INTRODUCTION
Acinetobacter baumannii is an increasingly recognized nosocomial pathogen. Because of its ability to survive on environmental surfaces and its intrinsic resistance to many antimicrobial agents, A. baumannii has become a challenging opportunistic pathogen in many medical centres worldwide. Genetic fingerprinting of A. baumannii is important for infection control studies and epidemiological investigations, and several methods have been employed for this purpose. Pulsed-field gel electrophoresis (PFGE) is often considered the most discriminative method for genotyping A. baumannii [Reference Gouby1, Reference Seifert and Gerner-Smidt2], but it is labour intensive, difficult for inter-laboratory comparisons, and may not be the most appropriate method to examine strains from geographically diverse areas. Amplified fragment length polymorphism analysis has been used to identify Acinetobacter at the species level and to effectively genotype clinical isolates from several geographic areas in Europe [Reference Koeleman3, Reference Spence4]. Two PCR-based fingerprinting methods, random amplified polymorphic DNA (RAPD) analysis and repetitive extragenic palindromic-sequence based PCR (rep-PCR), have also been used to type A. baumannii [Reference Koeleman3, Reference Bou5, Reference Huys6]. Inter-laboratory comparison is possible using standardized methods and reagents [Reference Grundmann7], however, multiple primers may be needed for optimal discrimination by RAPD analysis [Reference Koeleman3]. Multilocus PCR with electrospray ionization mass spectrometry has yielded results comparable with PFGE and has accurately genotyped Acinetobacter spp. from diverse geographic regions [Reference Ecker8, Reference Hujer9], however, the methodology may not be available for many laboratories. A multilocus sequence typing (MLST) scheme, using seven housekeeping genes, has been developed as a typing method for A. baumannii [Reference Bartual10]. The results with MLST have been comparable to PFGE, and MLST allows for the generation of an easily accessible databank necessary for inter-laboratory comparisons. However, the cost of sequencing seven genes per isolate may limit the usefulness of MLST for some laboratories. Sequencing of a single gene (adeB) has also been proposed as a method for typing A. baumannii [Reference Huys11]. Last, ribotyping has been used to identify species [Reference Gerner-Smidt12] and to characterize isolates of A. baumannii from different regions [Reference Landman13–Reference Tognim15], but it has been reported to be not as discriminatory as PFGE [Reference Seifert and Gerner-Smidt2].
In prior surveillance studies involving hospitals in Brooklyn, NY, USA [Reference Landman13, Reference Manikal14], most multidrug-resistant isolates of A. baumannii belonged to a small number of ribotypes but owing to uncertainty of the discriminatory power of this method, two aims were established for this investigation. First, the genetic relatedness of isolates belonging to the predominant ribotypes was further investigated by rep-PCR and PFGE. Second, the sequences of five chromosomal genes possibly contributing to antimicrobial resistance were assessed as a potential typing strategy. For comparative purposes, a group of susceptible isolates was also included for genotypic analysis.
MATERIALS AND METHODS
Bacterial isolates
A total of 43 clinical isolates of A. baumannii, obtained from Brooklyn-wide surveillance studies [Reference Landman16], were examined. These isolates were selected based on their susceptibility patterns to various classes of antimicrobial agents. Nineteen isolates belonged to one ribotype (28-5), and 12 to a second ribotype (35-2), which together accounted for >80% of carbapenem-resistant isolates [Reference Landman13, Reference Quale17]. Five fell into a third ribotype that represented a further 10% of resistant isolates. The remaining seven isolates (of unknown ribotype) were selected based on their susceptibility to most β-lactam antibiotics.
Genetic fingerprinting
Isolates were ribotyped using the Riboprinter™ Microbial Characterization System (Qualicon, Wilmington, DE, USA) using EcoRI as the restriction enzyme, according to the manufacturer's recommendations. Isolates were considered related if they had a similarity coefficient of ⩾90%; results were verified by visual inspection. Isolates were further characterized using rep-PCR with the ERIC-2 primer, as previously described [Reference Grundmann7]. Reactions were carried out using Ready-To-Go RAPD Analysis Beads (Amersham Pharmacia Biotech, Piscataway, NJ, USA) and established PCR conditions [Reference Grundmann7]. Isolates were considered related if there was a 0–1 band difference. Isolates were also analysed by PFGE using SmaI as the restriction enzyme, according to previously described methods [Reference Seifert and Gerner-Smidt2], and interpreted using standard definitions [Reference Tenover18].
DNA sequencing studies
An ongoing study involves the examination and expression of β-lactamases, porins, and efflux pumps in these isolates of A. baumannii. As part of this study, five genes (ampC, ompA, adeR, adeB, and abeM) were investigated. The gene encoding the chromosomal cephalosporinase AmpC was amplified using previously described primers and PCR conditions [Reference Huger19]. PCR products were cloned using the TOPO TA cloning kit (Invitrogen, Carlsbad, CA, USA) and DNA sequenced using the automated fluorescent dye terminator sequencing system (Applied Biosystems, Foster City, CA, USA). Results were compared to the gene encoding ADC-1 [20; GenBank accession no. AJ009979].
Four other genes that may contribute to antimicrobial resistance were also examined. The gene encoding for OmpA, a major porin in the A. baumannii cell membrane, was analysed using the following primers: ompA-F 5′-GGCTTGAGCTTGAACAACAA-3′ and ompA-R 5′-TGTTCAGCTAAAACAGTACGGC-3′; sequences were compared to GenBank no. AY485227. The gene encoding for AdeR, a regulatory gene for the adeABC complex, was amplified with the following primers: adeR-F 5′-AGCGTATGATGAGTTGAAGCA-3′ and adeR-R 5′-AATCCAGCCTTTTTCAATCG-3′; sequencing results were compared to GenBank no. CT025814. An internal 1142 bp product of adeB was amplified with the following primers: adeB-F 5′-CGGAAGGCATGGAGTTTAGT-3′ and adeB-R 5′-CTGCCATTGCCATAAGTTCA-3′. This segment includes 526 bp of the adeB sequence previously used for typing A. baumannii [Reference Huys11]. Sequences for adeB were compared to strain BM4454 (GenBank no. AF370885), as previously described [Reference Huys11]. Finally, the gene encoding for AbeM, an efflux pump protein belonging to the multidrug and toxic compound extrusion family, was amplified using the following primers: abeM-F 5′-TGCAACGCAGTTTCATTTTT-3′ and abeM-R 5′-CGATGTTTCATCGGCTTTTT-3′. Sequences for this gene were compared to GenBank no. AB204810.
Isolates were initially grouped according to ribotype (numerical designation) and then grouped according to rep-PCR type and PFGE group. Isolates related by ribotyping, rep-PCR, and PFGE were designated into a final group, identified by Greek letters.
Nucleotide sequence accession numbers
The nucleotide sequences of ampC in seven isolates were submitted to GenBank and assigned accession numbers EU118260, EU118261, EU118262, EU118263, EU118264, EU118265, EU118266. The sequences of ompA from five isolates have also been submitted: EU332795, EU332796, EU332797, EU332798, and EU332799.
RESULTS
All isolates were characterized by ribotyping, rep-PCR, and PFGE (Fig. 1). They were confirmed as A. baumannii by ribotyping which reliably identifies isolates to the species level [Reference Gerner-Smidt12].
Isolates of ribotype 28-5
Nineteen isolates were assigned to a ribotype pattern (28-5) that encompasses many of the carbapenem-resistant isolates gathered during our surveillance studies [Reference Landman13]. All were resistant to ceftazidime, and 16 were resistant to meropenem. Based on rep-PCR and PFGE patterns, two major subtypes were recognized (groups α and β; Table 1). Group α consisted of eight isolates with similar rep-PCR and PFGE types. Six of these isolates had a distinctive Pro238→Arg change in AmpC, and seven had an Ala insertion between amino acids 242 and 243 (Table 1). Consistent changes were found in OmpA (Gly52→Ser), AdeR (Val120→Ile and Ala136→Val), and AbeM (Val143→Ile) in the eight isolates. One isolate in this group had two additional changes in OmpA (GenBank no. EU332796), and had an ompA sequence that was identical to that of an isolate representative of the South East clone endemic to England [Reference Turton21, Reference Turton22]. While all eight α isolates carried a mutation in adeB at nucleotide 1974, two had additional mutations at nucleotide 1902, and would have been excluded from this group according the typing scheme using this gene [Reference Huys11].
* All had mutations leading to the following changes: S38T;N48E;A89L;S90A;D133E;F134I;D135P;G136D;V137L;N138S;R139Y;G140H;T141N; deletion142RGT144;S145D;A173G;A177F;E180K.
† Both had mutations leading to the following changes: G99A;D110N;S167P;F193S; only the first 209 amino acids could be determined.
‡ Two isolates had non-amplifiable adeR.
Ten isolates in ribotype 28-5 fell into a second distinct rep-PCR group and PFGE group, comprising the β group (Table 1). These 10 isolates shared changes involving OmpA and AdeR seen in the preceding α group. However, none of the isolates had the Val143→Ile change in AbeM. Only one of the 10 isolates possessed the Pro238→Arg amino-acid change and the Ala insertion seen in AmpC of the α group; this isolate may indeed be an intermediary between the α and β groups in this ribotype. All 10 isolates in the β group shared a single mutation in adeB at nucleotide 1974, as previously noted in the α group.
One isolate in ribotype 28-5 had a rep-PCR pattern and changes in OmpA and AdeR characteristic of the β group. However, this isolate had a unique PFGE pattern, and was therefore considered a unique strain. While this isolate possessed the mutation at nucleotide 1974 in adeB seen in the β group, it also harboured two other mutations, and would have been considered unique according the adeB typing scheme [Reference Huys11].
Isolates of ribotype 35-2
Twelve isolates were included in a second ribotype (35-2) that also represents a large percentage of carbapenem-resistant isolates in our city (Table 1). All were resistant to ceftazidime, and six were resistant to meropenem. Eight of these isolates were clearly related by rep-PCR and PFGE and comprised group ζ. These eight isolates also possessed an AmpC that closely resembled ADC-1, and had multiple changes in OmpA (17 amino-acid alterations and a deletion of three amino acids). The sequence of ompA from isolates in this group was identical to the ompA-2 allele from a strain recovered from France in 2001 and characteristic of two widely disseminated European clones [Reference Turton21, Reference Poirel23]. The eight ζ isolates also had distinctive amino-acid changes in AdeR (Leu142→Ile and Val243→Ile) and AbeM (Val171→Ile), and shared 22 mutations involving adeB.
Two identical isolates in ribotype 35-2 were unrelated to the preceding eight (ζ) isolates by both rep-PCR and PFGE. The ampC gene for these two isolates was non-amplifiable; using an internal primer (used for sequencing), the first 628 bp were able to be amplified, and revealed several unique changes (Table 1). These two isolates also had unique changes affecting AdeR and mutations involving adeB. The remaining two isolates in ribotype 35-2 were unique by rep-PCR and PFGE. One isolate had a premature stop codon in AmpC at amino acid 143 and had the signature changes in OmpA, AdeR, and AbeM of the α group. The remaining isolate had unique changes in the four genes.
Isolates of ribotype 66-3
Five isolates were classified as ribotype 66-3 (Table 1), the third largest ribotype among multidrug-resistant isolates. Four of these were resistant to ceftazidime and meropenem, and were related by rep-PCR and PFGE (group λ). Three λ isolates shared multiple distinctive mutations affecting AmpC (including Ala224→Asp, Pro244→Ala, and an insertion of a Leu) and OmpA; however, one of these three had unique mutations affecting adeB. The fourth isolate lacked the changes in AmpC seen in the other three isolates, but did possess the mutations affecting OmpA. Although the four group λ isolates had distinctive changes in AmpC and adeB, their changes in OmpA, AdeR, and AbeM were similar to those observed in the ζ group. The fifth isolate of ribotype 66-3 was unrelated by rep-PCR and PFGE, and had multiple unique changes affecting AmpC and OmpA.
Antimicrobial susceptible isolates
Seven isolates that were susceptible to piperacillin-tazobactam, ceftazidime, and carbapenems were selected for further analysis. Four isolates, gathered from patients at three different hospitals, shared similar ribotype (69-4), rep-PCR, and PFGE patterns (group ν; Table 1). These isolates also shared several unique mutations affecting AmpC, including the following amino-acid changes: Lys150→Gln, Ser167→Pro, Ser257→Lys, and Arg283→Phe. These four isolates had identical OmpA and AbeM, and three had identical AdeR alterations. There were 19 mutations in adeB that were characteristic of the four isolates in this group. The remaining three susceptible isolates each had unique ribotype, rep-PCR, and PFGE patterns. One isolate possessed similarities in OmpA compared to the λ and ζ groups, however, the changes in AmpC, AdeR, and AbeM were dissimilar. The remaining two unique isolates had distinct sequences involving these genes.
DISCUSSION
In prior surveillance studies involving the hospitals in Brooklyn, NY, it was documented that most of the multidrug-resistant A. baumannii isolates belonged to one of three ribotypes [Reference Landman13, Reference Manikal14]. In this report, further genotypic evaluation of isolates belonging to these three ribotypes was performed. It appears the multidrug-resistant A. baumannii isolates in our region fall into major groups with distinct predecessors. The isolates in one ribotype (28-5) possess similar mutations in ompA, adeB, and adeR (and overlapping mutations of ampC), but have two (α and β) subtypes with distinct rep-PCR and PFGE patterns and abeM sequences. While it is possible that ribotyping has less discriminatory power and these two subtypes are unrelated, it seems more likely that these two subtypes are derived from a common ancestor. Most isolates in the second (35-2) and third (66-3) ribotypes also had distinct rep-PCR, and PFGE patterns and ampC and adeB genes. However, these two share multiple changes in OmpA and unique mutations involving abeM and adeR. While these two types (ζ and λ) may be ancestrally related, it is also possible they represent unique strains that have acquired similar genetic changes.
Examination of a small number of β-lactam susceptible isolates did not reveal strains that were closely related to the multidrug-resistant isolates. None of the susceptible isolates appeared to be related to the α and β groups, and the susceptible strains generally possessed unique mutations affecting ampC, ompA, adeB, adeR, and abeM. When compared to the susceptible strains, it appears the multidrug-resistant isolates do not share the same genes associated with antimicrobial resistance.
Our results are analogous to the fingerprinting studies involving A. baumannii isolates originating from Europe, South Africa, and Israel [Reference Turton21, Reference van Dessel24]. As with our New York City isolates, multidrug-resistant nosocomial isolates from Europe and South Africa tend to fall into one of three large ribotypes, and within each ribotype are distinct genotypes (by PFGE) [Reference van Dessel24]. Moreover, remarkable similarities in ompA are observed in isolates from New York City and Europe [Reference Turton21]. Reported clonal groups carrying allele I of ompA include the South East clone, European clone II (corresponding to European ribogroup II [Reference van Dessel24]), T strain, and Oxa-23-clone I found in Europe and Israel [Reference Turton21]. Isolates belonging to our α and β genotypes (that share a single ribotype) have similar ompA sequences as these isolates. Clonal groups carrying allele II of ompA include European clone I (corresponding to European ribogroup I [Reference van Dessel24]), Oxa-23-clone 2, and the French strain AYE-VEB-1 [Reference Turton22]. This allele was also recovered in a distinctive group that included European clone III [Reference Turton21]. Two distinctive genotypic groups from New York City (ζ and λ, with different ribotypes) also share allele II of ompA. Moreover, many of the New York City isolates share sequences of adeB that have been documented in isolates from Europe [Reference Huys11, Reference Magnet, Courvalin and Lambert25]. Given these similarities, it appears the European and American isolates may indeed be ancestrally related. As a consequence of the worldwide distribution of multidrug-resistant A. baumannii, further development of MLST schemes [Reference Bartual10, Reference Turton21] will be essential to determine if isolates from geographically diverse areas share genetic lineages.
ACKNOWLEDGEMENTS
This study was supported by the National Institutes of Health (RO1 AI070246-01A1).
DECLARATION OF INTEREST
None.