Fitness cost of ceftazidime–avibactam resistance acquisition in carbapenem-resistant hypervirulent Klebsiella pneumoniae

Abstract

Background

Antibiotic-resistant Klebsiella pneumoniae (KP) poses a serious global public health threat. However, research on the resistance mechanisms and accompanying phenotypic changes in carbapenem-resistant hypervirulent KP (CR-hvKP) under antibiotic treatment remains limited. This study aims to investigate the resistance mechanisms of CR-hvKP to ceftazidime-avibactam (CAZ-AVI) and its concomitant phenotypic shifts, employing an in vitro induction assay.

Methods

Six ceftazidime–avibactam (CAZ–AVI)-susceptible CR-hvKP clinical isolates were subjected to in-vitro resistance induction. We used pulsed-field gel electrophoresis, whole-genome sequencing, biofilm formation, a Galleria mellonella infection model, and in-vitro competitive growth assays to characterize the virulence and adaptive changes of the isolates.

Results

CAZ–AVI-resistant CR-hvKP demonstrated enhanced biofilm formation capacity, but the G. mellonella infection model indicated a decrease in virulence of the drug-resistant strain. While resistant strains exhibited diminished competitive fitness in vitro, growth curves did not differ significantly. Genomic characterization identified both resistant and susceptible isolates as ST11, with resistant isolates exhibiting an expanded resistance gene profile, primarily involving KPC-2 variants. All strains carried typical virulence determinants, including iroE (a glycosidase gene within the salmochelin siderophore system), iucABCD (the aerobactin biosynthesis operon), and iutA (the gene encoding the outer membrane receptor for ferric-aerobactin).

Conclusions

CAZ–AVI resistance acquisition in CR-hvKP primarily occurs through KPC-2 mutations. Strains harboring such mutations exhibit enhanced biofilm formation capacity but attenuated virulence and competitiveness. Research into these adaptive changes will facilitate the development of improved clinical strategies for the treatment and control of carbapenem-resistant hypervirulent Klebsiella pneumoniae.

Introduction

Klebsiella pneumoniae (KP) is a major pathogen in both hospital- and community-acquired infections [1], with immunocompromised populations, including patients with diabetes [2] and children [3], being particularly susceptible to it. These infections can affect the respiratory, bloodstream, urinary, gastrointestinal, hepatobiliary, and central nervous systems [4], with infection sites varying according to strain virulence and migration capabilities [5]. With the widespread use of antibiotics in China, carbapenem-resistant (CR) hypervirulent (hv) KP (CR-hvKP) is increasingly reported [6]. KP employs multiple virulence factors for survival and pathogenesis. The hvKP virulence determinants identified to date include capsular polysaccharides, siderophores, virulence genes, virulence plasmids, lipopolysaccharides, and fimbriae, which function independently or synergistically to confer hypervirulent characteristics [7]. Among these, capsular polysaccharides and siderophores are considered crucial pathogenic determinants. The hvKP strains typically harbor virulence plasmids encoding multiple virulence genes, which continue to evolve during bacterial transmission. These virulence plasmids commonly carry genes such as prmpA, iucABCD, iutA, iroBCDN, and peg-344 [8], which are considered optimal laboratory markers for hvKP identification. Because of its high pathogenicity and transmissibility, CR-hvKP infections have become prevalent across multiple regions in China [9], Japan, South Korea [10], Europe [11], and Africa [12]. This pathogen has garnered increasing global attention [13] and may emerge as the next “superbug,” presenting a significant public health challenge.

HvKP and CRKP can evolve into CR-hvKP through the acquisition of plasmids carrying carbapenem resistance genes and virulence-encoding genes, respectively [6]. CR-hvKP exhibits resistance to multiple common antibiotics. According to the 2023 National Bacterial Resistance Surveillance Report from the China Antimicrobial Resistance Surveillance System, the national average carbapenem resistance rate of KP increased from 10.0% in 2022 to 10.8% in 2023. The resistance rate to third-generation cephalosporins declined from 36.9% in 2014 to 27.7% in 2022 and 2023 [14]. Our understanding of the mechanisms and dynamics of acquired resistance, particularly for novel antibiotics, remains incomplete, significantly limiting therapeutic options for CR-hvKP infections.

The novel combination ceftazidime–avibactam (CAZ–AVI) has been suggested as an effective alternative for treating infections caused by KP carbapenemase (KPC)-producing CRKP [15, 16]. Nevertheless, with the widespread use of CAZ–AVI, cases of resistance are increasingly reported [17, 18]. The resistance mechanisms and associated fitness costs remain insufficiently studied, complicating CR-hvKP treatment. As CR-hvKP poses a significant global clinical threat, understanding its pathogenicity- and resistance-related factors is crucial for effective treatment.

In this study, we aimed to investigate the resistance mechanisms of CR-hvKP to CAZ–AVI. To this end, we selected six CAZ–AVI-susceptible CR-hvKP isolates and generated resistant strains through in-vitro resistance induction. We subsequently conducted biofilm formation and in-vitro competitive growth assays and established a Galleria mellonella infection model. Growth patterns were analyzed and compared between susceptible and resistant isolates through growth curve analysis. We assessed phenotypic differences between resistant and susceptible strains and employed whole-genome sequencing (WGS) to investigate resistance mechanisms and virulence gene mutations in pre- and post-induction isolates. This study aims to clarify the resistance mechanism of CR-hvKP to CAZ-AVI and evaluate the changes in growth ability and virulence of the strains after resistance, with the hope of providing potential intervention targets for the clinical prevention, control and treatment of carbapenem-resistant and highly virulent Klebsiella pneumoniae.

Materials and methods

Sample collection

Six CR-hvKP isolates (F7692, F7624, F7644, F7678, F7837, and F7951) were collected from sputum or bronchoalveolar lavage fluid specimens of patients hospitalized in the intensive care unit or critical care department of a tertiary hospital in Jiangxi Province, China. Antimicrobial susceptibility testing using MIC Test Strip (Liofilchem® MTS™) demonstrated that the isolates were susceptible to CAZ–AVI.

The isolates were subjected to in-vitro drug resistance induction experiments, and the minimum inhibitory concentrations (MICs) were determined using MIC Test Strips (Liofilchem® MTS™). After 18 days of drug resistance induction, the six strains showed different degrees of CAZ–AVI resistance. The information of the isolates were shown in Table 1 (e.g., F7692 represents the susceptible strain before induction, and L7692 represents the corresponding resistant strain, and so forth).

Table 1.

Strains information

Antimicrobial susceptibility testing

According to the Clinical and Laboratory Standards Institute (CLSI, 2018) guidelines for antimicrobial susceptibility testing, the minimum inhibitory concentration (MIC) of antimicrobial agents was determined. The MIC is defined as the lowest concentration of an antimicrobial agent that completely inhibits bacterial growth [19]. The Epsilometer Test method was employed for detection: single colonies cultured for 16–18 h were selected and diluted to a 0.5 McFarland standard. The surface of Mueller–Hinton (MH) agar plates was uniformly swabbed with a cotton swab, followed by the application of MIC Test Strip. The plates were then incubated at 37 °C for 18–24 h, after which the susceptibility results were recorded.

In vitro-induced drug resistance assay

In vitro-induced drug resistance study was based on a previous report [20]. Six susceptible strains were inoculated onto Luria–Bertani (LB) agar plates and incubated at 37 °C for 16–18 h. LB broth containing sub-inhibitory concentrations of antibiotics (starting from 0.25 MIC) was prepared, followed by incubation at 37 °C and subculturing every 12–20 h. The concentration of the antibiotic in the medium was gradually doubled until the MIC of the test strain increased to within the resistant range. Using Klebsiella pneumoniae ATCC 700603 as the control strain. The six drug-resistant strains and the control strain were subjected to ten consecutive passages in antimicrobial-free LB broth, with MICs determined to assess the stability of their resistance.

Phylogenetic analysis

Pulsed-field gel electrophoresis (PFGE) was used to analyze the genetic relatedness between pre- and post-resistance induction isolates. Isolates showing ≥ 80% similarity were defined as belonging to the same PFGE type, with Salmonella enterica serovar Braenderup H9812 serving as the reference strain.

PFGE was carried out following the standard operating procedure by PulseNet International. In brief, genomic DNA from both CAZ–AVI-susceptible and -resistant isolates was embedded in agarose plugs and digested with XbaI. The digested plugs were segregated using the CHEF Mapper XA System (Bio-Rad, United States) at 6 V/cm for 18 h. Subsequently, the gel was stained with GelRed solution (Sangon Biotech, China) for 30 min, destained in ultrapure water. PFGE images were imported into the BioNumerics software v8.0 (Applied Maths, Sint Martens-Latem, Belgium) and obtained a tree diagram showing the genetic correlation among the isolates.

Biofilm formation assay

Biofilm formation ability was assessed using the violet staining method [21]. The isolates were cultured in LB agar plates at 37 °C for 16–18 h. Overnight cultures of the CR-hvKP strains were adjusted to 10⁸ colony-forming unites (CFU)/mL with phosphate-buffered saline (PBS) and then diluted 1:100 in LB broth. Two hundred microliters of the diluted bacterial suspension was transferred into a 96-well plate, using two hundred microliters of LB broth for the negative control. All experiments were performed in triplicate. Then incubated statically at 37 °C for 48 h. Following incubation, the bacterial suspension was removed, and the wells were washed with PBS and air-dried. The adherent biofilm was stained with 1% Crystal Violet Staining Solution (Solarbio, China) for 15 min. After removing the excess stain and washing with distilled water until colorless, the crystal violet was solubilized with absolute ethanol. The absorbance (OD) of each well was read at 570 nm using an enzyme marker. The criteria for defining different biofilm-forming capabilities are as follows: Using the average value of the negative control plus three times its standard deviation to define the critical value (Ac), the strains are categorized into the following four groups: (1) Strong biofilm formers (A > 4 × Ac); (2) Moderate biofilm formers (2 × Ac < A ≤ 4 × Ac); (3) Weak biofilm formers (Ac < A ≤ 2 × Ac); (4) Non-biofilm formers (A ≤ Ac).

G. mellonella larval infection model

A G. mellonella larval infection model was used to assess bacterial pathogenicity [22]. G. mellonella larvae were purchased from Keyun Biotech Company, China. For the experiments, we selected healthy larvae weighing 250–300 mg that exhibited active mobility and a strong turning ability. Overnight cultures of the KP isolates were harvested and cells were resuspended in PBS at a final concentration of 10⁸ CFU/mL. Ten microliters of bacterial suspension was injected into the hemocoel through the last left proleg of each larva. PBS-injected larvae served as controls. Each group comprised 10 larvae, and all experiments were performed in triplicate. The infected larvae were maintained at 37 °C, and mortality was monitored over 72 h with observations recorded every 24 h.

In-vitro competitive growth assay

In vitro competition experiments were conducted to evaluate the competitive fitness of the strains [23].Resistant and susceptible isolates were cultured in LB broth at 37 °C under shaking for 18 h. Bacterial suspensions were adjusted to 10⁸ CFU/mL, and resistant strains were mixed with their susceptible counterparts at a 1:1 ratio. The mixed bacterial suspensions were serially diluted and plated onto plain LB agar plates and CAZ–AVI-supplemented plates (The concentration of CAZ-AVI depends on the MICs of the susceptible bacterial isolates). After incubation at 37 °C for 16–18 h, colonies were counted, and the number of susceptible bacterial colonies was calculated by subtracting the colony count on CAZ-AVI-supplemented plates from the colony count on antibiotic-free LB plates. All experiments were performed in triplicate. Then competition curves were generated and analyzed using GraphPad Prism 9.5.

Assessment of growth ability

A growth assay was performed as previously described [24]. Bacterial suspensions were standardized to OD600 = 1 and diluted 1:1,000 in LB broth. Two hundred microliters of each dilution was dispensed into 96-well plates. All experiments were performed in triplicate. Growth was monitored over 24 h by measuring the absorbance at 540 nm every 30 min, using a microplate reader.

WGS

The six CAZ–AVI-resistant and six CAZ–AVI-susceptible CR-hvKP isolates were cultured overnight on standard agar plates. Single colonies were inoculated into LB broth and cultured under shaking. Bacterial precipitates were sent to the Bioengineering Institute (698 Xiangmin Road, Songjiang District, Shanghai China) for WGS using Pacbio. Raw sequencing data were processed and quality-assessed using Fastp, including quality trimming to obtain accurate clean reads. The Comprehensive Antibiotic Resistance Database (CARD), an enhanced version of the classical Antibiotic Resistance Genes Database, provides a foundation for drug resistance studies. Basic local alignment search tool (BLAST) analysis was performed to align the translated gene sequences against the CARD database, linking the genes with their specific resistance functions. The Virulence Factors Database (VFDB), designed for studying pathogenic factors in bacteria, comprises two main sections: SetA and SetB. SetB contains both verified virulence factor genes and predicted potential virulence factors, and SetA represents the core dataset of VFDB. Translated gene sequences were aligned against the VFDB using BLAST to link genes with their virulence factor annotations. Sequence alignments were conducted using SnapGene, and plasmid linear comparisons were performed using Easyfig. Whole-genome single-nucleotide polymorphism (SNP) alignment was performed using Snippy, and a high-resolution phylogenetic tree of the 12 CR-hvKP strains was constructed using neighbor-joining clustering by Roray and Fastree. What’s more, we performed multilocus sequence typing (MLST) on the 12 isolates.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 9.5. Continuous variables were compared using Student’s t tests. P-values < 0.05 were considered statistically significant.

Results

CR-hvKP morphologically changes after CAZ–AVI resistance acquisition

On LB plates, the clinical isolates exhibited a mucoid phenotype, characterized by large, opaque colonies with wet surfaces. After repeated subculturing in CAZ–AVI-supplemented LB broth, the isolates developed resistance and morphologically shifted to a non-mucoid phenotype, forming smaller, translucent colonies with dry surfaces (Fig. 1).

Fig. 1.

Morphological changes in the CR-hvKP isolates after in-vitro induction of resistance to CAZ–AVI. The induced resistant colonies were small and dry, compared to the large, opaque, and moist colonies of the susceptible strain

The genetics of CR-hvKP remain consistent after CAZ–AVI resistance acquisition

Antimicrobial susceptibility testing revealed diverse degrees of CAZ–AVI resistance among the isolates (Fig. 2A). To analyze the genetic relatedness between the pre- and post-resistance induction CR-hvKP strains, XbaI-digested PFGE analysis was performed, which revealed identical genetic backgrounds between the susceptible isolates and their resistant counterparts (Fig. 2B). Next, we aligned the Whole-genome sequences of the 12 CR-hvKP strains to construct a SNP developmental tree. The results showed that the strains had the same genetic background before and after the induction of CAZ–AVI resistance (Fig. 3). According to WGS analysis, all CR-hvKP isolates were identified as the most prevalent sequence type, ST11.

Fig. 2.

A MICs of ceftazidime-avibactam against CR-hvKP before and after in vitro resistance induction. All six susceptible strains developed resistance, with isolates F7692 and F7837 showing a MIC increase from 2 mg/L to 256 mg/L. B Genetic relatedness assessed by XbaI PFGE. The highly similar PFGE profiles between susceptible and resistant strains indicate that resistance evolved from the original isolates

Fig. 3.

Phylogenetic analysis of 12 CR-hvKP strains based on WGS data. Using the Bootstrap method, branch support (BP) was defined as: low (BP < 70), moderate (70 ≤ BP < 90), or high (90 ≤ BP < 100). The presence of high-support clusters (BP ≥ 90) among some strains indicates a close genetic relationship

CAZ–AVI resistance acquisition promotes biofilm formation

Biofilms are complex three-dimensional structures composed of bacteria and their secreted extracellular polymeric substances. The ability to form biofilms leads to increased resistance to host defense factors and antimicrobial agents and is considered a significant virulence characteristic. Different resistant strains exhibit variable biofilm formation capacity [25]. We evaluated the biofilm formation capability of the 12 CR-hvKP strains before and after in-vitro CAZ–AVI resistance induction. The calculated Ac value was 0.0411. Resistant strains, with a mean absorbance of 0.2159 (> 4 × Ac), were strong positive, whereas sensitive strains, with a mean absorbance of 0.1122 (2 × Ac < A ≤ 4 × Ac), were moderate positive. This demonstrates a stronger biofilm-forming capacity in the resistant strains (Fig. 4). It is well known that biofilm formation is a significant virulence factor. In this study, although the drug-resistant strains exhibited an enhanced ability to form biofilms, this was accompanied by a reduction in virulence (as demonstrated in the G. mellonella larval infection model). This suggests that the bacteria may have diverted more energy and resources to the construction of biofilms, thereby "sacrificing" resources otherwise allocated for the synthesis of virulence factors.

Fig. 4.

Overall comparison of biofilm formation capacity between susceptible and resistant isolates. The induced resistant strain exhibited a significantly higher OD₅₇₀ₙₘ value compared to the susceptible strain; Student’s t tests were performed for biofilm formation, **P < 0.01

CAZ–AVI resistance acquisition reduces virulence and in-vitro growth competitiveness

The virulence of the 12 CR-hvKP isolates was evaluated using the G. mellonella infection model. Standardized bacterial suspensions were injected into G. mellonella larvae, with PBS serving as a control, and larval survival was monitored for 72 h. Survival curve analysis revealed that susceptible strains F7624, F7644, and F7951 caused 100% larval mortality within 24 h, whereas susceptible strain F7837 induced complete larval death within 48 h (Fig. 5A, B). In comparison, resistant strains were associated with increased larval survival rates. These data suggested that the acquisition of CAZ–AVI resistance is accompanied with attenuated virulence.

Fig. 5.

A, B Kaplan–Meier survival analysis of the G. mellonella larval infection model. Larvae were infected by different CR-hvKP isolates, using PBS as a negative control. Each group contained 10 larvae. Survival curves were plotted using the Kaplan–Meier method. C In vitro competition assay. Susceptible and corresponding resistant strains were mixed at a 1:1 ratio and co-cultured on plates for 16–18 h. The final ratio of strains after co-culture was determined by plate counting. The susceptible strains demonstrated a competitive growth advantage over the resistant strains. D Comparison of growth curves between resistant and susceptible strains. Suspensions of 12 CR-hvKP strains were standardized, added to a 96-well plate, and monitored dynamically for 24 h (absorbance at 540 nm measured every 30 min). No significant difference was observed in the growth kinetics between susceptible and resistant strains. All experiments were independently repeated three times. Student’s t tests were performed for Growth curves. **P < 0.01, *P < 0.05, ns, not significant

The effect of CAZ–AVI resistance acquisition on CR-hvKP fitness was evaluated using competitive growth experiments. Mixed cultures of resistant and susceptible isolates were plated on selective and non-selective media and incubated for 16–18 h. The sensitive strains F7692, F7624, and F7951 exhibited significantly better growth than their resistant counterparts (Fig. 5C). These observations suggested a reduced competitive fitness and virulence in vitro associated with resistance acquisition.

Susceptible and resistant strains exhibit comparable growth capacity

To investigate the changes in growth rate of the bacterial strain after resistance induction by CAZ–AVI, the growth dynamics of both susceptible and induced-resistant isolates were monitored spectrophotometrically under identical initial inoculum sizes and culture conditions. Growth curves revealed no significant differences in growth kinetics between susceptible and resistant strains (Fig. 5D). In this study, although growth curves under monoculture conditions showed no significant differences, competition assays clearly revealed a decrease in fitness in the resistant strain. This finding suggests that the fitness cost associated with resistance mutations may not manifest as changes in the basal growth rate, but rather as a reduction in comprehensive competitive ability within complex ecological contexts. Thus, competition assays offer a more realistic picture of the adaptive challenges and evolutionary dynamics of drug-resistant bacteria in their natural habitats compared to growth curve assays alone.

CAZ–AVI resistance in CR-hvKP is conferred by KPC gene mutation

To further investigate the differences in pathogenicity and antibiotic resistance between susceptible and resistant isolates, pre- and post-induction strains were subjected to WGS. Common resistance genes in CR-hvKP include KPC-2, CTX-M-65, SHV-11, TEM-1, rmtB, and sul2 [26]. WGS analysis identified the resistance genes KPC, CTX-M, SHV, TEM, qnrS, aac (6')-Ib-cr and mphA in our isolates (Table 2).

Table 2.

Major drug resistance genes in CR-hvKP

KPC-2 is a common CAZ–AVI resistance determinant. In this study, the pre-induction strains F7692, F7644, F7678, F7837, and F7951 harbored KPC-2. The resistant strains L7692, L7678, L7837, and L7951 harbored KPC variants (KPC-87, KPC-33, KPC-71, and KPC-12, respectively). Strain F7644 lost its KPC-2 gene during resistance induction but acquired novel resistance genes, including TEM-1 and CTX-M-62. Strain F7624 evolved from a KPC-2-negative susceptible strain to a resistant strain carrying KPC-2. These results indicated that the distribution and content of common drug resistance genes differ among different bacterial strains (Fig. 6). The relatively low level of induced resistance in strain F7624 may be attributed to the absence of KPC-2 in the susceptible parent strain.

Fig. 6.

Distribution of common drug resistance genes in 12 CR-hvKP isolates. The bottom of the figure lists the names of the resistance genes and their respective classes. The left side shows the isolate name. Blue tiles indicate the presence of a gene in an isolate, while yellow tiles indicate its absence. The right side of the figure displays the count of common resistance genes per isolate, with different colors representing the various classes of resistance genes

WGS revealed that our isolates harbored virulence genes, including iroE, iucABCD, and iutA, which were predominantly located on plasmids (Fig. 7). The resistant strains had a similar or decreased virulence gene content compared to their parent strains, suggesting attenuated virulence in the resistant isolates, which was confirmed by the G. mellonella infection results.

Fig. 7.

Comparative analysis of virulence gene genetic context in plasmids of CR-hvKP isolates. Arrows indicate transcriptional directions of individual genes. Gray shading represents regions with nucleotide sequence identity ≥ 90.0%. The plasmids of the susceptible strains and their corresponding resistant strains demonstrate a high degree of similarity

Discussion

CR-hvKP isolates cause severe nosocomial infections worldwide and exhibit resistance to multiple antibiotics. Common mechanisms of carbapenem resistance include the acquisition of carbapenem-resistance plasmids by hvKP, acquisition of hypervirulence plasmids by CRKP, and simultaneous acquisition of hybrid plasmids carrying both hypervirulence and carbapenem-resistance determinants by KP. KP frequently undergoes genetic mutations in the course of transmission. [27]. KP can acquire resistance genes through plasmids and mobile genetic elements, leading to multidrug-resistant and extensively drug-resistant strains [28]. However, the mechanisms underlying their adaptive changes following the development of drug resistance remain poorly understood. In our study, CAZ–AVI-resistant mutants emerged in the clinical isolates within 18 days of stepwise exposure to increasing concentrations of CAZ–AVI, starting at 0.25 MIC. Although this in-vitro approach differs from clinical treatment protocols, the rapid resistance development was noteworthy. Antibiotic resistance acquisition often directly or indirectly reduces bacterial fitness and virulence [29]. The development of antibiotic resistance represents a double-edged sword: while essential for survival under selective pressure, it often incurs fitness costs, potentially resulting in decreased virulence, pathogenicity, and competitiveness, and compromised adaptation to new ecological niches. Our phenotypic analyses demonstrated that resistant strains exhibited enhanced biofilm formation capacity concurrent with increased resistance, while showing reduced virulence. Overall, the resistant isolates displayed diminished competitive fitness, although growth rates were comparable between resistant and susceptible strains.

The macroscopic phenotypic alterations were accompanied by genetic changes. In the isolates in this study, the majority of resistance genes were located on plasmid 2. Analysis of resistance determinants in the six paired isolates revealed that most resistant strains acquired resistance through KPC-2 variants, consistent with previous findings that KPC-2 mutations frequently confer CAZ–AVI resistance [30]. Genetic analysis revealed distinct KPC-2 variants across different strains: a base deletion in plasmid 2 of strain F7692 led to KPC-87; a base substitution in plasmid 2 of strain F7678 produced KPC-33; and a base insertion in plasmid 2 of strain F7837 resulted in KPC-71. Furthermore, the KPC-2 gene in plasmid 2 of strain F7951 appears to have undergone translocation, evolving into the KPC-12 variant found on plasmid 1 of strain L7951(Table 3). Notably, KPC mutants reportedly can revert to wild-type KPC-2 under carbapenem pressure. [6]This indicates that using carbapenems alone after KPC mutation may be clinically inappropriate as it may result in the reversion of KPC mutations to the wild type, resulting in more complex clinical conditions [31]. Additionally, other resistance-mediating genes, including TEM, CTX-M, and SHV, frequently mutate, contributing to antibiotic resistance [32]. The isolates also harbored quinolone resistance genes, such as qnrS and aac (6')-Ib-cr [33, 34].

Table 3.

KPC variants: nucleotide changes, amino-acid substitutions, and positions relative to KPC-2 reference

Analysis of virulence determinants using WGS revealed multiple virulence-associated genes, including iroE, iucABCD, and iutA. These siderophore-related genes enhance iron acquisition and bacterial fitness under iron-limited conditions, contributing to pathogenicity and transmission efficiency [7]. The isolates also harbored fimbrial genes (e.g., fimB) crucial for bacterial motility, adherence, and invasion, as well as capsular genes such as rmpA2, which regulate the production of capsular polysaccharides associated with hypermucoviscosity and increased virulence [35]. In this study, the observed attenuation of virulence and concomitant loss of virulence genes in CR-hvKP following the acquisition of resistance is potentially mediated by plasmid rearrangement. The proposed mechanism is as follows: the resistance and virulence plasmids may have fused into a single, large plasmid via homologous recombination or transposition, significantly increasing the metabolic burden on the bacterium. To alleviate this burden, homologous recombination between flanking repetitive sequences could lead to the excision and loss of the intervening virulence gene block. Provided the resistance genes reside outside this deleted region, the strain achieves a metabolic "load reduction" while maintaining its resistance. This represents an evolutionary strategy employed by the bacterium to optimize fitness under strong antibiotic selective pressure. Notably, heightened vigilance is warranted against the emergence of superbugs that concomitantly integrate high virulence and high resistance through such evolutionary processes.

The MLST results for all 12 isolates in our study were identified as ST11, which is the most prevalent CR-hvKP clone worldwide, with notable predominance in China [36]. The significantly higher prevalence of virulence plasmids among ST11 isolates than in other STs indicates that ST11 serves as a primary vector for virulence plasmid acquisition and transmission [37].

The evolution of antibiotic resistance within bacterial populations depends on multiple complex factors, including the fitness costs associated with resistance acquisition. These fitness costs result in susceptible strains demonstrating superior environmental adaptation and competitive advantage over resistant strains in antibiotic-free conditions, as corroborated by our in-vitro competition assay results. With the global dissemination of CR-hvKP, effective antimicrobial agents are crucial for infection control. Understanding the specific resistance mechanisms of clinical CR-hvKP isolates enables clinicians to tailor appropriate antibiotic therapy, thereby effectively eliminating infections and improving clinical outcomes. Virulence factor analysis may reveal potential targets for CR-hvKP drug development and novel therapeutic approaches, providing new methodological insights. Future investigations of the transmission mechanisms of hypervirulence and resistance genes will be instrumental in developing new therapeutic strategies and preventing dissemination.

Conclusion

The fitness cost associated with antibiotic resistance is a key aspect of bacterial evolution. This study evaluated the fitness cost of CAZ–AVI resistance in CR-hvKP, revealing that the acquisition of CAZ–AVI resistance in CR-hvKP is primarily mediated by mutations in the KPC gene, accompanied by reduced bacterial virulence and competitive fitness. Therefore, enhanced investigation into the fitness cost of CR-hvKP will contribute to a more comprehensive understanding of the trajectory of its resistance development, offering new insights for addressing the antibiotic resistance crisis.

Abbreviations

BLAST

Basic Local Alignment Search Tool

CARD

Comprehensive Antibiotic Resistance Database

CAZ-AVI

Ceftazidime–avibactam

CR

Carbapenem-resistant

CR-hvKP

Carbapenem-resistant hypervirulent KP

hv

Hypervirulent

KP

Klebsiella pneumoniae

KPC

KP carbapenemase

MIC

Minimum inhibitory concentration

MLST

Multilocus sequence typing

PFGE

Pulsed-field gel electrophoresis

SNP

Single-nucleotide polymorphism

VFDB

Virulence Factors Database

WGS

Whole-genome sequencing

Authors’ contributions

The study was designed by XHK. MH, HMG and SPX conducted experiments and wrote the original draft. YTL carried out the analysis, DDW and XHK revised the manuscript. QHZ played a key role in guiding the revision of the manuscript. All authors contributed to and approved the final version of the article.

Funding

This study was supported by the National Natural Science Foundation of China (82160553); Jiangxi Science and Technology Cooperation Project (20244BDF60008); The Science and Technology Plan of the Jiangxi Health Committee (grant No.: SKJP-220211638).

Data availability

All data generated or analyzed during this study are included in this published article and raw data will be available upon request. All sequencing data have been uploaded to NCBI, with the following accession numbers: PRJNA1280472.

Declarations

Ethics approval and consent to participate

The study was performed in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Nanchang University (Review Opinion No.: IIT [2024] Clinical Ethics Review No. 710). The Ethics Committee of the First Affiliated Hospital of Nanchang University waived the need for informed consent for the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.