Mdt(A), a New Efflux Protein Conferring Multiple Antibiotic Resistance in Lactococcus lactis and Escherichia coli

Abstract

The mdt(A) gene, previously designated mef214, from Lactococcus lactis subsp. lactis plasmid pK214 encodes a protein [Mdt(A) (multiple drug transporter)] with 12 putative transmembrane segments (TMS) that contain typical motifs conserved among the efflux proteins of the major facilitator superfamily. However, it also has two C-motifs (conserved in the fifth TMS of the antiporters) and a putative ATP-binding site. Expression of the cloned mdt(A) gene decreased susceptibility to macrolides, lincosamides, streptogramins, and tetracyclines in L. lactis and Escherichia coli, but not in Enterococcus faecalis or in Staphylococcus aureus. Glucose-dependent efflux of erythromycin and tetracycline was demonstrated in L. lactis and in E. coli.

Lactococci are important lactic acid bacteria used in the process of preparing fermented dairy products. Naturally found on plants and on parts of the body of cows, these bacteria are widely used as a starter culture in the dairy industry (37). Antibiotics used in animal husbandry have selected for antibiotic-resistant flora (44). Such resistant bacteria may contaminate milk and meat and persist in food made from their raw materials. Indeed, Lactococcus lactis subsp. lactis K214 isolated from a raw milk soft cheese has been shown to harbor a multiple antibiotic resistance plasmid (30). This plasmid, pK214, carries genes for chloramphenicol acetyltransferase (cat) and streptomycin adenylase (str), a tetracycline resistance gene [tet(S)], and a putative drug efflux gene previously named mef214 (30, 38).

Efflux proteins, membrane proteins distributed among gram-positive and gram-negative bacteria, are involved in transmembrane export of different substances such as heavy metals, organic solvents, dyes, disinfectants, and antibiotics (21, 22, 26, 35). Some efflux proteins are specific to a single class of drugs, while others may transport a variety of chemically different compounds. These proteins have been broadly classified into two groups: the ATP-binding cassette (ABC) transporters (10, 28) and the secondary transporters (29). Drug efflux proteins in pathogens can mediate resistance causing therapeutic failures.

Two multidrug transporters have been described in L. lactis. The first (LmrA) is a member of the ABC superfamily (3); the second (LmrP) is a proton-force-dependent transporter (4). Both transporters confer resistance to ethidium bromide, daunomycin, and tetraphenylphosphonium. Recent work shows resistance to macrolides, lincosamides, and streptogramins and tetracycline associated with LmrP expression (39).

Characterization of mef214 demonstrated that it mediated multiple drug resistance; hence, it has been renamed mdt(A) for multiple drug transporter. Mdt(A) is a plasmid-specified protein which is unusual in that it is related to the major facilitator superfamily (MFS) and yet contains two motifs C highly conserved in antiporters (41) and a putative ATP-binding site. The mdt(A) gene was cloned and studied for phenotype and function.

MATERIALS AND METHODS

Bacterial strains, growth conditions, and plasmids.

L. lactis subsp. lactis K214 bearing plasmid pK214 (30), L. lactis subsp. cremoris MG1363 (12) and LM0230 (9) were grown in M17 broth (Oxoid, Inc., New York, N.Y.) supplemented with 0.5% (vol/vol) glucose (GM17) at 30°C. Enterococcus faecalis JH2-2 (16) and Staphylococcus aureus RN4220 (19) strains were grown in brain heart infusion (BHI) broth (Oxoid) at 37°C. Escherichia coli strains DH5α (Life Technologies, Gaithersburg, Md.), AG100 (13), AG100A (Δacr) (27), and transformants were grown in Luria-Bertani (LB) broth at 37°C. Plasmid pUC19 (Life Technologies) and the shuttle vector pWM401 (43) were used as cloning vectors. Plasmids pK214 (30), pWM401, and pWVP6 were maintained in the strains by adding 20 μg/m of chloramphenicol per ml to the cultures, whereas the pUC vectors were maintained with 60 μg of ampicillin per ml. Solid media were prepared by the addition of 1.2% (wt/vol) agar (Oxoid) to broth.

Plasmid construction and transformation.

The mdt(A)-containing region from plasmid pK214 was amplified by PCR and cloned into vector pUC19 resulting in plasmid pUVP6. To allow transformation and expression of mdt(A) in both gram-negative and gram-positive bacteria, mdt(A) was additionally isolated from pUVP6 with XbaI and SphI restriction enzymes and inserted into the TetC determinant of the shuttle vector pWM401. This new plasmid, pWVP6, as well as the vector pWM401, was transformed into E. coli DH5α and AG100A by heat shock and into E. faecalis JH2-2, S. aureus RN4220, and L. lactis MG1363 and LM0230 by electrotransformation. Plasmids were transformed into L. lactis, E. faecalis, and S. aureus cells by electroporation in the Gene Pulser apparatus with the Pulse Controler (Bio-Rad Laboratories) as described previously (11, 17, 34). E. faecalis and S. aureus transformants were selected on BHI agar plates and Lactococcus transformants were selected on GM17 agar plates, each containing 10 μg of chloramphenicol per ml. E. coli cells were transformed by heat shock treatment (33).

Antibiotic susceptibility tests.

The MIC was determined with E-test strips (a gift of AB Biodisk, Solna, Sweden). The antimicrobial agents tested were azithromycin, clarithromycin, erythromycin, clindamycin, tetracycline, doxycycline, minocycline, nalidixic acid, ciprofloxacin, clinafloxacin, fleroxacin, norfloxacin, pefloxacin, sparfloxacin, quinupristin-dalfopristin, streptomycin, gentamicin, kanamycin, benzylpenicillin, piperacillin, amoxillin-clavulanic acid, imipenem, oxacillin, chloramphenicol, fusidic acid, ceftriaxone, cefepime, cefotaxime, cefoxitin, ceftizoxime, cephalothin, bacitracin, and rifampin. The MICs of lincomycin and spiramycin were determined by microdilution test according to NCCLS guidelines (25).

DNA techniques.

Plasmids were isolated from Lactococcus, Enterococcus, and Staphylococcus strains by the procedure of Anderson and McKay (2) by adding 10 μg of lysostaphin per ml to the lysozyme solution for the lysis of staphylococci. E. coli plasmids were isolated with Qiagen Miniprep Kit (Qiagen, Inc., Valencia, Calif.). Restriction enzyme digestions were performed according to the suppliers' directions. DNA was analyzed in 0.8% (wt/vol) agarose gels in TAE buffer (33). mdt(A) was amplified by PCR from plasmid pK214 (GenBank accession no. X92946) using Taq DNA polymerase in accordance with the manufacturer's protocol (Life Technologies). A ClaI restriction enzyme site was incorporated into both forward (mdt1, 5′-GATGATATCGATGACAATGCAATGATGG [positions 10163 to 10180 in pK214]) and reverse (mdt1R, 5′-TTCCAGATCGATATCAAACTGACTGTG [positions 12058 to 12041]) primers to facilitate cloning into pUC19. Prior to the ligations (33), the digested PCR product, DNA fragments, and the cloning vectors were purified from the agarose gel with the Qiaex II Gel Extraction Kit (Qiagen). Sequencing was performed at the Tufts University Core Facility using a 373 stretch DNA sequencer (Applied Biosystems).

Drug efflux by starved-cell assay.

Bacterial cells (5 ml) were grown to an A₆₀₀ of 0.4, washed once in 0.1 M HEPES (pH 7.0) supplemented with 0.9% (wt/vol) NaCl, and resuspended in the same buffer to obtain a final A₆₀₀ of 1.0. E. coli cells were starved for at least 2 h at 37°C. Lactococcus cells were starved at 30°C overnight. The starved cells were centrifuged and resuspended in HEPES- sodium salt at A₆₀₀ of 1.0 and incubated at 30°C. For tetracycline assays, radiolabeled tetracycline (0.2 μg of [³H]tetracycline per ml [specific activity, 0.81 Ci/mmol]; Dupont/NEN Research Products, Boston, Mass.) was added. After 10 min, the cell suspension was divided into two halves, and 0.4% (vol/vol) glucose was added to one-half to energize the cells. Then, 100- μl samples were taken at different times, resuspended in 6 ml of a saline solution (0.9% [wt/vol] NaCl solution containing 1 μg of tetracycline per ml). The samples were filtered through 0.45- μm-pore-size Gelman GN6 Metricel membrane filters prewet with the saline solution. The filters were washed twice with 10 ml of saline solution. Filters were air dried, and the radioactivity was assayed in a scintillation counter. Each strain was tested in triplicate.

For the assay of erythromycin efflux, the procedure was performed as described above with the following modifications. Bacterial cells (100 ml) were grown to an A₆₀₀ of 0.7, washed once, and starved overnight in 0.1 M HEPES supplemented with 0.9% (wt/vol) NaCl. The starved cells were resuspended in HEPES-sodium salt at an A₆₀₀ of 2.0, and [N-methyl-¹⁴C]erythromycin (0.2 μg/ml; specific activity, 54 mCi/mmol; Dupont/NEN Research Products) was added. After a 5-min incubation, 0.4% (vol/vol) glucose was added, and 1-ml samples were directly filtered and washed with 0.9% (wt/vol) NaCl containing 1 μg of erythromycin per ml.

Drug efflux by actively growing cells.

Efflux assays were performed as described previously (45) with the following minor modifications: the cells were grown in Mueller-Hinton broth without antibiotics to an A₆₀₀ of 0.4; the cultures were incubated with 100 μM CCCP (carbonyl cyanide m-chlorophenylhydrazone), 20 mM arsenate, or 20 mM cyanide, where appropriate, for 10 min before the addition of 0.2 μg of [N-methyl-¹⁴C]erythromycin or 0.4 μg of [³H]tetracycline per ml.

RESULTS

Mdt(A) protein.

The mdt(A) gene, 1,257 bp in length (positions 10534 to 11790 in pK214; GenBank accession no. X92946), specifies a putative 418-amino-acid protein with a calculated molecular mass of 45.6 kDa (Fig. 1). A putative ribosome-binding site capable of base pairing with the 3′ end of L. lactis 16S rRNA (UCUUUCCUCCA) (5) starts six nucleotides upstream of the ATG initiation codon. The promoter region of mdt(A) corresponds to the conserved sequence consensus of lactococcal promoters (8). Putative −10 and −35 promoter sequences exist at nucleotides 10449 and 10426, respectively. The two hexamers are separated by a 17-bp sequence, a TG dinucleotide is present at position −15, and an AT-rich region precedes the −35 sequence (Fig. 1).

FIG. 1.

Nucleotide and deduced amino acid sequence of mdt(A) and the flanking regions. Putative ribosomal binding site (RBS), promoter elements (−10 and −35) and 12-TMS (boxes), which were identified by the TopPred II program (7), are indicated. Motifs A, B, C, G, and H correspond to the motifs described previously (29) and are shown in boldface, as is the putative ATP and GTP binding site. The position numbers of the nucleotide sequence are indicated according to the complete nucleotide sequence from pK214 (GenBank accession no. X92946). The primers sequences used for PCR amplification are underlined.

Hydropathy analysis (20) predicts a highly hydrophobic protein with 12 membrane-spanning regions and hydrophilic sequences at both the N and the C termini of the protein. Motifs conserved among the members of the 12- and 14-transmembrane segment (TMS) family of the MFS (29) were localized in Mdt(A). Motif A has been located in the putative cytoplasmic loop between TMSs 2 and 3; motif B was located within TMS 4. Two motifs C were found in the Mdt(A) protein: the first within TMS 5, as described for other proteins of the 12- and 14-TMS family, and the second within TMS 9. Motif C is a part of the antiporter motif (gX₃GPXiGGxl) which is highly conserved in the fifth membrane-spanning domain of the antiporters but which is absent in the symporters and uniporters (14, 41). A motif H, previously identified in the 14-TMS family proteins only, was situated on the C-terminal side of each motif C: one in TMS 6 and one in TMS 10. Motif G, conserved only in the 12-TMS family proteins, has been identified in TMS 11.

An ATP-binding motif A (Walker A) (42) corresponding to the highly conserved residues [AG]X₄GK[ST] of the ABC transporters spanned the junction of TMS 8 and the subsequent loop. There were no identifiable Walker B motifs in the protein sequence. Similarity searches of protein data banks using BLAST (National Center for Biotechnology Information) and LALIGN (15) programs revealed a paucity of close homology. There was a 32.9% identity overall to Mef(A) from Streptococcus pyogenes (6) and 32.1% identity to Mef(E) from Streptococcus pneumoniae (36). The highest amino acid identity of Mdt(A) with Mef(A) was identified in the first two TMSs of the α- and β-domains (44% with TMSs 1 and 2 and 41% with TMSs 7 and 8). The other TMSs (TMSs 3 to 6 and TMSs 9 to 12) and the cytoplasmic loop of Mdt(A) shared only 27 to 29% identity with Mef(A). Additionally, Mef(A) did not harbor any ATP-binding site motif. Mdt(A) had less than 26% identity with the other known transmembrane proteins. The identity of Mdt(A) with the lactococcal multidrug transporters was 18.8% for LmrP (4) and 16.5% for LmrA (40).

Phenotypic expression of mdt(A).

The mdt(A) gene specified drug resistance in L. lactis and in E. coli (Table 1) but not in E. faecalis and S. aureus (data not shown). The most prominent phenotypic expression was observed when the mdt(A) gene was placed into the acrAB-deleted E. coli strain AG100A. Here, Mdt(A) conferred increased resistance to the 14-membered macrolides erythromycin and clarithromycin, to a 15-membered azithromycin, to a 16-membered spiramycin, to lincosamides clindamycin and lincomycin, and to the streptogramin combination quinupristin and dalfopristin. Of note was the increased level of resistance to the tetracyclines, i.e., tetracycline, doxycycline, and minocycline (Table 1). In E. coli DH5α, resistance to the macrolides and clindamycin was also detected, but no conclusion could be drawn for lincomycin or the streptogramin antibiotics; the MICs of these antibiotics for E. coli DH5α were higher than the highest MIC tested. A 20-fold increased resistance to tetracycline was noted, but little change was observed in doxycycline and minocycline susceptibility.

TABLE 1.

Susceptibility of E. coli and L. lactis strains to macrolides, lincosamides, streptogramins, and tetracyclines as determined by the E-test

Strainsc	MIC (μg/ml)a
	Macrolides				Lincosamides		Streptogramins (Q-D)	Tetracyclines
	ERY	CLR	AZM	SPMb	CLI	LINb		TET	DOX	MIN
L. lactis
MG1363	0.064	0.064	0.125	16	0.125	16	2	0.19	0.125	0.094
MG1363/pWM401	0.064	0.064	0.125	16	0.125	16	2	0.19	0.125	0.094
MG1363/pWVP6	2	0.125	3	32	0.125	64	4	4	0.19	0.19
MG1363/pK214	1.0	0.064	1.5	32	0.125	64	3	>256‡	24‡	24‡
LM0230	0.064	0.032	0.094	16	0.125	32	3	0.19	0.094	0.094
LM0230/pWM401	0.064	0.032	0.094	16	0.125	32	3	0.19	0.094	0.094
LM0230/pWVP6	1.5	0.094	3	32	0.125	64	4	2	0.25	0.125
E. coli
DH5α	32	24	3	256	96	>1,024	>32	0.75	1.5	0.75
DH5α/pWM401	32	24	3	256	96	>1,024	>32	48†	8†	1.5†
DH5α/pWVP6	>256	48	24	512	>256	>1,024	>32	16	2	1
AG100A	3	2	0.75	64	1.5	256	3	0.25	0.19	0.064
AG100A/pWM401	3	2	0.75	64	1.5	256	3	48†	1.5†	0.19†
AG100A/pWVP6	48	12	4	256	32	1024	>32	8	0.75	0.75

^aFourteen-membered erythromycin (ERY) and clarithromycin (CLR); 15-membered azithromycin (AZM); 16-membered spiramycin (SPM), clindamycin (CLI), lincomycin (LIN), quinupristin-dalfopristin (Q-D), tetracycline (TET), doxycycline (DOX), and minocycline (MIN). Results are representative of experiments repeated three times. †, Tet(C) (uninduced); ‡, Tet(S) (uninduced). 

^bDetermined by microdilutions. 

^cPlasmids: pWM401, cloning vector [cat, tet(C)]; pWVP6 [cat, Δtet(C)::mdt(A)], pWM401 harboring mef214 cloned into tet(C); pK214, original plasmid [mdt(A), cat, str, tet(S)]. 

In L. lactis, the gene mediated a greatly increased resistance to the 14-membered macrolide erythromycin, but much less so to clarithromycin. Resistance to a 15-membered azithromycin was found, with a lower but reproducible twofold-increased MIC of a 16-membered spiramycin, as well as of the streptogramin combination quinupristin and dalfopristin. An increased MIC of lincomycin, but not of clindamycin, was observed. As noted for E. coli DH5α, Mdt(A)-mediated resistance (10-fold) to tetracycline in L. lactis, but resulted in only 1.5- to 3-fold-increased resistance to doxycycline and minocycline (Table 1).

Drug accumulation assays.

Actively growing cells were used to determine the amount of [³H]tetracycline and [¹⁴C]erythromycin accumulated in L. lactis MG1363 and MG1363/pWVP6 and in E. coli AG100A and AG100A/pWVP6. The use of the acrAB-deleted AG100A allowed measurement of the effective amount of antibiotics extruded by Mdt(A), since acrAB is responsible for endogenous tetracycline and erythromycin efflux in E. coli (26, 27).

Both E. coli and L. lactis strains expressing the mdt(A) gene accumulated less tetracycline and erythromycin than the wild-type hosts (data not shown). To examine the energy dependence of the decreased level of antibiotic accumulated, E. coli cells were first starved for 2 h and L. lactis cells were starved overnight, and then the cells were divided into two samples: one was given glucose either 5 or 10 minutes after the addition of a radiolabeled drug. [¹⁴C]erythromycin accumulated in both E. coli and L. lactis strains (Fig. 2A and B). A small, but detectable energy-dependent decrease (10%) in [¹⁴C]erythromycin accumulation could be detected in the wild-type L. lactis strain MG1363 after the addition of glucose (Fig. 2A). In contrast, an easily detected decreased [¹⁴C]erythromycin accumulation was noted for the mdt(A)-expressing L. lactis strain carrying pWVP6, which accumulated 50% less [¹⁴C]erythromycin than the nonenergized cells after 20 min (Fig. 2A).

FIG. 2.

Accumulation of [N-methyl-¹⁴C]erythromycin and [³H]tetracycline in L. lactis (A and C) and E. coli (B and D). The closed symbols represent the starved cells and the open symbols indicate cells energized with 0.4% glucose at 5 min for erythromycin and at 10 min for tetracycline (arrows) for the wild-type strains (diamonds), and for the strains carrying pWVP6 with mdt(A) (triangles). The results are representative of experiments performed in triplicate.

The starved E. coli AG100A and AG100A/pWVP6 cells showed similar levels of erythromycin accumulation; however, after the addition of glucose, the mdt(A)-carrying AG100A/pWVP6 cells accumulated reproducibly less radiolabeled drug than the starved cells AG100A/pWVP6 or AG100A (Fig. 2B).

The [³H]tetracycline accumulation assay (Fig. 2C and D) showed no difference between starved or glucose-treated wild-type L. lactis MG1363 cells. Starved MG1363/pWVP6 showed active efflux of the accumulated drug following the addition of glucose (Fig. 2C). In the susceptible strain E. coli AG100A, the addition of glucose to the starved cells stimulated the uptake of tetracycline due to the active tetracycline uptake known in E. coli strains (23). The opposite phenomenon, i.e., reduced [³H]tetracycline accumulation, occurred when glucose was added to the starved AG100A/pWVP6 cells containing mdt(A) (Fig. 2D).

In other experiments, we noted that the presence or absence of 100 μM CCCP, 20 mM arsenate, or 20 mM cyanide had no effect on the accumulation of [¹⁴C]erythromycin in L. lactis MG1363, nor did these inhibitors affect [¹⁴C]erythromycin accumulation in mdt(A)-expressing L. lactis MG1363/pWVP6. A similar absence of effect of these inhibitors was seen in E. coli strains harboring mdt(A) (data not shown).

DISCUSSION

The amino acid sequence analysis of Mdt(A), specified by the 30-kb multiple antibiotic resistance plasmid pK214 from L. lactis, revealed a new type of protein with 12 putative transmembrane-spanning domains bearing the motifs A, B, C, G, and H that are conserved in the efflux proteins of the MFS. Of distinction, Mdt(A) also possesses two antiporter motifs (motif C) distributed among the α- and β-domains (TMS 5 and TMS 9). The duplication of these motifs could ensue from an evolutionary duplication forming the two domains, like that described for TetA (32). Additionally, at the end of a helix in the β-domain, the Mdt(A) protein also contains an ATP-binding motif as described for ABC transporters (10). Considering the small number of residues that denote this motif, it may have no ATP-binding properties. Further experiments are necessary to elucidate whether Mdt(A) binds to ATP.

Mdt(A) shared 32.9% overall amino acid identity with Mef(A) from S. pyogenes, with the highest identity scores in the two first TMSs (TMSs 1 and 2 and TMSs 6 and 7) of each α- or β-domain. There are no antiporter motifs in these segments that may increase the level of identity. These TMSs might play an important role in the export of macrolide antibiotics. However, Mdt(A) had a wider antibiotic resistance spectrum than Mef(A), which only recognizes 14-membered and 15-membered macrolides. In E. coli AG100A, deleted of the AcrAB pump, the Mdt(A) protein conferred resistance to 14-, 15-, and 16-membered macrolides, lincosamides, streptogramins, and tetracycline antibiotics. Some differences in resistance profiles were seen among other hosts. L. lactis harboring mdt(A) remained susceptible to clindamycin, whereas mdt(A)-carrying E. coli did not. Also, although plasmid pK214 replicated in E. faecalis JH2-2 and conferred resistances to tetracycline [tet(S)], chloramphenicol (cat), and streptomycin (str) (38), the mdt(A) gene did not cause a detectable phenotype in this host, even after induction attempts on gradient plates using different concentrations of erythromycin. The tet(S), cat, and str genes are promiscuous genes that can be expressed in heterologous gram-positive genera, e.g., Staphylococcus, Listeria, Enterococcus, and Lactococcus. However, the mdt(A) gene follows a lactococcal promoter which is known to be functional in E. coli (18) but is more stringently expressed in gram-positive bacteria.

The mechanism of resistance appears to be active efflux (Fig. 2); however, the source of energy mediating efflux specified by Mdt(A) remains unknown. The amino acid structure resembles a proton-motive-force pump, but it also contains a putative ATP-binding site. While glucose addition following starvation caused efflux, protonophores known to destabilize the proton gradient across the bacterial cell membrane had no effect on the Mdt(A) pump expressed in E. coli or L. lactis. Efflux proteins that were not affected by using CCCP or arsenate have been previously reported, such as AcrD in E. coli (31), AmrAB-OprA in Burkholderia pseudomallei (24), and Tap in Mycobacterium fortuitum and Mycobacterium tuberculosis (1).

The Mdt(A) protein offers a new look into the functioning of energy-dependent efflux systems in bacteria, as well as antibiotic resistance mechanisms. Its structure is different from any antiporters previously described. Given its location on a plasmid in L. lactis, it is likely to be a resistance gene that was not selected in medical environments but rather in animals. Resistant bacteria selected on food-producing animals may contaminate milk or meat and persist in fermented food such as cheeses and sausages made with such raw items. The increasing presence of resistances among pathogenic bacteria, as well as commensals, should encourage a prudent and appropriate use of antibiotics in both public health and agriculture.