Skip to main content
Download PDF

Unravelling the complexity of bovine milk microbiome: insights into mastitis through enterotyping using full-length 16S-metabarcoding

Animal Microbiome volume 6, Article number: 58 (2024) Cite this article

Abstract

Background

Mastitis, inflammation of the mammary gland, is a major disease of dairy cattle and the main cause for antimicrobial use. Although mainly caused by bacterial infections, the aetiological agent often remains unidentified by conventional microbiological culture methods. The aim of this study was to test whether shifts in the bovine mammary gland microbiota can result in initiation or progression of mastitis.

Methods

Oxford-Nanopore long-read sequencing was used to generate full-length 16S rRNA gene reads (16S-metabarcoding) to characterise the microbial population of milk from healthy and diseased udder of cows classified into five groups based on their mastitis history and parity.

Results

Samples were classified into six enterotypes, each characterised by a marker genus and several differentially-abundant genera. Two enterotypes were exclusively composed of clinical mastitis samples and displayed a marked dysbiosis, with a single pathogenic genus predominating and displacing the endogenous bacterial population. Other mastitis samples (all subclinical and half of the clinical) clustered with those from healthy animals into three enterotypes, probably reflecting intermediate states between health and disease. After an episode of clinical mastitis, clinical recovery and microbiome reconstitution do not always occur in parallel, indicating that the clinical definition of the udder health status does not consistently reflect the microbial profile.

Conclusions

These results show that mastitis is a dynamic process in which the udder microbiota constantly changes, highlighting the complexity of defining a unique microbiota profile indicative of mastitis.

Background

Mastitis is defined as the inflammation of the mammary gland and, although it can be caused by chemical, physical, or biological factors, the latter are the primary cause. While mastitis can sometimes be caused by archaea or viruses, bacteria are predominantly responsible for these infections [1]. Based on the severity of the symptoms, mastitis can be classified as clinical or subclinical, and both are associated with somatic cell counts (SCC) above 200,000 cells/ml [2]. Clinical mastitis is associated with visible alterations in the milk (such as lumps, clots, blood, or colour changes) and signs at the udder level (including inflammation, redness, heat, and pain), and in the most severe cases, systemic signs (such as fever, anorexia, and lethargy). Subclinical mastitis, on the other hand, does not produce visible alterations in the milk or udder, but is associated with productivity losses. In fact, bovine mastitis is one of the most significant diseases in dairy cattle due to its high prevalence and the economic losses it entails attributed to reduced milk production, milk wastage, and veterinary treatment expenses [3].

Microbiological culture methods continue to be the gold standard for the diagnosis of mastitis, but they have serious limitations since the aetiological agent is not identified in 40–50% of cases, either due to the growth of multiple types of colonies or to absence of significant growth [4, 5]. PCR-based molecular approaches provide higher sensitivity but only a limited number of target species can be detected. In recent years, high throughput sequencing (HTS) techniques have become more accurate and affordable, allowing the cost-effective study of whole microbial communities (microbiota) of practically any niche. Many studies have characterised the bovine milk microbiota in different situations, but only a few have addressed the question of how the milk microbiota behaves in the context of animal health and, specifically, in cases of mastitis [6,7,8,9,10]. However, the microbial composition of milk associated with the different presentations of mastitis is still far from being fully elucidated. To date, 16S rRNA gene sequencing remains the most extensively used HTS approach for the characterisation of bacterial microbiota associated with mastitis [11,12,13,14] and has helped to refute the common thought that milk is sterile [15] by the identification of several bacterial groups that went undetected with conventional culture-based methods [5, 11, 16]. However, most of the microbiome studies in milk are based on short-read sequencing technologies, which can only cover small regions of the 16S rRNA gene. This makes the results unreliable for microbiome profiling beyond the family taxonomic level because assignation rates to lower taxonomic levels are very low [17]. Therefore, the composition of the milk microbiome is still far from being completely characterised. Alternatively, long-read sequencing using Oxford Nanopore Technologies (ONT) provides a cost-effective method to cover the entire 16S rRNA gene, potentially allowing finer taxonomic resolution down to the species level compared to short-read sequencing. Thus, this study was designed to explore the complex microbiota structure of bovine milk under different conditions, such as milking phase, parity, and history of mastitis by sequencing of the complete 16S rRNA gene using ONT, to investigate the relationship between the bacterial community of milk and udder health status.

Methods

Experimental design and sample collection

The study was carried out in a Holstein–Friesian dairy cattle farm, located in Navarra (Spain), with 650 lactating animals housed in open lots with a roofed shelter. Bedding in the free-stall sheltered area consisted of straw that is tilled daily using a rototiller to turn over the surface (15–25 cm) to aerate it. In the open recreational area, the floor is made of concrete where cow manure accumulates. The diet of lactating cows remains constant throughout the year and consists of silage (grass, maize and dehydrated alfalfa) supplemented with beer bagasse and orange pulp, and concentrated feed.

Lactating cows are milked twice a day. The mean milk yield in the study period was 38.6 L of milk/animal/day, and the average days in milk was 160 days. The animals were dried out abruptly and, on average, the dry period lasted about 90 days. Mamyzin secado (100 mg penetamate iohydrate, 280 mg benzylpenicillin benetamine, and 100 mg framycetin sulfate; Boehringer Ingelheim) was applied to all the animals (blanket antimicrobial dry-cow therapy) until March 2022, when the drying management changed, and a selective drying protocol was initiated. The incidence of mastitis on the farm was 4.7% for clinical mastitis (with a high seasonality) and 20–22% for subclinical mastitis. Data on somatic cell counts (SCC) were collected every month from combined milk from the four quarters.

Milk sampling was carried out by the farm veterinary staff between May 2021 and September 2022. The animals were categorized into five study groups based on their parity, udder health status, and mastitis history at the time of sampling: healthy adult (multiparous) cows with no history of mastitis in the previous lactation (HC), healthy heifers (HH), cows recovered from a previous episode of mastitis (HR), cows diagnosed with clinical mastitis (CM), and cows with subclinical mastitis (SM). In all groups, samples were collected at three different phases of lactation: Initial (I): 10–25 days in lactation; Peak (P): 50–85 days in lactation for cows and 65–100 days for heifers; Late (L): over 180 days in lactation. Clinical mastitis was defined based on the presence of visible abnormalities in milk (e.g. colour alteration, flakes, and/or clots) and signs of udder inflammation (swelling, heat, pain, redness). Animals with clinical mastitis were sampled when visible milk and udder alterations were noted and before any antibiotic treatment. For all other animals, a California Mastitis Test (CMT) was performed on individual quarters before sampling to detect the presence of subclinical infections (positive CMT).

Milk samples were collected from individual quarters (quarter milk samples). The first streams of milk from each sampled quarter were discarded to minimise contamination, and, whenever possible, at least 30 ml of milk were collected into sterile plastic tubes. The samples were sent refrigerated (4 °C) to the laboratory within 24 h of sampling. A total of 281 samples from individual quarters were collected and distributed among the five experimental groups: 62 HC, 63 HR, 65 HH, 38 CM, and 53 SM (Table 1). Thus, 190 animals were considered healthy at the time of sampling (HC, HH, HR) and 91 were considered diseased (CM and SM). While most of the animals were sampled once, repeated samples were collected from 25 animals at different time points, either from the same quarter (n = 12) or from different quarters (n = 13).

Table 1 Distribution of the samples in the different experimental groups and milking phases
Full size table

Microbiological culture

Samples from CM and SM were analysed by conventional bacterial culture on COS Columbia agar with 5% sheep blood. Bacterial colonies were examined by Gram stain and identified by biochemical tests with a VITEK2 system (bioMeriéux).

16S rRNA gene high-throughput sequencing using Oxford Nanopore Technologies

Upon receipt, milk samples (30 ml) were processed using two centrifugation steps as a skimming treatment to prepare them for molecular analysis. Initially, samples were centrifuged at 4,500 g for 30 min at 4 °C, followed by the removal of the cream and the supernatant. Then, the obtained pellet was resuspended in 1 ml of cold, sterile phosphate buffer saline (PBS) (1x) and subjected to a second centrifugation at 10,000 g for 10 min at 4 °C. Again, the cream and the supernatant were discarded, and the pellet was resuspended in DNA/RNA Shield (Zymo Research Corp.) to stabilise the nucleic acids.

DNA was extracted using ZymoBIOMICS DNA 96 MagBead Kit (Zymo Research Corp.) following the manufacturer´s protocol, on the KingFisher Flex robot (Thermo Scientific). Negative controls (PBS only) were included and processed as above to rule out possible contamination during the laboratory processing. Libraries were prepared according to the 1–24 16S Barcoding (SQK-16S024) protocol (Oxford-Nanopore Technologies, ONT). This step involved PCR amplification of the complete 16S rRNA gene from all the bacteria in the samples. Cycling conditions consisted of an initial denaturation step at 95 °C for 1 min followed by 35 amplification cycles at 95 °C for 20 s, 55 °C for 30 s and 65 °C for 2 min, with a final extension at 65 °C for 5 min in a SimpliAmp thermocycler (Applied Biosystems by Thermo Fisher Scientific). Library preparation was performed following the standard ONT protocol. To improve DNA recovery, the incubation during the last elution step was carried out at 30ºC for 15 min. The DNA concentration was quantified with Qubit®2.0 (Invitrogen). Sequencing was carried out following the 1–24 16S Barcoding (SQK-16S024) protocol from ONT on R9 flow cells (FLO-MIN106D, ONT) using a MinION Mk1C instrument. The established sequencing parameters were a QScore greater than 8 and a fragment size ranging from 0.9 to 2.2 kb. The target was to achieve approximately 50.000 reads per sample.

Bioinformatic analysis

Reads were base-called on high accuracy mode (HAC), and barcodes trimmed using Guppy (v5.0.12 and v6.1.5). ONT adapters were removed with Porechop (v0.2.3) and sequence quality was assessed with FastQC (v0.11.9). Taxonomic assignment was performed with Kraken2 (v2.1.1) using RefSeq Standard (2021–05-17) as the reference database, with a confidence threshold set at 0.1. All these tools were freely available at Galaxy (https://usegalaxy.eu/).

Statistical analysis

Downstream analyses and graphical outputs were generated with different packages in R (v4.1.2) [18]. The first step was the removal of non-bacterial taxa and taxa with less than 10 read counts. Rarefaction curves were constructed using the ranacapa package (v0.1.0) [19] on samples to assess the sequencing effort. Phyloseq (v1.38.0) [20], microbiome (v1.16.0) [21], and stats (v4.3.2) [18] R-packages were used to estimate biodiversity, perform normalization and visualize the abundance of microbial taxonomic composition. A preliminary analysis was conducted considering the five working groups. Then, the Enterotype definition (clusters) was made in R following the EMBL-EBI tutorial to establish the analysis groups.

Microbial community diversity was estimated with both alpha and beta diversity metrics. For alpha diversity, estimations of species richness and evenness were calculated using Chao1 and Shannon indices, respectively. These metrics were compared across the different groups, milking phases, and enterotypes using the non-parametric Kruskal–Wallis test, followed by post-hoc pairwise comparisons with the Wilcoxon rank sum test. To address the variability in the number of reads between samples, normalisation was made using the total sum scaling (TSS). Beta diversity analysis was calculated in MicrobiomeAnalyst 2.0 [22]. The data filtering parameters set included a count prevalence threshold of 10% in samples, a low variance filter based on the inter-quartile range, and data scaled with TSS without rarefying or transforming. The Jensen Shannon Distance (JSD) metric was used to assess the dissimilarity matrix between pairs of samples, results were visualized with PCoA, and permutational multivariate analysis of variance (PERMANOVA) was performed to detect significant differences between enterotypes.

To look for differentially abundant genera among the different enterotype clusters, the Linear Discriminant Analysis Effect size (LEfSe) algorithm was used with a log LDA score of 5.0. Additionally, differentially abundant species with a score threshold of 4.0 were identified to assess potential functional differences. Both levels employed a p-value cut-off of 0.05, adjusted for the false discovery rate (FDR). The functionalities associated to the species identified by LEfSe were predicted using FAPROTAX, a stand-alone tool that uses a manually curated database to assign established metabolic or other ecologically relevant functions to prokaryotic clades [23].

Genera present with a relative abundance above 1% in at least 10% of the samples within each of the six enterotypes were identified using MicrobiomeAnalyst 2.0, and the results were visualized in R using the package microbiomeutilities (v1.00.17) [24]. Pearson’s correlation coefficients for all pairs of genera were calculated using the corrr (v2.1.6) package in R [25]. The correlogram, which illustrates the top 50 genera that show a significant correlation with at least one other genus, was created using the corrplot (v0.92) package [26].

Results

Conventional microbiology culture results differed between CM and SM

In milk samples from CM, the predominant microorganisms isolated were E. coli (39.5%, n = 15), followed by Bacillus spp. (18.4%, n = 7), and Streptococcus spp. (13.2%, n = 5). In SM, the most frequently isolated microorganisms were Corynebacterium spp. (20.8%, n = 11) and non-aureus staphylococci (NAS) (18.9%, n = 10). In the remaining samples, the microbiological culture led to a negative result (no growth) or contamination (defined as growth of more than two different microorganisms) accounting for 28.9% (10 negative and 1 contaminated) of the samples in CM and in 49.1% (21 negative and 5 contaminated) of samples in SM.

ONT sequencing captured most of the milk bacterial biodiversity

After removing non-bacterial taxa (0.06% of the total reads) and filtering out taxa with fewer than 10 read counts, a total of 13,127,110 reads remained, with a median of 49,243 reads per sample (Q1-Q3 of 44,894–50,218) and an average QScore of 20. A total of 2,965 species were identified, corresponding to 1,152 genera, 345 families, 150 orders, 65 classes, and 32 phyla. The rarefaction curves reached a plateau at approximately 40,000 reads (Additional file 1: Fig. S1), indicating that most of the biodiversity within the samples was captured and that the sequencing depth was therefore adequate.

Defined experimental groupings based on health status and parity did not adequately explain the observed biodiversity differences

A preliminary analysis of the alpha diversity (Chao1—richness and Shannon—diversity indices) showed higher microbial diversity in milk samples from healthy animals (HR, HH, and HR) compared to those from diseased animals (CM and SM). Similarly, microbial composition (beta diversity) was significantly different between groups (PERMANOVA, F = 15.496, padj = 0.001); specifically, between HC and HH, and between CM and the rest of the groups. However, the beta diversity plot showed that these differences were in some cases driven by several animals with microbiome profiles that markedly deviated from others within their corresponding experimental group. These differences were not consistently associated with the experimental grouping, health status or parity. Therefore, we used an alternative approach to classify the animals based on their microbiome, using Enterotyping.

Enterotyping analysis identified marker genera linked to milk from healthy animals and animals with mastitis

Enterotyping analysis revealed that six clusters best suited our dataset. Each cluster or enterotype was characterised by a predominant driver genus, as shown in Table 2. Enterotypes 1, 2, 3, and 5 were mainly composed of samples from healthy animals (HC, HR, HH; 189/262), whereas nearly all samples in enterotypes 4 and 6 (18/19) were from CM. Samples from SM were distributed in the different enterotypes where healthy animals predominated. The main contributor (marker) genera of each enterotype defined by Enterotyping were also found within the top driver genera identified by LEfSe analysis (Table 2). In samples of enterotypes 4 and 6, the bacteria identified as marker genera matched the organism isolated from the milk by bacterial culture; the marker genera of the remaining enterotypes would not grow in the media used for conventional bacteria isolation.

Table 2 Description of the different enterotypes
Full size table

Differences in diversity were very pronounced between the enterotypes and weaker between experimental groups or milking phases

Overall, statistically significant differences in alpha diversity indices (Chao1 and Shannon) (p < 0.05) were observed between enterotypes (Fig. 1A), where enterotypes 4 and 6 had lower richness and evenness compared to the other enterotypes. The different milking phases were not associated with overall changes in richness or evenness. However, enterotype 2 was the exception, where samples collected at late lactation showed lower richness (Chao1) compared to those from initial lactation (Additional file 1: Fig. S2A).

Fig. 1
figure 1

Species-level milk microbial biodiversity comparisons between enterotypes. A Boxplots showing the alpha diversity indices for richness (Chao1) and evenness (Shannon). Boxes represent the interquartile range (IQR; Q1 and Q3), with the line inside marking the median. Whiskers extend to the smallest and largest values within 1.5 times the interquartile range, with points beyond these considered outliers. Each point represents a sample's diversity score. A non-parametric Kruskal–Wallis test followed by post-hoc Wilcoxon rank sum test was performed for pairwise comparisons between enterotypes. Asterisks indicate statistically significance levels (*p = 0.05; ** p = 0.01; *** p = 0.001; **** p = 0.0001). B Bidimensional Principal Coordinate Analysis (PCoA) based on Jensen-Shannon Distance (JSD) dissimilarity matrix illustrating differences in the microbiome profiles among the different enterotypes. Colours represent different enterotypes, while ellipses indicate 95% confidence intervals of the multivariate t-distribution around centroids of each groupings with enterotype as factor. Differences were assessed statistically using PERMANOVA and pairwise comparisons results are provided in Additional file 2: Table S1

Full size image

Microbial diversity within enterotypes 1, 2, and 3 varied depending on the experimental groups. Within enterotype 1, animals with subclinical mastitis (SM) showed lower microbial richness compared to healthy animals (HC, HR, and HH). In the case of enterotype 2, a higher richness was found in healthy heifers (HH) compared to animals with subclinical mastitis (SM) and those recovered from clinical mastitis (HR). In enterotype 3, milk from healthy heifers (HH) not only showed a greater richness in comparison to animals with subclinical mastitis (SM) and those recovered from clinical mastitis (HR) but also showed higher richness and evenness when compared to healthy adult cows (HC) (Additional file 1: Fig. S2B).

Regarding the bacterial community structure, marked differences in beta diversity were observed among all enterotypes when compared pairwise (PERMANOVA, F = 106.99, p = 0.001) (Additional file 2: Table S1; Fig. 1B). Bacterial community composition generally remained similar throughout the different milking phases across all enterotypes, except for enterotype 2, where differences were observed between the initial and peak phases of lactation (PERMANOVA, F = 1.88, p = 0.006).

ONT sequencing allowed the definition of the bacterial community composition down to the species level

Firmicutes was the main phylum in all enterotypes (ranging from 68.8–99.2%) except for enterotype 4, where Proteobacteria predominated (95.4%), and Firmicutes, albeit being the second phylum, only accounted for 4.3%. In the rest of the enterotypes, Proteobacteria was the second predominating phylum followed by Actinobacteria and Bacteroidetes in the third and fourth positions, respectively. The compositional barplot at the genus level showed higher variation between enterotypes (Fig. 2). Enterotypes 1, 2, and 3 showed higher microbial diversity, in agreement with alpha diversity results, whereas in enterotypes 4, 5, and 6 a marked predominance of their specific driver genera was observed. Then, we analysed the species-level composition of the two enterotypes associated with clinical mastitis. Results showed that in all samples of enterotype 4, the concomitant presence of Escherichia coli, Klebsiella pneumoniae, and Salmonella enterica was consistently identified (Additional file 1: Fig. S3), and in enterotype 6, the predominant genus was Streptococcus, but six different species were identified (Additional file 1: Fig. S3). In enterotype 5, several Staphylococcus species were detected, the most abundant being Staphylococcus chromogenes and Staphylococcus aureus. Although most of the samples within each enterotype exhibited a similar composition, there were exceptions where individual samples showed slightly different profiles. For example, Pasteurella was more prevalent in two samples from enterotype 3.

Fig. 2
figure 2

Stacked bar-plot illustrating the relative abundance of bacterial genera in each of the six enterotypes. The ten most abundant genera are colour-coded as indicated in the legend and all other genera are grouped into "Other"

Full size image

Bacterial genera frequently present in the milk microbiome were identified and they varied in relative abundance across all enterotypes

A total of 31 genera were present in at least 10% of the milk samples of each enterotype with an abundance above 0.01% (Fig. 3). The most abundant four genera were Clostridium, present in 94.7% of the samples, Romboutsia (92.9%), Clostridioides (88.6%), and Paeniclostridium (85.4%). Relative abundance of the different genera varied across enterotypes, especially between healthy and mastitis-associated enterotypes (Fig. 3).

Fig. 3
figure 3

Bacterial genera present in > 10% of samples and a relative abundance > 1% within each enterotype. Colours in each cell represent the relative abundance of the genera, as indicated in the legend

Full size image

Correlation analysis identified potential positive and negative pairwise associations between genera

Most significant correlations were positive, with particularly strong correlations (ρ > 0.7) being exclusively observed among these positive associations. Detailed results for all pairwise correlations are available in Additional file 2: Table S2, and the top 50 genera that show a significant correlation with at least one other genus are illustrated in Additional file 1: Fig S4. A total of 127 genera demonstrated very strong positive correlations (ρ > 0.9) with at least one other genus. Examples of near-perfect positive correlations included Enterobacter with Klebsiella, Aromatoleum with Oryzomicrobium, and Salmonella with Shigella, each achieving a ρ of 0.99. Candidatus Sulcia produced the highest number of significant positive correlations, totalling 10, followed by Anaerococcus, Mycoplasma, and Citrobacter, each with 5.

Focusing on the marker genus of each enterotype (Table S3), Escherichia (enterotype 4) showed very strong correlations (ρ ≥ 0.8) with Kosakonia, Salmonella, Shigella, Rahnella, and Klebsiella. Romboutsia (enterotype 1) presented the strongest negative correlation with several genera such as Escherichia, Salmonella, Yersinia, Shigella, and Serratia, with ρ values around -0.4 and was positively associated with Turicibacter (ρ = 0.65). Fastidiosipila (enterotype 2) showed positive correlations with Mageeibacillus (0.86), Thermoclostridium (0.72), Mahella (0.71), and Flavobacterium (0.70). No correlations with ρ values above 0.4 were found for the marker genera of enterotypes 3 (Psychrobacter) and 5 (Staphylococcus).

Microbiome dynamics in mastitis recovery and variability across udder quarters show that clinical recovery and microbiome reconstitution do not always occur in parallel

Several animals sampled more than once provided insights into the recovery and re-establishment of the endogenous udder microbiome after an episode of clinical mastitis. For example, three animals (618, 554, and 90) had the same quarter sampled twice, firstly when diagnosed with clinical mastitis (CM) and secondly 1–3 months later when classified as having clinically recovered (HR). These animals exhibited dysbiosis due to Escherichia, Klebsiella, and Salmonella (animal 618) (Fig. 4A), Streptococcus (animal 554) (Fig. 4B) or Pasteurella (animal 90) (Fig. 4C). While animals 618 and 554 showed a microbiome profile that indicated complete recovery from the dysbiosis one and three months after the diagnosis of the clinical mastitis, respectively, animal 90 displayed only a tentative shift towards recovery one month later. Its microbiome still showed a significant presence of Pasteurella, and the diversity typical of a healthy microbial community had not yet been re-established.

Fig. 4
figure 4

Changes in bacterial genera relative abundance at two timepoints in quarters of animals sampled twice. A Animal 618; B Animal 554; and C Animal 90. The 15 most abundant genera are colour-coded as indicated in the legend, with all other genera grouped into "Other". The time interval between samples, in months, is specified below the name of each sample

Full size image

Functional characterisation of differentially abundant species identified unique functions within the enterotypes

LEfSe analysis identified 82 bacterial species as driver species across the 6 enterotypes (Additional file 2: Table S4). Among these, we identified both pathogenic and opportunistic bacteria in all enterotypes and predicted 89 functionalities. From this, 23 were unique to specific enterotypes (Additional file 2: Table S5). Every enterotype, except number 3, included at least one unique function. These unique functionalities helped identify the species that may play a crucial role in the determination of a particular enterotype. Unique functions in enterotypes 1 and 2 were not associated with mastitis-producing bacterial species, but several opportunistic and pathogenic species were detected. Conversely, enterotypes 4, 5, and 6 presented unique functions that were associated with mastitis-producing or highly pathogenic organisms (Additional file 2: Table S5).

Discussion

To investigate the relationship between the milk microbiota and the udder health status, long-read ONT sequencing has been used in this study to determine the full-length of the 16S rRNA gene of the bacterial community of milk from animals categorised into five study groups based on their parity, udder health status, and mastitis history. ONT sequencing offers a cost-effective approach to cover the entire 16S rRNA gene, which is a major challenge when using short-read sequencing. Thus, ONT offers potentially enhanced species-level resolution for microbiome profiling [17, 27]. In recent years, ONT sequencing technology has made significant advances, particularly in reducing error rates, and the mean QScore of 20 achieved in this study represents an accuracy of 99% [28]. Still, species-level inference is sometimes hampered by full-length 16S rRNA database limitations and the availability of adapted taxonomical assignation tools. The use of 16S ONT-metabarcoding is increasingly common for bacterial community profiling [29] but its application to studies of the bovine milk microbiome is still scarce [6, 30].

Our initial analysis of alpha diversity revealed a higher microbial richness and evenness in milk from healthy animals compared to those with mastitis, in agreement with most studies that have reported a loss of microbial diversity and an increase in pathogenic bacteria in the milk of animals with both clinical and subclinical mastitis [7,8,9, 31]. On the other hand, beta diversity analyses indicated significant differences in microbial compositions between the CM group and the other groups, but these differences were primarily driven by a subset of CM samples with clearly distinct microbiome profiles characterised by dysbiosis. Upon excluding these samples, the differences between the microbiomes of milk from healthy animals and those with mastitis were no longer apparent, suggesting that not all animals with mastitis (clinical or subclinical) show microbial profiles distinct from healthy individuals, as reported elsewhere [10, 12]. Therefore, we decided to classify animals based on the composition of their microbiome rather than our initial experimental groupings, to better understand the microbial dynamics underlying health and mastitis. To do so, we used "Enterotyping", an approach originally developed to identify distinct bacterial community clusters in the human gut microbiome [32] that has since been applied to various other biological niches [33,34,35]. This analysis resulted in the classification of the 281 milk samples into six enterotypes, each with different bacterial community composition and driven by specific differentially abundant genera. Two enterotypes (4 and 6) were exclusively composed of CM samples that displayed a marked dysbiosis, characterised by the predominance of a single pathogenic genus that wiped out the endogenous bacterial population. In contrast, the other samples with mastitis (all SM and half of CM) clustered with those from healthy animals across enterotypes 1, 2, and 3. Unexpectedly, dysbiosis was also detected in Enterotype 5, a small group exclusively formed by healthy animals.

Regarding the taxonomical composition of milk samples from healthy and diseased quarters, there is a general agreement on the four predominant phyla, i.e. Firmicutes, Proteobacteria, Bacteroidetes, and Actinobacteria. However, greater diversity has been reported at the genus level, with no consensus in the literature (reviewed by [1]). Although sequencing methodology and bioinformatic analyses may contribute to this lack of consensus at lower taxonomical levels, management and environmental factors like diet, seasonality, litter or stable [7] as well as individual factors like breed [14] have been shown to influence the bacterial composition of milk. To minimise the effect of such factors, this study was carried out on a single farm, ensuring that the observed bacterial diversity was not affected by differing management practices or environmental conditions. Logistical constraints prevented the sampling to be concentrated in a single season, but no clear effect of seasonality was observed.

The same four phyla described elsewhere prevailed in all the enterotypes described in our study. However, their proportions differed among enterotypes. For example, in enterotype 4, mostly composed of samples from animals with clinical mastitis, Proteobacteria clearly predominated over Firmicutes. This was due to the dominance of a single bacterial genus, Escherichia, that accounted for approximately 80% (range: 39.5–84%) of reads in all samples of enterotype 4. In the other enterotypes, E. coli was also present but at significantly lower concentrations. E. coli is recognised as a leading cause of acute clinical mastitis in dairy cattle worldwide [36]. Mammary pathogenic E. coli (MPEC) has been proposed as a new pathotype responsible for causing mastitis in dairy animals [37, 38], but as no specific virulence factors have been identified for all MPEC isolates, some consider E. coli isolated from mastitis-affected cattle to be environmental opportunistic pathogens [36]. Our results show that E. coli can invade the mammary gland, survive, and multiply in milk to such high concentrations that they displace the endogenous udder bacteria population leading to dysbiosis that results in clinical mastitis. Interestingly, in enterotype 4, E. coli always appeared in combination with Klebsiella pneumoniae and Salmonella enterica. Klebsiella pneumoniae is a known cause of environmental mastitis in cattle and is commonly found in manure and organic bedding [39]. However, Salmonella is not considered a causative agent of mastitis in dairy cattle, although it can be isolated from the mammary glands of dairy cows [40, 41]. The strong negative correlation of Escherichia (marker genera of enterotype 4) with Romboutsia (marker genera of enterotype 1) is interesting and suggests that large abundances of the latter might protect against E. coli proliferation. Another enterotype characterised by a clear dysbiosis and associated with animals suffering from clinical mastitis was enterotype 6, in which the marker genus, Streptococcus, accounted for more than 90% of all genera, and included several well-known mastitis-producing species like S. dysgalactiae, S. uberis or S. agalactiae [42].

Enterotype 5 was also characterised by the predominance of a single genus, Staphylococcus, that includes pathogenic species associated with clinical and subclinical mastitis [43, 44]. This enterotype included only six healthy animals, four heifers (HH) and two recovered cows (HR) in their second and fourth calvings, respectively. S. chromogenes (NAS), was the predominant species in the heifers and one of the cows (second calving). Some authors have described high Staphylococcus spp. abundances in heifers, especially at the initial period of lactation, but NAS infections in calving heifers only cause slight increases in SCC and mild, self-limiting inflammatory responses in the udder, without negatively affecting subsequent productivity [45, 46]. This could explain why these animals, despite the clear dysbiosis, did not show signs of mastitis. In the remaining cow (4 calvings) in Enterotype 5, S. chromogenes was not detected and S. aureus predominated instead. This cow was sampled just a few days before dry-off with antimicrobials. We could speculate that this animal may have developed mastitis had it remained in lactation.

Regarding the differentially abundant genera identified in enterotype 1 (Romboutsia, Clostridioides, Paeniclostridium, and Clostridium), they have already been detected in milk and are consistently more abundant in samples from healthy animals [7, 47,48,49], and only Clostridium has been associated with disease conditions or elevated SCCs [8, 13]. Similarly, Fastidiosipila (marker genus of enterotype 2) has never been linked to episodes of mastitis, but it has been isolated in cows with Bovine Digital Dermatitis (BDD) from interdigital lesions [50, 51]. The only differentially abundant genus of enterotype 2 known to be of interest as a mastitis producer is Pseudomonas but it has also been regarded as part of the milk microbiota of healthy animals [52, 53]. Both Psychrobacter and Jeotgalicoccus (enterotype 3) have been identified in milk, with Psychrobacter being considered part of its core microbiome [5, 11, 48, 54]. Although both genera have been detected in samples from healthy animals, some studies have reported higher abundances of these bacteria in cows with clinical and subclinical mastitis [2, 55].

Samples from animals with subclinical mastitis always clustered within enterotypes 1, 2 and 3, which mainly included samples from healthy animals. The observation that subclinical mastitis does not always lead to a significant shift in the taxonomic composition of milk microbial communities has been documented previously [7, 14]. Subclinical infections are more prevalent than the clinical form, infection can last longer, and, if undetected, they can easily spread within the herd [56]. However, subclinical mastitis are more difficult to diagnose, as there are no visibly apparent signs in the milk or the animals and increased SCC and decreased milk production are the only indicators. Besides, SCC can be affected by factors such as lactation number, milk yield, stress, and season [57]. Additionally, while bacteria are a recognised cause of mastitis, other factors such as immune responses, environmental stressors, and management practices also significantly contribute to the development of this condition [58]. Therefore, misclassification of certain animals within the experimental groups cannot be ruled out. Interestingly, half of the CM samples also exhibited a microbiome composition similar to that of milk from healthy animals and clustered within these three enterotypes. These findings suggest that depending on the causative agent, mastitis may not induce the same level of microbial composition disruption. Nevertheless, the microbial richness was generally lower in milk from animals with subclinical mastitis compared to their healthy counterparts within the enterotype, and microbial composition of milk from SM and CM samples within enterotypes 1, 2, and 3 was marginally different from that of HC (p = 0.052, data not shown), indicating changes in the proportions of the bacterial members instead of a complete shift in community composition. Alternatively, some of these animals might have been sampled during an intermediate situation between different conditions and their microbiome represented a transient profile that had not yet undergone significant alterations as observed in the animals sampled twice during the recovery process from an episode of clinical mastitis.

Although it is hard to draw conclusions from the small number of cases analysed, follow-up sampling of milk from the same udder quarters of the same animal—initially during clinical mastitis and then 1–3 months later, once recovery was diagnosed—showed that the process of microbiome re-establishment does not always coincide with the recovery of clinical symptoms. This process is probably influenced by factors such as the aetiological agent, the extent of infection/lesion, the animal immune response, the antibiotic treatments, the milking phase, and the overall management practices [58, 59]. For instance, E. coli typically causes acute infections of short duration (10–30 days), mainly occurring in the last two weeks of the dry period and in the first weeks postpartum [36], although reinfections may occur throughout lactation [60, 61]. In most cases, E. coli is eliminated by the host immune response [62, 63], often causing a significant physiological impact on the udder but leading to rapid recovery due to the availability of effective treatments [36]. Pasteurella spp. are opportunistic contagious pathogens rarely reported as a cause of bovine mastitis [64]. They are normally found in the upper respiratory tract of healthy livestock and spread to the mammary gland most probably occurs from cow to cow, although haematogenous or lymphogenic dissemination of Pasteurella multocida from the respiratory tract has also been reported [65]. Turni et al. reported a P. multocida strain associated with pneumonia in calves and mastitis in heifers [66]. Bovine mastitis caused by Pasteurella spp. do not generally respond well to antibiotic therapy [65] and this might explain the delay observed in microbiota recovery. Further studies that monitor over longer periods of time the milk microbiome of a larger number of animals are needed to fully understand the dynamics of the microbiota during the recovery process from mastitis associated with different pathogens.

The fact that certain genera were detected in the milk of all the enterotypes suggests that there might be a threshold above which disease development occurs, as would be the case of the opportunistic pathogen E. coli, a typical mastitis-causing microorganism. A higher abundance of other bacteria also present in most milk samples like Romboutsia might protect against the proliferation of pathogens. Corynebacterium reads were present in nearly all samples, including milk from healthy cows, which sometimes harboured five to ten times more Corynebacterium reads than samples from animals with mastitis. Corynebacteriaceae has been identified as one of the families composing the core microbiota of healthy cows [67]. Interestingly, Corynebacterium spp. were the most frequently isolated bacteria in SM samples in our study, but unfortunately, only milk samples from animals diagnosed with mastitis were subjected to microbiological culture.

Although the technique used (16S-metabarcoding) only allows taxonomic characterisation of the bacterial community, functional inference was attempted for the 82 bacterial species identified by LEfSe as driver species of each of the six enterotypes to interpret their role in the communities. It is noteworthy that the predicted unique functions in enterotype 2 (n = 15) were related to fermentative and cellulolytic activities, which may indicate that silage acted as a possible source of the bacterial community. These functions were associated with seven species that included two pathogens, Pseudomonas aeruginosa and Bacillus cereus, which can both produce clinical or subclinical mastitis in dairy cattle [52, 68]. Unique functions predicted for enterotype 4 were related to the adaptation of facultative anaerobic bacteria like E. coli, K. pneumoniae, and S. enterica to anaerobic metabolism [69, 70]. Shotgun sequencing would be necessary to assess these predictions and elucidate the corresponding functional pathways.

Conclusions

In conclusion, these results show that mastitis is a dynamic process in which the udder microbiota constantly changes, highlighting the complexity of defining a unique microbiota profile indicative of mastitis. There are situations where a single microorganism causes severe dysbiosis by disrupting the endogenous udder population and resulting in clinical mastitis. However, intermediate states where the udder microbiota transitions between a state of health and disease are frequent. The re-establishment of the milk microbiome after an episode of clinical mastitis may be slower than the clinical recovery. In these cases, the clinical definition used to determine the udder health status does not always reflect the microbial profile since clinical symptoms do not always correlate directly with substantial alteration of the milk microbiota. Animals with subclinical mastitis harbour a milk microbiota similar to that of healthy animals and, therefore, pose a challenge when attempting to identify them solely through microbiome analysis. Further studies are needed to investigate the possible role of the many opportunistic microorganisms present in bovine milk in the development of mastitis.

Availability of data and materials

All FastQ sequence files generated from this study can be accessed in the publicly available National Center for Biotechnology Information (NCBI) repository, BioProject number PRJNA1128873, under Short Reads Archive (SRA) accession numbers SRR29605949 – SRR29606229.

Abbreviations

SCC:

Somatic Cell Counts

HTS:

High Throughput Sequencing

ONT:

Oxford Nanopore Technologies

HC:

Healthy cows

HH:

Healthy heifers

HR:

Cows recovered from a previous episode of mastitis

CM:

Clinical mastitis

SM:

Subclinical mastitis

I:

Initial (10–25 days in lactation) 

P:

Peak (50–85 days in lactation for cows and 65-100 days for heifers)

L:

Late (over 180 days in lactation)

CMT:

Californian Mastitis Test

TSS:

Total sum scaling

JSD:

Jensen Shannon Distance

LEfSe:

Linear discriminant analysis effect size

FDR:

False discovery rate

References

  1. Guo W, Liu S, Khan MZ, Wang J, Chen T, Alugongo GM, et al. Bovine milk microbiota: key players, origins, and potential contributions to early-life gut development. J Adv Res. 2024;59:49–64.

    Article  PubMed  Google Scholar 

  2. Liu J, Liu H, Cao G, Cui Y, Wang H, Chen X, et al. Microbiota characterization of the cow mammary gland microenvironment and its association with somatic cell count. Vet Sci. 2023;10:699.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  3. Hogeveen H, Huijps K, Lam T. Economic aspects of mastitis: new developments. N Z Vet J. 2011;59:16–23.

    Article  CAS  PubMed  Google Scholar 

  4. Makovec JA, Ruegg PL. Results of milk samples submitted for microbiological examination in Wisconsin from 1994 to 2001. J Dairy Sci. 2003;86:3466–72.

    Article  CAS  PubMed  Google Scholar 

  5. Kuehn JS, Gorden PJ, Munro D, Rong R, Dong Q, Plummer PJ, et al. Bacterial community profiling of milk samples as a means to understand culture-negative bovine clinical mastitis. PLoS ONE. 2013;8:e61959.

    Article  CAS  PubMed  Google Scholar 

  6. Shinozuka Y, Kawai K, Kurumisawa T, Shimizu Y, Imanishi T, Ohno A, et al. Examination of the microbiota of normal cow milk using MinION nanopore sequencing. J Vet Med Sci. 2021;83:1620–7.

    Article  PubMed  PubMed Central  Google Scholar 

  7. Alessandri G, Sangalli E, Facchi M, Fontana F, Mancabelli L, Donofrio G, et al. Metataxonomic analysis of milk microbiota in the bovine subclinical mastitis. FEMS Microbiol Ecol. 2023;99:12.

    Article  Google Scholar 

  8. Gryaznova MV, Syromyatnikov MY, Dvoretskaya YD, Solodskikh SA, Klimov NT, Mikhalev VI, et al. Microbiota of cow’s milk with udder pathologies. Microorganisms. 2021;9:1974.

    Article  CAS  PubMed  Google Scholar 

  9. Hoque MN, Istiaq A, Clement RA, Sultana M, Crandall KA, Siddiki AZ, et al. Metagenomic deep sequencing reveals association of microbiome signature with functional biases in bovine mastitis. Sci Rep. 2019;9:13536.

    Article  PubMed  Google Scholar 

  10. Salman MM, Nawaz M, Yaqub T, Mushtaq MH. Milk microbiota of Holstein Friesian cattle reared in Lahore: association with mastitis. Saudi J Biol Sci. 2024;31:103984.

    Article  CAS  PubMed  Google Scholar 

  11. Oikonomou G, Bicalho ML, Meira E, Rossi RE, Foditsch C, Machado VS, et al. Microbiota of cow’s milk; distinguishing healthy, sub-clinically and clinically diseased quarters. PLoS ONE. 2014;9:e85904.

    Article  PubMed  Google Scholar 

  12. Pang M, Xie X, Bao H, Sun L, He T, Zhao H, et al. Insights into the bovine milk microbiota in dairy farms with different incidence rates of subclinical mastitis. Front Microbiol. 2018;9:2379.

    Article  PubMed  Google Scholar 

  13. Winther AR, Narvhus JA, Smistad M, da Silva DV, Bombelli A, Porcellato D. Longitudinal dynamics of the bovine udder microbiota. Anim Microbiome. 2022;4:26.

    Article  PubMed  Google Scholar 

  14. Steinberg RS, Silva e Silva LC, de Souza MR, Reis RB, da Silva PCL, Lacorte GA, et al. Changes in bovine milk bacterial microbiome from healthy and subclinical mastitis affected animals of the Girolando, Gyr, Guzera, and Holstein breeds. Int Microbiol. 2022;25:803–15.

    Article  CAS  PubMed  Google Scholar 

  15. Tolle A. The microflora of the udder: factors influencing the bacteriological quality of raw milk. Bull Int Dairy Fed. 1980;120:4.

    Google Scholar 

  16. Lima SF, Teixeira AGV, Lima FS, Ganda EK, Higgins CH, Oikonomou G, et al. The bovine colostrum microbiome and its association with clinical mastitis. J Dairy Sci. 2017;100:3031–42.

    Article  CAS  PubMed  Google Scholar 

  17. Johnson JS, Spakowicz DJ, Hong B-Y, Petersen LM, Demkowicz P, Chen L, et al. Evaluation of 16S rRNA gene sequencing for species and strain-level microbiome analysis. Nat Commun. 2019;10:5029.

    Article  PubMed  Google Scholar 

  18. R Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2021. https://www.R-project.org/.

  19. Kandlikar GS, Gold ZJ, Cowen MC, Meyer RS, Freise AC, Kraft NJB, et al. ranacapa: an R package and Shiny web app to explore environmental DNA data with exploratory statistics and interactive visualizations. F1000Res. 2018;7:1734.

  20. McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE. 2013;8:e61217.

    Article  CAS  PubMed  Google Scholar 

  21. Lahti L, Shetty S. Tools for microbiome analysis in R. Package version 1.16.0. 2017. https://github.com/microbiome/microbiome?tab=readme-ov-file.

  22. Chong J, Liu P, Zhou G, Xia J. Using MicrobiomeAnalyst for comprehensive statistical, functional, and meta-analysis of microbiome data. Nat Protoc. 2020;15:799–821.

    Article  CAS  PubMed  Google Scholar 

  23. Louca S, Parfrey LW, Doebeli M. Decoupling function and taxonomy in the global ocean microbiome. Science. 2016;353:1272–7.

    Article  CAS  PubMed  Google Scholar 

  24. Shetty S, Lahti L. Microbiomeutilities: utilities for microbiome analytics. Package version 1.00.17. 2020. https://github.com/microsud/microbiomeutilities/.

  25. Kuhn M, Jackson S, Cimentada J. corrr: Correlations in R. Package version 2.1.6. 2020. https://CRAN.R-project.org/package=corrr. Accessed 24 Jun 2024.

  26. Wei T, Simko V. R package “corrgram”: visualization of a correlogram matrix (version 0.92). 2021. https://github.com/taiyun/corrplot. Accessed 24 Jun 2024.

  27. Nygaard AB, Tunsjø HS, Meisal R, Charnock C. A preliminary study on the potential of Nanopore MinION and Illumina MiSeq 16S rRNA gene sequencing to characterize building-dust microbiomes. Sci Rep. 2020;10:3209.

    Article  CAS  PubMed  Google Scholar 

  28. Ewing B, Green P. Base-calling of automated sequencer traces using Phred. II. Error probabilities. Genome Res. 1998;8:186–94.

    Article  CAS  PubMed  Google Scholar 

  29. Kerkhof LJ, Dillon KP, Häggblom MM, McGuinness LR. Profiling bacterial communities by MinION sequencing of ribosomal operons. Microbiome. 2017;5:116.

    Article  PubMed  Google Scholar 

  30. Catozzi C, Ceciliani F, Lecchi C, Talenti A, Vecchio D, De Carlo E, et al. Short communication: Milk microbiota profiling on water buffalo with full-length 16S rRNA using nanopore sequencing. J Dairy Sci. 2020;103:2693–700.

    Article  CAS  PubMed  Google Scholar 

  31. Falentin H, Rault L, Nicolas A, Bouchard DS, Lassalas J, Lamberton P, et al. Bovine teat microbiome analysis revealed reduced alpha diversity and significant changes in taxonomic profiles in quarters with a history of mastitis. Front Microbiol. 2016;7:480.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Arumugam M, Raes J, Pelletier E, Le PD, Yamada T, Mende DR, et al. Enterotypes of the human gut microbiome. Nature. 2011;473:174–80.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  33. Tröscher-Mußotter J, Saenz JS, Grindler S, Meyer J, Kononov SU, Mezger B, et al. Microbiome clusters disclose physiologic variances in dairy cows challenged by calving and lipopolysaccharides. mSystems. 2021;6:e00856-21.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Sarpong N, Seifert J, Bennewitz J, Rodehutscord M, Camarinha-Silva A. Microbial signatures and enterotype clusters in fattening pigs: implications for nitrogen utilization efficiency. Front Microbiol. 2024;15:1354537.

    Article  PubMed  PubMed Central  Google Scholar 

  35. Ozga AT, Gilby I, Nockerts RS, Wilson ML, Pusey A, Stone AC. Oral microbiome diversity in chimpanzees from Gombe National Park. Sci Rep. 2019;9:17354.

    Article  PubMed  PubMed Central  Google Scholar 

  36. Goulart DB, Mellata M. Escherichia coli mastitis in dairy cattle: etiology, diagnosis, and treatment challenges. Front Microbiol. 2022;13:928346.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Shpigel NY, Elazar S, Rosenshine I. Mammary pathogenic Escherichia coli. Curr Opin Microbiol. 2008;11:60–5.

    Article  CAS  PubMed  Google Scholar 

  38. Richards VP, Lefébure T, Bitar PDP, Dogan B, Simpson KW, Schukken YH, et al. Genome based phylogeny and comparative genomic analysis of intra-mammary pathogenic Escherichia coli. PLoS ONE. 2015;10:e0119799.

    Article  PubMed  PubMed Central  Google Scholar 

  39. Schukken Y, Chuff M, Moroni P, Gurjar A, Santisteban C, Welcome F, et al. The “other” gram-negative bacteria in mastitis: Klebsiella, Serratia, and more. Vet Clin North Am Food Anim Pract. 2012;28:239–56.

    Article  PubMed  Google Scholar 

  40. Wood JD, Chalmers GA, Fenton RA, Pritchard J, Schoonderwoerd M, Lichtenberger WL. Persistent shedding of Salmonella enteritidis from the udder of a cow. Can Vet J. 1991;32:738–41.

    CAS  PubMed  Google Scholar 

  41. Asefa I, Legabo E, Wolde T, Fesseha H. Study on Salmonella isolates from fresh milk of dairy cows in selected districts of Wolaita zone. Southern Ethiopia Int J Microbiol. 2023;2023:6837797.

    PubMed  Google Scholar 

  42. Kabelitz T, Aubry E, van Vorst K, Amon T, Fulde M. The role of Streptococcus spp. in bovine mastitis. Microorganisms. 2021;9:1497.

    Article  CAS  PubMed  Google Scholar 

  43. Peton V, Le Loir Y. Staphylococcus aureus in veterinary medicine. Infect Gene Evol. 2014;21:602–15.

    Article  Google Scholar 

  44. Hoekstra J, Zomer AL, Rutten VPMG, Benedictus L, Stegeman A, Spaninks MP, et al. Genomic analysis of European bovine Staphylococcus aureus from clinical versus subclinical mastitis. Sci Rep. 2020;10:18172.

    Article  CAS  PubMed  Google Scholar 

  45. De Vliegher S, Fox LK, Piepers S, McDougall S, Barkema HW. Invited review: mastitis in dairy heifers: nature of the disease, potential impact, prevention, and control. J Dairy Sci. 2012;95:1025–40.

    Article  PubMed  Google Scholar 

  46. Wald R, Hess C, Urbantke V, Wittek T, Baumgartner M. Characterization of Staphylococcus species isolated from bovine quarter milk samples. Animals. 2019;9:200.

    Article  PubMed  PubMed Central  Google Scholar 

  47. Catozzi C, Sanchez Bonastre A, Francino O, Lecchi C, De Carlo E, Vecchio D, et al. The microbiota of water buffalo milk during mastitis. PLoS ONE. 2017;12:e0184710.

    Article  PubMed  Google Scholar 

  48. Frazilio DA, de Almeida OGG, de Oliveira CAF, Lee SHI, Corassin CH, Alves VF, et al. Metataxonomics contributes to unravel the microbiota of a Brazilian dairy. J Dairy Res. 2020;87:360–3.

    Article  CAS  PubMed  Google Scholar 

  49. Wang Y, Nan X, Zhao Y, Wang H, Wang M, Jiang L, et al. Coupling 16S rDNA sequencing and untargeted mass spectrometry for milk microbial composition and metabolites from dairy cows with clinical and subclinical mastitis. J Agric Food Chem. 2020;68:8496–508.

    Article  CAS  PubMed  Google Scholar 

  50. Bay V, Griffiths B, Carter S, Evans NJ, Lenzi L, Bicalho RC, et al. 16S rRNA amplicon sequencing reveals a polymicrobial nature of complicated claw horn disruption lesions and interdigital phlegmon in dairy cattle. Sci Rep. 2018;8:15529.

    Article  CAS  PubMed  Google Scholar 

  51. Bay V, Gillespie A, Ganda E, Evans NJ, Carter SD, Lenzi L, et al. The bovine foot skin microbiota is associated with host genotype and the development of infectious digital dermatitis lesions. Microbiome. 2023;11:4.

    Article  CAS  PubMed  Google Scholar 

  52. Moroni P, Nydam D V., Ospina PA, Scillieri-Smith JC, Virkler PD, Watters RD, et al. Diseases of the teats and udder. Rebhun’s Diseases of Dairy Cattle: Third Edition. 2018;:389–465.

  53. Mtshali K, Khumalo ZTH, Kwenda S, Ismail A, Thekisoe OMM. Exploration and comparison of bacterial communities present in bovine faeces, milk and blood using 16S rRNA metagenomic sequencing. PLoS ONE. 2022;17:e0273799.

    Article  CAS  PubMed  Google Scholar 

  54. Taponen S, McGuinness D, Hiitiö H, Simojoki H, Zadoks RN, Pyörälä S. Bovine milk microbiome: a more complex issue than expected. Vet Res. 2019;50:44.

    Article  PubMed  Google Scholar 

  55. Braem G, De Vliegher S, Verbist B, Heyndrickx M, Leroy F, De Vuyst L. Culture-independent exploration of the teat apex microbiota of dairy cows reveals a wide bacterial species diversity. Vet Microbiol. 2012;157:383–90.

    Article  PubMed  Google Scholar 

  56. Krishnamoorthy P, Goudar AL, Suresh KP, Roy P. Global and countrywide prevalence of subclinical and clinical mastitis in dairy cattle and buffaloes by systematic review and meta-analysis. Res Vet Sci. 2021;136:561–86.

    Article  PubMed  Google Scholar 

  57. Alhussien MN, Dang AK. Milk somatic cells, factors influencing their release, future prospects, and practical utility in dairy animals: an overview. Vet World. 2018;11:562–77.

    Article  PubMed  Google Scholar 

  58. Derakhshani H, Fehr K, Sepehri S, Francoz D, De BJ, Barkema HW, et al. Invited review: microbiota of the bovine udder: contributing factors and potential implications for udder health and mastitis susceptibility. J Dairy Sci. 2018;101:10605–25.

    Article  CAS  PubMed  Google Scholar 

  59. Cheng WN, Han SG. Bovine mastitis: risk factors, therapeutic strategies, and alternative treatments—A review. Asian-Australas J Anim Sci. 2020;33:1699–713.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  60. Hogan J, Smith KL. Managing environmental mastitis. Vet Clin North Am Food Anim Pract. 2012;28:217–24.

    Article  PubMed  Google Scholar 

  61. Wente N, Grieger AS, Klocke D, Paduch JH, Zhang Y, Leimbach S, et al. Recurrent mastitis–persistent or new infections? Vet Microbiol. 2020;244:108682.

    Article  CAS  PubMed  Google Scholar 

  62. Günther J, Petzl W, Bauer I, Ponsuksili S, Zerbe H, Schuberth HJ, et al. Differentiating Staphylococcus aureus from Escherichia coli mastitis: S. aureus triggers unbalanced immune-dampening and host cell invasion immediately after udder infection. Sci Rep. 2017;7:4811.

  63. Ganda EK, Gaeta N, Sipka A, Pomeroy B, Oikonomou G, Schukken YH, et al. Normal milk microbiome is reestablished following experimental infection with Escherichia coli independent of intramammary antibiotic treatment with a third-generation cephalosporin in bovines. Microbiome. 2017;5:74.

    Article  PubMed  Google Scholar 

  64. Milanov D, Aleksić N, Todorović D, Bugarski D. Pasteurella multocida mastitis in cow. Vet Glas. 2017;71:117–22.

    Article  Google Scholar 

  65. Swartz T, Petersson-Wolfe CS. Pasteurella spp.: a practical summary for controlling mastitis. https://vtechworks.lib.vt.edu/bitstream/handle/10919/75563/DASC-67.pdf?sequence=1Stevens/. 2016.

  66. Turni C, Dayao D, Aduriz G, Cortabarria N, Tejero C, Ibabe JC, et al. A Pasteurella multocida strain affecting nulliparous heifers and calves in different ways. Vet Microbiol. 2016;195:17–21.

    Article  PubMed  Google Scholar 

  67. Ma T, Shen L, Wen Q, Lv R, Hou Q, Kwok LY, et al. Pacbio sequencing revealed variation in the microbiota diversity, species richness and composition between milk collected from healthy and mastitis cows. Microbiology (N Y). 2021;167:000968.

    CAS  Google Scholar 

  68. Eid RH, Aref NE, Ibrahim ES. Phenotypic diagnosis and genotypic identification of Bacillus cereus causing subclinical mastitis in cows. Vet World. 2023;16:888–94.

    Article  CAS  PubMed  Google Scholar 

  69. Simon J, Kroneck PMH. Microbial sulfite respiration. Adv Microb Physiol. 2013;62:45–117.

    Article  CAS  PubMed  Google Scholar 

  70. Schubert C, Unden G. Fumarate, a central electron acceptor for Enterobacteriaceae beyond fumarate respiration and energy conservation. Adv Microb Physiol. 2023;82:267–99.

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

We would like to thank the clinical veterinarians of ALBAIKIDE S.A., in particular Jorge Eseverri, for their participation in taking samples and collecting the associated data, and the farmers for their collaboration in this study.

Funding

This work was funded by MCIN/AEI/10.13039/501100011033 (project PID2019-106038RR-I00) and by the Basque Government. L.U-A. is the recipient of a predoctoral grant (PRE2020-096275) funded by MCIN/AEI/10.13039/501100011033 and ESF Investing in your future.

Author information

Authors and Affiliations

  1. Animal Health Department, NEIKER – Basque Institute for Agricultural Research and Development, Basque Research and Technology Alliance (BRTA), Bizkaia Science and Technology Park 812L, 48160, Derio, Bizkaia, Spain

    Leire Urrutia-Angulo, Medelin Ocejo, Beatriz Oporto, Gorka Aduriz & Ana Hurtado

  2. Applied Mathematics Department, NEIKER – Basque Institute for Agricultural Research and Development, Basque Research and Technology Alliance (BRTA), Bizkaia Science and Technology Park 812L, 48160, Derio, Bizkaia, Spain

    José Luís Lavín

Authors
  1. Leire Urrutia-Angulo
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Medelin Ocejo
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Beatriz Oporto
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Gorka Aduriz
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. José Luís Lavín
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Ana Hurtado
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

A.H. and G.A. conceived the study. A.H. coordinated the study. L.U.-A. and B.O. performed laboratory analyses. L.U.-A., M.O., and J.L.L performed bioinformatic analyses. L.U.-A. and M.O., carried out statistical data analyses. L.U.-A. and M.O. prepared the figures. A.H., L.U.-A. and M.O. interpreted the data and wrote the manuscript. B.O., G.A., and J.L.L. contributed to manuscript revision. All authors read and approved the final version of the manuscript.

Corresponding author

Correspondence to Ana Hurtado.

Ethics declarations

Ethics approval and consent to participate

Sample collection was carried out by the farm veterinary practitioners strictly following Spanish ethical guidelines and animal welfare regulations (Real Decreto 53/2013). Ethical review and approval was waived for this study, as all examinations and treatments of the animals occurred as professional activities of the farm veterinarian in the framework of his veterinary services and medical care. Written consent was obtained from the farm owners before the start of the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Urrutia-Angulo, L., Ocejo, M., Oporto, B. et al. Unravelling the complexity of bovine milk microbiome: insights into mastitis through enterotyping using full-length 16S-metabarcoding. anim microbiome 6, 58 (2024). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s42523-024-00345-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s42523-024-00345-0

Keywords

Animal Microbiome

ISSN: 2524-4671