Antimicrobial Resistance, Virulence Factors and Plasmid Replicon Patterns of Klebsiella pneumoniae and Klebsiella grimontii Isolates from Bovine Mastitic Milk in the Northwest of Portugal: Pilot Genomic Characterization

TL;DR

This study characterizes the antimicrobial resistance and virulence of Klebsiella isolates from bovine mastitis in Portugal, revealing new resistance patterns and plasmid features.

Contribution

The study reports novel resistance profiles and plasmid replicon patterns in bovine mastitis-associated Klebsiella isolates.

Findings

K. pneumoniae-1DH1 showed resistance to multiple antibiotics including colistin and nitrofurantoin.

K. grimontii-2DH2 was found to carry a plasmid replicon with both blaCTX-M and astA genes.

Phylogenomic analysis showed K. pneumoniae-1DH1 clustered with human isolates, while K. grimontii-2DH2 grouped with environmental isolates.

Abstract

Background: Bovine mastitis (BM) remains an economically significant disease in the global dairy industry. Multidrug resistance (MDR) in Klebsiella pneumoniae and Klebsiella grimontii has increased in recent years, representing an area of concern related to BM. Methods: Bovine mastitis 1-DH1 and 2-DH2 isolates (n = 2) were investigated. Antimicrobial susceptibility testing was performed using the Neg-Urine-Combo98 panel. Antimicrobial resistance genes (ARGs), virulence factor (VF) genes and plasmid replicons were identified by whole-genome sequencing (WGS). Phylogenomic analyses were performed for a visual comparison of the genomes. Results: Phenotypically, isolates 1-DH1 and 2-DH2 were identified as K. pneumoniae-1DH1 and Klebsiella oxytoca, respectively; the latter was subsequently confirmed as K. grimontii-2DH2 by WGS. K. pneumoniae-1DH1 (20.0%, 5/25) exhibited phenotypic resistance to amoxicillin–clavulanic acid, ampicillin (AM), cefuroxime, colistin (Cl), and nitrofurantoin (Fd), whereas K. grimontii-2DH2 (8.0%, 2/25) was resistant to AM and nalidixic acid (NA). In total, 31 and 32 ARGs and 10 and 15 VF genes were identified in K. pneumoniae-1DH1 and K. grimontii-2DH2, respectively. Two plasmid replicons were detected in K. pneumoniae-1DH1 (one harboring the blaCTX-M ARG) compared to one in K. grimontii-2DH2 (encoding both the blaCTX-M ARG and the astA VF gene) isolates, respectively. Phylogenomic analysis showed that K. pneumoniae-1DH1 clustered most closely with human-derived K. pneumoniae isolates, whereas K. grimontii-2DH2 grouped with environmental K. grimontii isolates. Conclusions: Novel phenotypic resistance profiles were observed in the BM-associated K. pneumoniae-1DH1 (MDR, Cl and Fd) and K. grimontii-2DH2 (AM and NA) isolates. Notably, K. grimontii-2DH2 harbored a plasmid replicon carrying both the blaCTX-M ARG and the astA VF gene.

Figures

Full text

1. Introduction

Bovine mastitis (BM), defined as inflammation of the mammary gland, is one of the most significant diseases affecting the global dairy industry and is associated with substantial economic losses, involving both direct and indirect costs [1,2]. Bovine mastitis is commonly classified into clinical mastitis (CM) and subclinical mastitis (SCM). Economic burden associated with this disease arises from both direct and indirect costs: direct losses are primarily attributed to therapeutic intervention, milk disposal, and premature withdrawal of affected animals, whereas indirect losses are largely driven by milk yield reduction and quality [3,4]. Gram-positive organisms such as Staphylococcus aureus and Streptococcus agalactiae are typically considered contagious pathogens, spreading mainly during milking. In contrast, other streptococci (e.g., Streptococcus uberis) and coagulase-negative staphylococci are predominantly environmental. Environmental Gram-negative bacteria, including Escherichia coli (E. coli), Klebsiella spp., Pseudomonas spp., Acinetobacter spp. and Serratia spp., are well-known causes of clinical mastitis, although they may also be isolated from subclinical infections [5,6].

Klebsiella is a member of the Enterobacteriaceae family, and encompasses a wide diversity of phylogenetic lineages, genomic content, pathogenic properties and ecological niches [7]. The genus can be split into two major subdivisions: the Klebsiella pneumoniae species complex (KpSC) and the Klebsiella oxytoca species complex (KoSC), the latter of which includes the Klebsiella grimontii (k.grimontii) species [8]. K. pneumoniae has been increasingly associated with BM outbreaks worldwide [9,10,11,12,13,14,15,16,17,18]. Genotyping studies have revealed considerable genetic diversity within the K. pneumoniae population associated with BM worldwide. For example, 180 mastitis-associated K. pneumoniae strains from 11 U.S. states were classified into 61 sequence types (STs), 51 capsular locus (KL) types, and 12 lipopolysaccharide O-antigen serotypes [12]. Moreover, 42 strains of K. pneumoniae isolated from Scottish BM were classified into 31 different STs [19]. Research from China reported 239 and 68 mastitis-associated K. pneumoniae classified into 100 and 23 STs in the Hubei and North Jiangsu Provinces, respectively [20,21]. Regarding BM caused by K. grimontii, only a few recent studies have been reported [16,17], and no genotyping data are currently available. Therefore, the study of K. grimontii isolates represents a potentially important advance in understanding BM. K. pneumoniae is an increasingly significant bacterial pathogen capable of causing severe and potentially life-threatening infections, including urinary tract infections, pneumonia, enteritis, and septicemia in both humans and animals [17,22,23,24,25,26,27,28]. K. pneumoniae has been frequently detected in environmental studies [29,30]. In comparison, K. grimontii has been reported in respiratory and urinary tract infections, as well as gut colonization in humans [31,32,33,34] and has also been recovered from environmental sources [35].

Antimicrobial resistance (AMR) represents a major public health threat, with new and emerging resistance mechanisms continuously appearing and spreading worldwide [36]. According to the World Health Organization (WHO), AMR was directly responsible for an estimated 1.27 million deaths in 2019 and contributed to nearly 5 million additional deaths [37]. The emergence of AMR in bacteria within humans, animals or the environment can affect the health of all sectors, making it a critical One Health issue [38]. Resistant pathogens can cause disease in animals and humans, and many have zoonotic or foodborne transmission pathways. This highlights the importance of integrating veterinary diagnostic laboratory data into AMR monitoring programs under the One Health framework [39,40]. Consequently, the surveillance of zoonotic pathogens in animals is also a priority of the European Union [41]. The inappropriate use of antimicrobials has driven a significant global increase in multidrug resistance (MDR) in K. pneumoniae, raising public health concerns [42,43]. MDR in K. grimontii was also reported in a gut colonization study of patient-sourced isolates [33], and this species has been suggested as a potential vector for the dissemination of MDR genes through soils, sediment and aquatic systems [44].

In the context of BM, the use of antibacterial treatments has been identified as a major driver of resistance among bacterial populations in treated animals [45]. Additionally, AMR bacteria and ARG of animal origin might be transmitted to humans, either directly or via environmental contamination [45]. ARG may also spread horizontally through mobile genetic elements, raising the possibility that resistance genes from milk-associated bacteria could be transferred to the human gastrointestinal tract [46,47]. A strong correlation between MDR and biofilm formation in BM was also previously described [48,49], as well as the association of specific plasmid replicon types with MDR bacteria [50]. Taken together, these findings highlight the need for careful monitoring of AMR in Klebsiella spp. from dairy cows [51].

In Portugal, Klebsiella spp. has been associated with BM; however, no studies have reported on K. pneumoniae or K. grimontii specifically from BM isolates. Nevertheless, K. pneumoniae and K. oxytoca have been detected in Azorean milk tanks [52]. These species have been identified in companion animals, such as dogs and cats, in northern Portugal [53]. In contrast, several studies across different Portuguese regions have reported K. pneumoniae as a MDR pathogen in clinical isolates from hospitalized and non-hospitalized human patients, with the detection of multiple ARGs, virulence factor (VF) genes, plasmid replicons and biofilm formation [23,24,54,55,56]. Similar observations for K. pneumoniae and K. grimontii were reported from clinical samples collected in a neonatal special care unit [56]. Additionally, K. pneumoniae has been detected in environmental sources, including hospital sinks, sink drains and surface waters [55,57,58].

In this study, two isolates from raw milk (RM) obtained originating in DH from the Entre-Douro and Minho region of northwest Portugal were identified as K. pneumoniae (K. pneumoniae-1-DH1) and K. grimontii (K. grimontii-2-DH2). Isolates were phenotypically characterized for AMR, and genotypically characterized for ARG, VF genes and plasmid replicon patterns by whole-genome sequencing (WGS).

2. Results

2.1. Bacterial Phenotypic Identification and Antimicrobial Susceptibility Testing

The bacterial isolate 1-DH1 was phenotypically identified as K. pneumoniae (K. pneumoniae-1-DH1), while the isolate 2-DH2 was identified as K. oxytoca and later confirmed as K. grimontii (K. grimontii-2-DH2) by WGS, with both identifications assigned a confidence level of 99.99% using the Neg-Urine-Combo98 panel on the MicroScan WalkAway Plus system (Table 1). Antimicrobial susceptibility testing (AST) showed that K. pneumoniae-1-DH1 exhibited a 20.00% phenotypic resistance prevalence, displaying resistance to Aug-E and AM (β-lactams), Crm (cephalosporin), Cl (polymyxin) and Fd (nitrofuran) (Table 1). Based on this profile, K. pneumoniae-1-DH1 was classified as a MDR strain. In contrast, isolate 2-DH2 exhibited a lower resistance prevalence (8.00%), with confirmed resistance to AM (β-lactams) and NA (quinolone), along with an intermediate response (4.00%) to Crm (cephalosporin) (Table 1). Although not yet considered MDR, this resistance pattern suggests possible early development of multidrug resistance. Regarding antimicrobial susceptibility, K. pneumoniae-1-DH1 remained susceptible to 80.00% of the antimicrobials tested, including Cpe, Cft, Caz/CA, Etp, Imp and Mer (β-lactams); NA, Cp, Lvx, and Nxn (quinolones); AK, Gm and To (aminoglycosides); T/S (sulfonamide); AZT (monobactam); Cfx and Caz (cephalosporins); Fos (phosphonic acid derivative); Cft/CA (cephalosporin/β-lactamase inhibitor); and P/T (penicillin/β-lactamase inhibitor) (Table 1). Similarly, K. grimontii-2-DH2 demonstrated 88.00% susceptibility, remaining sensitive to Aug-E, Cpe, Cft, Caz/CA, Etp, Imp and Mer (β-lactams); Cp, Lvx, and Nxn (quinolones); AK, Gm and To (aminoglycosides); T/S (sulfonamide); AZT (monobactam); Cfx and Caz (cephalosporins); Cl (polymyxin); Fos (phosphonic acid derivative); Fd (nitrofuran); Cft/CA (cephalosporin/β-lactamase inhibitor); and P/T (penicillin/β-lactamase inhibitor) (Table 1).

2.2. Bacterial and Plasmid Genome Characteristics and Classification

Average nucleotide identity (ANI) analysis identified the whole-genome sequence of isolate 1-DH1 (contig_1) as K. pneumoniae, and the whole-genome sequence of isolate 2-DH2 (contig_1) as K. grimontii (Figure 1). Moreover, two plasmid replicons, IncFII_1_pKP91 and IncFIB(K)_1_Kpn3 were found in the whole-genome sequenced isolate K. pneumoniae 1-DH1 with identity percentage of 87.39% and 98.75%, respectively, using PlasmidFinder (Figure 1). The plasmid replicon IncFII_1_pKP91 was found in the whole-genome sequenced K. grimontii 2-DH2 isolate with identity percentage of 89.29% (Figure 1). Klebsiella pneumoniae isolate 1-DH1 and K. grimontii isolate 2-DH2 were namely K. pneumoniae-1-DH1 and K. grimontii-2-DH2, respectively.

2.3. Genomic Analysis of Bacterial Antimicrobial Resistance Genes

The ARGs oqxA, oqxB, acrA, acrB, acrD, mdtB, mdtC, kpnE, kpnF, kpnG, kpnH, ramA, marA, fosA6, ompK37, SHV-187, msbA, baeR and cpxA were identified in the whole-genome sequence of the K. pneumoniae-1-DH1 isolate with identity percentages ranging from 80.24% to 99.70% using ABRicate with the Comprehensive Antibiotic Resistance Database (CARD) (Table 2 and Table S2). All these ARGs were also detected in the K. grimontii-2-DH2 isolate, with identity percentages ranging from 80.06% to 99.89%, except for fosA6, ompK37 and SHV-187, which were not identified (Table 2 and Table S2). Additionally, fosA5, mdtN, bacA and oxy-6-1 were identified only in the K. grimontii-2-DH2 isolate with similar identity percentages (Table 2 and Table S2). Using the National Center for Biotechnology Information (NCBI) AMRFinderPlus database, oqxA6, oqxB20, fosA and blaSHV-187 were identified in K. pneumoniae-1-DH1, with identity percentages ranging from 97.62% to 99.65% (Table 2 and Table S2). These ARGs were also detected in K. grimontii-2-DH2 (82.97–99.89%), except for oqxA6 and blaSHV-187, which were not found (Table 2 and Table S2). Furthermore, oqxA10 and blaoxy-6-1 were identified additionally in K. grimontii-2-DH2 with similar identity levels (Table 2 and Table S2). Using ResFinder, fosA_5, blaSHV-187_1, oqxA_1 and oqxB_1 ARG were identified in K. pneumoniae-1-DH1 (97.62–99.65%) (Table 2 and Table S2). All four were detected in K. grimontii-2-DH2 (82.97–99.89%), except fosA_5 and blaSHV-187_1, which were absent (Table 2 and Table S2). In turn, fosA_2 and blaoxy-6-1_1 were detected only in K. grimontii-2-DH2 with similar identities (Table 2 and Table S2). Using Antibiotic Resistance Gene-ANNOTation (ARG-ANNOT), the (Flq)oqxA, (Flq)oqxBgb, (Fcyn)fosA6 and blaSHV-187 ARGs were detected in K. pneumoniae-1-DH1 (96.19–99.65%) (Table 2 and Table S2). These genes were also detected in K. grimontii-2-DH2 (81.31–99.89%), except for (Fcyn)fosA6 and blaSHV-187_1, which were not found (Table 2 and Table S2). In addition, (Fcyn)FosA5 and oxy-6-1 were identified only in K. grimontii-2-DH2 with comparable identity percentages (Table 2 and Table S2). Using MEGARes, the oqxA, oqxB, acrA, acrB, acrD, mdtB, mdtC, KpnE, KpnF, KpnO, ramA, marA, fosA, msbA, emrB, emrD, baeR and cpxAR ARGs were identified in both K. pneumoniae-1-DH1 (80.02% to 99.75%) and K. grimontii-2-DH2 (80.06 to 90.87%) (Table 2 and Table S2). Additionally, mdtN was also identified in K. grimontii-2-DH2 only, with similar identity percentages (Table 2 and Table S2).

Overall, both isolates carried a large set of ARGs, including oqxA, oqxA_1, oqxA6, oqxA10, oqxB, oqxB_1, oqxB20, oqxBgb, acrA, acrB, acrD, mdtB, mdtC, kpnE, kpnF, KpnG, KpnH, msbA, emrB, emrD, baeR, cpxA and cpxAR (Table 2 and Table S2). These genes are associated with resistance to diaminopyrimidines, fluoroquinolones, glycylcyclines, nitrofurans, tetracyclines, phenicols, quinolones, cephalosporins, penams, rifamycins, triclosan, aminoglycosides, aminocoumarins, acridine dyes, macrolides, carbapenems, penems, cephamycins, monobactams and nitroimidazoles (Table 2 and Table S2). All of these ARGs were present in both genomes, with the exception of oqxA6, which was not detected in K. grimontii-2-DH2, whereas oqxA10 and mdtN were exclusive to this isolate (Table 2 and Table S2). The regulatory genes ramA and marA, which are associated with efflux-mediated reduced permeability, were detected in both isolates and conferred resistance to carbapenems, cephalosporins, cephamycins, fluoroquinolones, glycylcyclines, monobactams, penams, penems, phenicols, rifamycins, tetracyclines and triclosan (Table 2 and Table S2). Regarding fosfomycin and beta-lactam inactivation mechanisms, fosA, fosA_5, fosA6, SHV-187 and blaSHV-187_1 were identified only in K. pneumoniae-1-DH1 (Table 2 and Table S2). In contrast, K. grimontii-2-DH2 carried fosA, fosA_5 and fosA6 but did not contain SHV-187 or blaSHV-187_1 (Table 2 and Table S2). The reduced-permeability gene ompK37 was detected only in K. pneumoniae-1-DH1 (Table 2 and Table S2). Finally, bacA, associated with target alteration and resistance to peptide antibiotics, was exclusive to K. grimontii-2-DH2 (Table 2 and Table S2).

2.4. Genomic Analysis of Bacterial Virulence Factor Genes

The VF genes yagZ/ecpA, yagY/ecpB, yagX/ecpC, yagW/ecpD, yagV/ecpE and ykgK/ecpR (all encoding E. coli common pilus structural subunits), fepC (E. coli ferrienterobactin ABC transporter ATPase), entA (E. coli 2,3-dihydro-2,3-dihydroxybenzoate dehydrogenase), entB (E. coli isochorismatase), and ompA (E. coli outer membrane protein A) were identified in the whole-genome sequence of K. pneumoniae-1-DH1 with identity percentages ranging from 80.70% to 90.24%, as determined by ABRicate (BLAST) against the virulence factor database (VFDB) (Table 2 and Table S2). In K. grimontii-2-DH2, yagZ/ecpA, entA, entB, and ompA VF genes were also detected, with identity percentages ranging from 80.36% to 83.29% (Table 2 and Table S2). Additionally, the VF genes ybtA (Yersinia pestis (Y. pestis) transcriptional regulator), ybtE, ybtT, and ybtU (Y. pestis biosynthetic proteins), ybtS (Y. pestis salicylate synthase Irp9), ybtP and ybtQ (Y. pestis inner membrane ABC transporters), ybtX (Y. pestis putative signal transducer), fyuA (Y. pestis pesticin/yersiniabactin receptor protein), and irp1 and irp2 (Y. pestis biosynthetic proteins) were uniquely identified in K. grimontii-2-DH2, with identity percentages ranging from 84.45% to 89.96% (Table 2 and Table S2).

2.5. Genomic Analysis of Bacterial Plasmid Replicons

Whole-genome sequencing of K. pneumoniae-1-DH1 revealed the presence of two plasmid replicons: IncFII_1_pKP91 (87.39% identity) carrying the blaCTX-M ARG (97.46% identity), and IncFIB(K)_1_Kpn3 (98.75% identity) as determined by ABRicate using the MEGARes database (Table 2 and Table S2). In K. grimontii-2-DH2, a plasmid replicon IncFII_1_pKP91 (89.29% identity) carrying the blaCTX-M ARG (87.78% identity) and the astA VF gene (88.70% identity) was detected (Table 2 and Table S2). Both isolates were also evaluated for plasmid replicons using the MOBScan server, and these or additional plasmid replicons were not identified in either genome.

2.6. Phylogenomic Analysis

Phylogenomic analysis revealed well-supported core genomes for K. pneumoniae-1-DH1 and K. grimontii-2-DH2 (bootstrap values predominantly 100%), each grouping with isolates from multiple sources (Figure 2). The core genome of K. pneumoniae-1-DH1 revealed two strongly supported clusters, A1 and A2 (bootstrap = 100% each) (Figure 2A). Cluster A1 comprised K. pneumoniae isolates from environmental, human, and cow’s milk sources, whereas cluster A2 included environmental and human-derived isolates (Figure 2A1,A2). Within cluster A2, two subclusters were strongly supported (bootstrap = 100% each): one grouping K. pneumoniae environmental and human isolates, and the other including the BM K. pneumoniae-1-DH1 isolate with human-sourced isolates (Figure 2A2). Overall, K. pneumoniae-1-DH1 was more closely related to K. pneumoniae human-derived isolates, including one from a Portuguese patient outside the study region, and more distantly related to K. pneumoniae environmental and cow’s milk isolates, suggesting a recent common ancestry with human-associated K. pneumoniae. The core genome of K. grimontii-2-DH2 revealed two strongly supported clusters, B1 and B2 (bootstrap = 100%) (Figure 2B). Cluster B1 comprised K. grimontii environmental-, BM-, animal-, and human-sourced isolates, whereas cluster B2 included environmental and human isolates (Figure 2B1,B2). Within cluster B1, two subclusters were identified (bootstrap = 100% and 93.40%): one grouping K. grimontii environmental isolates with the BM K. grimontii-2-DH2 isolate, and the other comprising environmental- and human-sourced isolates (Figure 2B1). Overall, K. grimontii-2-DH2 was more closely related to environmental K. grimontii isolates and more distantly related to K. grimontii human-sourced isolates.

3. Discussion

Two Gram-negative bacterial isolates obtained from RM samples of clinical BM cases from two DH isolates in the northwest region of Portugal were investigated: 1-DH1 and 2-DH2. Using the Neg-Urine-Combo98 panel on the MicroScan WalkAway Plus system, isolates 1-DH1 and 2-DH2 were phenotypically identified as K. pneumoniae (K. pneumoniae-1-DH1) and K. oxytoca, respectively. However, WGS later confirmed isolate 2-DH2 as K. grimontii (K. grimontii-2-DH2), indicating a misidentification by the panel-based phenotypic method. The use of this panel for phenotypic identification of Serratia spp. from BM isolates has been previously described, and a similar phenotypic misidentification was reported. The panel phenotypically identified one isolate as Serratia odorifera, while WGS confirmed the species as Serratia marcescens [6]. Furthermore, several studies have documented the limited ability of conventional phenotypic methods to accurately differentiate closely related species within the KoSC and other taxa from both human and animal sources. This reduced discriminatory performance has been noted for MicroScan panel systems (e.g., MicroScan Dried Overnight Positive ID Type 3 [PID3] for Gram-positive organisms and Negative ID Type 2 [NID2] for Gram-negative organisms), analytical profile index (API) kits (e.g., API 20E and API 50CH), and automated platforms such as Vitek II and VITK-JR30 [31,60,61,62,63,64,65].

Regarding AMR as determined by the Neg-Urine-Combo98 panel on the MicroScan WalkAway Plus system, K. pneumoniae-1-DH1 exhibited a phenotypic AMR prevalence of 20.00% (n = 5/25), with resistance detected to Aug-E and AM (β-lactams), Crm (cephalosporin), Cl (polymyxin) and Fd (nitrofuran). This isolate (K. pneumoniae-1-DH1) was phenotypically susceptible to the remaining antimicrobials (80.00%, n = 20/25). Based on these results, isolate K. pneumoniae-1-DH1 was classified as MDR [66], consistent with previous reports describing MDR K. pneumoniae isolated from BM and human clinical infections [48,49]. Isolate K. grimontii-2-DH2 (initially identified phenotypically as K. oxytoca but confirmed as K. grimontii by WGS), showed a lower phenotypic AMR prevalence of 8.00% (n = 2/25), with resistance to AM (β-lactam) and NA (quinolone). This profile does not meet MDR criteria, although it approaches the classification threshold. Additionally, the isolate K. grimontii-2-DH2 displayed intermediate resistance to Crm (n = 1/25), while remaining susceptible to the other antimicrobials tested (88.00%, n = 22/25). Overall, the present data indicate that antimicrobials such as Cpe, Cft, Caz/CA, Etp, Imp and Mer (β-lactams); NA, Cp, Lvx and Nxn (quinolones); AK, Gm and To (aminoglycosides); T/S (sulfonamide); AZT (monobactam); Cfx and Caz (cephalosporins); Fos (phosphonic); Cft/CA (cephalosporin/β-lactam); and P/T (penicillin/β-lactam) may represent suitable therapeutic options for BM caused by K. pneumoniae-1-DH1. BM associated with K. grimontii-2-DH2 may also be treatable with these antimicrobials, except for AM (β-lactam), NA (quinolone) and Crm (cephalosporin), which showed resistance or intermediate resistance in this isolate. It is important to note, however, that several of these antimicrobials are prohibited for veterinary use in the European Union due to their categorization as Category A by the European Medicines Agency (EMA), as they are reserved for human medicine. Phenotypic resistance of K. pneumoniae-1-DH1 to Aug-E, AM, Crm, Cl and Fd was identified in this study. Previous reports of K. pneumoniae isolated from BM described high to moderate levels of resistance to AM, sulfisoxazole, cephalothin, florfenicol, streptomycin, Cpe and T/S, while maintaining susceptibility to Mer, Cl, Cp, Fd and chloramphenicol [49,67,68]. Comparatively, the phenotypic resistance of K. pneumoniae-1-DH1 to Cl and Fd observed here has not yet been reported, suggesting that this is the first documented occurrence. Similarly, phenotypic resistance of K. oxytoca isolate 2-DH2, later confirmed by WGS as K. grimontii-2-DH2, to AM and NA was identified. Current literature reports K. grimontii isolates as phenotypically susceptible to AM, azithromycin, Cp, chloramphenicol, Cl, Cft, Gm, Mer, NA, sulfamethoxazole, Caz, tetracycline, tigecycline and trimethoprim [16]. Therefore, the phenotypic resistance of the BM-origin K. grimontii-2-DH2 isolate to AM and NA also appears to be reported here for the first time. However, it is noteworthy that an MDR K. grimontii strain has previously been isolated from a necropsied horse fetus [69].

Whole-genome sequencing was conducted for both BM isolates (1-DH1 and 2-DH2). Based on WGS data, isolate 1-DH1 was identified as K. pneumoniae and isolate 2-DH2 as K. grimontii, exhibiting an ANI of 98.91% and 99.15%, respectively, when compared with reference genomes of the corresponding species. The isolates were designated as K. pneumoniae-1-DH1 and K. grimontii-2-DH2. Previous WGS-based studies have characterized K. pneumoniae strains from BM, human, and environmental sources [12,14,29,55], while K. grimontii has been described through WGS in human and environmental isolates [31,32,33,34,35]; however, reports from BM remain scarce. Concerning ARGs, K. pneumoniae-1-DH1 carried oqxAA6BB20Bgb, acrABD, mdtBC, kpnEFGH, ramA, marA, fosAA5A6, ompK37, blaSHV-187, msbA, emrBD, baeR and cpxAAR, as detected by WGS. Several of these ARGs, namely oqxAB, acrAB, mdtBC, kpnEFGH, fosA6, blaSHV-187, msbA and baeR, among others, have previously been reported in K. pneumoniae isolated from BM [51,70]. To the best of our knowledge, oqxA6, oqxB20, oqxBgb, acrD, ramA, fosA_5, ompK37, emrBD and cpxAAR represent the first documented occurrence of these ARGs in K. pneumoniae from BM; however, oqxAB and fosAA5 have previously been reported in environmentally sourced K. pneumoniae isolates [71]. In K. grimontii-2-DH2, WGS identified the ARGs oqxAA10BB20Bgb, acrABD, mdtBCN, kpnEFGH, ramA, marA, fosAA2A5, msbA, emrBD, baeR, cpxAAR, bacA and blaOXY-6-1. A comprehensive review of the available literature did not identify any previously reported antimicrobial resistance gene (ARG) profiles for Klebsiella grimontii isolated from bovine mastitis, indicating that the ARG repertoire described here for BM K. grimontii-2-DH2 is reported for the first time. Nevertheless, the K. grimontii-specific β-lactamase genes blaOXY-6-1 and blaOXY-6-4 have previously been documented in human- and environment-derived isolates, and the efflux-associated gene oqxB has been reported in environmental strains [31,62,72]. Collectively, ARGs previously described in K. pneumoniae or K. grimontii from human, bovine mastitis, or environmental sources, as well as those identified in the present BM isolates, represent a potential concern within a One Health framework, as they reflect interconnected reservoirs and transmission pathways of antimicrobial resistance, consistent with prior One Health-oriented assessments [73,74]. With respect to virulence-associated genes, whole-genome sequencing revealed that K. pneumoniae-1-DH1 harbored the ecp (yag) fimbrial gene cluster (yagZ/ecpA, yagY/ecpB, yagX/ecpC, yagW/ecpD, yagV/ecpE, and ykgK/ecpR), as well as the iron acquisition genes fepC and entAB, and the outer membrane protein gene ompA. Many of these VF genes (along with others) have previously been reported in K. pneumoniae isolates from BM, humans, animals and powdered milk sources [75,76,77]. Additionally, the VF genes ybt, fyuA and irp have been documented in environmentally sourced K. pneumoniae strains [71]. No reports were found describing VF genes in K. grimontii isolates from BM, suggesting that the VF genes yagZ/ecpA, entAB, ompA, ybtAEPQSTUX, fyuA, and irp12 detected in the BM isolate K. grimontii-2-DH2 are documented for the first time here. However, the VF genes fyuA, traT and terC have previously been reported in K. grimontii recovered from environmental sources [72]. Concerning plasmid replicons, WGS revealed the presence of two plasmid replicons in K. pneumoniae-1-DH1: IncFII_1_pKP91, carrying the blaCTX-M ARG, and IncFIB(K)_1_Kpn3. Both plasmid types, along with several others, have been previously reported in K. pneumoniae isolates recovered from BM, human patients and powdered milk sources [50,76,78]. Nevertheless, no similar replicon profiles have been documented in K. pneumoniae strains isolated from environmental sources. In addition, studies focusing on newly acquired ARGs of K. pneumoniae utilized horizontal gene transfer through conjugative plasmids, finding that plasmid content such as conjugation machinery, transposons, VF and phages may contribute to the diversification and dissemination of plasmids containing ARGs previously described [79,80,81,82]. In K. grimontii-2-DH2, WGS identified the plasmid replicon IncFII_1_pKP91, harboring both the blaCTX-M ARG and the astA VF. Reports of plasmid replicons in K. grimontii, including FII(Y) detected in human sputum samples [31], IncHI2/HI2A and IncFII(Yp) carrying blaVIM-1 and mcr-9, or blaVIM-1 alone, identified in human urine sources [32] were previously documented. Furthermore, a study reported simultaneous gut colonization by K. grimontii and E. coli co-possessing the blaKPC-3-carrying pQil plasmid, where the ability of K. grimontii to transfer the blaKPC-3-pQil plasmid to other Enterobacterales, such as E. coli, was demonstrated [33], reinforcing the emergence of MDR bacteria carrying plasmid replicons that increase the risk of AMR worldwide. Additional plasmid replicons, including ColRNAI, FIA, IncFII(Yp) have been described in environmentally sourced K. grimontii isolates [72]. Based on the currently available literature, the plasmid replicon IncFII_1_pKP91 carrying both blaCTX-M ARG and astA VF gene in BM K. grimontii-2-DH2 isolate is reported here for the first time. Overall, plasmid replicons previously reported in K. pneumoniae or K. grimontii from human, bovine mastitis, or environmental sources, as well as those identified in the present BM isolates, represent a potential concern within a One Health framework, as plasmids are key mediators of antimicrobial resistance and virulence, a role that has been extensively highlighted in prior One Health-oriented studies [83].

In Portugal, there are studies focusing on environmental (e.g., Streptococcus spp. non-coagulase-negative Staphylococcus) and contagious (e.g., Corynebacterium spp., Staphylococcus aureus and Streptococcus agalactiae) BM agents [84]. Klebsiella spp. (environmental) has been associated with BM for a long time, though no studies have been reported specifically on K. pneumoniae or K. grimontii from BM isolates. Nevertheless, a previous study described the detection of Klebsiella species, specifically K. pneumoniae and K. oxytoca, in bovine milk sampled from bulk tanks in Azorean herds [52]. Additionally, both species have also been identified in companion animals such as dogs and cats in a study conducted in northern Portugal [53]. Despite the lack of BM-related data, several studies from different Portuguese regions have reported K. pneumoniae in human clinical isolates (from both hospitalized and non-hospitalized patients), where the species was frequently characterized as MDR and associated with multiple ARGs, VF genes, plasmid replicons, or biofilm formation [23,24,54,55,56]. Furthermore, similar findings concerning K. pneumoniae and K. grimontii were documented in clinical samples collected in a neonatal special care unit [56]. Environmental reports are also available, with K. pneumoniae being detected in hospital sink and sink drain samples, as well as in surface waters [55,57,58]. However, to date, no studies have described K. pneumoniae or K. grimontii from BM sources in Portugal.

Phylogenomic analysis revealed a well-supported core genome cluster for K. pneumoniae-1-DH1 and another for K. grimontii-2-DH2, each exclusively grouping with reference isolates of the corresponding species. Both core genomes were strongly supported, with bootstrap values predominantly reaching 100.00%. Within the K. pneumoniae-1-DH1 core genome, two major and highly supported clusters (A1 and A2; bootstrap = 100.00%) were identified. Cluster A1 comprised K. pneumoniae isolates from environmental, human and cow’s milk sources, whereas cluster A2 included isolates of environmental and human origin. Notably, within cluster A2 two strongly supported subclusters (bootstrap = 100.00%) were identified: one containing environmental and human-derived isolates, and a second grouping where the BM K. pneumoniae-1-DH1 isolate clustered with human-sourced isolates. Overall, K. pneumoniae-1-DH1 showed a closer phylogenetic relationship to human-associated K. pneumoniae isolates, including one recovered from a Portuguese patient unrelated to the sampling region, than to environmental or cow’s milk isolates. These findings suggest a recent shared ancestry between the BM isolate and human-derived K. pneumoniae strains. To our knowledge, no Portuguese studies have reported similar phylogenetic relationships between BM and human isolates. However, comparable patterns have been documented in other countries, including the United States, China and South Korea [12,85,86]. Moreover, a recent common ancestry between BM K. pneumoniae isolates and K. pneumoniae of environmental origin has also been previously described [12]. For K. grimontii-2-DH2, the core genome analysis revealed two strongly supported clusters (B1 and B2; bootstrap = 100.00%). Cluster B1 comprised K. grimontii isolates from environmental sources, the BM K. grimontii-2-DH2 isolate in this study, animal and human sources, whereas cluster B2 contained K. grimontii isolates from environmental and human origins. Within cluster B1, two strongly supported subclusters (bootstrap = 100.00% and 93.40%) were identified: one grouping environmental isolates with BM K. grimontii-2-DH2 isolate, and the other comprising environmental and human-sourced isolates. Overall, BM K. grimontii-2-DH2 isolate appears to be more closely related to an environmental-sourced K. grimontii isolate and more distantly related to human-sourced isolates. These results suggest a recent common ancestry between the BM isolate and environmental K. grimontii strains. To date, no similar studies have been reported in Portugal. Nevertheless, a wastewater-derived K. grimontii isolate closely related to human-sourced strains has previously been described [72], and phylogenetic analyses of K. grimontii isolates from urine samples of a catheterized patient collected over ~6 months have also been reported [34].

Additionally, resistant pathogens can cause zoonotic or foodborne illnesses in humans and animals through ingestion or direct contact [39,40,87]. In this context, it is essential that data from animal pathogens collected by veterinary diagnostic laboratories be incorporated into AMR monitoring programs as part of the One Health framework [39,40]. Several studies have investigated the potential role of K. pneumoniae across human, animal (including BM milk and dairy products) and environmental samples [55,57,58,88,89]. The possible involvement of K. grimontii in these three sectors has been suggested based on studies of isolates from human, animal, and environmental sources, including food, soil and sewage sludge [44,90,91,92]. According to the results obtained in this study, K. grimontii seems to be additionally implied in BM. These findings underscore the need for continued research on K. pneumoniae and K. grimontii within human, animal, and environmental contexts to prevent their dissemination, improve BM control, and ultimately support both animal and public health.

4. Materials and Methods

4.1. Samples

In this study, two Gram-negative bacterial isolates (1-DH1 and 2-DH2, n = 2) derived from RM samples of two cows with clinical mastitis from two dairy herds in the Entre-Douro e Minho region of northwest Portugal, collected in February 2025, were used. The affected cows presented altered milk with a yellowish, watery appearance and visible clots, edema of the affected mammary quarter, increased rectal temperature (>40 °C), and a strongly positive California Mastitis Test (score 3, with marked gel formation). These isolates are part of the microorganism collection of SVA Expleite, Ld., Fradelos, Portugal.

4.2. Bacterial Identification and Antimicrobial Susceptibility Testing

Identification and AST were performed as previously described [6]. Briefly, isolates were grown in MacConkey agar (Merck, Darmstadt, Germany) and incubated for 24 h at 37 °C [6]. Following incubation, the Neg-Urine-Combo98 panel (Beckman Coulter, Tokyo, Japan) was used for bacterial colony identification and antimicrobial minimum inhibitory concentration (MIC) testing using a MicroScan WalkAway Plus system (Beckman Coulter, Carnaxide, Portugal) [6,59,93,94,95]. The tested antimicrobials and MIC dilutions for antimicrobial interpretation, respectively, were as follows: amikacin (AK, 8–16), amoxicillin–clavulanate acid (Aug-E, 8/2, 32/2), ampicillin (AM, 4–8), aztreonam (AZT, 1–4, 16), cefepime (Cpe, 1, 4–8), cefotaxime (Cft, 1–16), cefotaxime–clavulanate acid (Cft/CA, 0.25/4–0.5/4, 4/4), cefoxitin (Cfx, 8), ceftazidime (Caz, 1–16), ceftazidime–clavulanate acid (Caz/CA, 0.25/4–0.5/4, 2/4), cefuroxime (Crm, 8), ciprofloxacin (Cp, 0.06, 0.25–1), colistin (Cl, 2–4), ertapenem (Etp, 0.12–0.5), fosfomycin (Fos, 8, 32), gentamicin (Gm, 2–4), imipenem (Imp, 2–4), levofloxacin (Lvx, 0.5–1), meropenem (Mer, 0.12, 2–8), nalidixic acid (NA, 16), nitrofurantoin (Fd, 64), norfloxacin (Nxn, 0.5–1), piperacillin–tazobactam (P/T, 8/4–16/4), tobramycin (To, 2–4) and trimethoprim–sulfamethoxazole (T/S, 2/38–4/76). Isolates were categorized as susceptible, intermediate, or resistant to each tested antimicrobial compound following the EUCAST guidelines [59]. Klebsiella, which were resistant to at least 3 different classes of antimicrobials, were considered as MDR [96]. Whole-genome sequencing was performed for both (1-DH1 and 2-DH2) isolates.

4.3. Bacterial DNA Extraction

Bacterial DNA extraction was performed as previously described [6]. Briefly, bacterial isolates were grown in brain–heart infusion (BHI) (Liofilchem, Roseto degli Abruzzi, Italy) followed by bead mill homogenization using a TissueLyser (Qiagen, Hilden, Germany) system [6]. Genomic DNA was then extracted using the Mag-Bind^®^ Universal Pathogen Core Kit-Tissue protocol (OMEGA BIO-TEK, Norcross, GA, USA) and quantified using the 1x ds DNA HS Assay Kit (TransGen Biothec, Beijing, China) in a Qubit™ fluorometer (Invitrogen, Carlsbad, CA, USA), following the manufacturer’s instructions [6].

4.4. Bacterial Whole-Genome Sequencing

Whole-genome sequencing was performed as previously described [6]. Briefly, genomic DNA from isolates 1-DH1 (42.70 ng/μL) and 2-DH2 (55.00 ng/μL) were sequenced by Plasmidsaurus (Bacterial Genome sequencing, London, UK) using Oxford Nanopore Technology (ONT, Oxford, UK) via a PromethION 24 instrument (ONT, Oxford, UK), equipped with an R10.4.1 flow cell as previously described. The Native Barcoding Kit 96 V14 (Oxford Nanopore Technologies, Oxford, UK) was used for library preparation following the manufacturer’s instructions [6,97], followed by custom analysis and annotation. The raw FASTQ reads were basecalled in super-accurate mode, using ont-doradod-for-promethion v7.4.12, applying a Q-score of minimum 10, with adapters and barcodes trimmed via MinKNOW v6.0.4. The quality control of the sequencing reads was performed using NanoPlot 1.43.0 [6,98]. Sequencing adapters and barcodes were removed using Porechop 0.2.4 [6,99]. NanoFilt v.2.8.0 was used to filter the reads, applying a minimum average quality score threshold of 10 [6,98].

4.5. Bacterial Whole-Genome Sequencing Analysis

Long-read sequencing data were quality-filtered by removing the bottom 5.00% of reads using Filtlong v.0.2.1 [100] with default parameters. The remaining low-quality reads were also removed (-mean_q_weight 10). Quality-controlled reads were then assembled using Flye v2.9.1 [101], with parameters optimized for high-quality ONT reads. The resulting assembly was then polished using Medaka v1.8.0. [102]. The resulting polished assembly was functionally and structurally evaluated. Genomes were annotated with Bakta v1.6.1 [103] and genome completion and contamination were measured with CheckM2 v1.1.0 [104]. Taxonomic identification was performed using MASH v2.3 against the RefSeq genome and plasmid database [105]. Results were confirmed by ANI using FastANI v1.34 [106] and the suspected species type strain (K. pneumoniae FDAARGOS_775 [T] and K. grimontii CIP111401T [T]), retrieved from GenomesDB [107]. Plasmid-related sequences were screened with PlasmidFinder 2.1 [108], along with MOBscan [109] server. Antimicrobial resistance genes and virulence factor-associated genes were screened with ABRicate v1.2.0 [110], using multiple databases: ResFinder [111], CARD [112] and MEGARes [113], and VFDB [114]. The two genome sequences of BM K. pneumoniae-1DH1 and K. grimontii-2-DH2 isolates were deposited in GenBank under the BioProject PRJNA1290908, with accession numbers Biosample SAMN49944116 for K. pneumoniae-1-DH1 isolate, and Biosample SAMN49944115 for K. grimontii-2-DH2 isolate.

4.6. Phylogenomics

For the phylogenomic reconstruction, representative complete genomes of K. pneumoniae and K. grimontii isolates were retrieved from the NCBI RefSeq database. K. pneumoniae and K. grimontii genomes were then used to generate a core-genome alignment, which served as the basis for inference of a maximum-likelihood phylogenetic tree along with K. pneumoniae-1-DH1 and K. grimontii-2-DH2 isolates, respectively. Briefly, genome annotation was performed with Bakta v1.11.1 [103]. Genomes were processed with Panaroo v1.5.2 [115], using strict filtering parameters to construct a high confidence pangenome. The core gene alignment was then used as a basis for phylogenomic inference. Phylogenomic reconstruction relied on IQTree3 v.3.0.1 [116]. The best-fit nucleotide substitution model was determined using the Bayesian Information Criterion (BIC), and to ensure topology robustness, 1000 replicates were performed. The resulting tree was visualized using the Interactive Tree of Life (iTOL) v7 [117]. Isolation source was mapped onto the tree to aid in the interpretation of clustering patterns.

4.7. Statistical Analysis

Descriptive statistics were performed using Microsoft^®^ Excel^®^ 2016. Categorical variables were summarized as percentages.

5. Conclusions

This study evaluated phenotypic identification and AST of BM Klebsiella spp. using the Neg-Urine-Combo98 panel on the MicroScan WalkAway Plus system. While AST results for Klebsiella spp. were consistent, phenotypic species identification was less reliable. Notably, the BM 2-DH2 isolate was misidentified as K. oxytoca by the panel, whereas WGS confirmed it as K. grimontii, demonstrating the superior specificity of WGS for species-level identification. In contrast, the BM K. pneumoniae-1-DH1 isolate was correctly identified by both phenotypic and genomic methods.

Phenotypic AST indicated that multiple therapeutic options are available for both isolates. The BM K. pneumoniae-1-DH1 isolate was susceptible to a broad range of antimicrobials, including Cpe, Cft, Caz/CA, Etp, Imp, Mer, NA, Cp, Lvx, Nxn, AK, Gm, To, T/S, AZT, Cfx, Fos and P/T. Similarly, the BM K. grimontii-2-DH2 isolate showed susceptibility to most tested antimicrobials, with the exception of resistance to AM and NA and intermediate resistance to Crm. Importantly, K. pneumoniae-1-DH1 exhibited resistance to Aug-E, AM, Crm, Cl and Fd, with Cl and Fd resistance representing an unreported profile, classifying it as MDR. The BM K. grimontii-2-DH2 isolate also displayed a previously unreported resistance phenotype to AM and NA.

WGS analysis revealed extensive resistance gene profiles. In K. pneumoniae-1-DH1, ARGs such as oqxAA6BB20Bgb, acrABD, mdtBC, kpnEFGH, ramA, marA, fosAA5A6, ompK37, blaSHV-187, msbA, emrBD, baeR and cpxAAR ARG were detected, with a novel resistance profile reported for oqxA6, oqxB20Bgb, acrD, ramA, fosA5, ompK37, emrBD and cpxAAR**.** In K. grimontii-2-DH2, WGS identified the ARGs oqxAA10BB20Bgb, acrABD, mdtBCN, kpnEFGH, ramA, marA, fosAA2A5, msbA, emrBD, baeR, cpxAAR, bacA and blaOXY-6-1, representing a ARG profile not previously reported in BM isolates. WGS VF gene analysis identified yagZ/ecpA, yagY/ecpB, yagX/ecpC, yagW/ecpD, yagV/ecpE, ykgK/ecpR, fepC, entAB and ompA genes in K. pneumoniae-1-DH1, consistent with previously reported profiles. In K. grimontii-2-DH2, the VF genes yagZ/ecpA, entAB, ompA, ybtAEPQSTUX, fyuA and irp12 were detected, representing a novel VF profile. WGS plasmid analysis revealed two replicons in K. pneumoniae-1-DH1, IncFII_1_pKP91 carrying the blaCTX-M ARG and IncFIB(K)_1_Kpn3, and one in K. grimontii-2-DH2, IncFII_1_pKP91, carrying both the blaCTX-M ARG and the astA VF, representing a novel plasmid replicon configuration for BM K. grimontii.

Within a One Health framework, the phylogenomic relationships observed in this study suggest potential connectivity among human, animal, and environmental reservoirs. Specifically, BM K. pneumoniae-1-DH1 clustered closely with human-derived K. pneumoniae isolates, whereas BM K. grimontii-2-DH2 exhibited greater phylogenetic affinity to environmental K. grimontii strains. These patterns underscore the potential public health and zoonotic relevance of bovine mastitis-associated Klebsiella spp.

This study has limitations as only one K. pneumoniae and one K. grimontii isolate from BM origin were investigated. In this line, future research should include larger-scale sampling of Klebsiella spp., particularly K. pneumoniae and K. grimontii, from BM, environmental sources (e.g., bedding, water, milking equipment), and potential human and animal reservoirs across DHs. Such comprehensive surveillance will improve understanding of local antimicrobial resistance patterns, support BM control measures, and inform strategies to mitigate the dissemination of zoonotic and MDR pathogens.