Since June 2002, the medical microbiology laboratory of the Funabashi Medical Center, Chiba, Japan, isolated three cefotaxime-resistant A. baumannii strains from inpatients in a neurosurgery ward. In the present study, we characterized the molecular mechanism of cefotaxime resistance in A. baumannii strains associated with a nosocomial infection episode.

Bacterial strains. Three cefotaxime-resistant A. baumannii strains, FM0209680, FM0300106, and FM0301433, were isolated from cultures of transtracheal aspirates from three different inpatients in the neurosurgery ward of the Funabashi Medical Center since June 2002. Biochemical identification of the isolates was performed with the API 20NE system (bioMérieux, Marcy l'Etoile, France) combined with a complementary test for the ability to grow at 44°C. Alternatively, sequencing of the 16S rRNA gene was performed by the method described by Sasaki et al. (46). β-Lactamase testing was performed on the basis of acidometry by using a commercial product (P/Case test; Nissui Pharmaceutical, Tokyo, Japan). The bacterial strains were stored in Casitone medium (Eiken Chemical, Tokyo, Japan) at room temperature until they were used.

MIC determinations. MICs were determined by a microdilution broth method with a WalkAway-96 SI system (NEG Combo 5J and NEG MIC 5J panels; Dade Behring, Sacramento, Calif.) with an inoculum of 10⁴ CFU per well. Susceptibility categories were determined according to the criteria of the National Committee for Clinical Laboratory Standards (35).

The ESBL plus panel (Dade Behring) with an inoculum of 10⁴ CFU per well was used as complementary test for MIC measurements. The panel was incubated for 18 h at 35°C, and then the results were assessed visually.

β-Lactamase study. The double-disk synergy test was used to screen for extended-spectrum β-lactamase (ESBL)-producing strains. Kirby-Bauer disks containing cefotaxime (30 μg), ceftazidime (30 μg), aztreonam (30 μg), and amoxicillin-clavulanic acid (20 μg-10 μg) for tests on Mueller-Hinton agar (BBL Microbiology Systems, Cockeysville, Md.) were obtained from Nissui Pharmaceuticals. The distance between the disks was adjusted so that synergy could be detected correctly (48), as shown in Fig. 1.

FIG. 1. Double-disk synergy test with CTX-M-2 ESBL-producing A. baumannii FM0209680. Disks: CTX, cefotaxime at 30 μg; CAZ, ceftazidime at 30 μg; AZT, aztreonam at 30 μg; CVA·AMPC, clavulanic acid at 10 μg (more ...)

The inducibility of AmpC β-lactamase production was tested by disk antagonism tests (30). Disks containing an inducing agent, cefoxitin at 30 μg or ceftazidime at 30 μg, were placed on Mueller-Hinton agar plates (BBL Microbiology Systems). The distance between the disks was adjusted so that the blunting phenomenon of the ceftazidime zone could be detected correctly.

PCR amplification and bla gene sequencing. The oligonucleotides used as primers for amplification and sequencing are shown in Table 1. A search for the bla_TEM (49), bla_SHV (41), bla_CTX-M-1 (3), bla_CTX-M-2 (24), and bla_CTX-M-9 (32) genes in the clinical isolates was performed by PCR amplification, as described previously (34). Detection of the ampC gene was performed as described by Bou and Martínez-Beltrán (7).

TABLE 1. Nucleotide sequences of the oligonucleotides used for PCR amplification and DNA sequencing

A resistance plasmid carrying the bla_CTX-M-2 gene was prepared from the Escherichia coli CSH2 (F⁻ metB, with resistance to both nalidixic acid and rifampin [18]) transconjugant. The 902-bp fragments corresponding to the main frame of the bla_CTX-M-2 gene were amplified by PCR with a set of primers (43) and were used for sequencing. The PCR products were labeled with fluorescent materials by using a BigDye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems, Foster City, Calif.). Both strands of the DNA sequences were analyzed with an ABI PRISM 377 XL sequencer analyzer (Applied Biosystems). The nucleotide and deduced amino acid sequences were analyzed and compared with sequences in a database by using the FASTA analysis programs available at the National Institute of Genetics (Mishima, Japan) website (http://www.nig.ac.jp/section/service.html).

Plasmid preparation and restriction endonuclease analysis. The resident plasmids of three A. baumannii strains and previously isolated Proteus mirabilis strains (34) were prepared by a conventional protocol (38). The plasmids were digested with EcoRI, and the restriction fragment profiles were compared by agarose gel electrophoresis.

Automated ribotyping and analysis. Because pulsed-field gel electrophoresis analysis failed to give clear results due to the persistent degradation of DNA, we used automated ribotyping for analysis of the genetic relatedness of the three clinical isolates. Ribotyping was performed with the restriction enzyme EcoRI and a RiboPrinter (Qualicon Inc., Wilmington, Del.), as described previously (13, 33). Briefly, a single colony from a 5% sheep blood agar plate was suspended in a sample buffer and heat treated at 80°C for 15 min. After addition of lysis buffer to release the DNA, the sample was loaded into the RiboPrinter system. Further processing including EcoRI digestion, agarose gel separation, transfer to a nylon membrane, and hybridization with a chemiluminescence-labeled DNA probe containing the rRNA operon from E. coli, which was carried out by automated instruments in 8 h. Output ribotype patterns with similarity coefficients of >0.93 were considered a single ribogroup and were automatically given a code number. Computerized ribotypes were finally exported for analysis in text files and imported into BioNumerics software (version 3.0; Applied Maths, Sint-Martens-Latem, Belgium) by using the Qualicon macro. Clustering analysis was performed by the unweighted pair group with arithmetic averages method (48), based on the Dice coefficient (16) for band matching, with a position tolerance setting of 1.0% (default values are 1% position tolerance and 0.5% optimization). Bands for analysis with the Dice coefficient were assigned manually, according to densitometric curves and the accompanying hard-copy photograph.

Identification of A. baumannii isolates and their clinical associations. Strain FM0209680 was derived from the culture of a transtracheal aspirate from a 24-year-old male patient (case patient 1) from whom the first cefotaxime-resistant A. baumannii strain was isolated on 19 June 2002. This patient was admitted to the neurosurgery ward on 14 June with a pneumonia-related condition and had a history as a patient in the same ward in October 1999 due to hemorrhaging of the brain caused by underlying cerebral arteriovenous malformation. Until his hospital discharge, on 6 January 2003, cefotaxime-resistant A. baumannii isolates were recurrently and predominantly detected in cultures of transtracheal aspirates from the patient. These data point out the great difficulty in eradicating the organisms. Strain FM0300106 was then recovered from the culture of a transtracheal aspirate from a 62-year-old female patient (case patient 2) who had been hospitalized in the neurosurgery ward due to brainstem infarction on 24 November 2002. The cefotaxime-resistant A. baumannii strain was first and predominantly detected in the specimen for culture taken on 6 January 2003, although it had not been found in a total of six samples for culture taken up to that time. Strain FM0301433 was recovered from the culture of a transtracheal aspirate from a 65-year-old male patient (case patient 3) who had been hospitalized in the neurosurgery ward due to cerebral hemorrhage on 27 December 2002. As was noted for case patient 2, the cefotaxime-resistant A. baumannii strain was first detected predominantly in the specimen for culture taken on 12 February 2003, although it had not been found in a total of seven specimens for culture obtained up to that time. In both case patients 2 and 3, the eradication of the cefotaxime-resistant A. baumannii strains was achieved within 3 weeks after the strains were first detected. Imipenem therapy was used for case patient 2 because a complication of pneumonia was a serious concern and the organisms disappeared. The cefotaxime-resistant A. baumannii strain was isolated from case patient 3 during prophylactic antimicrobial therapy with gentamicin and cefazolin. Administration of cefazolin may have induced colonization with this organism in case patient 3, in view of the fact that strain FM0301433 was resistant to cefazolin. The prophylactic therapy was then discontinued because no pneumonic clinical findings were noted, and, fortunately, the organism disappeared after 3 weeks.

The history of antimicrobial therapy before isolation of cefotaxime-resistant A. baumannii was untraceable for case patient 1 because this patient was transferred from another hospital to the Funabashi Medical Center. Sultamicillin, cefazolin, ceftriaxone, imipenem-cilastatin, cefoperazone-sulbactam, levofloxacin, clindamycin, gentamicin, and arbekacin were used prior to the isolation of A. baumannii in case patients 2 and 3. The sequence of isolation of the three cefotaxime-resistant A. baumannii strains, as well as the antimicrobial agents prescribed for the three case patients, are summarized in Fig. 2.

FIG. 2. Sequence of isolation of cefotaxime-resistant A. baumannii strains as well as the antimicrobial agents prescribed for three patients. Rectangles, periods of hospitalization; black triangles, times of isolation of cefotaxime-resistant A. baumannii from (more ...)

Three isolates, FM0209680, FM0300106, and FM0301433, which shared the same biochemical profile code, 0041073, by use of the API NE20 system, combined with a positive result for the ability to grow at 44°C as a complementary test, were eventually identified as A. baumannii. The 16S rRNA nucleotide sequences of these microorganisms shared 99% identity with that of A. baumannii type strain ATCC 19606 deposited in the EMBL database (accession number Z93435).

Antibiotic susceptibilities. The antibiotic susceptibility profiles of all three A. baumannii strains are shown in Table 2. These strains were broadly resistant to penicillins, cephalosporins, cephamycins, and monobactam but were highly susceptible to carbapenems. For all strains, addition of 4 μg of clavulanic acid per ml to the medium for antimicrobial susceptibility testing significantly decreased the MICs of cefotaxime from >128 μg/ml to 2 to 8 μg/ml, while the ceftazidime MICs decreased modestly from 8 μg/ml to 1 to 4 μg/ml.

TABLE 2. Antibiotic susceptibilities of A. baumannii clinical isolates

β-Lactamase production. The production of β-lactamase was initially judged by the P/Case test (37), which can discriminate between penicillinase (with 1.5 mg of benzylpenicillin as the substrate) and cephalosporinase (with 1.5 mg of cephaloridine and 75 μg of clavulanic acid as the substrate). According to the instructions of the manufacturer, the bacterial colonies are rubbed onto two indicator disks containing benzylpenicillin and cephaloridine with clavulanic acid, respectively. When the strain produces any class A β-lactamase, such as TEM- or SHV-derived β-lactamases or CTX-M-type enzymes, a disk containing benzylpenicillin turns yellow, while the other disk containing cephaloridine with clavulanic acid remains purple, because hydrolysis of cephaloridine by the class A β-lactamases is blocked in the presence of clavulanic acid. If the strain produces class C β-lactamases, the disk containing cephaloridine with clavulanic acid turns yellow, because these enzymes are not blocked by clavulanic acid. The results of the test suggested that all three A. baumannii strains produce both class A and class C β-lactamases.

The isolates suggested to produce class A β-lactamases were investigated further to determine whether they produce ESBLs by the double-disk synergy test. The presence of an ESBL was suggested by the synergy between amoxicillin-clavulanic acid and cefotaxime, ceftazidime, and aztreonam against all three strains. However, growth of scattered colonies was also noted within the expanded growth-inhibitory zones around the cefotaxime and aztreonam disks (Fig. 1). Enhanced production of β-lactamases as well as alterations in membrane protein composition might be implicated in this phenomenon. A significant decrease in the MICs of cefotaxime from >128 μg/ml to 2 to 8 μg/ml in the presence of 4 μg of clavulanic acid per ml (Table 2) suggested the production of some CTX-M-type class A β-lactamases by these isolates.

As described above, the P/Case test suggested that these isolates produce an AmpC cephalosporinase, and the ampC gene (7) was actually detected as a 662-bp PCR amplicon in all three strains, although no obvious inducibility of AmpC β-lactamases was detected in any of them by cefoxitin-ceftazidime disk antagonism tests (data not shown).

No amplicons for the bla_TEM, bla_SHV, bla_CTX-M-1, and bla_CTX-M-9 genes were detected in any of the strains except strain FM0301433, which produced an 824-bp amplicon for the bla_TEM gene.

Sequencing of bla_CTX-M-2 gene and plasmid analysis. In a preliminary search by PCR, 780-bp amplification products specific for the bla_CTX-M-2 genes were observed in all three A. baumannii strains. The nucleotide sequence of the 902-bp amplicon encoding a full-length bla gene, which was amplified with a set of PCR primers (43), was exactly the same as that of the bla_CTX-M-2 gene (GenBank accession number X92507), suggesting that all these strains certainly harbor the bla_CTX-M-2 gene. The nucleotide sequences of the bla_CTX-M-2 gene harbored by the A. baumannii strains were exactly the same as that of a P. mirabilis strain previously isolated in the urology ward of the Funabashi Medical Center (34). Moreover, the bla_CTX-M-2-bearing plasmids prepared from three A. baumannii strains were digested into four different EcoRI fragments of approximately 6.6, 6.4, 4.1, and 3.8 kb (data not shown). The restriction fragment profiles of the plasmids were exactly the same as those of plasmids previously isolated from P. mirabilis strains (data not shown).

Ribotyping. In order to investigate the genetic relationship among FM0209680, FM0300106, and FM0301433, the isolates were analyzed by automated ribotyping. Figure 3 shows the results of a clustering analysis performed with the ribotype patterns of the three isolates and some collection strains of the genus Acinetobacter, including strains with patterns related to those of European outbreak clones I and II (11, 17) and a newly described clone, clone III (S. Brisse, personal communication). The three A. baumannii strains shared the same fingerprint pattern, indicating a clonal lineage. Moreover, the patterns of the three strains showed a high degree of similarity (92%) with those of a clone II strain. Nevertheless, the patterns of all three strains were different from those of the European clones, which, it has been suggested (11), are probably widespread worldwide.

FIG. 3. EcoRI ribotypes of the three cefotaxime-resistant A. baumannii clinical isolates. A comparative analysis of these isolates with some collection strains of the genus Acinetobacter, including European outbreak clones I, II, and III of A. baumannii, is shown. (more ...)

This report describes the probable nosocomial transmission of CTX-M-2 β-lactamase-producing A. baumannii strains in a neurosurgery ward, where the majority of inpatients are represented by immunocompromised hosts who have undergone respiratory manipulations, such as a tracheotomy or mechanical ventilation, and where careful and strict monitoring is provided. For case patient 1, a CTX-M-2-producing A. baumannii strain was first isolated on day 5 after admission. During the 6-month hospitalization of this patient, cefotaxime-resistant A. baumannii strains were detected continuously, despite the administration of imipenem, ceftazidime, and levofloxacin, which consequently failed to eradicate the microorganism. For case patients 2 and 3, CTX-M-2-producing A. baumannii strains were detected at the sixth week after admission, and the patients were then colonized for 3 weeks. Imipenem treatment was used to eradicate A. baumannii from case patient 2 because a high risk of pneumonia was a serious concern in this patient. On the other hand, prophylactic therapy was discontinued in case patient 3 because no pneumonic clinical findings appeared and the microorganism consequently disappeared. As shown in Fig. 2, it seems very likely that an A. baumannii strain harboring the bla_CTX-M-2 gene was persistently transmitted from case patient 1 to case patient 3 via case patient 2 over a period of 9 months in the same neurosurgery ward.

The three A. baumannii strains, strains FM0209680, FM0300106, and FM0301433, shared the same ribotype pattern, suggesting a clonal lineage. For all three patients, multiple bacterial species other than A. baumannii were frequently coisolated throughout the periods of hospitalization, but fortunately, none was a CTX-M-2 producer. The genes for CTX-M-type β-lactamases are often encoded on plasmids and are easily transmitted among gram-negative bacilli, including members of the family Enterobacteriaceae (5, 38), and glucose nonfermenters, such as Pseudomonas aeruginosa (8). Since no CTX-M-2 ESBL-producing bacterial isolates, including A. baumannii, were previously detected in the neurosurgery ward, it is likely that case patient 1 already carried a CTX-M-2-producing A. baumannii isolate before his rehospitalization. However, in the Funabashi Medical Center, a previous outbreak of nosocomial infections in the urology ward, located on a different floor from the neurosurgery ward, involved 19 inpatients and was related to CTX-M-2-producing P. mirabilis (34). Agarose gel electrophoresis demonstrated that the bla_CTX-M-2-bearing plasmids found in both bacterial species had the same EcoRI digestion pattern. Therefore, this finding strongly suggests the possibility of the lateral transfer of the plasmids carrying the bla_CTX-M-2 gene between P. mirabilis and A. baumannii in the clinical setting described here.

A. baumannii strains which produce plasmid-mediated CTX-M-type β-lactamases or TEM- and SHV-derived ESBLs are still very rare. The class A β-lactamases PER-1 and VEB-1, which are genetically distant from the predominant TEM- and SHV-derived ESBLs and CTX-M-type enzymes, were first reported among nosocomial A. baumannii isolates in Turkish and French hospitals (41, 52). Similar strains have subsequently been isolated in France (40) and Korea (26). Among the CTX-M-type β-lactamases, the CTX-M-5 gene has been found in A. baumannii (GenBank accession number AF462635); however, no details have been published to date. A. baumannii usually produces a chromosomally encoded AmpC cephalosporinase, but this kind of enzyme generally cannot hydrolyze oxyiminocephalosporins, cephamycins, or carbapenems. Therefore, acquisition of plasmid-mediated enzymes with broad- and extended-spectrum substrate specificities could well allow this bacterial species to survive in present clinical environments. A. baumannii has become one of the major groups of bacteria that causes respiratory infections, especially among patients in intensive care units. Thus, the emergence of CTX-M-2-producing A. baumannii strains could become a serious clinical problem in Japan, because CTX-M-2-producing as well as CTX-M-1-producing microorganisms have already been frequently found among clinical isolates from humans (5, 25, 55) and cattle (47).

Early outbreaks due to β-lactamase-producing strains resulted in epidemics caused by isolates with only a single β-lactamase. However, more complex situations involving multiple-β-lactamase-producing microorganisms have been documented in recent nosocomial outbreaks (7, 9, 10, 56). The three A. baumannii strains reported in the present study produced two β-lactamases, the CTX-M-2 enzyme as well as the AmpC β-lactamase. Moreover, the third strain also produced a TEM-type penicillinase. In the present study, the induction of β-lactamase production in A. baumannii strains was suggested by the growth of colonies within the growth-inhibitory zones around cefotaxime and aztreonam disks. Such inducible class C β-lactamase production by A. baumannii was not detected by previously reported disk antagonism tests with cefoxitin and ceftazidime disks (30), although the presence of the ampC gene was confirmed by PCR. Thus, it is likely that some of the colonies appearing in the growth-inhibitory zone around the cefotaxime and aztreonam disks might be mainly constitutive CTX-M-2 producers. It is also possible that these colonies corresponded to AmpC-hyperproducing mutants in which the regulatory system of AmpC production was disrupted through mutations in the ampC promoter region or in genes implicated in the regulation of ApmC production, such as ampR, ampG, and ampD (28). These mutations might be easily induced in A. baumannii in the presence of cefotaxime, aztreonam, and clavulanic acid. Moreover, all three A. baumannii strains characterized in this study demonstrated resistance to cephamycins, such as cefmetazole, cefoxitin, and cefotetan. This may depend on a continuous and moderate level of production of chromosomal AmpC cephalosporinase as well as alterations in bacterial membrane permeability (44). Functional analysis of the Acinetobacter AmpC cephalosporinase and further characterization of the nature of the bacterial cells appearing around cefotaxime and aztreonam disks will be continued.

A. baumannii tends to accept foreign DNA as a means to adapt to environments hazardous for bacterial growth (15, 36) and has a high survival capacity in any environment with damp conditions and low temperatures. This inherent ability of A. baumannii may facilitate the development of a multidrug resistance profile with the widespread use of antimicrobial agents through the selection of strains with ever accumulating antibiotic resistance profiles in hospital environments. Thus, the emergence of CTX-M-2-producing A. baumannii strains not only may confer the potential for epidemics but also could serve as reservoirs for the plasmid-dependent CTX-M-2 enzyme for dispersal of the bla_CTX-M-2 gene among different gram-negative bacterial species. Indeed, although it might be difficult to detect the coproduction of class A β-lactamases like CTX-M-type enzymes in intrinsically cephalosporin-resistant A. baumannii, knowledge of the presence of these burdensome strains would seem to be very important in achieving suitable infection control measures in hospitals on a daily basis.

We are grateful to the following for their contributions to infection control at the Funabashi Medical Center: T. Karasawa, Department of Neurosurgery; Y. Yasutoshi, H. Tabeta, H. Kamii, and M. Shoji, Infection Control Team; and Y. Saito, N. Suwa, and M. Toyama, Medical Microbiology Laboratory. We also thank S. Brisse, Unité Biodiversité des Bactéries Pathogèenes Emergentes, Institut Pasteur, for allowing us access to the Acinetobacter ribotypes patterns from the Network for Automated Bacterial Strain Fingerprinting in Europe (GENE project, European Union Fifth Framework Programme).

This work was supported in part by grants H15-Shinkou-9 and H15-Shinkou-10 from the Ministry of Health, Labor and Welfare of Japan.