AcrAB Multidrug Efflux Pump Is Associated with Reduced Levels of Susceptibility to Tigecycline (GAR-936) in Proteus mirabilis

Abstract

Tigecycline has good broad-spectrum activity against many gram-positive and gram-negative pathogens with the notable exception of the Proteeae. A study was performed to identify the mechanism responsible for the reduced susceptibility to tigecycline in Proteus mirabilis. Two independent transposon insertion mutants of P. mirabilis that had 16-fold-increased susceptibility to tigecycline were mapped to the acrB gene homolog of the Escherichia coli AcrRAB efflux system. Wild-type levels of decreased susceptibility to tigecycline were restored to the insertion mutants by complementation with a clone containing a PCR-derived fragment from the parental wild-type acrRAB efflux gene cluster. The AcrAB transport system appears to be associated with the intrinsic reduced susceptibility to tigecycline in P. mirabilis.

The growing threat of acquired resistance in the Enterobacteriaceae (6, 10, 20) indicates the crucial need for new antibiotics for continued effective treatment of bacterial infection. Tigecycline, the 9-t-butylglycylamido derivative of minocycline, is a new antimicrobial agent belonging to a novel class of tetracyclines, the glycylcyclines (24). It has good activity against gram-positive pathogens, including penicillin-resistant Streptococcus pneumoniae, vancomycin-resistant enterococci, and methicillin-resistant Staphylococcus aureus (2, 5, 18). Tigecycline has good activity against most gram-negative pathogens, including Klebsiella pneumoniae and Escherichia coli (4, 18). Tigecycline also has good activity against organisms with a resistance determinant from the major facilitator family, including E. coli with tet(A), tet(B), tet(C), tet(D), and tet(M); S. aureus with tet(K) and tet(M); and Enterococcus faecalis with tet(M) (18). Proteus mirabilis is a notable exception to the activity of tigecycline, which routinely shows MICs of 4 μg/ml for the organism in tests.

It is important to identify current and emerging resistance mechanisms. Identification and mechanistic studies of bacterial resistance mechanisms can help to further reduce health care costs due to bacterial infection. Therefore, a study was performed to identify the mechanism responsible for the reduced susceptibility of P. mirabilis to tigecycline (M. A. Visalli, E. Murphy, S. J. Projan, and P. A. Bradford, Abstr. 41st Intersci. Conf. Antimicrob. Agents Chemother., abstr. C1-2019, 2001).

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains, plasmids, and transposons used in this study are listed in Table 1. A typical clinical isolate of P. mirablis, G151, was selected from a tigecycline phase II clinical trial. The E. coli strains used included Top10 (Invitrogen, Carlsbad, Calif.), Transformax EC100D pir-116 (Epicentre, Madison, Wis.), DM1 (Gibco Life Technologies, Rockville, Md.), and two laboratory strains, one wild type and the other an acrAB deletion mutant (17). The strains were grown in Luria-Bertani (LB) broth or agar in the presence of the following selective antibiotics when required: 50 μg of kanamycin/ml, 50 μg of ampicillin/ml, or 10 μg of gentamicin/ml.

TABLE 1.

Bacterial strains used in this study

Strain	Organism	Plasmid	Transposon	Characteristics	Genotype	Reference
G151	P. mirabilis			Clinical isolate		This study
GC 7020	P. mirabilis	pCLL3431		G151 with cloned AcrAB		This study
GC 6899	P. mirabilis		EZ::Tn	Insertion mutant		This study
GC 6900	P. mirabilis		EZ::Tn	Insertion mutant		This study
GC 7018	P. mirabilis	pCLL3431	EZ::Tn	Insertion mutant with cloned AcrAB		This study
GC 7019	P. mirabilis	pCLL3431	EZ::Tn	Insertion mutant with cloned AcrAB		This study
DM1	E. coli			Cloning strain	Fdam13::Tn9 (Cmr) dcm mcrB hsdRM+gal-2 ara lac thr leu (Tonr Tsxr)	Invitrogen
GC 7021	E. coli	pCLL3431			DM1 with cloned AcrAB	This study
AG100	E. coli			AcrAB deletion parent		17
GC7369	E. coli	pCLL3431		AG100 with cloned AcrAB		This study
AG100A	E. coli			AcrAB deletion strain	AcrAB−	17
GC 7368	E. coli	pCLL3431		AG100A with cloned AcrAB		This study
EC100Dpir-116	E. coli			Transposon rescue cloning strain	FmcrA Δ(mrr-hsdRMS-mcrBC) φ80lacZ ΔM15 ΔlacX74 deoR recA1 araΔ139 Δ(ara-leu)7697 galU ΔgalK λ rpsL nupG pir-116	Epicentre
Top 10	E. coli			Cloning strain	F−mcrA Δ(mrr-hsdRMS-mcrBC) φ80lacZ ΔM15 ΔlacX74 deoR recA1 araΔ139 Δ(ara-leu)7697 galU galK rpsL (Strr) endA1 nupG	Invitrogen
GC 7012	E. coli	pCLL3431		Top 10 with cloned AcrAB		This study

Antibiotic and susceptibility testing.

Antibiotic and substrate MICs were determined by broth microdilution using twofold dilution in Mueller-Hinton II broth (BBL, Cockeysville, Md.) according to the procedures established by the National Committee for Clinical Laboratory Standards (15). The following antibiotics, dyes, and detergents were used in this study: ampicillin, ceftriaxone, ciprofloxacin, novobiocin, ethidium bromide, chloramphenicol, erythromycin, acriflavine, and trimethoprim (Sigma Chemical Co., St. Louis, Mo.); imipenem (Merck, Rahway, N.J.); and tigecycline and minocycline (Wyeth Research, Pearl River, N.Y.). All substrates and antibiotics tested were prepared fresh on the day of testing.

Transposon mutagenesis.

To make the clinical isolate P. mirabilis G151 electrocompetent, cells were grown overnight in 25 ml of LB broth (without NaCl), 5 g of yeast extract/liter, and 10 g of tryptophan/liter at 37°C with shaking; 12.5 ml of the overnight culture was inoculated into 500 ml of fresh salt-free LB broth. The culture was grown at 37°C with shaking to an optical density at 600 nm of 0.6. The cells were chilled on ice for 15 min and then pelleted at 8,000 × g for 10 min at 4°C. The supernatant was removed, and the cell pellets were resuspended in 250 ml of ice-cold sterile 10% glycerol. The cells were pelleted and resuspended three more times in 10% glycerol in decreasing volumes of 100, 50, and, finally, 1 ml. The cells were snap frozen in a dry ice-ethanol bath in 110-μl aliquots and stored at −80°C. The electrocompetent cells were used the following day for transformation by electroporation with the Gene Pulser II system (Bio-Rad, Hercules, Calif.) using the EZ::TN <R6Kγori/KAN-2> transposon (Epicentre). The transposome system was used according to the manufacturer's protocol. The optimal electroporation settings with a cuvette gap size of 0.2 cm were 2.5 kV, 25 μF, 200 Ω, and ∼4.7 ms. Transformants were selected on LB agar plates (Difco Laboratories, Detroit, Mich.) with 2× agar (to inhibit swarming of the P. mirabilis colonies) and 50 μg of kanamycin/ml. The transformants were replica plated as previously described (8), using 50 μg of kanamycin/ml and tigecycline at a concentration of either 2 or 4 μg/ml.

Nucleic acid techniques.

Standard nucleic acid techniques were performed as described previously (21). Restriction enzymes (Roche Molecular Biochemicals, Indianapolis, Ind.) were used according to the manufacturer's instructions. Ligations were performed using the Fast-Link DNA ligation kit (Epicentre) according to the manufacturer's instructions. DNA sequencing of transposon insertion sites was performed using the Big Dye version 3 sequencing kit and the Applied Biosystems (Foster City, Calif.) automated DNA-sequencing system 3700. PCR was performed using the Failsafe PCR system (Epicentre). The PCR conditions were as follows: one cycle of denaturation at 94°C for 5 min followed by five cycles at 94°C for 3 min, 55°C for 30 s, and 72°C for 5 min, followed by 30 cycles at 94°C for 30 s, 55°C for 30 s, and 72°C for 5 min. A 5,281-nucleotide acrRAB gene fragment was amplified from genomic DNA of P. mirabilis G151 using the following primers derived from sequence generated in sequencing the transposon insertion: forward primer (5′-GCGTTTCTGGATGTTGCTCTT-3′) and reverse primer (5′-GATTACTTAGTTTGGTGCGGA-3′). This gene fragment was then ligated into the pCR 2.1-TOPO TA cloning vector (Invitrogen) according to the manufacturer's instructions. The resulting plasmid, pCLL3430, was then modified by cloning an 800-bp HindIII fragment containing the gentamicin cassette from pUCGm into the HindIII site in the multiple cloning site. The resulting plasmid, pCLL3431, was used in the complementation assays. Genomic DNA was prepared from LB broth cultures using the Puregene DNA isolation kit (Gentra Systems, Minneapolis, Minn.) according to the manufacturer's instructions.

DNA isolation and transposon mapping.

Transposon mapping was performed by rescue cloning the transposon and flanking chromosomal DNA. This was achieved by digestion of genomic DNA from insertion mutants using three individual restriction enzymes, EcoRI, PvuII, and BglII, none of which cut within the transposon. One microgram of fragmented DNA was self-ligated and then transformed into pir E. coli. Transformants were selected on LB agar containing 50 μg of kanamycin/ml. Plasmid DNA from the transformants was prepared using a QIAprep Spin Miniprep kit (Qiagen Inc., Chatsworth, Calif.) according to the manufacturer's instructions. Only the BglII-cleaved preparation generated a clone containing the full-length sequence of the disrupted gene.

Southern blot analysis.

Transposon insertion was confirmed by the digestion of the bacterial genomic DNA with restriction enzymes followed by electrophoresis in 1.0% agarose. The DNA fragments were then transferred to a nylon membrane (Hybond N+; Amersham Pharmacia Biotech, Piscataway, N.J.) using vacuum blotting. Hybridization and chemiluminescent detection were performed using the ECL Random-Prime Labeling and Detection System (Amersham Pharmacia Biotech). A 1-kb XbaI-XhoI fragment specific for kanamycin resistance gene sequences was isolated from the EZ:TN construct, labeled by enhanced chemiluminescence, and hybridized to the immobilized fragments.

Northern blot analysis.

Total P. mirabilis RNA was prepared using the Rneasy minikit (Qiagen). The RNA was electrophoresed and blotted as described by Sambrook et al. (21). Hybridization and detection were performed using the AlkPhos Direct labeling and detection system (Amersham Pharmacia Biotech). The probe was an 800-nucleotide internal EcoRI fragment generated from the acrRAB gene fragment of P. mirabilis in pCLL3430.

Bioinformatics.

Open reading frames from sequence data were translated using the EditSeq software program (DNAStar, Inc., Madison, Wis.) and used to perform a homology search with BLASTP (1).

Nucleotide sequence accession number.

The nucleotide and protein sequences of genes described from P. mirabilis are registered in GenBank under accession no. AY061647.

RESULTS

Transposon mutagenesis and mapping.

A typical clinical isolate of P. mirabilis, G151, for which the tigecycline MIC is 4 μg/ml, was selected for identification of the mechanism responsible for decreased susceptibility to tigecycline. Two independent transposon insertion mutants, GC 6899 and GC 6900, were isolated using the EZ::TN <R6Kγori/KAN-2> transposon (Epicentre). Southern blot analysis of DNA isolated from each transposon insertion mutant was performed using a kanamycin resistance gene probe. As shown in Fig. 1, the transposon was inserted into the P. mirabilis chromosome.

FIG. 1.

Genomic Southern blot showing transposon insertion into the chromosomes of two independently isolated insertion mutants. Lane 1, chromosomal DNA from the wild-type parent (G151); lane 2, positive control plasmid DNA (pCLL2300); lanes 3 and 4, chromosomal DNA from each of the insertion mutants (GC 6899 and GC6900). The presence of transposon insertion was detected by probing with the kanamycin resistance gene present in the transposable element.

The chromosomal DNA flanking the insertion site was rescue cloned, and nucleotide sequencing was performed. A BLASTP search of the translated open reading frames from each insertion mutant revealed insertion into a homolog of the acrB gene of E. coli. GC 6899 had an insertion at bp 1200, and GC 6900 had an insertion at bp 1925. Sequence analysis of the BglII rescue clones also revealed E. coli homologs of the acrA and acrR genes. Figure 2 is a linear view of transposon insertions and gene cluster arrangement.

FIG. 2.

Linear arrangement of P. mirabilis acrAB gene cluster.

Susceptibility profiles of insertion mutants.

The susceptibilities of the wild-type parental P. mirabilis strain G151 and the insertion mutants GC 6899 and GC 6900 to a number of antibiotics were determined (Table 2). Compared to those for the parent, the MICs of tigecycline for the two insertion mutants showed a 16-fold reduction, and the MICs of minocycline showed a 32-fold reduction. Transposon insertion did not affect susceptibilities to other antibiotics that are not substrates of the AcrAB efflux system, including ampicillin, ampicillin-sulbactam, ceftriaxone, and imipenem. The insertion mutants were also characterized by their increased susceptibilities to known AcrAB substrates, which included both antibiotics, such as ciprofloxacin, trimethoprim, novobiocin, and chloramphenicol, and dyes and detergents, such as ethidium bromide, acriflavin, and sodium dodecyl sulfate. As shown in Table 2, the MICs of all of the substrates tested for the insertion mutants showed decreases (range, 2- to 64-fold) compared to those for the wild-type parent.

TABLE 2.

Substrate profiles of strains expressing or lacking acrAB

Strain	Organism	acrABa	MIC (μg/ml)b
			TGC	MIN	AMP	SAM	CRO	CIP	IMP	NOV	EtBr	CHL	ERY	ACR	TRM	SDS
G151 WTc	P. mirabilis	+	4	32	0.5	0.5	≤0.06	0.06	2	16	2,048	64	>256	128	16	1,024
GC 7020	P. mirabilis	++	16	>64	>64	2	≤0.06	0.125	2	8	2,048	128	>256	64	16	512
GC 6899	P. mirabilis	−	0.25	1	0.25	0.5	≤0.06	≤0.015	4	≤0.25	128	8	16	16	≤1.0	128
GC 7018	P. mirabilis	++	16	>64	>64	4	≤0.06	0.06	2	8	2,048	128	>256	128	16	512
GC 6900	P. mirabilis	−	0.25	1	0.25	0.25	≤0.06	≤0.015	2	≤0.25	128	16	8	16	≤1.0	128
GC 7019	P. mirabilis	++	16	>64	>64	2	≤0.06	0.06	2	4	2,048	64	>256	128	16	512
DM1	E. coli	+	0.5	0.5	2	2	≤0.06	≤0.015	0.25	512	128	512	32	32	≤1.0	>2,048
GC 7021	E. coli	+	0.5	2	>64	>64	≤0.06	≤0.5	0.25	256	128	256	128	32	≤1.0	>2,048
AG100	E. coli	+	0.5	1	2	2	≤0.06	≤0.015	≤0.125	128	512	8	64	32	≤1.0	>2,048
GC 7369	E. coli	++	0.5	2	>64	64	≤0.06	≤0.015	≤0.125	128	256	8	128	16	≤1.0	>2,048
AG100A	E. coli	−	0.25	0.125	2	1	≤0.06	≤0.015	≤0.125	1	4	1	4	2	≤1.0	64
GC 7368	E. coli	++	1	4	>64	64	≤0.06	≤0.015	0.25	64	512	16	256	8	≤1.0	>2,048

^a+, wild-type acrAB present; ++, overexpression of P. mirabilis acrAB; −, not present.

^bTGC, tigecycline; MIN, minocycline; AMP, ampicillin; SAM, ampicillin-sulbactam (2:1); CRO, ceftriaxone; CIP, ciprofloxacin; IMP, imipenem; NOV, novobiocin; EtBr, ethidium bromide; CHL, chloramphenicol; ERY, erythromycin; ACR, acriflavine; TRM, trimethoprim; SDS, sodium dodecyl sulfate.

^cWT, wild type.

Cloning of the wild-type acrRAB gene complex and complementation studies.

Using the information obtained from the full-length nucleotide sequence generated by sequencing the insertion clones, a 5,128-bp acrRAB gene fragment from P. mirabilis G151 was amplified by PCR and ligated into the pCR2.1-TOPO TA cloning vector. The insert was then sequenced and compared to known homologs. This revealed that the acrB gene of P. mirabilis had 75% amino acid identity to the acrB genes found in E. coli, K. pneumoniae, and Enterobacter aerogenes. Interestingly, the acrB genes of E. coli, K. pneumoniae, and E. aerogenes have 85 to 88% identity to each other. The acrA gene of P. mirabilis also had 75% amino acid identity to the acrA genes found in E. coli, K. pneumoniae, and E. aerogenes.

The plasmid pCLL3430, containing a 5,128-bp acrRAB gene fragment from P. mirabilis G151, was altered by inserting a gentamicin resistance cassette into the multiple cloning site upstream of the acrRAB gene fragment. The resulting plasmid, pCLL3431, was then transformed into the wild-type parental P. mirabilis strain, the two insertion mutants, and four E. coli strains and selected for on LB agar containing 10 μg of gentamicin/ml. The susceptibilities of these strains were determined and are shown in Table 2. The overexpression of the cloned acrRAB gene fragment in the P. mirabilis wild-type parent (GC 7020) resulted in a fourfold elevation of the tigecycline MIC. The tigecycline MICs for P. mirabilis insertion mutants containing pCLL3431 (GC 7018 and GC 7019) showed a fourfold elevation over that for the wild-type strain. None of the strains containing pCLL3431 could be evaluated for a change in the ampicillin or ampicillin-sulbactam MICs because of the TEM-1 β-lactamase expressed as a selection marker on pCLL3431. The MICs of tigecycline and minocycline for all wild-type E. coli strains containing pCLL3431 and expressing the P. mirabilis acrAB genes showed no elevation. However, the tigecycline MIC for the E. coli acrAB deletion strain containing pCLL3431, GC 7368, showed a fourfold elevation compared to that for the parent strain. Again, all strains were tested with known substrates of AcrAB (Table 2). Insertion mutants containing pCLL3431 showed a wild-type parental phenotype for all substrates. No wild-type E. coli strains containing pCLL3431 showed a change in their susceptibilities to any of the substrates tested compared to their E. coli parent strains. The MICs of most substrates for the E. coli acrAB deletion mutant expressing the P. mirabilis AcrAB genes showed significant elevations compared to those for the parent (Table 2).

acrRAB expression.

To determine if the cloned acrRAB genes in the E. coli strains for which the tigecycline MIC was not elevated following transformation with pCLL3431 were being transcribed, Northern blot analysis of the wild-type parent strain (G151), GC 7020, and both insertion mutants, GC 6899 and GC 6900, along with their complements, GC 7018 and GC 7019, and two E. coli strains with and without pCLL3431 was performed. Transcripts were detected in all strains containing pCLL3431, including the E. coli strains (data not shown). The E. coli acrAB deletion mutant expressing the P. mirabilis AcrAB genes and its parent strain were not tested, as they were not available at the time of Northern blot analysis. The data suggest that expression of the acrRAB gene fragment from P. mirabilis in the various P. mirabilis strains was correlated with the observed MIC changes. However, the expression of the P. mirabilis AcrAB efflux system in wild-type E. coli strains did not affect susceptibilities to the various substrates tested.

DISCUSSION

Efflux systems have been identified as major contributors to bacterial resistance. This has furthered studies to identify numerous efflux systems in a broad range of organisms (25). The well-characterized AcrAB efflux pump in E. coli confers intrinsic resistance to many structurally diverse lipophilic compounds, including detergents, dyes, and antibiotics (7, 13, 14, 19). In this study, a homolog of the E. coli AcrAB efflux system was identified in P. mirabilis. This gene cluster appears to be responsible, at least in part, for the intrinsic reduced susceptibility to tigecycline in P. mirabilis, as shown by transposon mutagenesis and complementation studies.

The AcrAB efflux system in E. coli is known to transport hydrophobic substrates, including dyes, detergents, and antibiotics, directly out of the cell without accumulation in the periplasm (12). This efflux system was previously described as a tripartite complex consisting of AcrA, a periplasmic lipoprotein; an inner membrane transporter, AcrB; and an outer membrane channel, TolC (3, 12).

Members of several other genera, including E. coli, K. pneumoniae, and E. aerogenes, have close homologs of the AcrAB efflux system (11, 16, 22) yet do not show decreased susceptibility to tigecycline. The reason for the functional differences has yet to be found.

The transfer of resistance determinants among bacterial genera is a major concern for the spread of resistance. In this study, the possibility that the transfer and subsequent expression of the P. mirabilis AcrAB efflux system in different genera would result in decreased susceptibilty to tigecycline or other known substrates was investigated. When the AcrAB efflux system of P. mirabilis was expressed in various E. coli strains, wild-type strains did not show a change in their susceptibilities to any of the antibiotics or substrates tested. Although we did not express TolC from P. mirabilis in the E. coli clones expressing the P. mirabilis acrAB genes to determine the effect of the third component of this efflux system on tigecycline susceptibility, it is unlikely that the genes encoding TolC would be transferred to E. coli in a natural setting because they are located some distance from the AcrAB operon on the P. mirabilis chromosome. This suggests that the mobilization of the AcrAB pump from Proteus onto a plasmid does not pose an immediate threat of acquired resistance to tigecycline in E. coli.

It appears that even though the P. mirabilis AcrAB system was expressed in wild-type E. coli, it had no effect on susceptibilities to these substrates. Similar results have been previously reported in which the MexAB-OprM system from Pseudomonas aeruginosa was expressed in E. coli but showed no susceptibility changes. However, when the same P. aeruginosa efflux system was expressed in an E. coli strain containing an acrAB deletion, there were decreases in susceptibilities to a variety of substrates tested (23). We show a similar phenomenon in the E. coli acrAB deletion strain expressing the P. mirabilis AcrAB efflux system.

There are several possible explanations for the apparent lack of effect of P. mirabilis AcrAB expression in wild-type E. coli strains. First, there could be a conformational preference for the E. coli proteins over the P. mirabilis proteins. Regulation by the E. coli system might be preventing the expression of the P. mirabilis AcrAB efflux system. We looked at expression of the P. mirabilis AcrAB efflux system in E. coli via RNA levels. Since we did not look at actual protein levels, it is possible that the P. mirabilis transcripts are not being translated (9). Furthermore, it is possible that there are other, yet-undefined factors involved in this efflux system.

The reason for the differences between the MIC s for wild-type and acrAB deletion E. coli strains expressing the P. mirabilis AcrAB efflux system has not yet been explained and deserves further exploration. However, these experiments suggest that the simple transfer of acrRAB genes from P. mirabilis to other genera may not play a significant role in acquired resistance to glycylcyclines.

Although tigecycline is unaffected by classical tetracycline resistance determinants in E. coli, the identification of the AcrAB efflux pump in P. mirabilis, which is related to the reduced susceptibility of the organism to tigecycline, suggests that an efflux pump which can act upon tigecycline exists. The reason that the AcrAB pump acts on tigecycline in P. mirabilis but not in other genera has yet to be determined. Fortunately, there does not appear to be an immediate threat of the spread of resistance to tigecycline.