Strain identification was performed using MALDI-TOF MS (bioMérieux, France). Minimal inhibitory concentrations (MICs) were determined by broth microdilution following Clinical and Laboratory Standards Institute (CLSI) guidelines, with Escherichia coli ATCC 25922 as the quality control strain. MIC interpretation adhered to CLSI breakpoints, except for tigecycline (S ≤ 0.5 mg/L; R > 0.5 mg/L) and eravacycline (S ≤ 0.5 mg/L), which followed EUCAST and FDA breakpoints, respectively. Rapid carbapenemase detection utilized a colloidal gold immunoassay (CGI) kit (Gold Mountain River Tech Development Co., Beijing, China) and the carbapenemase inhibitor enhancement test.

Whole-genome DNA from the isolates carrying bla _KPC-2, and bla _KPC-190 was extracted using the QIAamp DNA minikit (Qiagen, France). PCR amplification of bla _KPC variants employed primers Kpc-rbs (5’-CTCCACCTTCAAACAAGGAAT-3’) and Kpc-rev (5’-ATCTGCAGAATTCGCCCTTCGCCATCGTCAGTGCTCTAC-3’), and the resulting products were cloned into plasmid pHSG398 downstream of the pLac promoter for phenotypic studies. Recombinant plasmids were electroporated into E. coli DH5α and K. pneumonia 13882, and constructs were sequenced using T7 promoter/terminator primers. Gene sequences were analyzed via NCBI tools.

Genomic DNA was sequenced on the Illumina MiSeq platform (paired-end, 2×300 bp) and supplemented with Oxford Nanopore long-read sequencing. Hybrid de novo assembly was conducted using Unicycler v0.4.8. Resistance genes, plasmid replicons, and sequence types were annotated using RAST v2.0, MLST, PlasmidFinder, and ResFinder. Single nucleotide polymorphism (SNP) analysis was performed with Snippy v4.4.3 using K. pneumoniae 5055 as a reference. The genetic environment of bla _KPC was examined with VRprofile, and plasmid drafts were generated via Proksee.

To evaluate the transferability of ceftazidime-avibactam resistance, a conjugation experiment was performed using Klebsiella pneumoniae LX02 as the donor strain and Escherichia coli J53 and EC600 as the recipient. Transconjugants were selected on Luria–Bertani (LB) agar plates containing ceftazidime-avibactam (1 μg/mL) and sodium azide or Rifampin (150 μg/mL). In addition, a transformation experiment was carried out by extracting plasmid DNA from K. pneumoniae LX02 using the phenol–chloroform method, followed by electroporation into E. coli DH5α as the recipient strain. Transformants were grown on LB agar plates supplemented with ampicillin (50 μg/mL), screened for the presence of the bla _KPC gene via PCR, and further analyzed through sequencing and comparison using the BLAST program (http://blast.ncbi.nlm.nih.gov/Blast.cgi) (Poirel et al., 2011).

The bla _KPC coding region (amino acids 25-293) was inserted into the pET-28a expression plasmid to generate an N-terminally His-tagged recombinant protein. Following transformation into E. coli BL21(DE3) cells, the expressed protein was isolated using immobilized metal affinity chromatography followed by size-exclusion purification (Pemberton et al., 2017). Protein quantification was achieved through UV absorbance at 280 nm, applying a molar extinction coefficient of 39,545 M^-1cm^-1, with final concentrations adjusted to 10-20 mg/mL.

Enzyme activity measurements were conducted spectrophotometrically under ambient conditions using purified protein preparations. Reactions were carried out in phosphate-buffered saline (pH7.4), with kinetic constants derived by nonlinear regression analysis of the Michaelis-Menten model (Winkler et al., 2015). Nitrocefin hydrolysis was assessed by tracking absorbance changes at 482 nm across a concentration gradient. Ceftazidime turnover rates were quantified by monitoring absorbance at 257 nm, while meropenem and imipenem degradation were both followed at 299 nm, both using identical experimental setups with appropriate substrate concentration ranges.

Inhibitor potency (IC₅₀) against KPC-2 and KPC-190 variants was evaluated using nitrocefin hydrolysis as reporter activity. Protein samples were pre-incubated with serial dilutions (0-30 μM) of avibactam for 10 minutes prior to substrate addition (100 μM final nitrocefin concentration). After 30 minutes incubation, 482 nm absorbance measurements were collected and dose-response curves generated using Prism analysis software.

K. pneumoniae LX01 and LX02 were sequentially isolated from a 65-year-old male patient hospitalized at the Second Affiliated Hospital of Xi’an Jiaotong University for hospital-acquired pneumonia (HAP) complicating acute liver failure. The patient had significant comorbidities including hypertension, and type 2 diabetes.

On admission, he was empirically treated with meropenem (1 g q8h for 5 d) for suspected Gram-negative infection. Despite therapy, persistent fever (38.5-39.2°C), leukocytosis (WBC 15.3×10⁹/L), and elevated procalcitonin (2.8 ng/mL) were observed, with K. pneumoniae LX01 (carbapenem-resistant) isolated from sputum cultures (≥10⁵ CFU/mL). Treatment was escalated to ceftazidime-avibactam (2.5 g q8h).

After 10 days, while fever subsided, inflammatory markers remained elevated (CRP 48 mg/L, PCT 1.2 ng/mL), and K. pneumoniae LX02 (ceftazidime-avibactam-resistant) was isolated. Rectal screening confirmed did not carry KPC-producing K. pneumoniae in the intestine before treatment. Final clearance was achieved with tigecycline (100 mg q12h for 4 d), with normalization of clinical (afebrile) and laboratory (WBC 8.7×10⁹/L, CRP <10 mg/L) parameters before discharge. A comprehensive timeline of the patient’s medication history is shown in Figure 1 .

Antimicrobial susceptibility testing revealed distinct resistance profiles for the K. pneumoniae clinical isolates carrying KPC-2 and its variants ( Table 1 ). The initial strain, LX01 (KPC-2), displayed high resistance to carbapenems (imipenem MIC = 32 μg/mL, meropenem MIC = 16 μg/mL). After 10 days of ceftazidime-avibactam therapy, the second isolate, LX02 (KPC-190), exhibited resistance to ceftazidime-avibactam (MIC > 64 μg/mL) and carbapenems (imipenem MIC = 2 μg/mL, meropenem MIC = 4 μg/mL). These findings suggest that the evolution from KPC-2 to KPC-190 resulted in partial restoration of carbapenem susceptibility, while maintaining high level resistance to ceftazidime-avibactam. The two isolates were resistant to all the β-lactams tested, quinolones, and aminoglycosides but they were susceptible to colistin and tigecycline. Concerning the BLICs, the isolate was resistant to CZA and susceptible to meropenem-vaborbactam, imipenem-relebactam and aztreonam-avibactam. Resistance to FDC (MIC = 8 mg/L) was also detected. In addition, there was a significant upward trend in the bla _KPC gene expression during treatment in K. pneumoniae LX02. This strain showed a 4-fold increase in gene expression when compared to the initially isolated K. pneumoniae LX01.

Comparative analysis revealed that the two isolates differed by 39 SNPs, which showed a high degree of homology ( Figure 1 ). Whole-genome sequencing analysis confirmed that the two strains belonged to clone ST11-KL64 with the iucABCD-iutA virulence cluster and rmpA/rmpA2 genes, which could be identified as hypervirulent K. pneumoniae ( Li et al., 2018) in association with the patient’s poor prognosis. In addition, several resistance genes were found, including the β-lactamase genes bla _CTX-M-65, bla _SHV-12, and bla _TEM-1B, the aminoglycoside resistance genes rmtB and aadA2, the fluoroquinolone resistance gene qnrS1, the fosfomycin resistance gene fosA3, the trimethoprim-sulfamethoxazole resistance genes dfrA14 and sul2. The bla _KPC-2 gene was detected in K. pneumoniae LX01. Notably, whole-genome comparative analysis also revealed one novel bla _KPC-2 variants in K. pneumoniae LX02, designated bla _KPC-190. Nucleotide alignment of bla _KPC variants revealed variations in both the Ω-loop and the 237-243 loop (D179Y and A243V) ( Figure 2 ).

The complete sequence was also obtained for plasmid characterization. bla _KPC-2-positive K. pneumoniae LX01 was used as a representative strain. This strain had five plasmids pVir-LX01, pKPC-LX01, p3-LX01, p4-LX01, and p5-LX01 with sizes of approximately 193, 136, 87, 12, 0.5kb. Virulence factors including iroN, rmpA, iucA, iucB, iucC, iucD, iutA, rmpA2 were identified on the virulence plasmid pVir-LX01. The bla _KPC was located on the IncFII-type plasmid, pKPC-LX01, which was reported to be one of the key vectors mediating transmission of bla _KPC (Yang et al., 2021). The multi-replicon enhanced the ability to replicate in a wider range of hosts (Li et al., 2019). bla _CTX-M-65, bla _TEM-1B, rmtB were also identified in this plasmid. In addition, several resistance genes including qnrS1, dfrA14 were carried by the plasmid p3-LX01. BLAST analysis revealed that the plasmid pKPC-LX01 had high coverage with the plasmid pNC75-5 in the NCBI database, which carries bla _KPC-2 isolated from clinical K. pneumoniae ( Figure 3 ).

Plasmid transformation assays yielded significant results ( Table 1 ). Transformants KPC-190-E. coli DH5a increased the MICs of ceftazidime-avibactam and ceftazidime by at least 256- and 256-fold, respectively compared with the recipient E. coli DH5α. KPC-190 mediated a 4-fold increase in MICs for imipenem, for meropenem, KPC-190 mediated an at least 4-fold increase in the MIC of imipenem. Notably, no bla _KPC-containing conjugants were detectable when either E. coli EC600 or J53 were used as recipients. The bla _KPC-containing plasmids in this study do not appear to be self-conjugative, potentially due to the absence of relaxase and T4CP.

The recombinant plasmids carrying bla _KPC were successfully introduced into the recipient strain K. pneumoniae ATCC 13882. bla _KPC-2 conferred resistance to most β-lactams including ceftazidime, cefepime, imipenem, and meropenem, but retained susceptibility to ceftazidime-avibactam ( Table 1 ). In contrast to the bla _KPC-2-positive clonal strains, the MICs of the bla _KPC-190 strains decreased at least 32-fold towards carbapenems and increased at least 64-fold to ceftazidime-avibactam. In general, the bla _KPC-2 variants conferred resistance to ceftazidime-avibactam and susceptibility to carbapenems compared to bla _KPC-2. Importantly, the bla _KPC-2 variants appeared to confer a higher ceftazidime-avibactam MIC in K. pneumoniae than in E. coli. This suggests that variations in strain backgrounds may influence the expression of bla _KPC variants. In addition, emerging carbapenemase inhibitor combinations, namely imipenem-relebactam, aztreonam-avibactam, and meropenem-vaborbactam, showed superior inhibitory performance against both bla _KPC-2 and variants.

The enzyme kinetics and IC₅₀ values of KPC-2 and KPC variants were determined. Enzymatic assays demonstrated that the hydrolytic activity (Kcat/Km) of KPC-190 against carbapenems was reduced compared to KPC-2. Hydrolytic activity against ceftazidime was significantly decreased for KPC-190 compared to KPC-2, indicating that the KPC-190 enzymes exhibited reduced hydrolysis and higher affinity for ceftazidime compared with wild-type KPC-2 ( Table 2 ). Notably, the KPC variants had almost no detectable hydrolysis activity of carbapenems (ertapenem, imipenem, and meropenem), indicating reduced activity to carbapenems. To evaluate the inhibitory activity of the inhibitor against the KPC-2 and KPC-190 enzymes, the 50% inhibitory concentration (IC₅₀) values for avibactam were also measured. The IC₅₀ values for avibactam were markedly higher for KPC-190 (0.13 μM) compared to KPC-2 (0.014 μM), indicating a significant reduction in avibactam binding affinity. This low affinity for avibactam, coupled with the retained hydrolytic activity against carbapenems, accounts for the resistance phenotype of KPC-190.

Whole-genome sequencing revealed that bla _KPC-190 is located on an IncFII plasmid within the translocatable unit: IS26-ISKpn8-bla _KPC-ΔISKpn6-ΔtnpR-IS26, which contrasts with previous studies where three different variants of Tn1721 harbouring bla _KPC-2 (referred to as A1-, A2-, and B-type) predominated in China ( Figure 4 ). Here, IS26 replaced Tn1721 to form a transposable structure with double IS26 flanking the ISKpn6-bla _KPC-ISKpn27 core structure, a configuration rarely observed in bla _KPC variants.

The ST11-K64 Klebsiella pneumoniae LX02 strain harbored a variety of virulence factors ( Table 3 ). Notably, it contained a large virulence plasmid approximately 200 kb in size, which carried genes conferring resistance to heavy metals such as copper and silver, alongside virulence-related genes responsible for capsular polysaccharide synthesis regulation (rmpA/rmpA2) and iron acquisition systems (iucABCD-iutA). This plasmid shared significant similarity with the widely disseminated pLVPK-like plasmids (reference sequence GenBank accession number CP094941) but lacked a ∼25-kb pathogenicity island that includes rmpA, fecAIR, and iroBCDN.

In addition to the plasmid, the bacterial chromosome encoded multiple virulence-associated gene clusters. These included the fim and mrk clusters, which are responsible for the production of type 1 and type 3 fimbriae, respectively. These fimbriae, in combination with capsule production, represented the primary virulence mechanisms of K. pneumoniae LX02. Interestingly, copy number variations (CNVs) were observed in some chromosomal genes, with a slight reduction in mrkA and an increase in pgaA.