Nationwide genome surveillance of carbapenem-resistant Pseudomonas aeruginosa in Japan

ABSTRACT Japan is a country with an approximate 10% prevalence rate of carbapenem-resistant Pseudomonas aeruginosa (CRPA). Currently, a comprehensive overview of the genotype and phenotype patterns of CRPA in Japan is lacking. Herein, we conducted genome sequencing and quantitative antimicrobial susceptibility testing for 382 meropenem-resistant CRPA isolates that were collected from 78 hospitals across Japan from 2019 to 2020. CRPA exhibited susceptibility rates of 52.9%, 26.4%, and 88.0% against piperacillin–tazobactam, ciprofloxacin, and amikacin, respectively, whereas 27.7% of CRPA isolates was classified as difficult-to-treat resistance P. aeruginosa. Of the 148 sequence types detected, ST274 (9.7%) was predominant, followed by ST235 (7.6%). The proportion of urine isolates in ST235 was higher than that in other STs (P = 0.0056, χ2 test). Only 4.1% of CRPA isolates carried the carbapenemase genes: blaGES (2) and blaIMP (13). One ST235 isolate carried the novel blaIMP variant blaIMP-98 in the chromosome. Regarding chromosomal mutations, 87.1% of CRPA isolates possessed inactivating or other resistance mutations in oprD, and 28.8% showed mutations in the regulatory genes (mexR, nalC, and nalD) for the MexAB-OprM efflux pump. Additionally, 4.7% of CRPA isolates carried a resistance mutation in the PBP3-encoding gene ftsI. The findings from this study and other surveillance studies collectively demonstrate that CRPA exhibits marked genetic diversity and that its multidrug resistance in Japan is less prevailed than in other regions. This study contributes a valuable data set that addresses a gap in genotype/phenotype information regarding CRPA in the Asia–Pacific region, where the epidemiological background markedly differs between regions.

CRPA isolates originated from nonoverlapping patients in 78 hospitals (the number of collected isolates per island was as follows: n = 30 from Hokkaido Island, n = 307 from Honshu Island, n = 7 from Shikoku Island, n = 32 from Kyusyu Island, and n = 6 from Okinawa Island).Source specimens were classified into one of the following JANIS categories (19,20): Oral/Endotracheal/Respiratory (n = 189), Urinary/Genital (n = 76), Blood/Fluid (n = 40), Digestive (n = 27), and Other (n = 50).Table S1 provides more detailed source categories and metadata for each isolate.Draft genome sequences of 382 CRPA strains were determined to identify the sequence type (ST), resistance genes, and mutations.

Antimicrobial susceptibility profiles and STs of CRPA in Japan
Table 1 summarizes the susceptibilities for antipseudomonal agents.Although the collected CRPA isolates were often nonsusceptible to antipseudomonal agents other than meropenem, the percentage of susceptible isolates remained 52.9% for piperacillintazobactam, 26.4% for ciprofloxacin, and 88.0% for amikacin based on breakpoints in CLSI M100-ed 32 in 2022 (Table 1).When focused on isolates from urine (CLSI M100-ed 33 in 2023), the proportion of amikacin-susceptible isolates was decreased to 75%.
Susceptibility data for individual isolates are provided in Table S1.Among the CRPA isolates, 27.7% (106/382) was deemed difficult-to-treat resistance P. aeruginosa (DTR-P.aeruginosa) in Tamma et al. 's definition (24), and 85.8% (328/382) was MDR in Magiorakos et al. 's definition (25).The susceptibility profiles of DTR strains are summarized in Table S2.Based on the CLSI M100-ed 32 breakpoints, tobramycin (80.2%) exhibited the highest susceptibility rate of DTR isolates among the three aminoglycosides analyzed.Only three CRPA isolates were resistant to colistin; two of the three colistin-resistant isolates (Table a Unless mentioned, criteria for S/I/R are identical between CLSI M100-Ed32 (2022) and M100-Ed33 (2023).-, Colistin susceptibility of P. aeruginosa is classified into either I or R in CLSI.
b Amikacin MIC against urine-associated isolates only (n = 72, linked to the source categories: urine, urine collected by catheter, urine obtained from indwelling catheter, catheterized urine, and urethral discharge in Table S1).
S1) were however DTR-P.aeruginosa and elicited resistance to at least one agent from each of the seven antimicrobial agent categories listed in Table 1.
Furthermore, we identified all the top 10 global high-risk clones (ST111, ST175, ST233, ST235, ST244, ST277, ST298, ST308, ST357, and ST654) (28), which collectively accounted for 17.0% of the total CRPA isolates in Japan (Table S1).Reyes et al. (8) proposed globally most common CGs (10 CGs in Fig. 1B).Those globally common CGs comprised 34.8%.Despite being among the top two most common CGs in the USA and Central South America, CG111 remains a minor CG in Japan (Fig. 1B).Surprisingly, the dominant clone CG463 (ST463), prevalent in Chinese clinical settings (7,29), was entirely absent in the 382 CRPA isolates collected in this study.The characteristic features of two major CG/STs in Japan, ST274 and ST235, are described in more detail with genotype information later in the result section.

bla IMP-98
IMP-98 differed from IMP-1 by only one site at the nucleotide level and likely originated from the bla IMP-1 gene by C637T nonsynonymous substitution, leading to P213S amino acid change (Fig. 3A).To deduce the origin of the bla IMP-98 region, we determined the complete genome sequence of the ST235 strain, JBBCAEG-19-0032, carrying bla IMP-98 using long reads.This revealed that strain JBBCAEG-19-0032 genome consists of one chromosome, and bla IMP-98 is embedded in a class 1 integron as the upstream region of bla IMP-98 contained specific motifs of attI1 (5′-GTTATGGAGCAGCAACGATGTTACGCAGCA GGGCAGTCGCCCTAAAACAAAGTTAGGC-3′: 7-bp core sites are underlined) (31).
A comprehensive analysis of the bla IMP-98 context using the ISFinder database (29) revealed that the class 1 integron is situated within a Tn402-related transposon, which is further nested within a TnAs3-related transposon.The archetype TnAs3 was originally identified in the genus Aeromonas (Fig. 3B) (35,36).However, the integron structure differs between TnAs3 and the TnAs3-related transposon in JBBCAEG-19-0032 in the gene cassette array composition and the absence of conserved attC/attI motifs of class 1 integron in TnAs3 (Fig. 3C).
Both transposon units in JBBCAEG-19-0032 possess intact terminal inverted repeats and are flanked by 5-bp target site duplication: CCCTG for the Tn402-related transpo son and TTATA for the TnAs3-related transposon, indicating their insertions through transposition events.Further comparison of the JBBCAEG-19-0032 chromosome with the PAO1 chromosome revealed that the TnAs3-related transposon is nested within a genomic island containing distant homologs of core genes of a recently defined mobile DNA element, the strand-biased circularizing integrative element (SE) (32,34,37) (Fig. 3B and D).The products of the SE core genes in JBBCAEG-19-0032 (IntA, encoding tyrosine recombinase; Tfp, tyrosine recombinase fold protein; IntB, large tyrosine recombinase; Srap, SE-associated recombination auxiliary protein) exhibited low identities (22.6%-28.5%) in BLASTp comparisons to homologs of a characterized SE, SE-6945, from the genus Vibrio (Fig. 3D) (34).However, IntA and IntB homologs of JBBCAEG-19-0032 possess the catalytic RHRY motif of the tyrosine recombinase (38)(Fig.S1), whereas Tfp and Srap homologs share secondary structures with known Vibrio homologs (Fig. S2 and S3).The in silico removal of a 37,833-bp segment from the JBBCAEG-19--0032 chromosome, with a putative 6-bp footprint normally formed on the right side of SE after SE integration (32,34,37), regenerated an intact form of the PA2049-equivalent gene encoding a hypothetical protein (Fig. 3E).This suggests that the 38-kb region was inserted via site-specific recombination.Hence, we designated the 37.8-kb insert as putative SE, SE-PaeJB0032.A BLASTn search against the NCBI-nr/nt database (2024-02-07) and a database containing 382 Japanese CRPA strains using the left end (intA region) of the SE as a query did not yield significant hits, indicating that SE-PaeJB0032 may have recently been integrated into the P. aeruginosa genome.
higher against E. coli DH5α carrying bla IMP-98 than the control strain, confirming that IMP-98 retains the wide substrate range of MBL.S1).Three colistin-resistant isolates did not possess any known colistin resistance mutations or genes.

Features of two major STs in Japan
Specifically, we aim to determine whether notable phenotypic or genotypic features are associated with ST274 (n = 37) and ST235 (n = 29).Table 3 presents the observation  a More specific source categories are shown in Table S1.
The possession rates of chromosomal mutations in oprD, mexAB regulator genes, and ftsI did not differ among STs (Table 4).However, the possession rates of carbapenemase genes bla IMP and bla GES differed significantly among STs.The possession rates of bla IMP and bla GES in ST235 were significantly higher than that in other STs: 34.5% in ST235 vs 1.7% in other STs (P = 1.4 × 10 −15 ).This trend was also observed for the possession rate of amikacin resistance genes: 58.6% in ST235 vs 2.0% in other STs (P < 2.2 × 10 −16 ).These characteristic features of ST235 were absent in ST274.Fluoroquinolone resistance determinants can be categorized into chromosomal resistance mutations (gyrA, gryB, parC, and parD) and transferrable quinolone resistance gene crpP (41).Although both types appeared at different proportions among STs (Table 4), the distribution trend differed between the two types.The proportion of isolates carrying gyr or par mutations was higher in the two major STs (75.4%, 49/65) than in others (28.5%, 90/316) (P = 5.6 × 10 −15 ), while the proportion of isolates carrying crpP was higher in nonST274 and nonST235 (63.0%, 199/316) than the two STs (34.8%, 23/66) (P = 4.6 × 10 −5 ).Surprisingly, crpP was present in 57.4% (58/101) of ciprofloxacin-susceptible isolates.Thus, its significance should be interpreted with caution.These observations validate the notion that ST235 exhibits an extraordinary resistance phenotype among STs, functioning as an epidemic high-risk clone (28).

DISCUSSION
Since the adoption of a global action plan on antimicrobial resistance at the WHO conference, regional surveillance on antimicrobial-resistant bacterial pathogens has commenced worldwide (8,(15)(16)(17)23).However, in the case of CRPA, most surveillance projects have not provided both MIC values of antimicrobial agents and genome sequences for the collected strains simultaneously, limiting data interpretation and reuse.Furthermore, comprehensive surveillance in a single country with a 10% prevalence rate of CRPA, such as Japan, has been lacking until this study.This study was aimed at addressing these gaps.
The ciprofloxacin and amikacin susceptibility rates regarding CRPA were 26.4% and 88.0%, respectively, in Japan.These values exceed 20.6% and 60.6% of pooled CRPA values reported in the Asia-Pacific region (CLSI breakpoint criteria before 2023) (16) and are then equivalent to and higher than 26.2% and 63.4% of the global CRPA population, respectively (17).Furthermore, the piperacillin-tazobactam susceptibility rate in Japan was 52.9%, which contrasts with the 25.3% susceptibility rate in the global CRPA population (17).This difference might reflect the difference in the carbapenemase possession rate in CRPA between Japan and other regions.Therefore, in Japan, MDR has not yet heavily prevailed in the P. aeruginosa population in clinical settings compared with other Asia-Pacific regions or the world average.Notably, fluoroquinolone resistance among CRPA is a global problem and is common even in non-high-risk clones.
Since several countries/regions normally categorized in the Asia-Pacific region are geographically isolated by the presence of sea, we expected to observe distinct features of the CRPA genotype in Japan and other Asia-Pacific regions.The striking feature was the predominance of CG463/ST463 in China (8) and its absence in Japan (Fig. 1B).Among the carbapenemase types, IMP was the most frequent (87.5%; 14/16) in Japan.This trend contrasts with the predominance of NDM in P. aeruginosa borne-carbapenemases in India (16), KPC in China (8), and VIM in Europe, Africa/Middle East, and Latin America (17).As a country with an average prevalence rate of meropenem-resistant CRPA, Japan was expected to have a low carbapenemase possession rate among P. aeruginosa strains.This study reveals that the carbapenemase possession rate in Japan is 4.2%, which is among the lowest compared with other regions where surveillance studies have been conducted.For instance, the carbapenemase possession rate of meropenem-resistant CRPA was 32.9% in the Asia-Pacific region from 2015 to 2019 (16), 1.9% for the USA in 2018-2019 (8), 69% for South and Central America in 2018-2019 (8), 32% for China in 2018-2019 (8), 57% in Singapore-Australia in 2018-2019 (8), and 21.9% for Europe in 2017-2019 (17).
The prevalence rate of carbapenemase is influenced by genetic, demographic, and environmental factors.The genetic factors include the type of STs constituting the P. aeruginosa population.As shown in Table 4, only so-called high-risk clones, represented by ST235, seem to have a trend toward capturing transferrable carbapenemase.This could be associated with the number of integrons possessed by the STs, as most reported MBL genes other than NDM are present as integron gene cassettes (28).In Japan, the proportion of high-risk clones is 17%, which is lower than that in China (30%, including KPC-producing ST463) and Singapore [60% in Reyes's surveillance (8)].Therefore, potential carbapenemase gene recipients might be fewer in Japan than in other regions.The demographic factor includes the low migration rate of carbape nemase-producing CRPA to hospitals in Japan from high-carbapenemase prevalence countries/regions.However, this factor is challenging to evaluate based on publicly available information currently.Environmental factors include the hospital's environmen tal condition that can prevent outbreaks and the resulting clonal expansion of carba penemase-producing CRPA within the country.The low occurrence of outbreaks is supported by the low frequency (7.6%) of nearly identical isolates in the 382 CRPA data set, as opposed to 19.2% in the Chinese CRPA data set.Previously, VIM-producing CRPA isolates were found in hospitals in Japan (42)(43)(44); however, they were not detected in the 78 hospitals participating in the JARBS from 2019 to 2020.Furthermore, according to the National Epidemiological Surveillance of Infectious Diseases by the Ministry of Health, Labor, and Welfare of Japan, the number of amikacin-nonsusceptible, carbapenem, and fluoroquinolone-resistant P. aeruginosa has been decreasing over a decade in Japan (45).Therefore, it is speculated that hospital environments are being managed to reduce outbreaks and horizontal transmission of carbapenemases in Japan.
The complete sequence determination of a bla IMP-98 -carrying strain revealed that the bla-containing integron was embedded in a Tn3-family transposon of Aeromonas origin and nested in a putative SE region in the chromosome, rather than in the plasmid.Previous studies have identified Pseudomonas as one of the top five representative host genera of SEs; however, to date, there is only one terminus-delineated SE from P. aeruginosa, SE-PaeBT2436, carrying the tmexCD-topJ cluster conferring tetracycline and tigecycline resistance (32).Since homologs of SE core genes could not be detected by nucleotide-based search in either the current NCBI database or the CRPA data set, the SE-PaeJB0032 type might have formed only recently, potentially capturing bla IMP-98 , in other taxa such as Aeromonas, and then jumping into the chromosome of the unique ST235 clone.
This surveillance revealed that two STs, ST274 and ST235, members of the top 10 globally most common CGs in CRPA, were also the most common in Japan.ST274 and CG274 members have frequently been detected in clinical settings worldwide (46) and are known for chronic infection and intrabody evolution involving MDR in patients with cystic fibrosis in Europe (47,48).Whether the ST274 isolates collected in this study are associated with a chronic lung infection remains unknown, but the "Oral/Endotra cheal/Respiratory" source category was most frequent in ST274 (Table 4).However, the proportion of this category did not significantly differ from other STs (P = 0.078).Conversely, the other major ST, ST235, has more striking features than ST274 in isolation source, nonsusceptibility, and resistance gene possession rate (Tables 3 and 4).ST235 was proposed as an epidemic high-risk clone due to its worldwide occurrence with an extremely drug-resistant phenotype (28,46).However, statistical data supporting this notion in a single surveillance has been lacking.Our data demonstrate that ST235 has a trend to acquire resistance genes and resistance phenotypes more frequently than other STs.The association of ST235 with urine is a novel finding, but the underlying mechanisms remain to be investigated.
Inactivating mutations in oprD were the most frequent meropenem resistance mechanism (found in 87.1%), as noted in previous genome surveillance performed in other regions, where 69% of meropenem-resistant CRPA possessed oprD-inactivating mutations and 22% possessed acquired carbapenemase (8).We show that in Japan, not only oprD-inactivating mutations but also efflux pump regulator inactivation and ftsI resistance mutations are more common carbapenem resistance mechanisms than the horizontal acquisition of carbapenemases.Therefore, future surveillance of P. aeruginosa should prioritize the study of chromosomal mutations.Surprisingly, 9.2% (35/382) of the CRPA isolates did not possess apparent inactivat ing mutations, known ftsI resistance mutations, or carbapenemase.A recent multicen ter study (9) reported that 14% of modified carbapenemase inactivation method (mCIM)-positive CRPA, constituting 4.6% of the total examined CRPA, did not possess known carbapenemases.Therefore, a fraction of the 9.2% of Japanese CRPA without a known resistance gene or mutation may harbor enzymes capable of partially degrading meropenem.The potentially relevant enzymes include variants of β-lactamase such as PDC and OXA (detected variants are listed in Table S1), as well as Zn 2+ -dependent imipenemase (49).However, pinpointing the causal loci of meropenem resistance in the 9.2% of Japanese CRPA is challenging with the current knowledge, as these variants have been poorly characterized to date.Furthermore, known resistance mutations could not account for the mechanisms of amikacin nonsusceptibility in 45.7% of amikacin-non susceptible isolates.These CRPA isolates must possess unreported resistance genes or mutations in their genomes.A comparative genomics approach is required to clarify this claim in the future to fully understand and control CRPA spread.
Limitations of this research include the inability to evaluate whether CRPA isolates are associated with colonization or infection (the cause of disease) in patients when collecting isolates from hospitals and the lack of investigation into the effectiveness of new antimicrobial agents such as cefiderocol, ceftolozane-tazobactam, and ceftazi dime-avibactam against CRPA isolates carrying chromosomal resistance mutations and carbapenemases.Further surveillance studies will require new study designs to address these limitations.

Conclusions
The first nationwide surveillance for CRPA in Japan demonstrates Japan to be a lowprevalence country for carbapenemase-producing CRPA.Although the CRPA largely consists of many minor STs in Japan, the presence of globally common CG/ST and all top 10 high-risk clones was noted.This underscores the need for continued and comprehen sive surveillance.A high-risk clone ST235 has a distinguished feature among STs to become DTR relatively easily and capture carbapenemase and aminoglycoside resistance genes.The carbapenem resistance mechanisms of CRPA cannot be fully explained by known resistance genes and mutations.The data set built in this study provides the national epidemiological features and serves as a foundation for future studies on global epidemiology, therapeutic options, and infection control.

Strains, media, and plasmids
CRPA candidates participating in our genome surveillance program [the Japan Antimi crobial-Resistant Bacterial Surveillance (JARBS)] and associated with isolation source and date information were transferred from hospitals to the NIID AMR center.As part of this program, hospitals were requested to transfer all isolates showing an imipenem or meropenem MIC ≥ 8 in the initial test, limited to one isolate per patient.A total of 668 viable isolates were received at the AMR-RC between 2019 and 2020.Since the JARBS-PA project commenced in mid-2019, the data set primarily consists of isolates identified in 2020.Subsequently, 273 isolates exhibiting a meropenem MIC < 8 upon remeasurements at AMR-RC were excluded from the data set to align with the definition of CRPA used in other surveillance studies (8,16,17).Additionally, three isolates lacked linked isolation source data, and 10 isolates with poor draft genome assembly quality (L90 > 150) were excluded.The final CRPA data set comprises 382 isolates with linked isolation source data.
Single-colony isolation was performed if necessary, using BD BBL BTB Lactose-Con tained Agar Medium (also called Drigalski agar) (Becton, Dickinson and Company, Franklin Lakes, NJ, USA).BD BBL Muller-Hinton II (Cation adjusted) Broth (Becton, Dickinson and Company) was used to culture P. aeruginosa strains in liquid.Table S1 contains the P. aeruginosa strains and associated metadata.

Antimicrobial susceptibility test
To obtain MICs of antipseudomonal agents, CRPA strains were propagated on Drigalski agar and incubated for 1 day at 35°C.Then, a single colony was inoculated into a MicroScan Neg MIC NF-1J panel using Prompt Inoculation System D (Beckman Coulter Inc., Brea, CA, USA).Inoculated panels were incubated at 35°C for 18 h in MicroScan Walkway40 Plus (Beckman Coulter Inc.) or MicroScan Walkway DxM 1096 (Beckman Coulter Inc.).To cover a therapeutically achievable range, the antimicrobial susceptibility testing was designed to measure major antimicrobials using four dilutions for amikacin and colistin and five dilutions for other antimicrobials.

Genome sequencing
Draft genome was determined for all 382 CRPA isolates.For this purpose, P. aeruginosa strains were cultured in Muller-Hinton II broth overnight at 35°C.Cells were collected in 1.5-mL tubes and resuspended in a lysis solution containing lysozyme and RNaseA and then incubated at 37°C for 1 h.After adding proteinase K and SDS, the reaction mixture was incubated at 55°C overnight.Genomic DNA was purified from the cell lysates using Agencourt AMPure XP beads (Beckman Coulter Inc.) following the protocol recommended by the manufacturer.NGS libraries were constructed using Enzymatics 5 × WGS fragmentation mix and WGS ligase reagents (Qiagen, Hilden, Germany), and then, libraries were sequenced on a HiSeq X Five platform in Macrogen Japan (Tokyo, Japan).Low-quality reads were removed from the original reads using fastp (54), after which the trimmed reads were assembled using the Shovill pipeline (https://github.com/tseemann/shovill/tree/master) set to default options.Draft genome assemblies fulfilled L90 < 150.
To determine the complete genome sequence of strain JBBCAEG-19-0032, genomic DNA was extracted using the Monarch HMW DNA Extraction Kit for Tissues (New England Biolabs, Ipswitch, MA, USA).Long-read sequencing was performed on the GridION platform (Oxford Nanopore Technologies, Oxford, UK).The library was prepared using the Rapid Barcoding Kit, and sequencing was performed using the R9.4.1 flow cell (Oxford Nanopore Technologies).Basecalling was performed using the high-accuracy base calling model of Guppy v5.0.12 (Oxford Nanopore Technologies).The complete sequence of JBBCAEG-19-0032 was obtained through long-read-only assembly using Flye v. 2.9-b1768, set to the "--nano-raw" option (55), followed by polishing with Illumina reads using Pilon v. 1.22 (56).

ETHICS APPROVAL
This study (JARBS) was reviewed and approved by the Institutional Review Board of the National Institute of Infectious Diseases (approval number 1251).The use of JANIS data was approved by the Ministry of Health, Labor and Welfare of Japan (approval number 1553).

ADDITIONAL FILES
The following material is available online.

FIG 1 (
FIG 1 (A) Minimum spanning tree of 382 meropenem-resistant CRPA isolates based on MLST profiles.Each node represents ST, otherwise clonal group (CG).STs with one allele difference were collapsed into one node as CG.Node size is proportional to the number of isolates.The number on the branch denotes the number of different alleles at the 7 MLST loci between two nodes.(B) Proportions of 10 globally most common CGs in Japan and other regions.Proportions of CGs in the USA, China, and South and Central America are based on Reyes et al. 's report (8).

FIG 2 6 FIG 3
FIG 2 Venn diagram showing the simultaneous occurrence of resistance mutations in meropenem-resistant P. aeruginosa.Numbers in circles represent the numbers of isolates having an acquired carbapenemase, a point mutation registered in AMRFinderPlus (30), or inactivating mutations.The "Mex" category indicates the presence of a mutation in at least one of three loci (mexR, nalC, and nalD) encoding a negative regulator for the mexAB operon.The "PBP3" category indicates the presence of a mutation in ftsI.The acquired carbapenemase/MBLs detected are GES-5, IMP-1, IMP-10, IMP-7, IMP-34, and IMP-98.

TABLE 3
Isolation sources observed in two major STs a

TABLE 4
Observation frequency of specific resistance phenotype, genes, or mutations in two major ST groups a χ 2 test P value after multiple testing corrections (false discovery rate) by the Benjamini and Hochberg method.The independence of two factors (phenotype/genotype and ST) was evaluated by using a two by three contingency table (df = 2) where the rows are the count of qualified isolates and the count of nonqualified isolates.DTR, difficult-to-treat resistance.b Breakpoint criteria follows CLSI M100-ed 32 (2022).c Acquired mutation in conserved genes.