AcrAB-TolC Directs Efflux-Mediated Multidrug Resistance in Salmonella enterica Serovar Typhimurium DT104

Abstract

Multidrug-resistant Salmonella enterica serovar Typhimurium definitive phage type 104 (DT104) strains harbor a genomic island, called Salmonella genomic island 1 (SGI1), which contains an antibiotic resistance gene cluster conferring resistance to ampicillin, chloramphenicol, florfenicol, streptomycin, sulfonamides, and tetracyclines. They may be additionally resistant to quinolones. Among the antibiotic resistance genes there are two, i.e., floR and tet(G), which code for efflux pumps of the major facilitator superfamily with 12 transmembrane segments that confer resistance to chloramphenicol-florfenicol and the tetracyclines, respectively. In the present study we determined, by constructing acrB and tolC mutants, the role of the AcrAB-TolC multidrug efflux system in the multidrug resistance of several DT104 strains displaying additional quinolone resistance or not displaying quinolone resistance. This study shows that the quinolone resistance and the decreased fluoroquinolone susceptibilities of the strains are highly dependent on the AcrAB-TolC efflux system and that single mutations in the quinolone resistance-determining region of gyrA are of little relevance in mediating this resistance. Overproduction of the AcrAB efflux pump, as determined by Western blotting with an anti-AcrA polyclonal antibody, appeared to be the major mechanism of resistance to quinolones. Moreover, chloramphenicol-florfenicol and tetracycline resistance also appeared to be highly dependent on the presence of AcrAB-TolC, since the introduction of mutations in the respective acrB and tolC genes resulted in a susceptible or intermediate resistance phenotype, according to clinical MIC breakpoints, despite the presence of the FloR and Tet(G) efflux pumps. Resistance to other antibiotics, ampicillin, streptomycin, and sulfonamides, was not affected in the acrB and tolC mutants of DT104 strains harboring SGI1. Therefore, AcrAB-TolC appears to direct efflux-mediated resistance to quinolones, chloramphenicol-florfenicol, and tetracyclines in multidrug-resistant S. enterica serovar Typhimurium DT104 strains.

Multidrug-resistant Salmonella enterica serovar Typhimurium definitive phage type 104 (DT104) emerged during the 1980s as a global health problem because of its involvement in diseases in animals and humans (6, 13, 15, 23, 24, 40, 42, 43). Multidrug-resistant strains of this phage type were first identified from exotic birds in the United Kingdom in the early 1980s and in cattle and humans in the late 1980s but have since become common in other animal species such as poultry, pigs, and sheep. The serovar Typhimurium DT104 epidemic clone has now spread worldwide, with several outbreaks since 1996 in the United States and Canada (6, 15, 23, 40).

Multidrug-resistant S. enterica serovar Typhimurium DT104 strains are commonly resistant to ampicillin (AMP), chloramphenicol (CHL)-florfenicol (FLF), streptomycin (STR)-spectinomycin (SPT), sulfonamides (SULs), and tetracyclines (TETs). The antibiotic resistance genes are clustered in part of a 43-kb genomic island, called Salmonella genomic island I (SGI1), between the thdF and int2 genes of the chromosome compared to the genome sequence of S. enterica serovar Typhimurium strain LT2 (8, 9). The antibiotic resistance gene cluster represents approximately one-third of SGI1 and is located at the 3′ end of the structure (8, 9). All resistance genes are clustered and are bracketed by two integron structures (4, 8-10, 13). The first integron carries the aadA2 gene, which confers resistance to STR and SPT, and a truncated sul1 (sul1delta) gene. The second contains the β-lactamase gene pse-1, which confers resistance to AMP, and a complete sul1 gene, which confers resistance to the SULs. Flanked by these two integron structures are the floR gene (4), also called floSt (7) or the cmlA-like gene (10), which confers cross-resistance to CHL-FLF, and the TET resistance gene tet(G). The floR and tet(G) genes code for efflux pumps of the major facilitator superfamily (MFS) with 12 transmembrane segments.

The emergence of decreased susceptibility to fluoroquinolones in multidrug-resistant S. enterica serovar Typhimurium DT104 strains is causing particular concern (11, 35, 43, 45). The emergence and spread of such serovar Typhimurium DT104 strains followed the licensing of enrofloxacin (ENR) for veterinary use in 1993 (43). This fluoroquinolone antibiotic has subsequently been used to treat cattle and poultry, resulting in the selection of quinolone-resistant strains (43). It has been reported that the mechanisms of decreased susceptibility to fluoroquinolones in serovar Typhimurium DT104 isolates involve point mutations in the quinolone resistance-determining region (QRDR) of the target gene, gyrA (12, 13, 43). Another important mechanism may be active efflux due to overproduction of the AcrAB efflux pump (22). We have previously shown that the AcrB multidrug transporter plays a major role in the high-level fluoroquinolone resistance of S. enterica serovar Typhimurium DT204 strains carrying multiple target mutations in gyrA, gyrB, and parC (5).

In the present study we analyzed the role of the AcrAB-TolC efflux system in the multidrug resistance of S. enterica serovar Typhimurium DT104 by constructing mutants with mutations in the AcrB transporter and the TolC outer membrane component of the efflux system. We investigated in particular the role of the efflux system in quinolone, CHL-FLF, and TET resistance. For resistance to CHL-FLF and TETs, it was our hypothesis that the AcrAB-TolC efflux system might interplay with the secondary specific transporters FloR and Tet(G), respectively.

MATERIALS AND METHODS

Bacterial strains.

Multidrug-resistant S. enterica serovar Typhimurium DT104 strains were isolated from cattle in Belgium (strain 543SA98) and France (strains BN10055, BN9945, and BN9181) (Table 1). Susceptible serovar Typhimurium DT104 control strain S/921495 was isolated from cattle in Scotland. Quinolone-susceptible serovar Typhimurium strain BN18 was isolated from a pigeon in France; and quinolone-resistant clones of this strain selected in vitro, strains BN18/21, BN18/41, and BN18/71, were used as controls for Western blot analysis for AcrA production. The MICs of ENR (Bayer AG, Leverkusen, Germany) for BN18 and the three clones were 0.125, 1, 4, and 16 μg/ml, respectively, and correlated well with the level of AcrA production (5, 22).

TABLE 1.

S. enterica serovar Typhimurium DT104 strains studied

Strain	Geographic origina	SG11	Quinolone MIC (μg/ml)b					MIC of other drugs (μg/ml)			Substitution in gyrA QRDR	WBc AcrA
			NAL	FLU	ENR	MAR	CIP	CHL	FLF	TET
BN10055	F	+	>1,024 (16)	64 (2)	4 (0.125)	2 (0.25)	2 (0.125)	256 (64)	128 (4)	128 (64)	Ser83Tyr	+++
BN10055acrB::kan		+	64	2	0.25	0.25	0.060	16	4	16	Ser83Tyr	NDe
BN10055ΔtolC::kan		+	32	1	0.060	0.125	0.060	16	4	8	Ser83Tyr	ND
543SA98	B	+	1,024 (16)	32 (1)	2 (0.125)	2 (0.25)	1 (0.125)	256 (64)	128 (16)	128 (64)	Ser83Phe	+++
543SA98acrB::kan		+	32	1	0.25	0.25	0.060	16	8	16	Ser83Phe	ND
543SA98ΔtolC::kan		+	16	1	0.125	0.125	0.060	16	8	8	Ser83Phe	ND
BN9945	F	+	512 (16)	32 (1)	1 (0.125)	1 (0.25)	0.5 (0.125)	256 (64)	128 (16)	128 (64)	—f	++
BN9945acrB::kan		+	32	1	0.25	0.125	0.060	16	8	8	—	ND
BN9945ΔtolC::kan		+	16	1	0.060	0.125	0.060	16	8	8	—	ND
BN9181	F	+	4 (0.5)	0.5 (0.03)	0.125 (0.008)	0.060 (0.030)	0.030 (0.015)	256 (64)	32 (2)	64 (32)	—	+
BN9181acrB::kan		+	0.5	0.060	0.008	0.030	0.004	16	2	16	—	ND
BN9181ΔtolC::kan		+	0.5	0.060	0.008	0.008	0.002	16	2	8	—	ND
Control strains
S/921495	S	−	4 (0.5)	0.5 (0.03)	0.125 (0.008)	0.060 (0.008)	0.030 (0.008)	8 (0.5)	8 (1)	2 (0.5)	—	+
S/921495acrB::kan		−	0.5	0.060	0.008	0.008	0.004	0.5	1	0.5	—	ND
S/921495ΔtolC::kan		−	0.5	0.060	0.008	0.008	0.004	1	1	0.5	—	ND
ST-1d	C	+	4 (0.5)	0.5 (0.03)	0.125 (0.008)	0.060 (0.008)	0.030 (0.004)	256 (64)	64 (4)	128 (32)	—	+
ST-1acrB::kan		+	0.5	0.060	0.008	0.008	0.004	16	4	16	—	ND
ST1ΔtolC::kan		+	0.5	0.060	0.008	0.008	0.004	16	4	8	—	ND

^aF, France; B, Belgium; S, Scotland; C, Canada.

^bValues in parentheses are MICs in the presence of the efflux pump inhibitor PAβN at 80 μg/ml.

^cWB, Western blotting with an anti-AcrA polyclonal antibody.

^dST-1 corresponds to strain S/921495 complemented with SGI1.

^eND, not determined.

^f—, no substitution.

MIC determinations.

MICs were determined by the standard agar doubling dilution method on Mueller-Hinton medium with inocula of 10⁴ CFU per spot. The MIC was defined as the lowest concentration of the drug that completely inhibited visible growth after incubation for 18 h at 37°C. The following antibiotics were used: nalidixic acid (NAL; Sigma, Steinheim, Germany), flumequine (FLU; Sigma), ENR (Bayer AG), ciprofloxacin (CIP; Bayer AG), marbofloxacin (MAR; Vétoquinol, Lure, France), CHL (Boehringer, Mannheim, Germany), FLF (Schering-Plough Animal Health, Kenilworth, N.J.), and TET (Boehringer). The following MIC breakpoints (c and C), defined by the Comité de l'Antibiogramme de la Société Française de Microbiologie (34) or the manufacturers, were used to classify strains as susceptible (MIC ≤ c), intermediate (c < MIC ≤ C), or resistant (MIC > C): NAL, 8 and 16 μg/ml; FLU, 4 and 8 μg/ml; ENR, CIP, and MAR, 1 and 2 μg/ml; CHL and FLF, 8 and 16 μg/ml; and TET, 4 and 8 μg/ml. The MICs of these antibiotics were also determined with 80 μg of the efflux pump inhibitor Phe-Arg-β-naphthylamide (PAβN; Sigma) per ml (30, 31).

PCR amplification and sequencing of QRDRs of gyrA, gyrB, parC, and parE genes.

The sequences of the primers used in the PCR amplifications of the QRDRs of gyrA, gyrB, parC, and parE are given in Table 2. Genomic DNA was extracted from overnight cultures in Luria-Bertani (LB) medium at 37°C by using the QIAamp DNA Mini extraction kit (Qiagen S.A., Courtaboeuf, France). PCR was performed with extracted genomic DNA in volumes of 100 μl containing 200 ng of template DNA, 100 pmol of each primer, 200 μM deoxynucleoside triphosphates, 1.5 mM MgCl₂, and 2 U of Taq polymerase (Promega, Madison, Wis.). After an initial denaturation step of 3 min at 94°C, amplification was performed over 30 cycles, with each cycle consisting of 1 min at 94°C, 1 min at the hybridization temperature (55°C for gyrA, 58°C for gyrB, and 52°C for parC and parE), and 1 min at 72°C, with a final extension step of 10 min at 72°C. Primers and free nucleotides were removed with a Qiaquick spin PCR purification kit (Qiagen S.A.). Nucleotide sequencing was performed with a Génome Express (Grenoble, France) sequencer. Sequence analysis was done by using the following programs: BLAST (http://www.ncbi.nlm.nih.gov/BLAST/), FASTA (http://www2.ebi.ac.uk/fasta3/), and CLUSTALW (http://www2.ebi.ac.uk/clustalw/).

TABLE 2.

Primers used for PCRs

Primer	Gene	Nucleotide sequence (5′-3′)
STGYRA1	gyrA	TGTCCGAGATGGCCTGAAGC
STGYRA12	gyrA	CGTTGATGACTTCCGTCAG
STGYRB5	gyrB	AACGGTCTGCTCATCAGAAAGG
STGYRB7	gyrB	GAAATGACCCGCCGTAAAGG
STPARC1	parC	ATGAGCGATATGGCAGAGCG
STPARC2	parC	TGACCGAGTTCGCTTAACAG
STPARE1	parE	GACCGAGCTGTTCCTTGTGG
STPARE2	parE	GCGTAACTGCATCGGGTTCA
Pr1	acrB	ACGGTATCGATAACCTGATG
Pr2	acrB	TTCTTCGCCGACAACCCAAT
P1TolC	tolC-kan	TCCCCATCCTTATCGGCCTGAGCCTGTCGGGGTTCAGCACACTAAGCCAGGTGTAGGCTGGAGCTGCTTC
P2TolC	tolC-kan	TCAATGCCGGAATGGATTGCCGTTATTGCTGTTGGCGCGAGCGGCGGTCGCATATGAATATCCTCCTTAG
k2	kan	CGGTGCCCTGAATGAACTGC
kt	kan	CGGCCACAGTCGATGAATCC
TolC3	tolC	AAGCACGCCTGAGCAACCCG
TolC4	tolC	AATGCCGGAATGGATTGCCG
TolC5	tolC	CTCTCTAGACGAACATCTCCAGCAGCCAC
TolC6	tolC	CTCGAGCTCGCACTGACCTTCGCCAAACT

Construction of acrB mutants.

S. enterica serovar Typhimurium acrB mutant strain LXA3, constructed by disrupting the acrB gene with the kanamycin resistance aminoglycoside 3′-phosphotransferase gene from plasmid pUC4K (Amersham Pharmacia Biotech, Orsay, France), was characterized previously (27). This strain was used to introduce the acrB::kan mutation in the strains used in this study by transduction with phage P22 (5, 41). The resulting acrB::kan mutants were selected on LB agar plates containing 50 μg of kanamycin per ml, and insertion of the kanamycin resistance gene into acrB was confirmed by PCR with primers Pr1 and Pr2 (Table 2).

Construction of tolC mutants.

Deletion of the tolC gene was performed by the method described by Datsenko and Wanner (14). Briefly, the kanamycin resistance gene kan flanked by FLP recognition target sites was amplified by a standard PCR with the template plasmid pKD4 and hybrid primers. These primers, P1TolC and P2TolC (Table 2), were synthesized with 20 nucleotides (nt) of priming sites 1 and 2 of pKD4 and with 50 nt of the 5′ and 3′ ends of the tolC gene. The 1.6-kb PCR fragment was purified and electroporated into antibiotic-susceptible strain S. enterica serovar Typhimurium DT104 S/921495, into which Red recombinase expression plasmid pKD46 was introduced. Transformants were selected at 37°C on LB agar medium containing kanamycin at 50 μg/ml. Homologous recombination between the genomic DNA and the PCR product resulted in the deletion of the tolC sequence from nt 70 to 1426 (1,356-bp deletion) and its replacement with the kan gene. This was confirmed by three different PCRs. Deletion of tolC in the transformants was first shown by a negative result by PCR with primers TolC3 and TolC4 (Table 2), whose sequences correspond to internal sequences of the deleted tolC gene. To confirm the deletion, a second PCR was carried out with primers TolC5 and TolC6 (Table 2), whose sequences correspond to sequences flanking the tolC deletion and which resulted in a 2,400-bp fragment for the parental strains and a 2,644-bp fragment when tolC was deleted and replaced by the kan gene flanked by FLP recognition target sites. The third control PCR, with primers k2 and kt (6) (Table 2), was used to detect the 471-bp kan fragment.

The tolC gene deletion was further transducted into other strains by using phage P22, as described above.

SGI1 isogenic strain construction.

Analysis of the S. enterica serovar Typhimurium LT2 genome revealed two AscI sites which flanked the thdF gene (also called the trmE gene), which has been shown to be the insertion site of SGI1 (8). No AscI sites were identified in the complete SGI1 sequence; therefore, the complete SGI1 sequence could be isolated on a 51-kb AscI fragment. Serovar Typhimurium DT104 strain 1948SA96 genomic DNA was cut with AscI and ligated into the MluI site of the suicide vector pMAKSACB (19). This ligation mixture was electrotransformed into Escherichia coli DH10B (Invitrogen), and colonies were selected on AMP-CHL at the permissive temperature. The plasmids from several colonies were isolated, and the presence of SGI1 was confirmed by restriction mapping with EcoRI and BamHI, targeting the ends of SGI1 by PCR, and sequencing across the cloning site. The plasmid, labeled pST-SGI1, was used to electrotransform antimicrobial-susceptible strain S. enterica serovar Typhimurium DT104 S/921495 with selection on AMP-CHL at the permissive temperature. No transformants could be obtained after several attempts, even though the pMAKSACB vector alone could be successfully maintained in serovar Typhimurium DT104 S/921495 at the permissive temperature. However, when a portion of the ligation mixture described above was electrotransformed into serovar Typhimurium DT104 S/921495, two colonies were isolated on AMP-CHL. PCR confirmed the presence of SGI1 but the absence of the sacB gene of the vector. Southern hybridization and further PCR analysis showed that the 51-kb AscI fragment was likely circularized and inserted into the genome by a single crossover event via the thdF end in one strain (strain ST-1) and the yidY end in the other strain (strain ST-2).

AcrA expression analysis by Western blotting.

acrA is cotranscribed with acrB, and AcrA expression is indicative of expression of the AcrAB efflux pump. Overnight cultures grown in LB medium at 37°C were harvested by centrifugation and resuspended at an A₆₀₀ of 10.0. Cells were diluted 1/2 in the sample buffer of Laemmli (28) and were heated for 10 min at 100°C. Whole-cell proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and were transferred at 0.8 mA/cm² to a nitrocellulose membrane. The membrane was washed three times with Tris-buffered saline (TBS; 0.15% NaCl, 10 mM Tris-HCl [pH 7.5]), saturated for 30 min at room temperature with TBS containing 1% skim milk, and incubated overnight at room temperature with an anti-AcrA polyclonal antibody (33) diluted 1/2,000 in TBS containing 0.05% Tween 20 (TBS-T) and 0.33% skim milk. After three washes in TBS-T, the membrane was incubated for 1 h with peroxidase conjugated to protein A (Sigma) diluted 1/1,000 in TBS-T. Finally, after three washes in TBS-T, the blot was developed by incubation at room temperature in a solution of TBS containing 0.06% 4-chloro-1-naphthol and 5 mM H₂O₂. The reaction was stopped by washes in distilled water.

RESULTS

Role of AcrAB-TolC in resistance to quinolones.

Representative strains of S. enterica serovar Typhimurium DT104 (strains BN10055, 543SA98, and BN9945) showing increasing levels of resistance to NAL and decreased susceptibilities to fluoroquinolones were studied for their mechanisms of resistance to these antibiotics (Table 1). Strain BN9945 exhibited high-level resistance to NAL (MIC, 512 μg/ml) and decreased susceptibilities to the fluoroquinolones ENR, MAR, and CIP (MICs, 0.5 to 1 μg/ml) without carrying any mutation in the QRDRs of target gene gyrA, gyrB, parC, or parE. This strongly suggests other quinolone resistance mechanisms such as active efflux. The levels of resistance to NAL, FLU, ENR, MAR, and CIP were slightly increased (two- to fourfold) in two other S. enterica serovar Typhimurium DT104 strains (strains 543SA98 and BN10055) carrying a single mutation at codon 83 of gyrA, leading to the amino acid changes Ser83Phe and Ser83Tyr, respectively. According to the clinical breakpoints, the latter strains were resistant or intermediate resistant to the fluoroquinolones ENR, MAR, and CIP (Table 1).

Evidence for active efflux was provided by Western blot analysis with an anti-AcrA polyclonal antibody, which showed overproduction of the AcrA periplasmic protein in the quinolone-resistant strains relative to the level of production by the susceptible control strains (Fig. 1). The level of production of AcrA correlated well with the level of resistance to NAL and fluoroquinolones; i.e., strains BN10055 and 543SA98 showed a higher level of production of AcrA than strain BN9945. These results support an active efflux mechanism due to overproduction of the AcrAB-TolC efflux system, which appears to be a primary mechanism of resistance in quinolone-resistant field strains of S. enterica serovar Typhimurium DT104.

FIG. 1.

Western blot analysis of AcrA protein production in S. enterica serovar Typhimurium control strains BN18 (lane 1), BN18/21 (lane 2), BN18/41 (lane 3), and BN18/71 (lane 4) and serovar Typhimurium DT104 strains S/921495 (lane 5), ST-1 (lane 6), BN9181 (lane 7), BN9945 (lane 8), 543SA98 (lane 9), and BN10055 (lane 10).

The role of AcrAB-TolC was further confirmed by inactivating the acrB and tolC genes, located at different chromosomal loci, in the strains studied. This resulted in 32- to 64-fold decreases in the levels of resistance to NAL and the fluoroquinolones (Table 1). The decreases in the levels of resistance were approximately the same for the acrB and tolC mutants, suggesting that AcrB is the major quinolone transporter interacting with TolC. Interestingly, the amount of decrease in resistance was the same whether the strains carried a mutation in gyrA or not. Nevertheless, inactivation of acrB or tolC did not result in a complete reversal of susceptibility to the level observed in the NAL-susceptible control strains (strains BN9181 and S/921495), suggesting that other not yet identified mechanisms may participate in quinolone resistance (Table 1).

Active efflux was also finally confirmed by use of the efflux pump inhibitor PAβN, which was effective in decreasing the MICs of NAL and fluoroquinolones for the S. enterica serovar Typhimurium DT104 field strains studied to the same extent obtained by inactivating their respective acrB and tolC genes (Table 1).

Role of AcrAB-TolC in resistance to other antibiotics.

Multidrug-resistant S. enterica serovar Typhimurium DT104 strains harbor a chromosomal antibiotic resistance gene cluster within SGI1 conferring resistance to AMP, CHL-FLF, STR-SPT, SULs, and TETs. Since some of these antibiotics, such as AMP, CHL, and TETs, are well-known substrates of the AcrB multidrug transporter, it was interesting to determine whether inactivation of acrB or tolC would affect the levels of resistance to these antibiotics in these strains. Antibiograms performed for the S. enterica serovar Typhimurium DT104 parental and mutant strains showed that only resistance to CHL-FLF and TET was affected in both the acrB and the tolC mutant strains (data not shown). As shown in Table 1, inactivation of acrB or tolC resulted in a 16- to 32-fold reduction in the level of resistance to CHL-FLF and in a 4- to 16-fold reduction in the level of resistance to TET, depending on the strain. As was the case for the decreases in the levels of resistance to NAL and fluoroquinolones, the decreases in the levels of resistance to CHL-FLF and TET were approximately the same for the acrB and tolC mutants. Interestingly, inactivation of acrB and tolC in S. enterica serovar Typhimurium DT104 strains harboring SGI1 resulted in an intermediate level of resistance to CHL (MIC, 16 μg/ml) and susceptibility to FLF (MICs, ≤8 μg/ml), according to the clinical breakpoints. In the case of the parental strains, the levels of resistance to FLF were two- to fourfold higher (MIC, 128 μg/ml) for the quinolone-resistant strains than for quinolone-susceptible strains BN9181 and ST-1 harboring SGI1 (MICs, 32 and 64 μg/ml, respectively) (Table 1). If this difference is significant, it could be due to the overproduction of AcrAB, as shown by Western blotting (Fig. 1). However, it was not evident for the CHL and TET resistance levels; i.e., the MICs were the same for the quinolone-resistant and quinolone-susceptible strains. The serovar Typhimurium DT104 acrB and tolC mutant strains harboring SGI1 showed intermediate resistance to TET (MIC of 8 μg/ml) or resistance to TET close to the clinical breakpoint (MIC, 16 μg/ml). Together, these results indicate that resistance to CHL-FLF and TET in multidrug-resistant serovar Typhimurium DT104 strains harboring SGI1 is highly dependent on the presence of the AcrAB-TolC efflux system.

This was further confirmed by studying a set of isogenic S. enterica serovar Typhimurium DT104 strains, a susceptible strain (strain S/921495), a strain in which SGI1 had been inserted in vitro (strain ST-1), and ST-1 with inactivated acrB and tolC genes. Serovar Typhimurium ST-1 exhibited 32-fold (MIC, 256 μg/ml), 8-fold (MIC, 64 μg/ml), and 64-fold (MIC, 128 μg/ml) increases in the levels of resistance to CHL, FLF, and TET, respectively, compared to those for its parental strain, S/921495 (Table 1). These resistance levels are comparable to those seen for multidrug-resistant serovar Typhimurium DT104 field strains harboring SGI1. Inactivation of the acrB and tolC genes in strain ST-1 resulted in the same decrease in the levels of resistance to CHL-FLF and TET observed for the multidrug-resistant serovar Typhimurium DT104 field strains harboring SGI1, thereby confirming the major role of the AcrAB-TolC efflux system in drug resistance.

To further show the effect of the simultaneous expression of the FloR or the Tet(G) transporter with the multicomponent efflux system AcrAB-TolC, we compared the MICs of FLF, CHL, and TET for control strains S/921495 and ST-1 to those for their acrB-derived mutants. The MICs of FLF, CHL, and TET for strain S/921495 [which lacks SGI1 and the floR and tet(G) genes] were 8, 8, and 2 μg/ml, respectively, whereas in the absence of the AcrB transporter, the MICs of FLF, CHL, and TET decreased to 0.5, 1, and 0.5 μg/ml, respectively. Complementation of this strain with SGI1 (strain ST-1) resulted in increases in the FLF, CHL, and TET MICs to 256, 64, and 128 μg/ml, respectively. Inactivation of the acrB gene in the latter strain decreased the FLF, CHL, and TET MICs to 16, 4, and 16 μg/ml, respectively (Table 1). By calculating the ratio of these MICs, i.e., the MIC in the presence of the transporter considered divided by the MIC in its absence, we obtained the results shown in Table 3. For each antibiotic, the ratio in the presence of the two types of transporters [FloR and AcrB or Tet(G) and AcrB] corresponded approximately to the multiplication of the ratio obtained for each transporter alone [FloR, Tet(G), or AcrB]. This indicates that the simultaneous expression of these transporters and a multicomponent efflux pump results in levels of drug resistance higher than those provided by each of them alone. Thus, the antibiotics FLF, CHL, and TET can be directly exported by the AcrAB-TolC efflux system, as well as efficiently transported to the periplasm by FloR or Tet(G).

TABLE 3.

Effect of simultaneous expression of single-component efflux pump FloR or Tet(G) and the multicomponent efflux system AcrAB-TolC

Transporter	Ratio of MICs of antibiotics for control strains
	CHL	FLF	TET
AcrBa	16	8	4
FloR and Tet(G)b	32	8	64
FloR, Tet(G), and AcrBc	512	64	256

^aMIC for strain S/921495/MIC for strain S/921495 acrB::kan.

^bMIC for strain ST-1acrB::kan/MIC for strain S/921495acrB::kan.

^cMIC for strain ST-1/MIC for strain S/921495::kan.

The efflux pump inhibitor PAβN was not as effective in decreasing the MICs of CHL, FLF, and TET to the same extent obtained by inactivating the acrB and tolC genes for most multidrug-resistant S. enterica serovar Typhimurium DT104 field strains studied (Table 1). By use of this efflux pump inhibitor, only the FLF resistance levels were decreased to the same extent obtained by inactivating acrB or tolC in strains BN10055, BN9181, and ST-1.

DISCUSSION

The mechanisms of decreased susceptibility to fluoroquinolones in S. enterica serovar Typhimurium DT104 isolates have principally been attributed to point mutations in the QRDR of the target gene, gyrA (12, 13, 43). Thus, it was thought that the mutation in gyrA constituted the major mechanism of resistance. Amino acid changes at Ser83 (to Phe, Tyr, or Ala) or Asp87 (to Gly, Asn, or Tyr) are the most frequently observed in NAL-resistant strains (12, 13, 21, 43). A previous study (22) showed the participation of active efflux in quinolone-resistant mutants of serovar Typhimurium selected in vitro. In such mutants quinolone selection resulted in a multiple-antibiotic-resistance phenotype with associated resistance to unrelated antibiotics (22). The efflux mechanism appeared prior to acquisition of target gene mutations in these in vitro-selected quinolone-resistant Salmonella mutants (22). In the first-step mutants lacking gyrA mutations, overproduction of the AcrAB efflux pump already resulted in an eightfold increase in the levels of resistance to most antibiotics studied, i.e., NAL, FLU, ENR, MAR, CHL, and FLF; and this accounted for clinical resistance to NAL and unrelated drugs, CHL and FLF. It was concluded that the primary mechanism of resistance to quinolones was active efflux (22). It was thus of interest to reassess more precisely the mechanisms of quinolone resistance in epidemic serovar Typhimurium DT104 strains.

The results of the present study are in accordance with previous observations for in vitro-selected quinolone-resistant Salmonella mutants, i.e., that active efflux by the AcrAB-TolC efflux system is also the main mechanism of resistance to quinolones in S. Typhimurium DT104 strains. First, quinolone-resistant S. enterica serovar Typhimurium DT104 strains, such as strain BN9945, exist without carrying any mutations in the QRDRs of the target genes. According to clinical breakpoints, strain BN9945 indeed displayed high-level resistance to NAL (MIC, 512 μg/ml) and reduced susceptibilities to the fluoroquinolones ENR, MAR, and CIP. This strain overproduced AcrAB, according to Western blot analysis with an anti-AcrA polyclonal antibody. Inactivation of the acrB and tolC genes in this strain further confirmed the importance of the AcrAB-TolC efflux system in quinolone resistance. Inactivation of the acrB and tolC genes in strains BN10055 and 543SA98 carrying a single point mutation in the QRDR of gyrA resulted in the same levels of susceptibility to quinolones observed for strain BN9945, which lacked any gyrA mutation. Thus, it appears from these data that the GyrA substitutions Ser83Tyr or Ser83Phe are of little relevance in mediating quinolone resistance and that efflux by the AcrAB-TolC efflux system is the main mechanism that mediates quinolone resistance in serovar Typhimurium DT104 strains. Our results are also in accordance with those of Oethinger et al. (38), who reported that in E. coli deletion of the acrAB locus made strains, including those with double gyrA mutations, hypersusceptible to fluoroquinolones and certain other unrelated drugs. To obtain high-level fluoroquinolone resistance, as shown previously for serovar Typhimurium DT204 strains (5), both active efflux by AcrAB-TolC and multiple mutations in the target genes gyrA, gyrB, and parC are probably required.

In this study, we showed that resistance to CHL and FLF and to TET is due to efflux mechanisms both by specific transporters encoded by the floR and tet(G) genes located on SGI1 and by the multidrug efflux system AcrAB-TolC. In addition, there were no significant differences between the MICs of CHL, FLF, and TET for either the acrB or the tolC mutants, thereby suggesting that AcrB is the main multidrug transporter that interacts with TolC and that is involved in resistance to these drugs. Moreover, the resistance levels observed for acrB and tolC mutants and for a susceptible strain complemented with SGI1 indicated strong functional interactions between the MFS transporters FloR and Tet(G) and the multicomponent AcrAB-TolC efflux system. Data from the present study indicate that the simultaneous expression of the AcrAB-TolC efflux system and the FloR or Tet(G) transporter results in multiplicative increases in the levels of resistance to CHL, FLF, and TET. This observation is in accordance with what was previously observed for E. coli (29, 39), i.e., a synergistic effect resulting from the simultaneous expression of the multicomponent efflux system AcrAB-TolC, where AcrB is an RND transporter that interacts directly with TolC, and transporters of the MFS family, such as FloR and Tet(G), with the energies of both systems depending on the proton motive force (12, 13, 21, 29, 39, 43, 48). Thus, the global mechanism of CHL, FLF, and TET export would be for the MFS transporter FloR or Tet(G) to transport the antibiotics from the cytoplasmic membrane to the periplasm, which would then be recovered by the multidrug transporter AcrB to be extruded out of the cell by TolC. This point of view is consistent with the recent determination of the three-dimensional structure of AcrB, which is organized as a trimer, in which the three-dimensional structure revealed the presence of vestibules that permit direct access of the substrates from the periplasm or the outer leaflet of the cytoplasmic membrane to its central cavity, with the large periplasmic domains of AcrB being involved in substrate recognition (16, 17, 20, 32, 36, 44, 46, 47). Then, the TolC channel, also organized as a trimer (26), would be recruited to form with AcrB a direct transit structure to the extracellular medium (2, 16, 25, 36, 37).

The results of our study have some practical implications. As shown in the present study, clinical levels of resistance to CHL and FLF and to TET are highly dependent on the synergy between AcrAB-TolC and FloR and Tet(G), respectively. Thus, targeting of AcrB (5, 30-32) or TolC (1, 3, 18, 25), rather than the other transporters, to develop efflux pump inhibitors would probably be sufficient to develop combination therapies with such inhibitors and CHL, FLF, or TET as well as fluoroquinolones.