Introduction

Pseudomonas aeruginosa is a common pathogen of nosocomial infection, with a large number of intrinsic antibiotic resistance genes and exogenously acquired resistance genes in the genome, increasing the difficulty of clinical treatment1,2. Recently, the WHO updated the list of bacterial priority pathogens, moving carbapenem-resistant P. aeruginosa (CRPA) from the “critical” to the “high” category, but innovative antibiotics targeting CRPA are still needed3. There are several main mechanisms of carbapenem resistance in P. aeruginosa, including the loss of the OprD protein, the overexpression of efflux pumps and/or AmpC-lactamases and the production of carbapenemase4. Among these, the most common is the acquisition of carbapenemase, and KPC-2 and VIM-2 were the most prevalent carbapenemases in P. aeruginosa worldwide at present5. Notably, up to 40% of carbapenemase-producing P. aeruginosa isolates harboring blaKPC-2 have been identified in China5.

Avibactam is a novel enzyme inhibitor with favorable efficacy in KPC-producing Enterobacteriaceae and P. aeruginosa when combined with ceftazidime6,7,8. However, with the widespread use of ceftazidime-avibactam (CZA) in clinical treatment, resistance to CZA has also emerged, and the susceptibility of CRPA to CZA has reached 38.2% globally as of 20219. CZA resistance in P. aeruginosa is associated with the production of metallo-β-lactamases (MBLs), changes in the Ω-loop of the AmpC enzyme, mutations or multiple copies of the KPC enzyme, and/or the overexpression of efflux pumps10,11. Among these, blaKPC mutations play a major role in the CZA resistance mechanism12. To date, the NCBI GenBank database lists 229 blaKPC variants, and compared with wild-type blaKPC (blaKPC-2 and blaKPC-3), blaKPC variants usually undergo amino acid substitution, insertion, or deletion12. These variants were frequently observed in Enterobacteriaceae bacteria but are less common in P. aeruginosa13,14. KPC-33, one of the most common variants, was commonly observed in Klebsiella pneumoniae and induces alterations in the hydrogen bonding structure of the omega-loop in KPC, resulting in resistance to CZA12. Compared with KPC-2, KPC-33 has a D179Y mutation in the Ω-loop, which reduces the inhibition of avibactam and enhances the efficiency of ceftazidime hydrolysis15.

The emergence of KPC-33 in P. aeruginosa was reported in our previous study16; however, the evolutionary path of blaKPC-33 in the CRPA has not been investigated. In this study, we observed the emergence of blaKPC-33 in the CRPA under drug pressure, leading to the emergence of resistance to CZA. We further elucidated the evolutionary trajectory of blaKPC mutations in P. aeruginosa. We also analyzed the genetic characteristics of the KPC-33-producing CRPA and the global phylogeography of KPC-producing P. aeruginosa.

Results

The development of clinical resistance in KPC-33-producing CRPA

An 86-year-old male was admitted to the hospital due to emphysema and fever after kyphoplasty. During the early stage of treatment, the patient was treated with meropenem, piperacillin‒tazobactam (PIP/TZP), and colistin; however, the inflammatory indices did not significantly improve. A carbapenem-resistant but CZA-susceptible P. aeruginosa strain (SRPA2863, KPC-2 positive) was isolated from the patient’s sputum on Day 46 after hospital admission. Owing to poor therapeutic effects, CZA (2.5 g q8h) and imipenem cilastatin sodium (2 g q8h) were used for anti-infective treatment. On Day 71, the patient’s condition had not improved, and a CZA-resistant P. aeruginosa strain (SRPA0656, KPC-33 positive) was isolated from his sputum. Consequently, PIP/TZP (4.5 g q8h) with colistin was used as a therapy. Considering the fluctuating inflammatory indices, PIP/TZP was discontinued, and imipenem cilastatin sodium (2 g q8h) was given. Unfortunately, the KPC-2-producing CRPA(SRPA3703) was isolated from sputum on Day 92. Despite continuous adjustments in the treatment regimen, the patient’s condition was not controlled, and he ultimately died as a result of respiratory failure and septic shock. The duration of CRPA during the course of antibiotic therapy is summarized in Fig. 1.

Fig. 1: History of P. aeruginosa isolation and course of clinical antimicrobial therapy.
figure 1

The light-yellow bar represents imipenem, the purple bar represents meropenem, the gray bar represents ceftazidime-avibactam, the green bar represents colistin, and the blue bar represents piperacillin-tazobactam. The black arrows indicate the time of P. aeruginosa isolation. IPM imipenem, MEM meropenem, PIP/TZP piperacillin-tazobactam, CST colistin, CZA ceftazidime-avibactam.

Full size image

Antimicrobial susceptibility testing (AST) revealed that the three CRPA strains were resistant to cephalosporins and carbapenems but susceptible to amikacin, colistin, and CZA (except for the blaKPC-33-positive strain SRPA0656, which was susceptible to amikacin and colistin but resistant to CZA) (Table 1). In addition, whole-genome sequencing (WGS) was performed to determine the genetic relationships among the three CRPA strains. MLST results revealed that all strains belonged to the same ST, ST463. SNP counts further showed that the three CRPA strains were highly homologous, with small SNPs (ranging from 4 to 8 SNPs) identified via whole-genome SNP analysis. In summary, we isolated three CRPA strains on Days 46, 71, and 92 after patient admission. The KPC-2-producing P. aeruginosa strains (SRPA2863 and SRPA3703) isolated after carbapenem treatment displayed CZA susceptibility but carbapenem resistance (CZA MIC = 1 μg/mL, imipenem MIC >128 μg/mL). In contrast, SRPA0656 (isolated 26 days after CZA treatment) had increased susceptibility to carbapenems and resistance to CZA (CZA MIC >128 μg/mL, imipenem MIC = 32 μg/mL).

Table 1 Antimicrobial susceptibilities of these isolates used in this study
Full size table

Characteristics of KPC-33-producing CRPA isolates

To further characterize the KPC-33-producing CRPA isolates, four additional strains were collected. Five KPC-33-producing CRPA strains were isolated from different patients, including four from sputum and one from blood (Supplementary Table 1). The results of antimicrobial susceptibility determination revealed that all KPC-33-producing CRPA strains were resistant to ceftazidime, cefepime, meropenem, imipenem, ciprofloxacin, and CZA, but susceptible to colistin and amikacin, indicating that these strains were multidrug-resistant (MDR) P. aeruginosa (Table 1).

WGS analysis further revealed that the KPC-33-producing CRPA isolates identified in this study could be categorized into two STs (ST463 and ST1076). The chromosomes of these strains were ~7 Mb in length, and the plasmids were of various sizes (Supplementary Table 2). We further confirmed that a total of eight antibiotic resistance genes (ARGs) confer resistance to five classes of antimicrobials, including those conferring resistance to β-lactams (blaKPC-33, blaPDC-374 and blaOXA486/395), aminoglycosides (aph(3’)-IIa), fosfomycin (fosA), fluoroquinolones (crp), and phenicols (catB7). The blaKPC-33 genes were localized on the plasmid in all the strains.

Characteristics of the plasmid carrying bla KPC-33 in P. aeruginosa

The WGS results revealed that blaKPC-33 was located on two different plasmids, type I and type II (Fig. 2 and Supplementary Table 2)17. The four plasmids harboring blaKPC-33 were identified as an uncategorized type (type I) (Fig. 2B). The complete sequences of pKPC33-ZYPA54 and pSYPA07-KPC-33 were obtained by using nanopore technology, and the plasmid contigs of pSRPA0656 and pSRPA5504 were highly matched with those of pSYPA07-KPC-33 and pKPC33-ZYPA54. All type I plasmids were highly similar to pZYPA01 (GenBank accession no. MZ050803.1, 100% query coverage and 100% nucleotide identity) and pR20-48 (GenBank accession no. CP138392.1, 99–100% query coverage, and 100% nucleotide identity) from clinical P. aeruginosa in China (Fig. 2B). BLASTn results revealed that the type II plasmid pSPPA1308 was highly identical to the backbone sequence of the plasmid pOZ176 (GenBank accession no. KC543497.1) with 94% query coverage and 100% nucleotide identity (Fig. 2A), the first IncP-2 plasmid to be identified18. Additionally, pSPPA1308 was similar to pLHL1-KPC-3 (GenBank accession no. CP081203.1) and the P9W plasmid unnamed1 (GenBank accession no. CP081203.1), which carry blaKPC-3 and blaKPC-2, respectively, with 100% query coverage and 100% nucleotide identity. Conjugation experiments were performed to confirm that the blaKPC-33-carrying plasmids pSRPA1308 and pSRPA5504 were successfully transferred into rifampicin-resistant PAO1, indicating that blaKPC-33 was located in a transferable plasmid. Compared with those of the PAO1Rif recipient strain, the resistance of the transconjugants to ceftazidime, cefepime, meropenem, imipenem, and CZA increased to varying degrees (Table 1). Notably, the elevated level of resistance to CZA in transconjugants was much greater than that in carbapenems.

Fig. 2: Alignment of the circular sequences of blaKPC-33-bearing plasmids in this study and other similar plasmids available from the NCBI database.
figure 2

A pSPRA1308 was used as the reference plasmid to perform the genome alignment with P9W plasmid unnamed1, pLHLl-KPC-3, and pOZ176. B pZYPA54-KPC-33 was used as the reference plasmid to perform the genome alignment of pSYPA07-KPC-33, pSRPA0656, pSRPA5504, pR20-24, and pZYPA01. The colored circles from the inside to the outside represent each plasmid, as shown in the right column. The solid regions demonstrate sequence similarity, whereas the gaps represent regions lacking sequence similarity.

Full size image

The genetic context of bla KPC-33

WGS analysis showed that the blaKPC-33 genes were located in two Tn4401-like transposons, with an identical core genetic platform, namely Tn6296 (∆ISKpn27–blaKPC-33–∆ISKpn6–korC-klcA). The Type I Tn4401-like transposons (IS26-∆ISKpn27–blaKPC-33–∆ISKpn6–korC-klcA-IS26-TpnA-IS26-IS26) were identical in the type I plasmid (Fig. 3). Four copies of IS26 (three intact and one truncated) were found in the blaKPC region from pSRPA5504, pSRPA0656, and pKPC33-ZYPA54, with one overlapping IS26 in the two IS26 units (IS26-Tn6376-IS26 unite and IS26-blaKPC-IS26 unite) (Fig. 3). Interestingly, in pSYPA07-KPC-33, the blaKPC core platform was translocated by IS26, resulting in the truncation of Tn6373 and forming a genetic context of IS26-∆ISKpn27–blaKPC-33–∆ISKpn6–korC-klcA-IS26 (Fig. 3). In the pSPPA1308 plasmid, Type II Tn4401-like transposons were composed of IS6100-∆ISKpn27–blaKPC-33–∆ISKpn6–korC-klcA-TpnR-TpnA. However, Tn6296 was truncated by the tnpR module of Tn1403 and IS6100, resulting in the formation of the ∆Tn1403-∆Tn6296-IS6100 region (Fig. 3).

Fig. 3: Genetic context alignment of blaKPC-33.
figure 3

Comparison of the genetic context of the blaKPC region from pSRPA1308, the blaKPC region from pSRPA5504, and the blaKPC region from pSRPA0656, pKPC33-ZYPA54, pSYPA07-KPC-33, and pZYPA01. The gray shading and squares indicate homologies between the corresponding genetic loci on each plasmid. Arrows indicate open reading frames, with arrowheads indicating the direction of transcription: red, antibiotic resistance-encoding genes; blue, transposon- and integron-associated genes; other genes are shown by gray arrows.

Full size image

KPC-33-mediated resistance to CZA and the growth rate of isolates carrying bla KPC-33

Cloning experiments were performed to investigate whether blaKPC-33 was responsible for CZA resistance in P. aeruginosa. blaKPC-2 and blaKPC-33 were expressed successfully in P. aeruginosa PAO1. The P. aeruginosa PAO1 strain carrying pGK-KPC-2 was resistant to ceftazidime, cefepime, meropenem, and imipenem, but was susceptible to CZA. In contrast, the P. aeruginosa PAO1 transformant carrying pGK-KPC-33 exhibited resistance to ceftazidime, cefepime, and CZA, but was susceptible to meropenem and imipenem (Table 1). Overall, these results indicate that blaKPC-33 is responsible for CZA resistance in P. aeruginosa.

To clarify the effect of blaKPC-33 on the growth rate of P. aeruginosa, we determined the growth curves of clinical KPC-PA strains and clonal strains in the absence of antibiotic pressure (Supplementary Fig. 2A–D). The relative growth rate of PAO1-pGK-KPC-33 was ~4.0% faster than that of PAO1-pGK-KPC-2 in logarithmic phase (0.9054 ± 0.005 vs. 0.8692 ± 0.027) (Supplementary Fig. 2A). The relative growth rate of SRPA0656 was 5.7% faster than that of SRPA3703 (0.6951 ± 0.012 vs. 0.6573 ± 0.010) and 5.8% faster than that of SRPA2863 (0.6951 ± 0.012 vs. 0.6570 ± 0.017) in clinical KPC-PA. The doubling time of strains carrying blaKPC-33 during the logarithmic phase was 3.4–10.7% shorter than that of strains carrying blaKPC-2 (Supplementary Fig. 2B). However, there was no significant difference between KPC-2-producing strains and KPC-33-producing strains in the stationary phase. Overall, in P. aeruginosa, strains carrying blaKPC-33 exhibit better growth during the logarithmic phase compared to strains carrying blaKPC-2.

Phylogenetic analysis of KPC-33-positive P. aeruginosa strains globally

A phylogenetic tree was constructed based on the core genomic SNPs of nine strains of clinical P. aeruginosa carrying blaKPC-33 (five from this study and four from the NCBI database) (Fig. 4A). The core genomes of the nine isolates were identified with a total of 53,450 SNPs (Supplementary Fig. 1). The phylogenetic tree revealed that the KPC-33-producing P. aeruginosa strains were distributed mainly in China (88.9%, 8/9) and Chile (11.1%, 1/9) (Fig. 4A). All P. aeruginosa strains harboring blaKPC-33 had seven or more resistance genes.

Fig. 4: Phylogenetic analysis of KPC-PA.
figure 4

A Phylogenetic relationships of KPC-33 producing P. aeruginosa in this study and NCBI database. The blue squares represent the corresponding drug-resistance genes. B Phylogenetic tree of 714 KPC-PA isolates collected from this study and NCBI database. The red-marked genomes were collected clinically in this study.

Full size image

Based on the phylogenetic tree, KPC-33-producing P. aeruginosa isolates could be categorized into three clades, and four STs, namely, ST463, ST235, ST1076, and ST654, were identified (Fig. 4A). In previous studies, SNPs <45 were defined as clonal dissemination; otherwise, they were defined as nonclonal dissemination19. Notably, there were few SNPs (4–39 SNPs) between the clinical strains SYPA07, ZYPA54, PA0107, R19-73, and SRPA0656, indicating a close phylogenetic relationship. SYPA07, ZYPA54, PA0107, and R19-73 originated from the same hospital, whereas SRPA0656 originated from a different hospital in the same province. These findings suggest that KPC-33-producing CRPA has been transmitted from hospital to hospital in certain regions.

Phylogeographic and phylogenetic analysis of KPC-producing CRPA strains globally

To investigate the phylogeographic relationships of the KPC-producing CRPA strains, basic information on a total of 714 KPC-producing CRPA strains was obtained from this study and the NCBI database (Supplementary Table 3). Most of the KPC-producing CRPA strains were isolated from Asia (64.1%, 458/714), followed by North America (17.51%, 125/714), South America (16.11%, 115/714), Europe (2.00%, 14/714), and other regions (0.28%, 2/714) (Fig. 5A). KPC-producing CRPA strains have been reported in 17 countries worldwide, with the majority of isolates detected in China (63.02%, 452/714), followed by the USA (17.29%, 124/717) and Colombia (7.53%, 54/717) (Fig. 5B). KPC-producing CRPA strains have been reported almost every year since 2007, particularly in China. Notably, KPC variants (KPC-33, KPC-31, KPC-35, KPC-90, and KPC-113) responsible for CZA resistance emerged after the Food and Drug Administration approved CZA for marketing in 2015. KPC-2-producing P. aeruginosa predominated among the KPC-producing CRPA strains (93.28%, 666/714), followed by KPC-3-producing P. aeruginosa (2.52%,18/714). Moreover, most of these KPC-2-producing CRPA strains were isolated from China (65.47%, 436/666), the USA (14.86%, 99/666) and South America (15.47%, 103/666) (Fig. 5B). MLST analysis showed that the 714 KPC-producing CRPA strains could be categorized into 52 different STs (Fig. 4B). The most common STs were ST463 (45.66%, 326/714), followed by ST235 (9.80%, 70/714).

Fig. 5: Global phylogeographic analysis of 714 KPC-PA strains from this study and the NCBI database.
figure 5

A Global distribution of KPC-PA. Different colors represent the number of KPC-PA strains. B Distribution of KPC-PA strains in different regions at different times.

Full size image

To trace the phylogenetic relationships between the CRPA strains carrying blaKPC-33 in this study and other KPC-producing CRPA strains globally, a phylogenetic tree was constructed using the assembled genomes of 714 KPC-producing CRPA strains from this study and the NCBI database (Fig. 4B and Supplementary Table 3). The phylogenetic tree showed that the KPC-producing CRPA strains could be categorized into five clades, and the seven clinical isolates in this study were distributed in Clade 1 and Clade 2. The phylogenetic tree further revealed that the STs of the KPC-producing CRPA strains isolated in China were diverse and dominated by ST463. Notably, the ST463 KPC-producing CRPA strains were detected only in China, and all the ST463 P. aeruginosa strains carried blaKPC-2 or its variants.

Discussion

In our previous study16, blaKPC-33 emerged from the hypervirulent ST463 CRPA, but there was no clear evidence to indicate the source of blaKPC-33. Here, we isolated KPC-2-producing CRPA strains (SRPA2863 and SRA3703) after carbapenem treatment, while KPC-33-producing CRPA strains (SRPA0656) were isolated after 26 days of treatment with CZA. The blaKPC-33 variant caused by the D179Y mutation of blaKPC-2 was driven mainly by the presence of CZA. Later, the strain was recovered to KPC-2 after carbapenem treatment. The patient had a history of CZA treatment, which indicates that KPC-33 was mutated from the wild-type KPC-2 in P. aeruginosa strains after CZA exposure. Available studies also suggest that CZA may be an independent risk factor for the emergence of KPC variants20,21,22 in bacteria. In this study, we further elucidated the pathway by which blaKPC-2 evolved into blaKPC-33 in ST463 CRPA strains, leading to CZA resistance.

KPC-33, as the most prevalent KPC-2 variant, was first identified in K. pneumoniae from Italy and then appeared in China and Greece22,23. KPC-33 is a single amino acid substitution at position 179 in the -ring, which increases the affinity of ceftazidime and attenuates the inhibitory effect of avibactam. This variation results in altered susceptibility of the bacteria to CZA and carbapenems: the “see-saw” effect15. AST revealed that, compared with the SRPA2863 and SRA3703 isolates, SRPA0656 exhibited >128-fold increase in the MIC of CZA and >2-fold decrease in the MIC of carbapenem. However, different from K. pneumoniae carrying blaKPC-3322, the KPC-33-producing CRPA strains in this study remained carbapenem-resistant, which may be related to the inactivation of OprD and the overexpression of efflux pumps in P. aeruginosa24,25. Furthermore, our study found that P. aeruginosa carrying blaKPC-33 exhibited better growth performance than those carrying blaKPC-2 during the logarithmic phase, which may also be an important reason for the continued emergence of KPC-33 mutants. Consequently, for KPC-producing CRPA strains under CZA stress, there is a high risk of developing resistance to both carbapenems and CZA, which may greatly limit the treatment options. Novel β-lactam‒β-lactamase inhibitor combinations, such as imipenem‒relebactam and meropenem‒vaborbactam, have been shown to be effective in vitro against KPC-producing Enterobacteriaceae, including KPC-33-producing K. pneumoniae. CZA in combination with other antibiotics has also been shown to have a favorable effect, and there have been reports that CZA being combined with aztreonam to successfully treat KPC-33-producing K. pneumoniae infections26. Nonetheless, we cannot ignore the risk of P. aeruginosa harboring blaKPC-33. Regular microbial culture and genotyping is needed to monitor dynamic changes in blaKPC in patients and adjust the treatment plan in a timely manner during the treatment of KPC-producing CRPA strains with CZA.

Consistent with Hu’s study27, we observed two blaKPC-bearing plasmid types. The type II plasmid is a common type of P. aeruginosa plasmid that is usually over 300 kb in size and has a narrow host range28. As previously reported27,29,30, the type I plasmid was closely related to the ST463 clones and always carried blaKPC-2. Although the T4SS was present in the type I plasmid, conjugation experiments demonstrated that it could not be transferred, which may be due to the absence of other mobile elements. In addition, blaKPC-33 is located in a different type of plasmid, but the core genetic platform of blaKPC-33, named Tn6296 with a structure of ∆ISKpn27–blaKPC-33–∆ISKpn6, was not changed and played an important role in the transfer of blaKPC. Compared with the pZYPA01 plasmid, the IS26-blaKPC-IS26 unit of the plasmid pKPC33-SYPA07 was inverted, and the IS26-Tn6373-IS26 unit lost IS26 and TpnA, which likely occurred due to the IS26-mediated inversion event. Moreover, IS26 can be used to form multicopy resistance genes by direct “copy and paste” or via the insertion of a “translocatable unit”31. Previous studies have indicated that IS26 mediates the formation of two copies of blaKPC-232. Briefly, the type II plasmid, as well as the Tn4401 transposons, may, contribute to the spread of blaKPC-33, leading to serious challenges in the prevention of extensive drug resistance in P. aeruginosa.

In addition, we collected information on P. aeruginosa strains harboring blaKPC-33 worldwide, all of which were isolated in China and Chile. Coincidentally, the global geographic analysis revealed high isolation rates of KPC-producing CRPA strains in both China and Chile, which is also consistent with other studies32,33. ST463 P. aeruginosa was the most prevalent type of KPC-producing CRPA strain in China, and the high carriage rate of blaKPC-2 offers additional possibilities for the emergence of KPC variants, which may explain why P. aeruginosa carrying blaKPC-33 was detected more frequently in China. In addition, there was evidence that the ST463 clone carries both exoU and exoS, which may be important reasons for the poor prognosis of patients30. The ST654 and ST235 clones, among the top ten global high-risk clones of P. aeruginosa, were associated with different acquired β-lactamases, IMP, and VIM MBLs in particular34. ST235 and ST643 KPC-producing CRPA strains are also prevalent worldwide and may have contributed to the spread of blaKPC to some extent. Therefore, we still need to strengthen surveillance to prevent the spread of high-risk clones carrying blaKPC, especially ST463 P. aeruginosa, which may become a future threat.

In summary, we report the global phylogeography and genetic characterization of carbapenem- and CZA-resistant KPC-33-producing CRPA strains. The use of CZA and the relatively high growth rate of blaKPC-33 may be important for inducing blaKPC-33 mutations in CRPA strains carrying blaKPC-2. In addition, the type II plasmid, as well as the Tn4401 transposons, may contribute to the spread of blaKPC-33 in CRPA strains, further limiting treatment options. Moreover, blaKPC-33 has emerged in high-risk clone ST463 CRPA strains. Therefore, it is necessary to monitor the resistance phenotypes of KPC-producing CRPA strains regularly and adjust antibiotic regimens in a timely manner when CZA is used to treat KPC-2-producing CRPA strains in the clinic.

Methods

Isolate data

We successively isolated KPC-33- and KPC-2-producing P. aeruginosa from a patient during treatment, including two KPC-2-producing strains (isolated on 26 August 2023 and 13 October 2023) and one KPC-33-producing strain (isolated on 21 September 2023). To study the prevalence of KPC-33-producing P. aeruginosa, four other P. aeruginosa strains carrying blaKPC-33 were collected from two tertiary hospitals in Zhejiang, China. These strains were identified using the MALDI-TOF MS system (bioMérieux, Marcy l’Etoile, France). The primers KPC-F and KPC-R were used to sequence the blaKPC genes (Supplementary Table 4). This study was conducted in accordance with the Declaration of Helsinki and had been reviewed and approved by the Research Ethics Committee of Zhejiang Provincial People’s Hospital (Approval no.: QT2024188).

Antimicrobial susceptibility testing

The minimal inhibitory concentrations (MICs) of the strains in this study were determined by the broth microdilution method. A total of eight antibiotics were tested, including cefepime, ceftazidime, imipenem, meropenem, amikacin, ciprofloxacin, colistin, and CZA. The results were interpreted according to the breakpoints recommended by the 2024 Clinical and Laboratory Standards Institute (CLSI) guidelines35, except for colistin, which was interpreted according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines (http://www.eucast.org/clinical_-breakpoints). P. aeruginosa ATCC 27853 was used as the quality control strain.

Whole-genome sequencing and analysis

Extraction of the Genomic DNA of isolates in this study was performed with QIAamp DNA MiniKit (Qiagen, Valencia, CA, USA). Oxford Nanopore (MinION system, Nanopore, Oxford, UK) and Illumina sequencing (NovaSeq system, Illumina Inc, San Diego, USA) were used respectively for genome sequencing and followed by de novo assembly with Unicycler v0.4.836. Prokka 1.11 was used to annotate the location and function of the genes37. Multilocus sequence type (MLST) was determined by using the MLST 2.1 (https://github.com/tseemann/mlst). ResFinder 4.1 and PlasmidFinder 1.3 at the Center for Genomic Epidemiology (http://www.genomicepidemiology.org/) were used to identify resistance genes and plasmid types. BLASTn v2.4.0 searches were conducted to find similar plasmids in this study38. The circular genome map and comparative genome map were completed by BRIG v0.95 and Easyfig v2.2.339,40.

Cloning experiments

Cloning experiments were conducted according to the methods described in a previous study41. Briefly, the wild-type blaKPC-2 gene and blaKPC-33 gene sequences containing the wild promoter were amplified from SRPA1308 and SRPA3703 as templates, respectively. The purified product was ligated with the pGK-1900 vector and transformed into E. coli DH5α. The recombinant plasmids pGK-KPC-33 and pGK-KPC-2 were also transferred into the PAO1 strain by electrochemical transformation experiments, respectively. Transformants were selected from Luria-Bertani (LB) agar plates supplemented with 50 μg/mL gentamicin, and further confirmed by PCR and Sanger sequencing. Primers were listed in Supplementary Table 4.

Plasmid conjugation experiments

According to previous studies42, the KPC-33-producing CRPA was used as the donor strain, and the rifampin-resistant P. aeruginosa PAO1Rif was used as the recipient strain for the conjugation assay. Bacterial suspensions of donor strain and PAO1Rif in the logarithmic phase of growth were 1:1 mixed and dropped onto filter membranes for overnight culture. Overnight cultures were screened on a selection medium (containing 16 μg/mL CZA and 800 μg/mL rifampin). The transconjugants harboring target plasmids were confirmed by PCR and antimicrobial susceptibility. Primers were listed in Supplementary Table 4.

Growth rate determination

We determined the growth rate of clinical and clonal strains in this study to investigate their fitness cost. As in the previous study43, three monoclones of clinical and clonal strains were grown overnight, and 200 μL of the overnight culture diluted at 1: 100 was added to a flat-bottom 100-well plate. Each overnight culture was repeated three times. The optical density of each culture at 600 nm was measured every 5 min for 20 h by Bioscreen C analyzer (Oy Growth Curves Ab.Ltd., Finland). The growth rate based on OD600 curves was calculated by an R script44. The growth rate results were analyzed using of one-way analysis of variance (ANOVA) as available in GraphPad Prism version 9. An adjusted P value <0.05 was considered significant.

Phylogenetic analysis

To further trace the phylogenetic relationship between blaKPC-33 in this study and blaKPC-positive P. aeruginosa isolates worldwide and to monitor the global phylogeography of KPC-PA, we evaluated the GenBank (https://www.ncbi.nlm.nih.gov/datasets/ genome, accessed December 19, 2023) for all available 27,716 P. aeruginosa genomes. And 707 assembled P. aeruginosa carrying blaKPC were screened. Phylogenetic trees were constructed using Roary and FastTree based on SNPs in the core genome and further visualized using iTOL v6 (https://itol.embl.de). ChiPlot (https://www.chiplot.online/map_plot.html) and Inkscape-1.3.2 was used for global geographic distribution mapping.