Emergence of ST463 exoU-Positive, Imipenem-Nonsusceptible Pseudomonas aeruginosa Isolates in China

ABSTRACT This study investigated the resistance mechanisms and the distribution and proportions of virulence genes, including exoU, in 182 imipenem-nonsusceptible Pseudomonas aeruginosa (INS-PA) strains collected from China in 2019. There was no obvious prevalent sequence type or concentrated evolutionary multilocus sequence typing (MLST) type on the INS-PA phylogenetic tree in China. All of the INS-PA isolates harbored β-lactamases with/without other antimicrobial mechanisms, such as gross disruption of oprD and overexpression of efflux genes. Compared with exoU-negative isolates, exoU-positive isolates (25.3%, 46/182) presented higher virulence in A549 cell cytotoxicity assays. The southeast region of China had the highest proportion (52.2%, 24/46) of exoU-positive strains. The most frequent exoU-positive strains belonged to sequence type 463 (ST463) (23.9%, 11/46) and presented multiple resistance mechanisms and higher virulence in the Galleria mellonella infection model. The complex resistance mechanisms in INS-PA and the emergence of ST463 exoU-positive, multidrug-resistant P. aeruginosa strains in southeast China indicated a challenge that might lead to clinical treatment failure and higher mortality. IMPORTANCE This study investigates the resistance mechanisms and distribution and proportions of virulence genes of imipenem-nonsusceptible Pseudomonas aeruginosa (INS-PA) isolates in China in 2019. Harboring PDC and OXA-50-like genes is discovered as the most prevalent resistance mechanism in INS-PA, and the virulence of exoU-positive INS-PA isolates was significantly higher than that of exoU-negative INS-PA isolates. There was an emergence of ST463 exoU-positive INS-PA isolates in Zhejiang, China, most of which presented multidrug resistance and hypervirulence.

(MexAB-OprM, MexCD-OprJ, MexEF-OprN, and MexXY-OprM) are all involved in carbapenem resistance in P. aeruginosa (3). The most common mechanism of imipenem resistance in P. aeruginosa is a combination of chromosomal AmpC production and a porin change. Indeed, a low level of AmpC enzyme production does not result in high-level carbapenem resistance, due to their limited potential to hydrolyze carbapenem drugs. However, their overproduction, together with reduced outer membrane porin permeability and/or efflux pump overexpression, contributes to high-level carbapenem resistance in this pathogen. P. aeruginosa can also obtain other b-lactamases by horizontal genetic transfer, including extended-spectrum b-lactamases (ESBLs), KPC, VIM, and metallo-b-lactamases (MBLs). The combination of these enzymes leads to high rates of carbapenem resistance in P. aeruginosa isolates (4).
Moreover, many Gram-negative bacteria, including P. aeruginosa, possess type III secretion systems (T3SS), which they utilize to introduce virulence factors directly into host cells. In P. aeruginosa, T3SS transport four secreted factors, exoU, exoS, exoY, and exoT. However, all of these factors may not be common to all P. aeruginosa strains. The exoU gene encodes a cytotoxic protein that rapidly destroys the cell membranes of mammalian cells by using its phospholipase activity (5). These virulence factors play important roles that may be involved in the genesis of acute lung injury, bacteremia, sepsis, and invasion of tissues. P. aeruginosa possesses a T3SS virulence mechanism (6).
Both carbapenem resistance and hypervirulence would probably result in poor clinical outcomes for infected patients. Therefore, this research is aimed at investigating the resistance mechanisms and distribution and proportions of virulence genes using whole-genomesequencing techniques.
According to the WGS results, there were 107 multilocus sequence typing (MLST) types detected among the 182 INS-PA isolates, with sequence type 463 (ST463) being the most frequently isolated (11/182), followed by ST244 (9/182), ST235 (7/182), and ST274 (7/182). The isolates did not present a concentrated evolutionary MLST type on the phylogenetic tree (Fig. 1A). There was no obvious prevalent sequence type in China in INS-PA isolates currently. Detailed information (STs, virulence genes, different regions, and infections) on all INS-PA isolates is shown in Fig. 1C.
Antimicrobial susceptibility of P. aeruginosa isolates. The antimicrobial susceptibilities of tested strains to common antibacterial agents are shown in Table 1. For IS-PA isolates, all of the antibacterial agents tested showed .70% susceptibility, except for piperacillin-tazobactam (TZP) and aztreonam (ATM). The rate of susceptibility of IS-PA isolates to aztreonam was 69.1%. Lower rates of susceptibility (,70%) were found for INS-PA isolates for all of the antibacterial agents tested, except for ceftolozane-tazobactam (C/T), amikacin (AMK), and tobramycin (TOB). However, INS-PA isolates from RTIs had higher rates of resistance to various antibacterial agents than INS-PA isolates from other infection sites on a numerical basis.
Virulence of exoU-positive strains. The cytotoxicity assay ( Fig. 4) showed that the cell inhibition rates of exoU-negative strains were significantly lower than those of exoU-positive strains (P , 0.05). Most cell inhibition rates of exoU-positive strains were .92% (the inhibition rate of the hypervirulent control strain, FAHZU24), and some even reached 100%. The cell inhibition rates of exoU-negative strains were mostly ,78% (the inhibition rate of the hypovirulent control strain, ZYPA09), with some even reduced to 0%.
Virulence and resistance analysis of ST463 strains. All of the ST463 strains (n = 11) collected from Zhejiang province were exoU positive. Six of the 11 ST463 strains were collected from intensive care units (ICUs), and the patient ages ranged from 42 to 72 years ( Table 4). Two of the 11 ST463 strains were collected from blood samples, which was significantly higher than for non-ST463 strains (12/171, P , 0.05). All of these ST463 isolates harbored bla OXA-486 , bla PDC-8 combined with PBP3 gene mutation, oprD gross disruption, and efflux pump upregulation, and 10 of them harbored bla KPC-2 ( Table 4). The imipenem MICs of the 11 strains were all .32 mg/mL, presenting high-level resistance to imipenem.

Epidemiology of INS-PA in China
Microbiology Spectrum Most of the ST463 exoU-positive strains collected from Zhejiang presented carbapenem resistance and hypervirulence. However, the evolutionary relationship was not obviously concentrated, based on the phylogenetic tree (Fig. 1B).

DISCUSSION
The resistance mechanisms of INS-PA are complex and associated with several AMR genes or resistance mechanisms. Because all of the isolates in this research were imipenemnonsusceptible strains rather than meropenem-or ertapenem-nonsusceptible isolates, the proportion of porin loss was higher than that of efflux pump upregulation numerically. The rate of metallo-b-lactamase (MBL)-harboring isolates was low, and most of the MBL-harboring isolates were collected from the southwest (4/6).
The sequence type (ST) distribution varied from region to region. ST235 has been reported as the most prevalent sequence type in single-center research conducted in the southwest over a 10-year period (7). It is noteworthy that ST235 was not detected in the southwest in the present study but was found in 4 isolates from the southeast and 3 from the east. ST235 with the hypervirulence gene is the most prevalent sequence type, with clones categorized as high-risk and widespread associated with poor clinical outcomes. In part, this is due to multilevel and high-level antibiotic resistance (8). In the present research, the ST463 strains that were isolated from Zhejiang Province had the highest proportion. Other research from Zhejiang also reported that ST463 with coexistence of exoS and exoU was the most prevalent ST type (9) and presented both multidrug resistance (MDR) and hypervirulence (10). Ten of the 11 ST463 isolates in this research harbored bla KPC-2 , resulting in high MICs for most of the b-lactam antibiotics tested, such as ceftazidime (CAZ) (.32 mg/L), cefepime (FEP) (.32 mg/L), piperacillin-tazobactam (.64 mg/L), ceftolozanetazobactam (.32 mg/L), and meropenem (.32 mg/L, except for one isolate). As there was no obvious evolutionary relationship observed, the ST463 strains collected from the same place tend to be the prevalent type rather than outbreak strains. The clone of ST463 should be noted in clinical practice as high risk, because it contains both the virulence gene and AMR genes. The present study found that there was no prevalent ST clone of INS-PA nationwide. Therefore, INS-PA is not spreading as a resistant clone in China. However, it should be pointed out that the higher proportions of exoU-positive isolates and ST463 strains in the southeast region could indicate that a resistant clone is spreading within certain regions. This finding was different from reported results from other countries, which had prevalent clones of resistant or high-risk P. aeruginosa, such as ST111, ST175, and ST235 (11). For example, ST175 is the most frequent extremely drug-resistant (XDR) high-risk clone detected in Spanish hospitals (12). However, the emergence of ST463 exoU-positive, multidrug-resistant P. aeruginosa strains in east China also indicates a challenge that may lead to failures of clinical treatment and a higher mortality rate. One retrospective cohort study in eastern China found that ST463 was predominant (48.0%) among 50 CRPA BSI cases and that the 28-day mortality was significantly higher for ST463 cases than for non-ST463 cases (66.7% versus 33.3%, P = 0.03) (13, 14). The reason given for the presence of ST463 with poorer outcomes was explained in the publication (14). Infections related to exoU-producing strains have been found to be associated with more severe clinical symptoms and poorer outcomes than infections caused by exoS-positive (exoU-negative) isolates (6). Moreover, drug-resistant P. aeruginosa strains, especially CRPA, contributed to poorer clinical outcomes simultaneously. One meta-analysis demonstrated a .2-fold-increased risk of mortality with multidrug-resistant P. aeruginosa (MDR-PA) (relative risk [RR], 2.34; 95% confidence interval [CI], 1.53 to 3.57) and a prolonged length of hospitalization compared to the risk of mortality and length of hospitalization from infections with susceptible P. aeruginosa strains (15). Thus, exoU and exoS virulence genes coexisting with the bla KPC resistance gene in ST463 CRPA may be an important intrinsic cause of the poor prognosis of clinical P. aeruginosa BSIs (14). Therefore, infections caused by hypervirulent and carbapenem-resistant organisms have clinical and economic consequences. Hospital-acquired resistant and MDR P. aeruginosa   infections will probably result in poorer clinical outcomes, and it will be necessary to rigorously monitor patients in clinical practice. The INS-PA strains were isolated from 13 provinces, which may not represent the whole of China, because both sequencing types and resistance mechanisms are different across regions. The IS-PAs were not tested by WGS, which may make the relationship between genotype and phenotype unreliable.

MATERIALS AND METHODS
Isolates from SMART in 2019. All of the P. aeruginosa isolates were collected during the Study for Monitoring Antimicrobial Resistance Trends (SMART) (16), which isolated pathogens from abdominal, urinary tract, blood, and respiratory tract specimens of patients from 16  Antimicrobial susceptibility testing. Antimicrobial susceptibility testing was performed at the Peking Union Medical College Hospital central laboratory using panels purchased from Thermo Fisher Scientific (Cleveland, OH, USA). MICs were interpreted using the CLSI breakpoints (17) (all antimicrobial agents except colistin [COL]) or the EUCAST breakpoint (colistin) (18).
Cytotoxicity assay. Cytotoxicity assays using A549 human pulmonary adenocarcinoma cells were conducted on all 46 exoU-positive strains and 46 exoU-negative strains selected according to region to evaluate the virulence. The high-cytotoxicity control was the exoU-positive/exoS-negative, ST235 Pseudomonas aeruginosa strain FAHZU24, and the low-cytotoxicity control strain was the exoU-negative/exoS-positive, ST236 strain ZYPA09. The cells were cultured in F-12K medium with 10% fetal bovine serum (FBS) at 37°C with 5% CO 2 . Amounts of 100 mL of fresh medium with 6 Â 10 3 cells/well were plated in 96-well plates and cultured for 24 h. Then, overnight cultures of single colonies from agar plates were diluted 10 3 times (3 Â 10 6 CFU/mL) with F-12K medium (including 10% FBS). One hundred microliters diluted bacterial culture was added into each well (multiplicity of infection of 50) and cultured at 37°C with 5% CO 2 for 3 h. Then, the supernatant was discarded and the cells washed with 100 mL fresh medium. Finally, 100 mL fresh F-12K medium (including 10% FBS) and 10 mL cell counting kit-8 (CCK-8) solution was added. The optical density at 450 nm (OD 450 ) was detected after 2.5 h of culture. Each group had 6 compound wells. The inhibition rate was calculated as follows: (A control 2 A experiment )/(A control 2 A blank ) Â 100, where A is absorbance. A higher inhibition rate indicated stronger cytotoxicity. GraphPad Prism 9 software was employed for statistical analysis, and a P value of ,0.05 was considered significant.
Galleria mellonella larva infection model. The virulence of exoU-positive ST463 strains was also tested by the Galleria mellonella larva infection model. Normal saline was used to adjust the bacterial suspension to 1 Â 10 6 CFU/mL, and 10 mL bacterial suspension was injected into each larva (n = 10 larvae/strain). Then, the larvae were incubated at 35°C and the number of surviving larvae was recorded once every 12 h for 48 h. Strain FAHZU24 is referred to as the hypervirulent control, and ZYPA09 as the hypovirulent control (10).
WGS. All imipenem-nonsusceptible isolates were sent for whole-genome sequencing (WGS). Bacteria cultured to stationary phase from single colonies were pelleted in 1Â Tris-EDTA buffer. DNA isolation used magnetic bead chemistry, and library preparation used the in-house method of Beijing Genomics Institute (BGI, Wuhan, China). Libraries were sequenced on a high-throughput Illumina sequencer in a 2 Â 150-bp paired-end configuration to a calculated coverage depth of Â100. WGS analysis. The CLC Genomics Workbench (Qiagen) was used for WGS analysis. On acquisition of fastq files, reads were trimmed for quality and adapter sequences, sampled to approximately Â100 coverage depth, and assembled de novo for downstream analysis. Genes encoding b-lactamases were identified by screening assemblies using the ResFinder database (https://cge.food.dtu.dk/services/ResFinder/), downloaded on 9 July 2020. Coverage and identity cutoffs were set to $35% and $72%, respectively, though any positively identified antimicrobial resistance gene that was ,100% for either parameter was confirmed by read mapping and/or examination at the amino acid level to identify a b-lactamase variant. Various genes of interest were analyzed for gross disruptions or previously characterized mutations by pairwise alignment to a reference sequence. oprD was tested for permeability, and the PBP3 gene (ftsI) was tested for target mutation. For detecting ampC regulation, ampD, ampDh2, ampDh3, dacB (pbp4), mpl, nuoN, and ampR were analyzed. To detect efflux regulation, nalD, mexR, nalC, and mexZ were analyzed. The exoU gene was tested for hypervirulence (19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29). Multilocus sequence typing determination was performed using CLC Genomics with schema downloaded from Pubmlst.org on 4 November 2020. The phylogenetic tree was constructed based on MLST of each isolate using GrapeTree (version 1.4.0, https://github .com/achtman-lab/GrapeTree) and clustered based on MLST, region, exoU-positive/-negative status, and infection site using the circlize package in R Studio (version 4.0.5).
Statistical analysis. SPSS (version 17.0; IBM) was used for all statistical analysis. Descriptive analysis was performed to calculate the susceptibility and proportion of each resistance mechanism. Student's t test was performed for the cytotoxicity assay for comparison between the cell inhibition rates of the exoU-positive group and the exoU-negative group. The Gehan-Breslow-Wilcoxon test was performed on the survival curves of clinical strains and control strains. Comparison of the resistance mechanisms and clinical characteristics between the exoU-positive and exoU-negative groups was evaluated using the chi-square test. P values of ,0.05 were considered to be statistically significant findings.