Efflux of chloramphenicol by the CmlA1 protein

Abstract

The cmlA1 gene cassette contains the cmlA1 gene, that confers resistance to chloramphenicol, as well as a promoter and translational attenuation signals, and expression of cmlA1 is inducible by low concentrations of chloramphenicol. The CmlA1 protein encoded by cmlA1 was localised in the inner membrane. Active efflux of chloramphenicol, additional to the endogenous efflux from Escherichia coli cells, was observed when the cmlA1 gene was present and the production of CmlA1 had been preinduced with subinhibitory concentrations of chloramphenicol. Both endogenous and CmlA1-mediated export of chloramphenicol was driven by the proton-motive force.

Chloramphenicol efflux, CmlA1 inner membrane protein

1 Introduction

Early studies on the plasmid-encoded determinants that confer resistance to chloramphenicol revealed that most encode chloramphenicol acetyltransferases [1]. However, ‘non-enzymatic’ chloramphenicol resistance that was induced by exposure to subinhibitory levels of chloramphenicol was also found [1,,3]. Reduced uptake of chloramphenicol was demonstrated in some cases and it was suggested that reduced entry of chloramphenicol into the cells was responsible for the resistance [4]. The first of these chloramphenicol resistance determinants to be characterised were found in the Inc P plasmid R26 [3] and in transposon Tn1696 on the Inc P plasmid R1033 [5]. Both plasmids were isolated from multidrug-resistant Pseudomonas aeruginosa strains. R26 is closely related to R1033 and contains a transposon equivalent to Tn1696[6]. Hence the chloramphenicol resistance genes are likely to be the same. The sequence of this gene, formerly cml or cmlA but now designated cmlA1, from Tn1696 predicts a polypeptide of 419 amino acids with a molecular mass of 44.2 kDa [7,8]. A shorter polypeptide of a size consistent with the estimated size of 31 kDa for the product visualised in minicells was predicted from the sequence of the cml determinant from R26 [9]. However, the sequences of cml (R26) and cmlA1 (Tn1696) can be aligned to reveal several frameshift errors in the former [7]. The CmlA1 polypeptide is highly hydrophobic, which may explain the failure of CmlA1 to run true to size. The cmlA1 gene is part of a mobile gene cassette and both a promoter and signals indicating that chloramphenicol-induced gene expression is regulated by translational attenuation were identified within the cassette [7]. This is consistent with the observation that regulation of CmlA1 induction is posttranscriptional [10].

Hydropathy plots of CmlA predict 12 transmembrane segments [7,8] and the CmlA1 protein was found to be weakly related to 12 other transmembrane segment transporters [8,11]. Hence, chloramphenicol resistance has been assumed to be effected by efflux of the antibiotic. To test this hypothesis, we have examined the intracellular location of CmlA1 and the uptake and efflux of chloramphenicol from Escherichia coli cells containing the cmlA1 gene, both with and without induction by chloramphenicol.

2 Materials and methods

2.1 Strains and plasmids

E. coli strains used were TG1 (supE, hsdΔ5, thi, Δ(lac-proAB), F′[traD36, proAB⁺, lacI^q, lacZΔM15]); N100 (recA3) [12] and AN718 (F⁻, uncA401, argH, pyrE, entA) [13]. The 1.9-kb HindIII fragment within Tn1696 that completely includes the cmlA1 gene cassette and hence the cmlA1 resistance determinant, the promoter, and the regulatory sequences for induction [7] was recovered, ligated to pUC18, previously cut with HindIII and treated with alkaline phosphatase [14]. After transformation into E. coli TG1 competent cells, recombinants were recovered by plating on LB agar containing chloramphenicol (20 μg ml⁻¹) and ampicillin (50 μg ml⁻¹). To ensure regulated expression of CmlA1 from the promoter in the cassette, a clone with the cmlA1 gene in the opposite orientation to the lac promoter in pUC18, designated pAMG1, was used for this study.

2.2 Maxicell analysis

Plasmid-encoded proteins were labelled with U-¹⁴C-amino acids (specific activity 1.85 MBq ml⁻¹; Amersham, UK) using a modified maxicell method [12]. After labelling, cells were pelleted by centrifugation, washed once, and part of the sample (whole cell extract) was resuspended in 50 μl of sample buffer [15]. Part of the sample was fractionated into inner and outer membranes, periplasm and cytoplasm [16]. Fractions were precipitated in 5% trichloroacetic acid, washed twice in acetone, and finally resuspended in sample buffer. All samples were then boiled for 3 min, cooled, and subjected to electrophoresis in sodium dodecyl sulfate (SDS)–polyacrylamide gels cast as 4% stacking and 10% resolving gels [15]. Following electrophoresis, the gels were treated with a fluorographic enhancer (En³hance; New England Nuclear), dried on 3MM paper, and exposed to Fuji RX X-ray film for 2–3 days at −80°C. The apparent sizes of maxicell proteins were estimated by comparison to prestained molecular mass standards (Bio-Rad Laboratories) run in the same gels.

2.3 Transport assays

Overnight Luria–Bertani broth cultures of TG1(pUC18) and TG1(pAMG1) were diluted 1/100 in fresh medium with or without chloramphenicol (1 μg ml⁻¹) and grown to an OD_530nm of 0.6. Transport assays were performed as described [17]. Cells were harvested and washed once in 1 vol. of assay buffer (50 mM KPO₄, pH 6.6, 1.0 mM MgSO₄), then suspended in assay buffer at an OD_530nm of 4.0. Cells were preincubated at 37°C for no more than 30 min before use in uptake assays. Samples of cells were energised with 20 mM d-glucose (10 min) and the assays were commenced by the addition of d-threo-[dichloroacetyl-1-¹⁴C]chloramphenicol (55 mCi/mmol; Amersham) at a final concentration of 5 μM. Samples were removed at intervals, diluted 100-fold in 100 mM KPO₄–100 mM LiCl, pH 6.6, filtered through 0.45 μm pore size GN-6 Metricel membrane filters (Gelman Sciences Inc.), and washed with 5 ml of the same buffer, using a filtration manifold and vacuum pump pressure of 18 psi. Filters were dried at 37°C and counted in a xylene-based scintillant (ACSII; Amersham). Total protein was estimated by a modified Lowry procedure [18] using bovine serum albumin as a standard.

2.4 Transport assays in starved cells

To test whether the proton-motive force (pmf) drives efflux of chloramphenicol, plasmid pAMG1 was introduced into the E. coli strain AN718 uncA which is defective in the coupling of ATP synthesis to electron transport [13]. Starved uncA cells which are subsequently provided with d-lactate as the sole carbon source, are able to generate an energised membrane state and drive the transport of some compounds independently of oxidative phosphorylation. Overnight LB cultures of AN718 and AN718(pAMG1) were diluted 1/100 into M9 minimal medium [14] supplemented with 0.4% glucose, 0.001% thiamine, 0.1% casamino acids, and 3% LB broth, with ampicillin at 20 μg ml⁻¹ if needed. Induced cultures also included chloramphenicol at 1 μg ml⁻¹. Cultures were harvested in the mid-exponential phase of growth, washed once in unsupplemented M9 minimal medium without a carbon source, and starved for 4 h in the same medium containing 5 mM 2,4-dinitrophenol as described [19]. Starved cells were washed four times with M9 minimal medium and suspended in assay buffer to a final OD_530nm of 4.0. After 30 min incubation at 37°C, 5 μM [¹⁴C]chloramphenicol was added. After an equilibration period, 2.5 mg ml⁻¹d-glucose or d-lactate were added as energy sources. Samples were removed and filtered and counted as described above.

3 Results

3.1 Location of the CmlA1 protein

A 1.9-kb HindIII fragment from Tn1696 that contains the cmlA1 gene cassette was cloned into pUC18. The plasmid used in this study, pAMG1, contains the cmlA1 gene in the opposite orientation to the lacZ promoter of pUC18 (Fig. 1). The maxicell method for detection of plasmid-encoded proteins was used to detect the CmlA1 protein. In a whole cell extract of TG1(pAMG1) (Fig. 2A, lane 3), the apparent molecular mass of CmlA1 was 34 kDa, consistent with the previous estimate [9]. The mature (29 kDa) and pre-processed (31 kDa) forms of the Bla_TEMβ-lactamase encoded by the bla_TEM gene of pUC18 are also visible. In fractionated extracts, the Bla proteins were found in the periplasmic fraction whereas CmlA1 was concentrated in the inner membrane fraction (Fig. 2B, lane 2), the predicted location for a major facilitator family transporter.

1

Structure of pAMG1. The cmlA1 gene cassette is represented as a part open and part filled (representing the 59-base element) box. The cmlA1 gene is depicted as an arrow below; the small arrow labelled ‘L’ represents the coding region of the nine amino acid leader peptide. The cmlA1 promoter is shown as a rightward-facing bent arrow. ‘H’ represents the HindIII sites flanking the cloned 1909-bp HindIII fragment and the hatched boxes are pUC18 sequences. The leftward-facing bent arrow is the lacZ promoter in the vector.

2

Identification and localisation of the CmlA1 protein. Proteins from whole cells (A) or fractionated extracts (B) labelled by the maxicell procedure were run in an SDS–polyacrylamide (10%) gel and detected by fluorography. A: E. coli N100 (lane 1), N100(pUC18) (lane 2), and N100(pAMG1) (lane 3). B: All samples are from E. coli N100(pAMG1). Proteins are from the total membrane fraction (lane 1), cytoplasmic membrane (lane 2), periplasmic fraction (lane 3), outer membrane (lane 4), and whole cell extract (lane 5). The positions of prestained molecular size standards run in the same gel are on the left and the position of CmlA1 is indicated by the arrows.

3.2 The cmlA1 gene confers resistance to chloramphenicol by active efflux

The presence of CmlA1 in E. coli cells reduced the accumulation of radiolabelled chloramphenicol (Fig. 3) relative to control cells. This reduced accumulation was most apparent in cells that had been preinduced with subinhibitory levels of chloramphenicol, though uninduced TG1(pAMG1) cells exhibited a slightly lower accumulation of chloramphenicol than TG1 containing only the pUC18 vector. The addition of the protonophore, 2,4-dinitrophenol (indicated by the arrows in Fig. 3), which dissipates the energised membrane state [20], caused an immediate influx of drug in all cases. This result reflects the presence of endogenous chloramphenicol efflux systems in E. coli[17,20,,22]. Both the endogenous efflux and the CmlA1-dependent efflux were dependent on an energised membrane.

3

Accumulation of chloramphenicol in E. coli TG1 cells. The uptake of radiolabelled chloramphenicol by TG1(pUC18) cells uninduced (▵) or induced with chloramphenicol (1 μg ml⁻¹) (▴), and by TG1(pAMG1) cells uninduced (◯) or induced (?), was assayed as described in Section 2. Cells were de-energised by the addition of 2 mM 2,4-dinitrophenol at the time indicated by the arrow. Values represent the average of two to four determinations.

To establish whether the CmlA1-dependent efflux leading to the reduced uptake of chloramphenicol was energised by the pmf, we examined transport in E. coli cells containing an uncA mutation [23]. The E. coli uncA mutant strain AN718 contains a defective F₀F₁-ATPase and cannot synthesise ATP via electron transport. In starved uncA cells, a distinction between energy derived from substrate level phosphorylation or from the pmf can be made on the basis of whether lactate is able to energise transport [19,23]. Whereas glucose can drive transport energised by ATP, derived from either substrate level phosphorylation or the pmf, lactate can only drive transport that is dependent on the pmf. Chloramphenicol transport was examined in AN718 and AN718(pAMG1) cells, following starvation and subsequent energisation with glucose (Fig. 4A) or lactate (Fig. 4B). Both glucose and lactate energised the endogenous as well as the additional CmlA1-mediated efflux of chloramphenicol, as indicated by the rapid reduction in accumulated chloramphenicol following the addition of glucose (Fig. 4A) or lactate (Fig. 4B) to starved cells. Thus, the energised membrane state must be derived from the pmf and not from oxidative phosphorylation, and chloramphenicol efflux mediated by the inner membrane CmlA1 transporter is pmf-dependent.

4

Efflux of chloramphenicol in starved E. coli AN718 cells. Cells were starved for 4 h in the presence of 5 mM 2,4-dinitrophenol prior to addition of radiolabelled chloramphenicol (5 μM). An energy source, 0.25% (w/v) glucose (A) or lactate (B), was added at the time indicated by the arrows. Cell-associated chloramphenicol concentrations are expressed as a percentage of the steady state (15 min time point) for AN718 uninduced (▵) or induced (▴), and AN718(pAMG1) uninduced (◯) or induced (?). Points represent the average of two or four determinations.

4 Discussion

We have shown that the CmlA1 protein encoded by the cmlA1 gene cassette is located in the inner membrane and confers resistance to chloramphenicol by pmf-driven active efflux of chloramphenicol from E. coli cells. A significant level of CmlA1-mediated efflux was only observed when the cells had been pregrown in the presence of a subinhibitory concentration of chloramphenicol and this is consistent with the fact that expression of the CmlA1 protein is regulated posttranscriptionally via a translational attenuation mechanism with chloramphenicol as the inducer [7,24]. The endogenous efflux of chloramphenicol from susceptible E. coli cells which is also pmf-driven appears to be mediated by the multi-drug exporters Mdf1 [22] and AcrAB [21]. In contrast, tetracycline is actively accumulated by susceptible E. coli cells [25] and the effect of the dissipation of the pmf by the addition of a protonophore causes the rapid egress of that drug.

Recently, genes encoding proteins that are related to CmlA1 have been identified. Two of these genes, cmlA2[26] and cmlA4[27], are, like cmlA1, part of a gene cassette and are also preceded by translational attenuation signals. As CmlA2 and CmlA4 are 90% and 98% identical to CmlA1, it seems reasonable to conclude that their mechanism of action is the same. A number of variants of a further gene, here designated cmlA3 but known variously as pp-flo, floSt or floR, has been found in several different bacterial species [28,,,31] and this gene confers resistance to florfenicol as well as to chloramphenicol. The CmlA3 protein is close to 50% identical to CmlA1, indicating that it too is likely to confer resistance via an efflux mechanism, albeit with a modified substrate specificity that permits the efflux of florfenicol.

Acknowledgments

This work was supported by a CSIRO/UTS Collaborative Research Grant. We thank Ann Abeles for supplying N100 and Ron Skurray for AN718.