Inducible β-Lactam Resistance in Aeromonas hydrophila: Therapeutic Challenge for Antimicrobial Therapy

Abstract

Despite the abundant amount of knowledge about inducible chromosomally mediated β-lactamases among Aeromonas species, extended-spectrum β-lactam-resistant A. hydrophila strains selected in clinical practice were rarely reported. In the present study, two strains of A. hydrophila, A136 and A139, with markedly different susceptibilities to extended-spectrum cephalosporins were isolated from blood and the tip segment of an arterial catheter of a burn patient. Another strain (A136m) was selected in vitro by culturing A136 in a subinhibitory concentration of cefotaxime, the β-lactam agent administered for the treatment of Aeromonas bacteremia in this patient. Typing studies by arbitrarily primed PCR and pulsed-field gel electrophoresis indicated a clonal relationship among strains A136, A136m, and A139. These strains were identified to be of DNA hybridization group 1. Wild-type strain A136 was resistant only to ampicillin and cephamycins, but A136m and A139 were highly resistant to the expanded- and broad-spectrum cephalosporins. The presence of increased β-lactamase activity in A139 suggests that A139 is a derepressed mutant which overexpresses β-lactamases. These results call attention to the use of β-lactam agents for the treatment of invasive Aeromonas infections.

Species of the motile mesophilic genus Aeromonas have been known to be pathogenic in immunocompetent and compromised persons (13). They were usually isolated from such clinical specimens as feces, blood, ascitic fluid, and wound discharge or pus (12, 17). Aeromonas infections also occur in hospital settings (24), where they are immersed in the incremental pressure of antibiotic selection. Antibiotic resistance will potentially become a problem among Aeromonas strains causing nosocomial infections.

The plasmid-mediated β-lactamases in enteric bacteria have been studied and characterized extensively. The mechanisms mediating antibiotic resistance in clinical Aeromonas species were elucidated recently. Chromosomally mediated, inducible β-lactamases were recognized as the major mechanism of antibiotic resistance (15). Aeromonas species were found to possess at least three inducible chromosomally mediated β-lactamases (34). The expression of genes encoding these three different β-lactamases was coordinated by a common regulatory pathway (2). Derepressed mutants that constitutively produce β-lactamases have been selected in vitro from A. hydrophila, A. veronii, and A. caviae (35). The evolution of β-lactam-sensitive Aeromonas strains into β-lactam-resistant mutants during β-lactam therapy has been described only for A. caviae (4). Here we report that the use of cefotaxime might promote the development of β-lactam resistance in clinical A. hydrophila strains from a burn patient.

Brief clinical history.

A 62-year-old female was initially healthy and suffered from a flame burn on 25 April 1995. The burn covered about 61% of her total body surface area. On the next day she was transferred to the burn center, which was a six-bed intensive care unit of the National Cheng Kung University Hospital, Tainan, Taiwan. After the initial standard care and escharotomy for her burn wound, she was intubated for pulmonary inhalation injury. Fever appeared on the second day of hospitalization, and gentamicin plus cephradine and then gentamicin plus ampicillin-sulbactam were administered intravenously. The fever persisted, and on the sixth day, the burn wounds deteriorated and A. hydrophila was isolated from the wound and the blood. On the seventh day, surgical debridement was performed. Intravenous cefotaxime was given according to the in vitro susceptibility report for the isolates described above. However, 3 days later A. hydrophila with the same antibiogram as the previous bacteremic strain was cultured from the blood. On the 15th day, nosocomial Candida albicans fungemia was detected. On the next day, a strain of A. hydrophila resistant to cefotaxime was isolated from the distal portion of the arterial indwelling catheter. The antimicrobial therapy was adjusted and ciprofloxacin plus fluconazole were given intravenously. The septic process was not halted, and acute renal failure and pulmonary edema occurred. On the 38th day, hypothermia, severe metabolic acidosis, and septic shock developed and she died on the following day.

Nosocomial bacteremia and wound infections caused by Aeromonas species were rarely encountered in our burn center. Since there was no clustering of other patients with nosocomial Aeromonas infection in the ward, no epidemiological survey was carried out.

MATERIALS AND METHODS

Bacterial strains and identification.

Although four strains of A. hydrophila were isolated from the patient during the hospitalization, only the first bacteremic strain (A136) and the strain from the arterial catheter (A139) were available. They were stored at −70°C for studies. Two reference strains of A. hydrophila were purchased from the Culture Collection and Research Center (CCRC), Hsinchu, Taiwan. CCRC 13018 is the same strain as ATCC 7966; CCRC 13881 is the same strain as ATCC 43414. Two randomly selected clinical isolates of A. hydrophila, A2866 and A4252, from different patients were used as the control group. The organisms were identified to the genus and species levels by conventional methods (14), and the identification was supplemented by the API-20E system (bioMérieux Vitek, Hazelwood, Mo.).

Genomic species identification.

In phenotypic strains of A. hydrophila, there are three DNA hybridization groups (HGs). The identification of HG1 is presumptively based on growth on dl-lactate (3). Identification of genomic species was further confirmed by rRNA gene restriction patterns and PCR sequencing of 16S rRNA.

(i) rRNA gene restriction patterns.

Aeromonas sp. strains A306 (HG1), A307 (HG2), and A308 (HG3) and plasmids pKK3535 and pGML were kindly provided by M. Altwegg (University of Zurich, Zurich, Switzerland). rRNA gene restriction patterns were used for characterization of genomic species identification. The method was based on one previously described by Lucchini and Altwegg (23). In brief, the genomic DNA was extracted, and aliquots were digested with SmaI (New England Biolabs, Beverly, Mass.). Fragments were separated in a 1.2% agarose gel in TBE (Tris-borate-EDTA) buffer, stained with ethidium bromide, and transferred to a nylon membrane, and the 567-bp HindIII fragment of pGML1 containing the rrnB operon of Escherichia coli was used as a probe. Methods of DNA hybridization have been described previously (33).

(ii) PCR sequencing.

Nucleic acids were extracted by simple mechanical lysis of bacterial cells as described by Kirschner et al. (16). Finally, a 5-μl aliquot was used in a PCR. Primers 5′-AGAGTTTGATCATGGCTCAG-3′ (forward) and 5′-GGTTACCTTGTTACGACTT-3′ (reverse) (6), at positions 8 to 27 and 1509 to 1491, respectively, of the E. coli numbering system, were used to generate a 1.4-kb fragment in the 16S rRNA gene. A 50-μl PCR mixture contained each deoxynucleoside triphosphate at a concentration of 200 μM, 1.25 U of Taq polymerase (Applied Biosystems Division, Perkin-Elmer Corp., Norwalk, Conn.), 1× PCR buffer and 0.5 μM (each) primer. The PCR protocol was performed by the protocol of Borrell et al. (6). Sequencing reactions were performed with the Taq DyeDeoxy terminator cycle sequencing kit (Applied Biosystems Division, Perkin-Elmer Corp.) with a GeneAmp PCR System 9600 (Perkin-Elmer Corp.) and a DNA Analysis System 373 Stretch (Applied Biosystems Division, Perkin-Elmer Corp.). Both strands of the gene were sequenced with primers 5′-TGGAGGAATACCGGTGGCGA-3′, at positions 704 to 723, and 5′-ATCTCTACGCATTTCACCGC-3′, at positions 702 to 683, as well as the primer pair used for PCR. The sequences obtained were compared to known sequences in the GenBank database and were interpreted by using the BlastN algorithm.

In vitro selection of mutants by a subinhibitory concentration of cefotaxime.

According to the medical record, cefotaxime was the β-lactam agent used prior to isolation of the resistant strain. It was likely that the use of cefotaxime offered the appropriate selective pressure for the emergence of a derepressed mutant. Thus, a subinhibitory concentration of cefotaxime was used in vitro to select resistant mutants. Wild-type strain A136 was grown overnight at 37°C with 0.2 μg of cefotaxime per ml (one-fourth the MIC) in Luria-Bertani (LB) broth. Cells that grew in the broth (strain A136m) were subcultured and then saved for further MIC determinations.

In vitro susceptibility test.

Initially, the in vitro tests for susceptibility to commonly used antibiotics were performed by the disk diffusion method. The diameters of the inhibition zones were measured and used for categorization of the strain as susceptible, intermediate, or resistant as described by the National Committee for Clinical Laboratory Standards (NCCLS) (26). The MICs for the Aeromonas strains were determined with the E-test strip (AB Biodisk, Solna, Sweden). Bacterial suspensions adjusted to a density equivalent to that of a 0.5 McFarland standard were used as inocula for determination of MICs with E-test strips. The interpretive breakpoint concentrations were in accordance with those of NCCLS (27).

Characterization of genotype.

Two clinical isolates (A136 and A139) from the same patient, A136m, A. hydrophila CRCC 13018 and CRCC 13881, and two randomly selected clinical isolates of the same species were investigated for genetic polymorphism. The techniques of arbitrarily primed PCR (AP-PCR) and pulsed-field gel electrophoresis (PFGE) were used to demonstrate the molecular similarity.

(i) AP-PCR.

Two primers, ERIC-1R (5′-ATGTAAGCTCCTGGGGATTCAC-3′) and ERIC-2R (5′-AAGTAAGTGACTGGGGTGAGCG-3′), were used for AP-PCR. The process of amplification was carried out in a GeneAmp PCR System 9600 (Perkin-Elmer Corp.). It was programmed for 4 cycles of 1 min at 94°C, 1 min at 37°C, and 2 min at 72°C, followed by 35 cycles of 1 min at 94°C, 1 min at 60°C, and 2 min at 72°C. The amplification products were separated by 1.2% agarose gel electrophoresis, stained with ethidium bromide, and visualized with a UV transilluminator.

(ii) Chromosomal DNA analysis by PFGE.

DNA embedded in agarose beads was prepared as described by Piggot et al. (29), with modifications. In brief, 30 ml of an overnight culture in LB broth was harvested and washed with 1× TE (10 mM Tris HCl [pH 7.5], 1 mM EDTA [pH 8.0]). The suspension was mixed with an equal volume of 1% low-melting-temperature agarose and 2 volumes of warm (42°C) paraffin oil. The mixtures were shaken vigorously for 2 min to form an emulsion. The emulsion was poured onto 10 ml of ice-cold 1× TE in a flask, and the contents of the flask were mixed by vigorous shaking for 5 min. The agarose beads were harvested and suspended in 15 ml of T10E (10 mM Tris HCl [pH 7.5], 10 mM EDTA [pH 8.0]), and the suspension was incubated at 37°C for 1 to 2 h with gentle shaking. The beads were incubated overnight at 40 to 45°C in solution C (1% sarcosyl, 0.4 M EDTA, 0.1 mg of protease K per ml), harvested by centrifugation, and resuspended in 15 ml of TE buffer (pH 8.0) containing 1 mM phenylmethylsulfonly fluoride and incubated for 2 h at room temperature with gentle shaking. The agarose beads were digested with 10 U of SpeI (New Englands Biolabs) for 18 h and were electrophoresed through a 1% agarose gel in TBE buffer at 8°C by using the contour-clamped homogeneous electric field system (Pulsaphor plus; Pharmacia LKB Biotechnology, Uppsala, Sweden). The conditions for electrophoresis were 150 V for 30 h, with pulse times ranging from 5 to 35 s. The DNA bands were visualized by staining the gel with ethidium bromide and were photographed. Bacteriophage lambda DNA concatemers (Gibco BRL, Gaithersburg, Md.) were used as size standards.

β-Lactamase preparation.

The bacteria were grown overnight with shaking in LB broth. The culture broth was diluted 20-fold in new flasks containing 100 ml of LB broth. Following 2 h of incubation at 37°C on an orbital shaker, cefotaxime was added as an inducer to one flask to a final concentration of 0.2 μg per ml (one-fourth the MIC), and incubation was continued for a further 4 h (or longer for the derepressed mutant) until the optical density was 0.7. The cells were harvested, washed with phosphate buffer (pH 7.0) once, and resuspended in the same phosphate buffer. The cells were then sonicated two to four times in 30-s bursts at an amplitude of 12 to 24 μm with intermediate cooling on ice. Major cell debris was removed by ultracentrifugation. The supernatant was used for the β-lactamase assays.

β-Lactamase activity.

The qualitative detection of β-lactamase was performed with the Cefinase disk (BBL, Becton Dickinson Microbiology Systems, Cockeyeville, Md.) according to the manufacturer’s instructions. The quantitative detection of β-lactamase was determined by a direct spectrophotometric assay in 1-cm-light-path cuvettes, with readings recorded at 30-s intervals for 5 min at a wavelength of 262 nm of optimal absorbance (28). Cephalothin at a concentration of 0.1 mM was used as a substrate. The protein concentration was determined by the method of Lowry et al. (22) with the use of bovine serum albumin as the standard. One enzyme unit is defined as the amount of enzyme that hydrolyzes 1 μmol of substrate/min/mg of protein.

RESULTS

In vitro susceptibility.

The results of in vitro susceptibility testing by the disk diffusion method were as follows: The initial isolate from blood (A136) and pus and the second isolate from blood were sensitive to gentamicin, netilmicin, amikacin, cefuroxime, cefotaxime and norfloxacin and resistant to ampicillin, ampicillin-sulbactam, cephalothin, cefoxitin, and cefmetazole. Another A. hydrophila isolate from the tip segment of an arterial catheter (A139), however, was resistant to cefuroxime and cefotaxime. The MICs for A136, A136m, A139 and two standard strains determined with E-test strips are presented in Table 1. The wild-type strain A136 had increased levels of resistance to most cephalosporins and carbapenem compared to those for the two strains from CCRC, although the MICs for A136 were still within the susceptible range. Among three β-lactam–β-lactamase inhibitor combinations, only piperacillin-tazobactam was active against the wild-type strain but inactive against A139. All Aeromonas strains tested were resistant to ticarcillin-clavulanic acid and amoxicillin-clavulanic acid. Tremendous increases in the MICs for A139 were found, ranging from 32- to several hundred-fold for several extended-spectrum β-lactam agents, including piperacillin, cefuroxime, cefotaxime, ceftriaxone, and ceftazidime. However, the MICs of amikacin and ciprofloxacin remained similar.

TABLE 1.

In vitro susceptibilities of five strains of A. hydrophila determined by E-test

Antimicrobial agent	MIC (μg/ml)
	A136	A139	A136m	CCRC 13018	CCRC 13881
Amikacin	2	2	2	2	1.5
Aztreonam	0.016	0.125	0.016	0.016	0.016
Cefaclor	>256	>256	>256	32	0.004
Ciprofloxacin	0.38	0.5	0.38	0.004	0.004
Cefotaxime	0.75	>256	>256	0.047	0.064
Ceftazidime	0.5	16	16	0.25	0.25
Ceftriaxone	2	>256	>256	0.047	0.047
Cefuroxime	2	>256	>256	0.75	0.75
Imipenem	3	12	1.5	0.75	0.75
Piperacillin	3	>256	>256	4	8
Amoxicillin-clavulanic acid	24	32	24	16	24
Piperacillin-tazobactam	1	>256	>256	2	2
Ticarcillin-clavulanic acid	>256	>256	>256	>256	>256

Selection of extended-spectrum β-lactam-resistant mutants in vitro.

The antimicrobial susceptibility of A136m, an in vitro-selected mutant, had patterns similar to those for A139 with two exceptions, aztreonam and imipenem. The MICs of these two antibiotics for A136m were eightfold less than those for A139 but were similar to those for the wild-type strain.

Genomic species identification.

All three phenotypic A. hydrophila strains were able to grow on the dl-lactate medium, which was characteristic for HG1. According to the rRNA gene restriction patterns, all three strains were identical, and major bands of HG1 were also found in these strains, as found in the reference strain of HG1 (data not shown). In addition, 16S rRNA analysis showed that all three strains were most closely related to A. hydrophila, with only one nucleotide difference (G to A) at nucleotide position 471 compared to the sequence of A. hydrophila (GenBank accession no. X87271).

Characterization of genotype.

The results of AP-PCR and PFGE are shown in Fig. 1 and 2, respectively. With primer ERIC-1R, the amplification products of A136, A136m, and A139 were identical and distinctly different from those of the other two clinical isolates and the two strains from the American Type Culture Collection (Fig. 1). Similar results were obtained by using primer ERIC-2R (data not shown). This was further confirmed by PFGE. Chromosomal DNAs were digested with SpeI, and the patterns for three strains (A136, A136m, and A139) were identical, whereas the patterns for reference strains and unrelated clinical strains were different (Fig. 2).

FIG. 1.

AP-PCR profiles for six strains of A. hydrophila obtained with primer ERIC-1R. Lane 1, A136; lane 2, A136m; lane 3, A139; lanes 4 and 5, two clinical strains of A. hydrophila, A2866 and A4252, respectively; lane 6, CCRC 13018; lane 7, CCRC 13881; lane M, 100-bp DNA ladder used as a molecular size standard.

FIG. 2.

PFGE separation of SpeI-digested chromosomal DNA from A. hydrophila. Lane 1, A136; lane 2, A136m; lane 3, A139; lanes 4 and 5, two clinical strains of A. hydrophila, A2866 and A4252, respectively; lane 6, CCRC 13018; lane 7, CCRC 13881; lane M, bacteriophage lambda DNA concatemers.

β-Lactamase activity.

With the Cefinase disk, β-lactamase activity can be detected in the overnight colonies and cell extracts of A139 and A136m but not in those of A136. Three strains (A136, A136m, and A139) were further examined for their β-lactamase activities against cephalothin. Cells extracted from A136, A136m, and A139 showed 0.04, 1.6, and 5.2 U of β-lactamase activity, respectively.

DISCUSSION

Aeromonas species are recognized as important infecting microorganisms for patients with liver cirrhosis and malignancy (13, 15, 17). Although the microorganisms are able to cause nosocomial infections, aeromonads rarely cause infections in burn patients. The colonization of burn wounds often preceded the occurrence of Aeromonas infections. Aquatic exposure is not essential for A. hydrophila infections in burn wounds (5), and suggestive clinical clues include a history of extinguishing the fire with dirty water or rolling in dirt (30). Aeromonas spp. had been shown to be susceptible to tetracycline, chloramphenicol, cephalosporins, aminoglycosides, and fluoroquinolones (19, 25). In the antibiotic era, incremental increases in the levels of resistance of clinical strains of Aeromonas to commonly used antibacterial agents have been observed (18). In the study described in this report, it was verified for the first time that a strain of A. hydrophila, the most common Aeromonas species causing human infections, isolated from a burn patient acquired resistance to extended-spectrum β-lactam agents during antibiotic therapy with cefotaxime but remained susceptible to aminoglycosides and fluoroquinolones.

Like enteric gram-negative bacilli, the emergence of resistance among aeromonads will be accelerated by the clinical use of antibiotics. Although plasmids encoding resistance to older cephalosporins were reported in environmental and clinical Aeromonas isolates (7), the present knowledge of the β-lactam resistance in Aeromonas species focused upon the chromosome-mediated enzymes. In these species, an uncommon character has been discovered: three chromosomally encoded, inducible β-lactamases are concurrently found in A. janaeii and A. salmonicida (10, 35). These enzymes processed hydrolyzing activities for cephalosporin, cloxacillin, and carbapenem, respectively. A recent report further demonstrated that three β-lactamases were simultaneously overexpressed in mutants and that their expression might be coordinated by a common regulatory system (2).

Initially, strain A136, isolated from blood, was susceptible in vitro to cefotaxime; however, under the selective pressure of the antibiotics, most likely cefotaxime, mutant A139 survived and became resistant to cefotaxime. Because of the temporal relationship, the increasing levels of resistance to broad-spectrum antibiotics, and the identical genetic clonality demonstrated by AP-PCR and PFGE, A139 was an in vivo resistant mutant of the wild-type strain. In addition, A139 had 130-fold more cephalothin-hydrolyzing activity than did A136. Therefore, the former was regarded as a derepressed mutant from A136, which overexpressed β-lactamase. A136m was an in vitro mutant selected under the pressure of cefotaxime. In spite of their resistance to cefotaxime, A136m and A139 possessed different susceptibilities to imipenem and aztreonam. In addition, A139 had threefold more cephalothin-hydrolyzing activity than A136m did. The mechanisms for phenotypic variability between those genetically identical strains were unclear. In a recent report, the frequency of in vitro production of resistant mutants in Aeromonas was about 10⁻⁷ to 10⁻⁹, suggesting that a point mutation was responsible for the generation of mutants (35). This appears to be compatible with the theory of coordinating regulation of three chromosomally mediated β-lactamases. The in vitro emergence of derepressed mutants is noted to be universal in A. hydrophila and A. caviae but to be temperature dependent in A. vernoii (35), and this suggests that inducible β-lactam resistance will be a common threat in treating Aeromonas species infections.

Inducible chromosomally encoded β-lactamases mediating resistance to extended-spectrum β-lactam agents have been well characterized in Enterobacter cloacae, Citrobacter freundii, and Pseudomonas aeruginosa (21). Clinically, the frequency of the development of inducible resistance during β-lactam treatment of Enterobacter infections varies from less than 20% to more than 70% (32), and the coadministration of an aminoglycoside with a β-lactam did not prevent the emergence of resistance to β-lactam agents (8). As for the genus Aeromonas, our clinical experience in treating Aeromonas bacteremia indicates that the emergence of resistance resulting from the selective pressure exerted by a β-lactam was a rare event. It is likely that the heavy colonization of Aeromonas species on the ischemic and damaged integument of the burn patient resulted in a high inoculum of bacteria immersed in a subinhibitory concentration of antibiotic, which favors the emergence of resistant mutants. The impact of simultaneous administration of aminoglycosides for the prevention of cephalosporin resistance during therapy with a cephalosporin for Aeromonas infections remains unknown.

AP-PCR has been widely used in the epidemiological typing of many pathogens, including Clostridium difficile (20), Legionella pneumophila (1), Streptococcus mutans (31), P. aeruginosa (11), and Vibrio cholerae (9). However, the application of AP-PCR for A. hydrophila typing has not been reported. In this study, we showed that, despite a small number of isolates, AP-PCR could easily demonstrate the same genetic fingerprints for A136 and A139 and could distinguish the former isolates from four genetically unrelated strains; this result was further confirmed by PFGE. Therefore, it is suggested that AP-PCR is a rapid and simple typing method for genotypic investigations of clinical isolates of A. hydrophila.

In conclusion, the present report demonstrated that the emergence of cephalosporin-resistant mutants from a wild-type strain might cause therapeutic failure in the treatment of invasive Aeromonas infections. Therefore, clinicians should closely observe clinical and microbiological responses to cephalosporins when patients with such infections are treated with these drugs.