Thirty-two clinical isolates of A. baumannii were collected from Southern Texas (24 isolates from Clinical Bacteriology Laboratory Services Section, Texas Department of State Health Services, Austin, TX; 8 isolates from Pathology Department of Baylor College of Medicine, Houston, TX). All isolates were collected from individual patients and from August 2010 to July 2013. Two carbapenem resistant A. baumannii collected from Chicago, IL (C1: H26, OXA-23 positive; C2: AC1204, OXA-24 positive) and one carbapenem resistant A. baumannii AB0057 collected from Washington, D.C. (OXA-23 positive; a complete genome sequence strain) were obtained from Lolans et al. [12] and Adams et al. [16] , respectively. A. baumannii ATCC 19606 was also included as a reference strain in this study. To ensure identity of A. baumannii, gyrB multiplex PCR was performed for each of the isolates as described [17] and their identities were confirmed by DNA sequence determination (data not shown). The clinical isolates were routinely grown in Luria-Bertani (LB) broth or LB agar plates.

All antibiotics used in this study were purchased from Sigma (Sigma, St. Louis, MO) except for meropenem (U S Pharmacopeia, Rockville, MD). Antibiotic susceptibility was determined by minimal inhibitory concentrations (MICs) using the broth dilution method as described [18] . Each antibiotic was added to divalent cation-adjusted Mueller-Hinton (MH) (Oxoid, Ogdensburg, New York) broth (pH 7.0) to achieve serial two-fold dilutions between 0.25 and up to 128 µg/mL using sterile 17 × 100 mm snapped-cap Falcon culture tubes (1 mL/tube; Fisher Scientific). Fresh overnight cultures of each isolate were diluted in saline to an optical density at 600 nm of 0.1 to 0.12 (approximately 1 × 10⁸ viable cells per mL). A portion of the adjusted cell suspension (3 μL for ~10⁵ cells) was inoculated to each MH broth containing antibiotics as indicated. The cell cultures were then incubated overnight (16 to 18 hrs) at 37˚C without shaking. MIC was defined as the lowest concentration that completely inhibited growth of the inoculums. MIC values were confirmed by three independent experiments. Antibiotic resistant breakpoints were used as suggested by Clinical and Laboratory Standards Institute (CLSI) cited by Peleg et al. [2] .

Presence of genes encoding carbapenemases known as Ambler Class A β-lactamase (bla_KPC2), Ambler Class B metallo β-lactamase (bla_IMP, bla_GIM, bla_SIM, bla_SPM, and bla_VIM), and Ambler Class D β-lactamase or oxacillinase (bla_OXA-23-type, bla_OXA-24-type, bla_OXA-51-type, and bla_OXA-58-type) was examined by multiplex PCR methods as described [1] [19] . The presence of an insertion sequence element (ISAba1) and its location in the upstream of bla_OXA-51-type was also examined by PCR methods as described [8] . All PCR amplified fragments from clinical isolates were used to determine DNA sequences of both strands from commercial DNA sequencing services (GENEWIZ, Inc., South Plainfield, NJ).

Genomic DNA from A. baumannii isolates A21 (bla_OXA-23 positive) and H1 (bla_OXA-24 positive) was extracted by Qiagen DNeasy kit (Valencia, CA). The genomic DNA was completely digested by BamHI or EcoRI and ligated to pGH240 (a bla gene of pBluescriptSK+ [Stratagene, La Jolla] was destroyed by inserting a chloramphenicol resistant gene cassette [20] into ScaI site) digested by BamHI or EcoRI, respectively. E. coli DH5α was then transformed with the ligation mixture by a standard chemical transformation method. Transformed E. coli DH5α carrying bla_OXA-23 or bla_OXA-24 was selected on LB agar plates containing chloramphenicol (15 μg/mL) and Ampicillin (100 μg/mL). Insert DNA fragments from the recombinant plasmids were confirmed by BamHI or EcoRI digestion and used to determine DNA sequences of both strands from commercial DNA sequencing services (GENEWIZ, Inc., South Plainfield, NJ).

Carbapenem susceptible A. baumannii ATCC 19606 grown in LB broth was concentrated at 10¹⁰ cells and spread on LB agar plates containing 50 μg/mL of rifampicin for 24 hours and one of colonies grown on the agar plates was used to grow in LB broth containing 50 μg/mL of rifampicin for 24 hours. The cells grown in LB broth containing 50 μg/mL of rifampicin were spread on agar plates containing 100 μg/mL of rifampicin. One of the colonies grown on the agar plates containing 100 μg/mL of rifampicin was subcultured on LB agar plates without any antibiotic and passaged three times on the same LB agar plates. The cells on the third passage was used to confirm stability of rifampicin resistance (MIC > 100 μg/mL) and used for a recipient strain. Two carbapenem resistant clinical isolates (A21 for OXA-23 positiveand H1 for OXA-24 positive) were confirmed as rifampicin susceptible (MIC 0.5 μg/mL) and used as donor strains for conjugative transfer of carbapenem resistance. Mating conditions for the donor and recipient strains were essentially identical as described previously [21] except for an extended-mating period (16 to 18 hours). The mating mixtures were spread on LB agar plates containing meropenem (5 μg/mL) or carbenicillin (100 μg/mL) and rifampicin (100 μg/mL). Individual transconjugants (3 colonies) were used for phenotype analysis and genomic DNA extraction.

Electroporation for transformation of A. baumannii ATCC 19606 was performed as described [22] . Briefly, electrocompetent cells were prepared by washing the cells (OD₆₀₀ of 0.8) 3 times with 10% (v/v) glycerol and concentrating the cells at 2 × 10¹⁰ cells per mL of 10% (v/v) glycerol. Plasmid DNA (up to 3 μg in 40 μL) extracted by Qiagen kit (Valencia, CA) mixed with the electrocompetent cells and placed in a pre-chilled sterile electroporation cuvette (2 mm electrode gap, Bio-Rad) following immediately pulse by using a Bio-Rad Gene Pulser (2.5 kV, 200 Ω, and 25 μF). The mixture was incubated at 37˚C for 1 hr with 1 mL of LB broth and spread the cells on agar plate containing carbenicillin (100 μg/mL) or meropenem (5 μg/mL). Natural transformation of carbapenem susceptible A. baumannii ATCC 19606 was also alternatively performed as described [23] . Briefly, 100 ng of plasmid DNA was added in a mixture containing 50 μL of fresh LB broth and 50 μL of culture in a stationary phase, followed by incubation at 37˚C for 1 hour and plating on LB agar plates containing carbenicillin (100 μg/mL) or meropenem (5 μg/mL). As a control for the natural competency of A. baumannii ATCC 19606, kanamycin resistant plasmid (pJHCMW1) was used for the natural transformation by selection on LB agar plates containing kanamycin (25 μg/mL) [23] [24] .

BamHI DNA fragment carrying bla_OXA-23 (4183-bp) from clinical isolate of A. baumannii A21 and EcoRI DNA fragment carryingbla_OXA-24 (6285-bp) from clinical isolate of A. baumannii H1 were deposited in GenBank accession numbers JN207493 and JN207494, respectively.

MIC levels to multiple antibiotics were determined for the 32 clinical isolates collected from Austin and Houston, TX including 3 representative carbapenem resistant clinical isolates from Chicago, IL, Washington, D.C., and the reference A. baumannii ATCC 19606. Results revealed that 28 isolates (87%) were resistant to all tested β-lactams including carbapenem (imipenem and meropenem). Four isolates were resistant to polymyxin B (MIC ≥ 8 μg/mL) and 3 of them were resistance to all tested antibiotics including polymyxin B (Table 1). The 3 isolates from Chicago, IL and Washington, D.C. were also resistant to all tested antibiotics except for polymyxin B. The reference strain ATCC 19606 was susceptible to all tested antibiotics (Table 1).

Multiplex PCR was performed to detect the carbapenemase encoding genes as described [1] [19] . Results revealed that a gene encoding OXA-51-type was detected in all 36 isolates. A gene encoding OXA-23-type was detected in 5 of the 36 isolates (3 from Austin, TX, 2 positive controls from Chicago, IL and Washington, D.C.). A gene encoding OXA-24-type was detected in 26 of the 36 isolates (17 from Austin, TX, 8 from Houston, TX, and a positive control from Chicago, IL) (Table 2). None of the other carbapenemase encoding gene was detected from any isolate in this study.

^aThe isolate AB0057 was obtained from Washington, D.C. [16] ; isolates (C1 and C2) were obtained from Chicago, IL [12] ; isolates of A1 to A24 were collected from Austin, TX; isolates of H1 to H8 were collected from Houston, TX. ^bMIC measurements were repeated three times with identical results. Note: aztreonam, ATM; carbenicillin, CAR; ceftazidime, CAZ; ciprofloxacin, CIP; gentamicin, GEN; imipenem, IMP; meropenem, MEM; polymyxin B, PXB; tetracycline, TCN.

^aPCR primer pairs used for bla_OXA-23-type was 5’-GATCGGATTGGA GAACCAGA-3’/5’-ATTTCTGACCGCATTTCCAT-3’, for bla_OXA-24-type was 5’-GGTTAGTTGGCCCCCTTAAA-3’/5’-AGTTGAGCGAAAAGGGGATT-3’ and for bla_OXA-51-type as described [1] .

Presence of the insertion sequence element (ISAba1) at upstream of the gene encoding OXA-51-type requires for production of significant carbapenem resistance in A. baumannii [8] . To understand involvement of the OXA-51-type in carbapenem resistance, presence andlocation, if any, of the ISAba1 was examined from the isolates as described [8] . Results revealed that the ISAba1 was detected in all isolates except for 8 isolates from Houston, TX. However, none of them was preceded to the gene encoding OXA-51-type.

To understand dissemination of carbapenemase encoding genes among the clinical isolates, two PCR-based genotype methods (REP-PCR and RAPD-PCR) proved as similar discrimination power as a pulse field gel electrophoresis (PFGE) method [25] were used. REP-PCR profiles for the 3 isolates carrying OXA-23-type from Austin, TX (A6, A15, A21) showed an identical pattern (Figure 1, lane 3) which was different from that of isolate carrying OXA-23-type from Washington, D.C. or from Chicago, IL (Figure 1, lane 1 and 2). REP-PCR profiles for all isolates carrying OXA-24-type from Austin and Houston, TX and Chicago, IL showed an identical pattern (Figure 1, lanes 4 and 5); interestingly, this genotype pattern was very similar as that of the isolate carrying OXA-23-type from Chicago, IL (Figure 1, lane 2).

RAPD-PCR profiles for the 3 isolates carrying OXA-23-type from Austin, TX showed an identical pattern (Figure 1, lane 8) which was different from that of isolate carrying OXA-23-type from Washington, D.C. or from Chicago, IL (Figure 1, lanes 6 and 7). As shown in REP-PCR profiles, RAPD-PCR profiles for all isolates carrying OXA-24-type from Austin and Houston, TX and Chicago, IL showed an identical pattern (Figure 1, lanes 9 and 10). The two genotype methods produced repeatedly consistent results, suggesting that a unique isolate carrying OXA-23-type is currently propagated in Austin, TX but the isolate carrying OXA-24-type in Austin and Houston, TX may be clonally disseminated from isolates carrying OXA-24-type in Chicago, IL or vice versa.

The genes encoding carbapenemase of OXA-23-type and OXA-24-type were cloned from the clinical isolates A21 and H1, respectively. BamHI fragment (4183-bp) from total DNA of the clinical isolate A21 conferred resistance to ampicillin (100 μg/mL) in E. coli DH5α and the insert DNA was used for DNA sequencing analysis. As shown in Figure 2(a), the gene encoding OXA-23 was a part of Tn2008 that was integrated into a plasmid

pACICU2 (GenBank accession number CP000865) and deduced amino acid sequence of the gene was 100% identical as previously reported OXA-23 (GenBank accession number EU594641) [26] . A newly designed forward PCR primer for upstream 500-bp from the start codon of OXA-23 (5’-GTATCTGAATTT CCACGTTT-3’) and a reverse primer for downstream 500-bp from the stop codon of OXA-23 (5’-AAGTGTACTTGATGGCTCAT-3’) were used to amplify a full length of bla_OXA-23 from the other two OXA-23 positive isolates (A6 and A15). DNA sequence analyses of the PCR fragments containing bla_OXA-23 from the two isolates were 100% identical as that from the isolate A21.

EcoRI fragment (6285-bp) from total DNA of the clinical isolate H1 conferred resistance to ampicillin (100 μg/mL) in E. coli DH5α and the insert DNA was used for DNA sequencing analysis. As shown in Figure 2(b), the gene encoding OXA-24 was flanked by XerC/XerD-like recombinases binding sites and surrounded by DNA sequences originated from multiple plasmids. Deduced amino acid sequence of the gene was 100% identical as previously reported OXA-24 (GenBank accession number GU199040) [13] . A newly designed forward PCR primer for upstream 500-bp from the start codon of OXA-24 (5’-AGGTGTGCATCATACTTCTA-3’) and a reverse primer for downstream 500-bp from the stop codon of OXA-24 (5’-TAAAGAATCAGATCGATTTT-3’) were used to amplify a full length of bla_OXA-24 from all the other OXA-24 positive isolates (24 isolates). DNA sequence analysis showed that all PCR fragments were 100% identical as that from the isolates H1.

Gene transfer experiment was performed by bacterial conjugation to test whether the carbapenem resistance of the clinical isolates was transferrable to a carbapenem susceptible A. baumannii. Each of carbapenem resistant but rifampicin susceptible clinical isolates (OXA-23 for the isolate A21, OXA-24 for the isolate H1; Table 1) was mated to the rifampicin resistant but carbapenem susceptible strain of A. baumannii ATCC 19606-rif-R.

Results revealed that countless transconjugants from each conjugation mating mixture grew on LB agar plates containing carbenicillin (100 μg/mL) or meropenem (5 μg/mL) and rifampicin (100 μg/mL) suggesting carbapenem resistance can be transferrable and may be mediated by a plasmid-specific gene.

To test whether the plasmid-specific gene is associated with carbapenem resistance plasmid DNA extracted from the clinical isolates (A21 and H1) was used for transformation of the carbapenem susceptible ATCC 19606 with selection on LB agar plates containing carbenicillin (100 μg/mL) or meropenem (5 μg/mL). Results revealed that each plasmid DNA from the carbapenem resistant clinical isolates produced countless transformants on the agar plates. Presence of the genes encoding OXA-23 and OXA-24 in the transconjugants and transformants (3 colonies for each) was confirmed by PCR methods as done above.

To understand roles of the carbapenemase encoding genes in β-lactam susceptibility the transconjugants of rifampicin resistant A. baumannii ATCC 19606 and the transformants of E. coli DH5α carrying bla_OXA-23 and bla_OXA-24 were used to measure β-lactam susceptibility. As shown in Table 3, MIC levels to carbapenem (imipenem and meropenem) and penicillin (carbenicillin) were significantly increased (up to 64-fold) in both A. baumannii ATCC 19606 and E. coli DH5α for both bla_OXA-23 and bla_OXA-24. However, MIC levels to monobactam (aztreonam) and cephalosporin (ceftazidime) were slightly increased or no difference in both A. baumannii ATCC 19606 and E. coli DH5αfor both bla_OXA-23 and bla_OXA-24. These results suggest that the transconjugants of A. baumannii ATCC 19606 are similar MIC levels for both bla_OXA-23 and bla_OXA-24 as that of their clinical isolates while the transformants of E. coli DH5α are significantly lower MIC levels for both genes than that of their clinical isolates.