In silico genomic analysis of resistome, virulome, and mobilome of β-lactamase-producing Klebsiella pneumoniae

Abstract

Klebsiella pneumoniae (K. pneumoniae) is a major multidrug-resistant pathogen associated with nosocomial infections, particularly among immunocompromised patients, and contributes significantly to global morbidity and mortality. This study provides a comprehensive genomic analysis of β-lactamase-producing K. pneumoniae isolates. Whole-genome sequences of clinical isolates were retrieved from the National Center for Biotechnology Information-Sequence Read Archive (NCBI-SRA) repository and systematically analyzed to characterize antimicrobial resistance (AMR) genes, virulence factors, mobile genetic elements, and defense systems. Clonal diversity was assessed using multilocus sequence typing and capsule typing. Pan-genome analysis revealed pronounced heterogeneity, with ST15 as the predominant sequence type. The presence of a large accessory genome indicates substantial genomic plasticity. In total, 244 unique AMR genes were identified, predominantly encoding β-lactamases and efflux pumps. Genes conferring resistance to penams, aminoglycosides, tetracyclines, and fluoroquinolones were also highly represented. Over 6203 insertion sequence (IS) elements were detected, mainly from the IS3 family. Integron-associated resistance genes were frequently observed, particularly those conferring resistance to β-lactams and aminoglycosides. Collectively, these findings highlight K. pneumoniae’s rapid genetic diversification and high potential for resistance, which are intrinsically linked to the complex dynamics of its resistome, virulome, and mobilome. This reinforces the urgent need for alternative therapeutic strategies, antimicrobial stewardship, and enhanced genomic surveillance to curb the increasing threat of AMR.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-56520-7.

Introduction

K. pneumoniae is a ubiquitous Gram-negative, rod-shaped, encapsulated organism that belongs to the Enterobacteriaceae family. It is a commensal organism commonly found in the gastrointestinal tract (GIT) of humans and animals, on mammalian mucosal surfaces, and in the environment. K. pneumoniae has long been recognized as an opportunistic pathogen associated with nosocomial infections such as pneumonia, meningitis, urinary tract infections (UTIs), liver abscesses, and endophthalmitis, predominantly targeting immunosuppressed individuals, thus accounting for one-third of all infections caused by Gram-negative pathogens¹. According to the World Health Organization (WHO), this bacterium is currently in the “critical priority pathogen” category among the ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp)². Over the past decade, antimicrobial-resistant (AMR) strains of K. pneumoniae have escalated at an alarming rate, leading to a surge in pan-drug-resistant strains globally^3–5. A report from the Global Burden of Diseases, Injuries, and Risk Factor Region (GBD) in 2019 stated that 1.27 million deaths worldwide were attributed to AMR, and 20% of these deaths were caused by K. pneumoniae infections⁶. Despite the use of antimicrobial therapy, fatality rates have increased to 50%, making it the leading cause of mortality and morbidity in healthcare settings^7,8.

Recently, Carbapenemase-producing K. pneumoniae (KPC) has become a pressing concern in antimicrobial therapy, owing to its high degree of resistance to all β-lactam antibiotics and extensive resistance to carbapenems^9,10. According to reports by the Centers for Disease Control and Prevention (CDC), approximately 80% of infections caused by the Enterobacteriaceae family are caused by KPC¹¹. This is particularly concerning due to the accumulation of multiple resistance mechanisms, including the production of β-lactamases, overexpression of efflux pumps, porin loss, and horizontal gene transfer (HGT), which collectively contribute to multidrug resistance (MDR) and reduced therapeutic efficacy^12,13. The dispersal of antimicrobial-resistant genes (ARG) in bacterial niches is driven by HGT facilitated by mobile genetic elements (MGEs). Integrons, plasmids, and transposons are comprised of MGEs. These constituents serve as carriers for ARGs, thus aiding their transfer across diverse bacterial pathogens^5,14. In addition to these elements, a wide array of virulence factors contributes to pathogenicity and evasion of host immune responses, including fimbriae, surface antigens (K and O types), outer membrane proteins (OmpA, OmpC, and OmpK), secretion systems, adhesins, and siderophores¹⁵.

The upsurge in KPC, extended-spectrum β-lactamases (ESBLs), and metallo-β-lactamases (MBL) strains of K. pneumoniae has become a significant matter of concern in clinical care settings. Two non-specific primary outer membrane porin proteins, OmpK35 and OmpK36, facilitate the passive diffusion of small-molecule hydrophilic substances and key therapeutic antibiotics, such as β-lactams and fluoroquinolones^16,17. Point mutations and disruptive insertions in coding sequences or promoter regions alter porin expression, thereby enhancing carbapenem resistance in clinical isolates. Strains that lack both porin proteins exhibit high levels of antibiotic resistance¹⁸. Managing these resistant strains is becoming increasingly difficult and poses a challenge for treating infections using therapeutic interventions. As specified in the CDC’s AMR action plan, among the four important initiatives, monitoring AMR organisms would help prevent the escalation and dissemination of multidrug-resistant (MDR) pathogens^19,20.

The advent of sequencing technologies has provided valuable insights into advancing gene profiling, deciphering functional mechanisms, and surveillance over traditional phenotypic and genotypic approaches²¹. Whole-genome sequencing (WGS) has emerged as a powerful tool for molecular diagnosis that provides accurate genetic information and high fidelity in organism characterization²². With the urgent public health concerns posed by K. pneumoniae, reliance on WGS has increased, as it now serves as the most reliable method for monitoring and tracking pathogenic strains²³. Contemporary sequencing methods provide in-depth data on the genotypic features of pathogens, including virulence factor identification, AMR genes, and serotypes. Additionally, WGS data can be used to estimate outbreaks caused by antibiotic-resistant strains, trace lineages, and transmission of AMR genes, thus mitigating diseases and the spread of infections²⁴.

The primary focus of this study was to identify key factors that confer resistance at the genetic level, such as ARGs, MGEs, virulence factors (VFs), CRISPR-Cas, and restriction-modification (RM) systems of β-lactamase-producing K. pneumoniae, from representative whole-genome sequences retrieved. This genomic landscape can aid future therapeutic research by providing a comprehensive understanding of emerging AMR threats and strategies to mitigate infections caused by MDR K. pneumoniae.

Results

Retrieval of genomic data, genome QC, pre-processing and De novo assembly

A total of 126 K. pneumoniae genomes were retrieved in the present study, and the metadata are provided in the Supplementary Table 1. The summarized details of the raw data statistics are provided in Supplementary Table 2a. Genome sequences were filtered based on the Phred quality threshold score of Q30, which is a widely used standard for depicting high-quality Illumina sequence reads. The benchmark score signifies a high accuracy of 99.9% and a minimal error rate of 0.1% (1 error per 1000 base calls), thus reflecting high-confidence base calls in Illumina sequencing. All redundant and duplicate reads, along with adapter sequences, were trimmed from the raw data. Thereafter, the overall quality of the sequences improved. A summary of the processed read statistics is provided in Supplementary Table 2b. De novo genome assembly of the raw sequences was performed using SPAdes, which resulted in an average genome size of approximately 5.72 Mb and an average GC content of 57.00%. The quality of genome assemblies was assessed by the N50 value, which denotes the shortest contig length required for 50% genome coverage. Statistical information related to the assembled genomes is compiled in Supplementary Table 3. Genome completeness and quality were assessed using BUSCO analysis against the relevant lineage dataset comprising 874 conserved orthologs. The majority of the genomes exhibited high completeness, with BUSCO completeness values ranging from 96.45% to 99.77%, and an overall average completeness of 98.59%. Most assemblies showed very low proportions of fragmented and missing BUSCOs, indicating high-quality genome assemblies suitable for downstream comparative genomic analyses. However, a few genomes displayed comparatively lower completeness (86.84–94.62%) with more fragmented BUSCOs, suggesting relatively lower assembly quality in those isolates (Supplementary Fig. S1). After the quality check, we filtered 118 sequences for further analysis.

Sequence typing and surface polysaccharide typing

MLST analysis revealed 117 K. pneumoniae genomes and one K. michiganensis genome. In the current study, we have considered only K. pneumoniae genomes and excluded others. This analysis using the Pasteur scheme identified 30 sequence types (STs), with ST15 the most dominant in the dataset (n = 32; 27.35%), followed by ST11 (n = 22; 18.80%) and ST11-1LV (n = 19; 16.23%) as shown in Fig. 1a. These outcomes reflect the significant clonal diversity among the K. pneumoniae strains. Capsule typing identified 24 discrete KL types. The most prevalent types were KL112 (n = 31; 26.49%) and KL25 (n = 21; 17.94%), followed by KL47, KL64, KL151, KL17, KL62, KL102, KL2, KL23, KL30, KL35, KL51, KL8, KL10, KL110, KL114, KL122, KL146, KL181, KL24, KL55, KL63, and KL9 depicted in Fig. 1b. The total number of O types detected was 9, with O1αβ,2α (n = 50; 42.73%) being the most prevalent, followed by O2α, O13, and O1αβ,2β (Fig. 1c). Comprehensive details of the sequence and capsule types are presented in Supplementary Table 4.

Fig. 1.

Distribution of sequence types and capsular types within the genomic dataset (a) Sequence type based on the Pasteur multilocus sequence typing (MLST) scheme, (b) K locus types, (c) O locus types.

Gene prediction and functional annotation

Genome annotation using Prokka identified that each genome contained an average of 5292 CDS, ranging from 4984 to 5672. Within the non-coding sequences, the annotation revealed an average of eight rRNA and 74 tRNA, as shown in Supplementary Table 5. The CDS counts classified under COG functional categories ranged from 4323 to 4669 in the 117 K. pneumoniae genomes. These findings highlight genomic architecture and functional heterogeneity, as illustrated in Fig. 2. The COG functional categorization demonstrated a predominance of genes linked to replication, protein translation, transcription, metabolic functions, transport (carbohydrate, amino acid, nucleotide, lipid, coenzyme), and signal transduction mechanisms in strains of K. pneumoniae, as outlined in Supplementary Table 6.

Fig. 2.

A heatmap showing the distribution of COG functional category profiles. The X-axis represents corresponding genomes while the Y-axis represents the COG categories.

Pan-genome and phylogenetic analyses

Pan-genome analysis revealed a total of 10,967 genes across all the genomes in the dataset. The gene distribution displayed 4017 core genes, 171 soft-core genes, 2205 shell genes, and 4574 cloud genes. Figure 3a shows the gene presence-absence matrix, and the distribution of genes is shown in Fig. 3b. A phylogenetic tree was constructed to investigate the evolutionary relationships between the 117 K. pneumoniae genomes. This is represented by a circular layout constructed using iTOL (Fig. 4). The tree revealed distinct clustering among genomes, suggesting significant genomic diversity. The tree shows multiple distinct clades, indicating a heterogeneous population structure rather than a single dominant lineage. The isolates segregated into several well-defined clusters, with closely related strains grouping together, suggesting shared evolutionary origins or potential epidemiological linkage. Notably, certain clusters comprised isolates with highly similar branch lengths, reflecting minimal genomic divergence and possible recent common ancestry. In contrast, other branches exhibited longer evolutionary distances, indicating greater genomic variability among these isolates.

Fig. 3.

Pan-genome and phylogenetic analysis (a) Gene presence-absence matrix of core and accessory genomes, (b) Pie-chart depicting the proportion of core and accessory genes across the genomes.

Fig. 4.

Phylogenetic tree of the isolates based on core gene alignment.

Assessment of AMR genes and virulence markers

Altogether, 157 distinct ARGs were identified from the CARD database across all genomes (Supplementary Table 7a). Figure 5 depicts heatmap illustrating the distribution of 42 prevalent ARGs detected in more than 30 genomes. Genes arnT, kpnE, kpnF, kpnH, ompK37, lptD, mdtQ, ompA, acrB, acrD, cpxA, emrR, eptB, marA, mdtB, msbA, CRP, H-NS were found in all genomes, whereas kpnG, acrA, and mdtC were present in > 95% of the genomes in the dataset. These genes encode efflux pumps, significantly contributing to resistance to aminoglycosides, penams, tetracyclines, and fluoroquinolones. Several other identified ARGs in the dataset conferred resistance to cephalosporins, fluoroquinolones, aminocoumarins, rifamycins, macrolides, and glycopeptides.

Fig. 5.

Heatmap illustrating the presence of antimicrobial resistance genes from the CARD database.

A total of 127 ARGs were classified from the NCBI AMRFinderPlus database based on their drug resistance mechanisms. Numerous identified ARGs were predicted to confer resistance to quinolones, β-lactams, cephalosporins, streptomycin, and trimethoprim. Other ARGs were also detected in a minimal number of diverse antibiotic classes. β-lactamase-encoding genes such as CTX-M (seven variants), OXA (nine variants), SHV (15 variants), and TEM (three variants), along with efflux pump-encoding genes, found in both databases, are primarily responsible for cephalosporin resistance. The ARG profiles of all the strains are shown in a heatmap in Fig. 6 and tabulated in Supplementary Table 7b.

Fig. 6.

Heatmap illustrating the presence of antimicrobial resistance genes from the NCBI AMRFinderPlus database.

A total of 170 virulence genes were identified in the VFDB database. The most prevalent virulence mechanisms in most strains are immunomodulatory activity, adherence, transport machinery, and metabolic pathways. Overall, 21 virulence genes (acrB, clpV/tssH, dotU/tssL, entB, entC, entE, fepA, fepB, fepC, fur, galF, hcp/tssD, iroE, iutA, ompA, rcsA, rpoS, ugd, yagV/ecpE, yagX/ecpC, and yagZ/ecpA) were identified across all genomes, highlighting the key drivers of K. pneumoniae pathogenicity. Figure 7 and Supplementary Table 7c shows the distribution of the VFs in the complete dataset.

Fig. 7.

Heatmap illustrating the presence of virulence determinants from the VFDB database.

Characterization of mobile genetic elements

Insertion sequence analysis revealed 6203 IS elements across all 117 genomes. Genome SRR37315449 exhibited the highest number of IS elements, with 94 identified. A total of 22 distinct IS families were detected in this analysis, with IS3, IS1, ISNCY, and IS5 being the predominant ones, followed by IS110, IS21, IS66, and IS6 (Supplementary Table 8). Additionally, a subset of unclassified IS elements, referred to as “new,” was reported, occurring 77 times in the dataset.

A total of 885 integrons were identified across 111 genomes, underscoring the broad prevalence of integron-associated genetic elements and suggesting notable diversity in HGT dynamics among isolates. Yielding a total of 334 attC recombination sites, attC was the most prevalent integron element among all genomes involved in gene cassette integration. intI integrase, a key component involved in integron regulation, was observed at a frequency of 108. Moreover, 131 protein-associated integron elements were found to confer resistance to various antibiotic classes. The key integron-mediated resistance determinants harbored genes linked to aminoglycoside resistance (aac(6′)-Ia, aac(6′)-Ib, ant(3′)-I, ant(3′′)-AadA1, and sat2), efflux-pump-mediated antiseptics and disinfectant resistance (qacE and qac-FL [both of the SMR family]), and resistance to β-lactams (bla_{OXA−1_like} and bla_IMP) (Table 1).

Table 1.

Distribution and functional classification of various integrons detected in the examined genomes of K. pneumoniae.

Category	Function	Integron Name	Frequency
Core integron structural elements	Gene cassettes recombination	attC	334
	Integrase	intI	108
Integron-associated resistance gene cassettes	β-lactam resistance	bla OXA−1_like	58
		bla IMP	1
	Aminoglycoside resistance	ant(3′′)-I	45
		ant(3′′)-AadA1	7
		aac(6′)-Ia	1
		aac(6′)-Ib	62
		sat2	8
	Trimethoprim resistance	dfrA12	26
		dfrA1	69
		dfrB	1
	Chloramphenicol resistance	cmlA	4
	Rifamycin resistance	rifampin-arr	2
	Macrolide resistance	ereA	1
	Efflux-pump-mediated antiseptics and disinfectant resistance	qacE (SMR family)	23
		qac_FL (SMR family)	2
Other integron-associated proteins	Miscellaneous proteins	Other proteins	131

Altogether, 41 replicon types were identified across all the K. pneumoniae genomes. The most abundant replicon types were IncR_1 (n = 80, 68.3%)), IncFIB(K)_1_Kpn3 (n = 65, 58.1%,), Col(pHAD28)_1(n = 60, 51.2%), IncFII_1_pKP91 (n = 52, 44.4%), and Col440II_1 (n = 44, 37.6%) (Fig. 8).

Fig. 8.

Distribution of the plasmid replicon types in K. pneumoniae genomes.

A total of 624 prophage regions were identified and classified into viral lineages based on their homology to known virus orthologous groups (VOGs). The majority of the prophages detected belonged to the Siphoviridae family (42.46%), followed by Myoviridae (36.21%), Podoviridae (11.53%), and unknown (9.77%) families, as depicted in Fig. 9.

Fig. 9.

Plot depicting the prevalence of taxonomic lineages of the prophage regions across all the genomes.

Identification of CRISPR-Cas and RM defense systems

A total of 81 genomes harbored CRISPR-Cas systems, with multiple Cas subtypes contributing to a cumulative count of 374 CRISPR-Cas loci. Both class 1 (CAS-I-E, CAS-IV-A) and class 2 (CAS-V-A, CAS-V-B, CAS-VI-B) systems were identified. The predominant subtype was CAS-V-A (n = 236) followed by CAS-V-B (n = 79), CAS-I-E (n = 35), and CAS-VI-B (n = 14). CAS-IV-A (n = 10) was found to have the lowest frequency among all the subtypes. Supplementary Table 9a provides overall information on the CRISPR-Cas systems detected across the genomes.

This study identified various classes of methyltransferases (MTases), enzymes that catalyze the transfer of methyl groups to specific substrates. They were identified across 117 genomes, ranging from 1 to 4 MTase(s) per genome. Restriction endonucleases (REases), which are responsible for recognizing and cleaving specific restriction sites in DNA, have also been found in all genomes. Various classes of REases were present across the genomes, ranging from 1 to 6 per genome, with specifics provided in Supplementary Table 9b.

Discussion

K. pneumoniae is a major Gram-negative pathogen in healthcare settings, particularly affecting immunocompromised patients, due to its rapid acquisition of multidrug resistance. Its genomic plasticity plays a key role in the spread of virulence factors and antibiotic resistance genes ²⁵. The genomic analysis of β-lactamase-producing K. pneumoniae in this study provides a multi-layered view of AMR, virulence potential, and genome plasticity within clinically relevant isolates. By integrating resistome, virulome, and mobilome profiling across the dataset, this work offers a systems-level perspective on the genetic architecture underlying multidrug resistance in this pathogen. Recent large-scale genomic surveillance studies in Salmonella have similarly highlighted the global dissemination patterns and temporal dynamics of AMR, emphasizing the role of clonal expansion and HGT in shaping AMR landscapes^26,27.

The presence of diverse sequence types (STs), KL, and O-types revealed genetic heterogeneity within K. pneumoniae genomes, with ST15 being the most prevalent sequence type associated with carbapenem resistance and hospital outbreaks; its predominance in the present dataset underscores its epidemiological significance^28,29. The association of dominant STs with specific capsular loci, particularly KL112 and KL25, reflects the adaptive interplay between clonal expansion and antigenic variation^30,31. Capsule diversity contributes to immune evasion, serum resistance, and biofilm formation, thereby enhancing persistence in clinical environments^32,33.

The comprehensive annotation of the genomes provides valuable insights into the conserved and variable features of K. pneumoniae. Consistent CDS content across the genome highlights the genomic stability of the organism’s core genome while still reflecting variations in the accessory genome. Such variations may contribute to virulence, AMR, and environmental adaptability among strains. Functional categorization based on COG classification revealed that a significant proportion of genes were associated with key cellular processes, such as replication, transcription, and translation, as well as metabolic processes, transport mechanisms, and signal transduction. These findings are consistent with those of previous genomic studies on K. pneumoniae by Ramirez et al., who demonstrated gene enrichment in COG categories related to important cellular and metabolic functions and processes³⁴. The pan-genome of K. pneumoniae revealed a highly diverse genetic repertoire, comprising a substantial core genome along with a wide array of accessory and cloud genes. The presence of a large accessory gene pool among the isolates is consistent with the pan-genome architecture previously described in K. pneumoniae, reflecting its extensive genetic diversity and adaptive capacity reported in larger-scale studies³⁵.

The phylogenetic analysis of β-lactamase-producing K. pneumoniae isolates revealed a genetically diverse population structure, with strains distributed across multiple distinct clades rather than forming a single dominant lineage. Closely clustered isolates with short branch lengths suggest recent common ancestry and possible clonal expansion in clinical settings, whereas the presence of distantly related branches indicates substantial genomic divergence among isolates³⁶. This heterogeneous clustering pattern implies that the dissemination of resistance is driven not only by clonal spread but also by independent acquisition of resistance determinants across diverse genetic backgrounds. Comparable evolutionary patterns have been observed in Salmonella enterica, where genomic analyses revealed dynamic shifts in dominant serovars and resistance profiles over time³⁷. Such diversity strongly supports the role of HGT in shaping the resistome and mobilome, enabling the spread of β-lactamase genes across unrelated lineages³⁸. Overall, these findings highlight the complex evolutionary dynamics of K. pneumoniae, in which both clonal expansion and genetic exchange contribute to the persistence and spread of MDR strains in clinical settings.

A key finding of this study is the extensive diversity and co-occurrence of β-lactamase genes, which constitute the primary mechanism of resistance in these isolates. Multiple variants of bla_CTX−M, bla_OXA, bla_NDM, bla_SHV, and bla_TEM were detected, frequently co-existing within individual genomes. This co-occurrence broadens the hydrolytic spectrum of β-lactamases, contributing to resistance to penicillins, cephalosporins, and carbapenems³⁹. The presence of carbapenemases such as OXA-type enzymes and NDM variants further exacerbates therapeutic challenges⁴⁰. This emphasizes the therapeutic potential of combining β-lactam with novel β-lactamase inhibitors against MDR K. pneumoniae, such as nanoparticles⁴¹. Importantly, the distribution of these genes across multiple sequence types indicates that β-lactam resistance is not confined to specific clonal backgrounds but is widely disseminated through MGEs. The emergence of plasmid-mediated resistance determinants such as mcr variants further underscores the global threat posed by transferable resistance genes⁴². Virulome analysis revealed a conserved core of 170 virulence genes, predominantly associated with immunomodulation, adherence, transport, and metabolic regulation. These factors facilitate host colonization, immune evasion, and biofilm formation, contributing to antimicrobial tolerance and persistent infection^43,44.

The resistance gene profiles identified in our isolates are consistent with phenotypic resistance patterns reported in previous studies of clinical K. pneumoniae. Published literature indicates that strains carrying bla_NDM−1/5 or bla_{OXA−48−like} genes often exhibit carbapenem minimum inhibitory concentrations (MICs) exceeding 32 µg/ml for antibiotics such as imipenem and meropenem⁴⁵. Similarly, ESBL-positive isolates harboring bla_CTX−M−15 have been associated with cefotaxime or ceftazidime MIC values ≥ 32 µg/ml. In contrast, chromosomally encoded SHV-1 β-lactamase typically confers ampicillin resistance with MIC values ranging from > 8 to > 64 µg/ml ⁴⁶. These observations, derived from previously published studies, are used here to support the functional relevance of the resistance determinants identified through genomic analysis. However, we note that direct phenotypic antimicrobial susceptibility testing was not performed on the isolates analyzed in this study, and we believe our results will be useful for experimental biologists to perform such analyses.

The widespread presence of MGEs, including IS elements, integrons, and prophages, illustrates the pivotal role of HGT in defining the genomic architecture of K. pneumoniae. The high abundance of IS3, IS1, ISNCY, and IS5 families supports ongoing genome plasticity, consistent with prior studies⁴⁷. Notably, IS3 and IS1 elements located upstream of β-lactamase genes, such as bla_OXA−232 and bla_CTX−M−15, may contribute to their transcriptional upregulation⁴⁸. Plasmid replicon typing revealed a highly diverse mobilome, with 41 replicon types identified, indicating extensive HGT among K. pneumoniae isolates. The predominance of IncR, IncFIB(K), and IncFII replicons highlights their central role in the dissemination of AMR. IncR plasmids, although non-conjugative, are frequently associated with MDR regions and are known to co-reside with conjugative plasmids, thereby facilitating the persistence and spread of key β-lactamase genes⁴⁹. Similarly, IncF-type plasmids are well-established vectors of carbapenemase and ESBL genes in K. pneumoniae, contributing to their global dissemination⁵⁰. The high prevalence of Col-type plasmids (Col(pHAD28) and Col440II) further suggests their auxiliary role in stabilizing resistance determinants, particularly through carriage of quinolone resistance genes and toxin-antitoxin systems^51,52. The frequent co-occurrence of multiple replicon types within isolates indicates plasmid compatibility and the potential for hybrid plasmid formation, thereby enhancing the accumulation and transmission of resistance and virulence traits. Collectively, these findings emphasize that the resistome of β-lactamase-producing K. pneumoniae is driven by a complex and heterogeneous plasmid pool, in which IncR and IncF backbones serve as key hubs for the maintenance and dissemination of AMR in clinical settings. Environmental studies, including analyses of hospital effluents, have further demonstrated that plasmid-mediated resistance genes are widely disseminated across clinical and environmental interfaces^53,54.

The diversity of integron-associated resistance genes across genomes implies exposure to variable selective pressures, likely driven by differing antibiotic regimens in medical facilities. Aminoglycoside resistance genes were most common, followed by those conferring resistance to β-lactams, trimethoprim, chloramphenicol, rifamycin, and macrolides. Notably, the integron-mediated qacE, found in over half of the genomes, encodes a small multidrug resistance (SMR) protein belonging to the transporter family. Commonly located on integrons and plasmids, qacE facilitates widespread dissemination of resistance determinants⁵⁵. The QacE efflux pump confers resistance to quaternary ammonium compounds (QACs), reducing the effectiveness of antiseptics and disinfectants and raising concerns about disinfectant efficacy in clinical environments. These findings underscore the need for enhanced hygiene strategies, including the use of disinfectants with alternative mechanisms of action, such as chlorine- or hydrogen peroxide-based agents, particularly in high-risk clinical settings⁵⁶. Additionally, the adoption of disinfectant rotation strategies, strict adherence to recommended contact times, and routine evaluation of disinfectant efficacy are essential to minimize selective pressure and mitigate the emergence of disinfectant-tolerant bacterial populations, particularly in K. pneumoniae, where disinfectant exposure has been shown to promote adaptive resistance and co-selection of AMR determinants⁵⁷. The detection of qacE also highlights the importance of incorporating genomic monitoring of disinfectant resistance determinants into infection control programs, enabling early identification of strains with reduced susceptibility and supporting evidence-based infection prevention policies^58,59. Furthermore, prophage analysis revealed that Siphoviridae was the most predominant viral lineage across the genomes, corroborating the findings of Bleriot et al., thereby highlighting the potential role of bacteriophages in mediating HGT and contributing to genomic diversification in K. pneumoniae⁶⁰.

The mobilome analysis provides mechanistic insight into the dissemination of resistance determinants. The high abundance of insertion sequences, particularly from the IS3 and IS1 families, indicates active genome restructuring and potential modulation of adjacent gene expression. These elements are known to influence the mobilization and upregulation of β-lactamase genes, thereby amplifying resistance phenotypes^61,62. Similarly, the widespread presence of integrons carrying aminoglycoside resistance genes, β-lactamase-associated cassettes, and disinfectant resistance determinants highlights their role as efficient platforms for gene capture and dissemination⁶³.

The widespread distribution of CRISPR-Cas and RM systems across all genomes highlights the importance of bacterial defense mechanisms in safeguarding genomic integrity and regulating HGT. Most CRISPR-Cas loci belonged to class 2 (subtype V-A), consistent with previous findings indicating the increasing prevalence of type V systems in Enterobacteriaceae. Although CRISPR-Cas systems are generally reported at low frequency in K. pneumoniae, particularly in MDR clinical isolates, our analysis revealed their presence across all genomes. This observation suggests that certain lineages (e.g., ST15 and ST11) may retain functional CRISPR arrays despite extensive acquisition of resistance determinants. Recent studies suggest that some K. pneumoniae strains, especially those with intermediate resistance profiles or plasmid-host equilibrium, may preserve CRISPR-Cas elements to maintain genomic stability by limiting acquisition of additional MGEs^64,65. Moreover, their persistence in MDR strains suggests that CRISPR-Cas activity may be modulated, allowing selective acquisition of beneficial MGEs while limiting the acquisition of deleterious DNA. Such regulation could enable the maintenance of essential resistance determinants without imposing excessive fitness costs associated with uncontrolled gene acquisition⁶⁶. The co-occurrence of CRISPR-Cas systems and MGEs plays a critical role in shaping the evolutionary trajectory of MDR strains by influencing plasmid compatibility, genomic stability, and the persistence of resistance genes within specific clonal lineages. The simultaneous presence of CRISPR-Cas systems and a diverse mobilome in the analyzed genomes suggests a regulated balance between defense against foreign genetic elements and the selective acquisition of beneficial genes via HGT. This balance likely enhances bacterial adaptability and promotes survival in antibiotic-rich clinical environments⁶⁷.

The detection of diverse RM systems, including multiple methyltransferases and restriction endonucleases, indicates an additional layer of genome defense⁶⁸. Together, CRISPR-Cas and RM systems form a multilayered barrier against foreign DNA, although their activity may be selectively attenuated in clinical strains in which the acquisition of ARGs confers a survival advantage. These findings indicate that defense systems in MDR K. pneumoniae are not merely protective but are likely finely regulated to optimize both resistance acquisition and genomic stability under selective pressure.

This study highlights the clinical relevance of MDR K. pneumoniae, characterized by the co-occurrence of carbapenemases, efflux pumps, and key virulence factors. The detection of genes conferring resistance to antiseptics and disinfectants highlights challenges for infection control and underscores the need to re-evaluate existing disinfection protocols. The detection of diverse MGEs further underscores the pathogen’s capacity to disseminate resistance. These findings support the potential of β-lactamase and efflux pump inhibitors as alternative strategies and reinforce the need for enhanced surveillance and judicious antimicrobial utilization. These findings also highlight the practical utility of genomic profiling in routine clinical microbiology laboratories. Integration of WGS into diagnostic pipelines can enable rapid identification of high-risk clones, such as ST15, along with simultaneous detection of critical resistance determinants, including β-lactamase genes and disinfectant resistance markers. In routine practice, this can be implemented through standardized bioinformatics pipelines incorporating tools for sequence typing, ARG detection, and plasmid profiling, thereby generating clinically actionable reports within relevant turnaround times. Such genome-informed surveillance can support outbreak detection, trace transmission pathways within healthcare settings, and guide targeted infection control interventions. Furthermore, incorporating genomic data into antimicrobial stewardship programs can facilitate evidence-based therapeutic decision-making by enabling clinicians to tailor treatment strategies to the specific resistance profiles of circulating K. pneumoniae strains. With decreasing sequencing costs and increasing availability of automated analysis platforms, implementation of WGS-based surveillance is becoming increasingly feasible in routine clinical microbiology laboratories.

Materials and methods

Genomic data collection

A total of 126 unannotated raw genomic sequences of β-lactamase-producing K. pneumoniae were retrieved from five BioProjects in the National Center for Biotechnology Information-Sequence Read Archive (NCBI-SRA) repository, using data selection criteria specific to clinical samples from developing countries. The particular dataset specifically includes strains evaluated for their resistance to carbapenems and β-lactam/β-lactamase inhibitor combinations. To evaluate the significance of the strains and illustrate the emerging drug resistance patterns, we compared the genomes with globally existing MDR strains of K. pneumoniae⁶⁹. All sequences were retrieved using the NCBI SRA toolkit⁷⁰. The overall workflow is illustrated in Fig. 10.

Fig. 10.

Schematic representation of the computational workflow for comprehensive genomic characterization of β-lactamase-producing Klebsiella pneumoniae.

Sequence quality check (QC) and pre-processing

First, the data quality of raw genomic sequences was analyzed using the FastQC toolkit, and subsequently, QC reports were consolidated using the MultiQC toolkit⁷¹. All adapter contaminants (including the sequencing library and Illumina universal adapters) along with low-quality regions and redundant sequences were trimmed using Trimmomatic v0.39⁷² with default parameters and a quality cut-off Phred score of Q30. Furthermore, to refine the dataset, low-quality reads and short sequences (< 50 bp) were eliminated. The read statistics of the raw and processed sequences were generated using Seqkit v2.10.0 ⁷³.

De novo genome assembly and sequence quality assessment

De novo whole-genome assembly of all pre-processed reads was performed using the St. Petersburg genome assembler (SPAdes) v3.15.5⁷⁴ with a K-mer range of K21-K71. Paired-end sequencing reads were used as input for genome assembly. Multiple steps were carried out by SPAdes, including generation of the de Bruijin graph, read error and anomaly correction, and scaffolding sequences as a part of the genome assembly pipeline. The genome assemblies were assessed using Quality Assessment Tool (QUAST) v5.3.0⁷⁵ with default settings, while the genome completeness was determined using Benchmarking Universal Single-Copy Orthologs (BUSCO) v5.8.3⁷⁶.

Multi-locus sequence typing (MLST) and surface polysaccharide typing

MLST and surface polysaccharide typing were performed using PathogenWatch v23.2.0 ⁷⁷. MLST profiling was based on the housekeeping genes of the Pasteur scheme (gapA, infB, mdh, pgi, phoE, rpoB, and tonB) from PubMLST⁷⁸. For surface polysaccharide typing, capsular polysaccharide loci (KL) and lipopolysaccharide O-antigen loci (O) were based on Kaptive databases⁷⁹.

Gene prediction and genomic annotation

Genetic features and functional annotations of the assembled genomes were predicted using Prokka v1.14.6 ⁸⁰. Prokka utilized BLAST and HMMER3 searches to annotate the genomes and Prodigal to identify coding sequences (CDS). Prediction of non-coding RNA elements and signal sequences was performed by integrating additional tools, including RNAmmer (rRNA), Aragon (tRNA), SignalP (signal peptides), and Infernal (ncRNA). Functional annotation of the predicted genes was carried out using COGclassifier v1.0.5⁸¹ to group them based on different Cluster of Orthologous Genes (COGs).

Pan-genome analysis

Pan-genome analysis was conducted using Panaroo v1.5.1⁸². Core gene alignment and gene presence-absence matrix were constructed using the annotated general feature format (GFF) files generated from the Prokka output. Pan-genome genes were categorized into four clusters: core, accessory, unique, and exclusively absent.

Phylogenetic analysis

A phylogenetic tree was constructed using IQ-TREE v3.0.0⁸³. The tool used core gene alignment generated with Panaroo as input and employed the General Time Reversal (GTR) substitution model for tree construction. The phylogenetic tree was visualized using the Interactive Tree of Life (iTOL) v7 with custom parameters to achieve consistent representation across the genomes.

Profiling of antimicrobial resistant genes (ARG) and virulence factors (VF)

The identification of ARGs and VFs was conducted using ABRicate v1.0.1⁸⁴ against three reference databases, viz., Comprehensive Antibiotic Resistance Database (CARD)⁸⁵, NCBI AMRFinderPlus database⁸⁶, and Virulence Factor Database (VFDB)⁸⁷. CARD and NCBI were used to detect ARGs, whereas pathogenicity-associated VFs were detected using VFDB.

Identification and analysis of mobile genetic elements (MGE)

All prokaryotes harbor autonomous transposable elements known as insertion sequences (IS). ISEscan v1.7.3 was employed to detect IS elements⁸⁸. This tool uses the profile Hidden Markov Model (pHMM) to recognize both conserved and novel transposase-like proteins (Tpases) in distantly related sequences. Integrons possess the ability to transfer genes and can consist of several cassettes containing accessory genes, the order of which is rearranged by a particular integrase enzyme. Integron sequences were identified using IntegronFinder v2.0.5 ⁸⁹. Additionally, plasmids present in the genomes were identified using ABRicate with the PlasmidFinder database. Finally, the prophage regions were recognized using Phigaro v2.4.0⁹⁰.

Identification of CRISPR-Cas and restriction-modification (RM) systems

CRISPR-Cas and RM systems act as protective barriers against bacteriophage and plasmid invasion. CRISPR-Cas loci were identified and classified using CRISPRcasIdentifier v1.1.0⁹¹, whereas RM systems were identified using DefenseFinder v2.0.0⁹². Both tools leverage machine learning algorithms, such as Extremely Randomized Trees (ERT) and HMM2019, for CRISPR-Cas subtype classification and for identifying conserved domains in phage-resistant proteins.

Conclusion

K. pneumoniae is a globally prevalent opportunistic pathogen associated with life-threatening infections and mortality. The evolution of novel MDR β-lactamase-producing strains has become an emerging health concern. This study provides a comprehensive overview of the genomic architecture of diverse STs of MDR K. pneumoniae, emphasizing the heterogeneity of its AMR genes, virulence determinants, and MGEs. The findings revealed a significant prevalence of carbapenem resistance and efflux pump genes co-occurring with a wide array of MGEs. The emerging hypervirulent clone, ST15, was notably enriched with gene clusters encoding yersiniabactin and aerobactin, which are key virulence determinants. A broad spectrum of VFs, instrumental in facilitating immune evasion, biofilm formation, and conferring antibiotic resistance, highlights the inherent challenges of clinical intervention. The genomes carry ARGs that confer resistance to penams, aminoglycosides, tetracyclines, and fluoroquinolones, predominantly via efflux pumps and β-lactamase production. Moreover, the identification of MGEs, such as integrons, insertion sequences, and prophages, underscores the pivotal role of HGT in mobilizing resistance determinants within bacterial populations. A large accessory genome that contributes to an open pan-genome window emphasizes the importance of HGT in the rapid acquisition of this pathogen’s adaptive traits. Therefore, this study highlights the pressing need to develop novel therapeutic interventions that specifically target β-lactamases, efflux pumps, and their associated regulatory elements. Sustained and improved surveillance, stringent infection control strategies, and antimicrobial stewardship of K. pneumoniae strains are vital for detecting emerging resistance patterns and managing potential outbreaks caused by high-risk MDR lineages.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

R.B.; Formal Analysis, Visualization, Writing-original draft preparation, Validation, Data curation, Formal Analysis. A.A.; Conceptualization, Writing-review and editing, Funding acquisition, Supervision.

Funding

Open access funding provided by Vellore Institute of Technology. This work was supported by the Indian Council of Medical Research (ICMR) (IRIS ID: 2021–10630). Author Dr. Anand Anbarasu has received research support from the ICMR.

Data availability

Raw sequence data related to this study are retrieved from the National Center for Biotechnology Information-Sequence Read Archive (NCBI-SRA) repository. All data generated or analyzed during this study are included in this article (and its Supplementary Information files).

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Informed consent

Not applicable.

Clinical trial number

Not applicable.