Introduction
The emergence of metallo-β-lactamases (MBL; also known as class B carbapenemases), which hydrolyze all classes of β-lactams except monobactams, in Gram-negative human pathogens represents a major threat for clinical chemotherapy and public health
[ 1]. The NDM-type carbapenemases, first identified in a strain of
Klebsiella pneumoniae isolated in 2008 in New Delhi, India
[ 2], are one of the most important MBL types because of their clinical relevance and international dissemination across all continents except Antarctica
[ 3]. To date, 16 variants of NDM (NDM-1 to NDM-16; no information on NDM-12, NDM-15 and NDM-16 is available in the GenBank database), which differ by one, two, or five amino acid substitutions at 14 positions (www.lahey.org/studies), have been identified mainly from Enterobacteriaceae isolated from nosocomial infections. In these NDM variants, the
blaNDM-1 gene is mostly located on conjugative plasmids, which facilitates its rapid dissemination
[ 4]. Detailed analysis of the genetic environment of the
blaNDM-1 gene in the chromosome or on plasmids of various Gram-negative bacterial species revealed the presence of a conserved structure comprising the complete or truncated insertion sequence element IS
Aba125 and the bleomycin (
bleMBL) resistance gene in the up- and down-stream regions of the
blaNDM-1 genes, respectively
[ 4].
Compared with the high prevalence of NDM in Gram-negative bacteria of human origin, there are only limited and sporadic reports of this type of carbapenemase in bacteria from non-human origins. In the environment, the
blaNDM-1 gene was first identified in multiple genera in 2011, including the opportunistic Enterobacteriaceae and non-fermentative Gram-negative bacteria, in the samples of both tap water and seepage water in New Deli, India
[ 5]. The
blaNDM-1 gene was then found in
K. pneumoniae isolated from a river in Vietnam
[ 6], as well as in
Acinetobacter baumannii isolated from sewage of hospitals in Beijing, China
[ 7]. Recently,
blaNDM-1-carrying
A.calcoaceticus and
A. junii were identified in environmental samples from livestock farms in China
[ 8]. In animals, a single isolate of NDM-1-producing
A. lwoffii and
A. baumannii has been reported from a chicken and a pig, respectively
[ 9, 10], and the
blaNDM-1 gene was also observed in clinical
Escherichia coli isolates recovered from cats and dogs in the USA
[ 11].
As there have only been a few reports on the carbapenemases in bacteria from food animals, we screened the clinical samples of diseased pigs from an animal diagnostic laboratory in China to investigate the presence of the carbapenem resistance gene in Gram-negative bacteria, and further characterized the genetic environment of the carbapenem resistance gene.
Materials and methods
Bacterial isolation and identification, and phenotypic and molecular detection of MBL in Gram-negative bacteria
A total of 334 lung samples were collected from the animal diagnostic laboratory of Foshan University in Foshan City, Guangdong Province, China, between July and August 2013. Each sample was isolated from an individual animal and spread on a brain heart infusion (BHI) agar plate containing 2 mg·L
−1 meropenem (Ouhe Technology Company, Beijing, China) and 30 mg·L
−1 vancomycin, then incubated for 18 h at 37°C. All colonies were selected and identified using Gram staining and sequence analysis of the 16S rDNA gene, using previously described primers
[ 12]. Species identification of Gram-negative bacteria was further conducted by MALDI-TOF MS (BrukerDaltonik GmbH, Bremen, Germany). The imipenem-EDTA double-disc synergy test and E-test using MBL strips (bioMérieux, Craponne, France) were performed to screen for MBL production in these isolates. The MBL-producing isolates were screened for known mobile MBL-encoding genes, NDM-, VIM-, SIM-, GIM-, AIM- and DIM-type β-lactamases were investigated using previously described PCR assays
[ 10].
Conjugation assay
Filter mating was performed with each of the original isolates using E. coli strain EC600 (a rifampicin-resistant strain) as the recipient. The original isolates and the recipient strains of E. coli EC600, both incubated overnight were adjusted to 0.5 McFarland standard in BHI broth, and 5 μL of parental and recipient strains were added to 1 mL of BHI broth and incubated for an additional 4 h at 37°C, respectively. Then, 20 μL of the parental strain and 60 μL of the recipient strains were incubated together on a microporous membrane (Millipore, Bedford, MA, USA) overnight for conjugation. The concentrations of antibiotics in the BHI agar plates used for the selection of transconjugants were 500 mg·L−1 for rifampin and 1 mg·L−1 for meropenem. Transconjugants were confirmed as MBL-producing isolates by PCR analysis, and pulsed-field gel electrophoresis (PFGE) was also performed to confirm that transconjugants were derivatives of the recipient strain EC600.
Antimicrobial susceptibility testing
The minimum inhibitory concentrations (MIC) of a range of antibiotics against the MBL-producing isolates, as well as their transconjugants, were determined using the broth microdilution method according to the Clinical and Laboratory Standards Institute document M100-S25 (2015) and the MIC were interpreted according to these standard
[ 13]. The
E. coli isolate ATCC25922 was used for quality control.
Molecular analysis of MBL-producing isolates
The genetic relatedness of MBL-producing isolates was investigated by
XbaI-PFGE for
E. coli and
SmaI-PFGE for
A. baumanii. The location of MBL genes in the original isolates was determined by S1-nuclease PFGE mapping and Southern blot analysis as described previously
[ 14]. Multilocus sequence typing (MLST) of the MBL-producing isolates was performed according to published protocols (mlst.warwick.ac.uk/mlst/dbs/Ecoli).
Sequence, assembly, and annotation of the flanking region of the MBL gene
The total DNA of MBL-producing isolates was extracted using the Wizard® Genomic DNA Purification kit (Promega, WI, USA), then subjected to whole genome sequencing. The library was constructed using the NEXT® Ultra™ DNA Library Prep kit (New England Biolabs, Ipswich, UK), and paired-end sequencing was conducted on an Illumina Hiseq2500 (Berry Genomics Company, Beijing, China). A draft assembly of the sequences was generated using CLC Genomics Workbench 5 (CLC Bio, Aarhus, Denmark). All contigs were searched for β-lactamase genes using standalone Blast analysis
[ 15]. Gap closure of the flanking regions of MBL genes was performed by PCR using a modified random primer walking strategy
[ 16]. The putative coding sequences of the flanking region of MBL genes were obtained using ORF Finder programs (www.ncbi.nlm.nih.gov/gorf/orfig.cgi) and the Vector NTI program (Invitrogen, CA, USA).
Results and discussion
In this study, six isolates classified as non-susceptible to meropenem (MIC≥4 mg·L
−1), obtained from six of the 334 lung samples from the diseased pigs, were confirmed as Gram-negative bacteria. These isolates were further identified by 16S rRNA gene sequencing and the Bruker Daltonik MALDI Biotyper Classification system as
E. coli (
n = 3, FSEC38, FSEC39, and FSEC69),
A. baumanii (
n = 2, FSAB08 and FSAB62) and
A. calcoaceticus (
n = 1, FSABC15). Imipenem-EDTA double-disc synergy tests and E-test MBL strip tests confirmed that the six isolates were MBL-positive, and PCR analysis using primers specific for mobile MBL genes revealed that all of the isolates were positive for
blaNDM. The 813 bp nucleotide sequence of
blaNDM in the six isolates showed 100% identity to that of the
blaNDM-1 gene on plasmid pKpANDM-1 of
K. pneumoniae 05-506 (GenBank accession no. FN396876). The detection rate (1.8%, 6/334) of the mobile MBL gene (
blaNDM-1) in the lung samples of diseased pigs in the current study was considerably higher than both that in the lung samples of pigs from commercial farms in Guangdong province during 2011–2012 (0.3%, 1/313), and that in various samples (cloaca, rectum, nasal cavity, and lymph nodes) from commercial farms in Shandong province in 2012 (0.3%, 1/396)
[ 9, 10]. To our knowledge, this is the first time that NDM-producing Enterobacteriaceae have been reported from food animals.
PFGE analysis of the three
blaNDM-1-carrying
E. coli isolates and two
blaNDM-1-carrying
A. baumanii isolates revealed distinct genomic heterogeneity in their
XbaI and
SmaI patterns, respectively. Comparison with the allelic profiles available in the aforementioned MSLT website identified
E. coli isolate FSEC69 as sequence type (ST) 48. ST48 has been associated with hospital-acquired extended-spectrum β-lactamase (ESBL)-producing
E. coli isolates from patients in Belgium
[ 17], and ESBL-producing
E. coli isolates in fecal samples of healthy humans in Tunisia
[ 18]. The sequence types of another two
E. coli isolates FSEC38 and FSEC39 were novel and were designated as ST5084 and ST5069, respectively.
The six isolates exhibited resistance and high MIC to almost all β-lactam antibiotics tested, and only one
E. coli isolate, FSEC39, showed borderline susceptibility to aztreonam (4 mg·L
−1). As the
blaNDM-1 gene cannot confer resistance to monobactams, a search for β-lactamase genes in the contigs of the two aztreonam-resistant
E. coli isolates revealed that the AmpC β-lactamases genes
blaCMY-2 +
blaDHA-1, and
blaampC were observed in FSEC38 and FSEC69, respectively. Three genes had previously been identified as being responsible for plasmid-mediated aztreonam resistance in the Enterobacteriaceae
[ 19]. It is noteworthy that no aztreonam-resistance gene was detected in the contigs of the three
Acinetobacter isolates, suggesting that an unknown monobactam resistance gene may exist, a possibility which needs further investigation. Moreover, all isolates exhibited MICs of 2 to 256 mg·L
−1 to ciprofloxacin, which classified the isolates as resistant or intermediate. All isolates, except FSEC39, exhibited resistance to tetracycline, and had MIC for colistin of≥4 mg·L
−1, while all isolates except FSAB62 exhibited resistance to gentamicin ().
S1-nuclease PFGE mapping and Southern blot analysis revealed that the gene blaNDM-1 was located on various-sized plasmids, ranging from approximately 47–200 kb, in five isolates and on the chromosomal DNA of A. baumanii FSAB08 (, ). However, only the blaNDM-1-carrying plasmids pEC39 and pEC69 were successfully transferred from FSEC39 and FSEC69, respectively, to E. coli EC600 using filter mating. Susceptibility tests revealed that both of the transconjugants presented resistance to all tested β-lactams except aztreonam and to gentamicin compared with recipient E. coli EC600, but remained susceptible to ciprofloxacin and colistin. In addition, one transconjugant EC600-69 also exhibited resistance to tetracycline and gentamicin and an elevated MIC for florfenicol. These results indicate that the determinants conferring resistance to three antimicrobials other than β-lactams can be co-transferred with the blaNDM-1 gene from the donor to the recipient E. coli EC600 ().
Analysis of the flanking regions of the
blaNDM-1 gene on two plasmids and in one chromosome of three
Acinetobacter isolates revealed that the
blaNDM-1 gene was located in an identical 9923 bp fragment, which exhibited 100% nucleotide sequence identity to the corresponding region of plasmid pNDM-BJ01 (JQ001791) isolated from human clinical
A. lwoffii strain WJ10621 from Beijing, China
[ 20]. In addition, a 5029 bp region including the
blaNDM-1 gene, its upstream IS
Aba125 and aminoglycoside resistance gene
aphA6, and its downstream gene
bleMBL also had 99.9% (5017/5029) nucleotide sequence identity to that in plasmid pAL-01 (JN616388) in chicken
A. lwoffii strain SGC-HZ9 ()
[ 10]. Further downstream of the
bleMBL gene, a truncated putative phosphoribosylanthranilate isomerase gene
trpF was identified, followed by
dsbC,
cutA1,
groES and
groEL, and
insE, encoding the oxidoreductase protein, then tolerance protein, heat chaperonin protein and transposase, respectively.
In the
blaNDM-1-carrying regions in plasmids from porcine
E. coli isolates, the nucleotide sequences of a 15914 bp segment of pEC38 (GenBank accession no. KT164808), a 10384 bp fragment of pEC39 and a 6406 bp fragment of pEC69 (GenBank accession no. KT164809) were obtained by whole genome sequencing and primer walking (). This 15.9 kb
blaNDM-1-harboring region in pEC38 had 100% nucleotide identity with the corresponding region of plasmid pKpn-SX04 (accession no. NG_041664) from
K. pneumoniae derived from a patient from the program of Chinese Antimicrobial Resistance Surveillance of Nosocomial Infections in 2012
[ 21]. In both cases, a disrupted IS
Aba125 (ΔIS
Aba125) resulted from the insertion of another insertion sequence, the IS
5 element. Small differences between these two plasmids existed: a truncation of the left direct target site (5′-CTAA-3′) and the 234 bp IS
Aba125 sequence was observed in pEC38. However, it was reported that the 7830 bp fragment, namely the 3′ region of the ΔIS
Aba125,
blaNDM-1,
bleMBL,
trpF,
dsbC,
cutA1,
groES,
groEL, and
insE, facilitated the horizontal mobilization of the
blaNDM-1 gene among Enterobacteriaceae
[ 21].
Both the 9693 bp and 5340 bp
blaNDM-1-carrying fragments of pEC39 and pEC69 had>99.9% nucleotide sequence identity with the corresponding region of plasmid pGUE-NDM (accession no. JQ364967) from
E. coli isolate GUE, a community-acquired strain from India ()
[ 22]. Downstream of the
blaNDM-1 gene, the
bleMBL,
trpF, and
dsbC genes were observed in pEC39, followed by the insertion sequence IS
CR1, and by a class 1 integron structure carrying three gene cassettes, including streptomycin/spectinomycin resistance gene
aadA2, the gene
orfF for a hypothetical protein, and the trimethoprim resistance gene
dfrA12. The sequence further downstream of
int1 also had a high degree of nucleotide identity (99.9%) to IS
26 present in pGUE-NDM. Instead of the intact or truncated IS
Aba125 element observed immediately upstream of the
blaNDM-1 gene in other cases in this study, a rifampicin resistance gene
aar-3 and chloramphenicol resistance gene
catB3 were observed in pEC69 ().
Conclusions
The results of this study confirm that blaNDM-1 is the only mobile MBL gene present in Acinetobacter spp. and E. coli isolates from food animals. Similar to the spread of blaNDM-1 in bacteria of clinical origin, both the plasmid and insertion sequence elements, such as ISAba125, IS5 and IS26, may be important in the dissemination of the blaNDM-1 gene and its conserved flanking regions in bacteria from food animals. Moreover, the frequency of identification of ESBL-carrying E. coli ST48 from both human fecal and clinical samples, as well as the high level of similarity of the genetic environment of the blaNDM-1 gene from the porcine E. coli isolate found in human Enterobacteriaceae, suggests the possibility of human-to-food animal transfer of this blaNDM-1-producing isolate. Thus, enhanced and continued efforts are needed to monitor carbapenemase producers in bacteria from food animals. It should be noted that carbapenems are not approved for use in food animals, and the blaNDM-1 gene was found to coexist with other genes conferring resistance to aminoglycosides, phenicols, and tetracyclines in isolates in this study, the co-selection and co-transfer of carbapenemase genes under the selective pressure imposed by these genes seems be key to a major role in their dissemination. Therefore, the prudent use of antimicrobial agents to decrease the opportunities for co-selection of carbapenemase genes in food animals is urgently needed.
Higher Education Press and Springer-Verlag Berlin Heidelberg