Characterization of an NDM-5 carbapenemase-producing Escherichia coli ST156 isolate from a poultry farm in Zhejiang, China

Strain features

Two hundred nineteen of the samples studied tested positive for E. coli, and E. coli isolates from all of these samples were obtained and characterized by antimicrobial susceptibility testing (AST) using the VITEK® 2 COMPACT system (BioMérieux, France). The highest overall levels of resistance were observed toward ampicillin, with 74.43% of all isolates resistant to this antimicrobial. High rates of resistance were also observed toward trimethoprim (54.34%), with lower levels of resistance observed toward piperacillin (1.83%), amikacin (2.29%), and amoxicillin (0.91%). No strain was determined to be resistant to tigecycline. One hundred eighty isolates (82.2%) were resistant to at least one antimicrobial agent, and 92 isolates (42.01%) were resistant to three or more antimicrobial agents. Of the 219 E. coli isolates, a carbapenem-resistant strain was identified that showed resistance toward ertapenem and imipenem, which is rare in poultry.

Characterization of the genome sequence of strain ECCRA-119

The genome of the ECCRA-119 isolate consisted of a single circular chromosome and three circular plasmids (Table 2, Figs. 1 and 2b). The chromosome sequence of ECCRA-119 was determined to be 4,893,130 bp in length, have a GC content of 50.77% and encodes 5042 proteins that account for 90.96% of the genome. The average depth of coverage was 210.5×, and 22 rRNAs, 87 tRNAs, and 2 CRISPRs were detected. Three plasmids in the ECCRA-119 isolate were identified, pTB201, pTB202 and pTB203. The plasmid pTB201, which is a combination of IncFII- and IncFIB-type plasmid, was determined to be 146,268 bp in length and have an average GC content of 51.35%. The plasmid pTB202, a p0111-IncN-type plasmid, was determined to be 139,629 bp in length and have an average GC content of 49.13%. In addition, the bla_NDM-5-harboring plasmid pTB203, an IncX3-type plasmid, was determined to be 46,161 bp in length and have an average GC content of 46.65%. Moreover, the three plasmids were characterized by S1-PFGE (Fig. 1a), the results of which were consistent with the whole genome sequencing analysis. Multilocus sequence typing (MLST) analysis classified E. coli ECCRA-119 as ST156, suggesting that E. coli ST156 strains have the potential to harbor bla_NDM-5-like genes.

Features	ECCRA-119
GenBank	CP029242	CP029243	CP029244	CP029245
Status	Chromosome	Plasmid	Plasmid	Plasmid
Genome size (bp)	4,893,130	146,268	139,629	46,161
G + C content (%)	50.77	51.35	49.13	46.65
No. of predicted coding sequences (CDS)	5042	199	170	64
rRNA	22	0	0	0
tRNA	87	0	0	0
No. of CRISPR regions	2

Seven virulence factors were detected in the whole genome sequence (Additional file 1: Table S2), four in the chromosome and three in the pTB201 plasmid, indicating the potential virulence of the ECCRA-119 isolate. These virulence factors are grouped into five classes (iss, gad, lpfA, iroN, and cma), which are related to serum survival, glutamate decarboxylase, long polar fimbriae, enterobactin siderophore receptor protein, and colicin M, respectively. In addition, one mercury resistance-related gene, merA, was identified on plasmid pTB201 (Additional file 1: Table S3).

Transferability of plasmids

Conjugation assays confirmed that bla_NDM-5 could be transferred between E. coli strains, with an observed transfer frequency of (1.39 ± 0.12) × 10^− 5. The antibiotic susceptibility testing results showed that the transconjugants, confirmed by PCR and sequencing, were resistant to meropenem (4 mg/L). The transfer regions of the three plasmids of strain ECCRA-119 were successfully identified (Figs. 1 and 2b) by oriTfinder, including the origin of transfer region (oriT), relaxase gene, bacterial type IV secretion system (T4SS) apparatus gene clusters and the type IV coupling protein (T4CP) gene. Plasmid pTB201 was observed to possess an oriT (52,884–52,969 bp in the plasmid), relaxase gene, T4CP and T4SSs, indicating a high potential for self-transferability [26]. Plasmid pTB202 was observed to harbor a relaxase but lacked an oriT, T4CP and/or T4SS, indicating that it is not a mobilizable plasmid [26]. Plasmid pTB203 possess a relaxase gene, T4CP and T4SSs, but lacked a typical oriT sequence, demonstrating its potential to be transferred to other bacteria [26], with its transfer ability having been confirmed experimentally.

Phylogenetic analysis of strain ECCRA-119 with other E. coli ST156 isolates

MLST analysis classified E. coli strain ECCRA-119 as ST156. Thus, we built a phylogenetic tree to determine its relationship among ST156 E. coli strains based on a SNP analysis (Additional file 1: Figure S1). We identified 52,076 SNPs from the 37 genome sequences available in GenBank. Of these, 17,953 and 34,123 were identified as core and noncore SNPs, respectively. We excluded the noncore SNPs for the further analysis and constructed a phylogenetic tree based on the genome-wide core SNPs. The core genome analysis identified 5 groups (Additional file 1: Figure S1). E. coli strain ECCRA-119 is grouped with the strains 174,900, SCEC020022 and VREC0575, which were isolated from Bangladesh, China and the United Kingdom, respectively. There were 7 group-specific core SNPs in this group. The number of strain-specific SNPs identified in strains ECCRA-119, 174,900, SCEC020022 and VREC0575 was 59, 71, 134 and 160, respectively. Interestingly, most isolates identified from the same region or source are not in the same lineage. Isolates from different countries were observed to be clustered together (strains 157–1949 and SE11). Similarly, strains isolated from different hosts (wild animals, livestock and poultry, and dog) clustered into the same branch (strains MOD1-EC5693, CVM N33633PS and MOD1-EC6498).

Phylogenetic and comparative analysis of pTB203 and other bla _NDM-5-harboring IncX3 plasmids

An SNP-based phylogenetic analysis was conducted using the 52 complete sequences of bla_NDM-5-harboring IncX3 plasmids available in GenBank (Fig. 2a). Among these sequences, 41 originated from bacterial strains from humans, 1 from a pig, 5 from geese, 1 from a vegetable, 1 from a layer hen, 1 from sewage, and 2 from unknown sources. Our results showed that the IncX3 plasmids have an extensive host range. Among these 52 plasmids, 43 were isolated in China and 33 were from E. coli. Five plasmids from geese became available in January 2019 but were not published. Among these plasmids, 9 published plasmids were selected and constructed by BRIG (Fig. 2b), including pVH1 (vegetable, China, 46,161 bp) [25], pNDM_MGR194 (human, India, 46,253 bp) [27], pECNDM101 (swine, China, 46,165 bp) [23], pEC463-NDM5 (human, China, 46,145 bp) [28], pBJ114–46 (human, China, 46,161 bp) [29], pEsco-5256cz (human, Czech, 46,161 bp) [30], pEc1929 (human, China, 46,164 bp) [31], pTB203 (layer hen, China, 46,161 bp, in this study), and pZSH6-blaNDM-5 (human, China, 46,161 bp) [32]. The results of BLAST homology analyses showed that these plasmids had more than 99.9% identity and 99.8% query coverage with each other. The comparative analysis of 9 bla_NDM-5-harboring IncX3 plasmids (~ 46 kb) revealed that these plasmids are highly similar to each other, possessing the same backbone that includes the IncX3 replication, bla_NDM-5 gene and conjugation/type IV secretion components. This result was further confirmed by the comparative analysis of 52 bla_NDM-5-harboring IncX3 plasmids, with the exception of pD2-NDM_1_1 (human, South Korea, 79,613 bp) (Additional file 1: Figure S2). The results of our analysis showed that bla_NDM-5-harboring IncX3 plasmids with an ~ 46 kb backbone have extensive host adaptability in Enterobacteriaceae.

Complete sequences of plasmids harboring bla _NDM variants from China

At present, 24 variant bla_NDM sequences are available in GenBank, all of which were aligned by ClustalX (Additional file 1: Figure S3 and S4). These sequences are 813 bp in length, with the exception of bla_NDM-18, and only 1–6 SNPs are observed among these sequences. In particular, the bla_NDM-5 gene has the closest homology with bla_NDM-17, bla_NDM-20 and bla_NDM-21 (Additional file 1: Figure S4). Relative to bla_NDM-5, bla_NDM-17, bla_NDM-20 and bla_NDM-21 contained point mutations at positions 508 (G → A), 809 (G → A), and 205 (G → A), generating amino acid substitutions Glu170Lys, Arg270His, and Gly69Ser, respectively. In China, 13 types of plasmids harbor bla_NDM genes with complete sequence are reported in GenBank, including bla_NDM-1, bla_NDM-4, bla_NDM-5, bla_NDM-6, bla_NDM-7, bla_NDM-9, bla_NDM-13, bla_NDM-14, bla_NDM-16, bla_NDM-17, bla_NDM-19, bla_NDM-20, bla_NDM-21 (Additional file 2: Table S4, Fig. 3). The bla_NDM-1 and bla_NDM-5 genes are the most prevalent bla_NDM variants in China, with humans being the primary host source. Additionally, seven bla_NDM-5-harboring plasmids have been detected in poultry and livestock in China, including 1 plasmid detected from swine in Sichuan in 2016, 1 plasmid detected from a layer hen in Zhejiang in 2017 (in this study), and 5 plasmids detected from geese in Jiangsu in 2018.

Comparative analysis of plasmids pTB201 and pTB202

We compared the plasmids pTB201 and pTB202 with the corresponding homologous plasmids from GenBank via BLAST analyses. The results showed that plasmid pTB201 shares homology with plasmid pSMS35_130 (CP000971), plasmid pJIE186_2 (JX077110) and the p300 iro gene cluster (AY205565) (Fig. 4a); plasmid pTB202 showed homology with plasmid p1079-IncFIB-N (MG825383) and part of plasmid pD90–3 (CP022453) (Fig. 4b). These comparisons revealed that these plasmids do not have full-length matching plasmids in the GenBank database, although they shared backbones with many other plasmids.

Sample collection and antimicrobial susceptibility testing

Using the sampling method proposed by Leon and Hassan [42, 43], 251 samples of chicken feces were collected from 12 large-scale chicken farms in Zhejiang province in 2017.

The E. coli isolate recovered was designated ECCRA-119 and showed resistance to meropenem (8 mg/L) and imipenem (4 mg/L). This isolate was selected for AST using the broth dilution method with the Biofosun® Gram-negative panel (Fosun Diagnostics, Shanghai, China). The criteria from the Clinical and Laboratory Standards Institute (CLSI) were used to interpret the results, and the US National Antimicrobial Resistance Monitoring System (NARMS) and European Committee on Antimicrobial Susceptibility Testing (EUCAST) protocol was used when CLSI standards were not appropriate. The panel of antimicrobial compounds tested included ampicillin, amoxicillin/clavulanic acid, tetracycline, doxycycline, gentamicin, spectinomycin, sulfisoxazole, trimethoprim/sulfamethoxazole, ceftiofur, ceftazidime, florfenicol, enrofloxacin, ofloxacin, colistin, meropenem, and mequindox from among livestock antibiotics, and ampicillin, ampicillin/sulbactam, tetracycline, chloramphenicol, trimethoprim/sulfamethoxazole, cefazolin, cefotaxime, ceftazidime, cefoxitin, gentamicin, imipenem, nalidixic acid, azithromycin, sulfisoxazole, ciprofloxacin, amoxicillin/clavulanate, Cefotaxime/clavulanate, ceftazidime/clavulanate, colistin, polymyxin B, minocycline, amikacin, aztreonam, cefepime, meropenem, levofloxacin, doxycycline, kanamycin, streptomycin, and gemifloxacin from among medical antibiotics.

Whole genome sequencing, assembly and annotation

After genomic DNA extraction and quality checks, a 20-kb fragment library was constructed for the sample when the concentration and purity met the sequencing requirements. Whole-genome sequencing was performed using a PacBio RS II instrument [44]. The assembly of the reads was performed following the Hierarchical Genome Assembly Process (HGAP) workflow [45]. In this process, the Celera Assembler, following the OLC algorithm, was used to assemble the sequences [46], and Quiver was used to optimize the assembly results [45]. The gene prediction and annotation of the genomes was performed using the NCBI Prokaryotic Genome Annotation Pipeline [47]. The complete genome of the ECCRA-119 isolate was deposited in GenBank under the accession numbers CP029242 (chromosome), CP029243 (plasmid pTB201), CP029244 (plasmid pTB202) and CP029245 (plasmid pTB203).

Sequence analysis

CRISPRfinder (https://crispr.i2bc.paris-saclay.fr/Server/) was used to search for CRISPR loci in the genome of the ECCRA-119 isolate [48]. MLST 2.0 (https://cge.cbs.dtu.dk/services/MLST/) was used to determine the ST [49]. The plasmid replicon types were identified using PlasmidFinder-1.3 (https://cge.cbs.dtu.dk/services/PlasmidFinder/) [50]. Acquired antimicrobial resistance genes were predicted using ResFinder (https://cge.cbs.dtu.dk/services/ResFinder/) [51]. VirulenceFinder (https://cge.cbs.dtu.dk/services/VirulenceFinder/) was used to identify the virulence factors [52], and oriTfinder (http://202.120.12.134/oriTfinder/oriTfinder.html) was used to identify the origin of transfer in the genome [53]. The genome was investigated for metal resistance genes using the Antibacterial Biocide and Metal Resistance Genes Database (BacMet) (http://bacmet.biomedicine.gu.se/) [54]. Easyfig [55] and BIRG [56] were used in the comparative analysis of the plasmids. Phylogenetic analysis of genome and plasmids was performed by KSNP based on the maximum-likelihood method [57]. Clustal X was used to perform the alignment analysis of bla_NDM based on nucleotide sequences [58]. The phylogenetic tree was generated used in MEGA X [59] and iTOL [60].

Conjugation assay

Plasmid conjugation experiments were performed on the ECCRA-119 isolate as described previously by Lin et al. [23, 61]. A rifamycin-resistant E. coli EC600 strain was used as the recipient in the plasmid conjugation assay to test the transferability of the carbapenem resistance gene and other resistance genes harbored by the ECCRA-119 isolate. Briefly, transconjugants were selected on LB agar plates (Landbridge., Beijing, China) supplemented with rifamycin (400 mg/L) (Sangon Biotech., Shanghai, China) and meropenem (4 mg/L) (J&K Chemical Ltd., Shanghai, China). The transfer frequencies were calculated by dividing the number of colony-forming units (CFUs) of transconjugants by the number of CFUs of the recipients. Genome DNA was extracted from the E. coli transconjugant using a bacterial DNA extract kit (Generay, Shanghai, China). The bla_NDM-5 primers (F: 5′-GTCTGGCAGCACACTTCCTA-3′; R: 5′- TAGTGCTCAGTGTCGGCATC-3′) were used to confirm that the transconjugant harbored the plasmid.

S1-PFGE

S1-PFGE was performed according to a standard protocol using the contour-clamped homogeneous electric field (CHEF) technique with 0.5 × TBE buffer [62]. Salmonella enterica serotype Braenderup H9812 was used as a size marker [63]. The gels were run at 6 V/cm and 14 °C with an angle of 120°, and initial and final pulses were set at 2.16 and 63.8 s, respectively. The running time was 16 h using the CHEF apparatus (CHEF MAPPER XA; Bio-Rad, USA).