A 71-year-old Japanese inpatient with no history of travel outside Japan was diagnosed with pneumonia in 2018 at a university hospital in Japan. K. oxytoca (KO_408) was initially identified from his respiratory culture as the causative agent by matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (Bruker Daltonics GmbH, Bremen, Germany).

Tf_Top10-NDM-5 and Tf_C600-NDM-5 were transformants of Escherichia coli Top10 and E. coli C600 with pKO_4-NDM-5, a plasmid derived from KO_408.

The antimicrobial susceptibilities of KO_408 and the transconjugant (Tc-NDM-5) were evaluated by broth microdilution with 15 antibiotics (cefazolin, cefotaxime, piperacillin, ceftazidime, cefpodoxime, cefepime, aztreonam, gentamicin, amikacin, minocycline, imipenem, fosfomycin, levofloxacin, sulfamethoxazole/trimethoprim, and meropenem) on DP31 dry plates (Eiken Chemical Co., Tokyo, Japan). The results of minimum inhibitory concentration (MIC) values were interpreted according to guidelines outlined in the Clinical Laboratory and Standards Institute document, M100 (30th edition). Quality control was performed using E. coli ATCC 25922. Modified carbapenem inactivation method (mCIM) testing was performed for KO_408, as previously described (Pierce et al., 2017). The major carbapenemase genes, including those encoding VIM-, IMP-, NDM-, KPC, and OXA-48-like carbapenemases, were screened (Dallenne et al., 2010; Ayibieke et al., 2018).

The transferability of the bla_NDM-5 gene was determined by the agar mating conjugal method using a previously described protocol with some modifications (Prah et al., 2021). The sodium azide-resistant E. coli strain J53 was used as the recipient strain. The recipient and donor strains (KO_408) were mixed in a ratio of 1:1 and inoculated on tryptone soya agar plates containing 0.05 μg/ml meropenem. The plates were incubated at 37°C overnight, and transconjugants (Tc-NDM-5) were selected on bromothymol blue lactose agar plates containing 2 μg/ml meropenem and 100 μg/ml sodium azide. The recipients were selected using only 100 μg/ml of sodium azide. The presence of the bla_NDM-5-containing plasmid was verified using PCR.

To determine the location of bla_NDM-5 in KO_408 and the size of the mobile element containing the carbapenemase gene, S1-nuclease pulsed-field gel electrophoresis and Southern blotting were performed. The protocol by Prah and colleagues (Prah et al., 2021) was followed with some modifications. Genomic DNA from KO_408, Tc-NDM-5(transconjugant), Tf_Top10-NDM-5 (transformant), and E. coli J53 were prepared in agarose plugs and digested with S1 nuclease (Takara Bio). DNA separation was performed on a CHEF-mapper XA system (Bio-Rad, Hercules, CA, USA) with a running time of 18 h, temperature of 14°C, field strength of 6 V/cm², angles of 120°C, initial switching time of 2.2 s, and final pulse time of 63.8 s. A lambda DNA ladder (Lonza, Rockland, ME, USA) was used as the size marker. Southern blot hybridization with digoxigenin-labeled bla_NDM-5 was used to determine the plasmid-carrying bla_NDM-5, in accordance with the manufacturer’s instructions for the DIGHigh Prime DNA Labeling and Detection Starter Kit II (Roche Diagnostics, Germany).

DNA for genomic sequencing of KO_408 was extracted using a NucleoBond HMW DNA Kit (Takara Bio, Shiga, Japan). DNA with low molecular weights (< 40 kb) was removed using a short-read eliminator (Circulomic, Japan) prior to nanopore sequencing library preparation. It met the quantity and quality requirements for both the Oxford Nanopore and Illumina library preparations.

A nanopore sequencing library was prepared using the native barcoding expansion 1–12 kit (EXPNBD104) and the SQK-LSK109 ligation sequencing kit. Sequencing was performed for 10 h using the MinION flow cell FLO-MIN106 R9.41 in a GridION X5 sequencer. The Illumina Nextera DNA Flex Library Prep kit was used to prepare the library for Illumina short reads, and sequencing was performed using the Illumina MiSeq (San Diego, CA, USA).

Low-quality reads (MinION Q < 10; MiSeq <30) and short reads (MinION length < 500 bp; MiSeq <10 bp) were filtered out. Read yields from Illumina MiSeq and Nanopore MinION and de novo assembly statistics are presented in Supplementary Tables S1 and S2, respectively. A hybrid de novo assembly was conducted using the Unicycler v0.4.8. and genes were predicted using RAST (Overbeek et al., 2014). Antimicrobial resistance gene, multilocus sequence type (ST), and plasmid replicon type analyses were performed using ResFinder, MLST, and PlasmidFinder tools available at the Center for Genomic Epidemiology server (https://cge.cbs.dtu.dk). Isfinder (https://www-is.biotoul.fr/) and Virulence Finder Database (https://http://www.mgc.ac.cn/cgi-bin/VFs/v5/main.cgi) were used to detect mobile elements and virulence factors, respectively.

To confirm the identity of KO_408 and delineate its lineage within the global context, average nucleotide identity (ANI) analysis of KO_408 with respect to NCBI reference genomes CP069911 (K. oxytoca) and CP022348(K. michiganensis) was initially assessed using FastANI (Jain et al., 2018). A cutoff of >95% ANI score was used as the standard for species demarcation (Jain et al., 2018). Next, all 38 completely assembled genomes of K. michiganensis available in the NCBI database as of February 1, 2022 were retrieved, and a further ANI analysis using CP022348 as a reference was conducted. Genomes of K. michiganensis with ANI values >95% and KO_408 (metadata on these genomes are presented in Supplementary Material) were annotated using Prokka (Seemann, 2014). Orthologous groups were built using a Roary pipeline (Page et al., 2015). The resultant core-genome alignment file from Roary was inputted into Iqtree to construct a phylogenetic tree using 1,000 bootstrapping replicates (Nguyen et al., 2015). The iTOL was used to visualize and annotate the trees (Letunic and Bork, 2019).

The phylogeny of the IncX3-bla_NDM-5 plasmid (pKO_4-NDM-5) and 17 other plasmids sharing high homology with pKO_4-NDM-5 in a BLASTn analysis was assessed using MEGA X (Kumar et al., 2018). BRIG was used to compare pKO_4-NDM-5 with MH78170 (Alikhan et al., 2011). The genetic environment of bla_NDM-5 on these plasmids and that of LC000627 was compared using EasyFig v2.1 (Sullivan et al., 2011).

A single colony of Tf_Top10-NDM-5 was passaged on tryptic soy agar plate (TSA) containing no antibiotics and incubated at 37°C for 12 h. Successive passaging (200 passages) at the same interval and conditions was maintained for 100 consecutive days. The presence of the bla_NDM-5-containing plasmid was analyzed after every tenth passage for selected colonies.

Overnight cultures of recipient E. coli C600, transconjugant TcEC187 (an E. coli C600 transconjugant containing IncX3 bla_OXA-181 plasmid) described in our previous study (Prah et al., 2021), and transformant Tf_C600-NDM-5 from TSA plates were diluted to McFarland 2 with PBS. Portions of the bacterial suspensions (200 μl) were transferred into 10 ml of sterile LB media in 18 mm diameter Pyrex tubes. The tubes were incubated at 37°C for 24 h in a Bio Shaker BR-21FP (TAITEC Co., Ltd., Japan) set to a speed of 200 rpm. The tubes were connected to a Taitec ODboxC (TAITEC Co. Ltd., Japan) for continuous measurements of the bacterial optical density (OD) at 600 nm every 30 min. Growth kinetic assays were performed in triplicate, and the mean OD was plotted against time to construct the growth curve.

Overnight cultures of E. coli C600, Tf_C600-NDM-5, and Tc1EC187 were diluted to McFarland 2 with sterile PBS. A 1:1 ratio of E. coli C600 and Tf-NDM-5, E. coli C600 and Tc1EC187, or Tc1EC187 and Tf-NDM-5 was prepared, and 30 μl was transferred into 3 ml of LB. The cultures were incubated at different time intervals (0, 6, 12, and 24 h) at 37°C with shaking (200 rpm). Cultures were serially diluted at the end of the incubation period, and 100 μl (10⁻⁵ diluted culture) was spread on LB agar plates with or without antibiotics. E. coli C600 and Tf-NDM-5 culture combinations and Tc1EC187 and Tf-NDM-5 were selected on LB agar plates with or without 2 μg/ml meropenem, whereas E. coli C600 and Tc1EC187 combinations were selected on LB agar plates with or without 8 μg/ml ampicillin. The plates were incubated at 37°C for 18–24 h, and the number of colonies counted. The experiment was performed in duplicate and the mean colony count was estimated.

KO_408 was highly resistant to all β-lactam classes of antibiotics, except the monobactam, aztreonam. Resistance to other class of antibiotics, including tetracycline and quinoline, was observed (Table 1). This clinical strain from Japan (KO_408) was mCIM-positive and harbored bla_NDM-5. From the conjugal analysis, KO_408 successfully transferred bla_NDM-5 to E. coli J53, a recipient strain. S1-PFGE and subsequent Southern hybridization with DIG-labeled bla_NDM-5 showed that bla_NDM-5 was localized on a mobile genetic element ∼44.9 kb in size (Supplementary Figure S1). The transconjugant strain Tc-NDM-5 also showed high resistance to most of the β-lactam antibiotics, except for aztreonam (Table 1). Tc-NDM-5 did not confer resistance to any of the antibiotics in the other classes.

The ANI analysis identified KO_408 as K. michiganensis with an ANI score of 99.19%. Its genome comprised a chromosome of 6,018,476 bp and six other circular plasmids ranging in size from 2,569 bp to 286,463 bp (Table 2). This comprised a total of 6,037 protein-coding sequences (CDS) and accounted for a coding ratio of 87.3% (Supplementary Table S1). Related to its multidrug-resistant phenotype, KO_408 harbored resistance genes to β-lactams (bla_NDM-5, bla_OXA-1, and bla_OXY-1-7), tetracycline (tet(A)), quinolone (aac(6′)-Ib-cr, qnrS1), and sulfonamide/dihydrofolate reductase inhibitors (sul1, sul1, sul2, sul3, and dfrA12) (Table 2). Most antibiotic resistance genes were plasmid-mediated, with the majority occurring in the IncHI2 plasmid (Table 2). The bla_NDM-5 gene was localized to the 44,878 bp IncX3 plasmid. The chromosomally encoded β-lactamase gene variant bla_OXY-1-7 was the only antibiotic resistance gene present on the chromosome (Table 2).

KO_408 was ascribed to the novel sequence type ST256. To investigate the phylogeny of KO_408 within the purview of global K. michiganensis collections, the resultant core-genome-based phylogenetic tree was resolved into two main groups, the KoI phylogroup with distinct subclades and the KoV phylogroup (Figure 1). The ANI range within these phylogroups relative to CP022348 was 97.50–97.78, KoV and 98.49–100, KoI (Figure 1). These ANI and phylogenetic analyses confirmed the intraspecies relatedness of the strains within these two phylogroups. Strains within the KoV phylogroup were mostly characterized by a new bla_OXY variant with >99% identity to bla_OXY-5-1 (Figure 1). The KoI phylogroup was split into three sub-lineages, which did not show any specificity with the occurrence of a particular OXY variant. The occurrence of the major carbapenemase gene was not limited to any of the phylogroups, with bla_NDM-1, bla_KPC-2, and bla_NDM-5 as the most frequently detected carbapenemase genes. KO_408 clustered with a subclade of KoI, which was closely associated with carbapenemase genes. This subclade predominantly contained strains with OXY-1-7 and a new OXY variant with >99% identity to bla_OXY-1-1(Figure 1).

Virulence factors of KO_408 and the other strains of K. michiganensis isolates were investigated. Genes encoding type 1 and 3 fimbriae, iron-chelating sideropores enterobactin (ent), salmochelin, and aerobactin were mostly present in all the K. michiganensis genomes (Supplementary Figure S2). Iron-chelating sideropore yersiniabactin and catalase (katA) were mostly limited to strains in the KoI phylogroup, whereas only a few strains in both KoI and KoV phylogroups harbored genes for allantoin utilization (Supplementary Figure S2). Nearly half of the strains (n = 15/33, 45.4%) including KO_408 harbored astA gene, a heat stable enterotoxin, whereas only a strain belonging to the KoI phylogroup harbored the bacterial toxin colibactin (Supplementary Figure S2).

We determined the phylogenetic relatedness of the IncX3-bla_NDM-5 plasmid of KO_408 (pKO_4-NDM-5) and 17 other plasmids with high homology to pKO_4-NDM-5 by BLAST analysis. pKO_4-NDM-5 shared a close genetic distance with seven plasmids from China, and these plasmids were recovered from diverse bacterial hosts and sources (Figure 2). A circular comparison of pKO-NDM-5 and MH781720 in Figure 3 shows that these IncX3 plasmids share a large, conserved scaffold containing many conjugative genes, as previously reported (Zhu et al., 2020). A structural difference that could have resulted from a deletion was observed around the genetic features closer to the bla_NDM-5 portion of pKO_4-NDM-5 (Figure 3). However, a closer view of the genetic environment of bla_NDM-5 in these plasmids showed previously reported genetic features. The bla_NDM-5 gene was immediately followed upstream by the insertion sequence ISAba125 and downstream by the bleomycin resistance gene ble_MBL. These were found to be bracketed within the insertion sequences IS5 upstream and IS26 downstream. This IS5 and IS26 conserved region was almost shared by the plasmid sequence of the IncN plasmid harboring the first reported case of bla_NDM-5 in Japan, but the sequence lacked the insertion sequence IS26 (Figure 4).

To understand the recent dissemination of IncX3-bla_NDM-5-carrying plasmids in different bacterial hosts, fitness, stability, and competitive assays were performed. The growth curves of transformant bearing the IncX3-bla_NDM-5 plasmid and the transconjugant bearing the IncX3-bla_oxa-181-containing plasmid were compared to those of their host (E. coli C600). The acquisition of these IncX3-containing plasmids did not impose any fitness burden on the host cells (Figure 5). When measuring the relative fitness of strains carrying IncX3 plasmids in competing for resources in the same culture environment, Tf-NDM-5 competed with its host and TcEC187. As shown in Figure 5B, Tf-NDM-5 was outcompeted by both its host cell and TcEc187, whereas TcEC187 had similar performance with the E. coli host, suggesting different competitive strengths of these IncX3 plasmids. Despite the low competitiveness of strains with IncX3-bla_NDM5- plasmids, the IncX3 bla_NDM-5-carrying plasmid showed strong stability in E. coli Top 10 in antibiotic-free culture without apparent plasmid loss for 200 consecutive passages (Figure 5).