Introduction
Mastitis is a prevalent and important disease in cattle worldwide, which has significant health, welfare, and economic impacts [Reference Petersson-Wolfe, Leslie and Swartz1–Reference Klaas and Zadoks3]. Mastitis control programmes have primarily focussed on contagious mastitis pathogens and not environmental opportunists [Reference Fu, Wen and Wang4]. Environmental mastitis can be caused by a large range of bacteria, with one of the most predominant and clinically significant being K. pneumoniae [Reference Fu, Wen and Wang4–Reference Zadoks, Griffiths and Munoz6]. K. pneumoniae is one of the most common coliforms causing clinical bovine mastitis, but has been found to be one of the most damaging when considering milk production [Reference Gröhn, Wilson and González7], treatment costs [Reference Fu, Wen and Wang4], and mortality rate [Reference Oliveira, Hulland and Ruegg5]. K. pneumoniae mastitis has a poor cure rate after antimicrobial treatment [Reference Oliveira, Hulland and Ruegg5, Reference Schukken, Chuff and Moroni8, Reference Fuenzalida and Ruegg9], and although it is commonly an environmental opportunist, lateral spread from diseased to healthy cattle is possible [Reference Schukken, Bennett and Zurakowski10].
Antimicrobial resistance (AMR) is an ongoing concern in both animal and human health settings. Due to the ubiquitous nature of K. pneumoniae, there are varied reports on the AMR potential in these strains. In some previous work, it has been shown that K. pneumoniae isolates from milk showed low levels of AMR [Reference Saini, McClure and Léger11–Reference Massé, Dufour and Archambault13]. However, there have been reports of multi-drug resistant K. pneumoniae isolated from bovine mastitis cases from several countries [Reference Koovapra, Bandyopadhyay and Das14–Reference Saishu, Ozaki and Murase16], including the United Kingdom [Reference Timofte, Maciuc and Evans17], most of which are extended-spectrum beta-lactamase producers. More concerning reports have also emerged describing carbapenem resistance in K. pneumoniae isolates from cattle milk [Reference Silva-Sanchez, Barrios-Camacho and Hernández-Rodriguez18, Reference He, Wang, Sun and Pang19], highlighting the importance of ongoing disease and AMR surveillance.
Whole-genome sequencing, in combination with diagnostic laboratory identification and antimicrobial sensitivity testing (AST), gives a comprehensive insight into bacterial epidemiology, virulence potential, and AMR. Here, these methodologies are used to characterise K. pneumoniae isolates from mastitis cases in Scotland between 2009 and 2021.
Materials and methods
Isolates and primary identification
Forty-two isolates, previously identified as K. pneumoniae from clinical or sub-clinical mastitis cases that had been stored on glycerol beads (SRUC Veterinary Services’ Pathogen Bank) at −80 °C, were revived by culture on Columbia agar supplemented with sheep blood (Oxoid, United Kingdom), incubated aerobically at 37°C for 24 h. Isolates were obtained from cattle showing signs of sub-clinical or clinical mastitis between 2009 and 2021. Further details of these isolates and a summary of clinical histories are available in Supplementary Table S1.
Isolates were identified on initial isolation using API 20E strips (Biomerieux, France) according to the manufacturer’s instructions. As part of this study, isolates were also identified post hoc by matrix-assisted laser desorption/ionisation time-of-flight mass spectrometry using a Microflex LT/SH Biotyper (MALDI-TOF MS) (Bruker, Germany). Briefly, single and well-isolated colonies were spotted directly in duplicate onto Disposable MBT Biotarget 96 target plates (Bruker, Germany) using the direct transfer method. Spots were then overlaid with 1 μl of α-Cyano-4-hydroxycinnamic acid (HCCA) matrix and allowed to dry, prior to transfer to the instrument. The generated spectra were then subject to best match identification and confidence scoring, with scores of ≥2.00 deemed as good identifications.
Whole-genome sequencing
Whole-genome sequencing was performed by Microbes NG (University of Birmingham, UK), as described previously [Reference Foster, Robb and Paterson20]. Briefly, genomic DNA was extracted using solid-phase reversible immobilisation beads, and genomic DNA libraries were prepared using the Nextera XT Library Prep Kit (Illumina) following the manufacturer’s protocol with the following modifications: input DNA is increased by twofold, and PCR elongation time is increased to 45 s. Libraries were sequenced using Illumina sequencers (HiSeq/NovaSeq) using a 250-bp paired-end protocol. Reads were trimmed using Trimmomatic version 0.30 [Reference Bolger, Lohse and Usadel21], using a sliding window quality cut-off of 15. Genome assembly was carried out de novo using SPAdes, version 3.7, with default parameters for 250 bp Illumina reads [Reference Bankevich, Nurk and Antipov22] and annotated by the National Center for Biotechnology Information (NCBI) Prokaryotic Genome Annotation Pipeline v6.6 [Reference Seemann23]. Contigs <200 bp were manually removed, and contaminant sequences were automatically removed during upload to NCBI via the NCBI Foreign Contamination Screen [Reference Astashyn, Tvedte and Sweeney24]. Genome assemblies were uploaded and analysed using Pathogenwatch [Reference Argimón, David and Underwood25] v22.1.1 (https://pathogen.watch/) and Kleborate v2.3.0 [Reference Lam, Wick and Watts26] incorporated therein. Resultant trees were annotated using the Interactive Tree of Life (iTOL) [Reference Letunic and Bork27, Reference Letunic and Bork28].
Antimicrobial sensitivity testing
AST using nine antimicrobials – amoxicillin–clavulanic acid 30 μg, ertapenem 10 μg, streptomycin 10 μg, enrofloxacin 5 μg, tetracyline 30 μg, cefotaxime 5 μg, cephalexin 30 μg, neomycin 30 μg, and cefquinome 30 μg (Oxoid, United Kingdom). The EUCAST disc diffusion method was used on Mueller–Hinton Agar (Oxoid, United Kingdom) using EUCAST [29], CLSI Vet [30], and CASFM Vétérinaire [31] guidelines. Namely, antimicrobial sensitivity was assessed for amoxicillin–clavulanic acid, ertapenem, streptomycin, enrofloxacin, tetracycline, cefotaxime, cephalexin, neomycin, and cefquinome. Isolates that were highlighted as potential extended-spectrum beta-lactamase (ESBL) producers by AST were subject to ESBL screening using a commercially available ESBL, AmpC, and carbapenemase activity kit (MAST, United Kingdom).
Results
Species identification of bovine mastitis isolates
Forty-two bovine mastitis isolates initially identified by MALDI-TOF MS and API 20E as being K. pneumoniae were subject to whole-genome sequencing by Illumina technology, and the resultant assemblies were analysed via Kleborate. The 42 isolates were confirmed, based on Kleborate and Type Strain Genome Server analysis [Reference Meier-Kolthoff and Göker32] of their genome sequence, as being K. pneumoniae (Supplementary Table S1).
Antimicrobial sensitivity testing
All isolates were susceptible to ertapenem and enrofloxacin (Table 1). Of note were two ESBL-producing isolates, which showed multi-drug resistance to amoxicillin–clavulanic acid, streptomycin, tetracycline, cefotaxime, cephalexin, and cefquinome. Excluding the ESBL-producing isolates, there was a prevalence of streptomycin (26.2%) and tetracycline (19.0%) resistance, with a notable number of isolates showing intermediate susceptibility to cephalexin (26.2%) and neomycin (21.4%).
Number and percentage of studied isolates which showed sensitive, intermediate, or resistant susceptibility profiles against the listed antimicrobial agents
Molecular characterisation of isolates
The 42 K. pneumoniae isolates originated from 28 dairy herd premises, 7 isolates came from the same farm, 4 from another farm, 2 from five other farms with the rest being single isolates from the remaining 21 premises. Their genome size varied from 5236065 to 5764556 Mb with a mean of 5454345 Mb (Supplementary Table S2). The G + C content ranged from 56.6 to 57.6 mol% with a mean of 57.2 mol% (Supplementary Table S2). The 42 isolates belonged to 31 different multi-locus sequence types (https://bigsdb.pasteur.fr/ ) each of which was represented by one or two isolates with the exception of seven ST107 isolates. One isolate (30881_C069646/2) belonged to a new sequence type – ST6855. The mean pairwise single nucleotide polymorphism (SNP) distance between isolates was 9234 SNPs (range 0–10665) (Supplementary Table S2). There were seven pairs of identical/near-identical isolates (separated by 0–7 SNPs). In four of these cases, the two isolates came from different farms (Figure 1).
Mid-point rooted, neighbour-joining tree showing phylogenetic relationships among K. pneumoniae mastitis isolates from Scotland. Tree constructed using pairwise SNP distances across 1972 K. pneumoniae core genes and produced by Pathogenwatch [Reference Argimón, David and Underwood25]. Multiple isolates came from Farms B (7 isolates) and V (5 isolates), and these are highlighted in green and yellow, respectively. ST107 isolates are highlighted in a black box. The presence of yersiniabactin is denoted by red star and antimicrobial resistance genes are shown by squares as being either present (filled black square) or absent (empty square).
AMR determinants
Ten different acquired AMR genes were present among the 42 isolates with the aminoglycoside phosphotransferases strA/APH(3″)-Ib and strB/APH(Reference Zadoks, Griffiths and Munoz6)-Id being the most abundant (present in 12 isolates) (Figure 1). All but one isolate carried the chromosomally encoded SHV β-lactamase with eight different alleles being present. None of these alleles carry known class-modifying mutations, i.e. conferring resistance to extended-spectrum β-lactams or β-lactamase inhibitors [Reference Lam, Wick and Watts26]. Two isolates, 30837_C348868 and 30838_C349123, were ESBL producers and encoded CTX-M-15. Despite originating from the same premises (Supplementary Table S2), these isolates are phylogenetically unrelated based on core genome SNPs (Figure 1).
Virulence determinants
Isolates were examined for the presence of the virulence-related genes encoding acquired siderophores (yersiniabactin, salmochelin, aerobactin), the genotoxin colibactin, the hypermucoidy locus rmpADC, and alternative hypermucoidy marker gene rmpA2. These were absent from all isolates except for ybST (encoding yersiniabactin) which was present in six isolates, of which only two are closely related (Figure 1 and Supplementary Table S1).
Prevalence of ST107
The most common multi-locus sequence type was ST107, with seven isolates being detected from six farms. These were investigated further by phylogenetic comparison with all 92 ST107 isolate genomes available from Pathogenwatch (accessed 17/01/2023) (Supplementary Table S3). Metadata showed that these 92 isolates came from 17 countries and from the continents of Australia (2 isolates), Africa [Reference Petersson-Wolfe, Leslie and Swartz1], Asia [Reference Saini, McClure and Léger11], Europe [Reference Letunic and Bork27], North America [50], and South America [Reference Petersson-Wolfe, Leslie and Swartz1]. These isolates were from cattle (44 isolates), humans (44 isolates), chickens [Reference Petersson-Wolfe, Leslie and Swartz1], and unknown sources [Reference Klaas and Zadoks3]. The resultant tree (Figure 2) showed that isolates from this study did not all cluster together within the ST107 global population. While each isolate was always most closely related to another study isolate, they were all also closely related (10–19 SNPs) to three bovine isolates from North America (Figure 2).
Mid-point rooted, neighbour-joining tree showing phylogenetic relationships among international isolates K. pneumoniae ST107. Tree constructed using pairwise SNP distances across 1972 K. pneumoniae core genes and produced by Pathogenwatch [Reference Argimón, David and Underwood25]. Geographical location of origin and host are indicated by a coloured key. Labels of study isolates are shaded grey, and UK is differentiated from Europe for the purposes of this figure.
Discussion
Klebsiella pneumoniae is a well-studied pathogen in the context of human infection [Reference Gorrie, Mirčeta and Wick33, Reference Wyres, Lam and Holt34], but less research has been carried out on bovine mastitis isolates [Reference Zheng, Gorden and Xia35]. The aim of this study was to carry out a systematic analysis of K. pneumoniae isolates causing sub-clinical and clinical mastitis in Scotland, focussing on comprehensive phenotypic identification, antimicrobial sensitivity, molecular characteristics, and population structure.
First, there was marked diversity in K. pneumoniae isolates by MLST, identifying 31 sequence types among the 42 strains. This finding has been mirrored in previous work, which also showed high diversity in K. pneumoniae genomes [Reference Zheng, Gorden and Xia35, Reference Yang, Xiong and Barkema36]. Klebsiella species are common inhabitants on the teat skin of dairy cattle [Reference Munoz, Bennett and Ahlström37]; thus, the sample collection method may have a part to play in the diversity of Klebsiella species found in milk samples [Reference Zadoks, Middleton and McDougall38] and could also be suggestive of an environmental source of infection [Reference Massé, Dufour and Archambault13]. Despite the heterogeneity, the same strains have been implicated in infections on the same units in this study, and in previous mastitis studies [Reference Munoz, Welcome and Schukken39, Reference Paulin-Curlee, Sreevatsan and Singer40], which may suggest a common source of infection.
Second, despite the diversity in this dataset by MLST, 7 of 42 isolates belonged to ST107. The frequency of the remaining STs was two or less, highlighting the diversity of K. pneumoniae strains on-farm. These ST107 strains were observed across six farms. These isolates were included in a meta-analysis with other ST107 strains globally, which appeared to be more related overall to other bovine isolates than other species. Indeed, other studies have found that ST107 is predominant [Reference Zheng, Gorden and Xia35, Reference Yang, Xiong and Barkema36, Reference Klaper, Hammerl and Rau41, Reference Cheng, Zhou and Nobrega42]. Other work has shown that ST107 are one of the most common strains associated with clinical mastitis [Reference Song, He and Yang43], which could potentially suggest lateral spread of the infection rather than being strictly environmental [Reference Schukken, Bennett and Zurakowski10]. This could be of clinical significance since it becomes more of a risk that infection could occur from other sources, such as milking equipment, increasing the likelihood of a herd-level infection rather than sporadic infection.
Third, virulence factors were considered using whole-genome sequencing data. The study isolates were examined for genes encoding yersiniabactin, salmochelin, aerobactin, colibactin, the hypermucoidy locus rmpADC and the alternative hypermucoidy marker gene rmpA2. Most of these virulence factors were not observed, except for ybST, which encodes yersiniabactin. This gene was found in six of the study isolates, from five different STs (ST37, ST458, ST111, ST1109 (n = 2), and ST976). Previous work has highlighted that many of the virulence genes mentioned previously are less common in bovine isolates in comparison to human isolates [Reference Song, He and Yang43–Reference Ludden, Moradigaravand and Jamrozy45] and are associated with increased virulence. However, we do not have the data resolution as part of this study to determine the role of yersiniabactin in virulence.
Fourth, the AMR phenotypes were well predicted by known AMR genes. Ten different acquired AMR genes were observed across the dataset, with the aminoglycoside phosphotransferases strA/APH(3″)-Ib and strB/APH(Reference Zadoks, Griffiths and Munoz6)-Id being the most prevalent (present in 28.6% of isolates). Indeed, 26.2% of the study isolates showed phenotypic resistance to streptomycin and 21.4% showed intermediate sensitivity to neomycin (both aminoglycoside antibiotics). Phenotypic resistance to tetracycline was also prevalent (19% of isolates), with tetA being observed in 4.8% of isolates, and both tetB and tetR genes being observed in pairs across 14.3% of isolates. Intermediate sensitivity to cephalexin was observed in over a quarter of isolates (26.2%), and all bar one isolates carried the chromosomally encoded SHV β-lactamase, showing eight different alleles. None of the eight alleles carry known class-modifying mutations linked to resistance to extended-spectrum β-lactams or β-lactamase inhibitors.
Finally, linked to the above, we identified two suspected ESBL-producing isolates from the same premises (encoding CTX-M-15) in this dataset by whole-genome sequencing, which was confirmed phenotypically. These isolates showed multi-drug resistance to amoxicillin–clavulanic acid, streptomycin, tetracycline, cefotaxime, cephalexin, and cefquinome. Critically, no study isolates showed carbapenemase activity, but there has been a series of recent work reporting carbapenem resistance in bovine mastitis isolates [Reference Silva-Sanchez, Barrios-Camacho and Hernández-Rodriguez18, Reference He, Wang, Sun and Pang19, Reference Saddam, Jamal and Rahman46, Reference Bonardi, Cabassi and Fiaccadori47], highlighting the importance of disease and AMR surveillance work.
Conclusions
To conclude, the genetic diversity in the described K. pneumoniae isolates is high with the most dominant ST107 potentially linked to lateral spread. The number of well-described virulence genes present was low, with ybST (encoding yersiniabactin) being observed in a small number of isolates. Higher levels of AMR to aminoglycosides and tetracycline were present, with two ESBL-producing isolates being observed. This study highlights the importance of monitoring bacterial isolates in clinical cases considering concerns about multi-drug resistance. This work should inform more extensive future studies including a larger number of farm units over a period of time.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0950268824001754.
Data availability
Whole-genome sequence data are available on Genbank (BioProject PRJNA943144, accession numbers and genome descriptive statistics are provided in Supplementary Table S3).
Acknowledgements
We thank the Institut Pasteur teams for the curation and maintenance of BIGSdb-Pasteur databases at https://bigsdb.pasteur.fr/.
Author contribution
Conceptualization: G.P., M.R.H., J.B., K.H., J.P.; Data curation: G.P., G.F., J.B., K.H., J.P.; Formal analysis: G.P.; Investigation: G.P., G.F., K.H., J.P.; Methodology: G.P., G.F., J.B., K.H., J.P.; Resources: G.P., M.R.H., J.P.; Software: G.P.; Supervision: G.P., M.R.H., G.F., J.P.; Visualization: G.P., J.P.; Writing – original draft: G.P., M.R.H., G.F., J.B., K.H., J.P.; Writing – review & editing: G.P., M.R.H., G.F., J.B., K.H., J.P.; Funding acquisition: M.R.H., J.P.; Project administration: M.R.H., J.P.
Competing interest
The authors declare none.
Funding statement
The whole-genome sequencing was funded by the Hannah Dairy Research Foundation, as part of the HDRF Small Grant Scheme (Principal Investigator: JP). MRH receives funding from the Scottish Government. JP, JB, and GF receive funding as part of the Scottish Government Disease Surveillance Programme.
Open access
