High Prevalence of ESBL-Producing Klebsiella pneumoniae Causing Community-Onset Infections in China

The aim of this work was to investigate the epidemiological and genetic characteristics of ESBL-producing Klebsiella pneumoniae (ESBL-Kp) causing community-onset infections. K. pneumoniae isolates were collected from 31 Chinese secondary hospitals between August 2010 and 2011. Genes encoding ESBL and AmpC beta-lactamases were detected by PCR. The isolates were assigned to sequence types (STs) using multi-locus sequence typing (MLST). Eleven ESBL-Kp strains were selected for whole-genome sequencing (WGS) for investigating the genetic environment and plasmids encoding ESBL genes. A total of 578 K. pneumoniae isolates were collected, and 184 (31.8%) carried ESBL genes. The prevalence of ESBL-Kp varied from different geographical areas of China (10.2–50.3%). The three most prevalent ESBL genes were blaCTX-M-14 (n = 74), blaCTX-M-15 (n = 60), and blaCTX-M-3 (n = 40). MLST assigned 127 CTX-M-14 and CTX-M-15 producers to 54 STs, and CC17 was the most prevalent population (12.6%). STs (23, 37, and 86) that were known frequently associated with hypervirulent K. pneumoniae (hvKP) account for 14.1% (18/127). Phylogenetic analysis by concatenating the seven loci of MLST revealed the existence of ESBL-producing K. quasipneumoniae (two strains) and K. varricola (one strain), which was further confirmed by WGS. This study highlights the challenge of community-onset infections caused by ESBL-Kp in China. The prevalence of STs frequently associating with hvKP should be of concern. Surveillance of ESBL-KP causing community-onset infections now appears imperative.


INTRODUCTION
Extended-spectrum β-lactamases (ESBLs) have increased dramatically among clinical Enterobacteriaceae isolates during last three decades. In the last two decades, CTX-M replaces SHV as the major type of ESBLs disseminating worldwide (Zhao and Hu, 2013;Calbo and Garau, 2015). Surveillance in Asia, Latin America, and European revealed dramatically increasing resistance to cephalosporins amongst Escherichia coli and Klebsiella spp., largely contingent on the spread of CTX-M ESBLs (Kumarasamy et al., 2010;Zhao and Hu, 2013). In most of the Europe countries, CTX-M-15 is the most prevalent ESBL type (Pitout, 2013), and recently also disseminated in America, Canada, and Latin America (Denisuik et al., 2013;Wang G. et al., 2013;Kazmierczak et al., 2015). In China, bla CTX-M-14 is identified as the most prevalent ESBL gene Yang et al., 2015).
It has been suggested that the frequent acquirement of plasmids harboring bla CTX-M is largely responsible for the increase of CTX-M-producing Enterobacteriaceae. The bla CTX-M genes are found to associate with certain replicon types of plasmids, mainly including IncF, IncI, IncN, IncHI2, IncL/M, and IncK groups (Zhao and Hu, 2013). The IncF group (FIA, FIB, and FII) is believed to play an important role in the dissemination of bla CTX-M-15 gene, and IncF, IncK and IncI1 largely contribute to the dissemination of bla CTX-M-14 gene. The bla CTX-M-3 gene is mainly harbored by plasmids of IncL/M and IncI1, and bla CTX-M-9 gene by IncHI2 plasmids (Zhao and Hu, 2013). Multiple mobile genetic elements, e.g., ISEcp1 and ISCR1, are involved in the mobilization of bla CTX-M genes as well (Shahid et al., 2012;Zhao and Hu, 2013).
Extended-spectrum β-lactamase-producing K. pneumoniae (ESBL-Kp) has recently become an important nosocomial pathogen (Arpin et al., 2009;Maina et al., 2012;Pitout, 2013;Lohiya et al., 2015). In the tertiary hospitals of China, approximate 50% nosocomial-acquired infections are caused by ESBL-Kp and bla CTX-M-15 and bla CTX-M-14 are the predominant genotypes Li et al., 2014;Yang et al., 2015). Recently, the prevalence rate of ESBL-Kp in communityacquired infections is increasing even causing invasive infection (Kassakian and Mermel, 2014;Toubiana et al., 2016). However, limited knowledge about the dissemination of community-onset ESBL-Kp on national scale in China, especially in settings of secondary hospitals or primary health care. This study was to investigate the epidemiological and genetic characteristics of ESBL-Kp isolates causing community-onset infections in 31 secondary hospitals across China.

Collection of Clinical Isolates
Isolates were collected from August 2010 to 2011 in 31 secondary hospitals locating in 11 provinces representing seven major geographic regions of China (listed in Supplementary Table S1). Patients were selected for this study using the criteria as previous described (Zhang et al., 2014). Bacterial strains were isolated from clinical specimens (urine, blood, sputum, abscesses, and secretions) and identified using API20 (bioMérieux, Durham, NC, USA). All the pure cultures were frozen at −80 • C and shipped to our laboratory for re-identification and further analysis as described previously (Zhang et al., 2014).
Detection of Genes Encoding β-Lactamase and K. pneumoniae-Carbapenemase (KPC) PCR amplification was used to detect β-lactamase genes (bla CTX-M , bla SHV , bla TEM , bla OXA-1 group, bla OXA-10 group, bla VEB , bla PER , bla GES , bla CMY-1 group, bla CMY-2 group, and bla DHA ) and KPC gene (bla KPC ). Primers used for PCR detection and sequencing were acquired from a previous study (Zhang et al., 2014). Sequencing results were analyzed online using Basic Local Alignment Search Tool (BLAST), and were further refined with use of a β-lactamase database 1 .

Multi-Locus Sequence Typing (MLST)
Multi-Locus Sequence Typing was performed on 127 ESBL-Kp (69 bla CTX-M-14 , 58 bla CTX-M-15 ) using the scheme of Institute Pasteur as described previously (Diancourt et al., 2005). New alleles and STs were assigned by the MLST database 2 . Clonal analysis of MLST data was performed using eBURST v3 (Feil et al., 2004). Clonal complexes (CCs) were defined as groups of two or more independent isolates that shared identical alleles at six loci. Each complex was named after the putative founder ST. A minimal spanning tree (MST) was generated by using BioNumerics v7.0 (Applied Maths, Sint-Martens-Latem, Belgium) to provide a graphical representation of the clonal distribution of ESBL-Kp. Neighbor-joining trees were constructed from concatenated sequences of seven MLST loci using the MEGA6 program with Kimura's two-parameter model (Tamura et al., 2011).

Whole-Genome Sequencing (WGS) and Data Analysis
Eleven ESBL-Kp were selected to WGS for further analysis of genetic environment of ESBL genes and plasmid characteristics, including one CTX-M-3-producing isolate, two CTX-M-9producing isolates, seven CTX-M-14-producing isolates, and one CTX-M-15-producing isolate. Genomic DNA was extracted using QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) and sequenced via HiSeq 2000 (Illumina, San Diego, CA, USA) with a 2 × 125 bp paired-end strategy. De novo assembly was generated by using the CLC Genomics Workbench, version 8.0.3 (CLC bio, Aarhus, Denmark). The plasmid analysis was performed as described previously (Zhou et al., 2015). In brief, the plasmids backbones were derived dependent on BLASTn. The contigs of each sample were blasted against the reference plasmid and plotted by BLAST Ring Image Generator (BRIG) (Alikhan et al., 2011).

Accession Numbers
All sequence data of six novel SHV variants are assigned in the GenBank database under the following accession numbers: KC688280, KC688281, KC688282, KC688283, KC688284, and KC688285.
Two AmpC genes CMY-2 and DHA-1 were detected in three and 32 isolates, respectively. All the 35 isolates carrying AmpC genes were also ESBL producers. Additionally, KPC-2 gene was detected in one strain.
ESBL-Producing K. varricola and K. quasipneumoniae were Identified by Phylogenetic Analysis Phylogenetic analysis by concatenating sequences of seven MLST loci showed that 54 STs detected in this study were split into three distinct clades, and most STs belong to clade I ( Figure 1B). It is known that K. pneumoniae consisted of three phylogenetic groups (KPI, II, and III), and has recently been reclassified as three different species (K. pneumoniae, K. quasipneumoniae, and K. variicola) (Brisse and Verhoef, 2001;Rosenblueth et al., 2004;Brisse et al., 2014). Nine isolates of clade I (ST17, ST290, ST322, ST875, ST896, ST1031, ST1522, and ST1798) and two isolates of clade II (ST138 and ST1031) were consequently sent to WGS for the species determination. Phylogenomic analysis assigned clade I to K. pneumoniae (KPI), and clade II to K. quasipneumoniae (KPII) (Figure 1B). The clade III was then supposed to be K. varricola. The two K. quasipneumoniae isolates carried bla CTX-M-9 (ST138) and bla CTX-M-14 (ST1031), respectively, and the one K. varricola strain carried bla CTX-M-15 (ST1181).

The Transmission Mechanism of CTX-M Genes
Plasmids and the genetic environment were analyzed to determine the transmission mechanism of CTX-M genes. The seven CTX-M-14-producing isolates (including K. pneumoniae ST17, ST1798, ST896, ST290, ST322, and K. quasipneumoniae ST1031) collected from different geographical regions shared a highly homologous plasmid carrying bla CTX-M-14 gene, of which the backbone was highly similar with pKP1-19 (accession number CP0012884) identified in a K. pneumoniae strain isolated in the environment in Australia (Supplementary Figure S2A). Of note, pKP1-19 did not encode any drug resistance genes. The seven strains shared an identical genetic environment structure as ISEcp1-bla CTX-M-14 -IS903 (Figure 2A). Both CTX-M-9 isolates (including K. quasipneumoniae ST138 strain and K. pneumoniae ST2316) harbored a similar CTX-M-9 plasmid showing a highly similar backbone with plasmid pFB2.1 carried by Pluralibacter gergoviae isolated in Malaysia (accession number CP014776) (Supplementary Figure S2B). Blasting the contigs harboring bla CTX-M-9 in GenBank revealed that the best match was pKPNDM1 (accession number JX515588; 99% coverage and 100% identity) carried by a Raoultella planticola strain isolated from a patient with post surgical operation infection. This was the only best match showed no less than 99% coverage of the contig. Notably, the R. planticola strain was isolated from the northwest of China (Gansu province), where the two K. pneumoniae were isolated. An ISCR1 located upstream of the bla CTX-M-9 gene ( Figure 2B).
The CTX-M-3 isolate harbored a plasmid carrying the CTX-M-3 gene, of which the backbone was highly similar to a K. pneumoniae plasmid pKF3-94 (accession number FJ876826) with CTX-M-15 gene detected from Zhejiang province of China (Supplementary Figure S2C). Blasting the 7050-bp contig carrying bla CTX-M-3 gene in GenBank revealed some best matches with 99% coverage and 99% identity, including pHS08204 (accession number KP125893; CTX-M-15 plasmid of   Figure S2D). Blasting the 10043-bp contig harboring the bla CTX-M-15 gene revealed the same batch of best matches as those of CTX-M-3-producing plasmids mentioned above ( Figure 2C).
The bla CTX-M-3 and bla CTX-M-15 shared an identical genetic environment. A truncated ISEcp1 disrupted by IS1 (at the end of the contig) was found at the upstream of bla CTX-M-3/15 genes, and genes encoding TEM-1, TnpR of Tn3, and TnpA of Tn3 located downstream of bla CTX-M-3/15 gene ( Figure 2C).

DISCUSSION
This study illustrated the epidemiological and genetic characteristics of ESBL-Kp causing community-onset infections in 31 secondary hospitals distributed in areas across China. Our results showed that the prevalence rate of community-onset ESBL-Kp (31.8%) was comparable to that of nosocomialacquired ESBL-Kp revealed by multiple studies across China (30.1-39.7%) Yang et al., 2013). This raises the concern that the dissemination of ESBL-Kp in community has become another challenge for the resistance control in China, which would be even more difficult to be controlled than that in health-care systems.
As the spread pattern of nosocomial-acquired ESBL-Kp, polyclonal dissemination without any predominant clones was identified in this community-onset study. However, the community-onset ESBL-Kp showed a different clonal distribution comparing to that of nosocomial-acquired ESBL-Kp in China. A multi-center study on tertiary hospitals in China showed that ST11 was the most prevalent ST (12.2%) among 74 STs identified in 155 ESBL-Kp , whereas ST17 was the most prevalent one in our study (12.6%). This is concordant with a previous study that ST17 was predominant in CTX-M-14-Kp and CTX-M-15-Kp more likely correlating to community-onset infections (Peirano et al., 2012). CTX-M-14 and CTX-M-15 were the predominant ESBL genotypes in our community-onset collections, which is the same as that found in the nosocomial studies. However, the population structure of CTX-M-14 and CTX-M-15 producers was different between our collections and the nosocomial-acquired isolates. A multi-center study identified that ST37, 5, 505, 11, and 23 were the major carriers of CTX-M-14, and CTX-M-15 mainly associated with two epidemic clones ST340 and ST15 . In our study, the major carriers of CTX-M-15 were ST86 (n = 7), ST873 (n = 6), and ST776 (n = 6), and the major CTX-M-14 producers consisted of ST17 (n = 12) and ST107 (n = 7). Neither CTX-M-15 nor CTX-M-14 associate with any known epidemic clones implies that different clones of community-onset ESBL-Kp would become epidemic in the future.
Of note, we found a set of ESBL-producing clones (ST23, ST37, and ST86) known to frequently associate with hvKP. Especially, ST86 was the most abundant clone among CTX-M-15 producers in this study, and is known to correlate to invasive infections such as bacteraemia, pneumonia with septic shock and liver abscess (Bialek-Davenet et al., 2014;Liao et al., 2014;Passet and Brisse, 2015). In this study, five of seven ST86 strains were isolated from sputum of children (aged from 15 days to 3 years) with community-acquired respiratory tract infections, and the other two were isolated from liver abscess and blood in adult patients. To our best knowledge, this is the first report of community dissemination of ST86 in Mainland China. Further study is needed to determine whether the ST86 strains are hvKP and why they emerged in respiratory tract infections in children.
Further investigations revealed various transmission mechanisms employed by different CTX-M genes in the community dissemination. We found that plasmids carried CTX-M-14 identified in seven phylogenetically diverse isolates collected from distinct geographical areas across China shared a similar backbone (Supplementary Figure S2). This suggests that a batch of plasmids sharing a similar backbone could be an important reservoir for bla CTX-M-14 in community in China. Interestingly, the segment carrying bla CTX-M-9 identified in our two isolates got a unique best match (99% coverage and 100% identity) to a part of R. planticola plasmid pKPNDM1. Moreover, the R. planticola strain was isolated from the same region (Gansu province) with the two isolates. This suggests that bla CTX-M-9 might not via the horizontal transfer of CTX-M-9-ISCR1 but that of the whole segment as found in the study spread in that region. Additionally, we found that an identical genetic environment ( ISEcp1-bla CTX-M -orf477-bla TEM-1 -tnpR-tnpA) shared by bla CTX-M-15 and bla CTX-M-3 identified in two phylogenetically diverse isolates collected from distinct geographical areas. This indicated that both bla CTX-M-15 and bla CTX-M-3 might spread mainly via the horizontal transfer of this structure in the community in China.
K. pneumoniae was previously classified into three phylogenetic groups (KPI, II, and III), which now are assigned to three different species (K. pneumoniae, K. quasipneumoniae, and K. variicola), respectively. K. quasipneumoniae (KPII) and K. variicola (KPIII) were previously thought to be more susceptible than K. pneumoniae (KpI) isolates to antimicrobial agents, and less frequently cause infections (Brisse et al., 2004;de Melo et al., 2011;Holt et al., 2015). K. variicola is an environmental bacterium mainly associated with plants. Rare reports are available of infections caused by K. quasipneumoniae and K. variicola (Holt et al., 2015). Maatallah et al. (2014) found that K. variicola infections correlate to higher mortality comparing to K. pneumoniae infections. Few studies showed the ESBL-producing KPII and KPIII strains can cause nosocomial infections (Valverde et al., 2008;Maatallah et al., 2014;Holt et al., 2015). The present study revealed the emergency of ESBL-producing K. quasipneumoniae and K. variicola causing community-onset infections. The lack of accurate identification methods for the three species may underestimate the severity of K. quasipneumoniae and K. variicola infections. Surveillance of ESBL-producing K. quasipneumoniae and K. variicola would be helpful to prevent the further prevalence.
In summary, this study highlights the challenge of ESBL dissemination in community in China. Various transmission mechanisms are responsible for the spread of the most common CTX-M genes in community. The prevalence of STs frequently associating with hvKP should be of concern. Surveillance of ESBL-KP causing community-onset infections now appears imperative.