them are characterized by a remarkably broad substrate compatibility (2). AcrB constitutes the major MDR efflux pump in E. coli, with constitutive expression in wild-type strains and stable overexpression due to drug exposure (3). Close homologs of AcrB are the E. coli MDR transporters AcrF and YhiV (MdtF), which supposedly play a minor role, since at least in vitro, their expression could only be triggered after AcrB inactivation (4, 5). Among substrates extruded by AcrB are “Gram-positive drugs,” including members from the relatively new class of fully synthetic oxazolidinones, such as linezolid and tedizolid (6), and the upcoming semisynthetic pleuromutilin derivatives (7). Agents from both classes target intracellular bacterial protein biosynthesis at the translational level. Only the oxazolidinones linezolid and tedizolid and the pleuromutilin retapamulin have been in use for human treatment, but new compounds are under development even against Gram-negative pathogens (7–10).

We evaluated the impact of RND-type transporters on the susceptibility to available oxazolidinones and pleuromutilins in E. coli using standard broth microdilution assays as described previously (11). We included wild-type acrB and efflux-deficient strains (3, 5, 12), mutants expressing acrF and yhiV, and eight AcrB substrate pathway mutants (11, 13, 14) (Table 1). The oxazolidinones tested comprised the experimental agents sutezolid (10) and radezolid (9) and the oxazolidinone-fluoroquinolone conjugate cadazolid (15). For completeness, we included previous data for linezolid (5, 13) and tedizolid (data for the latter are already available for wild-type E. coli, a ΔAcrB mutant, and the I38F I671T strain [6, 11]). Among the pleuromutilins, we tested the veterinary drugs tiamulin and valnemulin as well as retapamulin and their precursor pleuromutilin.

TABLE 1.

Strains and mutants used in this study

E. coli strain or mutant	Description	Source or reference
Strains
ATCC 25922	Reference E. coli strain (constitutive wild-type acrB expression)	DSMZa , no. DSM-1103
AG100	K-12 derivative [argE3 thi-1 rpsL xyl mtl Δ(gal-uvrB) supE44], parent of 3-AG100 (constitutive wild-type acrB expression)	12
3-AG100	AG100 derivative (wild-type acrB overexpressed)	3
2-DC14PS	AG100 derivative, acrAB knockout mutant (wild-type acrF overexpressed)	4
DKO20/1	2-DC14PS derivative, acrAB acrF double-knockout (wild-type yhiV overexpressed)	5
TKO	acrAB acrF yhiV triple knockout	5
AcrB mutantsb
F610A	Distal binding pocket mutant	13
F136A	Distal binding pocket mutant	13
F178A	Distal binding pocket mutant	13
F615A	Distal binding pocket (G-loop) mutant	13
F617A	Distal binding pocket (G-loop) mutant	13
F628A	Distal binding pocket mutant	13
G616N	Distal binding pocket (G-loop) mutant	14
I38F I671T	Entrance pathway mutant	11
ΔAcrB	Deletion mutant (whole acrB; resistance cassette removed)	This studyc
D408A	Transmembrane mutant (impeded proton motive force generation)	This studyc

^aLeibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures (https://www.dsmz.de).

^bChromosomal mutants, derived from E. coli 3-AG100.

^cRecombination method used for site-directed mutagenesis as described previously (13), with oligonucleotides given in the supplemental material.

High efflux compatibility was confirmed for both drug classes as demonstrated by ≥64-fold increased susceptibilities due to acrB deletion. Notably, overexpression versus constitutive acrB expression caused 16- and ≥128-fold increased resistance to radezolid and cadazolid, respectively (Table 2). To exclude efflux-independent effects due to smaller amounts of integral membrane proteins in a deletion mutant, we engineered an AcrB mutant D408A (Table 1). This mutation has been shown to prevent PMF generation without compromising the incorporation of functionless AcrB proteins (16, 17). We proved similar susceptibilities of the D408A and ΔAcrB mutants and the triple-knockout (TKO) mutant additionally lacking AcrF and YhiV (Table 2). The contribution of other transporters to MDR appeared negligible as shown by previous expression studies with acrB-deficient E. coli mutants (5, 6). Among the drugs tested, only the susceptibility to linezolid was significantly increased further in the ΔAcrB mutant by the PMF-inhibitor CCCP at 10 μM (Table 2), a concentration not far from the intrinsic MIC of the mutant (20 μM), but 5 μM CCCP did not show any impact (data not shown). Notably, resistance breakpoints for linezolid and tedizolid (4 μg/ml and 0.5 μg/ml, respectively, with Staphylococcus spp. [EUCAST clinical breakpoint tables v. 6.0]) were not reached in the efflux-deficient mutants. However, susceptibility increases appeared promising with radezolid, cadazolid, and the pleuromutilins regarding repurposing options for Gram-negative application under efflux inhibition.

TABLE 2.

Susceptibilities of E. coli strains expressing wild-type and mutated RND-type transporters

E. coli strain or mutant	MIC (μg/ml) for:a
	Oxazolidinones					Pleuromutilins
	Linezolid	Tedizolid	Sutezolid	Radezolid	Cadazolid (FQb conjugate)	Tiamulin	Valnemulin	Retapamulin	Pleuromutilin
Wild-type RND-transporter-expressing strains
ATCC 25922	256	>128	>512	16	2	256	64	32	256
AG100	256	32	512	8	2	128	32	16	128
3-AG100 (acrB overexpressed)	1024	>128	>512	128	>256	512	64	64	>256
With 25 μg/ml PAβN	128	16	128	16	2	8	4	2	4
With 100 μg/ml NMP	32	128	256	4	256	256	64	32	64
2-DC14PS (ΔacrAB acrF overexpressed)	256	>128	512	32	>256	256	64	64	256
DKO20/1 (ΔacrAB ΔacrF yhiV overexpressed)	8	32	32	1	256	64	16	16	16
AcrB drug pathway mutantsc
I38F I671T	16	16	32	2	64	128	128	32	32
F178A	64	>128	128	32	64	128	16	16	128
F615A	512	>128	512	64	32	32	4	8	32
G616N	512	>128	512	64	>256	256	64	32	256
F136A	1024	>128	512	64	128	128	16	16	128
F628A	512	>128	512	32	128	256	32	32	256
F617A	512	>128	512	64	128	128	8	16	256
F610A	256	32	512	8	32	32	8	8	32
Efflux-deficient mutantsc
ΔAcrB	16	2	8	0.5	0.25	0.5	0.25	0.125	0.5
With 25 μg/ml PAβN	8	0.5	4	0.5	0.06	0.125	0.03	0.03	0.25
With 100 μg/ml NMP	16	1	8	0.5	0.125	0.25	0.125	0.06	0.25
With 10 μM CCCP	4	1	4	0.5	0.125	0.25	0.125	0.06	0.25
D408Ad	16	4	8	0.5	0.5	1	0.5	0.25	0.5
TKO (ΔacrAB ΔacrF ΔyhiV)	8	1	4	0.25	0.5	0.25	0.125	0.06	0.125

^aMIC decreases >4-fold compared to the MIC detected with parental E. coli 3-AG100 (acrB overexpressed) are in boldface font; >, MIC determination limited due to drug precipitation at concentrations above the indicated values.

^bFQ, fluoroquinolone.

^cMutants derived from parental E. coli strain 3-AG100 (TKO from DKO20/1).

^dEfflux deficiency due to a point mutation preventing the proton generation for PMF.

The efflux pump inhibitor (EPI) 1-(1-naphthylmethyl)-piperazine (NMP) (18) enhanced the activity of oxazolidinones up to 32-fold but only up to 2-fold that of the pleuromutilin derivatives. In contrast, with Phe-Arg-β-naphthylamide (PAβN), an EPI with proven outer membrane activity (19), at least 16-fold enhancement was found with all of the pleuromutilins tested (Table 2). The differing EPI activities might give an additional hint on the drug pathway specificities.

The homologous transporter AcrF can almost completely replace AcrB (1, 4, 5), also with respect to the oxazolidinones and pleuromutilins tested (Table 2). In contrast, the substrate range of YhiV is narrower. For instance, linezolid is a poor substrate (5), and this was also the case with radezolid. However, the efflux of tedizolid, sutezolid, and the pleuromutilins appeared significantly less impeded as shown by MIC decreases in the yhiV-expressing mutant that were far from those occurring due to efflux deficiency (Table 2). Surprisingly, YhiV mediated high resistance to the fluoroquinolone-oxazolidinone conjugate cadazolid, whose activity was restored when knocking out yhiV (Table 2). The reason for the distinct specificities of YhiV remains to be elucidated.

Regarding AcrB substrate pathway mutations investigated in this study, we could broadly define those revealing a single compound effect, marginal effects on resistance to drugs from both classes, class-independent overall resistance decreases, and an oxazolidinone or a pleuromutilin specificity. A unique impact occurred from mutation F617A located at the so-called G-loop (switch-loop) separating the distal from the proximal substrate binding pockets (20). It significantly increased the susceptibility only to the largest member of the pleuromutilin derivatives, valnemulin (molecular weight [MW], 564.8), whereas minor (≤4-fold) or no effects were seen with the other agents tested (Table 2). Notably, adjacent G-loop mutation G616N severely affecting macrolide resistance in E. coli (14, 20) did not cause any significant alterations with either pleuromutilins or oxazolidinones. Similar marginal effects with drugs from both classes appeared from distal binding pocket mutations F136A and F628A. In contrast, mutation F610A located across the G-loop (Fig. 1) severely increased the susceptibility to nearly all drugs from both classes. A minor impact was only found on resistances to linezolid and sutezolid, with MIC decreases of no more than 4-fold (Table 2). Sutezolid differs from linezolid only by sulfur replacing the oxygen of the morpholine ring.

FIG 1.

AcrB binding state monomer (PDB 2HRT, chain B); visualization using PyMOL (Molecular Graphics System version 2.1.1, Schrödinger, LLC). (A) Part of the AcrB binding state monomer shown as cartoon (side view from the periplasmic cleft). PC1, PC2, PN1, and PN2, subdomains of the porter domain; Dd, docking domain; TM, transmembrane domain. (B) Enlarged part of the porter domain, with PC1 clipped. Blue sticks, phenylalanine side chains of the distal binding pocket; gray sphere, position of G616 (no side chain); red sticks, isoleucine side chains constituting a narrowing in the lower porter domain; red and black dashed-line ovals, residues critical for the efflux of oxazolidinones and pleuromutilin derivatives, respectively.

A significant oxazolidinone-specific impact appeared from double mutation I38F I671T located in a putative entrance channel of the lower porter domain (Fig. 1). It decreased the resistance to all of the tested oxazolidinones up to 64-fold, whereas the effects with pleuromutilin derivatives remained negligible (Table 2). It was shown previously that substitution of I38 and/or I671 by bulkier residues predominantly impedes the efflux of smaller drugs, whereas larger compounds, such as macrolides and rifamycins, supposedly use another entrance pathway (11). Notably, in contrast to that for the pleuromutilin derivatives (MW, >490), resistance to the smaller pleuromutilin (MW, 378.5) was significantly affected by I38F I671T (Table 2). Among the distal binding pocket mutations, the most remarkable oxazolidinone-specific impact occurred with F178A (Table 2, Fig. 1), whereas the largest pleuromutilin specificity arose from mutation F615A removing a phenyl residue from the G-loop (Fig. 1). In this mutant, the susceptibility to all pleuromutilins was significantly increased but not that to the oxazolidinones (except cadazolid) (Table 2).

Our study reveals evidence of different pathways for oxazolidinones and pleuromutilin derivatives through RND-type transporters. The knowledge of class-specific efflux determinants could elucidate structural requirements of competitive inhibitors or efflux-incompatible drugs.