Successive Emergence of Extended-Spectrum β-Lactamase-Producing and Carbapenemase-Producing Enterobacter aerogenes Isolates in a University Hospital

Patient and bacterial strain data.From July 2003 to May 2005, 62 clinical isolates of E. aerogenes resistant to expanded-spectrum cephalosporins were identified from four A.U.H. wards. Among these isolates, 23 (37.1%) were IPM susceptible (IPM-S), and 39 (62.9%) were IPM insusceptible (IPM-Ins). These isolates were collected from a group of 22 adult patients, consisting of 12 men (54.5%) and 10 women (45.5%) with a mean age of 67.1 years (range, 34 to 90 years). These patients had been hospitalized for periods of 7 to 181 days (median, 64.5 days). The time to acquisition of E. aerogenes colonization or infection ranged from 7 to 132 days (median, 46.7 days). The temporal occurrence of colonization in these patients showed the presence of one or more colonized and then infected patients on the wards at the time of colonization acquisition by a new patient and a constant rate of transmission throughout the study period.

Seven patients were admitted to the hepatology and gastroenterology ward, seven were admitted to the polyvalent intensive care unit (ICU), seven were admitted to the general and visceral surgery ward, and one was admitted to the respiratory ICU. Eight patients were admitted for digestive adenocarcinoma, seven for acute pancreatitis, five for multiple trauma, one for prostate adenocarcinoma, and one for posttraumatic leg necrosis. Eleven patients had received IPM plus an expanded-spectrum cephalosporin prior to the isolation of the organisms, and 11 patients had not received IPM prior to the isolation of the organisms, but these patients were treated with cefepime (FEP) or cefpirome (CPO) plus gentamicin (G).

The clinical isolates collected from these patients were identified as E. aerogenes by using an ID32 system (Bio-Rad, Marne-la-Coquette, France) according to the manufacturer's instructions. Table 2 shows the sources of these isolates.

Phenotypic detection of ESBL-, AmpC-, and MBL-producing isolates.Isolates were screened for ESBL production by using a double-disk synergy test (DDST). Mueller-Hinton medium was inoculated, and disks containing the standard 30-μg quantities of cefotaxime (CTX), FEP, CPO, ceftazidime (CAZ), and aztreonam (ATM) were placed 30 mm (center-to-center) from an amoxicillin-clavulanic acid (20 and 10 μg, respectively) disk. After disks were incubated, an enhanced zone of inhibition between any one of the β-lactam disks and the amoxicillin-clavulanic acid disk was interpreted as presumptive evidence of the presence of an ESBL. The same isolates were also screened for AmpC production by using a disk potentiation test on Mueller-Hinton agar containing, as the inhibitor, cloxacillin (250 μg/ml) (A.E.S. Chemunex, Bruz, France), according to the manufacturer's instructions. After the disks were incubated, the absence of the inhibition diameter from around the amoxicillin-clavulanic acid zone and the potentiation of the zone of inhibition around CTX, FEP, CPO, CAZ, and ATM indicated the possible presence of AmpC β-lactamase. Metallo-β-lactamase (MBL) production was detected by using the MBL Etest strips (AB Biodisk, Solna, Sweden) with the IPM-Ins E. aerogenes isolates, according to the manufacturer's instructions.

Antimicrobial susceptibility.MICs were determined for all isolates studied by using an Etest method (AB Biodisk, Solna, Sweden). The MICs were determined for the following antibiotics: cefoxitin (FOX), CTX, CAZ, FEP, CPO, ATM, IPM, G, tobramycin (T), netilmicin (N), amikacin (A), trimethoprim-sulfamethoxazole (SXT), and ofloxacin (OFX). Escherichia coli ATCC 25922 was used as the MIC reference strain. The MIC results were interpreted according to the Comité de l'Antibiogramme de la Société Française de Microbiologie guidelines (12).

Plasmid and chromosomal DNA preparation, Southern blotting, and hybridization.Plasmid and chromosomal DNA extraction from the clinical isolates was performed using a Qiagen plasmid kit (Qiagen, Courtaboeuf, France) and a GNOME kit (Qbiogene, France) according to the manufacturers' instructions. Plasmid and chromosomal DNAs were detected by 0.8% agarose gel electrophoresis. The sizes of the plasmids were determined by comparison with the plasmid standards pUC8 (2.678 kb), pBR322 (4.362 kb), RP4 (54 kb), and pIP173 (125.800 kb) (from Collection Institut Pasteur [CIP] strains).

Plasmid and genomic DNAs were digested with EcoRI, and the digested fragments were analyzed by electrophoresis using a 1-kb ladder (Promega, Madison, WI) as a DNA size marker. The restriction fragments were transferred from an agarose gel to a nylon membrane (Amersham Hybond-N⁺; GE Healthcare Bio-Sciences, Orsay, France) by the Southern method (4). A digoxigenin DNA labeling and detection kit (Roche Diagnostics, Mannheim, Germany) was used for labeling the probes bla_TEM, bla_SHV, bla_IMP-1, and bla_VIM-2 and for hybridization.

Isoelectric focusing (IEF).Sonicated extracts were prepared from cultures in brain heart infusion, by ultrasonic disintegration, as described previously (29). β-Lactamases with known isoelectric points (pI) (TEM-1, 5.4; TEM-24, 6.5; SHV-1, 7.6; SHV-4, 7.8; IMP-1, >9.5; and VIM-2, 5.3) were focused in parallel with the extracts. β-Lactamase activity was detected directly on the gel by nitrocefin (Oxoid Ltd., Basingstoke, United Kingdom) (29).

PCR for the detection of TEM and SHV genes.PCR was carried out with a final volume of 50 μl containing 25 μl Go Taq Green Master Mix (Promega, Madison, WI), 20 μM (each) primer, and 2 μl of plasmid DNA preparation. The cycling conditions were 2 min at 95°C, followed by 30 cycles of 95°C for 1 min, 55°C for 1 min, and 72°C for 1 min, and then 72°C for 5 min.

PCR for the detection of MBL (IMP-1 and VIM-2) genes.PCR was carried out with 1 μl of plasmid DNA. The reaction conditions consisted of predenaturation at 94°C for 2 min and then 30 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 1.5 min, and then 72°C for 5 min.

PCR mapping of the integrons (integron PCR), the IntL3 integrase gene (integrase gene PCR), and the aminoglycoside resistance gene.PCR amplification of the detection of class 1 integrons was performed with the primers 5′-CS and 3′-CS, using DNA as the template. Combinations of the 5′-CS primer and the bla_IMP-1 or bla_VIM-2 gene and the 3′-CS primer and the bla_IMP-1 or the bla_VIM-2 gene were used for the determination of the genetic content of class 1 integrons. For the detection of the intL3 and aac(6′)-Ib genes, PCR was performed with specific oligonucleotide primers. The annealing step was carried out at 57°C and 55°C for the intL3 and aac(6′)-Ib genes, respectively.

The primer sizes of the expected amplification products are listed in Table 1. Amplification reactions were carried out on a thermal Cycler 2400 instrument (Applied Biosystems, Foster City, CA). The PCR products and the base pairs (bp) DNA Smart ladder (Eurogentec, Belgium) were analyzed by 1% agarose gel electrophoresis with Tris-EDTA as the running buffer. The gels were stained with ethidium bromide, and a single observed band was photographed on a UV light transilluminator.

Sequencing of the bla_TEM, bla_SHV, bla_IMP-1, bla_VIM-2, aac(6′)-Ib, and intL3 gene-specific PCR products.PCR products containing these genes were purified with a QIAquick PCR purification kit (Qiagen, France) and subjected to direct sequencing on an ABI PRISM 377 automated sequencer (Applied Biosystems, Foster City, CA). The identities of the genes encoding TEM, SHV, IMP-1, VIM-2, the aac(6′)-Ib gene, and the IntL3 enzymes were resolved by sequencing the bla_TEM, bla_SHV, bla_IMP-1, bla_VIM-2, aac(6′)-Ib, and intL3 gene-specific PCR products with the same primers. The comparison of the determined nucleotide sequences with the sequence databases was performed with an updated version of FASTA and BLAST-2 software. Multiple sequence alignments were performed with a CLUSTAL program.

OMP analysis.The OMPs were isolated as sodium lauryl sarcosinate (1.7%) (Sigma-Aldrich, Germany), insoluble material from cell envelopes obtained by the sonication of bacteria grown in brain heart infusion broth (16). The electrophoretic analysis of OMPs was performed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), as described elsewhere (23). Western blotting analysis of SDS-PAGE-separated OMPs was carried out with the buffers and conditions described by Towbin et al. (38).

Western blotting analysis of proteins separated by SDS-PAGE was carried out by transfer to Immobilon membranes (Millipore, Saint-Quentin Yvelines, France). Membranes were then incubated sequentially with 5% skim milk-phosphate-buffered saline (PBS), anti-OmpK36 serum, alkaline phosphatase-labeled goat anti-rabbit immunoglobulin G (IgG) (Sigma-Aldrich, Saint-Quentin, France), and 5-bromo-4-chloro-3-indolylphosphate disodium-nitroblue tetrazolium (BCIP-NBT) (7). Incubations were carried out for 1 h, and washing steps with 0.05% Tween 20-PBS were included after each incubation step. Polyclonal antibodies directed against the E. coli porins were recognized by the E. aerogenes porins, as reported previously (28). Purification of the OmpK36 protein from selected strains and antiserum was performed according to previously published procedures (1).

Epidemiological typing.PFGE was performed as described previously (19), using SpeI (3 U/μl) as the restriction enzyme. Digested whole-cell DNA was separated on a 1% agarose gel (pulsed field certified; Bio-Rad, France) with a contour-clamped homogeneous electric field system (CHEF-DRII system; GenePath, Bio-Rad, France). The gel was stained with ethidium bromide and visualized with a Gel Doc 2000 system (Bio-Rad, France). A lambda ladder (Bio-Rad, France) was used as the molecular weight marker, and E. coli O157:H7 (Bio-Rad, France) and E. aerogenes 1846 and 1847 (CIP strains) were used as reference strains. Patterns were analyzed by visual comparison of the printed digital gel images and were interpreted as described previously (37). A percentage of similarity between each pulsotype was computed by Molecular Analyst Fingerprinting software (Bio-Rad, France), and Simpson's index of diversity (36) was used to analyze the PFGE results.

Antimicrobial resistance.One hundred percent of the strains were resistant to FOX (MIC, >256 μg/ml), CAZ (MIC, >256 μg/ml), SXT (MIC, >32 μg/ml>), and OFX (MIC, >32 μg/ml); 96.7% were resistant to T (MIC, 12 to 48 μg/ml) and N (MIC, 16 to 96 μg/ml); 93.6% were resistant to A (MIC, 32 to 64 μg/ml) and ATM (MIC, 48 to >256 μg/ml); 56.4% were resistant to IPM (MIC, 12 to >32 μg/ml); 29.1% were resistant to CTX (MIC, >32 μg/ml); and 22.5% were resistant to FEP and CPO (MICs, >32 μg/ml). Of the strains tested, 74.1% were intermediate to FEP (MIC, 8 to 32 μg/ml) and CPO (MIC, 12 to 32 μg/ml), 70.9% were intermediate to CTX (MIC, 8 to 32 μg/ml), 6.4% were intermediate to IPM (MIC, 8 μg/ml), 6.4% were intermediate to ATM (MIC, 24 to 32 μg/ml), and 3.2% were intermediate to A (MIC, 12 to 16 μg/ml).

Four phenotypes were defined on the basis of their resistance to β-lactams: phenotype a (Fox^r Ctx^r+i Fep^r+i Cpo^r+i Atm^r+i Ipm^r+i Caz^r, where “r + i” means insusceptible) included 61.2% (38/62) of the strains; phenotype b (Fox^r Ctx^r+i Fep^r+i Cpo^r+i Atm^r+i Caz^r) included 35.4% (22/62) of the strains; phenotype c (Fox^r Ctxⁱ Atmⁱ Caz^r, where “i” means intermediate) included 1.7% (1/62) of the strains; and phenotype d (Fox^r Ctx^r Atm^r Ipm^r Caz^r) consisted of 1.7% (1/62) of the strains.

Two aminoglycoside resistance phenotypes were also identified: phenotype e (T^r N^r A^r+i) included 91.9% (57/62) of the strains, and phenotype f (G^r T^r N^r) corresponded to 4.8% (3/62) of the strains.

ESBL-like phenotypes and MBL production.The DDST detected the presence of ESBLs in 60 of the 62 isolates studied. They were intermediate or resistant to CTX (MIC, 8 to 32 μg/ml), FEP (MIC, 8 to >32 μg/ml), CPO (MIC, 12 to 32 μg/ml), and ATM (MIC, 24 to >256 μg/ml) and resistant to CAZ (MIC, >256 μg/ml). Clavulanic acid inhibited the enzymes which hydrolyzed these substrates and subsequently enhanced the zones of inhibition of these substrates. Two of 62 (3.2%) isolates were DDST negative. They were sensitive to FEP and CPO and produced β-lactamases which were inhibited by cloxacillin, presumably indicating AmpC β-lactamase production. Twenty-six of the 39 IPM-Ins isolates were MBL positive by Etest, and the 13 remaining strains were MBL nonproducers.

Plasmid and chromosomal DNA encoding β-lactamases.All plasmid and chromosomal DNA were obtained from E. aerogenes isolates. These isolates had one or more plasmids with the following molecular sizes: 3.5, 6, 10, and 20 kb. Six- to twenty-kilobase plasmids were present in all ESBL-producing E. aerogenes isolates. The plasmid DNA obtained was recognized by bla_TEM-, bla_SHV-, and bla_IMP-specific primers used in this experiment, but not by bla_VIM-2.

Repeated attempts to detect the presence of plasmids capable of hybridizing with the bla_VIM-2 probe failed. Meanwhile, the fragments of the EcoRI-digested genomic DNA showed hybridization, suggesting a chromosomal location. On the other hand, repeated attempts to detect bla_TEM,bla_SHV, and bla_IMP-1 on plasmids were successful, confirming the specificity of the PCR assay.

Phenotypic characterization of β-lactamases: the problem of multiplicity within the phenotype.The MICs of the β-lactam drugs and the results of IEF analysis of the 62 isolates studied suggested ESBL expression in 60 isolates and MBL expression in 39 isolates. Two isolates (Table 2, no. 61 and 62) were ESBL and MBL nonproducers but were suspected producers of high-level chromosomally mediated AmpC β-lactamase. Five β-lactamase bands with pI values of 5.4 (5 isolates), 6.5 (39 isolates), 8.2 (55 isolates), >9.5 (24 of 39 IPM-Ins isolates), and 5.3 (2 of 39 IPM-Ins isolates) were identified (Table 2). These results suggested the presence of ESBL and MBL, for which the frequent association made it difficult to interpret resistance phenotypes.

PCR analysis was used initially to confirm and/or identify the β-lactamases observed by IEF. PCR using specific primers identified the presence of TEM (52/62 isolates), SHV (58/62 isolates), IMP-1 (24/39 IPM-Ins isolates), and VIM-2 genes (2/39 IPM-Ins isolates). The specific PCR products encoding enzymes with pIs of 5.4 (n = 5), 6.5 (n = 10), >9.5 (n = 10), and 5.3 (n = 2) were selected for sequencing (one or two strains were selected from those isolated at the beginning of the epidemic and one or two strains isolated at the end of the epidemic, and the others were selected at random during the epidemic). All specific PCR products encoding SHV, with a pI of 8.2, were sequenced because SHV-5 and SHV-12 have the same pI values.

Identification and nucleotide sequencing of the bla_TEM, bla_SHV, bla_IMP, and bla_VIM genes.The nucleotide sequences of the E. aerogenes bla_TEM gene were compared to the known sequences of bla_TEM-1, accession no. AF309824, and bla_TEM-24, accession no. X65253 (10) (98 to 100% identity). Two mutations, T₇₄₇C and G₉₁₄A, resulted in two amino acid substitutions, M182T and G238S, in the TEM-20 β-lactamase with a pI of 5.4 (2), and five additional amino acid substitutions were observed, as previously described (9), for TEM-24 with a pI of 6.5. The SHV gene sequences were compared with those of bla_SHV-5, accession no. AY386368, and bla_SHV-12, accession no. AY008838. The bla_SHV gene, which exhibited a 99 to 100% nucleotide homology, corresponded to the homologies observed for bla_SHV-5 (6) and bla_SHV-12 (22). Sequences for ESBL SHV enzymes, with a pI of 8.2, had a G-to-A mutation in the bla gene, resulting in amino acid substitutions G236S and E237K found in SHV-5 β-lactamase in four isolates. Two additional amino acid substitutions, G238S and E240K, were observed for SHV-12 β-lactamase in 54 isolates.

The nucleotide sequences of the bla_IMP-1 and bla_VIM-2 genes from our E. aerogenes strains were identical to those known for genes encoding IMP-1 (100% homology with sequences for Serratia marcescens, accession no. S71932) (31) and VIM-2 (99% homology with sequences for Pseudomonas aeruginosa, accession no. AF191564 and AY242984) (34), respectively. These enzymes with pIs of >9.5 and 5.3 corresponded to IMP-1 and VIM-2 β-lactamases, respectively.

OMP characterization.Thirteen IPM-Ins and MBL-nonproducing isolates as well as 7 IPM-S and MBL-nonproducing isolates were characterized by SDS-PAGE.

IPM-Ins and IPM-S strains demonstrated only two major proteins, of about 32 and 42 kDa. In contrast, E. coli HB101 produced four possible OMPs, of 45, 42, 38, and 32 kDa (Table 2). Three of thirteen IPM-Ins strains for which MICs of IPM were more than 32 μg/ml did not express the 42-kDa OMP but produced only the 32-kDa OMP, whereas the remaining 10 strains for which MICs of IMP were between 8 μg/ml and 32 μg/ml expressed both 32- and 42-kDa OMPs with a weak band of 42-kDa porin. Some minor nonspecific reactions were observed for the Western blot with endogenous E. coli porins, presumably because of the high degree of homology between enterobacterial porins.

Integrons and integrase PCR results.The use of the 5′-CS and 3′-CS primers allowed the detection of class 1 integrons and the sizes of any inserted gene cassette. The frequency of integron carriage was 100% (39/39) of the E. aerogenes IPM-Ins isolates. The inserted gene cassette sizes that were detected were 1 and 2 kb. The integrase gene PCR showed a frequency of 79.4% (31/39) of IntL3-positive IPM-Ins isolates. The intL3 gene was found in 79.1% (19/24) of isolates of IMP-1 producers, and both of the VIM-2-producer isolates were intL3 positive. The PCR mapping technique revealed that 94.8% (37/39) of IPM-Ins isolates carried the aac(6′)-Ib genes. Of these, 24 isolates were associated with bla_IPM-1, and two were associated with bla_VIM-2.

The sequences of intL3 and aac-(6′)-Ib products were 100% identical to previously published sequences (accession no. AF416297 [11] and accession no. 231133 [37a], respectively).

Sequence analysis of both the 1-kb and the 2-kb integrons revealed a structure with at least two gene cassettes, containing bla_IMP-1 and aac-(6′)-Ib.

Molecular epidemiology.Six SpeI patterns, named A to F, were found among the 62 E. aerogenes isolates. The major epidemic pattern A comprised 87% (54/62) of isolates genetically related to subtypes A₁ (1 isolate), A₂ (2 isolates), A₃ (1 isolate), and A₄ (1 isolate). The β-lactam phenotype a was present in 59.2% (32/54) of the isolates, phenotype b was present in 37% (20/54) of the isolates, and phenotypes c and d were present in 1.9% (1/54) of pulsotype A. Profiles B, C, D, E, and F corresponded to sporadic cases (only one isolate, except for C and E, which had two and three isolates, respectively) (Table 2). Pulsotype A was identified from patients on all wards included in this study. At the 73% similarity breakpoint, the patterns were categorized into five clusters (I to V). Cluster II contained 87% of the isolates. The percentage of similarity between isolates of this cluster reached at least 80% and belonged to the PFGE pattern A. For the whole set of isolates, the index of diversity was 0.10, indicating their homogeneity and their clonality.