Genetic and Phenotypic Characteristics of Carbapenem-Resistant Klebsiella pneumoniae Isolates from a Tertiary Hospital in Beijing

Introduction

Carbapenem-resistant Klebsiella pneumoniae (CRKP) comprise a set of Gram-negative bacteria that cause serious nosocomial infections, including pneumonia, meningitis, liver abscess, wound and urinary tract infections, especially in immunodeficient patients.^1,2 The rising prevalence and increasing multidrug resistance of CRKP are of critical concern.³ A World Health Organization report on antimicrobial resistance disclosed that CRKP has a worldwide geographic distribution, and that the prevalence of CRKP among K. pneumoniae isolates exceeds 50% in some patient groups.⁴ For example, carbapenem resistance among K. pneumoniae isolates in Greece increased from 0% in 2003 to 38.3% in 2010.⁵ Surveillance reports from CHINET show that the prevalence of meropenem resistance among K. pneumoniae isolates in China has increased annually, from 2.9% in 2005 to 26.3% in 2018.⁶

Carbapenems are used to treat severe infections caused by multidrug-resistant Enterobacteriaceae (including K. pneumoniae). The emergence of carbapenemases and extended-spectrum beta-lactamases (ESBL) in K. pneumoniae has reduced bacterial sensitivity to almost all beta-lactams,⁷ which confounds treatment and jeopardizes clinical outcomes.

bla_KPC-2 is the major carbapenemase gene in Enterobacteriaceae isolated from Chinese adults and children.^6,8 bla_NDM-positiveK. pneumoniae isolates often produce metallo-β-lactamase⁹ and demonstrate high-level resistance to all carbapenems.¹⁰ Understanding the types and distribution of K. pneumoniae antibiotic resistance genes (ARGs) facilitates hospital infection control and deserves attention.¹¹

We employed Illumina and Nanopore sequencing to analyze the genomic epidemiology of CPKP clinical isolates. Our findings characterized the genomic and resistance features of K. pneumoniae and revealed an evolving genotype in isolates from a single patient, and may guide prevention and treatment.

Methods

Bacterial Classification and Antimicrobial Susceptibility Testing

Samples were collected in a tertiary hospital in Beijing, China, from June to December 2021. Twenty-four CRKP isolates were obtained from sputum (11/24), urine (7/24), blood (3/24), catheters (1/24), secretions (1/24) and throat swabs (1/24). The samples were collected from 15 (62.5%) men and 9(37.5%) women between 50 and 94 years old. A fully automated analysis system (PHOENIX-100 [BD, USA]) was used for species identification and antimicrobial susceptibility testing. Susceptibilities to 18 commonly used antibacterial agents were determined; these included ampicillin, amoxicillin/clavulanate, ampicillin/sulbactam, piperacillin/tazobactam, cefazolin, cefepime, cefotaxime, ceftazidime, meropenem, imipenem, gentamicin, amikacin, ciprofloxacin, moxifloxacin, levofloxacin, trimethoprim/sulfamethoxazole, tetracycline, and aztreonam. Results were interpreted according to the breakpoints and reporting method of the Clinical Laboratory Standards Institute (2021 version, https://clsi.org).

Whole-Genome Sequencing Using Illumina and Nanopore

Genomic DNA was extracted from cultured bacteria. The Illumina sequencing library was prepared using NEBNext Ultra II FS DNA Library Prep Kit for Illumina (New England BioLabs, CA, USA) according to the manufacturer’s instructions. Sequencing was performed on the Illumina NovaSeq 6000 platform at Novogene company with a strategy of pair-end 150bp. Nanopore sequencing libraries were prepared using the Ligation Sequencing Kit SQK-LSK109 (Oxford Nanopore Technologies, Cambridge, UK). The prepared libraries were sequenced on a MinION MK1B sequencer (Oxford Nanopore Technologies, Cambridge, UK) with an R9.4 flow cell.

Dry Lab: Bioinformatics Analysis

The term dry lab denotes the bioinformatic portion of the technology.¹² Illumina and Nanopore sequencing data were filtered by Fastp¹³ (version 0.19.7) and NanoFilt (v2.8.0). Hybrid assembly of clean data was conducted with Unicycler (v0.4.8).¹⁴ The ARGs of CRKP were identified using the ABRicate (1.0.1) and Resfinder¹⁵ databases (https://cge.food.dtu.dk/services/ResFinder/). Kleborate (v2.0.1)¹⁶ was used for multilocus sequence typing (MLST). Plasmids were identified using the PlasmidFinder¹⁷ database. Genomes were annotated using Prokka (1.14.6).¹⁸ Core genome alignments were created by Roary,¹⁹ and maximum likelihood phylogenetic trees were generated using RAxML-NG²⁰ (v1.0.3), visualized using FigTree (v1.4.4) and the R ggplot2 package. Plasmid genome circle map visualization using proksee (https://proksee.ca).

Results

CRKP Isolation and Antimicrobial Susceptibility Testing

From June to December 2021, 24 CRKP isolates were obtained from sputum (11/24), urine (7/24), blood (3/24), catheters (1/24), secretions (1/24), and throat swabs (1/24). Specimens were primarily from the Intensive Care Unit (ICU) (17/24), followed by neurology (3/24), neonatology (2/24) and nephrology (2/24) wards. All strains were ESBL-positive. Three strains (BJ10-0010, BJ10-0013 and BJ10-0015) were isolated from urine, blood, and sputum of the same ICU patient in June, September, and October 2021, respectively. The samples were collected from 15 (62.5%) men and 9(37.5%) women between 50 and 94 years old. Most strains were resistant to multiple agents, including meropenem and imipenem (Table 1). All 24 strains (100%) were resistant to ampicillin, cefazolin, amoxicillin/clavulanate, ampicillin/sulbactam, meropenem, piperacillin/tazobactam, cefepime, cefotaxime and ceftazidime, but most (66.67%) were sensitive to amikacin. Interestingly, the three strains from the same patient showed slightly different levels of imipenem resistance. The MIC values of the 22 strains are provided in Supplementary Table S1.

Table 1 Antimicrobial Susceptibility of K. pneumoniae Isolates

Genomic Characterization of CRKP

The genomes of 24 CRKP strains were obtained by hybrid assembly using Illumina and nanopore sequencing, and have been deposited in NCBI under the BioProject accession number PRJNA874878. Whole genome MLST identified five sequence types. Most strains belonged to ST11 (19/24), and others belonged to ST15 (2/24), ST37 (1/24), ST403 (1/24) and ST1530 (1/24). Genomic annotation identified diverse carbapenemase- and β-lactamase-encoding ARGs. The bla_LAP-2 (13/24), bla_CTX-M-65 (7/24) and bla_SHV-12 (7/24) genes were chromosomal, while bla_KPC-2 (13/24), bla_TEM-1B (12/24), bla_OXA-1 (8/24), bla_NDM-5 (2/24) bla_SHV-12 (2/24) and bla_CTX-M-3 (1/24) bla_CTX-M-65 (1/24) were plasmid-encoded. The distribution of ARGs of strains of different STs is shown in Figure 1.

Figure 1 Distribution of antibiotic resistance genes, plasmids and STs of all strains. Colors represent different clades. Red represents subclade 1, blue represents subclade 2, and purple represents subclade 3. The antique white squares represent corresponding ARGs or replicons.

Distribution of Antibiotic Resistance Genes Among Plasmid Replicon Types

Plasmid incompatibility groups were determined according to replication sequences. The strains contained multiple plasmid replicon types, most of which distributed in ST11 strains. All ST11 strains containing plasmid replicons IncFII _(pHN7A8). IncFIB _(K), IncFIB _(pKPHS1) or IncFIB _(pNDM-Mar) were distributed exclusively in subclade1, while IncR was carried only in subclades 2 and 3. Twenty-five plasmids contained single replicons, and 14 carried multiple replicons. On average, each single-replicon plasmid carried 1.52 ARGs, and each multi-replicon plasmid encoded 2.71 ARGs. Plasmids containing multiple-replicons IncFIB_(K)/IncFII_(pHN7A8) and IncFIB_(K)/IncFIB_(pKPHS1) carried bla_KPC-2, bla_TEM-1B, and rmtB. The plasmids of BJ10-0013 included the IncFII _(K) replicon and encoded eight ARGs, the most of all plasmids.

Comparison of BJ19-0010, BJ19-0013 and BJ19-0015 from a Single Patient

These three strains were isolated from different samples from the same patient. Interestingly, BJ19-0010 and BJ19-0015 belonged to ST11 and were resistant to ciprofloxacin and levofloxacin, while strain BJ19-0013 belonged to ST1503 and was fluoroquinolone-sensitive. In the phylogenetic tree, BJ19-0010 and BJ19-0015 fell in subclade 1, while BJ19-0013 comprised a distinct branch. Strains BJ19-0010, BJ19-0013 and BJ19-0015 had diverse plasmid replicons. Interestingly, BJ19-0010 and BJ19-0015 shared IncFII _(pHN7A8), while BJ19-0010 carries IncFIB_(K)/IncFIB_(pNDM-Mar) and BJ19-0015 carries IncFIB _(K). Both BJ19-0010 and BJ19-0015 carried bla_KPC-2, bla_LAP-2 and bla_OXA-1. The bla_KPC-2 and bla_OXA-1 genes were plasmid-encoded, while the bla_LAP-2 gene was located on the chromosome. Strain BJ19-0010 also contained two copies of bla_TEM-1B that were located on different plasmids. However, BJ19-0013 carries IncX3 and IncFII_(K), which are different from the other two strains. The hybrid assembly showed that the IncFII_(K) plasmid pBJ1913-TEM had a length of 127942bp and carries the β- Lactams ARGs bla_TEM-1B and bla_CTX-M-3(Figure 2A). The other plasmid, pBJ1913-NDM5, had a length of 46068bp and carries the carbapenem-resistance gene bla_NDM-5 (Figure 2B), which is nearly identical with plasmid pR15_NDM-5 in Escherichia coli strain R15 isolated from wastewater treatment plant in China.

Figure 2 The genome circle map of plasmids pBJ1913-TEM (A) and pBJ1913-NDM5 (B) carried by strain BJ19-0013. The blue color in the outer circle represents the plasmid replicon protein, the red color represents the drug resistance gene region, and the Orange color represents the protein coding region. The inner-circle indicates the positive and negative GC skew of the plasmid in green and purple, respectively.

Discussion

Carbapenems are usually effective for the treatment of drug-resistant bacterial infections. However, CRKP can express plasmid-encoded carbapenemases, which inactivate most β-lactam antibiotics and confer carbapenem resistance.²¹ Our study showed that 24 CRKP strains were resistant to most tested drugs. The resistance rate to fluoroquinolones was relatively high, which was consistent with the results of a previous study.²² Three strains were isolated from different samples from the same patient within 4 months and carried different carbapenemase genes, which were located on separate plasmids. Genome annotation revealed that Tn3 was located upstream of the bla_KPC-2 gene on chromosomes of BJ19-0010 and BJ19-0015, while Tn552 was located downstream of bla_NDM-5 in BJ19-0013. The Tn3 transposon promotes the dissemination of AMR,²³ and Tn552 has a co-integrate resolution system homologous to Tn3 family elements,²⁴ which may have facilitated plasmid and ARG exchange among these three strains. A previous study showed that bla_KPC−2-positive K. pneumoniae isolates in China belonged primarily to ST11.²⁵ In this study, the majority of CRKP strains belonged to ST11, all of which contained bla_KPC-2. Most (68.42%, 19/24) ST11 strains were isolated from ICU patients; consequently, the monitoring of AMR strains in ICUs should be strengthened. In K. pneumoniae, the plasmid with an IncR replicon was identified as a multidrug resistance vector with variable ARGs.²⁶ ST11 strains in this study contained the IncR and IncFII _(pHN7A8) replicons with bla_CTX-M-65, bla_TEM-1B, bla_KPC-2, and rmtB genes.

Conclusion

Our study revealed the prevalence of CRKP strains isolated primarily from ICU patients. All 24 strains (100%) were resistant to β- Lactams and carbapenems and carried diverse plasmids with different ARGs. The rapid dissemination of ST11 strains highlights the necessity of strengthening CRKP surveillance and may provide a theoretical basis for infection control and treatment.