Characterization of ST11 and ST15 Carbapenem-Resistant Hypervirulent Klebsiella pneumoniae from Patients with Ventilator-Associated Pneumonia

Introduction

Klebsiella pneumoniae is a gram-negative opportunistic pathogen that can cause both community and hospital-acquired pneumonia, urinary tract infections, and bacteremia primarily in immunocompromised patients.¹ Most experts agree that carbapenems are an effective treatment for serious infections brought on by Enterobacteriaceae that produce extended-spectrum beta-lactamases (ESBLs).^2–4 Because of the wide and irrational use of antibiotics, the number of carbapenemase-resistant Enterobacteriaceae (CRE) infections is increasing per year and antibiotic resistance has become an international problem.^5,6 During 2005–2020, the China Antimicrobial Surveillance Network (CHINET, http://www.chinets.com) reported that K. pneumoniae resistance rates to imipenem and meropenem have rapidly increased from 3% and 2.9% to 23.2% and 24.2% respectively. There has been a significant increase in respiratory tract infections in intensive care units (ICUs) due to carbapenem-resistant K. pneumoniae (CRKP), which is considered a severe danger to public health as a pathogen associated with hospital-acquired infections (HAIs).^7–10 It is reported that 9–27% of ICU patients suffer from ventilator-associated pneumonia (VAP).¹¹ Meanwhile, treatment of CRKP-induced VAP in ICU patients is difficult because of the lack of multiple therapeutic drugs.

Hypervirulent K. pneumoniae (hvKP) is the primary cause of community-acquired illnesses such as meningitis, necrotizing fasciitis, endophthalmitis, liver abscesses, and pneumonia in people of all ages and health statuses. It is associated with severe morbidity and mortality and has been reported to have an increasing incidence rate globally.^12–15 When the string test is positive for hvKP, it is defined to have a hyper-viscous phenotype. K1 and K2 are the most frequent capsular loci for hvKP.¹² ST23 and ST258 strains account for the majority of hvKP strains.¹⁶ Antibiotics usually work well on hvKP, however infections brought on by carbapenem-resistant hvKP (hv-CRKP) have recently been documented increasingly frequently^17,18 To the best of our knowledge, the emergence of such a high-virulent, drug-resistant pathogen poses an even more significant challenge to clinical patient safety.

This investigation was conducted on the VAP outbreaks caused by ST11 and ST15 hv-CRKP in a teaching hospital’s Emergency Intensive Care Unit (EICU) in China. Furthermore, the clinical characteristics of patients as well as the outcome, resistance mechanisms, virulence characteristics, and genomic features of these strains were analyzed.

Materials and Methods

Isolates Collection and Identification

From March 2019 to September 2021, 19 hv-CRKP-VAP cases were admitted to the EICU of the Affiliated Hospital of Xuzhou Medical University. 19 non-repeated isolates of hv-CRKP were collected from these patients. Matrix-Assisted Laser Desorption / Ionization Time of Flight Mass Spectrometry (Bruker Daltonics, Billerica, MA, USA) was used to recheck the isolates of K. pneumoniae. All isolates were resistant to at least one carbapenems (imipenem, meropenem, ertapenem) and were identified as CRKP by the VITEK-2 compact automated microbiology analyzer (Biomerieux, Marcy L’Etoile, France). String tests were carried out to see if hvKP had the hypermucoviscosity phenotype. Furthermore, patients’ clinical data including gender, age, isolation date, sampling site, comorbidity, invasive operation, antibiotic treatments, and outcomes were also collected. This investigation was authorized by the Ethics Committee of the Affiliated Hospital of Xuzhou Medical University XYFY2020-KL084.

Testing for Antimicrobial Susceptibility

The minimal inhibitory concentrations of antibiotics, including amikacin, aztreonam, ceftazidime, ciprofloxacin, cefotaxime, cefepime, imipenem, meropenem, polymyxin B, sulfamethoxazole, tigecycline, and piperacillin/tazobactam were determined by VITEK-2 Compact system. The guidelines provided by the Clinical and Laboratory Standards Institute (CLSI, 2021) were followed for the interpretation of the results.¹⁹ Pseudomonas aeruginosa ATCC27853 and Escherichia coli ATCC25922 were used as quality control strains.

Detection of Antimicrobial Resistance Genes

For extracting the genomic DNA of all the isolates, the boiling method was applied. The ESBL genes (bla_TEM, bla_SHV,bla_CTX-M-1 group, bla_CTX-M-2 group, bla_CTX-M-8 group, and bla_CTX-M-9 group), carbapenemase genes (bla_KPC, bla_NDM,bla_VIM, bla_IMP, and bla_OXA-48) and AmpC-lactamase genes (bla_CMY and bla_DHA) were amplified by polymerase chain reaction (PCR) as described previously and the PCR products were sequenced and then on the sequencing results BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) was run.^20,21 The primer sequences are enlisted in Table S1.

Detection of Capsular Serotyping and Virulence-Associated Genes

Wzi gene was amplified by PCR and sequenced as previously described.²² To compare the findings and identify the capsular type of K. pneumoniae, the database (https://bigsdb.pasteur.fr) was consulted. 12 virulence-associated genes including aerobactin, rmpA2, rmpA, magA, kfu, allS, fimH, ybtS, entB, mrkD, iroN, and iucA were also amplified by PCR.^23,24 The primer sequences of Wzi and virulent genes are enlisted in Table S1.

Multilocus Sequence Typing (MLST)

gapA, infB, mdh, pgi, phoE, rpoB, and tonB are the seven housekeeping genes (expressed by K. pneumoniae) that were amplified by PCR and sequenced for all isolates by the MLST website’s methodology (https://bigsdb.pasteur.fr/klebsiella/primers-used). The sequencing results were compared with the database to identify sequence types.

Pulsed-Field Gel Electrophoresis (PFGE)

The homology of all strains was analyzed by PFGE assay.²⁵ Genomic DNA was digested with Xbal for 2 h at 37°C and the fragments were separated using the CHEF Mapper XA System (Bio-Rad) for 18h at 14°C. The molecular identifier was determined to be a Salmonella serotype Braenderup strain H9812.²⁶ Bionumerics software was utilized for the unweighted pair-group technique with arithmetic means (UPGMA) for cluster analysis. A greater than 85% similarity was considered as a closely related.^20,27

String Test

Isolates of K. pneumoniae were plated on blood agar and grown overnight at 37°C. A single colony was contacted and pushed outward using an inoculation loop. The string test was considered positive and the isolate showed a hypervirulent phenotype if the length of the viscous string was greater than 5 mm.²⁸ This experiment was repeated thrice.

Serum-Killing Assay

Three ST11-KL64 strains (KP3, KP9, KP13) and 3 ST15-KL112 strains (KP6, KP15, KP18) of bacteria were randomly selected for serum killing assay. The test was chosen in accordance with a previously reported study.^28,29 Serum was collected from 10 healthy individuals and was kept at −80°C. In phosphate-buffered saline (PBS), the mid-log phase bacteria were suspended in a concentration of 1×10⁶ CFU/mL and then combined with normal serum in a ratio of 1:3. The mixture was immediately plated on Mueller-Hinton agar after being incubated at 37°C for 0, 1, and 3 hours in order to count the number of bacteria. The results were graded on six levels, and an isolate was declared serum sensitive at grade 1 or 2, intermediately sensitive at 3 or 4, and resistant at 5 or 6. The experiment was repeated thrice. For hyper and low virulence control, ATCC43816 and ATCC700603 were employed, respectively.^27,30,31

Galleria Mellonella Larvae Infection Assay

Galleria mellonella larvae infection assay was performed to further assess bacterial pathogenicity.^20,32–34 The isolates and controls were the same as those in the serum-killing assay. PBS was used to adjust the overnight culture to the concentrations of 1×10⁵ to 1×10⁸ CFU/mL. Galleria mellonella larvae weighing 250—350 mg were used in this experiment. Ten microliter of various concentrations of each bacterium strain suspension was injected in 10 larvae through last proleg. Furthermore, 10 µL PBS was injected into 10 wax moths as an experiment control. Each larva was put onto a clean Petri plate and incubated at 37°C. For three days, the larval survival rates every 12 hours were recorded. This experiment was performed in three replicates.

Whole Genome Sequencing and Analysis

BGI Technology Service Co., Ltd sequenced the whole genome of strain KP3 in Shenzhen. The PacBio Sequel II system was used for third-generation sequencing. The whole genome was submitted to GenBank. BLAST (http://blast.ncbi.nlm.nih.gov) was utilized to compare plasmid sequences and the genes associated with drug resistance and virulence were predicted using ResFinder4.1 and VFDB 2019 databases. Map plasmid genome circles were assessed via the Gcview website.

Results

Patient’s Clinical Features and CR-hvKP Isolates

Nineteen isolates of hv-CRKP-VAP were collected from different patients during their stay at the EICU between March 2019 to September 2021. Before strain isolation, all the patients underwent invasive operations and received antibiotic treatment. The mortality rate was 73.68% (14/19). The isolates were acquired from different specimens including sputum (11/19, 57.89%), blood (4/19, 21.05%), bone marrow (1/19, 5.26%), cerebrospinal fluid (1/19, 5.26%), fibro bronchoscope lotion (1/19, 5.26%), and hydrothorax (1/19, 5.26%). The specifics of each clinical feature are listed in Table 1.

Table 1 Patient’s and Hv-CRKP Isolate’s Clinical Features

Antibiotic Sensitivity and Genetic Characteristics of Resistance Genes

Except for tigecycline and colistin, all the organisms were highly resistant to the tested antibiotics (Table 2). Carbapenemase genes were detected from all 19 isolates, including bla_KPC-2 (14/19, 73.68%), bla_OXA-232 (4/19, 21.05%), and bla_NDM-1 (2/19, 10.53%). KP19 contained both bla_KPC-2 and bla_NDM-1. Seven strains (36.84%) harbored the bla_CTX-M-1 group and the profitable products were identified as bla_CTX-M-107. The bla_CTX-M-65 variant was present in 12 (63.16%) of the strains that owned the bla_CTX-M-9 group. The ratios of bla_SHV and bla_TEM expression by CR-hvKP bacteria were 100% (19/19) and 73.68% (14/19), respectively. No isolate expressed bla_VIM,bla_IMP,bla_CMY,bla_DHA,bla_CTX-M-2, and bla_CTX-M-8 groups (Figure 1).

Table 2 Antibiotic Resistance Characteristics

Figure 1 19 hv-CRKP isolates’ antimicrobial resistance genes, virulence-related genes, and string test results. The blue box represents the presence of genes or a positive string test result.

Virulence-Associated Genes and Capsular Serotyping

Figure 1 displays the outcomes of 12 virulence-related genes. The fimH and entB genes were present in all 19 isolates. The following genes were mrkD (18/19, 94.74%), ytbS (18/19, 94.74%), iucA (17/19, 89.47%), rmpA2 (16/19, 84.21%), rmpA (11/19, 57.89%), iroN (10/19, 52.63%), and kfu (5/19, 26.32%). No isolate in this study expressed aerobactin, magA, and allS. Six distinct capsular serotypes for strains were found by PCR and sequencing of the Wzi gene, including KL64 (10/19, 52.63%), KL112 (4/19, 21.05%), KL47 (2/19, 10.53%), KL19 (1/19, 5.26%), KL62 (1/19, 5.26%), and KL140 (1/19, 5.26%) (Figure 2).

Figure 2 PFGE results, MLST type, capsular serotype, and date of isolate for 19 hv-CRKP isolates. Similarity >85% was considered closely related.

MLST and PFGE

In total four STs were identified, where ST11 (12/19, 63.16%) accounted for the majority of the strains, followed by ST15 (5/19, 26.32%), ST17 (1/19, 5.26%), and ST48 (1/19, 5.26%). In the PFGE experiment, 4 ST15 and 8 ST11 isolates indicated high homology (>85%), respectively (Figure 2).

Virulence-Related Characteristics

The string test revealed that all strains were positive (Figure 1). KP9 and the hypervirulence standard strain ATCC43816 are resistant to healthy, normal serum (grade 6). KP15 and KP18 strains were moderately sensitive (grade 4) to healthy serum, however, KP3, KP6, and KP13 as well as the classical K. pneumoniae isolate ATCC700603 were all susceptible to healthy serum (grade 1 or 2) (Figure 3). To further evaluate the virulence of these strains in-vivo, G. mellonella larvae were infected with the experimental strains and their survival rate was calculated after 72h of incubation. When the concentration of bacteria suspension was 1×10⁶ CFU/mL, the survival rate of classic K. pneumoniae isolates ATCC700603 was 73.33%, and that of hypervirulence control ATCC43816 was 6.67%, whereas, the survival rates of KP3, KP6, and KP13 were 16.67%, 6.67%, and 13.33%, respectively at 72h. KP9 and KP15 had a survival rate of 10% after 72h. Survival of G. mellonella reached 0% with KP18 at 24h (Figures 4 and S1). The result of the G. mellonella larvae infection assay suggested the strains in this study were hypervirulent.

Figure 3 Serum killing assay of 6 selected isolates. ATCC43816 and ATCC700603 were hypervirulence and low virulence controls, respectively.

Figure 4 Survival rate of G. mellonella larvae infected with 1×106 CFU/mL of each selected isolate. ATCC43816 and ATCC700603 were hypervirulence and low virulence controls, respectively. PBS was the experimental control.

Whole Genome Sequencing and Analysis

The genome sequence identified in this investigation was submitted to GenBank with the accession number CP116491, CP116492, CP116493, and CP116494. KP3 was discovered to have a 5521665bp chromosome and three plasmids (218592bp, 100630bp, and 85038bp). Plasmid 1 contained the virulence genes rmpA, rmpA2, and iucABCD, was quite similar to the virulence plasmid pLVPK (AY378100), and had a coverage of 91% and consistency of 99.05% (Figure 5A). Plasmid 3 carried bla_KPC-2, had 99.98% homology with NTUH-K2044-CR plasmid unnamed2 (MZ475710.1) and 100% with pKPC-CR-hvKP-C789 (CP034417.1), phvKP12-KPC (CP103319.1) and pJX2-2 (CP064248.1), and query coverages of 80–100% (Figure 5B).

Figure 5 Two plasmid circle maps were generated with the GCview website. (A) Plasmid 1 carried rmpA, rmpA2, and iucABCD and was highly similar to pLVPK. The white regions indicate absence. (B) Plasmid 3 carried blaKPC-2 and shared a high identity with CP034417.1, CP103319.1, CP064248.1, and MZ475710.1 by comparison.

Discussion

This study retrospectively investigated 19 hv-CRKP-VAP cases from March 2019 to September 2021 enrolled at the EICU of a Chinese teaching hospital and analyzed the outbreak characteristics of ST11-KL64 and ST15-KL112 hv-CRKP strain. Following the analysis, 13 risk variables associated with CRE infection were identified, including infection acquired from ICU, invasive procedures, mechanical ventilation, medical devices, CRE exposure, carbapenem use, β-lactam use, cephalosporin use, glycopeptide use, quinolone use, demographic patient characteristics, underlying disease or condition, and exposure to hospital care.³⁵ To our knowledge, outbreaks of HAIs due to CRKP are very commonly reported, especially in VAP due to intubation. PFGE assays revealed that the homology of the eight kpc-2-producing ST11-KL64 CR-hvKP strains was >90%. These strains were isolated from elderly patients hospitalized in the EICU between March 2019 to June 2020, most of whom had neurological diseases. All patients underwent invasive procedures and had a morbidity and mortality rate of 73.7%, suggesting that the consequences of hv-CRKP-induced VAP are severe. Recently, infections and outbreaks caused by this pathogen have been reported in many regions.^24,36,37 This pathogen requires more attention from clinicians and infection control physicians. Since the first report of K. pneumoniae carrying OXA-232 in Shanghai, in 2017, its epidemics have been reported in China.^38–40 In this study, the four strains of OXA-232-producing ST15-KL112 hv-CRKP had >92% homology. Different from the KPC-2-producing CRE strains (MIC ≥ 16ug/mL), these strains showed intermediary sensitivity to imipenem with MIC values of 2ug/mL, possibly because of more available drug options. 50% of patients with OXA-232-producing strains improved after treatment compared with the KPC-2-producing CRE strain. A deadly VAP epidemic induced by ST11 hv-CRKP was documented in a 2017 article and occurred in a Chinese hospital.¹⁸ The stains were highly transmissible, multidrug-resistant, and hypervirulent at the same time. Genomic analyses revealed that the virulence plasmid acquired by the hv-CRKP strains was the classic ST11 CRKP stains. Simply, hv-CRKP strains have high virulence and multidrug resistance, and therefore possess a great challenge to clinical treatment and are a severe threat to public health. Fortunately, no homologous strains in VAP patients were detected after July 2020 following intervention with stringent infection control measures.

hv-CRKP consists of two evolutionary mechanisms, one in which CRKP strains acquire virulence genes or virulence plasmids, and the other in which hvKP strains acquire chromosomal or plasmid-encoded drug resistance genes.¹⁶ The strains in this investigation fall in the former type. There are three main mechanisms responsible for K. pneumoniae’s resistance to carbapenem: enzyme production, efflux pumps, and pore protein mutations.⁴¹ In China, KPC-2 is the most common carbapenemase for CRKP, and ST11 is the most common sequence type.^10,24,36 This study comprises the majority of the KPC-2-producing ST11 K. pneumoniae strains. The hvKP strains are frequently used to describe the positive strains of the “string test”. However, it has also been reported that these lack hypervirulent phenotypes. Therefore, the string test cannot be the only indicator of the hvKP strain.⁴² A study suggested that rmpA, rmpA2, iucA, iroB, and peg-344 are the most accurate molecular markers to define hvKP because these genes are all present on virulence plasmids.⁴³ Capsular polysaccharide, siderophore system, lipopolysaccharide, outer membrane protein, and efflux pump are currently recognized hvKP pathogenic components, however, the capsule has been identified as a major virulence component.⁴⁴ The rmpA and rmpA2 genes increase muco-viscosity and pathogenicity and are linked to high capsule productivity.⁴² Here, all strains had a hyper-viscous phenotype and most of them contained rmpA or rmpA2 except KP4 and KP7, indicating that other mechanisms may also affect the hyper-mucoviscosity.⁴⁵ Aerobactin is encoded by the iucABCD gene, and its corresponding receptor is expressed by the iutA. iucA encodes a IucA ortholog that is thought to bind ahLys onto the primary carboxylates of citrate to produce aerobactin siderophore.^44,46 Aerobactin is considered the critical siderophore system of hvKP, expressed in >90% of its strains but only in 6% of classic K. pneumoniae.⁴⁴ No isolate from this study expressed aerobactin, further indicating that these strains were produced by CRKP obtaining virulence genes or virulence plasmids. The bla_KPC-2 and iucABCD were also identified on plasmids by whole genome sequencing of the KP3 strain.

In vitro serum killing assays and in vivo G. mellonella larvae infection assays were conducted to assess the virulence of these strains. Six strains were selected randomly for the virulence test and all displayed high virulence in G. mellonella larvae. However, the serum survival rate of KP3, KP6, and KP13 were low. Such inconsistency in the in vitro and in vivo virulence results are not unique. Shen P also reported a similar situation.⁴⁷ The underlying mechanisms behind these differences have yet to be determined. According to a study, humans and G. mellonella have different levels of K. pneumoniae virulence.⁴⁸ For the virulence investigation, ST11-KL64 and ST15-KL112 hv-CRKP were chosen. The results revealed that ST11-KL64 included both rmpA and rmpA2, while ST15-KL112 only contained rmpA2, but the findings from serum killing assays and G. mellonella larvae infection assays were identical. The G. mellonella larvae infection model, however, has been proposed as a preliminary discriminant between K. pneumoniae strains with comparable virulence. According to Tang, the LD₈₀ of K1 and K57 was 1.0×10⁶ CFU/mL while the LD₈₀ of K2, K5, and K20 was 1.0×10⁷ CFU/mL after 96 hours of inoculation, indicating that K1 and K57 were more virulent than K2, K5, and K20.⁴⁹ Insua also distinguished between K1 and K2 pathogenicity using G. mellonella larvae.⁵⁰ Although the G. mellonella infection model has many benefits, it has been reported that the murine model is better at differentiating between hypervirulent and standard K. pneumoniae.⁵¹ Nevertheless, one study showed that G. mellonella larvae are a crucial infection model in assessing the infection process, antioxidant response, and extent of cell damage of the K. pneumoniae pathogen. Furthermore, G. mellonella larvae may be appropriate models for antibiotic susceptibility investigations as well as host interactions with resistant K. pneumoniae.⁵²

This study illustrates that hv-CRKP has caused an outbreak in hospitals, and clarified the clinical and molecular characteristics of VAP caused by ST11 and ST15 hv-CRKP. It offers suggestions for clinical prevention, control, diagnosis, and therapy. However, this study has some limitations. Retrospective analysis prevents us from gathering more detailed clinical information and taking environmental samples to understand the route of the outbreak more accurately. Unfortunately, only a few strains were used in this single-center investigation and only the entire genome of one strain was read, sequencing the genomes of all the strains would allow for a deeper understanding of the genetic traits. In addition, we did not use the murine model to assess virulence, and using only G. mellonella larvae may not reveal the real hypervirulent natures of the studied strains.

Conclusions

Our findings show the prevalence of hv-CRKP in Chinese hospitals, with ST11-blaKPC-2 being the most common. After acquiring a virulence plasmid, the strain can infect patients across many systems, which significantly raises the mortality rate. The transmission mechanism of virulence plasmids, as well as variances in drug resistance and virulence among different subtypes, are concerns that should be addressed in the future. As a result, it provides substantial data support for infection control.