Occurrence of Efflux Mechanism and Cephalosporinase Variant in a Population of Enterobacter aerogenes and Klebsiella pneumoniae Isolates Producing Extended-Spectrum β-Lactamases

Abstract

We investigated the occurrence of multidrug resistance in 44 Enterobacter aerogenes and Klebsiella pneumoniae clinical isolates. Efflux was involved in resistance in E. aerogenes isolates more frequently than in K. pneumoniae isolates (100 versus 38% of isolates) and was associated with the expression of phenylalanine arginine β-naphthylamide-susceptible active efflux. AcrA-TolC overproduction in E. aerogenes isolates was noted. An analysis of four E. aerogenes isolates for which cefepime MICs were high revealed no modification in porin expression but a new specific mutation in the AmpC β-lactamase.

Enterobacter aerogenes and Klebsiella pneumoniae, two of the most prevalent nosocomial enterobacterial species, can frequently express a multidrug resistance (MDR) phenotype by the acquisition or high-level production of β-lactamases in combination with the overproduction of efflux pumps, associated or not with porin alterations in the outer membrane (1, 3, 7, 9, 13). Such modifications of envelope permeability, including efflux and influx (porin), in 28 Enterobacter aerogenes and 16 Klebsiella pneumoniae clinical isolates from Nîmes University Hospital, previously studied for extended-spectrum β-lactamase (ESBL) profiling and clonal identification, were investigated (11). Moreover, we ascertained the mechanism involved in cefepime resistance in four E. aerogenes isolates showing high-level resistance (MIC = 128 mg/liter).

Characterization of the MDR phenotype was carried out by determining the MICs of various structurally unrelated antibiotics selected for their capacities to be expelled and/or to enter cells via porins (Table 1). We used the twofold broth dilution method with Mueller-Hinton broth according to the CLSI guidelines: the tests were carried out with or without the broad-spectrum efflux pump inhibitor (EPI) phenylalanine arginine β-naphthylamide (PAβN). This diamine compound, used at a low concentration (26.3 mg/liter), can increase the activities of chloramphenicol, tetracycline, norfloxacin, and sparfloxacin against various enterobacterial isolates which overproduce efflux pumps such AcrAB-TolC (4, 9, 16). The test results confirmed the existence of a PAβN-susceptible efflux mechanism in all E. aerogenes and many K. pneumoniae strains, with marked decreases in the MICs of fluoroquinolones and chloramphenicol in the presence of PAβN (Table 1). Efflux pump activity was identified when the PAβN addition induced a threefold decrease in the MIC of an antibiotic molecule (4, 9). For the two fluoroquinolones, the PAβN addition had better efficiency in decreasing the MICs of sparfloxacin than those of norfloxacin. The observed difference in PAβN efflux inhibition between the two fluoroquinolones may reflect the respective locations of ligands inside the AcrB cavity and the flux competition (20). Yu et al. demonstrated, during previous cocrystalization analyses, the presence of different affinity sites located inside the AcrB pump and suggested that the benefit generated by the conjoint use of an EPI and an antibiotic is due to a variable synergistic effect (23). In addition, it is important that PAβN, a well-described EPI, can restore a high intracellular ATP concentration but cannot bypass other resistance mechanisms, such as a target mutation or altered permeability, existing in clinical isolates (5). Moreover, some isolates may also overproduce other efflux pumps less susceptible to PAβN (12).

TABLE 1.

Susceptibility data for E. aerogenes and K. pneumoniae clinical isolates and E. aerogenes and E. coli transformants^a

Strain	MIC (μg/ml) of:
	CM	CM + PAβN	TET	TET + PAβN	NOR	NOR + PAβN	SPAR	SPAR + PAβN	IPM	FEP	CAZ	FOX
E. aerogenes clinical isolates
EA1	1,024	128	8	8	256	256	256	8	1	32	>2,048	>1,024
EA2	512	64	8	4	256	256	128	8	1	32	>2,048	>1,024
EA3	1,024	128	8	8	256	256	128	8	1	16	>2,048	>1,024
EA4	1,024	128	16	8	512	512	128	8	1	32	>2,048	>1,024
EA5	>1,024	128	32	32	512	512	512	16	2	64	>2,048	>1,024
EA6	256	4	8	8	64	2	0.125	≤0.06	1	4	256	>1,024
EA7	1,024	128	16	16	256	256	128	4	2	64	>2,048	>1,024
EA8	256	64	2	2	128	128	64	4	1	64	>2,048	1,024
EA9	256	64	4	4	128	128	128	8	1	32	>2,048	>1,024
EA10	1,024	128	8	8	256	256	256	8	1	16	>2,048	>1,024
EA11	512	64	16	8	512	256	128	4	1	64	>2,048	512
EA12	1,024	128	16	16	512	512	128	8	1	128	>2,048	>1,024
EA13	1,024	64	8	8	512	512	256	8	2	64	>2,048	>1,024
EA14	512	64	16	16	512	512	128	8	2	64	>2,048	>1,024
EA15	512	64	16	8	512	128	128	8	0.5	16	>2,048	>1,024
EA16	512	64	2	2	512	128	64	8	1	64	>2,048	>1,024
EA17	1,024	64	32	8	>512	128	256	16	1	64	>2,048	>1,024
EA18	32	4	8	4	128	128	128	16	1	32	1,024	>1,024
EA19	1,024	64	8	8	128	128	128	16	2	64	>2,048	>1,024
EA20	1,024	64	16	4	512	128	256	16	1	64	>2,048	>1,024
EA21	1,024	128	16	8	512	128	512	32	2	128	>2,048	>1,024
EA22	1,024	128	8	4	256	128	256	8	1	64	>2,048	>1,024
EA23	512	64	8	8	512	256	128	4	1	64	2,048	256
EA24	512	64	8	8	512	128	256	16	1	128	>2,048	>1,024
EA25	512	64	4	4	256	128	128	8	1	64	>2,048	>1,024
EA26	256	32	2	2	256	128	32	4	1	2	1,024	1,024
EA27	1,024	128	16	8	256	128	256	8	1	128	>2,048	>1,024
K. pneumoniae ATCC 11296	4	2	1	1	0.5	0.125	≤0.06	≤0.06	0.5	0.125	0.5	8
K. pneumoniae clinical isolates
KP1	8	8	256	128	2	2	0.25	0.25	0.25	64	2,048	16
KP2	256	64	2	2	4	2	2	≤0.06	0.25	64	2,048	16
KP3	4	4	256	256	128	128	8	4	2	64	>2,048	32
KP4	4	2	256	256	128	128	8	2	1	64	>2,048	32
KP5	32	2	128	128	2	0.5	0.125	≤0.06	0.5	64	>2,048	16
KP6	32	2	256	256	1	0.25	≤0.06	≤0.06	0.5	16	32	16
KP7	32	2	128	128	0.25	0.25	0.25	≤0.06	0.5	64	>2,048	16
KP8	32	2	128	128	2	0.25	0.125	≤0.06	0.5	64	>2,048	8
KP9	64	8	128	128	1	≤0.06	≤0.06	≤0.06	0.5	64	>2,048	16
KP10	4	2	256	256	128	128	8	2	0.5	64	>2,048	32
KP11	8	4	4	4	64	64	8	2	0.5	64	>2,048	32
KP12	64	2	>512	256	8	0.25	≤0.06	≤0.06	1	16	16	64
KP13	128	4	>512	128	>256	32	2.1	0.5	1	2	16	32
KP14	16	2	>512	256	8	0.25	2	≤0.06	0.5	2	16	64
KP15	8	2	512	128	256	256	32	8	1	32	>2,048	64
KP16	16	4	128	128	1	0.5	0.125	≤0.06	0.5	64	2,048	32
E. aerogenes ATCC 15038	2	1	0.5	0.5	0.125	0.125	≤0.06	≤0.06	0.25	0.03	0.25	128
E. aerogenes transformants
ATCC 15038(pDrive)	ND	ND	ND	ND	ND	ND	ND	ND	0.5	0.25	0.5	ND
ATCC 15038(pACM204)	ND	ND	ND	ND	ND	ND	ND	ND	0.5	0.25	64	ND
ATCC 15038(pDrive-ampC)	ND	ND	ND	ND	ND	ND	ND	ND	2	32	64	ND
E. coli strains	ND	ND	ND	ND	ND	ND	ND	ND
DH5α	ND	ND	ND	ND	ND	ND	ND	ND	0.25	0.125	2	ND
DH5α(pDrive)	ND	ND	ND	ND	ND	ND	ND	ND	0.25	0.5	2	ND
DH5α(pDrive-ampC)	ND	ND	ND	ND	ND	ND	ND	ND	2	32	32	ND
DH5α(pACM204)	ND	ND	ND	ND	ND	ND	ND	ND	0.5	0.25	32	ND
ATCC 35218	ND	ND	ND	ND	ND	ND	ND	ND	0.125	0.03	0.125	ND
ATCC 35218(pDrive)	ND	ND	ND	ND	ND	ND	ND	ND	0.25	0.5	0.5	ND
ATCC 35218(pDrive-ampC)	ND	ND	ND	ND	ND	ND	ND	ND	0.5	32	16	ND

^aSusceptibilities of E. aerogenes and K. pneumoniae clinical isolates and E. aerogenes ATCC 15038, E. coli DH5α, and E. coli ATCC 35218 transformants with pDrive alone or pDrive bearing ampC (pDrive-ampC) to various antibiotics with and without PAβN. Abbreviations: CM, chloramphenicol; TET, tetracycline; NOR, norfloxacine; SPAR, sparfloxacine; IPM, imipenem; FEP, cefepime; CAZ, ceftazidine; FOX, cefoxitin; and ND, not determined. MICs were obtained from three independent measurements.

All strains were susceptible to imipenem, with MICs for E. aerogenes strains slightly higher than those for K. pneumoniae strains and MICs for all strains ranging from 0.25 to 2 mg/liter. High MICs of ceftazidime and cefoxitin were associated with the expression of an ESBL and the AmpC cephalosporinase, respectively. Two representative groups of isolates were selected to investigate (i) active drug efflux by studying the expression of the AcrAB pump element and measuring the accumulation of radiolabeled chloramphenicol and (ii) porin expression by using immunodetection. Eight isolates (E. aerogenes EA6, EA13, EA17, and EA20 and K. pneumoniae KP2, KP12, KP13, and KP14 [designations beginning with EA indicate E. aerogenes isolates, and those beginning with KP indicate K. pneumoniae isolates]) showing significant responses to PAβN, with some MICs for these strains decreasing, were selected for the examination of the efflux mechanism. Strains with susceptibility to imipenem corresponding to a MIC of 2 mg/liter (EA5, EA7, EA13, EA19, EA21, and KP3) were chosen for permeability studies. For the chloramphenicol accumulation test, the isolates picked out were EA6, EA13, EA17, KP13, and KP14. The method of measuring the uptake of radiolabeled chloramphenicol into fresh cells was adapted from previous studies in order to monitor the involvement of an active efflux mechanism in the intracellular drug concentration (8). Inhibition assays were performed by adding the energy uncoupler carbonyl cyanide m-chlorophenylhydrazone (CCCP), which collapses the membrane transporter activity, to evaluate the energy involvement in the active efflux for three exposure times (13). In comparison with the data for reference strains, the results displayed the inhibitory effects of CCCP on the efflux pumps of clinical isolates, with increases in the intracellular chloramphenicol concentrations (by two to five times) (Fig. 1). The antibiotic uptake by the reference strains in the presence or absence of CCCP indicated no active efflux. The PAβN showed an effect similar to that of CCCP on chloramphenicol accumulation when added during accumulation in selected isolates (data not shown).

FIG. 1.

Effect of the EPI on the intracellular accumulation of [¹⁴C]chloramphenicol. The assays were carried out in the presence or absence of CCCP. Values (expressed as counts per minute per unit of optical density [OD]) were obtained from duplicate independent experiments.

It was reported previously that the overproduction of AcrA and TolC pump elements is involved in the resistance levels of clinical isolates (4, 8, 9). The membrane proteins of selected isolates were extracted and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunodetection with polyclonal antibodies directed against AcrA and TolC components (9, 18). Immunoblots were stained by colorimetric detection with alkaline phosphatase-conjugated secondary antibodies. Three reference strains, EAEP289 (a Kan^s derivative of EA27, which is a clinical MDR strain), EAEP294 (EAEP289 acrA::Kan^r), and EAEP298 (EAEP289 tolC::Kan^r), were used as internal controls (18). Concerning K. pneumoniae, similar AcrA and TolC profiles were obtained with the ATCC reference strain and isolates. The selected E. aerogenes strains exhibited the overproduction of the tested efflux components, in comparison with the production of these components in the reference strain (Fig. 2C and D). Regarding K. pneumoniae isolates, an efflux mechanism was detected but the AcrAB-TolC pumps in these strains seemed to be less involved in resistance than those in E. aerogenes strains.

FIG. 2.

(A and B) Immunodetection of porins (A) and internal loop 3 (B) in E. aerogenes and K. pneumoniae strains. (C and D) Immunoblots of whole-cell extracts with polyclonal antibodies against AcrA (C) or TolC (D). Only the relevant parts of the gels are shown. MW, molecular weight marker; SM, size marker.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting methods were carried out as described previously using polyclonal antibodies directed against denatured porins and a polyclonal antiserum (F4) prepared against the internal loop 3 peptide sequence (7, 13). The results showed that all tested isolates expressed the major nonspecific porin (Fig. 2A). For E. aerogenes isolates, we obtained identical results with F4 antibodies, indicating a conserved sequence for the internal loop 3 porin epitope. However, F4 failed to recognize the peptides belonging to the loop 3 epitopes of porins from the tested K. pneumoniae strains.

In this clinical population of enterobacteria, the drug resistance was associated with the expression of PAβN-susceptible efflux and no alteration of porin synthesis was observed. It has been reported that these two mechanisms, influx (via porins) and efflux (via pumps), are controlled by the same complex regulation cascade (5, 10). However, they can be independently expressed (7, 8, 9, 10, 13). Regarding K. pneumoniae isolates, the signals corresponding to AcrAB-TolC components in the tested MDR strains did not show significant overstaining while a noticeable reversing effect of PAβN on these strains was obtained. These observations suggest that another efflux pump may be involved or, alternatively, that some changes occurring in antigenic sites may alter the immunodetection of the K. pneumoniae isolate efflux components.

Strikingly, the level of cefepime resistance in four E. aerogenes strains (EA12, EA21, EA24, and EA27) was especially high (MIC = 128 mg/liter). Three of the four strains were epidemiologically unrelated, and all were isolated after the use of broad-spectrum cephalosporin therapy (11). High levels of resistance to cephalosporins (up to céphèmes) are associated with the production of ESBLs or the hyperproduction of wild-type AmpC, but MICs have not exceeded 4 mg/liter (17). Thiolas et al. described a specific resistance phenotype of E. aerogenes due to a mutation in loop 3 of the Omp36 major porin that alters the pore properties (21). This substitution alters the antigenic site of the F4 antibody, and no signal was revealed by immunodetection with F4 (21). Positive F4 signals were obtained for the four E. aerogenes isolates tested here, indicating that the loop 3 epitope was preserved (data not shown). It was demonstrated in previous studies that cefepime is not efficiently removed by efflux and that PAßN has no effect on the MIC of cefepime for E. aerogenes, similar to its lack of effect for K. pneumoniae (7, 9).

Mutations in the AmpC cephalosporinase were recently reported to be the source of cefepime resistance in Escherichia coli, Enterobacter cloacae, and E. aerogenes (2, 6, 14, 15, 22). To evaluate the relationship between AmpC and cefepime resistance, the four E. aerogenes strains were transformed with plasmid pNH5 bearing an ampD gene (2). The MICs of cefepime for the transformants decreased by a factor of 128, confirming the involvement of ampC in the resistance. The four E. aerogenes isolates were singled out for the amplification and sequencing of the ampC to ampR genes with primers E1 (5′-TGCGTGTCATAACATTATCCG-3′) and E2 (5′-AACCCGTAGCCCAGGTAAAC-3′) as described previously (2). After the amplification of the genes and the sequencing of the PCR products, the sequences were compared to the E. aerogenes 97B sequence deposited in GenBank (accession no. AF211348). Three substitutions corresponding to the replacements Q-90-H, W-101-C, and L-108-Y, located on the H3 helix of the enzyme, were observed in the four strains. The final impact of these mutations on the functional structure of the enzyme is difficult to assess, but the H3 helix seems to be a region involved in cefepime resistance, as reported previously for the H10 helix (2). However, to confirm the involvement of mutations in the level of resistance to cefepime, PCR products were cloned with primers E1 and E2 into the pDrive PCR cloning vector (Qiagen) and recombinant plasmids were used to transform E. coli DH5α, E. coli ATCC 35218, and E. aerogenes ATCC 15038. For a control, pACM204, a plasmid bearing the wild-type ampC gene, was used to transform E. coli DH5α and E. aerogenes 15038 (19). The MICs for the transformants are presented in Table 1. The altered AmpC induced increases in cefepime resistance (raising the MICs by 32- to 258-fold compared to those for susceptible strains) and ceftazidime resistance (raising the MICs by 16- to 256-fold). No modification of the MIC of cefepime was observed with the wild-type AmpC, demonstrating the involvement of the AmpC variant in the high resistance levels. These mutations seem to induce key modifications in the structure of the AmpC β-lactamase, leading to changes in the catalytic properties that result in an extension of the enzymatic spectrum to include cefepime.

In the United States, 20% of K. pneumoniae infections and 31% of Enterobacter infections in intensive care units involve strains resistant to cephalosporins (17). Moreover, MICs of cefepime for E. aerogenes strains possessing ESBLs and overproducing AmpC are about 4 to 8 mg/liter (17). This work sheds new light on the occurrence of an active efflux mechanism and the identification of a new AmpC variant involved in high cefepime MICs for E. aerogenes. These two emerging mechanisms of resistance may become more frequent as broad-spectrum antibiotics are used. This is the first description of the association of these two resistance mechanisms in E. aerogenes. Such resistance to cefepime requires further attention to define the best options for detection and treatment.