Pseudomonas aeruginosa is one of the most common opportunistic pathogens. Despite a slight decrease in its isolation rate in recent years, it continues to be a significant etiological agent of healthcare-associated lower respiratory tract infections (Weiner et al., 2016). Carbapenem-resistant Pseudomonas aeruginosa (CRPA) presents considerable therapeutic challenges in clinical settings due to its inherent resistance mechanisms and strong environmental adaptability. According to data from the European Antimicrobial Resistance Surveillance Network (EARS-Net), the detection rate of CRPA was 18.1% in 2021. In the United States, the isolation rate of CRPA varied between 10% and 30% (Tenover et al., 2022), while in China it was 15.2% (Karlowsky et al., 2021). The 30-day overall mortality rate attributed to CRPA infection is above 18.1%, and there were significant geographical and infection-type differences in mortality; specifically, the mortality rate for lower respiratory tract infections attributed to CRPA can be as high as 19% (Reyes et al., 2023). Considering its significant threat to global health, the World Health Organization (WHO) reaffirmed the classification of CRPA as a critical priority pathogen in 2024 (Sati et al., 2025). Moreover, for better classification of the resistance profile of Pseudomonas aeruginosa, the additional new class of difficult-to-treat resistant Pseudomonas aeruginosa (DTR-PA) was introduced in 2018 (Kadri et al., 2018). DTR-PA is defined by strains that exhibit non-susceptibility to all of the following antimicrobial agents: piperacillin-tazobactam, ceftazidime, cefepime, aztreonam, meropenem, imipenem-cilastatin, ciprofloxacin, and levofloxacin (Tamma et al., 2024). The detection rate of DTR-PA has increased in recent years, presenting a significant threat to human health (Wu et al., 2024).

The resistance mechanisms of CRPA are multifaceted and can be primarily classified into three categories: β-lactamase production or overexpression, mutations in outer membrane proteins, and overexpression of efflux pump systems. Among these, non-enzymatic carbapenem resistance mechanisms are relatively prevalent (Falcone et al., 2024). Additionally, virulence genes are crucial in the pathogenicity of CRPA, which possesses various virulence factors such as lipopolysaccharide (LPS), quorum sensing, two-component systems, type VI secretion systems (T6SS), outer membrane vesicles, and the CRISPR-Cas system along with its regulatory mechanisms. These virulence factors increase bacterial pathogenicity via multiple mechanisms and cause harm to host cells (Letizia et al., 2025). Within the type III secretion system (T3SS), four key effectors—exoU, exoT, exoY, and exoS—play significant roles in bacterial pathogenesis by enabling the translocation of toxic proteins into host cells (Hu et al., 2013; Hardy et al., 2022; Nolasco-Romero et al., 2024). ExoS and exoU are considered as the major contributors in the pathogenicity of Pseudomonas aeruginosa. Notably, exoS⁺/exoU⁺ strains have been reported in 19 countries or regions worldwide (Song et al., 2023). A multicenter study conducted in China revealed that the majority of exoU+ strains display high virulence, and clonal clusters of ST463 exoU+ Pseudomonas aeruginosa were identified in southeastern China (Zhu et al., 2023). Such results emphasize the importance of considering local epidemiology since there is geographical diversity not only in resistance mechanisms but also in high-virulence genotypes.

Despite the growing recognition of the interplay between bacterial virulence and antimicrobial resistance, the epidemiological association between the resistance profile of CRPA and T3SS virulence genotypes in lower respiratory tract infections is not well elucidated. In order to further investigate the correlation between them, this study conducted WGS analysis on 106 clinical CRPA isolates obtained from a tertiary hospital in Ningbo, collected between July 2021 and May 2025. It also studied the association among T3SS virulence gene presence, bacterial virulence, antibiotic resistance, and molecular epidemiological patterns. This study aim to provide evidence for the optimization of dual surveillance strategies focused on high-virulence, DTR-PA strains.

According to the Global Burden of Disease (GBD) 2019, lower respiratory tract infections were the leading cause of infection-related mortality. More than 4 million people died worldwide from these infections in that year, and they became the infection syndrome with the highest mortality in the five super-regions (GBD 2019 Antimicrobial Resistance Collaborators, 2022). Pseudomonas aeruginosa is one of the important pathogenic bacteria in this field. It has strong drug resistance and poses treatment challenges, frequently resulting in serious complications, including ventilator-associated pneumonia. It poses a life threat to patients with pre-existing conditions and people with compromised immune systems. In addition, Pseudomonas aeruginosa exhibits a greater propensity for mutation and possesses more intricate drug resistance mechanisms compared to other gram-negative bacteria (Yang et al., 2025). Currently, there is a concerning global increase in Pseudomonas aeruginosa that produce KPC enzymes and isolates that carry the exoU and exoS virulence genes (Zhang et al., 2022; Zhang et al., 2022; Sun et al., 2023; Li et al., 2024). These strains have raised significant concern and vigilance due to their extensive drug resistance and high virulence.

In our study, all strains were found to carry the genes aph(3’)-IIb, catB7 and fosA, which are related to resistance to aminoglycosides, chloramphenicol and fosfomycin. The presence of these three genes indicated the potential existence of integrons. In addition, we detected the presence of various β-lactamases, including carbapenemases. Epidemiological data indicated that the majority of infected patients had prior exposure to β-lactamase inhibitors, specifically piperacillin/tazobactam before the detection of bla_OXA-50 strains. The MIC for the strains against piperacillin/tazobactam exhibited a general increase, with a drug resistance rate of 45.2%, which is close to the values reported in Japan and other countries (Yano et al., 2024). Traditional β-lactamase inhibitors exhibited limited inhibitory activity against bla_OXA-50(Horcajada et al., 2019). We detected 19 known subtypes of the bla_OXA-50 family and identified 2 new unknown subtypes. These findings together suggest that the bla_OXA-50 family may undergo functional differentiation through continuous genetic evolution and play a key role in the development of piperacillin/tazobactam resistance phenotypes. The extensive application of β-lactamase inhibitors may serve as a selective pressure, influencing the adaptive evolution and subtype diversification within this gene family. This mechanistic hypothesis offers a significant theoretical foundation for the optimization of inhibitors and the formulation of drug resistance monitoring strategies. Various carbapenemases have been identified in CRPA isolates across different geographical regions. Isolates from South America and Central America predominantly carried bla_KPC-2 and bla_VIM-2; bla_NDM-1 and bla_IMP-1 represent the predominant carbapenemase genes identified in Australia and Singapore (Reyes et al., 2023). However, bla_KPC-3 was found as the predominant enzyme type in this study, with only a single strain of bla_KPC-2 identified. This finding contrasts sharply with previous reports from China, where bla_KPC-2 was typically the main enzyme type observed (Reyes et al., 2023). Although some strains were resistant to carbapenems, these strains remained susceptible in vitro to other β-lactams, cephalosporins, or fluoroquinolones. In such cases, traditional β-lactams should be prioritized to avoid the utilization of new antibiotics. We also noticed a significant proportion of DTR-PA. Numerous studies have confirmed that ceftazidime-avibactam (CZA) exhibited substantial effectiveness against clinical strains of Pseudomonas aeruginosa responsible for lower respiratory tract infections, including DTR-PA (Corbella et al., 2022; Shields et al., 2025). However, the number of CZA-resistant strains is also increasing. In this study, the resistance rate to CZA reached 10%, which was lower than the 18.1% reported by the China Antimicrobial Surveillance Network in 2024 (CHINET.CRPA., 2025). CZA resistance is linked to: (I) mutations in β-lactamases, such as KPC, AmpC, CTX-M, and OXA-48; (II) the production of metallo-β-lactamases that are not inhibited by avibactam; (III) overexpression of hydrolases; (IV) heightened efflux pump activity, including AcrA/B–TolC and MexA/B–OprM; and (V) diminished cellular permeability (Horcajada et al., 2019). The high prevalence of KPC-3 in our study further explained the CZA resistance observed in certain strains, as previous research has documented the D179Y substitution in the Ω-loop of KPC-3, resulting in the KPC-31 variant linked to CZA resistance (Faccone et al., 2022). Therefore, laboratories should actively conduct in vitro AST of CZA, and clinicians should maintain vigilance regarding KPC mutations during CZA administration.

Previous studies have reported that the loss or mutation of oprD was associated with imipenem resistance. In contrast, the mechanisms of meropenem resistance exhibit greater diversity (Fratoni et al., 2024; Yano et al., 2024). Herein, we demonstrated that all strains examined harbored mutations within the oprD gene, and the majority of the amino acid alteration at these loci resulting from nonsense mutions have been previously documented to induce varying degrees of functional alterations in oprD (Nazari et al., 2023; Wang et al., 2025). Mutation in the oprD of Pseudomonas aeruginosa could enhance the adaptability of the bacterium to the host environment during infection (Skurnik et al., 2013). Additionally, other study has reported that inactivating mutations in oprD could arise even in the absence of carbapenems exposure (Wolter et al., 2008), suggesting that the loss of oprD protein function may confer a selective survival advantage. Additionally, mutations associated with efflux pump regulation constituted a comparatively minor proportion, even less than those related to acquired carbapenemase-mediated resistance mechanisms. However, it is challenging to establish a direct causal link because every strain possesses a variety of resistance mechanisms. Future studies will aim to further validate this hypothesis through in vitro and in vivo functional assays.

T3SS constitutes the fundamental component of the pathogenic mechanism in Pseudomonas aeruginosa. It functions as a “molecular syringe,” facilitating the direct delivery of toxins into host cells, and secretes four characteristic effector proteins: exoS, exoT, exoY, and exoU. In this study, we focused on exoS and exoU, which are key factors in the pathogenicity of Pseudomonas aeruginosa, and explored the association among different genotypic patterns, resistance to common antibacterial agents, and the mechanisms underlying this resistance. Our study found that the detection rate of exoS+/exoU- was the highest, aligning with the findings of Sarges, and Tang et al (Sarges et al., 2020; Tang et al., 2025). In addition, our findings indicated that exoU+/exoS- strains exhibited an increased resistance rate to ceftazidime, cefepime, and piperacillin/tazobactam. Previous study (Palzkill, 2018) has reported that KPC-3 showed increased kcat/KM values for ceftazidime hydrolysis compared to KPC2, therefore we speculate that a higher rate of drug resistance may be associated with the presence of these strains alongside certain subtypes of bla_KPC-3. We found that exoU+/exoS- group was likely to be significantly asscociated with the presence of ICU admission, bla_KPC-3, bla_OXA-488, bla_OXA-1032, while exoU-/exoS⁺ exhibited a significant association with the presense of blaOXA-486, blaOXA-904. These associated phenomena are trends worthy of attention, which provide clues for further research of the distribution of bacterial characteristics in specific cinical settings. Further, our analysis revealed that exoU, exoS and the aforementioned drug resistance genes were located on distinct scaffolds, and no structural association between them was supported by the current evidence. Nevertheless, this does not exclude the possibility that they may reside together on a single genetic element(such as a plasmid) that has not been fully assembled due to technical limitations of next-generation sequencing scan map. The fragmented nature of second-generation sequencing assemblies may result in the physical separation of co-located sequences across different scaffolds, thereby precluding reliable inference of their linkage. These speculations need to be confirmed through further experiments such as third-generation sequencing, thereby synergistically enhancing clinical adaptability.

The analysis of MLST and O-antigen serotypes revealed notable characteristics related to the population genetic structure and the distribution of virulence genes. The analysis of MLST and O-antigen serotypes revealed notable characteristics related to the population genetic structure and the distribution of virulence genes. Notably, the identification of high-risk clones ST233, ST244, and ST357 (Del Barrio-Tofiño et al., 2020) indicated the potential presence of evolutionary selection pressures associated with antimicrobial resistance or host adaptability within the hospital environment. While ST1076 and ST491 were relatively predominant, they did not establish a single dominant clone within the population, suggesting that the transmission of current strains may demonstrate a polyclonal nature. O-antigen serotypes were dominated by O11 and O6. The concentrated distribution of ST1076-O11 in ICU and the exclusive detection of ST493-O6 in EICU established a specific spatial distribution pattern. This may be closely related to variations in device usage patterns, antibiotic exposure, or patient immune status across different ICU. In this study, the majority of O11 strains were found to carry exoU and were predominantly DTR-PA. The direct involvement of the O11 antigen in the pathogenic process has yet to be conclusively established through gene knockout experiments, and the existing data do not exclude the possibility of it serving solely as a concomitant marker. It is worth noticing that the ST463 CRPA clone carrying the bla_KPC, exoU, and exoS genes has emerged and spread in East China (Hu et al., 2021; Li et al., 2024). The dispersed distribution of these strains within our hospital indicates a necessity for vigilance to mitigate the risk of cross-ward transmission through healthcare professionals or equipment.

The analysis of virulence gens showed a distinct differentiation between phylogroup A (exoS+) and phylogroup B (exoU+), which is consistent with the evolutionary patterns observed in PAO1 (phylogroup A) and PA14 (phylogroup B). Although phylogroup A strains were significantly predominant in number, the clustering of ST1076 and ST357 in phylogroup B may reflect that these ST strains have enhanced their invasive virulence through the acquisition of exoU in T3SS. The Galleria mellonella larva infection study further validated the high-virulence potential of ST1076, ST463, and ST357 strains carring exoU. This aligns with previous studies reporting a correlation between exoU+ strains and unfavorable outcomes in severe pneumonia (Hauser et al., 2002). However, in this model, we also observed different virulence levels within these three virulence genotypes, indicating that the factors influencing the virulence of CRPA are multifaceted. These factors may not only include the type and quantity of virulence genes carried by the strains but also their expression levels (Oliveira et al., 2021), G. mellonella has been used to study the virulence treatment of a wide range of microorganisms, including P. aeruginosa (Asai et al., 2023; Campo-Pérez et al., 2025), but this model still has certain limitations. Firstly, the lack of a standardized supplier for genetically uniform larvae introduces variability in experimental outcomes. We observed significant differences in mortality rates across different batches when exposed to the same bacterial strain, suggesting batch-to-batch inconsistency that may affect reproducibility. In addition, it may not accurately reflect the pathogenicity of the pathogen in mammalian hosts. This may be related to the essential differences in the immune system between G. mellonella and mammals, which leads to the inability of some virulence factors of the pathogen to be fully expressed or function in the model (Russo and MacDonald, 2020). Therefore, the results of this study mainly indicate the relative virulence of these strains, and their direct association with human clinical outcomes still needs to be further confirmed. The diversity of ST types (43 types) among exoU-/exoS+ strains indicated that the anti-phagocytic effect mediated by exoS (Sawa et al., 2014) was broadly preserved across strains with varying genetic backgrounds. The enrichment of ST463 in the eight exoU+/exoS+ double-positive strains was noteworthy, which might indicate that this ST type had acquired an additional virulence module via horizontal gene transfer (Song et al., 2023). This rare phenotype holds considerable importance in clinical practice, with studies indicating that the 28-day mortality rate for patients infected with ST463 exoU+/exoS+ strains was significantly elevated compared to the non-ST463 group (Hu et al., 2021). These data suggest that ST463 may evolve into a significant clinical threat.

In summary, this study offers novel insights for enhancing clinical infection control of CRPA and advancing laboratory detection methods. The prevalence trend of specific clones (e.g., ST1076-O11) in high-risk wards such as the ICU suggested that environmental microbial monitoring needs to be strengthened; the potential association between phylogroup B strains (exoU+/exoS- type) represented by the highly virulent PA14 and invasive infections necessitated the implementation of rapid classification protocols based on virulence genotypes in clinical laboratories. However, this study also presents several limitations: Firstly, our samples exclusively encompassed CRPA derived from lower respiratory tract sources. Secondly, although we observed that the co-occurrence of the exoU gene with resistance genes (e.g., bla_KPC-3) and some subtypes of the bla_OXA-50 gene, the mechanisms underlying these observations remain inadequately investigated. In the future, further studies should incorporate multicenter samples of various infection types. A thorough analysis of the three-dimensional association mechanisms among genetic lineages, virulence characteristics, and clinical outcomes should be analyzed in depth combined with clinical course data.