Role of Efflux Pumps in Adaptation and Resistance of Listeria monocytogenes to Benzalkonium Chloride

Abstract

In this study, potential mechanisms underlying resistance and adaptation to benzalkonium chloride (BC) in Listeria monocytogenes were investigated. Two groups of strains were studied. The first group consisted of strains naturally sensitive to BC which could be adapted to BC. The second group consisted of naturally resistant strains. For all adapted isolates, there was a correlation between the resistance to BC and ethidium bromide, but this was not the case for the naturally resistant isolates. To investigate the role of efflux pumps in adaptation or resistance, reserpine, an efflux pump inhibitor, was added to the strains. Addition of reserpine to the sensitive and adapted strains resulted in a decrease in the MIC for BC, whereas no such decrease was observed for the resistant strains, indicating that efflux pumps played no role in the innate resistance of certain strains of L. monocytogenes to this compound. Two efflux pumps (MdrL and Lde) have been described in L. monocytogenes. Studies showed low and intermediate levels of expression of the genes encoding the efflux pumps for two selected resistant strains, H7764 and H7962, respectively. Adaptation to BC of sensitive isolates of L. monocytogenes resulted in significant increases in expression of mdrl (P < 0.05), but no such increase was observed for lde for two adapted strains of L. monocytogenes, LJH 381 (P = 0.91) and C719 (P = 0.11). This indicates that the efflux pump Mdrl is at least partly responsible for the adaptation to BC.

The frequent use of disinfectants in food environments has raised concerns as to whether disinfectant-resistant isolates may emerge. Resistance to quaternary ammonium compounds (QACs) has been reported in many gram-positive as well as gram-negative bacteria associated with food (4). Cell membrane changes are believed to be responsible for this nonspecific increase in resistance. Efflux pumps, another important resistance mechanism, may be specific for one substrate or may transport a range of structurally dissimilar compounds (1). Such pumps can be associated with multiple drug resistance. In general, most genes coding for multidrug transporters are located on the chromosome, although some have been reported to be plasmid mediated (2, 7). It has been postulated that multidrug resistance efflux pumps may have been selected by bacteria to avoid the effects of toxic compounds present in their natural environment (2). However, once antimicrobial selective pressure is applied, mutants that overproduce these determinants can be selected, reinforcing their adaptive role in acquisition of resistance. It has been shown that a similar efflux pump mechanism can be induced by benzalkonium chloride (BC) in some strains of Listeria monocytogenes (1, 17, 20).

Two efflux pumps have been described in Listeria monocytogenes. An efflux pump, designated MdrL, can extrude antibiotics (macrolides and cefotaxime), heavy metals, and ethidium bromide (EtBr) (10). Another efflux pump, termed Lde, is associated with fluoroquinolone resistance and, in part, with resistance to acridine orange and EtBr (3). So far, research has indicated that the presence of the mdrL and lde genes, which perhaps are ubiquitous in L. monocytogenes, does not appear to be sufficient to impart resistance to disinfectants or antibiotics (3, 11, 14). Previous studies showed that L. monocytogenes exhibits morphological and physicochemical changes to the bacterial cell surface following adaptation to BC (20). It was also shown that after adaptation efflux pump activity was higher in originally sensitive strains than in strains possessing an innate resistance to BC. Alterations in membrane fatty acid composition after adaptation to BC were only demonstrated in the resistant strains tested and were not apparent in sensitive strains (20).

In previous work, several isolates of L. monocytogenes were characterized by automated ribotyping, pulsed-field gel electrophoresis, serotyping, and plasmid profiling, as well as their MICs for different sanitizers (14). However, the mechanisms involved in acquired and innate resistance remain unknown. The purpose of this research was to investigate the role of efflux pumps in adaptation of previously characterized isolates of L. monocytogenes that were either sensitive or resistant to BC and to examine the role of plasmids in natural resistance of L. monocytogenes to sanitizers. Although the genetic elements for control and expression of efflux pumps appear to be present in L. monocytogenes, little is known about the regulation of these putative mechanisms, either at the level of gene expression or the level of activity. However, the development of QAC resistance may be directly related to the efficiency of such an efflux system.

MATERIALS AND METHODS

Bacterial isolates and growth conditions.

Eight previously characterized isolates of Listeria monocytogenes (14) were used in this study. Sources of these isolates and their relative resistance to BC are presented in Table 1. Strains sensitive to BC (designated S) were adapted to tolerate higher BC concentrations with the protocol described by To et al. (20) and are designated as adapted strains (A). Bacteria were grown at 30°C on brain heart infusion agar (BHI; BD Diagnostic Systems, Sparks, MD), and prior to use, they were subcultured on two consecutive days in BHI broth and incubated with shaking at 30°C.

TABLE 1.

MICs of sanitizers and antibiotics for sensitive and adapted strains and naturally resistant, plasmid-cured strains of L. monocytogenes

Sensitivity group	Culture ID	MIC (mg/liter)a						Reference
		Clinicide	MCL	H2O2	AA	Gen	Kan
Sb	C511	0.77	1	18.75	2,500	2.25	6.25	14
	C511A	3.08	2	18.75	2,500	4.5	12.5	This study
	C719	0.77	1	18.75	2,500	2.25	6.25	14
	C719A	3.08	2	18.75	2,500	4.5	12.5	This study
	LJH381	0.77	1	9.4	630	0.56	>3.2	14
	LJH381A	3.08	2	18.75	2,500	2.25	6.25	This study
	LJH389	1.54	1	18.75	2,500	2.25	6.25	14
	LJH389A	3.08	2	18.75	2,500	4.5	6.25	This study
Rc	C716	3.08	4	18.75	2,500	5	25	14
	C716C	1.54	1	18.75	1,250	5	6.25	This study
	C717	3.08	4	18.75	2,500	5	25	14
	C717C	1.54	1	18.75	1,250	5	6.25	This study
	H7764	3.08	4	75	2,500	4.5	12.5	14
	H7764C	1.54	1	18.75	1,250	1.4	6.5	This study
	H7962	3.08	4	75	2,500	4.5	12.5	14
	H7962C	1.54	0.5	18.75	1,250	5.5	25	This study

^aData are from at least three independent measurements. Clinicide contains two quaternary ammonium compounds: 60% didecyl demethyl ammonium chloride and 40% benzalkonium chloride.

^bGroup S represents sensitive strains. Each strain has been adapted to benzalkonium chloride and now carries the original strain designation followed by A.

^cGroup R represents naturally resistant strains. Each strain has been treated with novobiocin to cure plasmids and now carries the original strain designation followed by C.

Antimicrobial agents.

Disinfectants used in the screening study included the following: BC (Sigma-Aldrich, St. Louis, MO); Clinicide, a mixture of two quaternary ammonium compounds as active ingredients (60% didecyl dimethyl ammonium chloride and 40% benzalkonium chloride; MTC Animal Health, Cambridge, ON, Canada); myristalkonium chloride (MCL; NASCHEM Inc., Mississauga, ON, Canada); bleach (Javex; Colgate-Palmolive, Toronto, Canada); acetic acid (AA; Fisher Chemicals, Fair Lawn, NJ); hydrogen peroxide, 30% (H₂O₂; Fisher Chemicals, Fair Lawn, NJ); and the intercalating dye EtBr (Fisher Chemicals). The antibiotics used were ampicillin sodium salt (Sigma-Aldrich, St. Louis, MO), gentamicin sulfate (Gen; Sigma-Aldrich), chloramphenicol (Sigma-Aldrich); tetracycline hydrochloride (Sigma-Aldrich), kanamycin monosulfate (Kan; Fisher Biotech, Fair Lawn, NJ), and novobiocin sodium salt (Sigma-Aldrich).

Susceptibility test.

MICs of disinfectants and antibiotics were determined using the broth dilution method (18). Strains of L. monocytogenes were tested in BHI using 96-well microtiter plates and an inoculum of 10⁴ to 10⁵ cells ml⁻¹. Growth was recorded after 24 h using a Victor² multilabel counter (Wallac Oy, Turku, Finland) set at 630 nm. The lowest concentration of disinfectant or antibiotic totally preventing growth was taken to be the MIC. In cases in which not all three replicates had the same results, the MIC test was performed once more.

Efflux pump inhibition with reserpine.

To determine activity of efflux pumps, the inhibitor reserpine (final concentration, 20 mg liter⁻¹; Sigma-Aldrich) was used. Changes to MICs of BC or EtBr were observed in either the absence or the presence of reserpine for all tested strains.

Plasmid curing.

Plasmid curing was performed according to the protocol described by Margolles and Reyes-Gavilan (8). Overnight cultures of both resistant and adapted strains of L. monocytogenes in tryptic soy broth (BD Diagnostic Systems, Sparks, MD) were inoculated (0.1 ml in 10 ml) into BHI broth containing novobiocin at a subinhibitory concentration (0.25 mg liter⁻¹). The culture was incubated for 24 h at 40°C, followed by 14 subcultures (1% inoculum) in the same medium containing novobiocin. After growth under these conditions, the cultures were plated on BHI agar, and individual colonies were subsequently picked, purified, and analyzed for the presence of plasmids according to the protocol described previously (14).

Nucleic acid extraction.

Overnight cultures grown on BHI agar were used for preparation of bacterial DNA. Single colonies were picked from plates, and cells were dispersed in 0.1 ml sterile water in microcentrifuge tubes. After boiling for 10 min in a water bath, samples were immediately placed on ice for 2 min and centrifuged for 6 min at 14,000 × g. The supernatant was used as a DNA template for PCR. RNA extraction for gene expression studies was performed using a QIAGEN RNeasy mini kit (QIAGEN, Germantown, MD).

PCR and sequencing of mdrL.

PCR amplification with primers MT1 (5′-AAGAATTCGA GCTGGTTG-3′) and MT2 (5′-AACTGCAGTGTGATACTTT-3′) was used to identify the mdrL gene in our population of listeriae (10). The PCR mixture consisted of buffer (10 mmol liter⁻¹ Tris-HCl, 50 mmol liter⁻¹ KCl, 2.5 mmol liter⁻¹ MgCl₂; pH 8.3), a 100 μmol liter⁻¹ concentration of each of the four deoxyribonucleoside triphosphates, 20 pmol of each of the two primers, and 0.5 U of AmpliTaq DNA polymerase in a total volume of 50 μl. For the amplification of the fragment, a MasterCycler thermocycler (Eppendorf-Netheler-Hinz GmbH, Hamburg, Germany) was programmed as follows: one denaturation cycle of 8 min at 94°C, 25 cycles (1 min at 94°C, 1 min at 48°C, and 1 min at 72°C), and an additional extension cycle of 5 min at 72°C. The expected size of the amplicon was 485 bp. PCR products were detected by 1.3% agarose gel electrophoresis in TAE buffer (40 mmol liter⁻¹ Tris-acetate, 1 mmol liter⁻¹ EDTA; pH 8.0) containing ethidium bromide (0.2 μg ml⁻¹). For sequencing, the PCR product was purified using the QIAquick PCR purification kit (QIAGEN Inc., Mississauga, ON, Canada) and sequenced at the Guelph Molecular Supercenter.

Gene expression profiling.

Real-time reverse transcriptase PCR (RT-PCR) was used to study gene expression of three genes. 23S rRNA was used as a reference gene. Primers for the 23S rRNA gene were forward (5′-GTGTCAGGTGGGCAGTTTG-3′) and reverse (5′-CATTCTGAGGGAACCTTTGG-3′) as described by Rudi et al. (15). New primers for mdrL were modified from those of Mata et al. (10) and now were mdrl1 (5′-TTTCGAGCTGGTTGGG-3′) and mdrl2 (5′-CACTAACGCGTGTGATACTTT-3′, and primers for lde were modified from those of Godreuil et al. (3) and now were lde1 (5′-ATCCTCATATAACTCAAGCG-3′) and lde2 (5′-CAATGGCTTTCGCACAA-3′). The PCR mix consisted of 1× RT-PCR buffer (Roche Diagnostics), 2.5 mM of Mn acetate, 4 mM MgCl₂, 300 μM of each deoxynucleoside triphosphate, 0.5 μM of each primer, 1 μl of 10,000-fold diluted SYBR Green I (Roche Diagnostics, Laval, PQ, Canada), 40 U of RNaseOUT (Invitrogen, Groningen, The Netherlands), 5 U Tth DNA polymerase (Roche Diagnostics), and 2 μl of sample in a final volume of 20 μl. Amplification was performed in a LightCycler (Roche Diagnostics). RT-PCR started with a reverse transcription step of 10 min at 55°C and a denaturation step of 1 min at 95°C, 40 cycles of 5 s of denaturation at 95°C and 10 s of annealing at 58°C, followed by a single fluorescence measurement and elongation for 25 s at 72°C. As a final step, melting curve analysis was performed between 65°C and 95°C. During amplification, the fluorescence was measured in channel F1/1. The quantification data, in terms of crossing point (Cp) values (the Cp value is expressed as a fractional cycle number and is the intersection of the log-linear fluorescence curve with a threshold crossing line), were determined using the second derivative method of the LightCycler software, version 5.3 (Roche Diagnostics). Relative expression of the efflux pump genes was determined by the following equation (13): ratio = (E_target^{ΔCp for target})/(E_reference^{ΔCp for reference}), with ΔCp for the target or reference determined from the difference of the control and sample values and where E described the amplification efficiency and was calculated from standard curves with the equation E = 10^(−1/slope) (13). The target was either mdrl or lde, and the reference was 23S rRNA. When performing an experiment, overnight cultures of L. monocytogenes were split in two. One half was diluted twofold with water (control), and one half was diluted twofold with BC to a final concentration of 2.5 mg liter⁻¹ sample. The cells were treated in this way for 30 min at 30°C. After treatment, RNA was extracted and real-time RT-PCR was performed. All experiments were performed in triplicate.

Statistical analysis.

Statistical analysis involving comparisons of sample means was performed using a one-way analysis of variance at a significance level of P = 0.05.

RESULTS

Strain characteristics.

Two main groups of strains were selected based on previous studies (14). The first group consisted of strains sensitive to BC (MIC ≤ 1 mg liter⁻¹). Those strains were adapted to BC over a period of 2 to 3 weeks. The MICs for BC of all adapted strains did not change after further multiple subculturing and storage at −80°C in the absence of BC for more than 1 year (data not shown). The four isolates adapted to BC were screened for resistance against disinfectants and antibiotics. The obtained data were compared with previously reported MICs for the strains prior to adaptation (Table 1, group S). Adaptation to BC led to a two- to fourfold increase of MICs to the antibiotics Gen and Kan and also to the disinfectants Clinicide and MCL, both consisting of QACs like BC. Resistance to chloramphenicol, ampicillin sodium salt, and tetracycline hydrochloride did not change (data not shown). Strain LJH 381A also exhibited an increased resistance to AA (fourfold increase in MIC) and H₂O₂ (twofold increase in MIC).

The second group consisted of strains which are naturally resistant to BC (MIC > 4 mg liter⁻¹). All of them contained two plasmids in comparison with one or two in sensitive strains of L. monocytogenes (14). To determine whether the observed BC resistance was plasmid mediated, plasmid curing was performed on the naturally resistant strains and MICs were measured again. After several subcultures, plasmids were cured in the three strains tested. Strains H7962C and C717C lost both plasmids, whereas H7764C lost only the large plasmid. After treatment, MICs of all selected L. monocytogenes strains for the same disinfectants and antibiotics were screened (Table 1, group R). Following plasmid curing, the strains (H7764C, H7962C, and C717C) of L. monocytogenes became four times more sensitive to BC and they became increasingly sensitive to all QACs (four to eight times more sensitive than parent isolates). There were also a twofold decrease in the MIC for AA and a fourfold decrease in the MIC for H₂O₂. The strain C716C retained all plasmids after treatment; however, its susceptibility to disinfectants and Kan increased in the same way as for strain C717C (Table 1, group R).

The presence of mdrL in the test strains was confirmed by PCR and sequencing (Fig. 1). Multiple sequence alignment was performed by using CLUSTAL X (1.81), and high similarity (96.4%) to the known mdrL gene sequence was observed.

FIG. 1.

Nucleotide sequence of PCR probe of mdrL gene (probe) and sequence of an internal 485-bp fragment of the mdrL gene (GenBank/EMBL/DDBJ AJ012115). The nucleotides which differ between the sequences are presented in boldface.

Efflux pump inhibition with reserpine.

Since previous studies indicated that the adapted strains were resistant to EtBr and tolerance to EtBr is commonly associated with efflux pumps, the activity of efflux pumps in L. monocytogenes was investigated using reserpine, a well-established inhibitor of efflux pumps among gram-positive microorganisms and nonfermenting gram-negative microorganisms. The MICs of BC and EtBr decreased two- to fourfold in all adapted and originally sensitive strains in the presence of reserpine (Table 2). The presence of reserpine did not affect the MIC of BC for the resistant strains (H7962, H7764, C716, and C717), but the MIC of EtBr decreased twofold in these strains. However, after plasmid curing, addition of reserpine further lowered the MIC to BC by two- to fourfold.

TABLE 2.

MICs of BC and EtBr for L. monocytogenes strains in the presence and absence of reserpine

Sensitivity group	Culture ID	MIC (mg/liter)a
		BC	BC + reserpine	EtBr	EtBr + reserpine
Sb	C511	0.78	0.3	25	12.5
	C511A	5	2.5	200	100
	C719	0.78	0.3	25	6.25
	C719A	5	2.5	200	50
	LJH381	0.78	0.3	6.25	3.25
	LJH381A	5	2.5	200	100
	LJH389	1.25	0.3	25	6.25
	LJH389A	5	1.25	100	50
Rc	C716	5	5	25	12.5
	C716C	1.25	0.6	25	12.5
	C717	5	5	25	12.5
	C717C	1.25	0.6	25	12.5
	H7764	6.25	6.25	25	12.5
	H7764C	1.25	0.6	25	12.5
	H7962	6.25	6.25	25	12.5
	H7962C	1.25	0.3	25	12.5

^aMIC data are from at least three independent measurements. Reserpine is an inhibitor of efflux pumps.

^bGroup S represents sensitive strains. Each strain has been adapted to benzalkonium chloride and now carries the original strain designation followed by A.

^cGroup R represents naturally resistant strains. Each strain has been treated with novobiocin to cure plasmids and now carries the original strain designation followed by C.

Gene expression profiling.

The efflux pump activities of four strains were further studied using gene expression profiling. Gene expression is described as the increase or decrease in expression of a target gene of a treatment compared to a control, and this is normalized against the expression of a stable reference (or housekeeping) gene in treatment and control groups. Thus, gene expression for two efflux pump genes was compared between split samples, with one sample being subjected to a 30-min treatment with BC and the other acting as an untreated control. 23 rRNA gene was used as a reference gene in the expression studies. The peak of gene expression was determined by preliminary estimation of the gene expression after 5, 15, 30, 45, and 60 min. The highest expression levels were observed at 30 min (data not shown). Analysis of variance on the obtained data after treatment for 30 min showed that for both sensitive strains, mdrL expression was significantly higher after adaptation to BC (P < 0.05) (Table 3, group S). Results for lde showed no significant difference for both strain LJH 381 (P = 0.91) and strain C719 (P = 0.11). For the naturally resistant strains, two opposite patterns were observed (Table 3, group R). Strain H7764 showed low expression levels for both the mdrL and lde genes; however, after plasmid curing, both genes exhibited a significant increase (20- to 30-fold) in gene expression levels. In contrast, strain H7962 showed intermediate expression of both genes, which significantly dropped in the case of mdrL after plasmid curing.

TABLE 3.

Relative gene expression data for two efflux pumps in strains of L. monocytogenes

Sensitivity group	Culture ID	Relative gene expression after 30-min contact with BCa
		mdrl	lde
Sb	C719	0.81 ± 0.46 (A)	1.48 ± 1.11
	C719A	3.55 ± 0.34 (A)	3.93 ± 1.78
	LJH381	0.94 ± 1.06 (B)	1.31 ± 1.85
	LJH381A	19.48 ± 9.44 (B)	1.46 ± 1.33
Rc	H7764	0.39 ± 0.29 (C)	0.38 ± 0.31 (D)
	H7764C	11.84 ± 2.01 (C, E)	8.57 ± 3.20 (D, F)
	H7764CA	0.53 ± 0.32 (E)	1.33 ± 0.33 (F)
	H7962	3.08 ± 1.39 (G)	2.94 ± 1.45
	H7962C	0.60 ± 0.23 (G, H)	1.32 ± 1.73
	H7962CA	3.41 ± 1.12 (H)	3.66 ± 1.30

^aFinal concentration of benzalkonium chloride was 2.5 mg liter⁻¹. Relative expression was calculated according to the equation provided in Materials and Methods. Results followed by different capital letters (in parentheses) are significantly different as determined by an analysis of variance (P > 0.05).

^bGroup S represents sensitive strains. Each strain has been adapted to benzalkonium chloride and now carries the original strain designation followed by A.

^cGroup R represents naturally resistant strains. Each strain has been treated with novobiocin to cure plasmids and now carries the original strain designation followed by C. The cured strains were consequently readapted to benzalkonium chloride and now carry the original strain designation followed by CA.

As both resistant strains became more sensitive to BC after curing the plasmids, they were subjected to adaptation against BC. The obtained resistance to BC after plasmid curing and subsequent adaptation reached the levels of the original resistance (data not shown). To investigate whether this adaptation was caused by higher expression of the efflux pumps, their expression was studied (Table 3, group R). In both cases, mdrL and lde returned to levels of expression observed prior to plasmid curing.

DISCUSSION

The aim of this study was to investigate the role of efflux pumps in the natural resistance and adaptation of L. monocytogenes to BC. Previous studies had already indicated that efflux pumps were involved in resistance or adaptation to certain antimicrobials. With the description of the two known efflux pump genes in L. monocytogenes, an extrusion mechanism was indicated, with the macrolide antibiotics, cefotaxime, heavy metals, and EtBr being extruded by Mdrl (10) and fluoroquinolone resistance and, in part, acridine orange and ethidium bromide resistance being due to Lde (3).

The selected strains for the study were divided into two groups: group S, strains which were naturally sensitive to BC, and group R, strains which were naturally resistant to BC. Within group S, the four selected L. monocytogenes strains could all be adapted to BC (Table 1). As observed, after adaptation to the QAC BC, adaptation to other QACs with a similar mode of action, such as Clinicide and MCL, also occurred. Cross-adaptation to antimicrobials with different modes of action, such as Gen, Kan, and the intercalating dye EtBr, was also observed. The phenomenon of cross-resistance between antibiotics and disinfectants has been already demonstrated in gram-negative (6) and in some gram-positive (19) bacteria. This supports the view that cross-adaptation can be related to nonspecific multidrug efflux pumps (12).

The role of efflux pumps in adaptation to BC was further investigated using reserpine, a potent inhibitor of efflux pumps. Reserpine is considered to inhibit members of the resistance nodulation division family, major facilitator family, and the ATP binding cassette (9). Two efflux pumps described in L. monocytogenes (MdrL and Lde) are members of the major facilitator family (family 2) (3, 10). After addition of reserpine, the MIC for the different strains decreased; indicating that efflux pumps play a role in the adaptation of L. monocytogenes to BC. To differentiate which one is responsible for adaptation, study of expression of the mdrl and lde genes in the selected naturally sensitive and adapted strains was carried out. It became clear that mdrL expression significantly increased after adaptation, but no significant increase was observed for lde expression. This was in agreement with previous studies indicating efflux pumps playing a role in adaptation of L. monocytogenes to BC (20). To our knowledge no previous studies on expression of these two genes have been performed. The results indicated that the efflux pump Mdrl is at least partly responsible for the adaptation to BC. It was also shown that low levels of Mdrl were already present in the naturally sensitive strains. This was in agreement with the reserpine studies (Table 2), which demonstrated that the MICs to BC also decreased for the sensitive strains after addition of reserpine, indicating a constitutive low level of efflux pump activity in these strains also.

Group R consisted of naturally resistant isolates. This group already had been shown to have high levels of resistance to selected antimicrobials (14) (Table 1). Addition of reserpine to these strains resulted in no drop in the MIC of BC, indicating that efflux pumps played no role in the natural resistance to this compound. Further results showed low and intermediate levels of expression of the genes encoding the efflux pumps for two selected resistant strains, H7764 and H7962, respectively. To gain further insight into this innate resistance, plasmids from the strains were cured. Antimicrobial resistance has been shown to be plasmid mediated in several instances (5, 16). After curing of plasmids, all four strains became more susceptible to QACs (Table 1, group R). However, a noticeable difference was observed between strains H7764C and H7962C. Whereas strain H7764C showed a decrease in MICs for Gen and Kan following curing of plasmids, strain H7962C showed an increase in MICs for these compounds. After addition of reserpine, the MICs decreased for BC, an indication that efflux pumps could play a role in resistance. The same treatment with novobiocin was also applied to the strains of L. monocytogenes belonging to group S and adapted to BC. Some strains lost plasmids and some did not, but this did not affect their MICs to any of the antimicrobial agents (data not shown). So, strains H7764C and H7962C were chosen for further study by gene expression profiling. The strain H7764C showed significant increases in expression of both mdrL and lde, confirming that efflux pumps play a role in resistance. However, strain H7962C showed a significant drop in mdrL expression. Further insight into the behavior of these two strains was obtained by adapting the cured strains to their original levels of BC resistance. Again, contrasting patterns were observed. Gene expression in both strains returned to levels similar to those obtained before plasmid curing. At the same time, there was no active efflux observed in these samples, supporting our previous statement that their tolerance to BC is due to some other mechanisms. These data suggested the existence of more than one mechanism of resistance to BC, making it difficult to predict how a bacterium may respond to sublethal exposure to an antimicrobial agent, even if data on related strains exist. The existence of more than one mechanism of resistance to BC in L. monocytogenes strains has been proposed by other researchers (1, 16, 18). Resistance to BC in naturally resistant strains of L. monocytogenes may be due to modifications in the cell membrane or due to the presence of one or more unknown types of efflux pumps. It is still unclear if resistance is linked with the presence of plasmids in these strains.

In conclusion, the results of this study suggest that adaptation of naturally sensitive strains of L. monocytogenes to quaternary ammonium compounds is at least partly caused by overexpression of the efflux pump Mdrl. Although previous studies indicated the role of efflux pumps in adaptation of L. monocytogenes to BC (20), no evidence showing the enhanced expression of particular genes had been presented. Additionally, results indicate that for naturally resistant strains both efflux pumps (Mdrl and Lde) may play a role in resistance, but other mechanisms may also be operating. Finally, this study has shown that not only the presence of the respective efflux pump genes but also their expression under different conditions should be investigated to develop strategies for efficient control of L. monocytogenes. This may in the future lead to methods to prevent or reduce such adaptation.