Molecular characterization of clinical IMP-producing Klebsiella pneumoniae isolates from a Chinese Tertiary Hospital

Background IMP-producing Klebsiella pneumoniae (IMPKpn) exhibits sporadic prevalence in China. The mechanisms related to the spread of IMPKpn remain unclear. Methods Carbapenem non-susceptible K. pneumoniae isolates were collected from our hospital. The genetic relatedness, antimicrobial susceptibility, as well as sequence types (ST) were analyzed by pulsed-field gel electrophoresis (PFGE), VITEK 2 AST test Kit, and multilocus sequence typing (MLST), respectively. S1-PFGE, Southern blot analysis and multiple PCR amplification were used for plasmid profiling. Results Between October 2009 and June 2016, 25 non-repetitive IMPKpn isolates were identified. PFGE results showed that these isolates belonged to 20 genetically unrelated IMPKpn strains. Diverse STs were identified by MLST. Most strains carried bla IMP-4, followed by bla IMP-1. Four incompatibility types of bla IMP-carrying plasmids were identified, which included A/C (n = 2), B/O (n = 2), L/M (n = 1) and N (n = 14), while type of other one plasmid failed to be determined. Conclusions The IMPKpn isolates exhibited sporadic prevalence in our hospital. IncN types of plasmids with various sizes have emerged as the main platform mediating the spread of the bla IMP genes in our hospital.


Background
The spread of carbapenemase-producing Enterobacteriaceae has been a major challenge both for treatment of individual patients and for policies of infection control [1]. IMP-1 is the first identified metallo-β-lactamase conferring carbapenem resistance and was described in 1988 in a Pseudomonas aeruginosa strain in Japan [2]. IMPtype carbapenemases can hydrolyze almost all β-lactams. The popular IMP-type carbapenemases have been widespread in non-fermenting Gram-negative bacilli, including Pseudomonas aeruginosa, Acinetobacter spp., and members of the Enterobacteriaceae family [3]. The bla IMP genes are usually located on large plasmids with different replicon types or incompatibility (Inc) types, such as IncA/C, L/M, N, and HI2, which have been commonly associated with the carriage and transmission of various bla IMP genes [4][5][6][7].
In the present study, a retrospective study of clinical carbapenem-non-susceptible enterobacteriaceae isolates from our hospital was performed. In total, 25 IMPKpn isolates were identified in this study. Phenotypic and genotypic characteristics of these isolates were further analyzed.

Bacterial isolates
All clinical Enterobacteriaceae isolates were collected from a 4000-bed tertiary-care hospital and were identified by VITEK ® MS (bioMérieux SA, Marcy-l'Etoile, France). The isolates exhibiting resistance or intermediate to any one of ertapenem or imipenem or meropenem will be defined as carbapenem non-susceptibility strain, and will be screened for bla IMP by PCR and subsequent amplicon sequencing using the primers IMP-1F: 5′-TGAGCAAGTTATCTGTATTC and IMP-1R: 5′-TTAGTTGCTTGGTTTTGATG [11]. E. coli ATCC 25922 was used as a quality control strain for antimicrobial susceptibility testing. The Salmonella ser. Braenderup strain H9812 was used as a reference standard for pulsed-field gel electrophoresis (PFGE). No ethical approval was obtained for using the clinical samples, because they were collected during routine bacteriologic analyses in public hospitals, and the data were anonymously analyzed.

Antimicrobial susceptibility testing
The MICs of clinical commonly used antimicrobial agents (listed in Fig. 1) were measured using VITEK 2 AST-GN09 and AST-GN13 test Kit (bioMérieux, Inc.) according to the manufacturer's instructions. All susceptibility results were interpreted according to the 2017 CLSI performance standards [23].

Plasmid and southern blot analyses
A bla NDM probe was generated by labeling the PCR product using the PCR DIG Probe Synthesis Kit (Roche Applied Sciences, Mannheim, Germany). The size and incompatibility types of bla IMP -carrying plasmids were analyzed by the S1-PFGE, Southern blot and multiple PCR as previously described [24,25]. The whole genomic DNA including the plasmid of the isolates was digested with 20 U of S1 nuclease at 37 °C for 20 min, and separated by PFGE. Then, the DNA fragments were transferred to positively charged nylon membranes (Roche Applied Science). Hybridization was carried out with the DIG Easy Hyb Granules (Roche Applied Sciences), and the detection was performed using the DIG Nucleic Acid Detection Kit (Roche Applied Sciences).

Prevalence and genetic relatedness of IMPKpn
The first IMPKpn isolate in our hospital was identified in October 2009. Since then until June 2016, 25 non-repetitive IMPKpn isolates have been recovered. The source of the IMPKpn isolates included 11 sputum, 6 drainges, 3 bloods, 2 urines, 1 catheter, 1 bile and 1 abscess. The 25 IMPKpn isolates were categorized into 20 PFGE types (type A to T). Type F contained 4 isolates, and types E and H contained 2 isolates, other 16 types contained only 1 isolate (Fig. 1). Isolates with same PFGE type were considered as the same strain. Therefore, a total of 20 IMP-Kpn strains with no genetic relationship were further analyzed.

Antimicrobial susceptibility
The antimicrobial susceptibility patterns of the isolates were listed in Fig. 1. All isolates presented resistance to ceftazidime, cefepime and piperacillin-tazobactam, and exhibited heterogeneous resistance patterns to carbapenem, amikacin and ciprofloxacin (Fig. 1).

bla IMP variants
PCR and subsequent amplicon sequence alignment revealed that most strains carried bla IMP-4 , and several bla IMP gene sequences displayed 100% identity with the published sequence of the bla IMP-1 (for 3 strains), bla IMP-26 (for 1 strains) and bla IMP-38 (for 1 strains) genes, respectively (Fig. 1).

Plasmid analysis
The bla IMP genes were located on plasmids with sizes ranged from approximately 30 to 320 kb (Fig. 1). Four incompatibility types of bla IMP -carrying plasmids including A/C, B/O, L/M and N were identified, whereas type of one plasmid was unable to be determined. In total, 14 IMPKpn strains carried the IncN plasmids with sizes ranging from 30 to 280 kb, while two A/C (120-and 310-kb), two IncB/O (170-and 220-kb) and one IncL/M (55 kb) plasmids were characterized.
In this study, PFGE analysis revealed that the majority of IMPKpn isolates belonged to different types, while an outbreak at small scales (type F strain) that covered only four patients was observed (Fig. 1), suggesting that the majority of IMPKpn isolates analyzed in this study were genetically unrelated strains. Meanwhile, most strains exhibited high ST diversity. Only three and two clones of IMPKpn strains belonged to ST11 and ST629, respectively, while other STs were only identified in single strain. Three strains were identified as new clones with novel STs (Fig. 1). Other reports also showed that the STs of IMPKpn are highly dispersed [5][6][7][8][9][10][11], suggesting the clonal diversity of IMPKpn isolates. Thus, the IMPKpn exhibited a significant sporadic prevalence across the world. Under the same medical condition and selective pressure, the KPC-2-producing K. pneumoniae spread throughout our hospital [24]. However, the IMPKpn strains only exhibited low prevalence and small scale of outbreaks at the same time. A study revealed that some biological characteristics, such as cell motility, secretion, DNA repair and modification, may contribute to the rapid spread of KPC-producing K. pneumoniae ST258 and ST11 [27]. Recent genomic analysis of K. pneumoniae has established that the genomic background was closely related to the pathogenicity of K. pneumoniae [28]. Therefore, the virulence characteristics, rather than their drug-resistant phenotype, may be the major driving force responsible for the emergence and widely spread of antibiotic-resistant pathogens. The urgent efforts are needed to reveal the genetic background of these pathogens, as well as its relationship with their pathogenic and transmission abilities, which is the essential first step to design intervention strategies preventing their spread.
In this study, most of the clinical isolates exhibited low-level resistance or even sensitivity to imipenem and meropenem, but high-level resistance to ertapenem (Fig. 1). Similar results have also been reported in other studies from China [9,13,14,16,20]. These studies showed that the effects of IMP hydrolase on different carbapenems may be diverse, and therefore, the carbapenem-sensitive phenotype does not mean negative for carbapenemase. In most laboratories in China, the production of carbapenemases was not usually detected by modified Hodge test [23] in clinical IMPKpn isolates with carbapenem-sensitive phenotype. The weak hydrolytic ability of IMPs to carbapenems may lead to an underestimation of the prevalence of IMPKpn isolates. It was reported that some IMP variants (such IMP-26) exhibited increased carbapenem-hydrolyzing activity [29]. However, the IMP-26-producing IMPKpn were sensitive to imipenem in this study (Fig. 1). Another study also showed that the IMP-26-producing IMPKpn displayed only low-level resistance to imipenem [14]. Therefore, the production of IMPs was not the only factor responsible for high-level carbapenem resistance, other resistance mechanisms may also confer this process, which include hyperexpression of efflux systems, and loss of the outer membrane channel OprD that allows the entry of carbapenems into the cell [30]. For clinical IMPKpn isolates, it is less clear to what extent the different resistant mechanisms have impact on the carbapenem-resistant phenotype.
Plasmid analysis in this study showed that the majority (13 of 20) of bla IMP -carrying plasmids belonged to IncN. Another study in China found the dissemination of IncN bla IMP -coding plasmids among K. pneumoniae of different genotypes [7]. These results suggested that IncN plasmids have a unique advantage and fitness in the spread of bla IMP genes. However, because most of the IncN plasmids had different sizes in this study (Fig. 1), it was possible that the dissemination of bla IMP genes in our hospital may not mediated by transfer of plasmids but other mechanisms.
The present study has several limitations. First, because this study was retrospective analysis and only limited patient information was available, thus the study focused on the characterization of phenotype and genotype of clinical IMPKpn isolates. Second, the genetic environment of bla IMP genes remained to be characterized in our future study. Another limitation of the study included the absence of analyzing other resistance-related determinants, the outer-membrane permeabilities, and efflux systems of IMPKpn isolates. Further studies are needed to address these limitations.
Authors' contributions KL, YM, LG, JA and LY performed the phenotypic and genotypic analysis of clinical isolates. JY designed the study, performed data analysis and drafted the manuscript. All authors read and approved the final manuscript.