Characterization and plasmid elimination of NDM-1-producing Acinetobacter calcoaceticus from China.

The presence of multidrug-resistant bacterial pathogens in the environment poses a serious threat to public health. The opportunistic Acinetobacter spp. are among the most prevalent causes of nosocomial infections. Here, we performed complete genome sequencing of the Acinetobacter calcoaceticus strain XM1570, which was originally cultivated from the sputum of a patient diagnosed with pneumonia in Xiamen in 2010. We identified carbapenem resistance associated gene bla(NDM-1) located on a 47.3-kb plasmid. Three methods--natural reproduction, sodium dodecyl sulfate treatment and nalidixic acid treatment--were used to eliminate the bla(NDM-1)-encoding plasmid, which achieved elimination rates of 3.32% (10/301), 83.78% (278/332), and 84.17% (298/354), respectively. Plasmid elimination dramatically increased antibiotic sensitivity, reducing the minimum bacteriostatic concentration of meropenem from 256 µg/ml in the clinical strain to 0.125 µg/ml in the plasmid-eliminated strain. Conjugation transfer assays showed that the bla(NDM-1)-containing plasmid could be transferred into Escherichia coli DH5α:pBR322 in vitro as well as in vivo in mice. The bla(NDM-1) genetic environment was in accordance with that of other bla(NDM-1) genes identified from India, Japan, and Hong-Kong. The multilocus sequence type of the isolate was identified as ST-70. Two novel genes encoding intrinsic OXA and ADC were identified and named as OXA-417 and ADC-72. The finding of bla(NDM-1) in species like A. calcoaceticus demonstrates the wide spread of this gene in gram-negative bacteria which is possible by conjugative plasmid transfer. The results of this study may help in the development of a treatment strategy for controlling NDM-1 bacterial infection and transmission.

Introduction

The widespread clinical use of antibiotics has induced rapid evolution of the amount and types of multidrug-resistant bacteria. Antibiotic resistance, which is acquired mainly by conjugative plasmid DNA transfer, is a threat to global health. In 2009, New Delhi metallo-β-lactamase (NDM-1), which confers high carbapenemase activity, was first detected in clinical samples from a Swedish patient of Indian origin in New Delhi, India [1]. Bacteria carrying the NDM-1-encoding bla NDM-1 gene are resistance to a broad spectrum of β-lactam antibiotics, including carbapenems, which are frequently used to treat infections caused by antibiotic-resistant gram-negative bacteria [2]–[4]. Most bla NDM-1 genes are located in plasmids [5] and bacterial conjugation is a key lateral DNA transfer mechanism, allowing easy transmission between different bacterial pathogens. The Swedish patient was found to be simultaneously infected with Klebsiella pneumoniae and Escherichia coli, both of which carried a plasmid containing the bla NDM-1 gene [1].

Acinetobacter spp. are nosocomial pathogens that have become increasingly common worldwide over the past few decades and account for the majority of human infections and hospital outbreaks [6]. Relevant species have been grouped together as the Acinetobacter calcoaceticus–Acinetobacter baumannii complex (A. baumannii, A. calcoaceticus, A.pittii and A. nosocomialis). However, it is difficult to identify these species clearly by routine diagnostic methods [7]. Clinical isolates of the A. calcoaceticus–A. baumannii complex are frequently resistant to last-line antibiotics, including cephalosporins, aminoglycosides, fluoroquinolones, and carbapenems [8], [9].

The lack of efficient drugs against multidrug-resistant pathogens poses a serious threat to public health. One feasible method to address this threat is to eliminate the drug-resistance plasmid, as such plasmids may be occasionally lost during natural reproduction or may be eliminated by exposure to physical or chemical treatments [10]–[12]. Here, we characterized a NDM-1-producing A. calcoaceticus isolate from China using whole genome sequencing and plasmid elimination assays.

Results

Basic Identification and Antimicrobial Susceptibility Testing

The multidrug-resistant pathogen XM1570 was isolated from a sputum culture and analyzed using the Vitek 32 microbial identification/susceptibility system (bioMérieux Inc., Durnam, NC, USA) with the Panel GNI+, which identified the isolate as a member of the A. calcoaceticus–A. baumannii complex. Mass spectrometry showed that strain XM1570 could be Acinetobacter sp Genospecies 3 (score value 2.322), A. calcoaceticus (score value 1.968), or A. baumannii (score value 1.96). The Acinetobacter strain XM1570 was resistant to all tested β-lactam antibiotics, including imipenem, meropenem, ertapenem, cefazolin, ceftazidime, cefotaxime, cefepime, aztreonam, ampicillin, and piperacillin. It was also resistant to chloramphenicol, but remained susceptible to amikacin, ciprofloxacin, tetracycline, trimethoprim–sulfamethoxazole, and colistin (Table 1).

Table 1

Susceptibility results of the A. Calcoaceticus XM1570, transconjugant (the E. coli DH5α:pBR322 strain after successful transformation), E. coli DH5α:pBR322, and plasmid-cured strain.

Antimicrobial agent	MIC (μg/ml)
	XM1570	transconjugant	DH5α:pBR322	plasmid-cured strain
Imipenem	>8 (R)	>8 (R)	< = 1 (S)	< = 1 (S)
Meropenem	>8 (R)	>8 (R)	< = 1 (S)	8 (I)
Cefazolin	>16 (R)	>16 (R)	>16 (R)	>16
Ceftazidime	>16 (R)	>16 (R)	< = 1 (S)	16 (I)
Cefotaxime	>32 (R)	>32 (R)	< = 1 (S)	32 (I)
Cefepime	>16 (R)	>16 (R)	< = 2 (S)	8 (S)
Aztreonam	>16 (R)	< = 2 (S)	< = 2 (S)	>16(R)
Ampicillin	>16 (R)	>16 (R)	>16 (R)	>16
Piperacillin	>64 (R)	>64 (R)	>64 (R)	32 (I)
Amoxicillin-Clavulanate	>16/8 (R)	>16/8 (R)	>16/8 (R)	>16/8(R)
Ampicillin-Sulbactam	>16/8 (R)	>16/8 (R)	>16/8 (R)	8/4 (S)
Piperacillin/Tazobactam	>64/4 (R)	>64/4 (R)	>64/4 (R)	32/4 (I)
Colistin	< = 0.5 (S)	< = 0.5 (S)	< = 0.5 (S)	< = 0.5 (S)
Trimethoprim-Sulfamethoxazole	< = 0.5/9.5 (S)	< = 0.5/9.5 (S)	< = 0.5/9.5 (S)	1/19 (S)
Chloramphenicol	>16 (R)	< = 4 (S)	< = 4 (S)	>16
Ciprofloxacin	< = 0.5 (R)	< = 0.5 (S)	< = 0.5 (S)	< = 0.5 (S)
Levofloxacin	< = 1 (S)	< = 1 (S)	< = 1 (S)	< = 1 (S)
Moxiflocacin	< = 1 (S)	< = 1 (S)	< = 1 (S)	< = 1
Tetracycline	< = 2 (S)	>8 (R)	>8 (R)	< = 2 (S)
Amikacin	< = 8 (S)	< = 8 (S)	< = 8 (S)	< = 8 (S)
Gentamicin	< = 2 (S)	< = 2 (S)	< = 2 (S)	< = 2 (S)
Δimipenem	256	128	0.25	0.0625
Δmeropenem	256	256	0.0625	0.125
Δertapenem	>512	128	0.03125	2

Note: MIC  =  minimum inhibitory concentration, R  =  resistant, S =  susceptible, I =  intermediate. MICs are based on the BD automated microbiology system (Panel NMIC/ID-4). ΔMICs are based on the broth microdilution susceptibility method following the standards of the Clinical Laboratory Standards Institute (document no. M100-S22, 2012).

Molecular and Genomic Analyses

PCR screening and sequencing revealed the presence of bla NDM-1 in strain XM1570. The genome of strain XM1570 was completely sequenced (Genbank accession number: AMXH01000000). We obtained 87 contigs with a total length of 4,076,308 bp, including one 47.3-kb circular contig identified as a bla NDM-1- containing plasmid (AMXH01000087.1) and another 93.8-kb circular plasmid (CM001803.1). A total of 3,959 coding sequences were identified on the contigs. The bla NDM-1 region in the plasmid was located in an island of medium G+C content (40.83%); the average G+C content of the strain was 38.82%.

Sequence analysis of the gyrB gene showed that strain XM1570 had the highest homology with A. calcoaceticus PHEA-2 (98%). Analysis of ropB showed that the strain had the highest homology with Acinetobacter genomosp. 3 strain CIP 70.15 (99%) and A. calcoaceticus PHEA-2 (98%). The complete genome of strain XM1570 showed ∼96% nucleotide homology to A. calcoaceticus PHEA-2 (CP002177.1). Taken together, these results indicated that XM1570 should be designated as species A. calcoaceticus.

The gene encoding the intrinsic OXA enzyme was identified in the whole genome sequence, with inclusion of a fragment of insertion sequence ISAba22 after the first 470 nucleotides of the bla OXA gene. The length of the inserting sequence was 1277 bp. The OXA-type was assigned OXA-417 (http://www.lahey.org/studies/; KM220587), which represented the bla OXA-213-like gene of A. calcoaceticus and shared 89.1% identity (30 amino acid substitutions) with bla OXA-213 protein (JN861781.1). Besides OXA and NDM-1, we identified a novel ADC gene in the genome (chromosome), which encoded an enzyme with a moderate degree of variability that was named bla ADC-72 (KJ885607). Sequence analysis of the bla ADC-72 gene showed that strain XM1570 shared 95% homology with A. calcoaceticus PHEA-2. The closest match was ADC-42 from an A. pittii isolate from Taiwan (99%) [13]. There were no ISAba sequences upstream of the bla ADC-72 gene. Multilocus sequence typing (MLST) of strain XM1570 revealed sequence type ST-70.

Plasmid Characterization

We attempted to use polymerase chain reaction (PCR) to identify the replicon type of the plasmid harboring the bla NDM-1 gene [14], [15]. However, our attempts were unsuccessful. We made comparisons with the replicon sequences of Enterobacteriaceae [16], but could not confirm the replicon type of the plasmid sequence harboring the bla NDM-1 gene (AMXH01000087.1). The plasmid lacked known replicon genes indicating that it may has an unknown replication mechanism. The search of the 22 replicase genes in the whole genome sequence indicated that replicase gene repB was encoded by the other (90 kb) plasmid of strain XM1570 (CM001803.1), which had 100% nucleotide identity to p3ABAYE0002 (A. baumannii; NC010404.1). Furthermore, the aci9 gene was found in the bacterial genome sequence, showing ∼87% nucleotide identity to that of the plasmid pA21 (A. baumannii; GU979001.1). Sequence analysis showed that the plasmid (CM001803.1) contained no antibiotic resistance genes but did contain other genes encoding proteins such as alcohol dehydrogenase, alkaline phosphatase, DNA-binding helix-turn-helix protein, inner membrane protein, Sel1 repeat protein, transcriptional repressor, integrase, site-specific recombinase, and transposase.

The bla NDM-1 genetic environment was systematically identified in our isolate, which was composed of the complete ISAba125 fragment containing the InsI1 transposase gene, the bla NDM-1 gene, the bleomycin resistance gene ble MBL, and the phosphoribosylanthranilate isomerase gene trpF. The bla NDM-1 gene exhibited 100% homology with that of the Hong Kong isolate HK-1 (E. coli) [17], and contained a 24-bp insertion sequence between the bla NDM-1and trpF genes compared with the India isolate KP05-506 (K. pneumoniae). We compared the genetic structures surrounding the bla NDM-1 of our isolate with that of isolates from India, Japan, and Hong-Kong (Figure 1), which showed that at least a remnant of the insertion sequence ISAba125 was present upstream of the bla NDM-1 gene. The entire ISAba125 element was only identified in our isolates, as the remnant ISAba125 sequences of all others were disrupted by different mobile elements (ISEc33, IS5, IS903 etc.), which contained the relevant transposase gene. Downstream of the bla NDM-1 gene, ble MBL was systematically identified in most isolates, with the exceptions of FN396876 and HQ451074, and the trpF gene was identified in all isolates, which encode the isomerase that catalyzes the third step of synthesis of the amino acid tryptophan. As an isomerase, it can rearrange parts of the molecule without additions or deletions, indicating that the trpF gene was a vital component of the bla NDM-1 module.

Figure 1

Schematic map representing the bla NDM-1-surrounding genetic sequences in the A. calcoaceticus XM1570.

The bla NDM-1 genetic environment was compared with previously described structures of other NDM-1 plasmids (p271A, pABC40, pNDM-1_Dok01, pKpANDM, pNDM-HK, and pABC85). Different insertion elements were identified among these Enterobacteriaceae. Genes and their transcription orientations are represented by arrows. Δ indicates the truncated genes.

Conjugation and Transformation of bla NDM-1

The frequencies of transfer in vitro were ∼10−3 for E. coli DH5α:pBR322. Mass spectrometry (data not shown) identified the bacteria that grew on agar plates containing meropenem and tetracycline as E. coli DH5α. PCR demonstrated successful transfer of the bla NDM-1 gene from A. calcoaceticus to DH5α, and sequencing results confirmed the absence of genetic mutation within the gene. Plasmid isolation revealed that the transconjugant acquired only the bla NDM-1-containing plasmid (47.3kb size; AMXH01000087.1; data not shown). The E. coli F25 strain failed to grow under these conditions.

A. calcoaceticus (donor) and DH5α:pBR322 or F25:pBR322 (recipients) were cocultivated in vivo in mice. The discharged bacteria curves for A. calcoaceticus, DH5α, and the transconjugants showed that the number of discharged bacteria peaked at 2 h post-transfer. The frequencies of transfer in vivo for DH5α: pBR322 were ∼10−6. Mass spectrometry, PCR and sequencing identified the bacteria as E. coli DH5α harbouring the bla NDM-1-bearing plasmid. Consistent with our in vitro results, there was no evidence of in vivo transformation of the F25 recipient strain. The minimum bacteriostatic concentration of meropenem was 0.0625 µg/ml for the DH5α:pBR322 recipient and 256 µg/ml for DH5α:pBR322 successfully transferred in vitro and in vivo. The E. coli DH5α:pBR322 strain after successful transformation acquired drug resistance to meropenem, imipenem, ertapenem, ceftazidime, cefotaxime, and cefepime. Resistance to antibiotics other than β-lactams was not transferred to the transconjugant.

Plasmid Elimination Analysis

The three methods used to eliminate bacterial plasmids were natural reproduction, sodium dodecyl sulfate (SDS) treatment, and nalidixic acid treatment. The plasmid-elimination rates were 3.32% (10/301) for natural reproduction, 83.78% (278/332) for SDS treatment, and 84.17% (298/354) for nalidixic acid treatment. The absence of the bla NDM-1-containing plasmid in elimination strains was confirmed by PCR and plasmid isolation. All plasmid-cured strains did not lose the other 93.8 kb plasmid (CM001803.1; data not shown). Antibiotic sensitivity was dramatically increased in elimination strains, which all exhibited a minimum bacteriostatic concentration for meropenem of 0.125 µg/ml, compared to 256 µg/ml for the clinical strain. The plasmid-cured strains lost resistance to carbapenem and some other β-lactams, including ceftazidime, cefotaxime, and cefepime, but maintained resistance to cefazolin, aztreonam, and ampicillin.

Discussion

Currently, the majority of NDM-producing Acinetobacter spp. have been isolated in China and the Middle East [18], [19]. Most of the isolates were identified as A. baumannii. Other NDM-producing Acinetobacter spp. have been identified in Chinese isolates, including A. pittii, A. lwoffii, and A. junii [19]–[21]. In the present study, NDM-1-production in A. calcoaceticus was identified for the first time.

The inappropriate use of β-lactamase antibiotics is causing an increasing public health threat of multidrug-resistant bacteria and poses a considerable challenge to the treatment of clinical infections. The development of appropriate methods to eradicate drug-resistant bacteria has become a focus of recent research. Performance of 3 different plasmid-elimination tests to remove the antibiotic resistance of A. calcoaceticus showed that the methods with SDS and nalidixic acid treatment were more efficient than the natural reproduction elimination assay. The plasmid-eliminated strains had completely lost resistance to carbapenem and some other β-lactams, including ceftazidime, cefotaxime, and cefepime.

The presence of carbapenem hydrolyzing class D β-lactamases is the most important factor mediating carbapenem resistance in Acinetobacter spp. The carbapenemase genes bla OXA-58 and bla OXA-23 are distributed worldwide. The new intrinsic A. calcoaceticus enzyme OXA-417 gene was identified in the genome of our isolate. The coding sequence of the bla OXA-417 gene was prematurely terminated by insertion of ISAba22, which resulted in inactivation of the protein carbapenem-hydrolyzing oxacillinase [22]. Intrinsic bla OXA genes, e.g. bla OXA-51 in A. baumannii only cause carbapenem resistance if an insertion sequence is present upstream (ISAba1-bla OXA-51). Given that no other insertion sequences are located upstream the gene, it is not clear if it can mediate carbapenem resistance even when functional.

Conjugation experiments by co-cultivation of the A. calcoaceticus strain with E. coli DH5α:pBR322 in vitro and in vivo in mice showed that the plasmid carrying the bla NDM-1 gene was successfully transferred to the recipient DH5α:pBR322 in vitro and in vivo. The minimum bacteriostatic concentration of meropenem for transduced bacteria was greatly increased, but was less than that of strain XM1570, suggesting that the degree of resistance was also dependent on physiological metabolic properties of the bacteria or by the contribution of additional genes to beta-lactam resistance, e.g. bla ADC [23].

NDM-1 can efficiently hydrolyze a broad range of β-lactams, but not monobactams, such as aztreonam [18]. In our study, after successful transfer of the bla NDM-1-carrying plasmid, E. coli strain DH5α:pBR322 remained susceptible to aztreonam. The plasmid-cured strain maintained resistance to aztreonam. An increasing number of reports have focused on Acinetobacter-derived cephalosporinases (ADCs), which are common among Acinetobacter spp. [24]. Some ADC β-lactamases had lower MICs of cefotaxime and aztreonam as well. Aztreonam is a weak inducer and weak substrate, but it can be hydrolyzed if sufficient amounts of ADC β-lactamases are produced [25]. Some ADC β-lactamases (ADC-33) can hydrolyze high levels of ceftazidime, cefepime, and aztreonam, which encouraged the classification of this enzyme as an extended-spectrum AmpC (ESAC) [23]. Because of the insertional inactivation of bla OXA-417, we speculated that bla ADC-72 conferred resistance to aztreonam, cefazolin, and ampicillin in the plasmid-cured strain.

This is the first report of NDM-1 -positive A. calcoaceticus from China. Our study provides molecular evidence for the emergence and dissemination of multidrug-resistant bacteria. bla NDM-1 in A. calcoaceticus XM1570 located on a highly mobile plasmid of indeterminable replicon type, further underlining the wide spread of that resistance trait. The results of this study also suggest a potential treatment strategy that may help in controlling NDM-1 bacterial infection and transmission.

Materials and Methods

Case Report

Sputum was collected from a patient diagnosed with pneumonia treated at a hospital in Xiamen, Fujian Province, China, in 2010. The male patient was initially diagnosed at a community clinic with a cold, and treatment with injection of clarithromycin, imipenem, and cilastatin sodium had no effect. He was subsequently transferred to a hospital, where lung cancer with right pleural metastasis and pneumonia was confirmed. According to the drug-resistance pattern, the patient was treated with levofloxacin via intravenous drip. The symptoms caused by pneumonia were attenuated after 1 week of treatment.

Bacterial Strains

A sputum culture yielded an Acinetobacter strain resistant to β-lactam antibiotics and carbapenems. The multidrug-resistant pathogen XM1570 was analyzed using the Vitek 32 microbial identification/susceptibility system (bioMérieux Inc., Durnam, NC, USA) with the GNI+ panel and matrix-assisted laser desorption ionization time-of-flight mass spectrometry.

Antibiotic Susceptibility Testing

Susceptibilities of the A. calcoaceticus XM1570, transconjugants (DH5α:pBR322 strain after successful transformation), transformants, and plasmid-cured strains were tested using the BD automated microbiology system (Panel NMIC/ID-4) (see Table 1). Susceptibility of these bacterial strains to imipenem, meropenem, and ertapenem was assessed using the broth microdilution susceptibility method as defined by the Clinical and Laboratory Standards Institute (CLSI) document no. M100-S22, 2012 [26].

Verification of bla NDM–1

A set of primer pairs was designed based on the open reading frame sequence of bla NDM–1 (product length, 831 bp): 5′-ATG GAA TTG CCC AAT ATT ATG CA-3′ (NDM-F) and 5′-TCA GCG CAG CTT GTC GGC CAT-3′ (NDM-R). The primers were synthesized by Beijing Genomics Institution. PCR-amplification was carried out in a total reaction volume of 50 µl containing 50 pmol of each primer and 2 µl of template (TaKaRa Bio, Inc., Osaka, Japan), as recommended by the manufacturer. The following specific cycling parameters were used: initial denaturation at 94°C for 3 min, followed by 35 cycles of denaturation at 94°C for 30 s, annealing at 53°C for 30 s, and extension at 72°C for 1 min, and a final extension step at 72°C for 7 min.

Whole Genome Sequencing, Assembly and Analysis

Bacterial cells were collected by centrifugation at 6,000×g for 15 min and genomic DNA was extracted using a Wizard Genomic DNA purification kit (Promega, Madison, WI, USA) according to the manufacturer's instructions. Complete genomic sequencing was carried out at the Beijing Institute of Genomics, Chinese Academy of Sciences, using a HiSeq 2000 system (Illumina, Inc., San Diego, CA, USA). Ultimately, 410 and 210 Mb of high-quality data was generated from two libraries. In both cases, read lengths were 90 bp. The paired-end reads were first assembled de novo using a Short Oligonucleotide Analysis Package (SOAP denovo v1.05). A. calcoaceticus strain XM1570 contigs were next manually connected by reference to paired-end relationships at the 2-kb level. Prior to functional annotation, putative protein-coding sequences were identified using Glimmer 3.0 online (www.ncbi.nlm.nih.gov/genomes/MICROBES/glimmer_3.cgi) and homologies to sequences in publically available databases were identified using the Basic Local Alignment Search Tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Gaps in plasmid sequences were bridged by PCR and confirmed by capillary sequencing. Sequences were submitted to Genbank (AMXH01000000). The MLST scheme was based on the allelic variations in seven housekeeping genes (cpn60, fusA, gltA, pyrG, recA, rplB, and rpoB) of the Acinetobacter whole genome sequences [27]. Allele sequences and sequence types (STs) were identified at http://www.pasteur.fr/recherche/genopole/PF8/mlst/Abaumannii.html.

To test the transferability of the bla NDM-1 gene, two recombinant bacteria (E. coli DH5α:pBR322 and E. coli F25:pBR322) were constructed and the conjugation-transfer assay was performed with A. calcoaceticus containing bla NDM-1 as the donor and E. coli DH5α or E. coli F25 as the recipient. For the in vitro conjugation and transformation assay, the donor and recipient bacteria were mixed at a ratio of 1:2 and then filtered through a microfiltration membrane, which was cultured in solid medium overnight with the cover containing the bacteria facing upward. Transconjugants were selected on solid lysogeny broth (LB) medium containing meropenem (16 µg/ml) and tetracycline (15 µg/ml), and verified by mass spectrometry (bacterial species) and PCR (bla NDM-1).

In vivo conjugation and transformation assays were performed in mice. The mice were denied food for 12 h and then fed donor and recipient bacteria mixed at a ratio of 1∶2. Gastric acid was neutralized with sodium hydrogen carbonate. Afterwards, the mice had access to food ad libitum. Mouse droppings were collected at 1, 2, 3, 4, and 5 h for analysis of excreted bacteria. The droppings were suspended in physiological saline. Transconjugants were selected by culturing suspensions on solid LB medium containing meropenem (16 µg/ml) and tetracycline (15 µg/ml) at suitable dilutions. Plasmid transfer was verified by mass spectrometry (bacterial species), PCR (bla NDM-1), and nucleotide sequencing. Antimicrobial resistance of the donor, recipient, and transconjugant strains was tested using the Phoenix100 system (BD Biosciences, San Jose, CA, USA) with the CLSI Broth Microdilution Susceptibility Method.

Plasmid Elimination Study

Elimination experiments were carried out in broth culture. Plasmid curing was performed as described previously [28] with slight modifications. For the natural reproduction method, 50 µl of A. calcoaceticus was inoculated into 5 ml of LB medium and incubated at 37°C for 8 h. Then, 50 µl of bacterial suspension was inoculated into 5 ml fresh LB medium at 37°C and incubated for 8 h. This latter procedure was repeated five times. After serial dilution, the bacterial suspension was inoculated onto solid LB medium at 37°C and incubated for 12 h. All colonies were inoculated onto solid LB medium with or without meropenem, and incubated at 37°C for 12 h. Plasmid elimination was calculated as the difference in the amount of growth on the two media divided by the amount of growth on solid LB medium.

For the SDS treatment method, 50 µl of A. calcoaceticus suspension was inoculated into five replicate samples of 5 ml of LB medium, to which 200, 100, 50, 25 or 12.5 µl of 10% SDS was added and then incubated at 37°C for 8 h. An SDS concentration one-half of that in which growth was first observed was chosen as the working concentration for the elimination test. The following steps were then performed: (1) 50 µl of A. calcoaceticus was inoculated into 5 ml of LB medium containing SDS and incubated at 37°C for 12 h; (2) 50 µl of bacterial suspension was inoculated into 5 ml of fresh LB medium and incubated at 43°C for 8 h; (3) repeat step 1; (4) repeat step 2; and (5) repeat step 1.

For the nalidixic acid treatment method, 50 µl of A. calcoaceticus suspension was inoculated into five replicate samples of 5 ml of LB medium, to which 10, 5, 2.5, 1.25 or 0.625 µl of 50 mg/ml nalidixic acid was added and the samples were incubated at 37°C for 12 h. A nalidixic acid concentration one-half of that in which growth was first observed was chosen as the working concentration for the elimination test. Then, 50 µl of A. calcoaceticus suspension was inoculated into 5 ml of LB medium and incubated at 37°C for 8 h. This procedure was repeated five times. Plasmid elimination was calculated as for natural reproduction. Plasmids of A. calcoaceticus XM1570 and the plasmid-cured derivative strain were isolated using the Qiagen Large-Construct Kit (Qiagen, Hilden, Germany). The absence of plasmids in elimination strains was confirmed by PCR. Antibiotic-resistance profiles of plasmid-harboring and plasmid-cured strains were tested using the BD Phoenix100 system. The MICs of carbapenems were confirmed by the CLSI Broth Microdilution Susceptibility Method [26].

Ethics Statement

The study protocol was approved by the Review Board of the Academy of Military Medical Sciences and Beijing Institute of Genomics. The patient provided oral informed consent because he was unable to write. The oral consent was documented by the attending physician. This consent procedure was approved by the Review Board of the Academy of Military Medical Sciences and Beijing Institute of Genomics. All animal experiments were conducted in accordance with the accepted standards of the Animal Care and Use Committee of Academy of Military Medical Sciences, China. The animal study protocol was approved by the Animal Care and Use Committee of Academy of Military Medical Sciences, China. All efforts were made to minimize suffering.