Introduction

Klebsiella pneumoniae is one of the clinically relevant priority pathogens responsible for hospital- and community-acquired infections such as liver abscesses, and urinary tract infections (UTIs)1,2. Multidrug resistant (MDR) K. pneumoniae high-risk clones are important in the spread of carbapenemases locally and globally, revealing a significant public health threat3,4,5,6. The major well-studied K. pneumoniae high-risk clones include those from the multi-locus sequence typing (MLST) clonal group 258 especially, sequence type (ST)-11, ST258, and ST5127.

Klebsiella pneumoniae normally colonises mucosal surfaces such as the gastrointestinal tract, which can harbour numerous K. pneumoniae clones with antimicrobial resistance (AMR), virulence determinants, and diverse plasmids8,9. Classical K. pneumoniae (cKp) mostly include MDR high-risk clones, some of which have been reported to co-harbour carbapenemase genes, which limit therapeutic options for K. pneumoniae infections10,11,12. Hypervirulent K. pneumoniae (hvKp) are known to be susceptible to most antimicrobials however, it is more virulent compared to the cKp13. Though the MDR-cKp and hvKp populations remain non-overlapping, traits associated with both pathotypes have been described in K. pneumoniae high risk clones (ST101, ST147)14,15. The ongoing global convergence of AMR, especially carbapenemase genes and virulence determinants in K. pneumoniae epidemic clones indicates an alarming evolution, which deserves continuous monitoring to control and prevent further spread16. Whole genome sequencing (WGS) is able to provide a more accurate understanding of the evolution of these epidemic clones17.

Klebsiella pneumoniae high-risk clones have been linked to several outbreaks in South African hospitals, particularly ST307 associated with incompatibility group (Inc)-X3 plasmids carrying beta (β)-lactamase gene encoding oxacillinase (blaOXA)-1814,5,15. Studies investigating K. pneumoniae high-risk clones in South Africa focused on a single clone, and associated AMR genes4,5,6. There is paucity of information on the distribution of K. pneumoniae high-risk clones linked to AMR, and virulence determinants. This study aimed to describe the genomic characteristics of K. pneumoniae MDR high-risk clones in Gauteng, South Africa using WGS.

Methods

Bacterial isolates, identification, susceptibility and genotyping

This descriptive study analysed clinical isolates of carbapenemase-producing K. pneumoniae collected between February 2021 and May 2022. Each isolate was resistant to at least one carbapenem (ertapenem, imipenem and meropenem). Isolates were selected from a total sample of 446 obtained from inpatient and outpatient’s urine (n = 194) as well as rectal screening (n = 252) at hospital admission. The susceptibility profiles of all 446 isolates were obtained. Urine isolates were identified using matrix assisted laser desorption ionization-time of flight mass spectrometry [MALDI-TOF MS, Bruker Daltonics, United States (US)]. Antimicrobial susceptibility testing (AST) was performed by disk diffusion method using the European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoints18. The rectal carriage isolates were cultured on Chrom-ID CARBA SMART (bioMérieux, Marcy l’Étoile, France) according to manufacturer’s instructions. Microbial identification of cultured isolates was confirmed using MALDI-TOF MS (Bruker Daltonics, US) at the Ampath National Reference Laboratory (Ampath-MDRC), Pretoria. Further analysis was performed at the Department of Medical Microbiology, University of Pretoria. The AST of rectal carriage isolates was performed using the VITEK® 2 system (bioMérieux-Vitek, Marcy-l’Étoile, France). String test was performed to determine mucoid phenotype19. Isolates were subjected to rapid polymyxin Nordmann/Poirel (NP) test, microbroth dilution testing, and PCR assays for phenotypic and genotypic detection of colistin resistance respectively. Carbapenemase genes [beta (β)-lactamase gene encoding New Delhi metallo-β-lactamase (blaNDM), K. pneumoniae carbapenemase (blaKPC), Verona integron-encoded metallo-β-lactamase (blaVIM) and blaOXA-48-like] were detected using PCR assays with published primers20,21. Selection criteria are described in Fig. 1. Since ST307 K. pneumoniae high-risk clone is a well-studied AMR clone, endemic to South African hospitals, representative ST307 (n = 36) and all non-ST307 (n = 287) isolates were typed using repetitive extragenic palindromic PCR (REP-PCR) to determine genetic relatedness among isolates. Patients’ demographic data were collected. Twenty-two carbapenem-resistant K. pneumoniae (CRKp) isolated from urine (n = 9) and rectal carriage (n = 13) were selected based on the presence of the PCR detected carbapenemase genes for short-read WGS to explore the genomic context of carbapenemase-producing K. pneumoniae isolates representing different K. pneumoniae lineages circulating in Gauteng, South Africa.

Fig. 1
figure 1

Description of selection criteria for the representative isolates for whole genome sequencing from the primary data.

Full size image

Genomic DNA extraction

Genomic DNA (gDNA) was extracted from K. pneumoniae cultures using the Zymo Quick-DNA™ Fungal/Bacteria Mini prep Kit [ZYMO Research, US] according to the manufacturer’s instructions. To ensure adequate quantity and quality of extracted gDNA, the yield, concentration, and purity were evaluated using the Nanodrop spectrophotometer (ND-1000). The gDNA was stored at − 20 °C.

Whole genome sequencing

The Qubit® 3.0 fluorometer (Invitrogen, Oregon, US) was used for quantification of the extracted gDNA to a concentration of > 10 nanogram/microlitre (ng/µL). Multiplexed, paired-end libraries (2 × 150 bp) were prepared using the Illumina DNA Prep kit (Illumina, San Diego, US), followed by sequencing on the Illumina NextSeq 550 platform (Illumina, San Diego, US) with 100 × coverage.

Bioinformatics analysis

Raw sequencing reads were analysed using the Jekesa pipeline v1.0 (https://github.com/stanikae/jekesa), which included the filtering of reads (Q, ≥ 20; length, ≥ 50) with Trim Galore v0.6.10 (https://github.com/FelixKrueger/TrimGalore), and de novo assembly with SPAdes v3.13.2 (https://github.com/ablab/spades)22,23,24. Sequence types were assigned using the BIGSdb platform curated by the Institute Pasteur (https://bigsdb.pasteur.fr/klebsiella)25. Raw reads were deposited in the National Center for Biotechnology Information (NCBI) under the Bio-Project number: PRJNA922902.

Pathogenwatch was used to infer isolates core genome (cg) MLST classification, capsular serotypes, and plasmid replicons26,27,28,29,30,31,32. The AMR and virulence determinants of the genomes were determined using Kleborate30,31.

Phage prediction was performed with the PHAge Search Tool (PHASTER) for the identification, annotation, and visualisation of prophage sequences33. The 22 genomes were run as a query on the Centre for Genomic Epidemiology MobileElementFinder v.1.0.2 to determine the mobile genetic elements (MGEs) associated with AMR and virulence determinants34. ResFinder database was used to determine the molecular basis of chromosomal mutations causing resistance on genomes35.

Phylogenetic analysis of the Klebsiella pneumoniae high-risk clones

Klebsiella pneumoniae publicly available reference genomes corresponding to the STs in this study from Australia, India, South Africa, United Kingdom, and the US, retrieved from Pathogenwatch were used to construct a phylogenetic tree27. The 22 genomes, reconstructed with ROARY, and aligned with MAFFET, were compared against K. pneumoniae reference genomes within the Pathogenwatch27. This was performed via BLASTn with species-specific parameter by default based on core gene pairwise SNP distances between genomes27. This was used to construct the neighbour joining trees to demonstrate the phylogenetic relatedness of the isolates27. The phylogenetic tree was visualised in FigTree v1.4.4, and annotated using the interactive tree of life (iTol) (https://itol.embl.de)36,37. The cgMLST and cg life identification number (cgLN) code information of the genomes were used to confirm the clonality of each ST26.

Ethics

Ethical approval was obtained from the Research Ethics Committee, Faculty of Health Sciences, University of Pretoria, South Africa (Ethics Reference No: 819/2020). Since clinical bacterial isolates were used for this study, the need for informed consent to participates was waived by the Research Ethics Committee, Faculty of Health Sciences, University of Pretoria, South Africa (Ethics Reference No: 819/2020). The study was performed according to the Declaration of Helsinki. The study methods adhered to the relevant guidelines and regulations.

Results

Baseline characteristics of sequenced isolates

A total of 22 CRKp isolates were analysed. Antibiograms of isolates are shown in Table 1. Among the CRKp isolates, 13 (59%) were obtained from rectal carriage and 9 (41%) from urine (Table 2). Urine isolates were collected from one (11%) outpatient and eight (89%) inpatients, of which five (63%) were admitted to the intensive care unit (ICU). Four of the CRKp isolates were resistant to colistin: minimum inhibitory concentrations (MICs) ≥ 64 µg per millilitre (µg/mL).

Table 1 Antimicrobial susceptibility profile and carbapenemase genes associated with the sequenced isolates.
Full size table
Table 2 Characteristics of the ST307 and non-ST307 Klebsiella pneumoniae clones in this study.
Full size table

Genomic assembly, quality control and phylogenetic tree of the Klebsiella pneumoniae high-risk clones

Genome sequencing of isolates generated a collective length of contigs from assembled genomes within a length range of 5.5 mega bp (Mbp) to 6.0 Mbp, with the total assembly genome length (N50) ranging from 34 kilo bp (Kbp) to 215 Kbp. Neighbour-joining phylogenetic trees are shown in Figs. 2 and 3. Core genome SNPs grouped the genomes into eight clonal groups and eight sublineages. The phylogenetic tree (Fig. 2) showed that the CRKp isolates clustered into nine distinct STs, five of which represented ≥ 3 isolates: ST2497 (n = 5), ST17 (n = 4) and ST307 (n = 3). Comparison of the virulence determinants, plasmids, and bacteriophage diversity of the isolates is shown in Fig. 3.

Fig. 2
figure 2

Phylogenetic tree of the Klebsiella pneumoniae high-risk clones showing accessory genome. The 22 study isolates reconstructed with ROARY and aligned with MAFFET compared against reference genomes within Pathogenwatch via BLASTn. The tree branches (n = 35) indicate isolate names with respective STs (colours) and 13 K. pneumoniae reference genomes from Pathogenwatch database. The outgroup was not shown. Isolate branches tips are coloured by the metadata columns showing accessory genome (showing the presence of carbapenemase genes in filled squares), heatmap of ESBL and other acquire AMR genes in purple and red, chromosomal mutation and efflux pump genes in filled circles.

Full size image
Fig. 3
figure 3

Neighbour-joining phylogenetic tree displaying the relationship of the Klebsiella pneumoniae high-risk clones based on virulence, bacteriophages and plasmids. Relationship among isolates sequenced in this study and similar strains obtained from the Pathogenwatch database (details in Fig. 1). Public K. pneumoniae genomes were obtained from Australia, India, South Africa, United Kingdom and the US. Clonality of the isolates were confirmed using cgMLST and cgLIN code information from Pathogenwatch. Isolate branch tips are coloured by metadata columns showing the presence of virulence determinants in filled star, the commonly detected bacteriophages in filled triangle and heatmap of plasmids in green and red filled squares.

Full size image

Among the 22 sequenced isolates, three (14%) were ST307 (Table 2). Commonly detected non-ST307 K. pneumoniae clones were ST2497 (5/22; 23%), followed by ST17 (4/22; 18%). The ST17 clones were only identified from rectal carriage isolates. Most of the isolates were assigned to capsular serotype KL64 (wzi64) (ST147, ST2497), and six O1 (ST2497, ST530) antigen types were identified (Table 2). Isolates belonging to ST307 clustered with ST2975 [ST307 single-locus variant (SLV)].

Klebsiella pneumoniae high-risk clones harbouring multiple carbapenemases and other antimicrobial resistance determinants

Among the 22 sequenced isolates, 10 isolates carried two or more carbapenemase genes (Fig. 2). The ST307 UP-KT-73CKP isolate harboured blaNDM-1, blaOXA-48 and blaOXA-181. All the ST2497 strains carried blaNDM-1 and blaOXA-232. The most prevalent carbapenemase genes were blaNDM-1 (n = 11) and blaOXA-181 (n = 11). Eight out of the 10 isolates co-harbouring carbapenemases have carbapenem MICs of 16 to > 32 µg/mL (Table 1).

All isolates also harboured extended-spectrum β-lactamase (ESBL) genes (e.g., blaCTX-M-15) with high prevalence as shown in Fig. 2. Two of the four colistin-resistant CRKp isolates carried mutations in the mgrB gene, while the other isolates possess an unknown colistin resistance mechanism. Plasmid-mediated colistin resistant genes mcr-1 to mcr-9 were not detected. Chromosomal mutations were observed for parC encoding topoisomerase IV, and gyrA (n = 21) encoding DNA gyrase. Defects in the outer membrane proteins [e.g., ompK37(n = 22)] were identified.

Klebsiella pneumoniae high-risk clones ST307 harbouring unusual virulence traits

Isolate ST307 UP-KT-139UKP K. pneumoniae strain exhibited a hypermucoviscosity phenotype. The ST307 UP-KT-151UKP harboured siderophore genes encoding yersiniabactin [ybt17 on integrative and conjugate element (ICEKp1)], and colibactin genotoxin (clb3) associated with hvKp (Fig. 3).

Plasmid replicons and bacteriophages associated with Klebsiella pneumoniae high-risk clones

Several plasmid replicons (Fig. 2), most commonly IncFIB(K) (n = 14), IncX3 (n = 13), and ColKP3 (n = 13) were identified. Each isolate harboured at least two intact bacteriophages, most commonly Klebsi_ST16_OXA48phi5.4_NC_049450 (n = 8) and Klebsi_3LV2017_NC_047817(36) (n = 8).

Genetic environment of the antimicrobial resistance and virulence genes

Antimicrobial resistance gene clusters, conferring resistance to multiple antimicrobial classes, flanked by MGEs were identified in 12 isolates (Table 3). In ST147 UP-KT-105CKP isolate, blaOXA-181 and qnrS1 encoding quinolone resistance on ColKP3 plasmid was coupled with virulence gene cluster irp2; fyuA on ICEKp1in the same contig. Further details regarding the genomic characteristics of the study isolates can be found in the supplementary file 1-Table S1.

Table 3 Mobile genetic elements associated with antimicrobial resistance and virulence genes among the Klebsiella pneumoniae high-risk clones.
Full size table

Discussion

The spread of carbapenemases causing outbreaks in South African hospitals has been linked to horizontal gene transfer (HGT), and clonal expansion via diverse K. pneumoniae clonal lineages, of which few have been investigated individually4,5,6. This study described the genomic characteristics of K. pneumoniae MDR high-risk clones in Gauteng, South Africa.

In this study, ST307 isolates were clonally related, and possessed similar traits as the South African ST307 from clade VI [e.g. blaOXA-181, IncX3, KL102 (wzi173), parC and gyrA]4,5. This is evidence that similar ST307 strains are circulating in South Africa. The majority (80%) of isolates co-harbouring carbapenemases showed high carbapenem MICs (16 to > 32 µg/mL). Additionally, evolutionary development was observed among ST307 isolates: ST307 UP-KT-73CKP isolate co-harboured multiple carbapenemase genes (blaNDM-1, blaOXA-48 and blaOXA-181), similarly to a recent Chinese study, where a ST307 isolate co-harboured blaKPC-2, blaNDM-1 and blaIMI-312. The ST307 UP-KT-73CKP also harboured plasmids such as IncX3 with the highest number of bacteriophages, which might aid in its acquisition of new AMR traits, and increased virulence via HGT38,39,40. The ST307 UP-KT-139UKP isolate lacked virulence determinants, but possess iutA, which is usually associated with hvKp. Similarly, a case study in India reported a clinical isolate exhibiting hypermucoviscosity phenotype, but lacked hypervirulence determinants [e.g. regulator of mucoid phenotype (rmp)-A]41. A ST307 UP-KT-151UKP isolate acquired ybt17 on ICEKp1 and colibactin genotoxin. These virulence traits have been associated with hvKp causing invasive and life-threatening infections such as sepsis42,43. Studies have shown the presence of colibactin among ST11 K. pneumoniae clone, but it has not been linked to ST307 until now44,45. Acquisition of virulence determinants can further enhance K. pneumoniae pathogenicity45. However, increased virulence of colibactin-carrying strains in the absence of a virulence plasmid has not been well established31.

The non-ST307 K. pneumoniae high-risk clone ST2497 (n = 5) exhibited a high level of AMR harbouring double carbapenemase genes (blaNDM-1 and blaOXA-232). The ST2497 strains were linked to other AMR mechanisms, and several MGEs, consistent with a case report in the US, in which the strain was resistant to numerous antimicrobials, except tigecycline and colistin46. The ST2497 in this study exhibited pandrug resistant (PDR) and extensively drug resistant (XDR) phenotypes. This is worrisome because three of these ST2497 strains were obtained from rectal carriage of patients, who might be colonised by CRKp. This may cause person-to-person transmission within the hospital and community settings or possible development of subsequent ST2497 infection47.

Klebsiella pneumoniae co-harbouring blaNDM-1 and blaOXA-232 was first reported in the US in 2014 in ST1448. It was later reported in other STs in Italy, Malaysia, South Korea and recently in France in ST249749,50,51,52. Resistance of K. pneumoniae strains harbouring blaNDM-1 and blaOXA-232 to the last-resort antimicrobials including the combination therapy like ceftazidime/avibactam have been reported46,52. The reported resistance might be because ceftazidime/avibactam is highly active against OXA-48-like carbapenemases but lacks activity against strains producing NDM and other metallo-β-lactamases53,54. There is a need for antimicrobials active against metallo-β-lactamases-producing strains such as cefiderocol or aztreonam with ceftazidime/avibactam combinations53,54.

The ST2497 associated plasmids IncHI1B(pNDM-MAR) and Col(pHAD28) have been linked to strains co-harbouring blaNDM-1 and blaOXA-23255. The ColE-type plasmid carrying blaOXA-232 was isolated from patients returning from India, indicating travel-related transmission56. The blaOXA-232 differs from blaOXA-181 by one amino acid substitution, but both share similar genetic environments57. The presence of OXA-232 with NDM in a region like South Africa, where OXA-181 is highly prevalent might lead to a significant treatment challenge. The combination therapy available presently in South Africa are partly effective against this mechanism of resistance58. The IncHI1B-like plasmid carrying blaNDM-1 was suggested to have been transferred from Acinetobacter spp. to K. pneumoniae via transpositional events47,59. The possible transfer of AMR genes by plasmids between diverse strains and species of bacteria are a threat to human health. This present study reported the occurrence of ST2497 co-harbouring blaNDM-1 and blaOXA-232 in South Africa. Limited therapeutic options available against this clone might result in poor patient outcomes and high mortality rates. As evidenced in a recent outbreak of NDM-1 and OXA-181 CRKp among neonates resulting in a 64% (9/14) mortality rate in South Africa60. Continuous genomic surveillance of carbapenem-resistant Enterobacterales (CRE) is needed to monitor and control the spread into the community.

The ST2975 clone isolated from an inpatient in the ICU originated from ST307 via a SLV (mutation in rpoB allele). This fact highlights the potential expansion of ST307 as a persistent nosocomial pathogen61. This ST2975 may pose a risk of acquiring additional AMR traits, which might result in effective transmission of extensive AMR in hospitals, and thus community spread61. The ST2975 and a PDR ST147 UP-KT-105CKP K. pneumoniae high-risk clone were phenotypically resistant to colistin (MIC = 64 µg/mL) however, colistin resistance mechanisms were unknown. This is worrisome because colistin in combination with other antimicrobials is currently used as a last-resort treatment for severe CRE infections in South Africa. The available β-lactam/β-lactamase inhibitor combinations (e.g. ceftazidime/avibactam) in South African hospitals are expensive, and not easily accessible62. There is a need for further investigation into the colistin resistance mechanisms. It is concerning that within the same contig of the PDR ST147, AMR genes and virulence determinants were found on a ColKP3 plasmid associated with other MGEs. Similar to an outbreak in Italy linked to NDM-1-producing ST147, which harboured numerous virulence determinants14. This might be an indication of evolutionary development through mutation, which might result in the emergence of a convergence clone (MDR-hvKp)13,63.

All the ST17 strains harboured blaOXA-181, blaCTX-M-15, several plasmids and capsular serotypes. The ST17 strains were exclusively obtained from rectal carriage, which might be related to colonisation. There has been an indication that ST17 infections occurs mostly by colonising strains, rather than causing outbreaks due to clonal expansion64. The divergence of ST17 to many sublineages with several AMR genes, virulence determinants, and plasmids has been attributed to the recombination of the K and O loci64.

This current descriptive study was limited by a small sample size, so it cannot be concluded that the K. pneumoniae clones were dominant or emerging. It was inconclusive that the urine isolates were solely from patients who had UTIs due to lack of access to the clinical information. The patients screened for CRKp colonisation had no documented history of exposure to the healthcare facilities, so it cannot be concluded that the colonisation was exclusively from the community settings. Short-read sequencing did not provide full description of the plasmids as compared to long-read sequencing. However, this study provided a detailed view of the genomic and evolutionary relationship of the circulating K. pneumoniae high-risk clones in Gauteng, South Africa using short-read WGS.

Conclusion

This study revealed the importance of diverse multiple clonal lineages, and HGT through MGEs especially plasmids in the spread of carbapenemases and virulence determinants in South Africa. This study highlights the importance of local genomic surveillance systems to characterise K. pneumoniae MDR high-risk clones. The generated data will aid in designing infection and prevention measures for limiting the spread of K. pneumoniae with carbapenemases in Gauteng, South Africa.