Modification of Outer Membrane Protein Profile and Evidence Suggesting an Active Drug Pump in Enterobacter aerogenes Clinical Strains

Abstract

Two clinical strains of Enterobacter aerogenes that exhibited phenotypes of multiresistance to β-lactam antibiotics, fluoroquinolones, chloramphenicol, tetracycline, and kanamycin were investigated. Both strains showed a porin pattern different from that of a susceptible strain, with a drastic reduction in the amount of the major porin but with an apparently conserved normal structure (size and immunogenicity), together with overproduction of two known outer membrane proteins, OmpX and LamB. In addition, the full-length O-polysaccharide phenotype was replaced by a semirough Ra phenotype. Moreover, in one isolate the intracellular accumulation of chloramphenicol was increased in the presence of the energy uncoupler carbonyl cyanide m-chlorophenylhydrazone, suggesting an energy-dependent efflux of chloramphenicol in this strain. The resistance strategies used by these isolates appear to be similar to that induced by stress in Escherichia coli cells.

Bacteria have developed various regulatory systems which coordinate their adaptive responses with the different environmental stresses to which they are exposed. Recently, the whole-genome transcriptional profiles, or “transcriptomes,” were determined in vitro for an Escherichia coli strain exposed to an inducer of the soxRS or marRAB system and an E. coli strain constitutively expressing MarA (7, 39). Under these conditions, the expression of several genes appeared to be significantly activated or down-regulated. These modulations of gene expression alter the sensitivities of the bacteria to a broad range of antibiotics (1, 2).

Enterobacter aerogenes is one of the more frequently described gram-negative bacteria responsible for nosocomial respiratory tract infections (5, 17). In the last 5 years, it has been shown that clinical isolates of this species, which are naturally resistant to aminopenicillins through their production of a chromosomal cephalosporinase, often express an extended-spectrum β-lactamase, TEM-24, which gives rise to resistance to β-lactam antibiotics (5, 17, 19, 31). Moreover, E. aerogenes exhibits acquired resistance to other families of antimicrobial agents. Previous studies have reported that clinical strains exhibiting an efflux process are resistant to β-lactam antibiotics, quinolones, tetracycline, and chloramphenicol (12, 29). Drug efflux can be coincident with a drastic reduction in drug uptake due to the loss of porin content (16, 22, 23, 29).

Two E. aerogenes clinical isolates, isolates 117 and 119, were selected from up to 100 strains isolated in our laboratory (12). These two strains exhibited a phenotype of multiresistance to β-lactam antibiotics, fluoroquinolones, chloramphenicol, tetracycline, and kanamycin comparable to that of E. aerogenes strains lacking nonspecific porins or expressing mutated porins (12, 29).

The aim of this work was to examine some factors that may contribute to the resistance to antimicrobial agents in these two E. aerogenes clinical strains.

MATERIALS AND METHODS

Bacterial strains, growth conditions, and antibiotic susceptibility tests.

Two E. aerogenes clinical strains, strains 117 and 119, were isolated during an epidemiological study carried out in the South Marseille Hospitals (Marseille, France) and were identified by using the API 20E identification system (BioMérieux). These isolates showed noticeable reductions in the levels of nonspecific porins and had high levels of resistance to several β-lactam antibiotics (12). E. aerogenes ATCC 13048 was used as a control strain. Bacteria were routinely grown in Luria-Bertani or Mueller-Hinton broth at 37°C.

Susceptibilities to antibiotics were determined by a twofold standard broth microdilution method (35). The results were scored after 18 h at 37°C and are expressed as the MICs (in micrograms per milliliter). The presence of β-lactamase was investigated by using clavulanic acid (2 μg/ml), as described previously (29).

Drug interaction was performed by the same dilution method described above. Various concentrations of antibiotics, alone or in the presence of phenylalanine arginine β-naphthylamide (PAβN) at 26.3 μg/ml, were tested (27). The results were scored after 18 h at 37°C and are expressed as the MICs.

Outer membrane protein isolation and immunoblotting.

Outer membrane proteins were prepared from cultures in the exponential growth phase in Mueller-Hinton broth by ultrasonic treatment, followed by ultracentrifugation and differential solubilization of the cytoplasmic material with sodium lauryl sarcosinate (0.3%). The final preparations were electrophoresed on sodium dodecyl sulfate (SDS)-polyacrylamide gels (10% [wt/vol] acrylamide, 0.1% [wt/vol] SDS). Immunodetection with antibodies was carried out after electrotransfer onto nitrocellulose, as described previously (18, 30). The OmpC, OmpF, OmpA, TolC, LamB, and OmpX bands from SDS-polyacrylamide gels were characterized by Western blotting with different polyclonal antibodies (42). Detection was then performed with alkaline phosphatase-conjugated Affiniti Pure goat anti-rabbit immunoglobulin G antibodies (Jackson ImmunoResearch).

The lipopolysaccharide (LPS) profile was determined after silver staining (44) of the whole-cell preparation, which had been treated with proteinase K (25) and run on an SDS-polyacrylamide gel with Tricine buffer (6, 21).

N-terminal amino acid sequence determination and detection of mutations in gyrA.

To determine the N-terminal sequences of the three unknown proteins of 85, 50, and 20 kDa, the bands from the SDS-polyacrylamide gels were transferred by Western blotting onto a polyvinylidene difluoride membrane (18, 33). These bands were excised, and their N-terminal sequences of at least 10 amino acids were determined by Edman degradation with a 492A protein sequencer (Applied Biosystems).

To detect mutations in the region of the E. aerogenes gyrA gene that corresponds to the quinolone resistance-determining region of the E. coli gyrA gene, DNA fragments of gyrA were amplified from chromosomal DNAs of the E. aerogenes isolates by PCR with one set of primers, primers EC gyrA6 and EC gyrA631R (47), and the sequences were determined.

Detection of chloramphenicol acetyltransferase.

A chemical chloramphenicol acetyltransferase assay was performed by the method reported by Walker and Brown (46). Acetyl coenzyme A and 5,5′-dithio-bis(2-nitrobenzoic acid) were used as reagents. A positive reaction was indicated by the development of a deep yellow color.

Measurement of chloramphenicol uptake.

Chloramphenicol uptake was measured as described previously (34). Exponential-phase bacteria in Luria-Bertani broth were removed by centrifugation and washed once in sodium phosphate buffer (50 mM; pH 7) supplemented with 5 mM MgCl₂. The pellets were suspended in the same buffer to obtain a density of 2 × 10¹⁰ CFU/ml. To abolish the energy of the bacteria and to inhibit the efflux pump competitively, 50 μM carbonyl cyanide m-chlorophenylhydrazone (CCCP) and 200 μM PAβN, respectively, were added 5 min before the addition of [¹⁴C]chloramphenicol.

Nucleotide sequence accession number.

The partial sequence of the E. aerogenes ATCC 13048 gyrA gene reported here appears in GenBank under accession number AF052255.

RESULTS

Antibiotic susceptibility.

The two clinical isolates E. aerogenes 117 and 119 had identical resistance profiles, with high-level resistance to the extended-spectrum cephalosporins (cefotaxime, ceftazidime), cefepime, and aztreonam. The presence of clavulanic acid, a β-lactamase inhibitor, had no effect on the MICs of any of the β-lactam antibiotics tested (Table 1). In contrast, the two isolates remained susceptible to imipenem. This type of resistance pattern has been shown to be related to the expression of a chromosomally encoded cephalosporinase of the Bush-Jacoby-Medeiros group 1 (38). Moreover, strain 117 produced a TEM-24 extended-spectrum β-lactamase of Bush-Jacoby-Medeiros group 2be, as described previously (12).

TABLE 1.

Susceptibilities of the E. aerogenes strains to a range of structurally unrelated antibiotics

	MIC (μg/ml)a
	ATCC 13048	117	119
Cefotaxime	0.5	256	256
Cefotaxime + clavulanate	0.25	512	256
Ceftazidime	1	>512	>512
Ceftazidime + clavulanate	1	>512	>512
Cefepime	0.25	64	32
Cefepime + clavulanate	0.25	64	32
Aztreonam	1	512	512
Aztreonam + clavulanate	1	512	512
Imipenem	0.125	0.25	0.125
Chloramphenicol	16	512	16
Tetracycline	2	8	8
Kanamycin	32	>128	>128
Gentamicin	1	8	16
Amikacin	4	32	16
Nalidixic acid	4	>256	>256
Norfloxacin	0.25	256	256
Ciprofloxacin	0.125	32	16

^aThe values are the means of three independent assays.

Both E. aerogenes strains had similar levels of resistance to quinolones, and their responses to tetracycline were comparable to that of the ATCC 13048 control strain. The high MICs of quinolones (nalidixic acid, norfloxacin, and ciprofloxacin) indicated the presence of mutations in the gyrA gene. Sequencing of the quinolone resistance-determining region of the gyrA gene showed that both of the quinolone-resistant E. aerogenes isolates had changes causing the amino acid substitution Thr-83 to Ile in GyrA, and an additional amino acid change, Ala-67 to Ser, was detected in isolate 119. These amino acid changes in GyrA were previously described in E. coli (11, 47). No changes were observed in ParC.

The chloramphenicol MIC was higher for strain 117 than for strain 119, which had the same susceptibility as strain ATCC 13048.

Analysis of outer membrane structure.

SDS-polyacrylamide gel electrophoresis analysis of the outer membrane proteins revealed a low level of expression of the major nonspecific porin in each of the clinical strains (strains 117 and 119) compared to that in reference strain ATCC 13048 (Fig. 1). The major porin, Omp 36, from the reference strain has been purified, and its functional properties as a porin channel were reported previously (18). In addition, the porins from isolates 117 and 119 were not found in the inner membrane or the cytoplasmic fractions, nor were they found in the culture media, thus indicating that the porin was correctly folded in the outer membrane. Results of immunoanalyses with polyclonal antibodies directed against the denatured porins (18) confirmed the reduced amounts of porin in the two clinical isolates relative to that in the susceptible strain (data not shown).

FIG. 1.

Outer membrane fractions of the various E. aerogenes strains. Outer membrane proteins were stained with Coomassie blue. E. aerogenes ATCC 13048 presented the major porin (18), while strains 117 and 119 presented decreased porin levels and an additional band at 46 kDa. Arrow, porin migration; dot, OmpA; triangles, the additional bands detected for the respective clinical isolates; asterisk, the position of an additional unknown product in isolate 119. Lane MWS, molecular weight standards (in kilodaltons); lanes 1, 2, and 3, outer membrane fractions of E. aerogenes ATCC 13048, 117 and 119, respectively.

In contrast to the decrease in porin levels, we observed a normal level of OmpA expression (Fig. 1) in association with overproduction of other outer membrane proteins in isolates 117 and 119. Among these, we observed a marked increase in a product migrating at about 18 kDa. In order to identify this protein, the band was excised from the gel and the N-terminal sequence was determined by microsequencing. The sequences of the first 11 residues were identical for both strains and showed identity with the N-terminal sequence of OmpX from Enterobacter cloacae (43). In addition, we detected OmpX expression in the two isolates (data not shown) by using polyclonal antibodies directed against OmpX, as described previously (28).

The sequence analyses carried out on the 46-kDa product revealed an N terminus identical to the first 12 residues of E. coli LamB. In addition, immunodetection carried out with specific polyclonal antibodies confirmed the production of LamB in both clinical isolates. The N-terminal sequence of the additional product that migrated at about 80 kDa and that was detected only in the outer membrane of isolate 119 showed a 52% identity with the amino-terminal sequence of the putative FepA of Salmonella enterica (CAD 05061). Isolates 117 and 119 exhibited core oligosaccharide (LPS) profiles that were the same as the LPS profile of strain ATCC 13048, but they appeared to express a modified high-molecular-mass O-antigenic LPS side chain, which may be evidence of a major alteration of the outer membrane ultrastructure (data not shown).

Effect of PAβN on susceptibilities to various antibiotics.

PAβN (27, 28), an antibiotic efflux pump inhibitor, was assayed for its ability to increase the antibiotic susceptibilities of the E. aerogenes strains. The presence of PAβN in the incubation medium at a concentration which did not affect the growth of the E. aerogenes strains tested resulted in up to a fourfold decrease in the chloramphenicol MIC for both E. aerogenes ATCC 13048 and 119 and up to an eightfold decrease in the chloramphenicol MIC for isolate 117 (Table 2). PAβN had no effect on the MIC of cefepime (36), acriflavine, norfloxacin, or tetracycline for the two isolates or for the reference strain tested, suggesting that efflux is not the major mechanism involved in the resistance to these molecules. Alternatively, the efflux pump inhibitor used may not be capable of blocking antibiotic export in these strains.

TABLE 2.

MIC results by use of PAβN in combination with various antimicrobial agents

Antimicrobial agent and inhibitor	MIC (μg/ml)a
	ATCC 13048	117	119
Cefepime	0.25	64	32
Cefepime + PAβN	1	128	32
Chloramphenicol	16	512	16
Chloramphenicol + PAβN	4	64	4
Acriflavine	128	128	128
Acriflavine + PAβN	128	128	128
Norfloxacin	0.5	256	256
Norfloxacin + PAβN	0.5	256	256
Tetracycline	2	16	8
Tetracycline + PAβN	2	16	8

^aThe values are the means of three independent assays carried out in the absence or presence of PAβN at 26.3 μg/ml.

Effects of CCCP and PAβN on intracellular chloramphenicol accumulation.

As no chloramphenicol acetyltransferase activity was detected in the two clinical isolates (data not shown) and because of the effect of PAβN on the MIC of chloramphenicol, drug efflux (15) could be involved in the case of strain 117. Consequently, we decided to investigate the uptake of radiolabeled chloramphenicol and to compare the intracellular accumulation of this antibiotic in the two clinical isolates and sensitive strain ATCC 13048. Since chloramphenicol efflux is energy dependent, incubation with an uncoupler (CCCP) will clearly indicate whether an active efflux pump is involved in determining the intracellular concentration of [¹⁴C]chloramphenicol. No significant CCCP-sensitive efflux was observed for strain ATCC 13048 or 119. In contrast, an energy-dependent efflux was evidenced in isolate 117: the addition of CCCP, which collapses the proton gradient across the cytoplasmic membrane (Fig. 2), induced a noticeable increase in the intracellular level of radiolabeled chloramphenicol, reaching about 240% of the basal level observed for the resistant isolate in the absence of the uncoupler. In strain 117, in the presence of PAβN, which acts by competitive inhibition of efflux pump systems such as AcrAB-TolC (28), the amount of chloramphenicol accumulated reached about 180% of the intracellular level obtained in the absence of CCCP (Fig. 2).

FIG. 2.

Uptake of chloramphenicol by E. aerogenes isolate 117. The intracellular accumulation of [¹⁴C]chloramphenicol in E. aerogenes isolate 117 was determined in the absence (cross) and in the presence (square) of CCCP or in the presence (circle) of PAβN. Each point is the mean of three assays.

DISCUSSION

The modulation of envelope permeability plays a role in the resistance to antimicrobial agents of gram-negative bacteria such as E. coli (32, 37) and Klebsiella pneumoniae (24). In E. aerogenes, resistance can result from the loss or reduction of a major porin or from the expression of a porin altered in the constriction area, which decreases the diameter and modifies the electrostatic field or the expression of an efflux mechanism (10, 12, 13, 18, 29, 45). The implications of this complex phenotype, which includes a deficit in porin synthesis and multidrug resistance, has been studied in two clinical multidrug-resistant E. aerogenes strains isolated in our laboratory (12).

Despite the reduced level of the major porin, analysis of outer membrane fractions indicated a normal structure for the E. aerogenes porin: a normal molecular mass and normal porin assembly in the outer membrane. In addition, the incorporation of OmpA into the membrane appeared to be unaffected. The reduction in the amount of major porin in the E. aerogenes outer membrane was associated with overproduction of two known outer membrane proteins, OmpX and LamB. Although the precise function of OmpX has yet to be elucidated, it has been reported that overexpression of OmpX causes a decrease in the quantities of cellular porins via a reduction in the level of porin gene transcription (43). LamB production could be explained in part by coregulation with major porin expression at the level of transcription and/or at the level of translation (14, 20). The synthesis of LamB, which is usually induced by maltose (8, 9), was constitutive in our clinical strains. Similar results have been reported in part for organic solvent-tolerant E. coli mutants and for some K. pneumoniae clinical isolates. The E. coli mutants overexpressed three proteins: an unknown 77-kDa protein, OmpX, and LamB (3). Moreover, a K. pneumoniae strain with a loss of porin expression or a reduced level of porin expression in association with LamB overexpression has been reported (4).

The transcriptional profile analyses of E. coli responses to (i) superoxide stress and sodium salicylate and (ii) MarA overexpression indicate increased levels of expression of several genes. The marA transcriptional activator of the multiple-antibiotic resistance cascade and the AcrAB-TolC pump involved in antibiotic efflux were particularly up-regulated (7, 39, 40). Several other genes, including ompX and lamB, were also considerably activated under these conditions, while other genes, such as ompF, ompC, and some of the genes involved in LPS biosynthesis, were down-regulated (7, 39).

It is interesting that in the two resistant clinical strains we clearly detected overproduction of OmpX and LamB, a severe reduction in the level of porin synthesis, modification of the LPS structure such that the phenotype was similar to a Ra phenotype (26, 41), and expression of an efflux mechanism. The phenotypes of the bacteria identified here suggest that the genetic response reported in vitro (7, 39) may be of biological relevance in clinical E. aerogenes isolates. Obviously, several of these elements are essential to the mechanism that allows E. aerogenes to show high levels of resistance to various classes of antibiotics. This strategy may be a major contributing factor to the emergence of the multidrug resistance phenotype in response to antibiotic use in hospitals.