In total, 94 carbapenem-resistant K. pneumoniae isolates were collected from clinical specimens in a tertiary hospital in Lishui, China, between January 2017 and December 2023. The presence of the bla_NDM gene was confirmed by PCR and sequencing; only 27 bla_NDM-positive strains were included in this study.

Species identification was performed using VITEK MS. MICs were determined by the broth microdilution method according to CLSI 2024 standards; the results with tigecycline were interpreted according to FDA guidelines. The quality control strains included E. coli ATCC 25922, P. aeruginosa ATCC 27853, and S. aureus ATCC 25923.

PCR amplification was performed to detect a broad panel of resistance genes, including carbapenemases (bla_NDM, bla_KPC, bla_VIM, bla_IMP, bla_GES, bla_SME, bla_IMI, bla_SIM, bla_GIM, bla_SPM, bla_OXA-23, bla_OXA-24, bla_OXA-58), extended-spectrum β-lactamases (bla_CTX-M, bla_TEM, bla_SHV), and plasmid-mediated quinolone resistance genes (qnrA, qnrB, qnrC, qnrD, qnrS, qepA, aac (6′)-Ib-cr). Primers were designed based on the conserved regions of the target genes using Primer Premier 5.0 software. The specific primer sequences are listed in Supplementary Table S1. All primers were synthesized by Qingke Biotechnology Co., Ltd. (Hangzhou, Zhejiang Province, China). Products were analyzed by 1% agarose gel electrophoresis at 120 V for 30 min and subsequently sequenced. The sequences were compared against the NCBI BLAST database for confirmation.

Multilocus sequence typing was performed by PCR amplification and sequencing of housekeeping genes using primers and conditions provided by the K. pneumoniae MLST database (https://bigsdb.pasteur.fr/klebsiella/). Allele profiles were assigned to sequence types.

Bla_NDM-carrying plasmids were transferred into Trans1-T1 competent cells or J53 cells via transformation or conjugation. Transformants/conjugants were confirmed by VITEK MS and PCR. Plasmids were extracted using an EasyPure MiniPrep Kit and typed by PCR targeting plasmid replicons.

Genomic DNA from the bla_NDM/bla_KPC-positive ST11 strain CRKP26 was extracted using an EasyPure Bacteria Genomic DNA Kit. Whole-genome sequencing was performed using a hybrid strategy of combining next-generation sequencing (NGS) on an Illumina NovaSeq platform with third-generation single-molecule sequencing on an Oxford Nanopore platform. Libraries with different insert sizes were constructed for each platform, and the sequencing reads were subsequently assembled to generate the complete genome. Annotation was performed with Prokka, and gene function was analyzed using the GO and KEGG databases. KL typing of CRKP26 was performed using Kaptive. Comparative mapping of the plasmids was performed using the Basic Local Alignment Tool (BRIG), a visualization tool for genomic comparison. Differential genes were analyzed via BLAST.

All animal procedures were approved by the Institutional Animal Care and Use Committee of Lishui University (protocol number 2025D179).

To evaluate the in vivo virulence of the strains, we performed a median lethal dose (LD₅₀) assay using BALB/c mice and a modified Karber’s method. Specific pathogen-free male BALB/c mice (6–8 weeks old and weighing 25–30 g) were obtained from the Experimental Animal Center of Lishui University. Briefly, 6–8-week-old male BALB/c mice were randomly assigned to various experimental groups, with six mice per group. The control strain NTUH-K2044, a well-characterized hypervirulent Klebsiella pneumoniae isolate originally obtained from a patient with pyogenic liver abscess in Taiwan, and the test strain CRKP26 were cultured, and bacterial suspensions were prepared in saline at an initial concentration of 1 × 10⁷ CFU/mL as the stock solution. This was followed by two-fold serial dilutions to achieve different challenge doses, and the actual inoculum concentration for each dilution was verified by plate counting. Mice were inoculated via intraperitoneal injection with 0.1 mL of the corresponding bacterial suspension, while control group mice received an equal volume of sterile saline. Following inoculation, the mice were monitored continuously for 7 days, with survival status and time of death recorded meticulously. The LD₅₀ values for each strain were calculated based on observational data. According to the criteria established by the Chinese expert consensus on the diagnosis, treatment, and prevention of hypervirulent carbapenem-resistant Klebsiella pneumoniae infections, an LD₅₀ value≤1 × 10⁶ CFU defines a hypervirulent strain.

The mice were housed under standard laboratory conditions with free access to food and water and acclimatized for 7 days before the experiments. Bacterial inocula were prepared by culturing the strains overnight in Luria–Bertani broth, followed by centrifugation and resuspension in sterile normal saline. The concentration was adjusted to 1 × 10⁷ CFU/mL on the basis of the OD₆₀₀ and confirmed by plating. Each mouse was intraperitoneally injected with 100 μL, 200 μL or 300 μL of bacterial suspension. The control mice received normal saline. Groups of 10 mice were used per strain. The mice were monitored at 6-h intervals for 60 h post-injection. Survival was recorded, and Kaplan–Meier survival analysis was performed.

Suspensions of the strains ATCC 13883, a type strain of Klebsiella pneumoniae obtained from the American Type Culture Collection (ATCC), NTUH-K2044 and the test isolate strain CRKP26 were adjusted to 0.5 McFarland turbidity and then diluted 1: 100 in phosphate-buffered saline containing 5% serum to approximately 1 × 10⁶ CFU/mL. Suspensions were incubated at 37 °C with shaking at 80 rpm for 20 min prior to use. Human neutrophils were isolated from heparinized whole blood by density gradient centrifugation. Following erythrocyte lysis and washing, the neutrophils were resuspended in PBS containing 10% serum at 1 × 10⁶ cells/mL and used within 1 h. For the killing assay, 300 μL of neutrophil suspension was mixed with 300 μL of bacterial suspension in the experimental group, while 300 μL of PBS was mixed with 300 μL of bacterial suspension in the control group. The samples were incubated at 37 °C with shaking at 180 rpm for 2 h. Following incubation, 0.1% saponin solution was added for neutrophil lysis, the samples were diluted 1: 100, and 10 μL aliquots were plated on blood agar plates. The plates were incubated at 37 °C overnight for colony enumeration.

The serum survival assay was performed as follows. Briefly, bacterial cells from the mid-log phase culture were harvested and adjusted to a concentration of 1.5 × 10⁶ CFU/mL. The bacterial suspension was subsequently mixed with normal human serum (NHS) at a volumetric ratio of 1: 3. The mixture was then incubated at 37 °C for 2 h to allow bactericidal activity. Following incubation, the reaction mixture was serially diluted in sterile phosphate-buffered saline (PBS). To enumerate the viable bacteria, aliquots of the dilutions were plated onto LB agar plates and incubated overnight at 37 °C. The surviving bacteria were quantified by counting the resulting colonies.

All analyses were performed using SPSS 25.0. p < 0.05 was considered to indicate statistical significance. Statistically significant differences are indicated as follows: ns, not significant; p > 0.05; *: p ≤ 0.05; **: p ≤ 0.01; ***: p ≤ 0.001.

Between January 2017 and December 2023, 27 NDM-producing K. pneumoniae isolates were identified among 94 NDM-producing Enterobacteriaceae samples collected from a tertiary hospital in Lishui, China. Carbapenemase gene analysis revealed that 15 isolates (55.6%) carried bla_NDM-1 and 12 (44.4%) harbored bla_NDM-5. Two isolates (7.4%) coproduced both bla_KPC-2 and bla_NDM-1. Additional β-lactamase genes detected included bla_SHV in 25/27 isolates (92.6%), bla_TEM in 16/27 (59.3%), and bla_CTX-M in 10/27 (37.0%). The full set of resistance genes detected in each isolate, including carbapenemases, ESBLs, and plasmid-mediated quinolone resistance genes, is provided in Supplementary Table S2.

All 27 isolates were resistant to cephalosporins (ceftazidime, cefotaxime, and cefepime) and carbapenems (meropenem and imipenem). The MICs of carbapenem ranged from 4 to 64 μg/mL for imipenem and from 8 to 256 μg/mL for meropenem (Table 1). The isolates showed variable resistance to aztreonam (15/27, 55.6%), levofloxacin (14/27, 51.9%), and amikacin (3/27, 11.1%) but remained uniformly susceptible to polymyxin B and tigecycline (both 27/27, 100%). The two dual-carbapenemase producers displayed distinct levels of carbapenem resistance: strain CRKP26 exhibited extreme resistance, with meropenem and imipenem MICs of 256 and 64 μg/mL, respectively, which were significantly higher than those of strain CRKP11 (16 and 8 μg/mL, respectively). Strain CRKP26 showed pan-β-lactam resistance, including resistance to aztreonam (MIC ≥256 μg/mL) and amikacin (MIC ≥1,024 μg/mL), and remained susceptible only to colistin (MIC ≤0.5 μg/mL) and tigecycline (MIC = 0.5 μg/mL) (Table 1).

MLST analysis revealed high clonal diversity, with 19 distinct sequence types (STs) identified among the 27 NDM-producing isolates, in contrast to the predominant clonal expansion of KPC-2-producing ST11 strains observed during national surveillance. The most prevalent STs were ST11, ST340, and ST485, each accounting for 3 isolates (11.1%), followed by ST1131 and ST218, with 2 isolates (7.4%) each. The remaining 14 sequence types were each represented by a single isolate (Table 1). Among the isolates coproducing KPC-2 and NDM-1, CRKP26 belonged to ST11 and was identified as capsular type KL64, whereas CRKP11 belonged to ST17 and capsular type KL25.

To assess plasmid transferability, conjugation and transformation experiments were conducted as previously described (Huang et al., 2020). NDM-carrying plasmids from 26/27 isolates (96.3%) were successfully transferred to recipient strains, indicating high horizontal transfer potential. Plasmid replicon typing revealed that IncX3 was the predominant type in 25/27 isolates (92.6%), with one isolate (CRKP11) harboring an IncX3/IncFII hybrid replicon (Table 1). The remaining isolate failed to yield a transferable plasmid in conjugation assays and therefore could not be assigned a specific replicon type.

Analysis of the bla_NDM genetic context revealed a conserved core structure (bla_NDM-bla_MBL-trpF) with variable flanking regions. Ten distinct genetic contexts were identified: seven variants in bla_NDM-1-carrying isolates and three in bla_NDM-5-carrying isolates. With respect to CRKP26, one of the two dual-carbapenemase producers, bla_KPC-2 was located on an IncFII/IncR plasmid within a NTE_KPC-I chimeric transposon structure (Huang et al., 2020), and bla_NDM-1 was located on an IncX3 plasmid with the typical ISAba125-bla_NDM-1-bla_MBL-trpF-dsbC genetic context.

We performed whole-genome sequencing on CRKP26. The genome comprised a 5,529,569-bp chromosome with 57.31% GC content and 5,292 predicted ORFs, along with six plasmids ranging from 5.6 to 198.7 kb in length. Chromosomal analysis revealed virulence genes, including genes encoding siderophores (iroN, iutA, iron-enterobactin, and yersiniabactin), and a type VI secretion system (T6SS). The chromosome also harbored the β-lactamase gene bla_SHV-11 and a glycine-aspartate (GD) mutation at positions 137–138 in the OmpK36 porin gene, a mutation previously associated with reduced carbapenem permeability (Wong et al., 2019). The whole-genome sequencing data for strain CRKP26 have been deposited at the National Microbiology Data Center (NMDC) under accession number NMDC60216001.

Whole-genome sequencing identified six plasmids with distinct sizes, incompatibility groups, and genetic features, which are summarized in Table 2. pCRKP26-KPC was a IncFII/IncR hybrid plasmid. Bioinformatics analysis revealed that this plasmid harbored bla_KPC-2, bla_SHV-66, rmtB, bla_TEM-1b, and bla_CTX-M-65, conferring resistance to carbapenems, extended-spectrum cephalosporins, and aminoglycosides (Figure 1A). pCRKP26-NDM was a IncX3 plasmid carrying bla_NDM-1 and ble_MBL, which confer resistance to β-lactams, including carbapenems (Figure 1B). pCRKP26-vir is a IncHI1B plasmid encoding the hypervirulence determinants iucABCD and iutA. pCRKP26-3 carried the quinolone resistance gene qnrS1 and the β-lactamase gene bla_TEM-215. However, no major resistance or virulence genes were detected in pCRKP26-5 or pCRKP26-6, which contained only replication and mobilization genes.

Comparative sequence analysis revealed that pCRKP26-KPC shared 100% nucleotide identity (100% coverage) with pNC75-5 (GenBank: CP163294.1) from a K. pneumoniae strain isolated in Jiangxi Province (Figure 1A), suggesting interprovincial dissemination. In contrast, pCRKP26-NDM was identical (100% identity, 100% coverage) to pEC150 (GenBank: NZ_MH328008.1), an NDM-1 plasmid previously identified in Enterobacter cloacae from the same intensive care unit in 2015 (Figure 1B), indicating interspecies horizontal transfer within the healthcare facility.

For clarity, the local genetic environments of bla_KPC-2 and bla_NDM-1 are further illustrated as a linear comparison (Figure 1C) (Luo et al., 2024).

The CRKP26 virulence plasmid was compared with three reference virulence plasmids: the classical pNTUH-K2044 (KpVP-1), the prototypical pLVPK virulence plasmid (KpVP-2), and the KpVP-3-type virulence plasmid (CP 132963) (Figure 2).

The CRKP26 virulence plasmid (pCRKP26-vir, 198,670 bp), belonging to the IncHI1B incompatibility group, retained the aerobactin biosynthesis and transport system (iucABCD-iutA-def) and capsule regulatory genes (rmpA, rmpC, and rmpD). The retained rmpA gene shared 84.89% nucleotide identity with the reference gene in pNTUH-K2044 (97% coverage).

pCRKP26-vir retained the core aerobactin system (iucABCD-iutA) and capsule regulators (rmpA) but lacked the accessory salmochelin (iroB/C/D/N) and ferric citrate uptake systems (fecI/R/A). This deletion pattern suggests that primary iron acquisition mechanisms are preserved while secondary iron uptake pathways are lost.

The in vivo virulence of the strains was assessed by determining the median lethal dose (LD₅₀) in a murine model. The reference hypervirulent strain NTUH-K2044 exhibited high virulence with an LD₅₀ of 2.53 × 10² CFU. Consistent with the definition of hypervirulence (LD₅₀ ≤ 1 × 10⁶ CFU) outlined by the Chinese expert consensus, the test strain CRKP26 also demonstrated a hypervirulent phenotype, with a calculated LD₅₀ of 5.55 × 10⁵ CFU. Although the LD₅₀ value of CRKP26 was approximately 3-log₁₀ higher than that of NTUH-K2044, indicating a moderately attenuated lethality, it unequivocally qualifies as a hypervirulent strain (Figure 3A).

The virulence of CRKP26 was evaluated using mice intraperitoneally infected with bacteria at three concentrations over a 60-h observation period (Supplementary Figures S1–S3). At 1 × 10⁶ CFU, NTUH-K2044 caused rapid mortality, with a survival rate of only 20% by 48 h; however, CRKP26 showed attenuated virulence, resulting in a 90% survival rate, and the survival rate after ATCC 13883 infection was 100% through 60 h (Supplementary Figure S1). At 2 × 10⁶ CFU, NTUH-K2044 resulted in 100% mortality by 48 h, whereas CRKP26 demonstrated intermediate virulence, with 40% survival at 60 h, which was significantly greater than that of ATCC 13883, which resulted in a 100% survival rate (p < 0.01; Supplementary Figure S2). At the highest concentration (3 × 10⁶ CFU), both NTUH-K2044 and CRKP26 caused 100% mortality within 24 h, demonstrating comparably high virulence. In contrast, 90% survival was observed through 60 h after ATCC 13883 infection, with CRKP26 infection resulting in a significantly lower survival rate (p < 0.0001; Supplementary Figure S3).

In a neutrophil killing assay, the survival rate of ATCC 13883 was 19.1%, while the survival rates of NTUH-K2044 and CRKP26 were significantly greater (91.9 and 90.7%, respectively). Compared with ATCC 13883, CRKP26 was significantly more resistant to neutrophil-mediated death (p < 0.001; Figure 3B). In the serum bactericidal assay, the survival rate of ATCC 13883 was less than 1%, whereas the survival rates of NTUH-K2044 and CRKP26 were high (91.7 and 93.2%, respectively). Serum resistance was significantly greater for CRKP26 than for ATCC 13883 (p < 0.0001; Figure 3C). Both in vivo and in vitro virulence assessments demonstrated that the virulence of CRKP26 was significantly greater than that of ATCC 13883 but slightly lower than that of the hypervirulent strain NTUH-K2044.