Molecular Characteristics of Klebsiella pneumoniae Isolates From Outpatients in Sentinel Hospitals, Beijing, China, 2010–2019

Background: Klebsiella pneumoniae is an opportunistic pathogen associated with community-acquired and nosocomial infections. Since 2010, K. pneumoniae testing has been included into an existing diarrhea-syndrome surveillance system for estimating the prevalence of K. pneumoniae in diarrhea-syndrome patients, assessing antibiotic susceptibility, and investigating molecular characteristics of K. pneumoniae. Methods: Klebsiella pneumoniae strains were isolated from stool specimens from diarrhea-syndrome outpatients in Beijing, China. Isolates were tested for antibiotic susceptibility, and phylogenetic relationships were explored though whole genome sequence analysis. Multi-locus sequence type (MLST) alleles were extracted from the whole genome sequence (WGS) data. A maximum likelihood tree was generated by MEGAX. Genomes were annotated by Prokka; core genes were produced by Roary; a maximum likelihood phylogenetic tree was generated using FastTree. Results: Forty-four K. pnuemoniae strains were isolated from 2010 to July 2019; of these 37 were K. pneumoniae and seven were K. variicola. Antibiotic susceptibility testing showed that all 44 strains were sensitive to gentamicin, imipenem, amikacin, meropenem, kanamycin; 97.73% were sensitive to cefoxitin andlavo-ofloxacin; the highest antibiotic resistance rate was 79.55%, which was to ampicillin. We found three extended-spectrum beta-lactamase (ESBL) producing strains; we identified high-virulence ST types, including ST307 and ST65; and we found that ST23 has been the epidemic clone since 2010. MLST and core genome sequence analysis showed two distinct clusters of 44 K. pnuemoniae; 40 alleles were identified in core genome sequence analysis, while 36 alleles were identified in MLST typing. Conclusions: There is an urgent need for epidemiological and molecular studies to understand the dynamics of antibiotic resistance and virulence gene transmission to guide strategies for K. pneumoniae surveillance. WGS analysis provided high discrimination power and reliable and robust data useful for molecular epidemiology.

Since 2010, K. pneumoniae testing has been included in an existing enteric pathogen surveillance system focused on diarrhea-syndrome outpatients of all ages in 245 sentinel hospitals of the 16 districts of Beijing (Lu et al., 2017). The aim of the system is to monitor the prevalence of K. pneumoniae in diarrhea-syndrome outpatients, assess antimicrobial resistance, and explore molecular characteristics of community-acquired K. pneumoniae infection strains.
Isolated K. pneumoniae strains were tested for antibiotic susceptibility, deoxyribonucleic acid (DNA) extraction, wholegenome sequencing (WGS) analysis, and determination of their molecular characteristics.

DNA Extraction and WGS
DNA was extracted by QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). Quantification of extracted genomic DNA (gDNA) was determined on a NanoDrop spectrophotometer, with verification by agarose gel electrophoresis and fluorometric analysis (Qubit2.0).
Multiplexed paired-end libraries (2 × 300 bp) were prepared for DNA sequencing using the NEBNext R Ultra TM DNA Library Prep Kit for Illumina (NEB, USA). Sequences were determined on an Illumina PE150 platform with 100 × coverage at Beijing Novogene technology Co., Ltd.
Raw sequencing data were checked for quality, trimmed, and assembled de novo into contiguous segments using CLC Genomics Workbench version 10.1.1 (CLC, Bio-QIAGEN, Aarhus, Denmark) and SPAdes version 3.13 (Bankevich et al., 2012).
MLST analyses were performed using seven housekeeping genes (gapA, infB, mdh, pgi, phoE, rpob, and tonB) to characterize diversity and epidemiology of K. pneumoniae isolates (Diancourt et al., 2005). WGS data were used to generate MLST assignments for each isolate; unknown STs were sent to the Klebsiella pneumoniae MLST database at the Pasteur Institute (https:// bigsdb.pasteur.fr/klebsiella/klebsiella.html). Genotyping analysis was based on MLST sequences; maximum likelihood trees were generated by MEGA-X (Kumar et al., 2018).

Annotation and Core Genome Analysis
Genomes were annotated by Prokka, a tool for rapid prokaryotic genome annotation (Seemann, 2014). Phylogenetic analyses were produced by Roary, a tool that rapidly builds large-scale pan genomes and identifies core genes (shared by all strains) and accessory genes (Page et al., 2015). A maximum likelihood phylogenetic tree was generated by FastTree version 2.1.10 (Price et al., 2010) to assess relatedness among genomes in the isolated bacteria and to approximate the species tree.

RESULTS
Surveillance led to isolation of 1 to 11 K. pneumoniae strains each year, identifying 44 K. pneumoniae strains from 25,411 stool specimens in 10 years, the detection rate was 0.17% (44/25,411).

MLST Results and MEGA Analysis
MLST of the 44 strains revealed 36 different sequence types (STs), including ST23, which has been detected five times in Beijing in the most recent 10 years, and ST4447, ST4448, ST4449, ST4450, ST4451, and ST4452, which were seen for the first time in the global database. The maximum likelihood tree identified 36 MLST alleles, and 44 strains were disambiguated into two clonal groups: Cluster M1 (containing 7 strains, K. variicola strains) and Cluster M2 (containing 37 strains, K. pneumoniae strains) (Figure 1).

Core Genome Analysis
The whole genome sequence of the 44 strains identified 3,428 core genes. In maximum-likelihood phylogenies trees, these core genome sequences showed 40 allele differences that grouped into two distinct clusters: cluster C1 (containing7 K. varricola strains), and cluster C2 (containing 37 K. pneumoniae strains) (Figure 2).

DISCUSSION
K. pneumoniae has been reported to be a leading cause of hospital associated infections and a common cause of communityacquired infections in many countries (Pendleton et al., 2013;Moradigaravand et al., 2017;Musicha et al., 2017). Beijing outpatient-based diarrhea-syndromes surveillance detected K.
pneumoniae every year since 2010 demonstrating the existence of community-acquired infection caused by K. pneumoniae. Detection of five ST23 strains from 2010 to 2019 further demonstrated that the ST23 strain has persisted in Beijing throughout these years. Our results should alert public health officials since ST23 of K. pneumoniae has well-known virulence and is able to cause severe disease in otherwise healthy individuals (Turton et al., 2007;Brisse et al., 2009;Holt et al., 2015). It typically carries all four acquired siderophore systems as well as rmpA (Brisse et al., 2009). K. pneumoniae ST23 is the most predominant sequence type causing invasive communityacquired infections in Asia (Chung et al., 2012). Surveillance also detected an ST65 strain, which carries colibactin and rmpA (Brisse et al., 2009), and which is associated with lethal infections in humans and marine mammals (Liao et al., 2014). The three community-acquired K. pneumoniae ESBLproducing strains (ST307, ST4452, and ST17) that were identified in the most recent 2-years period provide a significant signal of drug resistance in the population. All three ESBLs producing strains harbor blaCTX-M-15 and blaTEM-1B antibiotic resistance genes. CTX-M-15 belonged to the CTX-M-1 group, and is widespread in east Asia (Bonnet, 2004). The blaCTX-M was first reported in 1990 in a cefotaxime resistant E. coli strain isolated from the fecal flora of a laboratory dog (Bauernfeind et al., 1990). Since then, the CTX-M enzymes have formed a rapidly growing family of ESBLs distributed over wide geographic areas and among a wide range of clinical bacteria, particularly among members of the Enterobacteriaceae family (Bonnet, 2004). Outbreaks have been described in several countries (Yan et al., 2000;Baraniak et al., 2002). Since 1999, CTX-M has been reported to have become the most frequent ESBL in the Enterobacteriaceae in China (Chanawong et al., 2002;Xiong et al., 2002;Wang et al., 2003). Notably, the K. pneumoniae ST307 ESBL-producing strain has a novel lineage with potential to become an epidemic or "high-risk" clone. It has been recognized as a candidate for becoming one of the most clinically-relevant clones since its worldwide emergence during recent years (Villa et al., 2017). The ST307 lineage displays an association with CTX-M-15-and Carbapenemase (KPC)-producing encoding plasmids (Villa et al., 2017). The K. pneumoniae ST307 detected in our study did not harbor the blaKPC gene, but KPC producing factor could be acquired through horizontal plasmids transfer. The ability of this clone lineage to acquire novel genetic features may contribute to its increased persistence in the environment and highlights its potential public health threat of dramatically disseminated multiple drug resistance among bacteria.
MLST and core genomes sequences consistently differentiated 44 K. pneumoniae into 7 K. varricola strains and 37 K. pneumoniae strains. However, the core genome sequences increase discriminatory power for bacterial pathogen subtyping. For example, BJ2013086-S7 strain is very close to BJ2014039-S57 in an MLST molecular phylogenetic tree (see Figure 1), however, BJ2013086-S7 was separated from BJ2014039-S57in the phylogenetic tree generated by the core genome, and was closer to ST23 strain. Similar distinction was made for BJ2013261-S8 and BJ2019005-S58 stains, BJ2019059-S97 and BJ2013082-S34, and BJ2019061-S99 and BJ2014199-S15. Since WGS consists of sequencing chromosome information, both inherited from ancestors and their mutations, in theory, this powerful tool can deduce the chains of potential cross transmission of K. pneumoniae infection (Croucher and Didelot, 2015) and facilitate study of the population structure and pathogen evolution (Bialek-Davenet et al., 2014;Struve et al., 2015;Zhou et al., 2017). However, its discriminatory power relies on reliable, and robust, and long-term WGS data from different geographic areas. It will be valuable to establish a K. pneumoniae identification network for information sharing.
This study suffers two main limitations. First, the sentinel surveillance could have under-estimated the prevalence of K. pneumoniae in diarrhea-syndrome outpatients since K. pneumoniae in most of the circumstance is not the predominant causative-pathogen. Second, lack of comparation with molecular characteristics of hospital-acquired K. pneumoniae infection strains encourages more effort should be made to provide complete molecular spectrum in future studies.

CONCLUSIONS
Outpatient-based diarrhea-syndrome surveillance in Beijing China identified 3 ESBLs-producing strains in 2018 and 2019 that had not been detected previously. We identified high virulence ST types, such as ST307 and ST65, and we showed that ST23 has been the epidemic clone since 2010. There is an urgent need for epidemiological and molecular studies to understand the dynamics of antibiotic resistance and virulence gene transmission to guide strategies for K. pneumoniae surveillance. WGS analysis provides high discrimination power, and reliable and robust data for molecular epidemiology.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

ETHICS STATEMENT
The study was approved by the Ethics Committee of the Beijing Center for Disease Prevention and Control.

AUTHOR CONTRIBUTIONS
BL participated in data analysis and drafted the manuscript. CL and HL managed the bio information analysis. XZ and YH carried out the molecular genetic studies. HY participated in sample isolation. LJ and YT managed the strains and data collection. MQ and QW participated in the design of the study. All authors read and approved the final manuscript.