Novel Variant of New Delhi Metallo-β-lactamase, NDM-20, in Escherichia coli

The spread of carbapenem-resistant Enterobacteriaceae (CRE) mediated by New Delhi metallo-β-lactamase (NDM) poses a serious challenge to clinicians and has become a major public health concern. NDM has been evolving into variants that possess different hydrolysis activity toward antibiotics, so as to affect treatment strategy. In addition, very few studies on NDM variants have focused on animal-derived bacterial isolates. Our study reports a novel NDM variant, NDM-20, in an isolate of Escherichia coli CCD1 recovered from the food animal swine in China. The isolate that was assigned to ST1114, exhibited high level resistance to all β-lactams tested, including aztreonam and carbapenems. The gene of blaNDM-20 was located on an IncX3-type plasmid, surrounded by multiple insertion sequences. Sequencing analysis demonstrated that blaNDM-20 contained three point mutations at positions 262 (G→T), 460 (A→C), and 809 (G→A), compared with blaNDM-1, and just one point mutation at position 809 (G→A), relative to blaNDM-5. Functional analysis revealed that the blaNDM-20 transformant, DH5α+pHSG398/NDM-20, exhibited a higher resistance to ertapenem than that of blaNDM-1 transformant DH5α+pHSG398/NDM-1. Kinetic parameter analysis showed that NDM-20 had increased enzymatic activity against some penicillins and cephalosporins but decreased carbapenemase activity relative to NDM-5. The identification of NDM-20 further confirms the evolution and prevalence of NDM variants in bacteria of food-animal origin.


INTRODUCTION
The global dissemination of carbapenem-resistant Enterobacteriaceae (CRE) has become a major threat to public health. With the increasing global prevalence of New Delhi metalloβ-lactamase (NDM)-producing bacteria, which are highly disseminated in China (Khan et al., 2017), available antibiotics are becoming less effective in the treatment of drug-resistant bacterial infections. Since NDM was first reported in India (Yong et al., 2009), the sequences of 18 NDM variants  have been reported. NDMs may be evolving diverse hydrolytic activity toward β-lactams, resulting from amino acid substitutions or insertions (Khan et al., 2017), subsequently, it may affect treatment strategy. In addition, these bacteria producing NDM variants are disseminated not only among humans, but also in food animals . Even though food-producing animals serve as the major reservoir of NDM-producing bacteria (Madec et al., 2017), apart from our previous description of NDM-17 (Liu et al., 2017), studies on NDM variants of animal-derived bacterial isolates are limited. So, it is important and necessary to monitor NDM evolution in food animals. Here, a novel NDM variant, NDM-20, was identified in an Escherichia coli strain isolated from swine, and was subjected to in-depth molecular characterization.

Isolation and Identification of the Bacterial Strain and Phenotypic Screening for MBL Production
To better understand the prevalence of CRE, a routine surveillance of antimicrobial resistance in bacteria of foodproducing animal origin was carried out in Shandong province, China, in 2016. The swine anal swab samples were collected and enriched in 5 mL of brain heart infusion broth before being plated onto CHROMagar KPC medium (CHROMagar, Paris, France) for CRE selection. Then, the resistant strains were identified on MHA medium containing 1 mg/L of meropenem. Strain CCD1 was isolated from a swine fecal swab sample from a commercial farm. Identification of the bacterial species was performed by MALDI-TOF MS (Bruker Daltonik, Bremen, Germany) and 16S rRNA sequencing, as described previously (Wang et al., 2013). To detect carbapenemase production, the modified Hodge test was carried out using imipenem and meropenem disks, and for MBL production test, the combined-disk was performed by imipenem with and without EDTA.

Antimicrobial Susceptibility Testing and Detection of Antibiotic Resistance Genes
Antimicrobial susceptibility testing of isolates was carried out using the broth microdilution method, and the results in triplicate with best repeatability were interpreted according to the CLSI and EUCAST (version 7.1) guidelines (Clinical and Laboratory Standards Institute [CLSI], 2015;EUCAST, 2017). The tested isolates and antimicrobial agents are listed in Table 1, and E. coli ATCC 25922 was used as a quality control strain. Carbapenemase-encoding genes bla KPC , bla VIM , bla IMP , bla NDM , bla OXA−23−like , bla OXA−58−like , bla OXA−48−like , bla IND , bla KHM , bla DIM , bla GIM , bla SIM , and bla SPM , and β-lactamase-encoding genes bla TEM , bla SHV , and bla CTX−M were screened as previously described (Dallenne et al., 2010;Poirel et al., 2011;Zhang et al., 2017b), and bla NDM was screened by primer preA and preB in Supplementary Table S1. PCR amplicons were sequenced and then analyzed against the GenBank database.

Conjugation Assay and Location Analysis of bla NDM
To determine the transferability of bla NDM , transconjugation was attempted using E. coli J53 as the recipient (Liu et al., 2017). Transconjugants were selected on MacConkey agar supplemented with 100 mg/L sodium azide and 1 mg/L meropenem, and confirmed by PCR and sequencing. The transfer frequency was determined from the ratio of transconjugants: donors. Location analysis of bla NDM was performed by S1-PFGE and Southern blotting, as described previously (Borgia et al., 2012).

Resistome and Plasmid Analysis
To further confirm the resistance gene profile of strain CCD1 and to determine the genetic background of bla NDM , the genomic DNA of strain CCD1, including plasmid pNDM-20, was extracted and subjected to WGS, as reported previously (Liu et al., 2017;Wang et al., 2017). A 250-bp paired-end library was constructed using a NEXT Ultra DNA Library Prep kit (New England Biolabs, E7645L) and an Illumina Hiseq 2500 system (Bionova Biotech, Co., Beijing, China). The raw sequencing data were then de novo assembled and examined for the presence of resistance genes using CLC Genomics Workbench 9.0 (CLC Bio, Aarhus, Denmark) against resistance gene database based on CARD. SRST2 was used for MLST analysis of strain CCD1 using the sequence reads (Inouye et al., 2014).

Cloning of bla NDM Genes and Purification of the NDM Proteins
To compare the resistance abilities of bla NDM-20 , bla NDM-5 and bla NDM-1 , the entire bla NDM-20 , bla NDM-5 , and bla NDM-1 ORFs were cloned into pHSG398, resulting in recombinant plasmids pHSG398/NDM-20, pHSG398/NDM-5, and pHSG398/NDM-1, respectively (Liu et al., 2017). The plasmids were then transformed into E. coli DH5α. To assay the effect on enzymatic activity caused by Arg270His, bla NDM-20 and bla NDM-5 minus their signal peptide regions were also cloned into the pET28a expression vector, and the resulting plasmids were transformed into E. coli BL21 (DE3) (TransGen Biotech, Beijing, China) (Liu et al., 2017). The corresponding primers were described in Supplementary Table S1. All the transformants were confirmed by PCR and sequencing analysis, and their sensitivity to β-lactams were determined as described above. Purification and identification of the NDM proteins was carried out as per the manufacturer's instructions in the following order: (1) purification of recombinant NDM proteins using Ni-nitrilotriacetic acid agarose; (2) cleavage of His-tags using Turbo TEV protease (Accelagen, San Diego, CA, United States); (3) collection of untagged proteins by an additional passage over Ninitrilotriacetic acid agarose; (4) SDS-PAGE analysis to assess the purity of the protein products; (5) measurement of protein concentrations by using a Pierce bicinchoninic acid protein assay kit (Thermo Scientific, Waltham, MA, United States); and (6) monitoring β-lactamase activity using nitrocefin (Oxoid, Ltd., Basingstoke, United Kingdom) (Liu et al., 2017).

Enzyme Kinetics Assay
To assay the catalytic properties of NDM-20 and NDM-5, a kinetic study was conducted. A SpectraMax M5 multidetection microplate reader (Molecular Devices, Sunnyvale, CA, United States) was used to measure initial hydrolysis rates in 30 mM phosphate buffer (30 µM Zn 2+ , pH 7.0) at 25 • C. The hydrolysis rates were then used to determine the K m and k cat values, as well as the k cat /K m ratio, using a Lineweaver-Burk plot. All data used to determine the hydrolysis rates were collected from three individual experiments, and wavelengths and extinction coefficients were applied as previously described (Crowder et al., 1998;Queenan et al., 2010).

Accession Numbers
The sequence of bla NDM-20 gene has been deposited in GenBank under accession no. KY654092 (sequence of bla NDM-20 in Supplementary data 1). The complete nucleotide sequence of plasmid pNDM-20 (sequence of pNDM-20 plasmid in Supplementary data 2) has been deposited as GenBank accession no. MF458176.

Strain Features
Escherichia coli strain CCD1 was isolated from a swine fecal swab collected from a commercial pig farm in Shandong province, China. The susceptibility testing results showed that E. coli CCD1 exhibited high level resistance to all β-lactams examined ( Table 1), including aztreonam, carbapenems and combinations of penicillins and β-lactamase inhibitors, but susceptible to gentamicin and colistin. The modified Hodge test was positive for carbapenemase, and subsequent positive result from the MBL combined-disk test further suggested MBL production in E. coli CCD1.
The de novo assembled genome of CCD1 showed that the number of contigs was 87 and length of N50 was 110,756 bp. PCR and sequencing analysis identified the presence of a novel NDM-encoding gene, designated bla NDM-20 (GenBank accession no. KY654092). Relative to bla NDM-1 , bla NDM-20 contained three point mutations at positions 262 (G→T), 460 (A→C), and 809 (G→A), generating amino acid substitutions Val88Leu, Met154Leu, and Arg270His, respectively (Supplementary Figure S1). Homology and phylogenetic analysis of the amino acid sequences of all NDM variants showed that NDM-20 shared the closest evolutionary relationship with NDM-5 (Figure 1). Notably, only one difference (Arg270His) was observed between the two variants, with this being the first description of a substitution at residue 270. In addition, sequence analysis confirmed the presence of ampC, bla CMY (aztreonam resistance), and sul2 (sulfonamide resistance) in strain CCD1 (Seral et al., 2012). However, no other ESBL-encoding genes were detected.
MLST analysis classified E. coli CCD1 as ST1114, suggesting that E. coli ST1114 strains have the potential to become a reservoir for the bla NDM-20−like gene. A previous study showed that ST1114 E. coli isolates are the second largest reservoir of mcr-1 (Matamoros et al., 2017), while our previous study identified two E. coli ST1114 isolates from chickens that were positive for both mcr-1 and bla NDM-5 .

Transferability of bla NDM-20 and Plasmid Analysis
Conjugation assays confirmed that bla NDM-20 could be transferred between E. coli strains, although the transfer FIGURE 1 | Phylogenetic relationships between New Delhi metallo-β-lactamase (NDM) variants based on amino acid sequences. The tree was generated in MEGA7 and iTOL using the maximum likelihood method. Increased/similar carbapenemase activity of NDM variants was determined from previously reported data and compared with NDM-1. The numbers on the branches signify represent branch lengths.
No resistance genes other than bla NDM-20 and ble MBL were identified on pNDM-20, supporting the observation that the transconjugants only exhibited resistance to β-lactams, and remained sensitive to aztreonam and other classes of antibiotics. Homology analysis revealed 99% nucleotide sequence identity between pNDM-20 and IncX3-type plasmid pNDM-MGR194 (GenBank accession no. KF220657). pNDM-MGR194 was the first reported bla NDM-5 -harboring plasmid, and suggested to be responsible for the evolution and distribution of bla NDM-5 amongst bacteria (He et al., 2016). To date, multiple pNDM-MGR194-like plasmids have been identified in bacteria isolated from both humans and animals in China. These plasmids which are 46,161-46,165-bp in length showed >99% nucleotide sequence identity to pNDM-MGR194, including pAD-19R (GenBank accession no. KX833071), pEC50-NDM7 (GenBank accession no. KX470735), pNDM-QD28 (GenBank accession no. KU167608), and pNDM5_IncX3 (GenBank accession no. KU761328) (Figure 2). Therefore, the identification of pNDM-20 provides further evidence that IncX3-type bla NDMharboring plasmids have become established in bacterial strains originating from both humans and animals, and may circulate via the food chain (He et al., 2016;Liu et al., 2017).
Of note, pNDM-20 had the same backbone structure as the two bla NDM-5 -harboring plasmids (653 and 655) from the ST1114 E. coli strains mentioned above. Further analysis showed that the two bla NDM-5 -harboring plasmids could be typed as IncX3 using PlasmidFinder software 1 (Figure 2). Given these plasmids have the same genetic background and plasmid group/type and were all isolated from ST1114 E. coli strains, it is possible that bla NDM-20 evolved from bla NDM-5 through a series of mutations.

Functional Analysis of bla NDM-20
To determine the specific function of bla NDM-20 , bla NDM-20 , the reference gene bla NDM-5 (containing only a single nucleotide difference at position 809) and bla NDM-1 were cloned, and the corresponding transformants were examined to determine their antibiotic sensitivity profiles. As shown in Table 1, compared to DH5α+pHSG398, expression of bla NDM-20 , bla NDM-5 , and bla NDM-1 in E. coli DH5α reduced susceptibility to almost all tested β-lactams, including penicillins, cephalosporins, carbapenems, and combinations of penicillins and β-lactamase inhibitors, except aztreonam. In addition, the MICs of piperacillin was not affected by the addition of β-lactamase inhibitors tazobactam. In general, the bla NDM-20 transformant DH5α+pHSG398/NDM-20 exhibited similar β-lactam resistance to bla NDM-5 transformant DH5α+pHSG398/NDM-5, although several slight differences were observed. Compared with DH5α+pHSG398/NDM-1, the DH5α+pHSG398/NDM-20 exhibited a fourfold elevation in MIC of ertapenem, indicating that bla NDM-20 should enhance resistance to ertapenem. All our findings demonstrated that bla NDM-20 was not as effective as bla NDM-5 , but conferred higher resistance to carbapenems than bla NDM-1 .

Enzyme Activity Analysis
To further characterize NDM-20 and explore the influence of the amino acid substitutions on enzymatic activity, NDM-20 and NDM-5 were successfully purified (>90%) and their kinetic parameters were examined. NDM-20 was able to hydrolyse all β-lactams tested, except aztreonam ( Table 2). The k cat /K m ratios for ampicillin, cefotaxime, and cefoxitin for NDM-20 were higher than those for NDM-5, but carbapenems k cat /K m ratios for NDM-20 were lower than that of NDM-5. These findings resulted from the lower K m values for some penicillins and cephalosporins and higher values for carbapenems for NDM-20 compared with NDM-5. These results demonstrated that NDM-20 had higher enzymatic activity against some penicillins and cephalosporins and lower activity against carbapenems relative to NDM-5, caused by its enhanced affinity for some penicillins and cephalosporins and depressed affinity for carbapenems.
NDM-20 contains an Arg270His substitution relative to NDM-5. This substitution may be responsible for the decreased carbapenemase activity of NDM-20 relative to NDM-5. Two other amino acid substitutions (Val88Leu and Met154Leu) were identified in NDM-20 compared with NDM-1. The Val88Leu substitution was reported to decrease the hydrolytic activity of NDM, while the Met154Leu substitution appeared to increase the carbapenemase activity (Nordmann et al., 2012). The active site of NDM-1 is formed by two loops, loop3 and loop10, and crucial amino acid residues binding to two zinc ions, including His120, His122, His189, His250, Cys208, and Asp124 (Green et al., 2011), and the residue 270 was located out of the enzyme active center (Figure 3). Based on this information, the substitution at residue 270 would not affect the conformation of enzyme active site. It is also possible that the Arg270His substitution may change the protein conformation of NDM-20, thereby altering its hydrolytic activity (Zou et al., 2015). However, further experiments are required to confirm which, if any, of these mechanisms contribute to the altered activity of NDM-20 compared with NDM-5.

CONCLUSION
Our study reports a novel NDM variant, NDM-20, in a ST1114 E. coli isolate recovered from food animal in China. Compared to NDM-5, NDM-20 enhances the hydrolytic activity toward some penicillins and cephalosporins, but depresses the carbapenemase activity, which may affect drug strategies for CRE infection. Both the identification of NDM-20 from swine and NDM-17 from chicken suggest that the food animals have become the reservoir of bacteria producing NDM, which may drive NDM evolution to enhanced enzyme activity toward some or all β-lactams. Furthermore, bla NDM-20 positive bacteria possess the potential capability of dissemination in not only animals but human, becoming a worldwide spread risk. It is time to monitor bacteria producing NDM-20 and other NDM variants in food animals, and coordinate globally to prevent or slow down the NDM evolution.

AUTHOR CONTRIBUTIONS
JS was responsible for the study design. ZL, JL, XW, and DL assisted in data collection. ZL and YK completed the data interpretation. ZL, YW, and JS wrote the report. All authors revised, reviewed, and approved the final report.

FUNDING
The study was supported by grants from the National Natural Science Foundation of China (31530076, 31422055, and 81661138002) and the National Basic Research Program of China (2013CB127200).