Molecular characterization and antimicrobial susceptibility of Acinetobacter baumannii isolates obtained from two hospital outbreaks in Los Angeles County, California, USA

Background Antibiotic resistant strains of Acinetobacter baumannii have been responsible for an increasing number of nosocomial infections including bacteremia and ventilator-associated pneumonia. In this study, we analyzed 38 isolates of A. baumannii obtained from two hospital outbreaks in Los Angeles County for the molecular epidemiology, antimicrobial susceptibility and resistance determinants. Methods Pulsed field gel electrophoresis, tri-locus multiplex PCR and multi-locus sequence typing (Pasteur scheme) were used to examine clonal relationships of the outbreak isolates. Broth microdilution method was used to determine antimicrobial susceptibility of these isolates. PCR and subsequent DNA sequencing were employed to characterize antibiotic resistance genetic determinants. Results Trilocus multiplex PCR showed these isolates belong to Global Clones I and II, which were confirmed to ST1 and ST2, respectively, by multi-locus sequence typing. Pulsed field gel electrophoresis analysis identified two clonal clusters, one with 20 isolates (Global Clone I) and the other with nine (Global Clone II), which dominated the two outbreaks. Antimicrobial susceptibility testing using 14 antibiotics indicated that all isolates were resistant to antibiotics belonging to four or more categories of antimicrobial agents. In particular, over three fourth of 38 isolates were found to be resistant to both imipenem and meropenem. Additionally, all isolates were found to be resistant to piperacillin, four cephalosporin antibiotics, ciprofloxacin and levofloxacin. Resistance phenotypes of these strains to fluoroquinolones were correlated with point mutations in gyrA and parC genes that render reduced affinity to target proteins. ISAba1 was detected immediately upstream of the blaOXA–23 gene present in those isolates that were found to be resistant to both carbapenems. Class 1 integron-associated resistance gene cassettes appear to contribute to resistance to aminoglycoside antibiotics. Conclusion The two outbreaks were found to be dominated by two clonal clusters of A. baumannii belonging to MLST ST1 and ST2. All isolates were resistant to antibiotics of at least four categories of antimicrobial agents, and their antimicrobial susceptibility profiles correlate well with genetic determinants. The results of this study will facilitate our understanding of the molecular epidemiology, antimicrobial susceptibility and mechanisms of resistance of A. baumannii obtained from Los Angeles hospitals.


(Continued from previous page)
Conclusion: The two outbreaks were found to be dominated by two clonal clusters of A. baumannii belonging to MLST ST1 and ST2. All isolates were resistant to antibiotics of at least four categories of antimicrobial agents, and their antimicrobial susceptibility profiles correlate well with genetic determinants. The results of this study will facilitate our understanding of the molecular epidemiology, antimicrobial susceptibility and mechanisms of resistance of A. baumannii obtained from Los Angeles hospitals.
A few years ago, we examined the phenotypic and molecular characteristics of 20 representative outbreak isolates of A. baumannii obtained from Los Angeles County (LAC) hospitals [14]. Subsequently, in vivo virulence screens identified among these isolates a hypervirulent strain (LAC-4) which was used to develop a mouse model of A. baumannii-associated pneumonia [23]. Recently, we reported the complete genome of LAC-4 [24]. In this study, we determined antimicrobial susceptibility and performed detailed molecular analyses of 38 additional isolates of A. baumannii derived from two hospital outbreaks in Los Angeles County. Our report provides critical insight into the development of antimicrobial drug resistance, clonal dissemination and evolution of clinical isolates of A. baumannii in LA County hospitals.

Epidemiology and definitions of case and control patients/isolates
A. baumannii isolates were collected as part of two separate healthcare-associated outbreak investigations that occurred in two Los Angeles County hospitals in 2007 and 2009 respectively. Hospital A (Outbreak X) is a large tertiary care teaching facility that provides general and specialized medical services. The outbreak occurred in the burn unit, a six-bed unit with two rooms located across the hall from each other. Hospital B (Outbreak Y) is a medium sized tertiary care facility that provides general and specialized medical services. The outbreak occurred in the medical intensive care unit. In both outbreaks isolates from case patients (thus the isolates are called case isolates) were cultured after the onset of new symptoms, e.g. fever, increased sputum. Control patients had A. baumannii cultures during the outbreak period from the same facility and, based on epidemiologic data, these cultures were determined to be unrelated to the outbreak (thus these isolates are called control isolates). Control isolates were used as a baseline for comparison of current strains circulating within the facility. Environmental cultures were collected from multiple unit surfaces of Hospital B, including patient room sinks, faucets and door handles as well as ventilator supplies and equipment. Environmental Cultures were not collected from Hospital A. A. baumannii isolates are shown in Table 1.

Additional bacterial strains
Quality control strains used in antimicrobial susceptibility testing [Escherichia coli (ATCC #25922) and Pseudomonas aeruginosa (ATCC #27853)] were purchased from American Type Culture Collection (ATCC, Manassas, VA). Additionally, A. baumannii strains AYE [12] and CCF-9 [25] were used as positive controls for Global Clones I and II (formerly known as International Clones I and II), respectively. All strains were stored in 20 % glycerol at -80°C.

Species-level confirmation of A. baumannii isolates
All isolates were previously determined to be A. baumannii by hospital clinical laboratories. Species of these isolates were further confirmed based on sequence analysis of 16S-23S rRNA gene intergenic spacer (ITS) regions using modified methods [25] based on Chang and coworkers [26].

Pulsed field gel electrophoresis (PFGE)
Bacterial genomic DNA plugs were made from A. baumannii isolates according to PulseNet USA standardized laboratory protocol (http://www.cdc.gov/pulsenet/pathogens/index.html) for PFGE of E. coli O157:H7, Shigella, and Salmonella with following modifications. Genomic DNAs (in agarose plugs) released from lysed cells of all isolates were digested with ApaI (40 U/μl) for 5 h at 30°C . Samples were run on 1 % SeaKem Gold pulsed field agarose gels in 0.5 × TBE using a Bio-Rad CHEF-Mapper XA Electrophoresis System in 0.5 × TBE for 18 h with an initial pulse at 5 s and final pulse at 13 s.  [14,25,27]. The genomic DNA banding patterns of the various isolates generated by PFGE were analyzed for clonal relationships using BioNumerics software, version 5.1 (Applied Maths, Austin, TX), using the UPGMA (unweighted pair group method with arithmetic mean) clustering method and the Dice similarity coefficient with 1.5 % band matching tolerance and 0.5 % optimization.

Tri-locus multiplex PCR
Tri-locus multiplex PCR was performed using genomic DNA from these isolates to determine clonal relationships (Global Clones I, II or III). Briefly, six pairs of PCR primers designed by Turton and colleagues [28] were used to perform two sets of multiplex PCR amplifications of ompA, csuE and bla OXA-51-like gene fragments for each of the isolates using 5 Prime Taq polymerase MasterMix (Gaithersburg, MD). PCR parameters consisted of an initial denaturation at 94°C for 3 min, 30 cycles of denaturation at 94°C for 45 s, annealing at 57°C for 45 s, extension at 72°C for 1 min, and a final extension at 72°C for 5 min. The isolates were designated as A. baumannii Global Clones I, or II based on criteria established for Group 2 and Group 1 strains, respectively [28].

Multi-locus sequence typing (MLST)
The MLST scheme described by Diancourt and coworkers [29] was performed and the scheme defined as Pasteur's MLST scheme is now hosted at http:// pubmlst.org/abaumannii/ site of PubMLST database. Briefly, the sequences of internal fragments of seven house-keeping genes cpn60 (encoding 60-kDa chaperonin), fusA (protein elongation factor EF-G), gltA (the citrate synthase), pyrG (CTP synthase), recA (homologous recombination factor), rplB (50S ribosomal proteins L2) and rpoB (RNA polymerase subunit B) were determined. PCR amplification was performed in separate reactions of a final volume of 50 μl containing 20 μl Taq polymerase MasterMix, 1 μl forward and reverse primers (10 μM) and 2 μl template DNA (10 ng/μl). PCR amplification was carried out in the GeneAmp 9700 system without a mineral oil overlay. Thermocycler parameters included an initial melt at 94°C for 2 min, followed by 35

Antimicrobial susceptibility testing
Antimicrobial susceptibility of the A. baumannii isolates were determined using the broth microdilution protocols of Clinical Laboratory Standards Institute [30] against a total of 14 known antibiotics according to methods described previously [14]. Cefotaxime, ceftazidime, ceftriaxone, gentamicin, levofloxacin, ciprofloxacin, piperacillin and polymyxin B were purchased from Sigma-Aldrich (St. Louis, MO). Meropenem was from US Pharmacopeia (Rockville, MD). Amikacin, cefepime, gatifloxacin, imipenem and tobramycin were purchased from Fisher Scientific (Tustin, CA).

Sequencing of QRDRs of gyrA and parC genes
The quinolone resistance determinant regions (QRDRs) of A. baumannii genes gyrA and parC were sequenced based on the methods described previously [14], with the exception of a new forward PCR/sequencing primer (5′-GTGTGCTTTATGCCATGCAC-3′) for the gyrA QRDR. Sequence comparisons with the amino acid sequences of wild type A. baumannii GyrA (accession No. X82165) and ParC (accession No. X95819) QRDR regions [31,32] were made to identify mutated amino acids.

Detection of class 1 integrons
The presence or absence of Class 1 integrons in all 38 hospital isolates was determined by PCR using previously reported primers specific for the intI1 genes: intI1L primer, 5′-ACATGTGATGGCGACGCACGA-3′; intI1R primer, 5′-ATTTCTGTCCTGGCTGGCGA-3′ [33]. Amplification was performed using genomic DNA as template as described previous for the QRDR of the gyrA and parC genes [14], with the following parameters: an initial template denaturation at 94°C for 2 min; 36 cycles of 1 min denaturation at 95°C, 30 s of annealing at 53°C, and 2 min of extension at 72°C; final extension at 72°C for 10 min. Similarly, primers specific to the 5′ and 3′conserved segments (CS) of class 1 integrons (5′ CS primer, 5′-GGCATCCAAGCAGCAAG-3′; 3′ CS primer, 5′-AAGCAGACTTGACCTGA-3′) were also chosen to amplify the variable regions, as previously described by Levesque and coworkers [34]. After the PCR products were confirmed by agarose gel electrophoresis analysis, successful amplicons were sequenced as previously described for the QRDR of gyrA and parC genes. DNA sequences were analyzed using NCBI BLAST program to identify the entities of the genes within the variable regions.

Carbapenamase gene detection
The presence of four prevalent OXA carbapenemase genes (bla OXA-23-like , bla OXA-24-like , bla OXA-51-like and bla OXA-58-like ) for all isolates was determined using established procedures described by Woodford and colleagues [35]. Briefly, the multiplex PCR reactions using 4 pairs of published primers [35] were performed in the GeneAmp 9700 system. The resulting agarose gel electrophoresis revealed the presence or absence of a given bla OXA allele based on the predicted amplicon sizes [35].

Detection of ISAba1
These hospital isolates were subsequently tested for the presence of ISAba1 element, using procedures described by Segal and coworkers [36]. Additionally, for those isolates that were confirmed to harbor bla OXA-23 , PCR mapping experiments were performed to determine whether there was an ISAba1 element upstream of the bla OXA-23 gene using the ISAba1 forward primer and the bla OXA-23-like gene reverse primer, 20 ng template DNA, with the following PCR parameters: an initial denaturing step at 94°C for 5 min, followed by 30 amplification cycles of 94°C for 25 s, 52°C for 40 s and 72°C for 50 s, with a final extension step at 72°C for 6 min.

Results and discussion
Epidemiology All isolates were confirmed as belonging to A. baumannii species based on sequence analysis of the 16S-23S rRNA intergenic spacer (ITS) region (data not shown). Outbreak X (n = 6 cases) and Outbreak Y (n = 8 cases) were defined as hospital-acquired A. baumannii infections clustered in time and location. Thirty-five A. baumannii clinical isolates, obtained from two large general acute care hospitals (>300 beds) as part of two respective outbreak investigations, were collected and analyzed from twenty-one unique patients. Specimen sites included respiratory (resp, sputum, endotracheal tube, and bronchoalveolar lavage) (n = 19), blood (n = 7), catheter tip (n = 4), wound (n = 4), and urine (n = 1) ( Table 1). Three additional isolates were collected from the hospital environment in Outbreak Y (sink faucet, towel dispenser and ventilator tubing in the same room). Outbreak X occurred in the burn intensive care unit. All six cases (i.e., case patients) were admitted for severe thermal injuries. Outbreak Y occurred in the critical care unit of another hospital and involved cases who were admitted for a variety of medical diagnoses including respiratory failure, altered consciousness, dehydration, abdominal distress, end stage renal disease, and stroke. The remaining six patients in Outbreak X and the remaining patient in Outbreak Y were considered controls (i.e., control patients). The control patients in Outbreak X were admitted to intensive care units (n = 4) or closed monitored units (n = 2), a step-down unit from intensive care, for conditions that include wounds, cancer, trauma, sepsis, and necrosis. The control patient in Outbreak Y was admitted to the telemetry unit for fever and hypotension. For the twenty-one total patients, the average age was 56 years (ranging from 19 to 85 years); 76 % (n = 16) were male and 24 % (n = 5) were female; 48 % of patients died (n = 10) with cause of death determined to be related to A. baumannii in three cases. The average length of stay prior to date of first positive A. baumannii culture was 25 days (range 1-109 days).

Pulsed field gel electrophoresis analysis
Pulsed field gel electrophoresis demonstrated relative relatedness of case isolates as compared to control isolates in each outbreak respectively. In Outbreak X, eight isolates from four cases (Cases 1-4) had indistinguishable isolate patterns and were designated Type X-A ("X" was added as a prefix to distinguish the PFGE types from those of Outbreak Y), while one isolate from Case 4 was designated Type X-A1 due to 1 band difference (Table 1 and Fig. 1 PFGE). The five isolates from the remaining two cases (Case 5 and Case 6) were designated Type X-A3, were indistinguishable from each other and had three band differences from the major pattern of the other cases (Table 1 and Fig. 1). Among the control isolates, two isolates of Control 1 and one isolate of Control 3 showed three band differences from the major pattern of the cases and were designated Type X-A2 and Type X-A5, respectively. Two isolates of Control 2 and the other isolate of Control 3 had four band differences from the major pattern and were designated Type X-A4 and Type X-A6, respectively (Table 1 and Fig. 1 PFGE). The isolates of the remaining three controls had greater than seven band differences from the major pattern of the cases, thus are considered not clonally related to those described above and were designated Types X-B, X-B1, X-C and X-D (Table 1 and Fig. 1). In Outbreak Y, four cases (Cases 1-4) were indistinguishable and designated Type Y-A (Table 1 and Fig. 1). One case (Case 5) and one control (Control 1) each had one band difference from the major pattern of the cases and were designated Types Y-A1 and Y-A2. All 3 patterns (Y-A, Y-A1, and Y-A2) are between 1-3 band apart so they are considered closely related but are not the same, according to Tenover et al [27]. Two cases (Case 6 and Case 7) were indistinguishable from each other but had greater than seven band differences from the other cases and were designated Type Y-B, and the remaining case (Case 8) was greater than seven band different from the major patterns of the cases and was designated Type Y-C (Table 1 and Fig. 1). The isolates with Types Y-B and Y-C are deemed not clonally related to Types Y-A to Y-A2. The three environmental isolates were also designated Type Y-A, and had 0 band difference from the major pattern of the cases (Table 1 and Fig. 1). Of note, in Outbreak X, one case (Case 4) and two controls (Control 3 and Control 5) had isolates collected from different sites on different days that, though they were from the same patient, differed from each other (Table 1). Among 38 isolates obtained from two hospital outbreaks, there were two clonally related clusters: one cluster consisting of isolates with Types X-A1 to X-A6 (n = 20) and the other cluster of isolates with Types Y-A to Y-A2 (n = 9) (Fig. 1). The remaining six isolates collected from Outbreak X were all from controls, while the remaining three isolates from Outbreak Y were from three cases.

Typing of the isolates based on trilocus multiplex PCR and MLST
To determine which Global Clone (previously known as European Clone or International Clone) each of the A. baumannii isolates belongs to, trilocus multiplex (TLM) PCR was performed. Our results showed that, consistent with PFGE profiles (Fig. 1), 20 isolates associated with Outbreak X (14 case isolates, 6 control isolates) were found to belong to Global Clone I and they are all clonally related based on PFGE patterns (Fig. 1). In contrast, all 12 isolates from Outbreak Y (including case, control and environmental isolates) were found to belong to Global Clone II and all but three (LAC-236. LAC-237 and LAC-238) were clonally related (Fig. 1). Interestingly, six control isolates from three control patients in Outbreak X were also typed to Global Clone II (Fig. 1). Fig. 1 The dendrogram of PFGE profiles of the 38 A. baumannii isolates and comparison of PFGE types, trilocus multiplex PCR types and MLST sequence types. The dendrogram includes PFGE patterns of 38 isolates and the comparison was performed with the BioNumerics software v5.1 using the UPGMA (unweighted pair group method with arithmetic mean) clustering method and the Dice similarity coefficient with 1.5 % band matching tolerance and 0.5 % optimization. The PFGE types were assigned, first based on the outbreaks (X or Y), then based to the degree of similarity according to guidelines of Tenover and colleagues [27] To confirm TLM typing results, 13 representative A. baumannii isolates from the outbreaks were examined for their MLST sequence types (ST) based on the Pasteur Institute Scheme [29]. The MLST results with select isolates showed those isolates typed to Global Clone I were confirmed to be ST1 while those isolates typed to Global Clone II were confirmed to be ST2 (Fig. 1). These results confirmed that for these isolates, TLM PCR assays accurately typed them to respective Global Clones, consistent with their sequence typing. In summary, the first outbreak (X) involved the spread of 20 isolates assigned to PFGE types X-A to X-A6 which belong to ST-1, the second outbreak (Y) resulted in the spread of 9 isolates assigned to PFGE types Y-A to Y-A2 which belong to ST-2. Isolates with Y-B, Y-C patterns are not related to Y-A, Y-A1 and Y-A2 isolates even though they were also typed ST-2.
MLST, TLM and PFGE genotyping methods are frequently used for epidemiological studies and the surveillance of A. baumannii outbreaks [37,38]. PFGE has been considered the gold standard for outbreak investigations due to its higher discriminatory power. For years, PFGE had been the leading standard for typing of A. baumannii isolates [38]. However issues related to interlab exchange of data and reproducibility has raised concerns about reliability and portability of the results. While the TLM approach is simple to perform and has higher throughput, it has failed to resolve strains belonging to unusual lineages [24]. On the other hand, MLST is highly discriminating and uses the DNA sequences of several conserved housekeeping gene fragments [39]. It has emerged as the preferred method to type A. baumannii [40][41][42]. Our experience indicates that a bipartite interrogation using both PFGE and MLST approaches is ideal in examination of molecular epidemiology of A. baumannii isolates.

Antimicrobial susceptibility
To determine the antimicrobial susceptibility of these A. baumannii isolates, minimal inhibitory concentrations (MICs) were determined for a panel of 14 antibiotics against the 38 isolates obtained from the two outbreaks in Los Angeles County hospitals. Based on the MICs obtained for each antibiotic for the two groups of isolates (Outbreaks X and Y), MIC range, MIC 50 and MIC 90 values are summarized in Table 2. With the exception of polymyxin B which is active against all isolates, MIC 50 values of all antibiotics tested fell within resistant ranges as defined by CLSI [43]. Specifically, all 38 isolates were resistant to cefepime, cefotaxime, ceftazidime, ceftriaxone, ciprofloxacin, levofloxacin and piperacillin (based on MIC ranges, Table 2), some may even be called XDR (extremely drug-resistant) (data not shown) as defined by Magiorakos and colleagues [44]. Additionally, while the X group of isolates (outbreak X) appeared to be more heterogeneous than those of Y group (outbreak Y) because of higher number of antibiotics with wide range of MIC values, the MIC 50 and MIC 90 profiles were quite similar between the two groups ( Table 2).
There have been few reports on molecular epidemiology and antimicrobial resistance of isolates of A. baumannii obtained from hospitals in Los Angeles metropolitan area. In one of such reports, we characterized genetic and antimicrobial resistance profiles of 20 representative outbreak isolates of A. baumannii derived from hospitals in Los Angeles County, CA [14]. Using PFGE technique, these 20 multidrug resistant A. baumannii isolates were classified into eight epidemiologically distinct groups, which were consistent with isolates' antimicrobial susceptibility profiles [14]. Armed with more sophisticated molecular typing methodologies, such as MLST, we recently typed nine of these LAC isolates to ST10 (including LAC-4), five of them to ST417, two of them to ST241, while four of them confirmed as ST2 [24]. Although all isolates from the present study belong to ST1 and ST2 thus may not be genetically related to those 20 isolates examined previously, the increase in resistance phenotypes of the 38 isolates is quite remarkable ( Table 2), considering the two batches of isolates were collected ten years apart.
For example, all of the case isolates in the present study were resistant to all three aminoglycoside antibiotics tested (data not shown), while 25 % of isolates (5/20) from the previous batch were susceptible to amikacin and tobramycin [14]. Secondly, with the exception of three isolates (LAC-236, -237, and -238), all case isolates from the present study were resistant to both carbapenems (data not shown); in contrast, 50 % and 45 % of isolates from previous batch were susceptible to imipenem and meropenem, respectively [14].
Recently, Miyasaki and coworkers examined in vitro activity of antibiotic combinations against a panel of multidrug resistant strains of A. baumannii collected from Cedars-Sinai Medical Center in Los Angeles, CA [45]. In their report, among 102 hospital-acquired MDR A. baumannii isolates, 76 belonged to a major clone A while eight to a minor clone B, as characterized by repetitive PCR amplification using the DiversiLab Acinetobacter Fingerprinting Kit [45]. Unfortunately, because MLST tests were not performed by Miyasaki and colleagues, the genetic relationships of their isolates to those described in our studies (Fig. 1) [14] are unknown. In a study that reports immuno-protection against XDR A. baumannii infection in mice, Luo and coworkers described five isolates collected from in-patients at Harbor-UCLA Medical Center, Los Angeles, CA, in 2009 [46]. These five isolates were found to exhibit resistance phenotypes to at least one antimicrobial agents of all but 2 categories of antimicrobial agents [46].

Mechanisms of resistance
Our antimicrobial susceptibility testing indicated that all isolates were resistant to the three fluoroquinolone (FQ) antibiotics, with the exception of one control isolate which was intermediate to gatifloxacin (data not shown). Point mutations in the QRDR of genes encoding the GyrA and ParC subunits of DNA gyrase and DNA topoisomerase IV, respectively, are the primary mechanism of high-level bacterial resistance to fluoroquinolones (FQs) [31,32]. These mutations and the resultant amino acid changes led to an altered target protein structure and reduced fluoroquinolones binding affinity [47]. To confirm if this is the case for these A. baumannii isolates, the QRDRs of the gyrA and parC genes of the 38 A. baumannii isolates were amplified via PCR and sequenced. Sequencing results indicated that with exceptions of control isolates LAC-222 and LAC-225, all the remaining 36 isolates possess point mutations that led to amino acid substitutions in both GyrA and ParC polypeptides (Table 3). In LAC-222, only one mutation that occurred in gyrA gene was found, which explains the intermediate phenotype of LAC-222 to gatifloxacin (data not shown).
These results are consistent with the previous findings that mutations in both GyrA and ParC are required for high-level FQ resistance phenotypes [31]. Indeed, we have previously reported that presence of mutations in both gyrA and parC genes in A. baumannii isolates is associated with high-level resistance to fluoroquinolones [14,25]. Similarly, other investigators have found that possessing mutations in both gyrA and parC genes is strongly correlated with high level resistance to ciprofloxacin in A. baumannii isolates [18] or A. pittii isolates [48].
To assess the role of Class 1 integrons and the associated antibiotic resistance gene cassettes in antibiotic resistance phenotypes in the isolates, we performed PCR screens against the 38 isolates for the presence of Class 1 integrons by amplifying the integrase gene. Our results revealed that 31 of the 38 isolates harbored the integrase gene (Table 3) and that the seven isolates without integrase gene were all derived from control patients (Table 1 and Table 3). Subsequent conserved segment amplification via PCR, followed by DNA sequencing, indicated that all amplified conserved segments contain three antibiotic resistance genes sharing the same gene identity and organization: aacA4, catB8 and aadA1 (Table 3). Since aacA4 [aka aac (6')-Ib] encodes aminoglycoside 6'-N-acetyltransferase, which confers resistance to amikacin and tobramycin, among others [49,50], the presence of this gene (Table 3) correlates well with the susceptibility patterns of these 38 A. baumannii isolates because four of the seven control isolates being negative for class 1 integron were found to be susceptible to at least two of the three aminoglycoside antibiotics (data not shown). The presence of the integrase gene in 31 of the 38 isolates provides a critical surveillance parameter that is indicative of acquisition of antibiotic resistance genes.
Integrons have been shown to be an important type of mobile elements contributing significantly to the dissemination of antibiotic resistance genes in Gram-negative bacterial pathogens [51,52]. It was interesting to note that the majority of A. baumannii clinical isolates tested in this study harbor Class 1 integron in their genomes and the seven isolates without it were control isolates, indicating that the control isolates had not yet acquired the Class 1 integron. The fact that outbreak isolates from two hospitals shared that same antibiotic resistance Absence of mutations in gyrA or parC genes is denoted as "-" Presence or absence of class 1 integrase gene is denoted as "+" or "-" Empty spaces, no tests were performed gene array suggests the prevalence of this particular type of Class 1 integron in Los Angeles area. Such a gene array, aacA4-catB8-aadA1, was also found in many MDR A. baumannii clinical isolates collected from prison inmates [25]. This same array of gene cassettes has been identified in clinical isolates of A. baumannii worldwide: in 25 of the 30 A. baumannii isolates in Korea [53], in 47 of the 65 A. baumannii isolates in Taiwan [54], in all 30 clonal isolates of carbapenemresistant A. baumannii involved in a hospital outbreak in Latvia [55], and in 26 clinical isolates of A. baumannii obtained in central Ohio, U.S.A. [56]. Carbapenem resistance in A. baumannii has been suggested to contribute to increased risk of mortality in patients infected with this bacterium [57]. In this study, the majority of A. baumannii isolates tested were resistant to both imipenem and meropenem, with a few isolates exhibiting intermediate or susceptible phenotypes, suggesting widespread resistant strain dissemination in the LAC hospital setting. Multiplex PCR reactions using primer pairs specific to four OXA carbapenemase genes (bla OXA-23-like , bla OXA-24-like , bla OXA-51-like and bla OXA-58-like ) were performed against these 38 isolates. As anticipated, bla OXA-51-like gene was detected in every A. baumannii isolates (Table 3). Additionally, 29 out of 38 isolates (76 %) also harbor the bla OXA-23-like gene (Table 3). Furthermore, none of the isolates harbors bla OXA-24-like or bla OXA-58-like gene. Results from the ISAba1 element PCR screens indicated that ISAba1 was present in all of the 38 isolates (Table 3). Subsequent PCR screens against the 29 bla OXA-23 positive strains using ISAba1 forward and bla OXA-23 reverse primers confirmed that all of these 29 isolates possess ISAba1 element upstream of the bla OXA-23 gene (Table 3), which is consistent with the carbapenem resistant phenotypes of these isolates. The remaining nine isolates without the bla OXA-23 gene were either susceptible or intermediate to at least one carbapenem tested (data not shown). Remarkably, 20 isolates of the clonal cluster (LAC-201 to -220) from Outbreak X and nine isolates of the clonal cluster (LAC-227 to -235) from Outbreak Y all possess ISAba1-linked bla OXA -23 gene (Fig. 1, Table 3) and they all are resistant to both carbapenems (data not shown).
Carbapenems are considered the last resort to treat infections caused by A. baumannii. In this study, we showed that the majority of case isolates and some control isolates were resistant to both carbapenems, which is of great concern. The carbapenem susceptibility phenotypes in these 38 A. baumannii isolates are clearly correlated with the presence or absence, in the genomes of the isolates, of ISAba1-linked bla OXA -23-LIKE genes. Presumably, the ISAba1 element likely provides exogenous promoter functions to over-express the linked genes, as demonstrated previously [41,58,59]. In contrast, among the 20 nosocomial outbreak isolates of A. baumannii reported previously by us, 50 % and 45 % were susceptible to imipenem and meropenem, respectively [14], suggesting rapid dissemination of carbapenem resistant genetic determinants over a ten-year span. In fact, the LAC-4 strain, whose genome was completely sequenced, and 19 other LAC isolates from the 2008 study all lack bla OXA -23-like , bla OXA -24-like , and bla OXA -58-like genes, even though ISAba1 was detected in all of these isolates (data not shown). On the other hand, MDR A. baumannii strains isolated recently were often found to be resistant to carbapenems. For example, Hammerum and coworkers described eight carbapenem-resistant A. baumannii isolates in a Denmark hospital. These strains were typed using Pasteur MLST Scheme to five ST2 isolates, two ST1 isolates and one ST158, all of which harboring either bla OXA -23 or bla OXA -72 (which is bla OXA -24-LIKE ) genes, consistent with carbapenem resistance phenotype [40]. Additionally, 26 carbapenem-resistant A. baumannii clinical isolates from two Italian hospitals were characterized and typed to ST2 using MLST and all possess bla OXA -82 genes that are located downstream of ISAba1 elements [60]. Finally, Teo and coworkers identified 49 MDR A. baumanniii clinical isolates that were resistant to carbapenems and all isolates were found to harbor both bla OXA -51 and bla OXA -23 genes, in accordance with carbapenem resistant phenotypes [61].

Conclusions
This report presents the phenotypic and molecular analyses of 38 isolates of A. baumannii collected from two hospital outbreaks in Los Angeles County. Based on genetic patterns and molecular tests, one outbreak (Outbreak X) was dominated by a cluster of 20 clonally related isolates derived from Global Clone I or ST1 (MLST Pasteuar Scheme) while the other (Outbreak Y) was dominated by a different cluster of nine clonally related strains belonging to Global Clone II or ST2. While every single isolate (including environmental isolates from the hospital) is multidrug resistant, all isolates were found uniformly resistant to seven antibiotics belonging to cephalosporin, penicillin and fluoroquinolone classes. There was good correlation between antimicrobial susceptibility phenotypes and the presence of resistance genetic determinants. In particular, it is alarming that nearly all of these cluster isolates (except three that were obtained from control patients) were resistant to both imipenem and meropenem, which can be attributed to presence of bla OXA -23 genes downstream of the ISAba1 in the genomes of these isolates. The results presented here provide clinically relevant insights into antimicrobial susceptibility, clonal dissemination and mechanisms of resistance in Los Angeles area hospitals.

Ethics approval and consent to participate
Because the clinical isolates were not isolated by coauthors who had no direct interactions with patients from whom these isolates were isolated and the isolates do not contain patient identifiable information, ethics review and approval by institutional review board (IRB) was waived. Consent to participate is not applicable.

Consent for publication
Not applicable.

Availability of data and materials
All data is contained within the manuscript. Clinical isolates of A. baumannii will be made available upon requests from Dr. H. Howard Xu.