Identification and Characterization of a Novel aac(6′)-Iag Associated with the bla IMP-1–Integron in a Multidrug-Resistant Pseudomonas aeruginosa

In a continuing study from Dec 2006 to Apr 2008, we characterized nine multi-drug resistant Pseudomonas aeruginosa strains isolated from four patients in a ward at the Hiroshima University Hospital, Japan. Pulsed-field gel electrophoresis of SpeI-digested genomic DNAs from the isolates suggested the clonal expansion of a single strain; however, only one strain, NK0009, was found to produce metallo-β-lactamase. PCR and subsequent sequencing analysis indicated NK0009 possessed a novel class 1 integron, designated as In124, that carries an array of four gene cassettes: a novel aminoglycoside (AG) resistance gene, aac(6′)-Iag, bla IMP-1, a truncated form of bla IMP-1, and a truncated form of aac(6′)-Iag. The aac(6′)-Iag encoded a 167-amino-acid protein that shows 40% identity with AAC(6′)-Iz. Recombinant AAC(6′)-Iag protein showed aminoglycoside 6′-N-acetyltransferase activity using thin-layer chromatography (TLC) and MS spectrometric analysis. Escherichia coli carrying aac(6′)-Iag showed resistance to amikacin, arbekacin, dibekacin, isepamicin, kanamycin, sisomicin, and tobramycin; but not to gentamicin. A conjugation experiment and subsequent Southern hybridization with the gene probes for bla IMP-1 and aac(6′)-Ig strongly suggested In124 is on a conjugal plasmid. Transconjugants acquired resistance to gentamicin and were resistant to virtually all AGs, suggesting that the In124 conjugal plasmid also possesses a gene conferring resistance to gentamicin.


Introduction
Pseudomonas aeruginosa strains generally carry intrinsic resistance to various antimicrobial agents. This organism is susceptible to a limited number of drugs including some of the b-lactams, e.g. ceftazidime and imipenem (IPM), and the AGs, e.g. amikacin (AMK) and isepamicin (ISP). However in nosocomial settings, acquired resistance to such anti-pseudomonal agents is frequent and involves more than one antibiotic class. The acquisition of metallo-b-lactamases (MBLs) is often selected by mobile genetic elements as cassettes inserted into integrons, which confer a multi-drug resistance profile against virtually all anti-pseudomonal b-lactams as well as other classes of antibiotics such as the AGs [1]. During the last decade, the acquired MBLs have emerged; these broad-spectrum b-lactamases raise a serious concern with respect to antimicrobial chemotherapy as well as to control propagation of multi-drug resistant P. aeruginosa (MDRP) [1]. Therefore, in Japan, a new infectious disease control law was issued by the government in 1999 defining MDRP that meet MIC criteria for resistance to IPM$16 mg/ml, ciprofloxacin (CIP)$4 mg/ml and AMK$32 mg/ml. The appearance of such strains signals reports for emergence of MDRP.
In the Hiroshima region, in 2004 we began monitoring MDRP in clinical isolates in eight major hospitals [2,3]. Within three years the genetic mechanism of IPM resistance in MDRP in this region changed; the ratio of MBL-positive strains significantly increased in MDRP, reaching 80% during 2006 [2]. Investigation of the genetic content of MBL integrons in MDRP showed most of them carried resistance genes conferring resistance to AGs, suggesting acquisition of the MBL integron facilitated the recent propagation of multi-drug resistance determinants in P. aeruginosa in Hiroshima.
Here we describe the identification of a class 1 integron containing a novel acetyltransferase gene in its variable region in an MDRP strain that caused an outbreak of infection in the Hiroshima University Hospital. We report here the structure of this gene and characterization of the gene products.

Bacterial strains, plasmids and primers
In 2008, nine clinical P. aeruginosa strains, NK0001-NK0009, were isolated from four patients in a ward of Hiroshima University Hospital. Pseudomonas aeruginosa PAO1 resistant to rifampicin was used as the recipient for the conjugation experiment [2]. P. aeruginosa 060123 was used as a positive control as a bla IMP-1 carrier [2]. Escherichia coli XL-II and M15 were used as host strains for the recombinant plasmid and expression of aac(69)-Iag, respectively. The plasmid pGEM-T Easy (Promega, Tokyo, Japan) was used as a sub-cloning vector and for sequencing analysis. Plasmid pQE30 (Qiagen) was used for expression of the recombinant protein. Primers used in this study are listed in Table 1.
PCR detection and characterization of the variable region of the bla IMP-1 -containing integron The structure of the variable region of the bla IMP-1 -containing integron was determined using a PCR mapping approach with primers designed from the panel of integron sequences isolated in the Hiroshima region during 2004-2006 (Table 1) [2]. PCR amplification was performed using Takara EX Taq DNA polymerase (TaKara Tokyo, Japan) using 25 cycles: denaturing at 98uC for 10 sec; annealing at 50uC for 30 sec; and polymerization at 72uC for 1 min. DNA sequencing of the PCR products was performed using a CEQ Genetic Analysis System (Beckman Coulter Inc., Fullerton, CA). Comparison of experimentally

Alignment and phylogenetic analysis of the aminoglycoside acetyltransferase (aac) gene
The amino acid sequences of the aac genes were obtained from NCBI (http://www.ncbi.nlm.nih.gov/). Multiple sequence alignments for the 119 aac genes was performed using the ClustalW program [6]. A phylogenetic tree of the aligned sequences was generated with the neighbor-joining method using MEGA4 software [7]. The ribosomal protein of human (Genbank accession number BAA04887) was used for the root of the phylogenetic tree.

Conjugation experiments and plasmid purification
The conjugation experiment was performed using the filtermating method [8]. Donor and recipient were P. aeruginosa clinical isolate NK0009 and rifampicin (RIF)-resistant Pseudomonas aeruginosa PAO1 (PAO1Rp), respectively. After filter-mating, the filter was incubated on NAC agar for 6 hours and was washed using 300 ml sterile water. The collected material was incubated on LB agar containing 500 mg/ml RIF, 16 mg/ml IPM and 16 mg/ml AMK or 8 mg/ml TOB for 24 hours; and then the transconjugants were selected.

Southern blot hybridization
Southern blot hybridization was performed using PFGEseparated DNA after transfer to a Hybond N + membrane using an ECL direct nucleic acid labeling and detection kit (Amersham-Pharmarcia). A 587-bp PCR-generated fragment was internal to the bla IMP-1 gene and used as the probe to detect the bla IMP-1 gene [2].
Purification of the AAC(69)-Iag from E. coli The following set of primers was designed and used to amplify the aac(69)-Iag gene: 59-CCCCGGATCCATGAGCAAGT-TAGG-39 (forward) and 59-GGGGGTCGACGACTCTGCT-GCGG-39 (reverse). Restriction enzyme sites (underlined) were introduced for the in-frame expression of recombinant proteins in the pQE30 expression vector. The PCR was performed using the same conditions as described above. After the sequence confirmation, a 0.6-kbp BamHI-SalI fragment was inserted into the same site of pQE30 to construct pQE30-aac(69)-Iag. The recombinant protein was expressed as an N-terminal 66 Histagged fusion protein under the control of the T7 promoter in the plasmid. E. coli M15 harboring pQE30-aac(69)-Iag was grown at 37uC in LB medium with 100 mg/ml ampicillin and 25 mg/ml KAN. The expression of the recombinant protein was induced adding 1 mM isopropyl-b-D-thiogalactoside when the optical density at 600 nm reached around 0.8, and it was incubated at 37uC for an additional 5 hours. The culture was centrifuged and re-suspended in 10 ml lysis buffer. After sonication on ice, 2 ml 50% Ni-NTA slurry was added and shaken for 1 h at 4uC. The slurry was then washed with 10 ml wash buffer three times and eluted with 1 ml of elution buffer. The eluted protein was dialyzed in 0.1M sodium phosphate buffer (pH 6.8) for 2 h at 4uC. The purity of the protein was confirmed using SDS-PAGE on a 15% gel; and subsequent staining with Coomassie Brilliant Blue.

TLC and mass spectrometry analysis
The enzymatic acetylation of AGs was performed using the method described previously [9]. The reaction mixture was 1 mM AGs, 1 mM acetyl-CoA, and the purified recombinant protein (500 mg/ml) in a final volume of 20 ml 10 mM phosphate buffer (pH 7.0). The reaction mixture was incubated at 37uC for 12 h. Aliquots of the reaction mixture were applied to silica gel TLC 60F 254 (Merck, Ltd. Japan) and developed with 3:2:1 methanol:ammonium hydroxide:chloroform. The AGs and their acetylated products were detected spraying with 0.2% ninhydrin solution in ethanol. Alternatively, the reaction mixture was analyzed by a matrix-assisted laser desorption ionization (MALDI) -time of flight (TOF) mass spectrometer (Biflex IV, Burker Daltonics) or an electrospray ionization (ESI)-MS/MS spectrometer (LTQ Orbitrap XL, Thermo Fisher Scientific). In the MALDI-MS analysis, the samples (1 ml) were co-applied with an equal volume of the matrix, a-cyano-4-hydroxycinnamic acid dissolved in a mixture of CH 3 CN: 0.1% TFA (1:1) as a saturated solution onto the sample plate and allowed to dry before insertion into MS. The MALDI-MS spectra were obtained as an ion of [M+Na]+using a reflectron positive mode.

Kinetic studies of AAC(69)-Iag
The AAC(69)-Iag activity was determined spectrophotometrically measuring the increase in A 412 due to the formation of 5-thio-2-nitrobenzoate (TNB, 15,570M 21 cm 21 ), resulting from the reaction between the acetyl-coenzyme A (Acetyl-CoA) and 5,59dithiobis-(2,2)-nitro benzoic acid (DTNB) [10]. Kinetic assays were performed in a 200 ml reaction mixture containing 50 mM phosphate buffer (pH 7.0), 200 mM Acetyl-CoA, 2 mM 5,59dithiobis-(2,2)-nitro benzoic acid (DTNB), 0.6 mg AAC(69)-Iag and AG substrates, and monitored continuously with a plate reader (VarioScan, ThermoFisher Scientific). The reactions were performed at 37uC. Enzyme activities were calculated from the initial rate. For the estimation of the kinetic parameters, a Lineweaver-Burk plot was used. One unit of enzyme activity is defined as the amount of enzyme catalyzing the formation of 1 mmol TNB per min at 37uC.

Nucleotide sequence accession number
The nucleotide sequence reported in this study was deposited in the EMBL/GenBank/DDBJ databases under accession number AB472901.

Epidemiology of a nosocomial outbreak of P. aeruginosa
From Feb to Nov 2007, a nosocomial outbreak caused by P. aeruginosa occurred in a ward of the Hiroshima University Hospital. During that period, nine clinical P. aeruginosa strains (NK0001-NK0009) were isolated from four patients who overlapped during their stay periods. Comparisons of the MICs of various antimicrobial agents including imipenem showed very similar MIC profiles; the strains were classified as MDRP. Among the nine strains only the last isolate from the period, NK0009, produced metallo-b-lactamase using the SMA test. Genotyping of the nine P. aeruginosa isolates using PFGE showed identical migration patterns using SpeI-digested genomic DNA from the isolates (Fig. 1a).

Detection of a new integron in P. aeruginosa NK0009
We analyzed NK0009 for the metallo-b-lactamase genes using PCR universal primer sets to identify the three types of metallo-blactamase genes: bla IMP-1 , bla VIM-2 or bla SPM ; and then performed aac(69)-Iag with the blaIMP-1-Integron in an MDRP subsequent direct sequencing. The strain was positive for bla IMP-1 (data not shown). Our recent molecular epidemiological study in the Hiroshima region demonstrated there are at least six variants of the integron gene cassette arrays (from type A to F) in bla IMP-1containing class 1 integrons found in MDRP [2]. Clustering analysis of PFGE patterns of all of the P. aeruginosa bla IMP-1 positive strains isolated during 2004-2006 in the Hiroshima region together with NK0009 suggested NK0009 belonged to a cluster of P. aeruginosa with integron type A [2]. This integron contains a single bla IMP-1 gene cassette between the 59-CS and the 39-CS. Our previous study demonstrated that there is a correlation between integron type and genotype [2]. Therefore, PCR scanning analysis of NK0009 genomic DNA with primer sets covering integron type A was performed. Among seven sets of PCR amplicons, three consecutive amplicons were unexpectedly longer (shown as 2, 3, 4 in Table 1 and Fig. 2b). This indicated NK0009 possesses additional DNA in the framework of integron type A.

Drug resistance mediated by AAC(69)-Iag
To investigate the AG resistance activity of AAC(69)-Iag, a DNA fragment containing aac(69)-Iag was amplified using primers intl1-1014r and imp1-500r, and cloned into pGEM T-Easy to generate pKK1. The recombinant was transformed into E. coli XL-II. Escherichia coli harboring pKK1 showed resistance to AMK, KAN, ABK, DBK, TOB, SISO, and ISP; but was susceptible to GEN ( Table 2). This indicated aac(69)-Iag is involved in AG resistance.

Location of In124 including the aac(69)-Iag
To investigate whether the integron including aac(69)-Iag was located on a plasmid, a conjugation experiment was performed using rifampicin-resistant P. aeruginosa PAO1 (PAO1Rp) as a recipient and a transconjugant was obtained. MICs of IPM for NK0009 and its transconjugant were .512 mg/L and 128 mg/L, respectively ( Table 2). Those for AMK and GEN were 256 mg/L and .512 mg/L for NK0009 and 64 mg/L and .512 mg/L for the transconjugant, respectively. Conversely, the MICs of IPM, AMK, and GEN for PAO1Rp were 1 mg/L, 1 mg/L, and 0.5 mg/L, respectively. The PFGE patterns of the transconjugants were identical to the recipient, and were distinct from NK0009 (Fig. 1a). This indicated that the resistance to IPM, AMK, and GEN of NK0009 was transferred to the recipient PAO1Rp. The resistances to CIP and LVX were not transferred (data not shown). Southern hybridization using a bla IMP-1 probe demonstrated that both the wild type and the transconjugant had the same hybridized bands (Fig. 1b, arrow heads).

Aminoglycoside acetylation by AAC(69)-Iag
To investigate potential acetylase activity, we incubated the purified recombinant AAC(69)-Iag with various AGs in the presence or absence of acetyl coenzyme A; and the reaction mixtures were analyzed using TLC. As shown in Fig. 5 conversion of all tested AGs using the purified protein were observed in the presence of acetyl coenzyme A. These AGs all possess a 69-NH 2 [25,26]. To further analyze the AAC(69)-Iag activity, each reaction mixture of AG in the presence or absence of acetyl coenzyme A were analyzed using MS (Table 3). In the absence of acetyl coenzyme A, the parent ion of amikacin in the reaction mixture had a m/z = 608.5 that corresponds to amikacin Na. After incubation in the presence of acetyl coenzyme A, the parent ion at m/z = 608.5 was converted to m/z = 650.5. This mass indicated that AMK was modified adding one acetyl moiety (m/z = 42).  Table 4). The BX and BCX ions represented the product ions of ring A fission (Fig. 6). The glycoside residue on the C6-O of the 2-deoxystreptamine was observed to undergo significant decomposition at the C2-C3 and O-C1 bonds ( Table 4). All AGs have free C69-amine group on ring C, except for GEN-C1, which has a secondary amine group on ring C (GEN is a mixture of C1, C2 and C1a) (Fig. 6, Table 4). C69-amine group in ring C is the target of acetyl modification by AAC(69)-Iag. In an analysis of ions of samples treated with AAC(69)-Iag, the m/z value of P, (P-18/17), BC, (BC-18/17), BCX, and (C-18) containing ring C was 42 daltons higher than corresponding ions of the sample without AAC(69)-Iag (Table 4). Mass shifts of 42 daltons reflect a structural change introducing an acetyl moiety in ring C. Conversely, the m/z values of AB, BX, and B were the same value between samples treated with/without AAC(69)Iag, suggesting no modification moiety in ring A and ring B. These results indicate AAC(69)Iag introduces an acetyl moiety on the C69 primary amino group in ring C.

Kinetic studies of AAC(69)-Iag
AAC(69)-Iag demonstrated the expected broad substrate specificity for AGs with free amino groups at their C69 positions. Paromomysin substitutes a hydroxyl group for an amine group at position 69, and therefore was not a substrate (Data not shown). We performed the kinetic studies for AAC(69)-Iag (Table 5). All AGs, with a free 69 amino group tested in this research were substrates. The turn-over rate (k cat ) for all substrates were relatively low (0.10 to 0.79 s 21 ), while the K m s varied (3.03 mM to 209 mM). The specificity constant (k cat /K m ) determined for AGs varied by a factor of up to 160. Using the kinetic constants, AG substrates can be divided into three classes. The first class include AMK and ISP which displayed a lower specificity constant (k cat /K m values . From the relative specificity constant (k cat /K m ), SISO and ABK were found to be the best substrates, and AMK and ISP were the worst substrates.

Discussion
Since nine MDRP with identical PFGE patterns were isolated in the same ward from four patients whose hospitalization periods overlapped, it is most likely that this was an outbreak from a single MDRP clone. During nosocomial spread, an MDRP strain  appeared to acquire the class 1 integron containing a bla IMP-1 gene in the hospital. The structure of the integron of NK0009 was different from other bla IMP-1 -containing integrons recently isolated in the Hiroshima region [2]. The first cassette at the 59 end of the NK0009 bla IMP-1 -containing integron was a novel AG acetyltransferase gene, aac(69)-Iag, and the second was a bla IMP-1 followed by truncated forms of bla IMP-1 and aac(69)-Iag.
The conjugation experiment and subsequent Southern hybridization using a gene probe for bla IMP-1 suggested this integron is on a conjugal plasmid. NK0009 was resistant to GEN and this resistance was also transferred to PAO1Rp (Table 2). However, E. coli harboring aac(69)-Iag showed resistance to a variety of AGs; but not to GEN. Previous studies of substrate profiles of 24 different AAC(69)-I indicated that substrates for AAC(69)-I are TOB, AMK, NTM, DBK, SISO, KAN, and ISP; but GEN was not included as a preferred substrate [18,27,28,29]. This suggests AAC(69)-Iag does not contribute to GEN resistance where other AG resistance genes conferring resistance to gentamicin may be present on the plasmid carrying bla IMP-1 and aac(69)-Iag.
Our in vitro acetylation assay using TLC and mass spectrometry showed AAC(69)-Iag modified most of the AGs tested in the presence of acetyl coenzyme A (Fig. 5, Table 3). In the case of GEN, C1a and C2 were acetylated by AAC(69)-Iag but C1 was not (Table 3). Therefore exposure of cells to gentamycin including all GEN C compounds may have resulted in their death due to the activity of remaining un-acetylated compounds C1.
Though a variety of AAC(69)-I enzymes have been identified in Gram-positive and -negative microorganisms, a limited number of kinetic studies on AACs have been performed. Steady-state kinetic parameters for drugs are useful to compare the drugs and to infer the mechanism of inactivating activity to drugs. We evaluated the kinetic parameters of AAC(69)-Iag for various AGs using recombinant AAC(69)-Iag with a 66 histidine tag at the Nterminal. Kinetic study indicated the OH groups of AGs at positions 29, 39, and/or 49 affect the binding to AAC(69)-Iag. Among kinetic parameters, Km is estimated as a useful indicator of relative substrate affinity of the enzyme and, in some cases, the kinetic efficiency (kcat/Km) or turnover rate (kcat) can be used to probe the substrate specificity of the enzyme. AMK is different from KAN in the modification of N-1 of the 2-deoxystreptamine ring with a 2-hydroxybutyryl amine (HBA) group. ISP differs from KAN with two modification sites, the one is a 2-hydroxypropionyl amine (HPA) group on the N-1 site of 2-deoxystreptamine ring and the other is cystamine conjugated to 2-deoxystreptamine ring on C6-O, instead of the 3-deoxy-3-aminoglucose ring in KAN. AMK and ISP that showed lower substrate affinity and had a less effective catalytic rate of inactivation. These results suggested that only substitutions in the 6-aminohexose ring do not affect the binding to AAC(69)-Iag. Both the OH groups on the 6aminohexose and 1-N-acylation on 2-deoxystreptamine appeared to influence the activity of acetyl transfer on the AAC(69)-Iag region specifically.
Likewise, E. coli XL-II harboring AAC(69)-Iag was resistant to ABK similar to the E. coli DH5aharboring AAC(69)-Iaj (8 mg/ml and 4 mg/ml, respectively). This suggests AAC(69)-Iag can inactivate ABK like AAC(69)-Iaj. Among the tested AGs, ABK showed significantly higher Kcat and Kcat/Km values with a low steady-state affinity, indicating lower stability levels to AAC(69)-Iag. Hence highly efficient acetylation of ABK by AAC(69)-Iag may contribute to a moderate level resistance to ABK even though the acetyl-ABK retain ca.10% antibiotic activity to ABK.
In conclusion, we report here the identification of a novel aminoglycoside 69-N-acetyltransferase gene aac(69)-Iag from a class 1 integron of a bla IMP-1 -containing MDRP strain. Retrospectively, the ten month-prevalence of an MDRP clone in a hospital appeared to result in the genesis of NK0009 possibly by horizontal transfer of the bla IMP-1 -containing plasmid. Frequent monitoring and control for the presence and spread of P. aeruginosa in hospitals is important to prevent spread of new resistant strains in the hospital setting.