Molecular Epidemiology of Antimicrobial Resistance, Virulence and Capsular Serotypes of Carbapenemase-Carrying Klebsiella pneumoniae in China

This study analyzed genomic data of 4643 strains of carbapenemase-carrying Klebsiella pneumoniae (KPN) in China by using the Kleborate software package. The data showed rich diversity in carbapenemase-carrying KPN genomes, which contain not only 152 sequence types but also 90 capsular serotypes. In 2013, the transfer of carbapenemase to hypervirulent Klebsiella pneumoniae (HvKP) of KL1 and KL2 occurred, and since 2014, the propagation of carbapenemase into mammals, poultry, and insects has been detected. The ST11 capsular serotype had a reversal of the prevalence of KL47 and KL64 in 2016, with KL64 replacing KL47 as the dominant serotype. Colibactin is a very suitable indicator to differentiate KL1-type HvKP and classic Klebsiella pneumoniae. The most prevalent yersiniabactin of KL1 is ybt1 ICEKp10, and that of ST11 carbapenem-resistant KPN(ST11-CRKP) is ybt9 ICEKp3. The virulence genes of KL1 carbapenem-resistant hypervirulent KPN (KL1-CRHvKP), as well as ST65- and ST86-type KL2-CRHvKP, were not lost after carbapenemase was obtained.


Introduction
The Klebsiella pneumoniae (KPN) genome has a size of about 5.5 Mbp, encoding 5000~6000 genes, with rich genetic diversity [1]. Classic KPN (CKP) is an opportunistic pathogen that can colonize the human gastrointestinal tract, skin, oral cavity, and nasopharynx [2,3]. CKP usually infects neonates, the elderly, and immunocompromised patients and is the leading cause of hospital-acquired infection and neonatal sepsis [4,5]. In contrast, hypervirulent KPN (HvKP) often infects healthy individuals with normal immunity, causing community-acquired infection. Furthermore, the infections are invasive, including purulent liver abscesses, endophthalmitis, and meningitis [6].
CKP is a key host and transmission vector of clinically important antimicrobial resistance genes [7]. The spread of carbapenem-resistant Klebsiella pneumoniae (CRKP) has become an urgent global public health problem [8,9]. HvKP infection is dominated by two lineages: CG23 and CG65 [10][11][12], with CG23 being the most common. Compared with the rich genetic diversity of CKP, the CG23 core chromosome has limited nucleotide variation, and the acquired virulence genes are highly conserved [13,14]. The enhanced or high virulence is associated with specific capsular serotypes, such as KL1 and KL2, and with accessory genes carried by mobile elements [15], such as genotoxin colibactin, as well as yersiniabactin, aerobactin, and salmochelin that encode the additional siderophore system [16,17]. High virulence is also associated with rmpADC, which is a hypermucus phenotype-related gene [18,19]. KL1-type HvKP acquiring carbapenemase and ST11-CRKP acquiring HvKP virulence genes appeared in Zhejiang Province, China, in 2015 and 2017, respectively [20,21]. The emergence of a fusion strain CRHvKP, with high virulence and carbapenem resistance, has become a significant clinical problem.
The long-term and nationwide molecular epidemiological monitoring data for carbapenemase-carrying KPN in China are limited. However, with the reduction in wholegenome sequencing costs, the National Center for Biotechnology Information (NCBI) has shared public whole-genome data of bacteria from various countries and regions. Furthermore, analyses of antimicrobial resistance, virulence, and mobile elements, as well as the development and promotion of multiple types of software, have facilitated the analysis of these large data sets. In this study, we downloaded NCBI genome data of 37,322 strains of KPN and conducted a molecular epidemiological analysis of antimicrobial resistance, virulence, and capsular serotypes of carbapenemase-carrying KPN in China. This genomic data set represents the largest and most geographically diverse KPN sets in China to date.

Results
As of April 20, 2022, the genomic data for a total of 37,322 strains of KPN have been downloaded. The data for Klebsiella quasipneumoniae and Klebsiella variicola were excluded from the analysis results of Kleborate Species, and the data for 37,264 KPN strains were collected. Through the collection of BioSample information, a total of 6107 KPN strains submitted by China were obtained by screening, of which 4643 were carbapenemasecarrying strains. The strain collection spanned from 2008 to 2022. The strain collection spanned from 2008 to 2022. A total of 3162 strains have provincial distribution information and are distributed in 23 provinces or municipalities directly under the central government. The three regions with the most genomic data were Zhejiang Province (986 strains), Beijing City (572 strains), and Sichuan Province (520 strains). A total of 4428 KPN strains were derived from Homo sapiens and 38 from mammals, poultry, and insects ( Figure 1). We acknowledge the limitations of these genome sequencing data, and it is difficult to avoid selection bias from researchers in the selection of sequencing strains. Therefore, we chose to span a wider range of regions and times and use the phenotype of carbapenem to minimize data bias. There are 152 ST types of carbapenemase-carrying KPN, of which the dominant types are ST11 (3497, 75.3%), ST15 (311, 6.7%), ST2407 (50, 1.1%), ST45 (47, 1.0%), and ST37 (42 0.9%) (Figure 2a). Fifty-nine strains had polymorphism variations of a single housekeeping gene, five strains had gene polymorphism of two housekeeping genes, and one strain had gene polymorphism of three housekeeping genes.  There are 152 ST types of carbapenemase-carrying KPN, of which the dominant types are ST11 (3497, 75.3%), ST15 (311, 6.7%), ST2407 (50, 1.1%), ST45 (47, 1.0%), and ST37 (42, 0.9%) (Figure 2a). Fifty-nine strains had polymorphism variations of a single housekeeping gene, five strains had gene polymorphism of two housekeeping genes, and one strain had gene polymorphism of three housekeeping genes. Observation results from different provinces and municipalities indicated that ST11 was the most common serotype in all regions, and the types of carbapenemase were slightly different in different regions. For example, KPC-2 was the absolute dominant type in Zhejiang Province (933/986, 94.6%); there were a total of 43 OXA-48 strains, of which 38 strains (88.4%) were detected in Beijing; there were 249 NDM-5 strains, of which 111 strains (44.6%) were detected in Sichuan Province. The three regions with more than 500 strains were Zhejiang Province (986 strains), with a KL64:KL47 ratio of 556:225; Beijing (572 strains), with a KL64:KL47 ratio of 105:337; and Sichuan Province (520 strains), with a KL64:KL47:KL25 ratio of 243:65:46.
There were 125 strains of KPN that carried genes related to polymyxin resistance, of which 94 strains carried mgrB mutations of a two-component system, 4 strains carried pmrB mutations of a two-component system, 10 strains carried the mcr-1 gene, of which 1 strain was an imperfect hit (either <100% identity or <100% coverage), 11 strains carried mcr-8, of which 6 strains were imperfect hits, 2 strains carried two mcr-8 imperfect hits, and 4 strains carried mcr-9. A total of 21 strains carried genes related to tigecycline resistance, of which 1 strain carried modification enzyme Tet(X4), and 20 strains carried RND-type pump gene clusters, including 2 strains of tmexCD1-toprJ1 (mcr-8, NDM-5), 16 strains of tmexCD1-toprJ1, and 2 strains carried two tmexCD1-toprJ1 imperfect hits.

Aerobactin
The detection ratio of antimicrobial-sensitive and antimicrobial-resistant KL1 aerobactins nearly had no change, and they were all dominated by the iuc-1 lineage. The proportion of antimicrobial-resistant KL2 iuc-1 was obviously changed compared with that of sensitive KL2, decreasing to 26.1%, 50.0% of which had no iuc-1. In the antimicrobialresistant ST11-CRKP, 48.4% of the strains contained iuc-1, and 41.4% of ST15-CRKP contained iuc-1 (Table 1).

Colibactin
Both antimicrobial-sensitive and antimicrobial-resistant KL1 colibactins were dominated by the clb 2 lineage. Most KL2s, regardless of being antimicrobial-sensitive or antimicrobial-resistant, had no colibactin. The few strains with colibactin had clb 3. Both antimicrobial-sensitive and antimicrobial-resistant CKP were rarely detected with colibactin (Table 1).

Salmochelin
The detection results of salmochelin were quite similar to those of aerobactin. Both antimicrobial-sensitive and antimicrobial-resistant KL1s were dominated by the iro 1 lineage, and the detection ratio of antimicrobial-resistant KL2 iro 1 was obviously decreased, and most of them had no salmochelin. Salmochelin was not detected in most of the antimicrobial-sensitive and antimicrobial-resistant CKP (Table 1).

Discussion
Multidrug-resistant CKP clones are highly diverse, with whole-genome diversity driven by frequent gains and losses of plasmids and phages. Capsular polysaccharide diversity results from chromosomal recombination. The highly virulent clones are seldom subject to chromosomal recombination and therefore have low plasmid and wholegenome diversity [22]. Furthermore, capsular serotypes seldom deviate from KL1, KL2, and KL5 [1]. This may be due to the overproduction of capsular polysaccharides in many HvKP strains interfering with DNA uptake [23]. In this study, the carbapenemasecarrying KPN had 90 capsular serotypes. With the exception of ST11 and ST15, a total of 835 strains were dispersed among 150 ST types. There were also 26 polymorphic variants consisting of one, two, and three housekeeping genes. This indicates that carbapenemases

Discussion
Multidrug-resistant CKP clones are highly diverse, with whole-genome diversity driven by frequent gains and losses of plasmids and phages. Capsular polysaccharide diversity results from chromosomal recombination. The highly virulent clones are seldom subject to chromosomal recombination and therefore have low plasmid and whole-genome diversity [22]. Furthermore, capsular serotypes seldom deviate from KL1, KL2, and KL5 [1]. This may be due to the overproduction of capsular polysaccharides in many HvKP strains interfering with DNA uptake [23]. In this study, the carbapenemase-carrying KPN had 90 capsular serotypes. With the exception of ST11 and ST15, a total of 835 strains were dispersed among 150 ST types. There were also 26 polymorphic variants consisting of one, two, and three housekeeping genes. This indicates that carbapenemases were widely transferred between KPN subtypes. In China, carbapenemase resistance genes have been transferred to community-based hypervirulent KL1 and KL2 HvKP since 2013. KL1 is dominated by KPC-2, and KL2 is dominated by NDM-1, followed by KPC-2. Since 2014, the propagation of carbapenemase into chickens, migratory birds, flies, mice, and pigs has been detected, with metalloenzyme NDM-5 being the most common.
The ST11 capsular serotype had a reversal of the prevalence of KL47 and KL64 in 2016, with KL64 replacing KL47 as the dominant serotype. CRKP KL47 and KL64 prevalence were also reversed in 2016 in bloodstream infection in a tertiary hospital in China, with KL64 gradually replacing KL47 [24]. In this study, the data of carbapenemase-carrying KPN from many regions of China showed that although the proportion of KL47 decreased, it is still the second largest capsular serotype and has not been eliminated.
Colibactin is a genotoxin expressed by the pks genomic island (clbA-clbS). It is associated with DNA double-strand breaks, chromosomal abnormalities, and cell cycle arrest. The survey data of a tertiary university hospital in Egypt showed that KL1 was greatly correlated with the co-carrying of colibactin [25]. A total of 66% of colibactin-positive KPN strains in Taiwan are of KL1 type [26]. The data in this study showed that colibactin can distinguish hypervirulent KL1 from classic KPN, making colibactin a suitable indicator for discriminating between KL1-type HvKP and CKP. In contrast, most KL2 does not carry colibactin. Similar to previous findings [14], KL2 is genetically more diverse than KL1. In this study, of the 6107 strains of KPN from China, a total of 336 strains were KL1, and 187 strains were KL2. KL1 had 287 strains (85.4%) of KL1 were ST23, whereas KL2 had 8 of 11 14 ST types, with the most prevalent ST65 accounting for 59 strains (31.6%). Resistant carbapenemase-carrying KL2 showed a large decrease in aerobactin, salmochelin, and rmpADC compared with sensitive KL2, which was associated with the different ST-type composition ratios of resistant and sensitive KL2. For antimicrobial-resistant KL2 KPN, ST14 and ST25 were more prevalent (23 strains, 54.8%), and ST65 and ST86 were less prevalent (14 strains, 33.3%). In contrast, for sensitive KL2 KPN, ST65 and ST86 were more prevalent (77 strains, 59.7%), and ST14 and ST25 were less prevalent (22 strains,17.1%). ST65 and ST86 carried more virulence genes than ST14 and ST25. The virulence genes of KL1-CRHvKP and ST65 and ST86 KL2-CRHvKP were not lost after carbapenemase was obtained.
Iron acquisition is a key virulence factor for HvKP, the siderophore production of which increased by six-to ten-fold compared with CKP [27]. Among the siderophores of aerobactin, yersiniabactin, and salmochelin, aerobactin is considered the decisive genetic feature and potential antivirulence target of HvKP [28]. However, the data of this study showed that 48.4% of ST11-CRKP and 41.4% of ST15-CRKP also carried the iuc-1 lineage, which was consistent with the KL1 lineage. ICEKp is the most common mobile genetic element in KPN and is usually integrated at an aspartate (Asn) tRNA site. The Ybt locus is often mobilized by ICEKp, and ybt and ICEKp are diverse. Each ICEKp carries a different cargo gene cluster at the right end, and ICEKp10 often carries the clb locus and is associated with ybt lineages 1, 12, and 17 [17]. In this study, the common yersiniabactin in KL1 was ybt1 ICEKp10, the most common yersiniabactin in ST11-CRKP was ybt9 ICEKp3, and KL2 and ST15-CRKP were represented by diverse ybt and ICEKp.
OmpK35 and OmpK36 are the major non-specific porins of KPN, through which nutrients and hydrophilic β-lactam antibiotics can diffuse across the membrane [29,30]. Their most common mutations are premature emergence of stop codons and truncated frameshifts, and strains lacking these two porins have enhanced antimicrobial resistance [31]. The existence of porin deletion, along with carbapenem, is often reflected as high levels of resistance. There are many heterogeneities in the truncation and mutation of OmpK35 and OmpK36. In this study, 4643 carbapenem-carrying KPN strains had 98 types of mutants, 6107 strains of KPN from China had 146 types of mutants, and 37,264 strains of KPN had 651 types of mutants. Although there are many types of mutations, there are only a few prevalent mutants. OmpK35-17% OmpK36GD was the most common (3893/37,264, 10.4%), of which 89.4% (3482/3893) were type ST11. The second most prevalent mutant was OmpK35-25% (3518/37,264, 9.4%), of which 86.0% are ST258. The truncation and mutation of OmpK35 and OmpK36 are spontaneous mutations, and their coexistence with carbapenemase is largely related to the spread of regional dominant clones.
In this study, 79.2% of OXA-48 was detected in Beijing and 44.6% of NDM-5 in Sichuan Province. Unlike other regions, KL47 was the predominant capsular serotype in Beijing. Although the data in this study covered 23 provinces or municipalities across China, the amount of data in each region varied greatly, making it difficult to analyze differences in molecular epidemiology between regions.

Klebsiella pneumoniae Genome-Wide Data
The Klebsiella pneumoniae genome-wide data as of 20 April 2022, were downloaded from the NCBI Assembly database.

BioSample Information Collection
All BioSample information identified as KPN by Species was collected, including collection date, geographic location, host, and isolation source.

Conclusions
Through this large sample molecular epidemiological analysis of carbapenemasecarrying KPN in China, we found that carbapenemases have been widely transferred in CKP subtypes and are infiltrating into KL1, KL2 HvKP, and non-human hosts. The KL1 and KL2 HvKP acquire carbapenemases, which are truly carbapenem-resistant and hypervirulent strains, posing a serious threat to the survival rate of infected hosts. It is imperative to limit transmission among human hosts, non-human hosts, and environments. In addition to antimicrobial resistance, monitoring of high virulence should also be given attention. We must take measures to strengthen the colonization screening of antibioticresistant and/or highly virulent strains in hospitalized patients, conduct regular screening of farm animals and the environment, and rationally use antibiotics when infection is indicated. The definition and screening indicators of high-virulence KPN have been controversial in academic circles. In this large sample analysis, we found that colibactin can be used as an indicator to distinguish KL1 HvKP from CKP.