IMP-1 encoded by a novel Tn402-like class 1 integron in clinical Achromobacter xylosoxidans, China

Achromobacter xylosoxidans strain A22732 is isolated from a pneumonia patient in China and produces carbapenemases OXA-114e and IMP-1, which are encoded by chromosome and plasmid, respectively, and confer resistance to multiple ß-lactam antibiotics including carbapenems. The blaIMP-1 gene together with aacA7 and orfE is captured by a novel Tn402-like class 1 integron in a conjugative IncP-1ß plasmid. In addition to the intrinsic integron promoter PcW, there is still a blaIMP-1 gene cassette-specific promoter. This is the first report of carbapenemase-encoding IncP-1ß plasmid in clinical bacterial isolate.

negative E. coli transconjugant, designated A22732-IMP-EC600, was obtained by conjugal transfer, indicating that A22732 harbored a conjugative IMP-encoding plasmid, which was designated pA22732-IMP. As determined by a modified CarbaNP test 6 , strain 22732-IMP-EC600 had class B carbapenemase activity, while A22732 probably expressed class B/D carbapenemases (Fig S1), being consistent with the above PCR/sequencing results.
The minimum inhibitory concentration (MIC) values (Table 1) were determined for A22732, 22732-IMP-EC600, and EC600. A22732 and 22732-IMP-EC600 show almost identical drug resistance profiles. These two strains are highly resistant to penicillins, aztreonam, and ephalosporins tested, but they remain susceptible to the fluoroquinolone, furane, aminoglycoside, and sulfanilamide drugs tested. Both A22732 and 22732-IMP-EC600 are resistant to imipenem and meropenem, but the MIC values of imipenem and meropenem (both .516) against A22732 are much higher than those (4 and 2, respectively) against 22732-IMP-EC600, which is consistent with the fact that A22732 expresses two different arbapenemase IMP-1 and OXA-114e while 22732-IMP-EC600 produces only a single one IMP-1.
The entire pA22732-IMP sequence had .70% query cover and .80% identity with the above IncP-1ß plasmids. pA22732-IMP was aligned with four representative plasmids R751 (accession number U67194) 8 , pB10 9 (AJ564903), pB8 (AJ863570) 10 , and pAKD31 (JQ436721) 11 through pairwise whole plasmid sequence comparison (Fig 2). R751 is from a clinical Enterobacter aerogenes isolate and represents the prototype IncP-1b drug-resistance plasmid, pB10 is isolated from a bacterial community residing in activated sludge compartment of a waste-water treatment plant, and pB8 and pAKD31 are from agricultural soil bacteria. The above IncP-1ß plasmids were selected for further analyses, because they harbored different forms of Tn501-like or Tn402-like elements (see below).
The backbone regions are highly conserved in pA22732-IMP, R751, pB10, pB8, and pAKD31. All the former four plasmids harbor two accessory modules, namely Tn501-like element and Tn402-like integron, each of which has different gene organizations among these four plasmids (Fig 2). The above two accessory modules are inserted into downstream of trfA and that of traC2, respectively, which are the targeting locations typically interrupted by mobile elements in IncP-1b plasmids. By contrast, only Tn501-like element rather than Tn402-like integron is found in the corresponding region of pAKD31 (Fig 2).
Tn501-like loci can be identified in all the above five plasmids (Fig 3). The Tn501-like element of pB10 is flanked by a 5-bp target site (direct repeats of TGCCT), but those of all the other four plasmids leave no trace of insertion.
pB10 harbors an greatly extended Tn501-like mer locus, because tnpA is interrupted by IS1071 (which is further interrupted by Tn1721-like tetracycline-resistance locus) plus a Tn5393c-like streptomycin resistance locus, thereby leading to partial deletion of tnpA. Loss of IRi and tnpAR as well as 3'-terminus deletion of orf2, because of insertion of IS1071 plus IS21 into orf2, is observed for the Tn501like mer locus of pAKD31. The Tn501-like mer locus of pA22732-IMP has undergone loss of IRi, tnpAR, and orf2 as well as an inversion event of the whole locus.
Both R751 and pB8 carry a 'cryptic' Tn501-like element, which is composed of IRi, orf1, orf2, a merR remnant and IRt in the absence of all the other features identified for the ancestral Tn501-like mer locus. In addition, R751 has acquired two copies of IS4321 (IS4321L and IS4321R), which flank the merR remnant and orf2, respectively. By contrast, evolution of pB8 involves insertion of a Tn501-like quaternary ammonium compound resistance (qacF) locus into orf2, thereby disrupting this gene.
The pB8 and pB10 integrons, each of which contain 59-CS, drug resistance gene cassettes, 39-CS, and completely or partially truncated tni module, appear to undergone all the above three steps of evolution. By contrast, the evolution step II (generation of 39-CS) is mostly likely omitted for the pA22732-IMP integron, while the R751 integron might represent a primitive Tn402-like integron due to absence of evolution steps II and III (truncation of tni module).
The Tn402-like integron of pA22732-IMP is inserted into the traC-parA intergenic region, leaving parA and its downstream gene upf31.0 intact. The parA upstream or around region represents a hot spot for Tn402 targeting, most likely due to it contains the multimer resolution site II for recognition by Tn402 transposase 20,21 . Interestingly, the insertion of Tn402-like integrons into the hotspot target region leads to further deletion of parA from pB8 and R751, and that of parA/upf31.0 from pB10.
The pA22732-IMP, R751 and pB8 integrons contains IRi and IRt of Tn402, but only IRi rather than IRt is identified for pB10. The lack of IRt in pB10 might due to the above-mentioned deletion removing of parA/upf31.0. The 5-bp target site (direct repeats of AGCAT) is still intact to flank the pA22732-IMP integron, but all the other three integrons do not leave traces of insertion.
Integrase IntI1 recognizes two different types of recombination site attI1 (integron attachment site) and attC (recognition site for integrase), and it catalyzes integration or excision of gene cassettes through site-specific recombination commonly between one attI1 site and one or more attC sites 22,23 . One attI1 site and three attC sites, upstream of aacA7, orfE, bla IMP-1 and tniA, respectively, are indentified in the pA22732-IMP integron (Fig 4a). These sites are long inverted-repeat-containing sequences of variable length and sequence, and each inverted repeat begins with a core sequence RYYYAAC and ends with an inverted core sequence GTTRRRY as described previously 24,25 , which would form imperfect cruciform structures and be required for capture of aacA7, orfE, and bla IMP-1 .
Expression of integron gene cassettes. The pA22732-IMP integron gene cassettes aacA7, orfE and bla IMP-1 , but not intI1 and tniA, are organized in the same transcriptional direction (Fig 5a). PCR generates an amplicon ranged from 59-untranslated region (59-UTR) of aacA7 to 59-terminal of the bla IMP-1 coding region, when using A22732 genomic DNA as template (Fig 5a). The positive RT-PCR amplification with the same primer pair, using cDNA sample generated from A22732 total RNA as template (Fig 5b), indicates that aacA7, orfE, and bla IMP-1 are transcribed into a single RNA transcript and thereby, these three genes constitutes a single operon aacA7-orfE-bla IMP-1 (Fig 5a). two broken-line arrows represent primary RNA transcripts transcribed for the aacA7-orfE-bla IMP-1 operon and the bla IMP-1 gene, respectively. Line with filled circles at both termini indicates location of primer pair plus expected PCR amplicon. b) PCR and RT-PCR. cDNAs generated from total RNA of strain A22732, and genomic DNA of A22732 were used as templates for RT-PCR and PCR, respectively. c) Primer extension. Primer extension assay of the RNA transcript of aacA7 or bla IMP-1 was done for A22732 cultured with addition of increasing amounts of imipenem. Lanes C, T, A and G represent Sanger sequencing reactions. The transcription start of aacA7 or bla IMP-1 is indicated by the arrow with nucleotide T or G, respectively, and the minus number under arrow indicate the nucleotide position upstream of the aacA7 or bla IMP-1 start codon. Representative data from at least two independent biological replicates are shown. Integrons act as natural gene expression platforms due to the presence of an intrinsic promoter (Pc) that is recognized by RNA polymerase to drive transcription of inserted gene cassettes that generally do not have their own promoters. At least eight distinct types of Pc promoter, PcS (strong, TTGACA-N 17 -TAAACT), PcW (weak, TGGACA-N 17 -TAAGCT), PcH1 (hybrid 1, TGGACA-N 17 -TAAACT), PcH2 (Hybrid 2, TTGACA-N 17 -TAAGCT), PcSS (super-strong, TTGATA-N 17 -TAAACT), PcIn42 (TTGGCA-N 17 -TAAACT), PcIn116 (TTGACA-N 17 -TGAACT), and PcPUO (TCGACA-N 17 -TAAACT) have been described for class 1 integrons 26 . As detected by primer extension (Fig 5c), a transcription start site (nucleotide C) is located at 227-bp upstream of aacA7 start codon (i.e. a 227-bp 59-UTR of aacA7), validating presence of the PcW promoter to drive aacA7-orfE-bla IMP-1 transcription. The PcW promoter can be found for all the pA22732-IMP, pB8, R751, and pB8 integrons.
Remarkably, the primer extension assay detects another transcription start site (nucleotide G) which is located at 155-bp upstream of bla IMP-1 start codon. This assay discloses presence of an internal promoter (TTGCCA-N 16 -TATCAT) driving transcription of the bla IMP-1 gene cassette (Fig 5c). Being very rare, the internal gene cassette-specific promoter is an extra element in evolution of antimicrobial resistance phenotype and act independent of Pc promoter 27 .
In addition, the primer extension assay shows that addition of increasing amounts of imipenem during bacterial cultivation has no effect on promoter activity of either aacA7-orfE-bla IMP-1 or bla IMP-1 (Fig 5c), validating constitutive expression of the aacA7-orfE-bla IMP-1 operon and the bla IMP-1 gene cassette. Notably, constitutive transcription of integron gene cassettes has been suggested previously 23,28,29 .
Concluding remarks. We present the first complete sequence of IMP-encoding plasmid from A. xylosoxidans, and this is also the first report of identification of a carbapenemase-encoding IncP-1ß plasmid from a clinical bacterial isolate. The detected bla IMP-1 gene is captured by a novel class 1 integron with a novel gene cassette array in the IncP-1ß plasmid pA22732-IMP. The class 1 integron is embedded in a Tn402-like transposon and inserted into pA22732-IMP by transposition. Most of the characterized IncP-1ß plasmids are isolated from bacteria in agricultural soils or waters and frequently associated with mercury resistance 7,[9][10][11]13 . Only a few of them, e.g. pA22732-IMP and R75, are of clinical origins and harbor an array structure of multiple resistance gene cassettes. The IncP-1ß plasmids thus could represent important vehicles for spreading clinically relevant resistance determinants across a number of bacterial species. The transfer of bla IMP to A. xylosoxidans, which already carried bla OXA-114 intrinsically, would lead to more severe drug resistance, making high difficulty in timely choosing sensitive antibiotics for treatment. The IMP-producing A. xylosoxidans should be taken seriously as the surveillance target especially in East Asia countries such as China and Japan.

Methods
Bacteria isolation and identification. Fresh sputum specimens were sampled from the indicated patient and inoculated onto Mueller-Hinton agar for bacterial isolation. The use of human specimens and all related experimental protocols were approved by the Committee on Human Research of Chinese People's Liberation Army General Hospital and carried out in accordance with the approved guidelines, and moreover the informed consent was obtained from the indicated patient. Single colony of each bacterial strain tested was subjective for species identification by VITEK 2 (BioMérieux), Bruker MALDI Biotyper, and 16s rRNA gene sequencing. For determination of16S rRNA gene sequence, the almost complete coding region of 16S rRNA gene was amplified by PCR with the universal primers 27f (AGAGTTTGATCCTGGCTCAG) and 1492r (TACCTTGTTACGACTT) 30 and then sequenced on ABI 3730 Sequencer.
PCR detection of bla genes. All known carbapenemase and ESBL genes as listed in Table S1 were subjected to PCR detection. Primer pair GTCCAAGACCGGCAACTC/CACCAGCAGGATCGACAG was designed from whole-genome shotgun sequences of strain A22732 (data not shown) for amplifying the DNA fragments containing the whole coding region of bla OXA-114 , because all the available primers gave negative amplification. All amplicons were sequenced on ABI 3730 Sequencer with the same primers for PCR.
Conjugal transfer. Plasmid conjugal transfer experiments were carried out with rifampin-resistant E. coli EC600 being used as recipient and bla IMP -positive A. xylosoxidans as donor. 3 ml of overnight cultures of each of donor and recipient bacteria were mixed together, harvested and resuspended in 80 ml of Brain Heart Infusion (BHI) medium. The mixture was spotted on a 1 cm 2 filter membrane that was placed on the BHI agar plate, and then incubated for mating at 37uC for 12-18 h. Bacteria were washed from the filter membrane and spotted on the Muller-Hinton agar plate containing 1500 mg/ml rifampin and 100 mg/ml ampicillin for selection of the bla IMP -positive E. coli transconjugant.

Determination of minimum inhibitory concentration (MIC).
The MIC values of indicated bacterial strains were tested by using VITEK 2 according to manufacturer's instructions, and antimicrobial susceptibility was judged by Clinical and Laboratory Standards Institute (CLSI) standard.
Determination of plasmid DNA sequence. The chromosome DNA-free plasmid DNA was isolated from the cell cultures of the bla IMP -positive E. coli transconjugant using a Qiagen large construct kit, and then sequenced by using whole-genome shotgun strategy in combination with Illumina HiSeq 2500 sequencing technology. The contigs were assembled with Velvet, and the gaps were filled through combinatorial PCR and Sanger Sequencing on ABI 3730 Sequencer. The genes were predicted with GeneMarkS and further annotated by BLASTP against Uniport and NR databases.
RNA isolation and reverse transcription (RT)-PCR. Bacteria were cultured overnight in Mueller-Hinton broth (BD) with or without addition of 2 mg/ml imipenem (Sigma). Total RNAs were extracted from harvested bacterial cells using TRIzol Reagent (Life Technologies). RNA quality was monitored by agarose gel electrophoresis, and RNA quantity was determined by spectrophotometry. The contaminated DNA in the total RNA samples was removed by using Amibion's DNA-free TM Kit. cDNAs were generated by using 5 mg of RNA and 3 mg of random hexamer primers in a 40 ml reaction mixture. The cDNA samples were generated by RT from total RNAs. Genomic DNA and cDNA were used as the templates for PCR and RT-PCR, respectively, with the primer pair TGTTTGATGTTATGGAGCAG/ AGCCGTAAATGGAGTGTC. To ensure that no contamination of genomic DNA in the RT reactions would occur, RT-PCR of negative controls was performed using the 'cDNA' sample generated without reverse transcriptase as template. Reactions containing primer pairs without templates were also included as blank controls.
Primer extension assay. The [c-32 P] ATP end-labeled primer CAGTCATAACAAGCCAT or CATACTTTTCCTTTTCTAACGG, which was complementary to aacA7 or bla IMP-1 transcript, respectively, was annealed with total RNA sample of strain A22732 for primer extension assay as described previously 31 . For different cell cultures (lanes) in a single experiment, equal amounts of total RNA were used as starting materials. The corresponding end-labeled primer was also used for sequencing the PCR amplicon generated by the primer pair TGACGATGCGTGGAGACC/CAGTCATAACAAGCCAT or TGTTTGATGTTATGGAGCAG/AGCCGTAAATGGAGTGTC. DNA sequencing was carried out using the AccuPower & Top DNA Sequencing Kit (Bioneer). Primer extension products and sequencing materials were analyzed on an 8 M urea-6% polyacrylamide gel electrophoresis. Radioactive species were detected by autoradiography.
Nucleotide sequence accession number. The complete sequence (File S1) of plasmid pA22732-IMP was submitted to the GenBank nucleotide sequence database under accession number KJ588780. www.nature.com/scientificreports