Molecular Characteristics of an NDM-4 and OXA-181 Co-Producing K51-ST16 Carbapenem-Resistant Klebsiella pneumoniae: Study of Its Potential Dissemination Mediated by Conjugative Plasmids and Insertion Sequences

ABSTRACT

The carbapenem-resistant Klebsiella pneumoniae (CRKP) strain GX34 was recovered from the respiratory tract of an elderly male with severe pneumonia, and only susceptible to amikacin, tigecycline, and colistin. Complete genome suggested that it belonged to K51-ST16 and harbored plasmid-encoded NDM-4 and OXA-181, located on IncFIB plasmid GX34p1_NDM-4 and ColKP3/IncX3 plasmid GX34p4_OXA-181, respectively. A series of transconjugants generated in the plasmid conjugation assays, including Escherichia coli J53-N1 (harboring a self-transmissible and bla_NDM-1-producing plasmid Eco-N-1-p), J53-N2 (harboring a bla_NDM-4-producing plasmid and a helper plasmid GX34p5), and J53-O (harboring a bla_OXA-181-producing plasmid), could be stably inherited after 10 days of serial passage and no significant biological fitness costs were detected. Furthermore, we first reported the bla_NDM-1 gene, derived from bla_NDM-4 mutation (460C>A) under meropenem pressure, could be in vitro transferred into a self-conjugative, recombined plasmid Eco-N-1-p of J53-N1. Eco-N-1-p was mainly recombined by GX34p4_OXA-181 (40,449 bp, 75.16%) and GX34p1_NDM-4 (8,553 bp, 15.89%), in which IS26 and IS5-like probably played a major role. Eco-N-1-p could be transferred into the conjugation recipient K. pneumoniae KP54 and make the latter sacrifice fitness. The retention rates of bla_NDM-1 remained high stability (>80% after 200 generations). The comparative genomic analysis of GX34 and those carrying bla_NDM-4 or bla_OXA-181 genes retrieved from the NCBI RefSeq database showed all bla_NDM-4 (26/26, 100.00%) and bla_OXA-181 (13/13, 100.00%) were surrounded by IS26. The immediate environment of bla_NDM-4 and bla_OXA-181 in GX34 and some retrieved strains shared identical features, hinting at their possible dissemination. Effective measures should be taken to monitor the spread of this clone.

INTRODUCTION

The past 2 decades have witnessed the rapid rise of carbapenem-resistant Enterobacterales (CRE) worldwide, which was highlighted as a significant threat by the WHO (1). In the SMART surveillance program assessing the global prevalence of carbapenemase-producing Enterobacterales (CPE), bla_OXA-48-like and bla_NDM genes were the second and third most commonly identified carbapenemase-encoding genes, accounting for 20.09% and 19.42%, respectively (2). New Delhi Metallo-β-lactamase (NDM) encoded by bla_NDM is a metallo-β-lactamase capable of hydrolyzing almost all β-lactams except aztreonam. NDM-4 differs from NDM-1 only by a single amino acid substitution (Met154Leu) and possesses increased hydrolytic activity toward carbapenems and cephalosporins (3). It was documented that NDM-4 located on a self-transmissible IncFIA plasmid in an Escherichia coli strain led to cryptic transmission in a Chinese hospital, facilitating further expansion of bla_NDM-4 and posing a potential threat to public health (4). Besides, OXA-181, encoded by bla_OXA-181, showed a slightly higher hydrolysis activity than OXA-48 (5). Certain high-risk clones, e.g., Klebsiella pneumoniae ST14 and E. coli ST410, have been associated with the global dispersion of OXA-181 (6).

The co-existence of dual carbapenemases (NDM and OXA) in CPE have been reported elsewhere, e.g., NDM-1 and OXA-232 in K. pneumoniae, and NDM-5 and OXA-181 in E. coli (7–10). However, the bla_NDM-4 and bla_OXA-181 coproducing CRKP strains are rarely documented in China and their dissemination remained unclear (11). Herein, we reported the co-occurrence of NDM-4 and OXA-181 in a K. pneumoniae strain in China. Revealing its potential resistance and dissemination mechanism will play a critical role in preventing the expansion of this clone.

RESULTS

Brief medical history of the patient and resistance phenotypes of GX34.

In our surveillance of the prevalence of CRE isolates (unpublished data), 91 CRE strains were collected from September 2019 to November 2021 at the Fourth Affiliated Hospital of Guangxi Medical University, Liuzhou, Guangxi, China. Forty-two of 91 CRE strains harbored bla_NDM, including 21 bla_NDM-5, 20 bla_NDM-1, and one bla_NDM-4. An NDM-4 and OXA-181 coproducing K. pneumoniae GX34 strain was identified from the bronchoalveolar lavage fluid (BALF) in a 79-year-old male patient with severe pneumoniae. We did not detect other strains carrying bla_NDM-4 and/or bla_OXA-181. Antimicrobial susceptibility testing (AST) results revealed that the GX34 strain conferred resistance to aztreonam, imipenem, meropenem, piperacillin-tazobactam, and cefepime, and susceptibility to amikacin, tigecycline, and colistin (Table 1). ECIM and mCIM tests indicated that it produced metallo-β-lactamase. The symptoms of the patient were significantly improved after treatment was transferred from ertapenem to tigecycline.

TABLE 1.

Carbapenemase and ESBL genes located on plasmids and MICs (mg/L) of GX34 K. pneumoniae strain and its transconjugants^a

Strains	Species	Carbapenemase genes	ESBL genes	MIC (mg/L)
				ATM	IMP	MEM	TZP	FEP	AMK	TIG	CST	LEV	TOB
GX34	K. pneumoniae	blaNDM-4, blaOXA-181	bla CTX-M-15	≥64	≥16	≥16	≥128	≥32	16	2	≤0.5	≥8	≥16
E. coli J53	E. coli			≤1	≤0.25	≤0.25	≤4	≤0.12	≤2	≤0.5	≤0.5	≤0.12	≤1
J53-N1	E. coli	bla NDM-1	bla SHV-12	≥64	≥16	≥16	≥128	16	≤2	≤0.5	≤0.5	≤0.12	≤1
J53-N2	E. coli	bla NDM-4		≤1	≥16	≥16	≥128	16	≤2	≤0.5	≤0.5	≤0.12	≤1
J53-O	E. coli	bla OXA-181		≤1	2	1	8	≤0.12	≤2	≤0.5	≤0.5	1	≤1
KP54	K. pneumoniae		blaCTX-M-65, blaTEM-1D, blaSHV-12	≥64	≤0.25	≤0.25	≥128	≥32	≥64	2	≤0.5	≥8	≥16
KP54-N	K. pneumoniae	bla NDM-1	blaCTX-M-65, blaTEM-1D, blaSHV-12	≥64	≥16	≥16	≥128	≥32	≥64	2	≤0.5	≥8	≥16

^aESBL, extended spectrum β-lactamase; ATM, aztreonam; IMP, imipenem; MEM, meropenem; TZP, piperacillin-tazobactam; FEP, cefepime; AMK, amikacin, TIG, tigecycline; CST, colistin; LEV, levofloxacin; TOB, tobramycin. Bold of MICs mean resistance.

Genetic features of GX34.

Whole-genome sequencing (WGS) showed that the genome of GX34 consisted of a chromosome (5,298,818 bp) and five plasmids, namely, GX34p1_NDM-4 (193,355 bp), GX34p2 (201,292 bp), GX34p3 (4,686 bp), GX34p4_OXA-181 (51,478 bp), and GX34p5 (5,250 bp), as shown in Table 2. Plasmid replicon analyses suggested that both GX34p2 and GX34p4_OXA-181 were hybrid plasmids for they belonged to IncFIB/IncFII and ColKP3/IncX3 types, respectively. Plasmid types of GX34p1_NDM-4, GX34p3, and GX34p5 belonged to IncFIB (pKPHS1), Col440I, and Col440II, respectively. Sequence types (ST) and capsule serotype analyses revealed that GX34 belonged to ST16 and K51. The isolate harbored multiple acquired determinants of resistance to β-lactams (bla_NDM-4, bla_OXA-181, and bla_CTX-M-15), quinolones (qnrS1, aac(6’)-Ib-cr), aminoglycosides (aac(6’)-Ib-cr, rmtB) and macrolides (mph(A)), detailed in Table 2.

TABLE 2.

Antibiotic resistant genes of GX34 K. pneumoniae and J53-N1 E. coli (carrying Eco-N-1-p plasmid) strains

Strains	Resistant genes
GX34	Position	Chromosome	GX34p1_NDM-4	GX34p2	GX34p3	GX34p4_OXA-181	GX34p5
	Size (bp)	5,298,818	193,355	201,292	4,686	51,478	5,250
	genes	bla SHV-194-like	blaNDM-4, rmtB	blaOXA-1, aac(6′)-Ib-cr, mph(A), blaCTX-M-15		blaOXA-181, qnrS1
J53-N1	Position	Chromosome	Eco-N-1-p
	Size (bp)		53,814
	genes		blaNDM-1, blaSHV-12

Transferability of bla_NDM-4 and bla_OXA-181 genes.

The results are shown in Fig. 1. Conjugation experiments showed that bla_NDM-4 was successfully transferred into E. coli J53 from GX34 strain at a frequency of approximately 1.34 × 10⁻⁷ cells per recipient. The transconjugant harboring bla_NDM-4 of E. coli J53 was named J53-N2 (Fig. 1A). The PFGE patterns of the two plasmids harbored in J53-N2 strain matched those of GX34p1_NDM-4 and GX34p5, respectively, suggesting the transmission of GX34p1_NDM-4 required the helper plasmid GX34p5 (Fig. 1A; Fig. S1). However, despite repeated efforts cautiously, no transconjugants carrying bla_OXA-181 alone were obtained if not under meropenem pressure.

FIG 1.

Formation, growth curves, and resistance plasmids stability of the transconjugants of K. pneumoniae GX34 strain. (A) Transconjugants were obtained from the recipient strain E. coli J53 or K. pneumoniae KP54 strain in conjugation experiments. (B) Growth curves of the transconjugants and the recipient strain E. coli J53 and KP54. (C) Stability of the bla_OXA-181- and bla_NDM-1/4-carrying plasmids along the 10-day serial passage.

To further evaluate the transferability of bla_NDM-4 and bla_OXA-181, GX34 and E. coli J53 were mixed at the ratio of 1:3 under 0.5 mg/L meropenem since the patient was treated with carbapenem antibiotics. We obtained the transconjugant E. coli J53-N1 twice under the same mating condition on different days, which carried a novel plasmid Eco-N-1-p, harboring bla_NDM-1 and bla_SHV-12 at frequencies of approximately 3.33 × 10⁻⁸ cells per recipient, and the transconjugant E. coli J53-O that carried one plasmid harboring bla_OXA-181 at frequencies of approximately 1.48 × 10⁻⁷ cells per recipient.

To explore the transferability of the plasmid Eco-N-1-p, conjugation experiments were performed with an amikacin-resistant K. pneumoniae KP54 as the recipient. Eco-N-1-p could be transferred into KP54 at 1.93 × 10⁻⁷ cells per recipient (Fig. 1).

The above results revealed that bla_NDM-4 and bla_OXA-181 could be transferred separately, especially after the inappropriate use of carbapenems (Fig. 1A; Fig. S1).

Fitness cost and plasmids stability of K. pneumoniae GX34 and transconjugants.

To investigate the effect of the acquiring of bla_NDM-1/4 or bla_OXA-181-carrying plasmids on biological fitness cost, the growth rates were measured. No significant differences (all P > 0.05) in the growth rates were observed between the recipient strain E. coli J53 and the homologous transconjugants (J53-N1, J53-N2 and J53-O) harboring bla_NDM-1/4 or bla_OXA-181 plasmids (Fig. 1B). By comparison, statistically significant growth inhibition was observed in KP54-N compared with KP54 (P = 0.0149) (Fig. 1B), showing that the bla_NDM-1-carrying KP54-N strain may sacrifice fitness.

The stability of transconjugants (J53-N1, J53-N2, J53-O, and KP54-N) was determined by a passage experiment (Fig. 1C). The retention rates of bla_NDM-4 or bla_OXA-181 remained 100% after approximately 200 generations, showing that the plasmid harboring bla_NDM-4 or bla_OXA-181 could be stably inherited. Although KP54-N gradually lost its bla_NDM-1-carrying plasmid after 6 days, its retention rates remained relatively stable (>80% after 200 generations).

Genetic environment analysis of bla_NDM-4.

The bla_NDM-4 gene was located on the plasmid GX34-p1_NDM-4 (CP104797.1), which belonged to the IncFIB group and also harbored an aminoglycosides resistance gene rmtB. The BLASTn analysis revealed that GX34-p1_NDM-4 exhibited high sequence similarity to other bla_NDM-4-carrying plasmids: pNDM4-191773 (CP080366.1, 99.98% identity and 43% coverage) and pSECR18-2374C (CP041930.1, 99.97% identity and 43% coverage) (Fig. 2B). Both plasmids belonged to IncFII group and were harbored in two ST16 K. pneumoniae strains. Furthermore, the genetic environment of our bla_NDM-4 was completely identical to those of the bla_NDM-4 in pNDM4-191773 and pSECR18-2374C, namely, ΔIS26-groL_1-groS-cutA-dsbD-trpF-ble-bla_NDM-4-ΔISAba125-ΔIS26 (9,378 bp in size). The comparisons of linear maps also revealed that the bla_NDM-4 surroundings in the above three plasmids had identical sequences (Fig. 2D).

FIG 2.

Comparative genomic analysis of resistance plasmids. (A) Circular alignment analysis of plasmid sequences of GX34p4_OXA-181, GX34-p1_NDM-4 and Eco-N-1-p. (B) Circular comparison between GX34-p1_NDM-4 in the current study and other homologous plasmids, pNDM4-191773 and pSECR18-2374C, available in the NCBI database. The outermost circles indicate the plasmid GX34-p1_NDM-4 with genes annotated. (C) Circular alignment analysis of plasmid sequences of GX34p4_OXA-181, CP090300.1 and three identical resistance plasmids (MN227183.1, MG893567.1, and KP400525.1). (D) Linear alignment of bla_NDM-4-bearing structure on GX34p4_OXA-181, Eco-N-1-p, GX34-p1_NDM-4, pNDM4-191773, and pSECR18-2374C, respectively. ORFs are indicated by arrows. Sequences of shared homology between five plasmids are marked by gray shading.

For comparative genomic analysis, a total of 25 bla_NDM-4-harboring plasmids worldwide were retrieved from the NCBI genome database. As shown in Fig. 3, these plasmids are mainly harbored in E. coli (13, including three strains of ST405 and 2 of ST101) and K. pneumoniae (seven, including four strains of ST16). The maximum-likelihood phylogenetic tree was built to illustrate their relationship. The strains harboring CP080366.1, GX34p1_NDM-4 (this study), CP050167.1, CP028588.1, CP032878.1, and CP022226.2, collected in China from 2015 to 2020, were classified into a cluster. The plasmid types included IncX3 (11/26, 42.31%), IncFII (6/26, 23.08%), lncFIA (4/26, 15.38%), lncFIB (2/26, 7.69%), lncX3/lncA/C2 (1/26, 3.84%), lncHI2 (1/26, 3.84%), and lncA/C2 (1/26, 3.84%), respectively. The prediction of transferability revealed that four plasmids, CP041642.1, AP018139.1, CP095684.1, and CP050167.1, consisted of four modules mediating self-transmissible mobile genomic element (MGE) typically, and 15 other plasmids consisted of three modules without oriT (Fig. 3).

FIG 3.

Comprehensive information and evolution relationship among 26 bla_NDM-4-bearing plasmids. Plasmids were colored according to strains, STs, plasmids replicons, sample, putative transferability (four modules: origin of transfer site [oriT], relaxase gene, gene encoding type IV coupling protein [T4CP], and gene cluster for bacterial type IV secretion system [T4SS]), year, country, and genetic context of bla_NDM-4. The maximum-likelihood phylogenetic tree was built by Neighbor-Joining Algorithm from an alignment generated by MEGA-X. The Interactive Tree of Life (https://itol.embl.de) was used for visualization. The bootstraps are shown as blue circles on the branches.

The bla_NDM-4-surrounding environments in 26 plasmids were further analyzed. A total of 14 bla_NDM-4 genes were flanked by two IS26 elements, thus forming a pseudo-composite transposon, similar to ISAba125-formed composite transposon consisting bla_NDM-1 except for the inserted sequence types. The nine pairs of IS26 elements of the above pseudo-composite transposons were mainly located at the 1,100 bp upstream and 7,500 bp downstream of bla_NDM genes, respectively. By comparison, 12 bla_NDM-4 genes were flanked by a single IS26 element.

Eleven bla_NDM genes were sided by the small truncated ISAba125s, which were located at closer sites to bla_NDM than IS26 (10/11) and IS5(1/11).

Genetic context analysis of bla_OXA-181.

In GX34, the bla_OXA-181 gene was located in a hybrid plasmid GX34p4_OXA-181, and flanked by a putative composite transposon consisting of two copies of IS26 elements, as shown in Fig. 4A.

FIG 4.

Comprehensive information of the assembled draft genomes from 13 bla_OXA-181-bearing plasmids. (A) Comparison of bla_OXA181-bearing plasmids from this study with 12 complete resistance plasmids downloaded from the RefSeq database in NCBI. The maximum-likelihood phylogenetic tree was built by Neighbor-Joining Algorithm from an alignment generated by MEGA-X. Host, type of plasmids, strain, putative transferability, ST, history of foreign travel, year, and sample were shown in different colors. Linear alignment of bla_OXA-181-bearing structure also was performed to evaluate transferring-related ISs. (B) Comparison of bla_OXA-181-bearing K. pneumoniae from this study with four complete genomes downloaded from the RefSeq database. The maximum-likelihood phylogenetic tree was built by RaxML from an alignment generated by SNPs and filtered to remove recombination using Gubbins. The similarity between core genomes of bla_OXA-181-bearing K. pneumoniae is defined as the coverage of homology regions for query core genome (row-wise) and subject core genome (column-wise).

The structure of the bla_OXA-181-harboring genetic context was ΔIS3000-IS26-qnrS1-hin-ISKpn19-bla_OXA-181-ΔIS3000-IS26 (14,843 bp in size). The hin gene encoding DNA-invertase further facilitates the activation of bla_OXA-181-harboring genetic context. The identical genetic environment surrounding bla_OXA-181 was also found in the plasmid pOXA181_EC14828 (KP400525.1), located in the first-reported OXA-181-producing E. coli strain in China (Fig. 2C).

Furthermore, over 30 plasmids were detected in K. pneumoniae and E. coli strains worldwide, and exhibited 100% identity and 100% coverage to the GX34p4_OXA-181 plasmid via online BLASTn analysis, detailed in Table S1.

A total of 12 bla_OXA-181-carrying strains from China having complete data were enrolled for further analysis, including five strains of E. coli (five isolated from humans and one from pig), five K. pneumoniae (human), one Pseudocitrobacter faecalis (human), and one Morganella morganii (human), as shown in Fig. 4A.

The phylogenetic tree of the 12 bla_OXA-181-carrying plasmids and GX34p4_OXA-181 was established. All these plasmids belonged to the IncX3 group. CP090300.1 presented 99.98% identity at 95% coverage to GX34p4_OXA-181 (Fig. 2C). Putative transferability revealed that two of 13 plasmids (CP090300.1 and CP084242.1) consisted of four modules mediating the self-transmissible MGEs, and 10 of 13 harbored three modules (Fig. 4A). The IS3000 near bla_OXA-181 in all 13 plasmids were truncated and close to IS26 (13/13, 100.00%), ISKpn19 (13/13, 100.00%), and ISKox3 (12/13, 92.31%), respectively. Double IS26s flanking bla_OXA-181 (13/13) formed composite transposon and direct repeats (DR, 11 GT and two AC) revealed transpositions happened. The structure ΔISEcp1-bla_OXA-181-ΔlysR-Δere (Tn2013) existed in all 13 plasmids. Moreover, qnrS1-mediating quinolones resistance also was close to bla_OXA-181 (12/13, 92.31%).

The maximum-likelihood phylogenetic tree of five bla_OXA-181-bearing K. pneumoniae was built (Fig. 4B), suggesting a better homology (99.00% coverage) between GX34 and K191773 (from Jiangsu province in China) and showing the potential dissemination of GX34.

Recombination of NDM-1-producing Eco-N-1-p mediated by insertion sequences.

The plasmid Eco-N-1-p, harbored in J53-N1 and obtained in conjugation experiment by using 0.5 mg/L meropenem (Fig. 1), was different from any plasmid of GX34. It was sequenced to illustrate its derivation and formation mechanism. The results of nanopore long-read sequencing demonstrated Eco-N-1-p (53,814-bp, CP104795.1) harbored bla_NDM-1 rather than bla_NDM-4, as confirmed by PCR and Sanger sequencing. Moreover, bla_SHV-12 also existed in the Eco-N-1-p plasmid and mediated resistance to aztreonam. Comparative genomic analysis showed that Eco-N-1-p was mainly derived from GX34p4_OXA-181 (40,449 bp, 75.16%) and GX34p1_NDM-4 (8,553 bp, 15.89%), and minorly from chromosomes of GX34 (3813 bp, 7.09%) and GX34p2 (776 bp, 1.44%), as detailed in Fig. 2A. Eco-N-1-p exhibited the most closely match to plasmid pNDM1_140542(CP103366.1) of a Citrobacter freundii strain (100.00% identity and 98% coverage).

The proposed mechanism for the formation of Eco-N-1-p was shown in Fig. 5, in which insertion sequences probably played a major role. RT-qPCR was used to confirm the proposed mechanism by the expression changing of recombination-related insertion sequences, resistance genes and replicons. Results showed that the expression levels of IS26 (2.82 ± 0.16-fold) and IncFIB (3.84 ± 0.64-fold) in mixed culture (GX34 and E. coli J53) processed with 0.5 mg/L meropenem significantly increased compared without. However, the expression levels of IS3000, IS1R, IncX3, bla_OXA-181 and bla_SHV showed no significant difference with or without meropenem (Fig. 6).

FIG 5.

Recombination of NDM-1-producing Eco-N-1-p mediated by insertion sequences. Proposed formation mechanism of Eco-N-1-p is as follows by sequence alignment. Step 1: IS1R-like forming translocatable unit (TU) was infused to GX34p4_OXA-181 to generate cointegrate 1 while bla_OXA-181 was removed by IS26-bla_OXA-181 TU, including ColKP3. Step 2: bla_SHV-12 from chromosome was mediated IS26 by TU and integrated into cointegrate 1 to form cointegrate 2 whileΔISEc63 alone was lost. Step 3: IS5-like from chromosome mobilized surrounds of bla_NDM-4 (about 7738 bp) to form IS5-like-bla_NDM-4 TU with a single base substitution (460C>A). Then IS5-like-bla_NDM-1 TU was inserted cointegrate 2 to generate Eco-N-1-p.

FIG 6.

Expression of resistant genes, replicons, and insertion sequences relative to the housekeeping gene rpoB in strains exposed to various concentrations of meropenem. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. MEM, meropenem.

DISCUSSION

The emergence of K. pneumoniae carrying dual carbapenemases is rarely reported in China (11). Herein, we identified a bla_NDM-4 and bla_OXA-181-co-harboring CRKP strain from an elderly male, which belonged to K51-ST16, an international cephalosporin- and carbapenem-resistant clone (12). Besides, it also harbored aac(6′)-Ib-cr, conferring resistance to tobramycin and amikacin, and bla_CTX-M-15, producing extended-spectrum beta-lactamase (ESBL), posing an additional challenge to treatment (13). As mentioned previously, bla_NDM and bla_OXA genes have a risk of rapid dissemination (6–9, 11, 14), and therefore, the transferability of bla_NDM-4 and bla_OXA-181 harbored in the CRKP strain was investigated in the study.

The diverse transconjugants of our isolate GX34 under different conditions suggested its capacity for potential dissemination. In the plasmid conjugation assays of GX34 with E. coli J53 as the receptor, with or without sub-MIC meropenem, the different transconjugants were obtained, including J53-N1, J53-N2 and J53-O, hinting the diverse dissemination means of bla_NDM-4 and bla_OXA-181 genes. J53-N1 contained only one self-transferable recombined IncX3-type plasmid Eco-N-1-p. The comparative genomics suggested that Eco-N-1-p was possibly generated by multistep recombination, including the interchanging of the surroundings of bla_NDM in GX34p1_NDM-4 with those of bla_OXA in GX34p4_OXA-181. To our knowledge, this recombination phenomenon was first reported. Previous studies found the plasmid rearrangement was mostly mediated by reverse transcriptase (e.g., ltrA) or insertion sequences (e.g., IS26, IS903B), accompanied by the transfer of resistance or hypervirulence genes (15, 16). Fig. 5 showed the proposed formation mechanism of Eco-N-1-p plasmid, which might be mediated by insertion sequences due to the recombination sequences flanked by insertion sequences (e.g., IS26 and IS5-like) and the rise in the expression level of insertion sequences (IS26 and IncFIB). There are generally two means of IS26-medicated intermolecular recombination: cointegrates using the host homologous recombination under the help of recA gene that probably acts as resolvase, and translocatable units through transposases (17). In the current study, the overexpression of the transposase of IS26 in GX34p1_NDM-4 might suggest the recombination was medicated by the translocatable units. Although no significant increase in the expression of IS5-like and bla_NDM, the sequence analysis suggested that IS5-like and bla_NDM might participate in the evolution of Eco-N-1-p. The overexpression of IncFIB would provide a dynamic structure in replication and increase the recombination rates. The bla_NDM-1 in Eco-N-1-p differed from bla_NDM-4 by a single nucleotide substitution (460C>A), which might be caused by the meropenem pressure. The similar resistance phenotypes of the bla_NDM-4- and bla_NDM-1-harboring K. pneumoniae strains might make this substitution unnoticed and underestimated. The genetic shift between different bla_NDM genes has been previously documented, e.g., a change from NDM-1 (2018 to 2019) to NDM-9 (2020) ST147 K. pneumoniae strains was observed in Italy during two sequential outbreaks, and the genomic and phylogenetic analyses suggested relatedness of these strains (18). Our study also showed that the bla_NDM-1 derived from bla_NDM-4 could be mobilized into a recombined and self-transmissible plasmid and subsequently the latter was transferred into other strains via conjunction. The J53-N2 harboring bla_NDM-4 was obtained with the helper plasmid GX34p5, suggesting GX34-p1_NDM-4 could not be self-transferred. By comparison, J53-O carrying bla_OXA-181 only contained one plasmid identical to that in GX34p4_OXA-181, revealing its self-transferability. Interestingly, the chromosomal narrow-spectrum bla_SHV-194-like gene was also found to turn into the ESBL-encoding bla_SHV-12 with bla_NDM mutation (bla_NDM-4 to bla_NDM-1) in Eco-N-1-p plasmid, posing an extensive threat to the emergency of antibiotic resistance during the treatment.

The genome characteristic of bla_NDM-4-carrying CKRP has been rarely documented. In our study, the similar structure groL_1-groS-cutA-dsbD-trpF-ble-bla_NDM of GX34-p1_NDM-4 was found in the contigs carrying bla_NDM (over 1/3 in 7,148 contigs) and flanked by ISAba125, IS3000, and IS26 (>90%), respectively, suggesting the potentiality of insertion sequence-mediated horizontal transfer of this structure, which was consistent that mobile genetic elements played a significant role in the global dissemination of bla_NDM by Acman et al. (19). The STs of above 7 bla_NDM-4-carrying K. pneumoniae retrieved from NCBI database majorly belonged to ST16 (4/7), possibly suggesting these high-risk clones promote their dissemination. The immediate environment of bla_NDM-4 indicated that the gene was close to IS26, IS5, and truncated ISAba125. However, the truncated ISAba125 was nonfunctional, hinting that IS26 and IS5 elements probably mediated the mobilization of bla_NDM-4 and gradually replaced ISAba125. Increasing risk of dissemination could be driven by IS26 via forming pseudo-composite transposons or mediating the evolution of recombined, self-transferring plasmids (20). Although GX34p1_NDM-4 only contained three classical modules mediating self-transferability, it could be transferred among bacteria of different species (e.g., from E. coli to K. pneumoniae) and then disseminated. More importantly, GX34p1_NDM-4 and the plasmids (CP080366.1, CP050167.1, CP028588.1, CP032878.1, and CP022226.2) were classified into a cluster, suggesting there was a potential horizontal spread.

The bla_OXA-181 gene, which likely originated from the chromosome of Shewanella xiamenensis, was first described and mostly detected in K. pneumoniae (6). OXA-181-producing E. coli and K. pneumoniae in China were documented in 2015 and 2020, respectively (21, 22). The bla_OXA-181-carrying plasmids, including GX34p4_OXA-181 and over 30 plasmids worldwide in NCBI database, shared 100% identity and coverage, suggesting its horizontal spread (6). Plasmid phylogenetic tree and sequence analyses also suggested the plasmids of MN227183.1, MG893567.1, and KP400525.1 were identical plasmids from different provinces in China (Fig. 2C and 6A). Tn2013, composed of bla_OXA-181, ISEcp1 (located upstream) and ΔlysR-Δere (located downstream) (6), existed in all 13 complete bla_OXA-181-carrying plasmids in China. Genetic surroundings of bla_OXA-181 revealed ISEcp1 belonging to Tn2013 was always truncated by IS3000 (100%), implying that IS3000 most likely mediated the movement of bla_OXA-181. Further, 13 of 13 bla_OXA-181-harboring plasmids possessed two IS26 elements and DR, suggesting that bla_OXA-181 be horizontally transferred via IS26-mediated transposition, resulting in the dissemination of bla_OXA-181. Putative transferability analyses suggested that 12 of 13 bla_OXA-181-harboring plasmids (more than three molecules) have potential transferability as GX34p4_OXA-181 and GX34-p1_NDM-4. The maximum-likelihood phylogenetic tree of five bla_OXA-181-bearing K. pneumoniae suggested GX34 and K191773 (ST16 K. pneumoniae) isolated sputum and reported in China (11), existed a closer homology and those all carried bla_NDM-4 and bla_OXA-181. Likely, GX34 is similar to SECR18-2374 (ST16 and coproducing NDM-4 and OXA-181) in resistance reported by Kim JS in South Korea (10). SECR18-2494 was found in the same hospital as SECR18-2374 and both had an identical XbaI PFGE. That indicated dissemination of coproducing NDM-4 and OXA-181 K. pneumoniae has already happened with a potential means.

In conclusion, we identified a CRKP strain cocarrying bla_NDM-4 and bla_OXA-181, which could be transferred into K. pneumoniae or other Enterobacterales strains in diverse ways. A self-transmissible recombination plasmid Eco-N-1-p was generated in which IS26 and IS5-like probably played a major role. Moreover, bla_NDM variants (460C>A) from bla_NDM-4 to bla_NDM-1 occurred under meropenem pressure. The comparative genomics suggested that the bla_NDM-4 and bla_OXA-181 genes were always surrounded by IS26 and identical surroundings of bla_NDM-4 or bla_OXA-181 were found on other genomes retrieved from NCBI database, indicating their potential dissemination mediated by IS26. This represented a threat to public health and effective measures should be taken to prevent the dissemination of this clone.

MATERIALS AND METHODS

Identification of K. pneumoniae GX34 strain and its clinical data collection.

The K. pneumoniae GX34 isolate, collected from our surveillance of the prevalence of CRE isolates, harbored the bla_NDM and bla_OXA-48-like genes encoding carbapenemases by using PCR. Species identification was determined by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) (Bruker Daltonik, Bremen, Germany), as described previously (23).

AST.

In vitro susceptibility tests were determined by the N335 susceptibility cards and Vitek-2 system (bioMérieux, France), including amikacin, minocycline, doxycycline, ceftazidime, cefepime, piperacillin-tazobactam, aztreonam, levofloxacin, ciprofloxacin, and sulfamethoxazole/trimethoprim. The MICs of imipenem, meropenem, tigecycline, and colistin were determined by the microdilution broth method (bio-KONT, Ltd. China) with E. coli ATCC 25922 as the quality control strain, as we previously described (24).

The breakpoint of tigecycline was defined by the U.S. Food and Drug Administration (FDA) on Antimicrobial Susceptibility Testing (25). The results of other antimicrobial agents were interpreted following the standards of the Clinical and Laboratory Standards Institute (CLSI, 2021) (26). The production of carbapenemases was determined using the modified carbapenem inactivation method (mCIM) and EDTA-modified carbapenem inactivation method (eCIM), as recommended by CLSI 2021 (26).

WGS and bioinformatic analysis.

WGS was performed using Illumina HiSeq 2500 platform (for GX34) and nanopore sequencing method on MinION flow cells (for GX34, and E. coli J53-N1, the transconjugant of GX34, harboring a self-transmissible and bla_NDM-1-producing plasmid Eco-N-1-p). Raw reads were filtered to remove the low-quality sequences and adaptors using skewer (27) and PoreChop (https://github.com/rrwick/Porechop), respectively. De novo assembly was conducted via SPAdes Genome Assembler v3.13.1 (28) and Unicycler (29). Gene prediction for 38 genomes, including one from this study (GX34) and 37 retrieved from the NCBI genome database (Table S1), was performed using Prokka 1.12 (30). Insertion sequences were identified using the ISfinder database (31). The antimicrobial resistance genes, multilocus sequence types (MLST) and plasmid replicon were analyzed via the CGE server (https://cge.food.dtu.dk/services/). K type was determined using Kaptive tool (32). The single nucleotide polymorphism (SNP) was determined using Snippy (https://github.com/tseemann/snippy). Linear alignments of bla_NDM-4- and bla_OXA-181-bearing structures were generated using genoPlotR and gggenes in R-4.1.2. Core genome sequences were compared using blastn + 2.13.0. The transferability of bla_NDM-4 or bla_OXA-181-carrying plasmids was evaluated by oriTfinder (33).

Plasmid conjugation assays.

As shown in Fig. 1, the plasmid conjugation experiments were performed for K. pneumoniae strain GX34 and J53-N1, as described previously (24). Azide-resistant E. coli J53 and amikacin-resistant K. pneumoniae KP54 were used as the recipient strains, respectively, and their resistance features were shown in Table 1. KP54 was recovered from the urine specimen of a patient with the urinary tract infection visiting China-Japan Friendship Hospital in 2020 and its IncR-type plasmid was knocked out.

Briefly, both GX34 and recipient E. coli J53 were adjusted to a McFarland standard of 0.5 and mixed at a ratio of 1:3, and a 0.1-mL aliquot of mixture was transferred into LB broth with or without meropenem, respectively. After an 18-h incubation at 37°C, 150-μL cultures were streaked onto China blue agar (CBA, addition of rosolic acid as a pH indicator) plates containing both azide (150 mg/L) and meropenem (1 mg/L), to screen the bla_NDM or bla_OXA-carrying transconjugants. Similarly, J53-N1 and recipient KP54 were mixed and transconjugants carrying bla_NDM-1 were also selected on CBA plates containing both amikacin (16 mg/L) and meropenem (1 mg/L).

The above transconjugants were confirmed by PCR, pulsed-field gel electrophoresis (PFGE) (34), and antimicrobial susceptibility testing. Frequencies of conjugation transfer were calculated by the number of transconjugants per recipient.

Growth curve assay and plasmid stability of GX34 and transconjugants.

The growth curve assay and plasmid stability were performed as previously described but with slight modifications (35). Growth curves of the recipients J53, KP54 and transconjugants were measured during the exponential growth phase as the maximum increase in optical density over time. Briefly, the recipients E. coli J53, KP54 and transconjugants were inoculated and shaken at 200 rpm overnight at 37°C in 10-mL LB broth. The overnight cultures were diluted and incubated to reach the logarithmic phase. Then the cultures were diluted and incubated at 37°C for 25 h to measure the optical density values (OD₆₀₀). The growth curves were estimated by R-4.1.2 using a one-way analysis of variance (ANOVA) followed by Tukey’s multiple-comparison tests.

The stability of GX34 and transconjugants were also evaluated. GX34 and transconjugants were grown in broth and were transferred daily (24-h intervals) at a 1:1,000 dilution into fresh LB broth for 10 consecutive days (approximately 200 generations). The cultures at the second, fourth, sixth, eighth, and 10th days were serially diluted and streaked onto the antibiotic-free LB agar. Approximately 50 colonies were randomly selected to identify the retention of bla_NDM-1/4 and bla_OXA-181 using PCR. All the above experiments were conducted in triplicate on different days.

mRNA expression of recombination-related insertion sequences, resistance genes, and replicons.

The expressions of the recombination-related insertion sequences (36), resistance genes, and replicons of GX34 and E. coli J53 strains treated with sub-MIC meropenem during the exponential growth phase at 37°C, were measured using quantitative reverse transcription-PCR (RT-qPCR) on the Applied Biosystems QuantStudio 5 real-time PCR system (Thermo Fisher Scientific). The sub-MIC meropenem was used since the patient was administered with carbapenems. The total RNA was extracted by using EASY spin bacterial RNeasy minikit and reverse-transcribed into cDNA using PrimeScript RT reagent kit (Qiagen, Germany). The qPCR was used to quantify gene expression using SYBR Premix Ex Taqv (TaKaRa, Japan). The rpoB gene was used as the internal control. Primer sequences are listed in Table S2. The fold changes in the expression of these genes were calculated by 2^–ΔΔCT as previously described (37). All tests were repeated in triplicate.

Comparative genomic analysis of bla_NDM-4 and bla_OXA-181 carrying plasmids.

To better understand the features of genetic environments surrounding bla_NDM-4 and bla_OXA-181 genes, we searched the RefSeq database on NCBI and obtained 25 intact plasmids harboring bla_NDM-4 worldwide, and 12 intact plasmids harboring bla_OXA-181 from China as of March 15, 2022. The blasting was performed.

Phylogenetic analysis.

Multiple sequence alignments from relevant plasmids were built by Neighbor-Joining Algorithm from the alignment results generated by MEGA-X (38). The maximum-likelihood phylogenetic tree of bla_OXA-181-bearing K. pneumoniae strains in this study with four complete genomes was built by RaxML (39) from the alignment generated by SNPs and filtered to remove recombination using Gubbins v2.4.1 (40). The iTOL (https://itol.embl.de) was used for visualization.

Statistical analyses.

GraphPad Prism 8.2.1 was used for data analyses. Data are expressed as mean ± standard deviation. Significant differences were assessed using a one-way analysis of variance, with P < 0.05 being considered statistically significant.

Ethics statement.

Permission for using the information in the medical records of the patient and the K. pneumoniae isolates for research purposes was granted by the Ethics Committee of the China-Japan Friendship Hospital (2022-KY-054).

Data availability.

Sequences of GX34 and Eco-N-1-p were deposited to National Center for Biotechnology Information (NCBI) in BioProject: PRJNA881721.