An Outbreak of Carbapenem-Resistant and Hypervirulent Klebsiella pneumoniae in an Intensive Care Unit of a Major Teaching Hospital in Wenzhou, China

Carbapenem-resistant, hypervirulent Klebsiella pneumoniae (CR-hvKP) has recently emerged as a significant threat to public health. In this study, 29 K. pneumoniae isolates were isolated from eight patients admitted to the intensive care unit (ICU) of a comprehensive teaching hospital located in China from March 2017 to January 2018. Clinical information of patients was the basis for the further analyses of the isolates including antimicrobial susceptibility tests, identification of antibiotic resistance and virulence gene determinants, multilocus sequence typing (MLST), XbaI-macrorestriction by pulsed-field gel electrophoresis (PFGE). Selected isolates representing distinct resistance profiles and virulence phenotypes were screened for hypervirulence in a Galleria mellonella larvae infection model. In the course of the outbreak, the overall mortality rate of patients was 100% (n = 8) attributed to complications arising from CR-hvKP infections. All isolates except one (28/29, 96.6%) were resistant to multiple antimicrobial agents, and harbored diverse resistance determinants that included the globally prevalent carbapenemase blaKPC−2. Most isolates had hypervirulent genotypes being positive for 19 virulence-associated genes, including iutA (25/29, 86.2%), rmpA (27/29, 93.1%), ybtA (27/29, 93.1%), entB (29/29, 100%), fimH (29/29, 100%), and mrkD (29/29, 100%). MLST revealed ST11 for the majority of isolates (26/29, 89,7%). Infection assays demonstrated high mortality in the Galleria mellonella model with the highest LD50 values for three isolates (<105 CFU/mL) demonstrating the degree of hypervirulence of these CR-hvKP isolates, and is discussed relative to previous outbreaks of CR-hvKP.


INTRODUCTION
Klebsiella pneumoniae is a Gram-negative pathogen causing hospital-acquired infections, especially for immunocompromised patients in the intensive care units (ICU) (1)(2)(3). Variants called hypervirulent K. pneumoniae (hvKp) have evolved (4) which are becoming more prevalent. Diagnostically, hvKp tend to be hypermucoviscous due to increased secretion of capsular polysaccharide, and metabolically more robust through the acquisition of genes for more efficient iron acquisition from host tissues (2). Isolates of hvKP usually belong to sequence type ST23 and their capsule is usually of the fucosylated K1-type (5). There is also mounting evidence that hvKp secrete factors which impact the epithelia and microbiome, such as the cell-cycle modulating non-ribosomal peptide colibactin and the channel-forming bacteriocin microcin E492 (6,7), respectively. These factors modify the host niche and might thereby contribute to the observed increase in virulence (6,8). The significance of these features has been established in the measurement of virulence phenotype in vitro and in vivo (9)(10)(11), and whole genome sequence analyses document the range of genes for iron-acquisition, enhanced capsule secretion, and toxin secretion. These include iron-binding siderophores: aerobactin, yersiniabactin, and salmochelin, as well as TonBdependent transporters for iron (Kfu), iron dicitrate (Fec), and ferric siderophores (FyuA) (7,12).
Carbapenem-resistant K. pneumoniae (CRKP) are another global threat given that until recently carbapenems were last line drugs for Klebsiella infections (13)(14)(15). Carriage of the gene bla KPC−2 is the main molecular mechanism of carbapenem resistance in K. pneumoniae in Asia, especially in China, and carriage of the bla KPC−2 is mainly associated with strains of the clonal complex 258 (CC258). Among CC258, ST11 is the major sequence type (ST) from Asia, especially China (16).
Recently, a lineage of K. pneumoniae that simultaneously exhibits hypervirulence and carbapenem-resistance often refer to as "CR-hvKp" (12,24,(26)(27)(28). These strains are of major concern: not only are they hypervirulent and multidrug-resistant, but they are also highly transmissible (24,25,29,30). Sequence analysis of a recent fatal outbreak of ST11 CR-hvKP strains in a Chinese hospital revealed that the several isolates belonged to a single clone, and yet with slight variation in pulsed-field gel electrophoresis (PFGE) patterns between the strains suggestive of extraordinarily high mutation rates (24). The observations in that study underscore the urgent need for a better understanding of CR-hvKp.
In this study, we describe an outbreak of ST11 K. pneumoniae strains that are both carbapenem-resistant and hypervirulent, isolated from the ICU of the First Affiliated Hospital of Wenzhou Medical University. We sought to investigate the molecular and epidemiological features of these strains during the outbreak, especially on the characteristics of virulence factors and antimicrobial resistant determinants. The data from this outbreak confirmed that long-term carriage of CR-hvKp can give rise to in host evolution and an increase the repertoire of virulence factors carried. Hence, monitoring the nosocomial dissemination of CR-hvKp, providing effective strategies for antiinfective prophylaxis and outbreak management are essential.

Collection of K. pneumoniae Clinical Isolates From ICU
The First Affiliated Hospital of Wenzhou Medical University is a leading teaching, three-grade hospital in Wenzhou, a city in Zhejiang province, China. The hospital features 3,000 beds, 64 departments and 89 wards. It is one of the largest health care centers in south Zhejiang province, has responsibility for the medical care of an estimated 30 million population.
A total of 29 K. pneumoniae isolates were collected from eight patients from the ICU from March 2017 to January 2018. Patient information including age, gender, length of ICU stays, diagnosis, and outcomes is obtained from the Electronic Medical Records. The study and consent procedure were approved by the Ethical Committee of the First Affiliated Hospital of Wenzhou Medical University.
All isolates were isolated from various clinical specimens and identified as K. pneumoniae by VITEK-2 automated microbiology analyzer (bioMérieux, Marcy l'Etoile, France).

Antimicrobial Susceptibility Testing (AST)
Minimal inhibitory concentrations (MICs) of antimicrobial agents (gentamicin, tobramycin, ampicillin, cefazolin, ceftazidime, ceftriaxone, imipenem, ertapenem, levofloxacin, and ciprofloxacin) were determined by the agar dilution method and interpreted according to the Clinical and Laboratory Standards Institute (document M100-S27). Furthermore, ASTs of tigecycline and polymyxin B were performed by the broth microdilution method and interpreted by the recommendation of the European Committee on Antimicrobial Susceptibility Testing clinical breakpoints (http://www.eucast.org). Escherichia coli ATCC25922 and Pseudomonas aeruginosa ATCC27853 served as the quality control strains for susceptibility testing.

Molecular Epidemiology Analysis
Multilocus sequence typing (MLST) was performed for phylogenetic analyses according to the protocols on the Pasteur Institute MLST website for K. pneumoniae (https://bigsdb. pasteur.fr/klebsiella/klebsiella.html).
All strains were subjected to PFGE analysis. Strain DNA were digested through XbaI for 2 h at 37 • C. Electrophoresis was conducted at 14 • C for 18.5 h using the Bio-Rad R CHEF-Mapper XA machine (Bio-Rad, California, USA). Salmonella enterica serotype Braenderup H9812 was used as the molecular marker (33). Reports-phylogenetic tree (UPGAMA) was constructed to determine the hierarchical representation of different linkage levels between the strains. Chromosomal DNA restriction patterns were interpreted using previously established guidelines (34), with a phylogenetic dendrogram analysis indicated at 85% demonstrating strain relatedness.

Galleria mellonella Larvae Infection Assay
The virulence capability was evaluated through the commonly used in vivo infection model, the wax moth Galleria mellonella (7,24,(35)(36)(37). According to the antimicrobial resistant profile and hypermucoviscous genotype reported previously, five strains were picked out for the further infection model. Two strains (Kp1 and Kp15) were resistant and hypermucoviscous as well, two strains (Kp4 and Kp21) were resistant but nonhypermucoviscous and one (Kp6) was susceptible and nonhypermucoviscous. The methods were described previously (35). In the present study, each infection group includes eight larvae and were injected with 10 µL of bacterial suspension with bacteria concentration ranging from 10 5 to 10 7 CFU per larva. As a control group, 10 µL PBS was injected in parallel. All larvae were placed in Petri-dishes and kept at 37 • C in the dark. The number of dead larvae was recorded with notes on any melanization and lack of motility 24, 48, and 72 h after injection, respectively. All experiments were repeated in biological triplicate as described and results were not considered if two or more larvae in any of the control groups died. The survival curves were plotted by the GraphPad Prism 5.0 program and the Kaplan-Meier method was used for statistical analysis. P < 0.05 was considered statistically significant. For each isolate, the 50% lethal dose (LD 50 ) was analyzed using the probit analysis program by SPSS 22.0 and the results were expressed as log 10 LD 50 .
The retrospective clinical data showed that the first case (patient 8) was a male patient admitted to the ICU on 25th March 2017 with the diagnosis of pulmonary infection. On day first after hospitalization, three ST11 K. pneumoniae (Kp27, Kp28, and Kp29) isolates were retrieved from fecal, pus and sputum samples, respectively, and the patient died of fatal infection at day 15 ( Figure 1, Table 1). The second case was patient 4, a 69-year-old man who underwent surgery for a brainstem tumor, and was admitted to the ICU on 1st June 2017 (Figure 1, Table 1). Patient 4 had the longest hospitalization length of stay (228 days), dating from June 2017 to January 2018. On 2nd June, K. pneumoniae was isolated from a sputum sample from patient 4. In the course of the ICU stay, patient 4 yielded 11 independent Klebsiella samples, with positive sputum samples recovered for 115 days (Figure 1). Administration of antibiotics (cefoperazone, meropenem, and tigecycline) did not resolve the ongoing respiratory infection. Two fecal samples from patient 4, in June and November, were also positive for Klebsiella (Figure 1) but were subsequently identified as a distinct sequence type (ST37, Figure 2), while other isolates belonged to ST11. Ultimately, an ST11 strain was isolated from the patient's blood in January 2018, immediately preceding death. Within this time frame, ST11 K. pneumoniae were isolated from other six patients in the ICU concurrently during this hospitalization period (Figure 1), indicating a potential of nosocomial transmission. The infection timeline is detailed in Figure 1.
To control this outbreak, all patients were placed in contact isolation. The patients were arranged in single rooms with disinfection of 2% chlorhexidine baths daily and the affected patients' rooms were disinfected by glutaraldehyde thoroughly. The staff members in the ICU were screened for signs and symptoms of infections daily. The outbreak was effectively controlled and besides these eight patients, no further infected patients were found in the ICU.

Epidemiological Analysis and Virulence Characteristics
Multilocus sequence typing (MLST) analysis revealed that 26 isolates (26/29, 89.7%) retrieved from eight patients were identified as ST11 K. pneumoniae, indicating that this was the dominant ST type in these ICU deaths (Figure 2). In addition, three isolates belonged to either ST37 (patient 4, Kp10, Kp16; 6.9%) or ST375 (patient 3, Kp6; 3.4%). The similarities seen in the MLST phylogram indicated close relationships between some of the isolates. For example, the Kp1 isolate from the initial blood sample drawn from patient 1 was very closely related to the isolates Kp2 and Kp3 (Figure 2) drawn from both patient 1 and patient 2 in November (Figure 1). The subsequent urinary tract infection of patient 2 in December (Figure 1) was attributable to a distinct pair of ST11 isolates, Kp4, and Kp5 (Figure 2). The PFGE results showed that 29 isolates were divided into five different PFGE clusters (A to E). Cluster D and E accounted for 20.6% (6/27) and 65.5% (19/27) of ST11 isolates, respectively, indicating that these strains were highly homologous. The two ST37 isolates belonged to the same cluster (cluster A). Kp6 (ST375) and Kp7 (ST11) formed separate monotypic clusters, cluster B and C, respectively (Figure 2). According to previous definition, the ST11 K. pneumoniae should be considered potentially hypervirulent (23,24,38). This, together with the high mortality rate (8/8 patients, 100%) in the outbreak led us to address the characteristics of these new isolates.

Extracellular Capsule
Only two isolates (Kp2, Kp7) lacked the transcriptional regulator of mucoid phenotype rmpA, a gene found responsible for upregulation of the synthesis of capsular polysaccharide (39). All of the isolates were negative for rpmA2, which has been reported as a trans-acting activator for the capsular polysaccharide biosynthesis (39). The wzy-like gene magA was involved in polysaccharide synthesis in K1 capsulated strains (40), and the biofilm-promoting fucose synthase wcaG (41) is required for decoration of the capsule polysaccharide with fucose. Both genes were absent from all isolates (Figure 3). While the ST11 strains were of indeterminate capsule type (i.e., not of the typical K1 capsule type) and only Kp6 (ST375) was identified as K2 capsule type (Figure 3). Interestingly, only four ST11 (Kp1, Kp6, Kp15, and Kp18) were string-test positive, which was correlated with a hypermucoviscous phenotype in previous studies (2,43).

Fimbrial Adhesin
All ST11 isolates were positive for fimH and mrkD suggesting that they can express type 1 and type 3 fimbrial adhesion (45,46), which mediate binding to the extracellular matrix and play an important role in biofilm formation (47).

Nitrogen Metabolism
All isolates were found to be negative for the allS transcriptional activator (Figure 3), suggesting they were incapable of allantoin utilization to compete for nitrogen access in host niches (48). Most of the isolates were positive for ureA (23/29), which enables the use of urea as a nitrogen source in host niches (49). FIGURE 2 | PFGE results and MLST typing for 29 K. pneumoniae isolates. Genomic DNA from each of the 29 K. pneumoniae isolates was digested using XbaI and the digests subjected to pulsed-field gel electrophoresis to generate diagnostic genomic DNA fragmentation fingerprints. The dendrogram of the PFGE profiles is clustered by the UPGAMA on the basis of the Dice similarity by the Quantity One software package 4.6. The sequences of seven housekeeping genes (gapA, infB, mdh, pgi, phoE, rpoB, and tonB) were analyzed for the purpose of MLST typing. PFGE results showed that 29 isolates were divided into five different PFGE clusters (A-E) as represented by different colors. Cluster D and E accounted for 20.6% (6/27) and 65.5% (19/27) of ST11 isolates, respectively, indicating that those strains were highly homologous. The two ST37 isolates belonged to the same cluster (cluster A). Kp6 (ST375) and Kp7 (ST11) formed separate monotypic clusters, cluster B and C, respectively. Asterisk indicates the string test of isolate was positive.
Polymyxin-resistance was measurable in two of the isolates, Kp12 (patient 4, ST11) and Kp21 (patient 5, ST11), but not in any others. Nevertheless, these two strains were negative for mcr-1, the most prominent cause of polymyxin resistance in China (51,52). However, other resistance mechanism such as modifications of two-component systems (PmrA/PmrB and PhoP/PhoQ) and inactivation of the mgrB gene (53) might be present, but would require to be further investigated. Interestingly, patient 5 did not respond to polymyxin treatment, despite Kp19 (ST11) being deemed polymyxin-sensitive ex vivo (Table S3). Similarly, for the glycylcycline antibiotic tigecycline, patients 1, 4, and 6 did not respond to intravenous treatment with the antibiotics, despite all isolates being deemed sensitive to tigecycline ex vivo (Table S3).

Hypervirulence Measurements Through the Death of G. mellonella Larvae
Five isolates were collected based on the antimicrobial resistance profile and hypermucoviscous phenotype: Kp6 (patient 3, ST375, hypermucoviscous, antibiotic-sensitive), Kp1 (patient 1, ST11, hypermucoviscous, carbapenem-resistant, bla KPC−2 positive), and Kp15 (patient 4, ST11, hypermucoviscous, carbapenemresistant, bla KPC−2 positive), Kp4 (patient 2, ST11, nonhypermucoviscous, carbapenem-resistant, bla KPC−2 positive), and Kp21 (patient 5, ST11, non-hypermucoviscous, carbapenemresistant, bla KPC−2 positive). Among the five selected strains, the Kp1 isolate from the initial blood sample drawn from patient 1 is hypermucoviscous as determined by string-test and resistant to beta-lactams, carbapenem and quinolones. During the outbreak this strain was most likely responsible for the infection of patient 1 possibly transmitted to patient 2 (Figure 1, Table S3). The isolate Kp6 was ST375, showed a positive string test as well as acquisition of genes for iutA, which predicted it could be hypervirulent. Kp21 was isolated from a urinary catheter of patient 5 before expiration. From our analysis it was determined to be non-hypermucoviscous, but extremely drug-resistant. It was therefore of interest to determine the virulence phenotypes of Kp1, Kp6, and Kp21. In addition, Kp4 and Kp15 were selected randomly from each of the two groups with hypermucoviscous phenotype and non-hypermucoviscous phenotype.
All isolates caused a time-dependent death of the larvae. When the dose of bacteria was altered (10 5 , 10 6 , 10 7 CFU), K. pneumoniae-induced lethality was found to be dependent on the inoculum density ( Figure S1). After 72 h of infection, the mortality of the larvae infected with 10 7 CFU bacterial suspension was higher than that infected with 10 5 CFU bacterial suspension. The observed mortality of the larvae was higher than 60% after 72 h (Figure 4) with the most virulent being Kp4 (100% mortality), Kp21 (100% mortality), and Kp6 (75% mortality). There was no significant difference among these groups (P > 0.05).

DISCUSSION
Our study reported an outbreak of ST11 K. pneumoniae in the ICU of a teaching hospital where multiple isolates showed highly multiple-drug resistance phenotypes including carbapenem-resistance. Sampling from patients showed continuing colonization with K. pneumoniae. Isolates were taken under a surveillance protocol aimed at patient care, which means sampling bias interferes with drawing strong conclusions on the extent of in host evolution vs. transmission and re-infection by multiple ST11 K. pneumoniae strains. Nonetheless, several important points can be made from the data gathered from this ICU outbreak.
Firstly, this study is consistent with observations of long-term colonization by carbapenem-resistant K. pneumoniae reported by three other studies of hospital-acquired infection (54)(55)(56). As the ICU outbreak reported in the United States of America, two patients who survived the outbreak remained colonized by carbapenem-resistant K. pneumoniae for up to 4 years. A single clone was detected in samples from various sites (throat, groin, rectum) of the patients, and the documented characteristics of in host evolution over these years suggested recombination and gene loss events, particularly affecting the plasmids carried by the long-term colonizing K. pneumonia (56). In this outbreak in the First Affiliated Hospital in Wenzhou, patient 4 remained colonized by carbapenem-resistant K. pneumoniae, with positive sputum samples recovered for 115 days.
Secondly, most K. pneumoniae (26/29, 89.7%) in this outbreak were assigned to ST11 lineage which was reported as being prevalent in the wide spread of KPC-producing K. pneumoniae in China (13). In early 2011, a nosocomial outbreak of bla KPC−2 positive ST11 CRKP infection was observed in the north of China (57). Subsequently, several outbreaks of bla KPC−2 positive ST11 CRKP were reported in multiple hospitals in China (58)(59)(60)(61), echoing the prevalence of those stains in the mainland of China. In our study, all ST11 CRKP harbored bla KPC−2 and bla SHV−11 , and most were also positive for bla TEM−1 and bla CTX−M, consistent with previous reports. ST11 CR-hvKP has been recently reported elsewhere in China, causing fatal infections and high mortality in other hospitals (24,38,61,62).This seemingly recent increase in occurrence of the ST11 CR-hvKP in various Chinese hospitals is of grave concern, as it indicates either the presence of these strains within the community, or the high potential of ST11 K. pneumoniae to repeatedly yet independently acquire hypervirulence genes in the distinct hospitals to cause serious outbreaks. The isolates presented in our study carried more resistance determinates in their genotypes, and showed correspondingly more multi-drug resistant phenotypes, compared to other reports in the literature. This indicates that increasingly serious threats are emerging.
Thirdly, our study highlights some short-comings in realtime diagnostics of drug-resistant phenotypes and genotypes in the case of K. pneumoniae detection. Technologically, it is now feasible to apply partial-targeted or whole genome sequencing (WGS) to patient samples in order to determine drug resistance profiles. However, there were multiple instances in this outbreak where isolated bacteria were sensitive to the antibiotic used ex vivo, nevertheless, the infection caused by these isolates did not respond to antibiotic treatments effectively in vivo. For example, in the case of patient 4 administration of antibiotics including tigecycline was ineffective in treating the respiratory infection, despite the observation that ex vivo, the isolates from patient 4 were tigecycline-sensitive. Another example was in the case of patient 5, polymyxin B also failed to treat the infection although the isolates was sensitive ex vivo. For future outbreaks, introducing WGS analysis should be considered as it can identify pathogenic virulence factors, diagnostic for hvKP variants guiding specific hospital outbreak control measures and treatment plans to be actioned (24,60,62,63). For instance, due to the high transmission nature of hvKP (24,64), time-sensitive treatment and strict isolation measures should be performed immediately after detection to prevent further possible transmission reducing. Furthermore, as hvKP often present as abscesses, drainage of these sites of infection should be considered (2), and strict glycemic control may help clinicians to reduce the possibility of septic metastatic complications (43).
Finally, with the emergence of drug-resistances in hypervirulent strains of K. pneumoniae, it is incumbent on the community to have clearer definitions of the term "hypervirulence". The accepted clinical definition of hvKP was based on manifestations that include liver abscesses, metastatic meningitis and endophthalmitis (2). In some studies, hypervirulent strains were identified by their hypermucoviscous phenotype through string test (26,38,65). Alternatively, hypervirulence gene signatures have been developed through genome-wide association studies that mostly identified an increased capability of iron acquisition and likely production of a hypermucoviscous phenotype (2, 6-8, 11, 12, 66). Ultimately, the definition of a hypervirulent phenotype is a greatly increased level of virulence. Larvae from Galleria mellonella, a wax moth, are becoming more widely used to study the pathogenesis of K. pneumoniae infections and the efficacy of antimicrobial drugs and phage-based therapies (24, 35-37, 67, 68). It is an experimentally tractable model and the similarities between these insects and vertebrate innate immune responses make it suitable for screening and comparative analysis of bacteria isolated from human patients (69). The results presented here show that the K. pneumoniae infection of G. mellonella larvae induced dose-dependent and time-dependent lethality. Importantly, even with strains that covered a range of drugresistance and hypermucoviscous phenotypes, similarly high mortality rates were observed in the animal model of infection, with no significant difference between these isolates. This data suggests that hypervirulence is not directly predictable from genetic signatures indicative of hypermucoviscosity or iron-acquisition phenotypes. Our data support the proposition that "hypervirulent" could be defined through calculated LD 50 value drawn from a standardized Galleria model assay (37). The threshold value used to denote hypervirulence should be addressed by the Klebsiella research community.
In conclusion, our study described the outbreak of ST11 K. pneumoniae that exhibited hypervirulence and multidrug resistance in one of the largest comprehensive hospitals in China. With the prevalence of such CR-hvKp, severely challenges have appeared in clinical treatments and pose a substantial threat to the world's population. Sustained efforts are required to limit the transmission of these strains, and prevent future outbreaks.

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