bla KPC-2 overexpression and bla GES-5 carriage as major imipenem/relebactam resistance mechanisms in Pseudomonas aeruginosa high-risk clones ST463 and ST235, respectively, in China

ABSTRACT Pseudomonas aeruginosa high-risk clones pose severe threats to public health. Here, we characterize the imipenem/relebactam (IR) resistance mechanisms in P. aeruginosa high-risk clones sequence type 235 (ST235) and ST463 in China. Minimum inhibitory concentrations (MICs) were determined, and Illumina short-read sequencing was performed for 1,168 clinical carbapenem-resistant P. aeruginosa (CRPA) isolates. The gene copy number and expression level were analyzed by Illumina sequencing depth and reverse transcription-quantitative PCR, respectively. Resistance conferred by bla GES-5 was evaluated by cloning experiments. ST463 and ST235 accounted for 9.8% (115/1,168) and 4.5% (53/1,168) of total isolates, respectively, and showed high frequencies of extensively drug-resistant and difficult-to-treat resistant phenotypes. The overall IR-resistant rate in CRPA was 21.0% (245/1,168). However, the IR resistance rate was 81.7% (94/115) in ST463-PA and 52.8% (28/53) in ST235-PA. Of the ST463 isolates, 92.2% (106/115) were Klebsiella pneumoniae carbapenemase-producing P. aeruginosa (KPC-PA), and all 94 IR-resistant ST463-PA produced KPC-2. Compared to IR-susceptible ST463 KPC-2-PA, IR-resistant ST463 KPC-2-PA exhibited significantly higher bla KPC-2 copy numbers and expression levels. In ST463 KPC-2-PA, 16 mg/L relebactam resulted in additional fourfold reductions in imipenem MIC50/90 values compared to 4 mg/L relebactam. In ST235, 1.9% (1/53) carried bla IMP carbapenemase and 54.7% (29/53) carried bla GES carbapenemase. Other than the IMP producer, all 27 IR-resistant ST235-PA produced GES-5. Cloning experiments revealed that imipenem resistance in bla GES-5-carrying PAO1 transformants was generally unaffected by relebactam. In conclusion, IR-resistant CRPA isolates in China were mainly distributed in P. aeruginosa high-risk clones ST463 and ST235. The major underlying IR resistance mechanisms were bla KPC-2 overexpression and bla GES-5 carriage.

Relative to other CRPA infections, a substantial increase in mortality has been demonstrated for ST235 and ST463 CRPA infections, especially in patients with blood stream infections (BSIs) (3)(4)(5).The higher mortality rates observed could be largely explained by two common clonal characteristics of ST235 and ST463.First, because of horizontally acquired and mutation-driven resistance, ST235-PA and ST463-PA often exhibit a multidrug-resistant (MDR), extensively drug-resistant (XDR), or pandrug-resist ant (PDR) profile (6,7), which limits treatment choices and could lead to inappropriate empirical antibiotic therapy.Second, ST235-PA and ST463-PA have been reported to harbor a large number of virulence determinants, thus leading to more severe infections (6,8).Considering the limited activity of traditional antipseudomonals and the pathoge nicity of ST235 and ST463 isolates, evaluations of new antimicrobial activity are urgently needed.
Imipenem/relebactam (IR), a novel β-lactam-β-lactamase inhibitor (BLBLI) combina tion, has been developed and applied clinically to treat CRPA infections.Relebactam exerts strong inhibitory effects on many class A and class C β-lactamases and is almost unaffected by OprD inactivation or efflux pump extrusion (9,10).Consequently, IR has shown considerable antibacterial activity against P. aeruginosa.In a multicenter study involving 4,927 MDR-PA isolates from adult patients, although 13.5% of samples produced metallo-β-lactamases (MBLs), nearly 70% of isolates were still susceptible to IR (11).
Currently, although many studies emphasize the in vitro activity of IR against P. aeruginosa, the IR resistance mechanisms in CRPA high-risk clones remain largely unexplored.In this study, we evaluated the IR resistance profiles in clinical CRPA isolates from China, with a particular focus on the IR activity against ST235 and ST463 CRPA strains.Notably, we identified a surprisingly high IR resistance rate in ST235 and ST463 CRPA isolates.The underlying IR resistance mechanisms were further explored in detail.

Bacterial isolates and antimicrobial susceptibility testing
A total of 1,168 clinical CRPA isolates, defined as resistant to either meropenem or imipenem according to the Clinical and Laboratory Standards Institute (CLSI) breakpoints (12), were included in this study.These isolates were obtained from two separate epidemiological survey programs.In the first survey conducted in a tertiary hospital in Hangzhou, China, 839 nonduplicated CRPA strains isolated from various specimens were collected during 2011-2020.In the second survey of patients diagnosed with hospital-acquired pneumonia from 33 tertiary hospitals in China, 329 nonduplicated CRPA strains isolated from sputum or bronchoalveolar lavage fluid were collected during 2019-2020.A nonduplicated CRPA strain was defined as one CRPA strain per patient during a single hospitalization.
For all 1,168 CRPA isolates, the minimum inhibitory concentrations (MICs) of amikacin, imipenem, meropenem, IR (relebactam fixed at 4 mg/L), ceftazidime, cefepime, ciprofloxacin, levofloxacin, piperacillin/tazobactam, aztreonam, fosfomycin, and polymyxin B were determined using the broth microdilution method with Thermo Scientific Sensititre System.To facilitate comparison between different antimicrobials in ST463 KPC-2-producing and ST235 GES-5-producing strains, the MICs of imipenem and IR (relebactam fixed at 4 mg/L or 16 mg/L) were simultaneously obtained by the agar dilution method described in CLSI standard M07 (13).Escherichia coli ATCC 25922, P. aeruginosa ATCC 27853, and Klebsiella pneumoniae ATCC 700603 and ATCC BAA-1705 were used as quality controls.For all antimicrobials except fosfomycin, CLSI breakpoints (12) for P. aeruginosa were adopted for result interpretation.For fosfomycin, CLSI breakpoints (12) for Enterobacterales were adopted.The MDR/XDR/PDR phenotypes and the difficult-to-treat resistant (DTR) phenotype were defined according to the criteria established by Magiorakos et al. and Tamma et al. respectively (14,15).
For KPC-2-producing strains, the bla KPC-2 copy number was further calculated based on WGS.For each sample, a SAM alignment file was obtained by mapping Illumina raw reads to references (bla KPC-2 gene and seven MLST alleles) using bowtie2 version 2.2.5 (https://github.com/BenLangmead/bowtie2)with the option "-I 0 -X 800 --fast -q." After transforming the alignment results into BAM files, the sequencing depths of the bla KPC-2 gene and seven MLST alleles were analyzed by BEDTools version 2.30.0 (https:// github.com/arq5x/bedtools2)with the option "bedtools coverage -mean." Finally, the bla KPC-2 copy number of each strain was calculated as the depth of bla KPC-2 divided by the average depth of the seven MLST alleles.

Quantitative determination of gene expression level
To compare the expression levels of bla KPC-2 , mexA, mexD, mexE, and mexY genes in clinical isolates with different resistance phenotypes, we extracted the total RNA of selected strains using Total RNA Kit I (Omega Bio-Tek, Georgia, USA).Reverse transcrip tion was conducted using PrimeScript RT Reagent Kit (Takara, Beijing, China).Quantita tive PCR was performed in biological triplicates using TB Green Premix ExTaq (TaKaRa, Beijing, China).Each biological triplicate was assayed in technological triplicates.The housekeeping gene rpoD was chosen as the endogenous reference for normalization.The primers used for quantitative PCR are listed in Table S1.

Cloning experiments
The bla GES-5 and bla GES-15 genes with the same upstream promoter sequences were amplified from clinical strains R16-15 and R20-99, respectively.The cloned sequences were homologously recombined into the pGK1900 plasmid.The constructed plasmids were then transformed into both E. coli DH5α and P. aeruginosa PAO1.Transformants verified by Sanger sequencing were eventually selected for antimicrobial susceptibility testing.The primers used for cloning experiments are listed in Table S1.

Statistical analysis
Statistical analysis was performed using the Fisher's exact test, the Mann-Whitney U test, and the Kruskal-Wallis test in SPSS version 25.The results were considered statistically significant when P < 0.05.
With a fixed relebactam concentration of 4 mg/L, IR MICs were determined and compared with imipenem MICs to explore the antibacterial activity of IR against CRPA isolates.In general, 1,147/1,168 (98.2%) isolates were resistant to imipenem with an MIC 50/90 of 16/>128 mg/L; however, only 245/1,168 (21.0%) isolates were resistant to IR with an MIC 50/90 of 2/32 mg/L (Table 1).The addition of 4 mg/L relebactam led to an eightfold decrease in imipenem MIC 50 value and at least an eightfold decrease in imipenem MIC 90 value (Table 1; Fig. 1).The overall results demonstrated the notable in vitro activity of IR against clinical CRPA isolates.
The phylogenetic tree was then constructed for IR-resistant isolates, and the clonal characteristics were analyzed.Among 245 IR-resistant isolates, 62 different STs were identified (excluding five strains with unknown STs).In 16 of these 62 STs, the IR-resist ant strain numbers were more than one (Fig. S3).As shown in Fig. 2, ST463-PA and ST235-PA were still the most prevalent, together accounting for 49.8% (122/245) of total IR-resistant isolates.Notably, differing from the low-level IR resistance observed in most non-ST463 non-ST235 isolates, a significantly higher IR resistance level (MIC 32/4 to >128/4 mg/L) was observed in the majority of ST463 and ST235 isolates, especially when excluding MBL producers (Fig. 2).In terms of the effect of 4 mg/L relebactam, restoration of imipenem susceptibility was only achieved in 7.8% (9/115) of ST463-PA and 18.9% (10/53) of ST235-PA isolates, which was considerably lower than the average (Table 1).Altogether, these resistance profile differences indicated the contribution of clonal characteristics to IR resistance in ST463-PA and ST235-PA.

Molecular epidemiology and resistance phenotype of ST463-PA and ST235-PA
To further describe the molecular epidemiology and interpret the IR resistance data of ST463 and ST235 isolates, genomic analysis was subsequently conducted.As listed in The relationships between carbapenem carriage and IR resistance phenotype in ST463-PA and ST235-PA were also assessed.In brief, ST463 and ST235 non-CPPA isolates were all susceptible or intermediate to IR, but the IR resistance phenotype in ST463 and ST235 CPPA isolates varied.Specifically, in ST463, 2/2 (100%) bla KPC-33 -carrying P. aeruginosa were IR intermediate, while 94/104 (90.4%) bla KPC-2 -carrying P. aeruginosa were IR resistant (Fig. 3a); in ST235, other than one IMP-15 producer, 2/2 (100%) bla GES-15 -carrying P. aeruginosa were IR intermediate or IR susceptible, and 27/27 (100%) bla GES-5 -carrying P. aeruginosa were IR resistant (Fig. 3b).Overall, this analysis strongly indicated the potential involvement of KPC-2 and GES-5 in IR resistance in ST463-PA and ST235-PA.
Finally, to further verify our hypothesis, the relative bla KPC-2 expression level was quantitively analyzed.In total, 18 bla KPC-2 -carrying ST463 isolates were selected from three phenotypic groups (≥64-fold, 16-fold, and 4-fold imipenem MIC decrease with 4 mg/L relebactam).These strains shared similar genetic elements surrounding bla KPC-2 and the same bla KPC-2 promoter sequences.As shown in Fig. 5d, the relative bla KPC-2 expression level differed significantly among these three groups (3.1 vs 4.5 vs 12.6, P < 0.0001), showing that bla KPC-2 overexpression was related to the attenuated inhibitory effects of 4 mg/L relebactam.

IR resistance mechanisms in ST235 GES-5 producers
Mechanisms of IR resistance were also investigated in detail for ST235 GES-5-PA.As illustrated in Table 1 and Fig. 4b, neither 4 mg/L nor 16 mg/L relebactam remarka bly altered the MIC 50 /MIC 90 values or the cumulative inhibition curves of imipenem, suggesting that IR resistance in ST235 GES-5-PA could not be attributed to bla GES-5 overexpression.
Subsequently, the impacts of bla GES allelic variants on imipenem and IR resistance were analyzed.The imipenem MICs were generally higher in ST235 GES-5 producers (32 mg/L to 128 mg/L) than in ST235 GES-15 producers (16 mg/L).The IR resistance rate was 100% (27/27) for bla GES-5 -carrying ST235-PA isolates, with MIC 50 /MIC 90 values of 32/32 mg/L.In contrast, all bla GES-15 -harboring ST235-PA isolates were susceptible or   Cloning experiments were therefore conducted to confirm that IR resistance in ST235 GES-5-PA was mainly attributable to bla GES-5 carriage (Table 2).Compared to transform ants carrying the empty vector, bla GES-5 -carrying transformants displayed 8-to 16-fold increases in imipenem MIC.In bla GES-5 -harboring PAO1 transformants, the addition of 4 mg/L and 16 mg/L relebactam only caused a slight decrease in imipenem MIC.These results indicated that relebactam exerted poor inhibitory effects on GES-5 carbapene mase activity in P. aeruginosa, thus resulting in IR resistance in ST235 GES-5-PA.

DISCUSSION
Given the increasing clinical availability of IR and the severe threat of P. aeruginosa high-risk clones, an understanding of IR resistance in P. aeruginosa high-risk clones is urgently needed.In this study, we explored the IR resistance profiles of 1,168 clinical CRPA isolates from China, which revealed high-risk clones ST463 and ST235 as the major contributors to IR resistance.ST463 and ST235 clones in this study were characterized by KPC and GES carbapenemase carriage, respectively.Further experiments and genomic analysis demonstrated that bla KPC-2 overexpression combined with OprD inactivation mediated high-level IR resistance in ST463-PA, while bla GES-5 carriage mediated relatively low-level IR resistance in ST235-PA.
Several studies have reported the in vitro or in vivo dynamic evolution of IR resistance in P. aeruginosa, mainly focusing on the contribution of the mutational resistome.In a stepwise resistance development study (20), Gomis-Font et al. obtained IR-resistant mutants of wild-type and hypermutable PAO1 strains by serial passage.Most mutants only exhibited low-level IR resistance, which was associated with OprD inactivation and MexAB-OprM overexpression.High-level IR resistance could be achieved if an additional mutation (PBP1a T680A) was acquired.In another in vitro study (19), Gomis-Font et al. conducted the same experiments on three P. aeruginosa high-risk clones.Similarly, only borderline resistance was identified, which was caused by MexB or ParS struc tural mutations.Recently, Shields et al. reported the emergence of IR nonsusceptibility following the treatment of MDR-PA infections for the first time (21).Out of 19 patients treated with IR, isolates not susceptible to IR were recovered from five patients.All isolates displayed moderate IR resistance related to MexAB-OprM and/or MexEF-OprN mutation, which could be reverted by PAβN inhibition.Overall, the promising antibacte rial effects of IR in non-CPPA could be concluded.
Generally, IR resistance related to KPC enzymes has seldom been reported.Until 2022, four studies investigating IR resistance development in KPC-producing K. pneumoniae (KPC-KP) were published.In one in vitro multistep selection study (29), IR-resistant mutants were selected for five KPC-KP strains.WGS and quantitative PCR confirmed OmpK36 inactivation, KPC allele mutations, and increased bla KPC copy number as the major IR resistance mechanisms.In other studies describing the in vivo evolution of IR resistance (30-32), Gaibani et al. isolated three IR-resistant KPC-KP strains, which were compared with the corresponding parental KPC-KP strains susceptible to IR.An increased bla KPC copy number was confirmed in all IR-resistant strains, manifesting the rapid resistance evolution in KPC producers.Similar to the findings described above, we demonstrated that increased bla KPC-2 copy number and bla KPC-2 expression level were also associated with IR resistance in P. aeruginosa.Concerningly, ST463 KPC-2-PA isolates in our study were never exposed to IR before collection, but they still showed an extremely high IR resistance rate (90.4%).Under these circumstances, clinical treatment choices would be greatly constrained.
Although GES producers were found among IR-resistant P. aeruginosa in previous studies (27,33), the causal relationship between GES and IR resistance is uncertain.With GES-15 as the reference, the low-level IR resistance mediated by GES-5 in P. aeruginosa was confirmed by cloning experiments in this study.This is consistent with the enzymatic properties of GES-20 (identical to GES-5 with the leader sequence removed).According to Hujer et al. 's study (34), GES-20 possesses a constrained Ω loop; thus, relebactam is positioned in an unstable conformation, which impairs the inhibitory effect of relebac tam.
The poor outcome associated with ST235 and ST463 CRPA, mainly caused by their hypervirulence and MDR/XDR/PDR profiles, is concerning.Generally, hypervirulence is defined as the ability to cause higher mortality and more severe acute infections (1).ST235 CRPA isolates typically possess a hypervirulent exoU+ genotype and various resistance-related mutations.In a study on P. aeruginosa bacteremic pneumonia, a 29.8% increase in 5-day mortality and a 34.4% increase in 30-day mortality were noted in patients infected with ST235-PA (4).Similarly, harboring both exoU and exoS genes, the ST463 clone also exhibits a hypervirulent phenotype.In a recent study by Zhang et al. (35), the virulence level of different strains was assessed by cytotoxicity assay and the Galleria mellonella larvae infection model.The included ST463-PA isolates showed hypervirulence comparable to that of ST235-PA.Moreover, in a clinical retrospective cohort study conducted in East China (5), ST463 CRPA accounted for 48.0% of total CRPA BSI cases and 69.6% of total death cases, highlighting its serious threat.Thus, therapeutic options should be carefully considered to improve clinical outcomes in patients infected with ST235-PA and ST463-PA.
In conclusion, our data highlighted the distinct IR resistance characteristics between CRPA isolates from high-risk clones and other STs.Of particular concern is the associa tion of high-risk clones with IR resistance phenotype and XDR/DTR phenotype, which may worsen clinical outcomes and restrict therapeutic options.Using whole-genome analysis methods and quantitative PCR, we identified bla KPC-2 overexpression as the major driver of high-level IR resistance in ST463-PA.In addition, we also confirmed that bla GES-5 carriage mediated IR resistance in ST235-PA.Overall, this epidemiological survey systematically deciphered the IR resistance mechanisms in P. aeruginosa global high-risk clone ST235 and regional high-risk clone ST463.

TABLE 1
In vitro activities of imipenem and imipenem/relebactam against Pseudomonas aeruginosa isolates in this study a