Genetic Diversity of Polymyxin-Resistance Mechanisms in Clinical Isolates of Carbapenem-Resistant Klebsiella pneumoniae: a Multicenter Study in China

Polymyxin-resistant CRKP is a serious public health threat whose resistance mechanisms should be under continuous surveillance. Here, we collected 662 nonduplicate CRKP strains across China to identify the carbapenemase and polymyxin resistance genes and epidemiological features. ABSTRACT Polymyxin has been the last resort to treat multidrug-resistant Klebsiella pneumonia. However, recent studies have revealed that polymyxin-resistant carbapenem-resistant Klebsiella pneumonia (PR-CRKP) emerged due to the mutations in chromosomal genes or the plasmid-harboring mcr gene, leading to lipopolysaccharide modification or efflux of polymyxin through pumps. Further surveillance was required. In the present study we collected PR-CRKP strains from 8 hospitals in 6 provinces/cities across China to identify the carbapenemase and polymyxin resistance genes and epidemiological features by whole-genome sequencing (WGS). The broth microdilution method (BMD) was performed to determine the MIC of polymyxin. Of 662 nonduplicate CRKP strains, 15.26% (101/662) were defined as PR-CRKP; 10 (9.90%) were confirmed as Klebsiella quasipneumoniae by WGS. The strains were further classified into 21 individual sequence types (STs) by using multilocus sequence typing (MLST), with ST11 being prevalent (68/101, 67.33%). Five carbapenemase types were identified among 92 CR-PRKP, blaKPC-2 (66.67%), blaNDM-1 (16.83%), blaNDM-5 (0.99%), blaIMP-4 (4.95%), and blaIMP-38 (0.99%). Notably, 2 PR-CRKP strains harbored both blaKPC-2 and blaNDM-1. The inactivation of mgrB, associated significantly with high-level polymyxin resistance, was mainly caused by the insertion sequence (IS) insertion (62.96%, 17/27). Furthermore, acrR was inserted coincidently by ISkpn26 (67/101, 66.33%). The deletion or splicing mutations of crrCAB were significantly associated with ST11 and KL47 (capsule locus types), and diverse mutations of the ramR gene were identified. Only one strain carried the mcr gene. In summary, the high IS-inserted mgrB inactivation, the close relationship between ST11 and the deletion or splicing mutations of the crrCAB, and the specific features of PR-K. quasipneumoniae constituted notable features of our PR-CRKP strains in China. IMPORTANCE Polymyxin-resistant CRKP is a serious public health threat whose resistance mechanisms should be under continuous surveillance. Here, we collected 662 nonduplicate CRKP strains across China to identify the carbapenemase and polymyxin resistance genes and epidemiological features. Polymyxin resistance mechanism in 101 PR-CRKP strains in China were also investigated, 9.8% of which (10/101) were K. quasipneumoniae, as determined via WGS, and inactivation of mgrB remained the most crucial polymyxin resistance mechanism, significantly related to high-level resistance. Deletion or splicing mutations of crrCAB were significantly associated with ST11 and KL47. Diverse mutations of the ramR gene were identified. The plasmid complementation experiment and mRNA expression analysis further confirmed that the mgrB promoter and ramR played a critical role in polymyxin resistance. This multicenter study contributed to the understanding of antibiotic resistance forms in China.

PR-CRKP strains in China were also investigated, 9.8% of which (10/101) were K. quasipneumoniae, as determined via WGS, and inactivation of mgrB remained the most crucial polymyxin resistance mechanism, significantly related to high-level resistance. Deletion or splicing mutations of crrCAB were significantly associated with ST11 and KL47. Diverse mutations of the ramR gene were identified. The plasmid complementation experiment and mRNA expression analysis further confirmed that the mgrB promoter and ramR played a critical role in polymyxin resistance. This multicenter study contributed to the understanding of antibiotic resistance forms in China.
KEYWORDS polymyxin resistance mechanism, carbapenemase-resistant Klebsiella pneumoniae, antibiotic resistance, mgrB inactivation, K. quasipneumoniae, genetic diversity, multicenter study, polymyxin-resistant Klebsiella pneumoniae C arbapenem-resistant Klebsiella pneumoniae (CRKP) was highlighted as a notorious pathogen of various infections (1). Polymyxin (polymyxin B and colistin) has been the last-resort treatment for such infections, even with an unfavorable side effect profile, including nephrotoxicity and neurotoxicity (2). However, the emergence of polymyxin resistance has been documented in the setting of increased use and imposed negative impacts on its usage in CRKP treatment (2,3).
Moreover, Klebsiella quasipneumoniae was a newly designated species in Klebsiella spp. Few reports described its polymyxin-and carbapenem-resistance molecular features since it was not generally distinguished from K. pneumoniae by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) and routine biochemical methods in the clinical laboratory (13).
The increasing rate in clinical isolates of polymyxin-and carbapenem-resistant K. pneumoniae (PR-CRKP) warrants further investigation into their epidemiology and underlying molecular mechanisms. Here, we conducted the study aiming to determine the prevalence of PR-CRKP in six provinces/cities in China, their clonal relationships, and the molecular features via whole-genome sequence (WGS) and bioinformatics analysis.
Polymyxin resistance mechanisms. The polymyxin resistance mechanism is shown in Table 2 Table 2. The acrR gene mutation was the most frequently identified (71, 70.29%), followed by mgrB (27, 26.73%), ramR (11, 10.89%), and arnB (7, 6.93%).   Fig. 3. The nucleotide substitution (T-35C) at the promoter region of the mgrB gene (-10 region) was identified in two strains via BPROM. In addition, all 101 PR-CRKP strains were divided into the mgrB-mutation group or not. The Fisher exact test and Mann-Whitney U test showed that the mgrB inactivation was significantly relevant with a higher polymyxin resistance level (P , 0.001).
The efflux pump AcrAB-related genes involved in polymyxin resistance were also evaluated. The ISkpn26-inserted acrR genes, the repressor of the AcrAB, were observed in 67 strains at the same site (1277 nt) (67/101, 66.33%) (Fig. 5C). The PR-CRKP strains were then divided into two groups with or without ISkpn26-mediated insertion of the acrR gene, and it revealed that ST, especially ST11, was significantly associated with this insertion (P , 0.001). Furthermore, one missense mutation (L34F) and two shift mutations (81Ffs, 121Gfs) were also detected in the acrR gene. The ramR gene, the negative regulator of ramA, which can positively regulate efflux pump expression, was the second-mostfrequently mutated gene. The amino acid deletions occurred at amino acids 66 and 111 and were overall missing from amino acids 1 to 38. One stop codon was detected at amino acid 122 (Q122*) of the ramR gene in 2 strains. One missense mutation was found in kpnE, an efflux pump-related multidrug resistance gene. No deleterious mutation was identified among the acrA, acrB, kpnF, ramA, soxS, and soxR genes.
In the genes related to LPS synthesis and modification, the missense mutations were identified in the arnB (6 strains) which was one part of arnBCADTEF operon and yciM (1 strain).
The mcr gene was detected only in strain CJFH07 (mcr-8.2), in which the missense mutation in mgrB and crrB and disruption of acrR also played a major role in polymyxin resistance (14). However, the underlying resistance mechanism of 26 PR-CRKP strains (25.49%) remained elusive.
Complementation experiment and mRNA expression. The complementation experiment of the ramR and mgrB genes was performed for strains CJFH05 and CJFH01, whose resistance only resulted from the mutation of ramR and the insertion of the mgrB promoter region, respectively. The polymyxin MIC of CJFH05 fell from 8 mg/ mL to 1 mg/mL, whereas the MIC of CJFH01 decreased from 256 mg/mL to 0.5 mg/mL after the complementation experiment with the ramR and mgrB promoter, while the MICs of the control strains were the same as those of the original strains. The mRNA assessment is shown in Fig. 6. The expression of the polymyxin resistance genes (crrA/ B/C, pmrA/B/C, and arnT) and efflux pump genes (acrA/B and ramA) reduced to the same level as ATCC 13883 after the plasmid was complemented, suggesting that the mutations in ramR or mgrB played a critical part in polymyxin resistance.
surveillance (10,16,17). Predominant molecular characteristics of PR-CRKP strains may differ geographically and over time (10,16,18,19). Therefore, in the study, the genomic survey of clinical PR-CRKP was conducted to investigate the genetic alterations in these isolates circulating in China.
Most PR-CRKP strains were recovered from the lower respiratory tract, similar to a previous survey in China (20). Furthermore, a considerable number of PR-CRKP strains were isolated from children (22.77%) and elderly patients (41.58%).
K. quasipneumoniae, as a novel Klebsiella sp., was commonly misidentified as K. pneumoniae in the clinical laboratory (21). WGS can be used to distinguish the species, and its distribution may be related to the regions and infection sites (13). However,
The dysfunction of the crrCAB system might mediate the polymyxin resistance (41,42). In the present study, if the whole crrCAB system was deleted, as observed in almost

Polymyxin-Resistant, Carbapenemase-Resistant Klebsiella pneumoniae
Microbiology Spectrum all ST11 K. pneumoniae strains, the polymyxin MIC remained unchanged. The regulatory pathway by which crrAB regulated polymyxin resistance was only via crrC (41,43). Consequently, crrCAB cannot contribute to polymyxin resistance if the whole system is removed. However, the splicing mutations of crrB and crrC may lead to the loss of the crrCAB function. The role of the crrCAB system in the polymyxin resistance of K. pneumoniae requires further in-depth study. In addition, previously unreported novel mutations were identified in our study; crrB (V193G), crrC (P68T), and crrC (107-to 129-amino acid [aa] deletion) were reported. As for other TCS-related genes, the mutation in pmrB (R256G) was identified in our study, but its role in polymyxin resistance was controversial. It was considered a deleterious mutation according to the PROVEAN result (44). However, Sampaio reported that the substitutions in pmrB (T246A and R256G) were not involved in polymyxin resistance (45). Campos also documented that the expression of the pmrAB system with the above-described mutations had no significant influence on polymyxin resistance (15). Therefore, in our study, we considered it a nondeleterious mutation.
The inactivation in the repressors of efflux pump AcrAB genes has been documented to contribute to polymyxin resistance (10,11). The efflux function of AcrAB would be hyperactive if acrR, the repressor of the AcrAB, was inactivated (11). In the present study, the inactivated acrR genes inserted coincidently at 1277 nt by ISkpn26 were present in about 65.69% of PR-CRKP strains, and all of them belonged to ST11. A total of 25.74% (26/101) strains only harbored this acrR gene mutation and showed low-level polymyxin resistance. Similar to our results, Yang et al. demonstrated that ISkpn26 was always inserted into the acrR gene, resulting in acrR inactivation and then acrAB gene overexpression and multidrug resistance, and the insertion was mainly relevant to the ST11-KPC-2-producing CRKP (46). In their study, several K. pneumoniae isolates presenting an ISKpn26 interrupted the acrR gene and caused only a limited alteration of colistin MIC after exposure with 1-(1-naphthylmethyl)-piperazine (NMP), which was an efflux pump inhibitor. It seemed that the truncated acrR played only a small part in polymyxin resistance. Taken together, ST11 CRKP, as the predominate ST type in China, should be highlighted due to the overactivated efflux pump mediated by the ISkpn26-inserted acrR genes. However, the underlying mechanism remains elusive, and further study should be performed.
The ramR gene might repress the expression of ramA, which can regulate other efflux pumps, such as AcrAB, and then LPS modification and eventually mediate the decreased susceptibility to polymyxin in K. pneumoniae (47,48), while in our study, the complementation experiment of the ramR gene was employed, indicating that the mutation of ramR will result in polymyxin resistance. Molecular epidemical studies of the ramR gene in PR-CRKP have rarely been reported, and the present study showed that mutations in the ramR gene were diverse, containing deletion, missense mutation, nonsense mutation, and truncation. This deserves more attention.
The arnBCADTEF operon, lpxM and yciM, mediated LPS biosynthesis and modification (3,15,22). In the deleterious mutations in the above-mentioned genes, the mutations in arnB were prevalent, similar to the results of a previous study (14). However, in the present study, though predicted as deleterious mutations by PROVEAN, the genes lpxM (T229S), lpxM (N7I), and yciM (N212T) were 100% similar to the reference sequences in GenBank (yciM [N212T]: NCBI reference sequence WP_002901781.1; lpxM [T229S] and lpxM [N7I]: NCBI reference sequence: WP_072353980.1) after alignment using the NCBI BLASTp online tool. Consequently, we defined them as nondeleterious variations. Multiple new mutations were identified in various genes in this study, but the significance of these mutations and their contribution to polymyxin resistance remained unclear.
In the current study, only one isolate carried the mcr gene, showing that polymyxin resistance was most associated with chromosome variation. However, the polymyxin resistance mechanisms of the 26 isolates remained unclear, indicating the presence of an alternative mechanism regulating the expression of these genes, which prompts the demand for exploration of polymyxin resistance.
Our study was limited by several factors. First, the majority of the isolates were collected from a few provinces/cities in China. Therefore, the national surveillance program should be further developed. Second, the underlying resistance mechanism of 26 PR-CRKP strains remained unclear, and we need to investigate this further.
In summary, the current study demonstrated the genetic diversity of the underlying polymyxin resistance mechanism of 101 PR-CRKP strains collected in China. Polymyxin resistance in the study is multifaceted and attributed to mutations in chromosomal genes leading to lipopolysaccharide modification or efflux of polymyxin through pumps. With few transmissible mcr genes identified, the prevalence of polymyxin-resistant strains was low in this unit. Furthermore, the high IS-inserted mgrB inactivation, the close relationship between ST11, the deletion or splicing mutations of crrCAB, and specific features of PR-K. quasipneumoniae constituted notable features of our PR-CRKP strains. The current study will help us to understand the diversity of polymyxin resistance in PR-CRKP circulating in China.

MATERIALS AND METHODS
Ethics approval. The use of the K. pneumoniae isolates for research purposes was approved by the ethics committee of the China-Japan Friendship Hospital (2022-KY-054). The written informed consent from participants was exempted, and the privacy of involved subjects was not affected.
Strain collection and antibiotic sensitivity test (AST). A total of 662 unduplicated CRKP strains, confirmed as K. pneumoniae by MALDI-TOF MS (Bruker Daltonics, Germany), were collected from 8 hospitals in 6 provinces/cities across China. Their MIC of carbapenems, polymyxin, and tigecycline were determined by BMD. A MIC of polymyxin of .2 mg/mL was defined as resistance recommended by CLSI M100 guidelines (49), a MIC of tigecycline of .2 mg/mL was defined as nonsusceptible according to the FDA (https://www.fda.gov/). The MIC of other antibiotics, including amikacin, aztreonam, levofloxacin, ciprofloxacin, tobramycin, and sulfamethoxazole/trimethoprim, were determined via the Vitek-2 system with N335 susceptibility cards (bioMérieux, France).
Epidemiological and clinical data. Medical reports were accessed to collect demographic variables for all patients; 46 (6.95%) non-PR-CRKP cases were lost for follow-up, and their data were unavailable. The following data were retrospectively collected: gender, age, and place of residence.
Definition. The strains with a MIC of .16 mg/mL were defined as having high-level polymyxin resistance as previously described (50). Deleterious mutations were defined as the mutations of certain genes predicted to affect protein function using PROVEAN.
WGS. The whole genomes of PR-CRKP strains were sequenced using an Illumina NovaSeq PE150 at the Beijing Novogene Bioinformatics Technology Co., Ltd. Raw reads were filtered to remove low-quality sequences and adaptors. De novo assembly was conducted using SOAP de novo 2.04, SPAdes 3.10.0, and ABySS 1.3.7. The assembly results were integrated with CISA 1.3 software. The gap in preliminary assembly results was filled using Gapcloser 1.12.
Identification of mutations in proven or putative polymyxin resistance-related genes in K. pneumoniae. The mutations in the mgrB gene and its promoter region were identified using NCBI-blast1 2.11.0 with K. pneumoniae MGH 78578 (GenBank accession number CP000647) and K. quasipneumoniae strain ATCC 700603 (GenBank accession number CP014696) as reference strains.
Phylogenetic reconstruction. For the phylogenetic analysis, high-quality single-nucleotide polymorphisms (SNPs) were obtained using snippy 3.2-dev against MGH 78578. The alignment file was filtered from variants with elevated densities of base substitutions as a putative recombination event using Gubbins 2.4.1. Then snp-sites 2.5.1 was used to reduce the filtered alignment to the core polymorphic sites. Next, the core alignment output was used to create a randomized accelerated maximum likelihood (RAxML) tree with RAxML 8.2.12 under the general time reversible (GTR)-Gamma model. The tree layout was graphically edited using iTOL 5.6.
Statistical analysis. The differences between the ST type and crrCAB, the KL type and crrCAB, and the resistance level and related genes were evaluated via the x 2 test or Fisher's exact test for categorical variables and the Mann-Whitney U test or continuous variables, as appropriate, using the IBM SPSS Statistics software packages 22.0 (SPSS, Inc., Chicago, IL). A P value of ,0.05 was considered statistically significant.
Complementation experiment. Apramycin-resistant pDK6 was used as the vector as previously described, but with modifications (51). The ramR and mgrB sequences were amplified by PCR via PrimeSTAR Max DNA polymerase (TaKaRa, Dalian) with primers ramR-F/R (forward [F]: AGCCAAGCTTGCATGCCTGCA GCCGGGTTCGATACTGCGATAA, reverse [R]: ACAGGAAACAGAATTCGAGCTCCGGCATTAAGAACAAACGGCA) and mgrB-F/R (F: AGCCAAGCTTGCATGCCTGCAGAGCCAGCGATGCCAGATTTA, R: ACAGGAAACAGAATTCG AGCTCCGCCAATCCATAAGATAGCCAC). The plasmid was linearized and PCR products were digested using the double-digestion method. Then, T4 DNA ligase (NEB, USA) was selected for ligating the vector at 16°C overnight, and the plasmid was transformed into CJFH05 and CJFH01 and then inoculated on 30 mg/mL apramycin-containing LB ager. The MICs of polymyxin were determined. The control strains were CJFH05 and CJFH01 with empty vector pDK6 transformed named CJFH01-pDK6 and CJFH05-pDK6, respectively. mRNA extraction and expression assessment using digital droplet PCR (ddPCR). Total RNA was extracted from strains CJFH01-pDK6, CJFH01-mgrB, CJFH05-pDK6, and CJFH05-ramR using the RNeasy minikit (Qiagen) according to the manufacturer's instructions and reverse-transcribed into cDNA via a RevertAid first-strand cDNA synthesis kit with DNase I (Thermo Fisher Scientific, USA). The ddPCR was performed following the manufacturer's protocol using QX200 ddPCR EvaGreen supermix (Bio-Rad, USA), a QX200 droplet generator (Bio-Rad), a QX200 droplet reader (Bio-Rad), and QuantaSoft software (Bio-Rad). The rpoB gene was used as the internal control. K. pneumonia ATCC 13883 was selected as the control strain. The primers used refer to previous studies (14). The relative expression (RE) of each strain was calculated as follows: Expression ratio ER ð Þ ¼ AVE copy numbers detecting gene =copy numbers rpoB

RE ¼ ER testing strains =ER ATCC 13883
ER is the average of the ratio of the detecting gene to the rpoB gene for each strain. The RE is the ER of detecting strains divided by the ER of ATCC 13883.
Data availability. All genome sequences have been submitted to the National Center for Biotechnology Information (NCBI) database with BioProject accession number PRJNA839599 and BioSample accession numbers SAMN28546397 to SAMN28546497. The original data presented in the study are included in the article. Further inquiries can be directed to the corresponding author.

ACKNOWLEDGMENT
This work was supported by the Chinese Academy of Medical Sciences (CAMS) Innovation Fund for Medical Sciences (grant number CIFMS 2021-I2M-1-030).