Genomic architecture of fourth-generation cephalosporin-resistant Klebsiella pneumoniae Species Complex (KpSC) from patients and their hospital environment in Uganda

Abstract

The emergence of fourth-generation cephalosporin-resistant (4GCR) K. pneumoniae species complex (KpSC) is a human health concern due to limited therapeutic options and association of the bacteria with severe morbidity and mortality. This study investigated the genomic characteristics of 4GCR K. pneumoniae species complex strains from patients and their hospital environment in Uganda. Twenty-seven isolates were obtained from two tertiary healthcare hospitals in Uganda. Whole genomic sequence (WGS) analysis revealed dominance of phylogroup Kp1 (70.4%). Isolates were highly diverse genetically, representing 15 clonal groups and 10 different serotypes, with isolates of CG17 (6/27) and serotype O5 (25%) as the most common. Core genome SNP-based phylogenetic comparison placed Ugandan strains together with strains from African lineages; however, with a few strains clustering with global references. Most strains carried multiple resistance genes, particularly CTX-M-15 (24/27) and aminoglycoside-modifying enzymes. A search for virulence factors revealed that most isolates carried few virulence genes, particularly those associated with hypervirulence. A yersiniabactin loci was detected in a subset of Kp1 isolates. Ten plasmid replicons and multiple insertion sequences were detected that may mediate resistance dissemination. These findings provide evidence of 4GCR KpSC in clinical settings and patient environments in Uganda. This underscores the need for ongoing genomic surveillance and antimicrobial stewardship to enable early detection of resistant strains and prevent localized clusters from escalating into widespread outbreaks in hospitals across the country.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-026-05232-z.

Introduction

The global rise of antimicrobial resistance (AMR) among Enterobacteriaceae poses a significant public health threat, with K. pneumoniae species complex (KpSC) recognized as a leading cause of healthcare-associated infections [1–3]. The emergence of fourth-generation cephalosporin-resistant (4GCR) K. pneumoniae species complex has been particularly concerning due to limited therapeutic options and association with severe morbidity and mortality. In healthcare settings, both patients and the hospital environment serve as reservoirs and transmission hubs for resistant strains, yet the mechanisms driving persistence remain incompletely understood, particularly in low- and middle-income countries (LMCIs). Whole-genome sequencing offers a powerful approach to delineate the genetic architecture underlying the K. pneumoniae species complex, resistance traits, and to trace potential transmission pathways, and to inform infection prevention and control. K. pneumoniae species complex is differentiated into seven phylogroups, and they include: K. pneumoniae sensu stricto (Kp1), K. quasipneumoniae subsp. quasipneumoniae (Kp2), K. variicola (Kp3), K. quasipneumoniae subsp. similipneumoniae (Kp4), K. variicola subsp.tropicalensis (Kp5), K. quasivariicola (Kp6), and K. africanensis (Kp7) [4].

Among the members of the K. pneumoniae species complex (KpSC), K. pneumoniae (Kp1)has emerged as the most important pathogen in both community and healthcare settings, largely due to its remarkable ability to acquire and disseminate resistance determinants [5–7]. This organism is implicated in a wide spectrum of infections, including bloodstream infections, urinary tract infections, ventilator-associated pneumonia, and surgical site infections, with particularly high prevalence in intensive care units and among immunocompromised patients [8]. The clinical burden of K. pneumoniae infections is compounded by the increasing incidence of multidrug-resistant strains, which significantly limits the effectiveness of standard antimicrobial therapies and is associated with elevated morbidity, prolonged hospital stays, and increased mortality rates [8, 9]. K. quasipneumoniae subsp. Similipneumoniae, K. quasipneumoniae subsp. quasipneumoniae and K. variicola have the capacity to persist in the hospital environment and accumulate multidrug-resistance determinants, including carbapenemases, polymyxins, and tigecycline, making them notable members of the species complex [6, 10].

In particular, strains resistant to fourth-generation cephalosporins (4GCR) have attracted considerable attention. Fourth-generation cephalosporins, such as cefepime, represent critical therapeutic options for severe Gram-negative infections due to their broad-spectrum activity, stability against certain beta-lactamases, and relative safety profile [11]. The emergence of 4GCR K. pneumoniae therefore signifies a substantial clinical threat, as treatment alternatives are often restricted to last-resort agents, including carbapenems, polymyxins, and newer combination therapies, which may be less accessible in LMICs [12]. The spread of 4GCR phenotypes has been documented worldwide, with notable outbreaks in Europe, Asia, and sub-Saharan Africa, highlighting both the pathogen’s adaptability and the global nature of the AMR crisis [13].

The epidemiology of 4GCR K. pneumoniae is shaped by a complex interplay between human hosts, healthcare environments, and bacterial genomic evolution [14]. Hospitals, particularly high-dependency units, serve as amplifiers and reservoirs of resistant strains, where colonized patients, contaminated medical equipment, and environmental surfaces contribute to pathogen dissemination [15]. The risk of transmission is exacerbated by factors such as prolonged hospitalization, invasive procedures, and inadequate infection prevention and control measures. Evidence indicates that colonization of patients often precedes clinical infection, suggesting that asymptomatic carriage within the gut microbiome or on mucosal surfaces may serve as a hidden reservoir for resistant K. pneumoniae [16]. Concurrently, the hospital environment, including sinks, ventilators, and bed rails can harbor viable bacteria for extended periods, facilitating indirect transmission between patients [17]. In addition to the antibiotic selection pressure, more isolates with combined hypervirulence (hv) and 4CGR have been detected and have reached public attention as “superbugs” with the possibility of causing untreatable invasive infections. The convergence of the classical K. pneumoniae (cKp), which is commonly responsible for nosocomial infections, and the hypervirulent type (hvKp) associated with community-acquired diseases and has increasingly infiltrated healthcare environments is posing a more serious threat, further causing challenges for its management [18].

Despite the critical importance of understanding 4GCR KpSC, there remains a paucity of evidence on the extent of the problem caused by such bacteria in LMICs, where resource limitations often constrain both diagnostic and surveillance capacity. Most studies have been concentrated in high-income countries, potentially overlooking the diversity and epidemiological dynamics of strains circulating in Africa and other resource-limited regions [19]. In Uganda, for instance, hospitals are increasingly reporting multidrug-resistant K. pneumoniae, yet detailed investigations into the genomic determinants of resistance, virulence, and persistence remain limited [19–23]. The lack of such data hampers efforts to implement targeted antimicrobial stewardship programs and infection control measures and limits the ability to identify potential outbreaks before they escalate into public health emergencies.

This study, therefore, aimed to characterize the genomic features of 4GCR KpSC isolated from both patients and the hospital environment in tertiary healthcare facilities in Uganda. By employing WGS, the investigation sought to elucidate the distribution of resistance determinants, the presence and diversity of mobile genetic elements, and potential epidemiological linkages within clinical settings. The findings are expected to enhance our understanding of the genetic and ecological factors driving the emergence and persistence of resistant KpSC strains, thereby informing strategies to mitigate their spread and improve patient outcomes.

Materials and methods

Study sites

The study was conducted in two hospitals located in the Masaka district of Central Uganda: Masaka Regional Referral Hospital (referred to as Masaka Hospital) and St. Joseph’s Hospital Kitovu (referred to as Kitovu Hospital). Masaka Hospital is a public tertiary healthcare facility operated by the Government of Uganda and serves as one of the regional referral hospitals in the country. In contrast, Kitovu Hospital is a private not-for-profit community hospital located in Kitovu town, a neighborhood of Masaka City, and is owned by the Roman Catholic Diocese of Masaka.

Sample collection, Klebsiella isolation, antimicrobial susceptibility testing

The study was conducted from July 2020 to March 2021 in two tertiary hospitals in Uganda. Patients were divided into two categories: those who kept livestock and those who did not have livestock at home. Individuals who reported having livestock were recruited for the study. Rectal swab samples were collected from these recruited patients, referred to as index patients. Additionally, rectal swab samples were taken from caregivers of these patients at the hospitals. Follow-up visits were made to the patients’ homes, where samples were also collected from relatives who were present during the visits. In addition, samples were obtained from the livestock found at the patients’ homes, including rectal swabs from cattle, goats, sheep, and pigs. Cloacal swabs were collected from poultry, including chickens, turkeys, and ducks. Samples were also taken from various surfaces within the patients’ homes, such as kitchen surfaces. Furthermore, swabs were collected from healthcare workers at the two hospitals, along with samples from hospital surfaces, including equipment and benches. Environmental samples, including water, were also collected from both hospital and community settings, as previously described [24]. All samples were processed for isolation and identification of ESBL-producing KpSC on McConkey agar containing 2 µg/ml cefotaxime using methods as described in our previous studies [24]. KpSC were presumptively identified based on culture appearance and later confirmed biochemical reactions on API-20E tests (Biomeriaeux) and upon sequencing. The presumptive colonies were purified and confirmed as ESBL by the double-disk synergy method [25].

ESBL-producing isolates were subjected to susceptibility testing by the disc diffusion method to a panel of 19 antibiotics as described in our previous studies [21]. The zones of inhibition were measured and recorded in millimeters and were interpreted using the standardized breakpoints outlined in the Clinical and Laboratory Standards Institute M100 (33rd Edition) guidelines for antimicrobial susceptibility testing. These breakpoints were used to classify the isolates as susceptible (S), intermediate (I), or resistant (R) to the tested antimicrobial agents. The susceptible (S), intermediate (I), or resistant (R) results were presented separately. Colistin susceptibility testing was performed using the microdilution broth method. Bacterial isolates were cultured and adjusted to a 0.5 McFarland standard, then diluted in cation-adjusted Mueller–Hinton broth to achieve a final inoculum of approximately 5 × 10⁵ CFU/mL. Serial two-fold dilutions of colistin sulfate were prepared in 96-well microtiter plates, and each well was inoculated with the standardized bacterial suspension. Plates were incubated at 35 ± 2 °C for 16–20 h. The minimum inhibitory concentration (MIC) was defined as the lowest concentration of colistin that inhibited visible bacterial growth. MIC interpretation was performed using the EUCAST breakpoints (version 12.0, 2022), where isolates with MIC ≤ 2 µg/mL were considered susceptible and those with MIC > 2 µg/mL resistant.

Strains that showed resistance to more than two classes of antimicrobial agents were classified as multidrug-resistant [26]. In total, 56 KpSC that were ESBL positive were isolated. Of these, twenty-seven isolates were resistant to cefepime, a fourth-generation cephalosporin, and were deemed fourth-generation cephalosporin (4GCR) and were selected for whole genome sequencing.

DNA extraction, library preparation, and whole genome sequencing

DNA was purified from the K. pneumoniae species complex strains and subjected to whole genome sequencing (WGS) on Illumina NEXTSEQ 2000 (Illumina, Inc., San Diego, CA, USA). DNA extraction, WGS, and sequence reads processing were performed following previously described protocols [21]. The sequences were deposited into NCBI under a bioproject PRJNA1307811.

Genome assembly, quality check, and annotation

Raw reads were evaluated for quality using FastQC v0.12.1 [27] program and subsequently trimmed with Btrim v0.2.0 [28] to remove low-quality base calls at both the beginning and end of the reads, as well as to remove Illumina barcodes. De novo draft genome assemblies were produced using Newbler v2.7 (Roche Diagnostics). Minimum thresholds for contig size and coverage were set at 200 bp and 49.5+, respectively. Taxonomic assignment of the assembled genome was conducted using the Genome Taxonomy Database (gtdb-tk R226) [29, 30]. All genomes identified as belonging to the Klebsiella pneumoniae species complex (KpSC) were further confirmed using Kleborate version 3.1.3, which accurately assigned the species [31].

Genetic diversity analyses, antimicrobial resistant genes, and virulence genes profiling

The genomes of the Klebsiella pneumoniae species complex (KpSC) were analyzed using Kleborate v3.1.3 to explore their genetic structure [31]. Kleborate v3.1.3 categorized the isolates into multi-locus sequence types (MLST), including prediction of the K (capsule) and O antigen (LPS) serotype and wzi allele. To further verify the sequence types (ST) and wzi allele assignments, the genomes were queried in the allele and profile database of the Klebsiella PasteurMLST database (https://bigsdb.pasteur.fr/klebsiella/). The scgMLST629_S locus from this database was selected for the classification of the strains into their distinct profiles, which included cgMLST assignment and KpSC phylogroups designation (Kp1, Kp2, and Kp4). Phylogroup Kp1 denotes strains that belong to the Klebsiella pneumoniae, and Kp2 denotes a group that includes strains that belong to Klebsiella quasipneumoniae subsp. quasipneumoniae, and phylogroup Kp4 denotes strains belonging to Klebsiella quasipneumoniae subsp. similipneumoniae. Clonal groups (CG) and sublineages were assigned using the Life Identification Number (LIN)-based system in the Pasteur Institute Klebsiella MLST Database. Genome sequences were uploaded to the database and compared with the closest references available. Each genome is assigned a hierarchical LIN code that indicates its position based on progressively finer thresholds of genomic similarity to the references in the database. Clonal groups and sublineages were assigned to each genome based on the LIN code that corresponds to the references in the Pasteur Institute Klebsiella MLST Database.

Kleborate v3.1.3 uses Mimimap2 and CARD v3.2.9 AMR database to determine antimicrobial resistance profiles of the input genome sequence data. This was used to screen for different forms of antimicrobial resistance determinants present in the genomes. The determinants were categorized as acquired resistant genes, SNP mutations associated with resistance, gene truncations, and intrinsic β-lactamases. The predicted mutation-associated with antimicrobial resistance included Omp mutations associated with resistance to carbapenems, SHV mutations related to β-lactam resistance, Col mutations associated with Colistin resistance, and Flq mutations (GyrA-S83I; ParC-S80I) associated with fluoroquinolone resistance. The antimicrobial resistance score was assigned to isolates based on the combination of genes for ESBL, carbapenemase, and colistin resistance genes within their genome as described in the Kleborate manual. Presence of hypervirulence determinants like yersiniabactin (ybt), colibactin (clb), and aerobactin (iuc) were determined using Kleborate v3.1.3. Additional virulence determinants were also determined by querying the genome sequences into the virulence factor database [32]. Profiling of mobile genetic elements (MGEs) and their relationships with acquired antimicrobial-resistant genes was performed using MobileElementFinder v1.0.3 [33]. Contigs that had antimicrobial resistance genes flanked by mobile gene elements were extracted using samtool faidx [34]. The contigs were further annotated with Prokka and CARD database to determine the position of the AMR gene. The annotated contigs were visualized in proksee to map the position of the AMR genes relative to the position of the mobile genetic element [35].

Phylogenetic tree construction from core genome SNPs

A phylogenetic tree was constructed using REALPHY v113 software [36]. REALPHY v113 utilizes Bowtie2 to map sequences against references, generating multiple alignments that are then used to construct the maximum likelihood phylogenetic tree using PhyML version 3.3.20220408 [37]. The sequence GCA_033336135_1 (ASM3333613v1) downloaded from GenBank was used as the reference strain for building the tree. Meanwhile, sequence GCA_050435165_1 (ASM5043516v1) of Klebsiella aerogenes was used as an outgroup. For comparison of the phylogenetic relationship of the Uganda strain with global contemporary strains, genome sequences from different countries were retrieved from BV-BRC database (https://www.bv-brc.org/). The global reference sequences were chosen mainly based on the multilocus sequence type (MLST) profiles found among the Ugandan strains. To compile the reference dataset, we conducted searches on bv-brc database to identify publicly available genomes of contemporary global strains that represented the same MLSTs identified in the Ugandan strains. We applied filtering criteria based on geographical location to ensure that the reference dataset captured the global genetic diversity relevant to the strains detected in Uganda for phylogenetic comparison. In total, 257 sequences representing Africa, Europe, Asia, and North America strains were retrieved and included for comparison with Uganda strains (Additional file Table S1). The phylogenetic tree was imported into Interactive Tree of Life (iTOL) v6 [38] and then annotated with the associated metadata. GrapeTree v1.5.0 [39] was used to visualize the Minimum spanning trees from the core genome SNP profile to show the relationship between clonal groups, phylogroups, and hospital sources of the strains.

Statistical analysis

The total count of antimicrobial-resistant classes and resistant genes identified for each strain was calculated. The strains were subsequently grouped based on the hospitals from which they were isolated. A comparison was made between the total counts of antimicrobial-resistant classes and resistant genes detected in strains from Kitovu Hospital and those from Masaka Hospital using Mann-Whitney U test. The analysis was performed in XLSTAT v2024.4.0 (1424) (Lumivero, LLC, Denver, Colorado, US).

Results

Phenotypic resistance profile

A total of 1880 samples were collected from health care and community settings, these included index patients (n = 293), caretakers, and patient relatives (during home visit) rectal swabs (n = 304), animal rectal swabs (n = 105), environmental swabs (n = 158), hospital and home stead surface swabs (n = 966), and cloaca swabs (n = 54). Phenotypically, 56 KpSC isolates were ESBLs, and 27 of these were resistant to Cefepime, a 4th generation cephalosporin, and were selected for sequencing. Out of the 27 isolates studied, 11 4CGR strains were collected from Kitovu Hospital, while 16 were collected from Masaka Hospital. At Kitovu Hospital, the samples were sourced from human rectal/stool swabs (n = 7), surface swabs (n = 3), and a water swab (n = 1). The patients were primarily located in the maternity ward (n = 5) and surgery ward (n = 2). Two samples came from surface swabs within the surgery ward (ward toilet and ward floor), and one sample came from a surface swab of an equipment (Additional file Table S2). Most patients from Kitovu Hospital came from an urban setting, whereas only one patient was from a rural area. In contrast, the samples from Masaka Hospital were predominantly rectal swabs (n = 14), with only two samples taken from surface swabs. These samples were collected from patients in the children’s ward (n = 7), female ward (n = 3), male ward (n = 2), nutrition ward (n = 1), a patient caretaker (n = 1), as well as from the community environment (n = 2). All twenty-seven 4CGR strains were resistant to Aztreonam and to the following cephalosporin antibiotics: Cefotaxime, Cefepime, Cefotaxime, Ceftazidime, and Cefuroxime. Twenty-five of the 27 (92.59%) 4CGR strains were resistant to Nitrofurantoin, Trimethoprim-sulfamethoxazole, Ciprofloxacin, and Amoxicillin/Clavulanate (Fig. 1). A high number of strains were further phenotypically resistant to Tetracycline (66.67%), Gentamicin (70.3%), Ceftriaxone (81.48%), and Chloramphenicol (74.07%) antibiotics. In contrast, a significantly lower proportion of the strains were resistant to Cefoxitin (29.63%), Streptomycin (22.22%), Ertapenem (18.528%), and Imipenem (11.11%) (Fig. 1).

Fig. 1.

Antimicrobial susceptibility of 27 4CGR strains isolated from patients and hospital environment in two hospitals in Uganda. Blue-filled circles indicate resistance. Empty circles indicate susceptibility

Phylogroup composition of the KpSC 4CGR strains

The dominant phylogroup, Kp1, corresponding to Klebsiella pneumoniae accounted for 19 (70%) of the isolates. Six isolates (22%) belonged to phylogroup Kp4, and two strains (7%) were classified as phylogroup Kp2. The distribution of the three phylogroups according to hospital can be seen in Additional file Fig. S1.

Clonal groups (CG), Sequence Types (STs), and Sublineages (SLs)

The 4CGR KpSC strains were genetically diverse, grouping into fifteen clonal groups. The dominant clonal group was the CG17, which comprised 6 strains (Fig. 2A). This group corresponded to sequence type (ST17) and sub-lineage SL17 and was predominantly associated with healthcare-acquired infection (n = 5) and less frequently associated with community-acquired infection (n = 1). The CG17 group was found to be distributed in both hospitals (Fig. 2B). Other common clonal groups were CG14089 and CG34, both represented by three strains. CG14089 corresponds to sequence type ST1016-2LV and sublineage SL12200. All three CG14089 strains belong Kp4 phylogroup, and were from healthcare-acquired infections. Clonal group CG34 corresponds to sequence type ST34 and sublineage SL34 and belongs to phylogroup Kp1. All three CG34 strains came from hospital-acquired sources of infections, and rectal swabs from patients at Masaka Hospital. The remaining strains belonged to clonal groups represented by one or two strains each. A full overview can be seen from Additional file Table S2.

Fig. 2.

The minimum spanning tree illustrates the various clonal groups associated with the phylogroups Kp1, Kp2, and Kp4. In Fig. 2A, the color coding distinguishes the phylogroups: blue represents Kp1, grey indicates Kp4, and orange denotes Kp2. Conversely, in Fig. 2B, the color scheme reflects the different hospitals, with blue representing Masaka hospital and grey signifying Kitovu hospital. The numbers in square brackets indicate the total number of strains for each phylogroup and hospital, respectively

Serotype profile

Ten different O serotypes were detected among the 27 4CGR strains. The O5 serotype was the most abundant, with 7 strains (27%). The strains carrying the O5 serotype largely belong to the Kp1 phylogroup (n = 6). Only one strain of Kp4 phylogroup belongs to the O5 serotype group, and none in the Kp2 phylogroup. The O5 serotype was restricted to strains of clonal group CG17 of the Kp1 phylogroup and CG1224 of the Kp4 phylogroup. Six strains of the O5 serotype occurred in healthcare-associated infections, with only one strain coming from a community-acquired source of infection. The O1αβ,2α serotype was the second most abundant as it occurs among 4 strains and was restricted to the Kp1 phylogroup. Strains with O1αβ,2α serotype belonged to CG607 (n = 2) and CG14 (n = 2) groups. The CG607 strains were isolated from a surgery ward, and CG14 strains were from a children’s ward. The remaining serotypes were observed in only two or fewer strains each. A full overview of the distribution of serotypes according to phylogroup, ST-types, hospital, and origin of isolate can be seen in Additional file Table S2.

K locus capsular profile

Thirteen different K locus types were identified among the twenty-seven strains studied. The KL25 type was the most common, found in a total of eight strains, all of which belonged to the Kp1 phylogroup. Strains of KL25 were present in all sample types and were isolated from patients, household kitchen items, hospital equipment, and water surfaces. Only two clonal groups were associated with the KL25. Six KL25 strains belong to CG17 clonal group, while two KL25 strains belong to CG607 clonal group. The KL22 was detected in four strains; unlike other K locus types, which were observed within a single phylogroup, KL22 was found in both the Kp1 (n = 3) and Kp4 (n = 1) phylogroups. The KL22 strains in the Kp1 phylogroup were associated with healthcare-associated sources of infections, whereas the single KL22 isolate from the Kp4 phylogroup was linked to a community-associated infection. The remaining K locus types were observed in two or fewer strains, and the reader is referred to Additional file Table S2 for information on these types.

Antimicrobial-resistant determinants

When classified according to Kleborate v3.1.3, two strains had a resistance score of zero (No ESBL and no carbapenemase genes); instead, they carried chromosomally based beta-lactamase genes, whereas the remaining 25 strains carried a resistance score of one (presence of ESBL genes), with diverse resistance gene classes. Strain DKB0042, with a resistance score of zero, carries a chromosomally based beta-lactamase OKP-B-2 gene. Another Strain, DKB0761, with a resistance score of zero, carries another chromosomally based beta-lactamase SHV-11 allele. Strain DKB0114, isolated from the male ward at Masaka hospital, had only one resistance gene class (CTX-M-15). The remaining strains had multiple resistance gene classes of six or above. The number of resistance gene classes only marginally varied (p = 0.061) across the strains recovered from the two hospitals, with a similar median value of 7 resistance gene classes (Fig. 3). However, there was a significant difference (p = 0.009) in the number of resistance genes carried by strains from the two hospitals. The 4CGR strains isolated from Masaka Hospital had a median number of resistance genes of 9, compared to the median number of 12 resistance genes observed among the strains isolated from Kitovu Hospital (Fig. 3).

Fig. 3.

The Mann-Whitney U test shows differences in the antimicrobial resistance profiles of KpSC strains from two hospitals. Figure 3A illustrates the variations in the number of antimicrobial resistance classes between strains from Kitovu Hospital and Masaka Hospital. Figure 3B illustrates the differences in the number of antimicrobial-resistant genes found in the genomes of KpSC strains from both hospitals

The strain exhibited a diverse array of acquired aminoglycoside-resistant genes. Among these, three variants of acquired aminoglycoside acetyltransferases were identified. The aac(6’)-Ib-cr.v2 gene was the most prevalent, found in 70.37% of the strains, and was distributed within phylogroups Kp1 and Kp4. In contrast, the aac(3)-IIa.v1 gene, present in 48.15% of the strains, was limited to phylogroup Kp1 (Fig. 4). The aac(3)-IId gene was restricted to clonal group CG14 within phylogroup Kp1. The strB.v1 and strA.v1 genes demonstrated a broader distribution, as they were detected across all three phylogroups. A total of 55.56% of the strains had strB.v1 gene comparable to 48.15% of the strains that had strA.v1 genes. The aadA16 gene, found in 40.74% of the strains, primarily occurred in phylogroup Kp1, with only one instance in phylogroup Kp4. Conversely, the aadA2 gene was predominantly present in Kp4, with just a single isolate from Kp1 carrying this gene.

Fig. 4.

Heatmap showing the distribution of antimicrobial resistance (AMR) determinants identified among the 4CGR strains. Each row represents an individual strain, while columns represent AMR genes detected. The colored circle cells indicate the presence of specific resistance determinants, and empty circle cells indicate the absence of the AMR determinants. Metadata associated with each strain, including hospital of origin, sample type, phylogroup, K-locus, and clonal group, are displayed alongside the heatmap to illustrate epidemiological and genomic relationships between isolates

The qnrS1, qnrB1.V2, and qnrB6 genes were identified as the three acquired Fluoroquinolone-resistant genes. The qnrS1 gene was the most prevalent, detected in 44.44% of the strains, and displayed a wider distribution across all three phylogroups (Fig. 4). In contrast, the qnrB6 gene was present in 22.22% of the strains, primarily found within the Kp1 phylogroup, with only one strain from Kp4 carrying this gene. The qnrB1.V2 gene was found in only two strains belonging to the Kp1 phylogroup.

The rifamycin gene arr-3 was most prevalent in the Kp1 strain and was less frequently detected in other phylogroups. The Extended-spectrum beta-lactamase (ESBL) gene CTX-M-15 was identified in all three phylogroups, while just one strain of Kp1 possessed the CTX-M-14 gene. Other acquired beta-lactamase genes detected included TEM-1D.v1, LAP-2, and OXA-1. Overall, TEM-1D.v1 was the most frequent, occurring in 70.37% of strains and was widely distributed across all phylogroups (Fig. 4). Besides, ten different chromosomal non-ESBL genes were detected. Five were the OKP variants, and another five were SHV variants. The OKP gene variants were OKP-B-2 (n = 3), OKP-B-21(n = 1), OKP-B-45 (n = 1) and OKP-B-6 (n = 1) that occurred only among the strains of Kp4 phylogroup. In contrast, OKP-B-45 and OKP-A-12 (n = 2) were carried only by strains of phylogroup Kp2. The SHV gene variants were detected only among strains of phylogroup Kp1. The SHV-11 gene was detected in 9 strains of phylogroup Kp1. Other variants include SHV-186 (n = 3), SHV-27 (n = 2), SHV-1 (n = 2) and SHV-28 (n = 3). All the strains with SHV-28 were isolated from children’s ward, while the two strains carrying SHV-1 were isolated from surgery ward.

Virulence determinants

The strains demonstrated very few virulence-associated gene loci, with virulence scores of zero (n = 13) and one (n = 14). A virulence score of zero indicates the complete absence of the virulence-associated loci yersiniabactin (ybt), colibactin (clb), salmochelin (iro), and hypermucoidy (rmpA), and a virulence score of one is linked to the presence of only the yersiniabactin (ybt) gene locus, which was found exclusively among strains of phylogroup Kp1. A total of eight strains carried the ybt16 gene locus, which is shared solely among clonal groups CG17 and CG14. Both ybt14 and ybt9 were identified in three strains each, with ybt14 occurring in two strains of CG607 and one strain of CG10192. Finally, ybt9 was detected in clonal groups CG340, CG307, and CG873 (Fig. 5).

Fig. 5.

Distribution of virulence loci of the yersiniabactin gene and other virulence determinants among the strains. The pink circles are the yersinibactin gene loci. The blue circles are other virulence factors detected within the genomes of the different strains

Other virulence factors identified included the enterobactin synthase subunit F (entF) gene, which occurred in all the strains. The fimD gene, which encodes the outer membrane usher protein for type I fimbriae, was detected in 20 strains of the Kp1 phylogroup except DKB0262. The fimD gene was observed in only two strains of Kp4, and none of the Kp2 strains. A total of 17 strains were detected with the acriflavine resistance protein B (acrB) gene. This gene was distributed across all three phylogroups. The immune modulation gene (KP1) was observed in 12 strains. The mrkC gene, which is involved in fimbrial biogenesis, was only observed in strains from the Kp4 and Kp2 phylogroups. The siderophore esterase (iroE) gene was found in 12 strains of Kp1 and one strain of Kp4.

Diversity and distribution of mobile genetic elements and plasmid replicons

A total of ten plasmid replicons were identified across the 27 genomes (Fig. 6). The IncFIB(K) and IncFII(K) plasmid replicons were found to be widely distributed in genomes from both hospitals. IncFIB(K) was present in 21 genomes and was detected across all phylogroups. In contrast, IncFII(K) was observed in 15 genomes, primarily within the Kp1 phylogroup and to a lesser extent in the Kp4 phylogroup. The phylogroup Kp2 exhibited a limited variety of plasmid replicons, with only IncFIB(K) and IncFIA(HI1) detected in its two strains. Additionally, IncFIB(pKPHS1) and IncY were exclusively found in the genomes belonging to phylogroup Kp4.

Fig. 6.

Plasmid replicons characteristics of KsPc isolated from the two hospitals in Uganda

Most of the genomes were found with IS903 (n = 23), IS5075 (n = 19), IS6100 (n = 19), ISEc9 (n = 15), and ISKpn26 (n = 18) mobile genetic elements (Additional file Fig S2). These insertion sequences were widely distributed in all the phylo-groups occurring in both strains derived from community and hospital sources. Contrary, only the genomes of phylogroups Kp1 and Kp4 were found to harbor ISKpn26. The IS5 and ISSpu2 insertion sequences were limited only to strains of phylogroup Kp2. Two strains (DKB0114 and DKB0042) had only n = 3 mobile genetic elements within their genomes, which is less comparable with other strains. We observed that clonal groups and sequence type largely determined the distribution of some of the insertion sequences. The ISKox1 was mainly present within the genomes of CG14089 of phylogroup Kp4 and CG34 of phylogroup Kp1. All the n = 9 genomes with ISKpn24 belong to only two clonal groups, CG607 and CG17 of the Kp1 phylogroup. Other insertion sequences that only occur within the genomes of Kp1 were ISKpn12, ISKpn13, ISKpn25, ISKpn1, and ISCfr1 (Additional file Fig S2).

Several of the insertion elements were carried together on contigs with antimicrobial resistance genes. Most of the contigs with ISKpn19 insertion sequences also carry dfrA12, qnrS1, qacE, aadA2, sul1, and sul2 antimicrobial resistance genes. Similarly, contigs with IS6100 insertion also had qacE, sul1, qnrB6, aadA16, aac(6’)-Ib-cr, dfrA27, ARR-3 antimicrobial resistant determinants (Fig. 7). The ISEc9 was carried together with blaTEM-1B, sul2, aph(6)-Id, aph(3’’)-Ib, and blaCTX-M-15 AMR genes on the same contigs. Meanwhile catA1 gene was found mainly in contigs carrying ISCfr1 insertion elements (Fig. 7). aph(3’’)-Ib and aph(6)-Id genes were carried on contigs with transposable element Tn6082. tet(A) and qnrB1were carried on contigs with transposable element Tn5403.

Fig. 7.

Circular map of contigs showing the relationship between antimicrobial resistance genes and their corresponding mobile genetic elements: IS6100 (A), ISEcp1 (B), ISKpn19 (C), and ISCfr1 (D). The green color shows the position of the mobile genetic element on the contig. Red color indicates the position of AMR genes

Phylogenetic relationship with global contemporary strains

A phylogenetic comparison of Ugandan strains with 257 contemporary global strains revealed that strains clustered according to ST-type. Within each ST-TYPE, Ugandan strains generally clustered with strains of African origin, while only n = 5 strains clustered with strains isolated outside Africa (Fig. 8). Five of the Uganda strains that belong to ST17 were seen to cluster together. These five strains were closely related to strains from South Africa as they shared a common node. Another strain, DKB1372 of ST17, was found in a different cluster, which predominantly had strains from Kenya, Tanzania, and Nigeria. ST34 strains from Africa were not common, and only one sequence from Nigeria was obtained and retrieved. Three strains (DKB0024, DKB1213, and DKB0243) of the ST34 from Uganda did not cluster with the sequence from Nigeria (NFYJ01); instead, they were closely related to a reference strain from the United Kingdom (CAKNEE01). Strains DKB0121 and DKB0120 of ST607 were found to be closely related to reference strains from Kenya (JBBGGR01 and JAJHYD01), Tanzania, and South Africa. For ST14, the two Uganda strains (DKB1202 and DKB0262) of ST14 clustered with one strain previously isolated from Uganda and two strains that were isolated from Kenya and South Africa.

Fig. 8.

Phylogenetic tree illustrating the genetic relatedness and clustering patterns of Ugandan isolates with the broader global contemporary population reference strains. Nodes highlighted in red indicate the Ugandan strains generated in the current study, while the remaining nodes represent global reference strains

Discussion

The present study provides a comprehensive characterization of multidrug-resistant (MDR) Klebsiella pneumoniae species complex (KpSC) isolates exhibiting resistance to fourth-generation cephalosporins from two major hospitals in Uganda. These were recovered from patients with both urban and rural backgrounds with small but insignificant differences in their resistance. This may reflect widespread antibiotic use in both community and agricultural contexts, as was suggested based on high resistance levels in E coli recovered from the same settings [24]. This underscores the growing challenge of AMR in resource-limited settings, as pointed out by others [40], where surveillance is often sporadic and empirical therapy remains common. As expected, these 4GCR strains were often MDR. The observed resistance to nitrofurantoin, trimethoprim-sulfamethoxazole, tetracycline, and gentamicin reflects the multidrug-resistant nature of these isolates and poses a significant therapeutic challenge, particularly in settings with limited access to newer antibiotics. Notably, a substantial proportion of strains retained susceptibility to carbapenems, suggesting there is low carbapenem resistance which is often the last resort antibiotic [41]. These isolates were also largely sensitive to cefoxitin which has increased stability against bacterial beta-lactamase enzymes, making it effective against some bacteria that resist other cephalosporins [42]. The AST evaluation showed that several strains of the Klebsiella pneumoniae species complex were resistant to colistin, which is particularly concerning since colistin is often considered a last-line treatment for multidrug-resistant (MDR) Gram-negative bacterial infections. The resistance to colistin leaves few, if any, effective treatment options for managing MDR bacterial infection that will emerge in Uganda. This increases the risk of treatment failures and may lead to higher mortality rates among patients, and it is an important clinical, epidemiological, and public health concern.

These 4GCR KpSC strains were mostly of Kp1 type, consistent with previous reports that K. pneumoniae sensu stricto is the principal pathogen within the KpSC [43]. The Kp4 and Kp2 strains were identified at lower frequencies. Their presence emphasizes the genetic diversity of clinical KpSC isolates. It also underscores the potential role of these less frequent types in hospital outbreaks and their potential to transmit MDR attributes to other more common pathogens. Clonal analysis revealed the presence of 15 distinct clonal groups, with CG17 (ST17, SL17) emerging as the dominant clone, largely associated with healthcare-acquired infections and known among high-risk clones [19, 21, 44]. The presence of CG17 in both hospitals suggests that certain high-risk clones may be circulating regionally [19, 21]. The detection of CG14089 and CG34 among multiple wards reinforces the heterogeneous nature of MDR KpSC populations in clinical settings and underlines the possible transmission within the hospital settings, especially with poor infection prevention [45].

Genomic-based serotyping revealed significant diversity, with O5 being the most prevalent serotype, primarily linked to healthcare-associated Kp1 isolates. The restricted distribution of certain serotypes within specific clonal groups suggests a potential role of serotype-clone associations in hospitals. The predominance of O5 and O1αβ,2α serotypes, particularly among pediatric and surgical wards, may have implications for future vaccine design and targeted interventions, given the growing interest in O-antigen-based vaccines for K. pneumoniae [46].

Analysis of antimicrobial resistance determinants revealed a complex repertoire of acquired resistance genes, with strains harboring up to nine distinct resistance gene classes. Two strains carried chromosomally-based genes that mediate resistance to β-lactams. The OKP-B-2 gene found in strain DKB0042 encodes a chromosomally-based beta-lactamase enzyme that primarily confers resistance to penicillins and cephalosporins that have been reported elsewhere [47, 48]. The widespread occurrence of aminoglycoside-modifying enzymes, including aac(6’)-Ib-cr.v2 and aac(3)-IIa.v1, along with plasmid-mediated quinolone resistance genes such as qnrS1, underscores the multidrug-resistant phenotype of these 4GCR strains. ESBL genes, notably blaCTX-M-15 and blaTEM-1D.v1, were prevalent across all phylogroups, corroborating the phenotypic resistance patterns observed, which was not unexpected given these were 4GCR. Although some strains exhibit phenotypic resistance to colistin, genomic analysis did not detect any known resistance determinants, such as mcr genes or chromosomal mutations in pmrA/pmrB, phoP/phoQ, and mgrB genes. As observed by [49] Klebsiella pneumoniae strains harbor a diverse set of colistin-resistant determinants, including novel, previously undescribed mutations potentially associated with colistin resistance. Some of these novel mutations that have recently been associated with colistin resistance included mgrB^Ile13fs, pmrB^Gly207Asp, phoQ^His339Asp, and ramA^Ile28Met [49]. Since our 4GCR KpSC collection did not include known acquired resistance genes like mcr-1 or chromosomal mutations in genes such as pmrA/pmrB, phoP/phoQ, and mgrB, it is possible that these strains possess uncharacterized mechanisms of colistin resistance, warranting further detailed genomic analysis and SNP typing.

The co-localization of resistance genes with mobile genetic elements, such as ISKpn19, IS6100, and ISEc9, suggests a high potential for horizontal gene transfer, facilitating the dissemination of resistance determinants both within and between hospitals and the community [50]. This genomic plasticity is particularly concerning, as it may enable rapid adaptation and spread of high-risk clones in Ugandan healthcare settings [21, 24]. Virulence profiling indicated limited virulence potential, with most isolates carrying only the yersiniabactin locus (ybt) and lacking other classical virulence-associated loci such as colibactin and salmochelin. The restriction of ybt loci to phylogroup Kp1, particularly CG17, may confer a selective advantage in the hospital environment, enhancing persistence without imposing a high fitness cost. This finding aligns with global observations that hypervirulent determinants are less prevalent in MDR hospital-associated KpSC strains, suggesting that resistance rather than hypervirulence drives the epidemiology of nosocomial infections in this context [19, 21].

The distribution of plasmid replicons further illustrates the genetic flexibility of these isolates. The predominance of IncFIB(K) and IncFII(K) plasmids across multiple phylogroups highlights their role in maintaining resistance gene arrays and mediating inter-strain gene transfer. The restricted distribution of certain plasmids and mobile genetic elements among specific clonal groups has been described elsewhere [51]. This may indicate that the genomic context and clonal background influence the acquisition and retention of these elements, contributing to the differential resistance profiles observed between the two hospitals. Notably, isolates from Kitovu Hospital exhibited a higher median number of resistance genes compared to Masaka Hospital. This may reflect differences in antibiotic use between the two institutions but may also be an accidental effect of the dominating clones circulating in the hospitals.

Phylogenetic comparisons with global KpSC strains demonstrate that Ugandan isolates cluster predominantly with African strains, particularly those from Kenya, South Africa, and Tanzania, confirming previous reports [19, 22, 52]. This regional clustering underscores the transnational circulation of high-risk clones within the continent, likely facilitated by human mobility. Interestingly, certain sequence types, such as ST34, exhibited closer relationships with European strains than with other African isolates, suggesting occasional long-distance dissemination events [53].

Conclusions

This study characterized 27 4GCR KpSC from two hospital environments in Uganda, highlighting the importance of these resistant strains as fourth-generation cephalosporins are critical antibiotics used for the treatment of severe hospital-associated infections. Although the number of isolates analyzed in this study was limited and sampling was restricted to two selected hospitals, the findings provide important preliminary insights into the presence and diversity of 4GCR KpSC in Uganda. The detection of these resistant strains underscores the need for strengthened antimicrobial resistance surveillance, improved infection prevention and control practices, and continued genomic monitoring to better understand the dissemination of 4GCR KpSC in healthcare settings. Future studies with a broader geographic scope will be essential to more comprehensively characterize the burden, transmission patterns, and public health implications of 4GCR KpSC in Uganda.

Authors' contributions

D.K.B., F.W.M., J.E.O., conceptualized and designed the study. D.K.B., A.W., and S.A. undertook the sample collection, isolation, identification, and susceptibility testing of the bacteria. G.W. performed bioinformatics analysis. G.W., D.K.B., HC, FWM, YMHH, J.E.O. wrote the manuscript and revisions. All authors read and approved the final manuscript.

Funding

This study was funded by Makerere University Research and Innovations Fund (RIF) from the Government of Uganda to DKB for research on “Unravelling the burden and transmission dynamics of antimicrobial resistance in humans, animals and the environment”.

Data availability

The sequences were deposited into NCBI under a bioproject PRJNA1307811 (https://www.ncbi.nlm.nih.gov/search/all/?term=PRJNA1307811).

Declarations

Ethics approval and consent to participate

Ethical approvals were obtained from Makerere University School of Public Health Higher Degrees Research and Ethics Committee (HDREC #760) and Uganda National Council for Science and Technology (UNCST) (#HS649ES). The study also obtained administrative approval from the Uganda Ministry of Health (#ADM.185/130/01). Written informed consent was obtained from each eligible study participant with assurance for confidentiality of information. The study was performed in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards, and the relevant guidelines and regulations.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Clinical trial number

Not applicable.