The Rumen Microbiota Contributes to the Development of Mastitis in Dairy Cows

To exclude the effect of experimental period on milk composition, we detected the milk composition of eight other Holstein Frisian cows with the same standards used to establish the SARA model (cows with similar lactation durability and body weight). The results showed that there were no significant changes in milk yield, milk fat, milk protein, lactose, urea nitrogen, or SCC of healthy cows during the trial period (see Supplementary Figure S1). To evaluate the HCD-induced SARA model in cows, the dry matter intake (DMI) and milk yield were measured weekly. After 8 weeks of feeding with HCD, the DMI and milk production were significantly reduced compared to the control groups (Fig. 1A and B). In addition, the pH of the rumen fluid was noticeably reduced, and a pH < 5.8 was sustained for more than 3 h at different periods of time in the cows fed an HCD for 8 weeks (Table S1 in the supplemental material), indicating that SARA model was efficiently established (21). In addition, the levels of pH in feces were also significantly lowered. However, there were no changes in the blood pH of SARA cows compared to the control cows (Table S1).

Milk composition analysis showed that the contents of milk fat, milk protein, fat/protein ratio and dry matter were markedly decreased in SARA cows compared to the control cows (Fig. 2A, C–E). The content of lactose in milk from SARA cows was unchanged (Fig. 2D), and urea nitrogen was significantly increased compared to the milk from control cows (Fig. 2F). Importantly, SCC and serum amyloid A (SAA), two important indicators of mastitis, also significantly increased after cows suffered from SARA (Fig. 2G and H). In addition, the levels of proinflammatory cytokines, including tumor necros factor-α (TNF)-α, interleukin-1β (IL-1β), and interleukin-6 (IL-6), were clearly increased in the mammary gland (Fig. 2I–K) and in milk (Fig. 2L–N) from cows suffering from SARA. Furthermore, histopathological analysis of SARA cows showed a large amount of inflammatory cell infiltration, thickening of the alveolar wall, mammary gland destruction in the mammary gland and a higher inflammation score (Fig. 2O and P).

Neutrophil translocation from blood to milk must occur through the blood-milk barrier, which is a specific structure that prevents foreign matter from entering the mammary gland from the blood or external environment (22). To evaluate the permeability of the blood-milk barrier, we detected the expression of tight junction proteins that make up the blood-milk barrier. The results showed that the expression of Claudin-1, Claudin-3, Occludin, and Zonula occludens-1 (ZO-1) was significantly reduced in SARA cows (Fig. 3A–D). In addition, blood-derived proteins in milk, including IgG and lactate dehydrogenase (LDH), serve as indicators of blood-milk barrier integrity (23). Moreover, the production of IgG and LDH was clearly increased in both the blood and milk of SARA cows compared to the control cows (Fig. 3E–H). These results suggested that SARA impairs the structure and function of the blood-milk barrier in dairy cows.

LPS, the main component of the cell walls of Gram-negative bacteria, is an important inflammatory substance leading to mastitis. Evidence has increasingly shown that LPS -derived from the rumen of SARA cows can translocate to the bloodstream. During the lactation period, the mammary gland blood volume of cows accounts for 8% of the total body blood volume. Therefore, we assessed whether cow mastitis induced by SARA was associated with LPS migration from the rumen to the mammary gland via the bloodstream. We found that the levels of LPS in the mammary gland, milk, lacteal veins, tail veins, and rumen fluid were all significantly increased in SARA cows (Fig. 4A–E). LPS is a strong activator of innate immune responses and is recognized by toll like receptor 4 (TLR4), which then activates nuclear factor-κB (NF-κB) to induce the transcription of TNF-α, IL-1β, and IL-6. The expression of TLR4, phosphorylated NF-κB p65, and phosphorylated inhibitor of NFκB was upregulated in the mammary glands of SARA cows compared with control cows (Fig. 4F). These results suggest that SARA of cows leads to rumen-derived LPS entering the mammary gland through blood circulation, damaging the blood–milk barrier, and inducing inflammation of the mammary gland in cows.

The liver is an important metabolic and immune organ of ruminants, playing a vital role in transferring LPS to the bloodstream (24). In contrast, overproduction of LPS can also promote the occurrence of inflammatory responses and the aggregation of immune cells, thus interfering with the metabolism of substances in the liver (25). SARA cows showed severe pathological damage, including inflammatory cell infiltration and liver cell injury ballooning (Fig. 5A and B), and upregulated levels of AST (Fig. 5C). Plasma levels of ALB, GLOB, TP, and GLU were also assessed as indicators of the metabolism and internal organ status of animals (26). The data showed higher levels of ALB and lower levels of GLU, but no changes of GLOB and TP in plasma were observed in SARA cows (Fig. 5E–H). In addition, LPS entering the blood from the rumen needs to cross the rumen wall and/or intestinal tract; thus, rumen and gut epithelial permeability was assessed in cows from the control and SARA groups. The histological changes suggested that the rumen epithelium of SARA cows was incomplete, and greater numbers of immune cells infiltrated into the rumen epithelium of SARA cows (Fig. 5I and J). Furthermore, the detection of tight junction proteins of the rumen barrier indicated that the protein expression levels of Claudin-1, Claudin-3, Occludin, and ZO-1 were all significantly reduced in SARA cows compared to control cows (Fig. 5M–P). Moreover, histopathologic changes in intestinal tissues indicated gut epithelium desquamation and severe cellular damage in SARA cows compared to control cows, and SARA cows showed significantly higher epithelial damage scores in the intestine (Fig. 5K and L). In addition, the expression of Claudin-1, Claudin-3, Occludin, and ZO-1 in the intestinal barrier was significantly reduced in SARA cows compared to control cows (Fig. 5Q to T). These results suggested that SARA cows have increased permeability of the rumen and intestinal barrier, inducing the release of rumen-derived LPS into the bloodstream, and subsequent liver damage.

Milk, rumen fluid, and fecal samples from the same 8 Holstein Frisian cows on day 0 and week 8 after feeding an HCD were used to detect the microbiota community using PCR amplification of the V4 region of the bacterial 16S rRNA gene from all 48 samples. The 16S rRNA sequencing resulted in 3,456,642 Taxon Tags, with an average of 72,013 Taxon Tags per sample. For operational taxonomic units (OTUs) at a 3% distance, 9,495 different phylotypes were detected among all of the samples. Rarefaction curves for 48 samples indicated that the sampling depth was sufficient to characterize the majority of bacterial diversity (see Supplementary Figure S2 in the supplemental material). To explore whether mastitis induced by SARA was associated with changes in the microbiota, the bacterial richness and diversity in rumen fluid, feces, and milk were compared between control and SARA cows. The results showed that the estimators of community richness (Chao 1 and ACE) and diversity (Shannon index and Simpson) of milk, rumen fluid, and feces in SARA cows were significantly reduced compared to the control cows (Fig. 6A and B; see Supplementary Figure S3). In addition, a principal coordinate analysis (PCoA) plot, based on the weighted- UniFrac distance matrices, indicated that the microbiota composition of rumen fluid, feces, and milk from control cows was clearly separated from that of SARA cows (Fig. 6C).

The top 10 phyla and the top 30 genera in relative abundance were used to analyze the bacterial community structure of rumen fluid, milk, and feces of cows. At the phylum level, Proteobacteria, Firmicutes, Bacteroidetes, and Tenericutes were the four dominant phyla in all samples by relative abundance (Fig. 7A). t Test analysis showed that Proteobacteria was significantly increased in the rumen fluid microbiota, whereas Firmicutes, Tenericutes and other low abundance phyla, including Spirochaetes, Gracilibacteria, Euryarchaeota, Fibrobacteres, and Kiritimatiellaeota were significantly reduced in SARA cows compared to control cows (see Supplementary Figure S4A in the supplemental material). The changes in the milk microbiota were similar to the changes in the rumen microbiota, as shown by the significant increase in Proteobacteria, while other phyla, including Firmicutes, Bacteroidetes, Tenericutes, and Actinobacteria, were significantly reduced in SARA cows compared to the controls (see Supplementary Figure S4B). In feces, the relative abundance of Proteobacteria was significantly increased, while the relative abundance of Melainabacteria was significantly reduced in SARA cows compared to control cows (see Supplementary Figure S4C).

In addition, the sequence results showed that Stenotrophomonas, Bacteroides, Sphingononas, unidentified Ruminococcaceae, and Succinivibrio were the most prominent genera in all samples (Fig. 7B). Significant differences in the abundances of milk, rumen fluid, and feces genera between control and SARA cows were further assessed by t test analysis. The results showed that Stenotrophomonas and Succinivibrio were markedly increased, while the other 10 genera were significantly reduced in rumen from SARA cows compared to those from control cows (see Supplementary Figure S4D in the supplemental material). The changes in the milk microbiota were also similar to the changes in the rumen microbiota, as shown by the relatively increased abundances of Stenotrophomonas, Sphingomonas and Brevnudimonas, while the other detected 14 genera were notably reduced in milk from SARA cows compared with those from control cows (see Supplementary Figure S4E). The microbiota in feces indicated that Stenotrophomonas, Succinivibrio, unidentified Prevatellaceae, unidentified Erysipelotrichaceae, Unidentified Clostridiales, Roseburia, Acetitomaculum, Oscillibacter, Anaeroplasma, and Marvinbryantia were significantly increased, and the other five genera were reduced in the SARA group (see Supplementary Figure S4F). These results showed that the changes in the microbiota of milk were similar to those in the rumen fluid samples after SARA. This indicated that there may be a certain correlation between the milk microbiota and rumen microbiota of dairy cows.

To identify the specific bacteria in the rumen associated with mammary gland inflammation, we performed a biomarker analysis using linear discriminant analysis (LDA = 4.0) effect size (LEfSe) and a cladogram generated from LEfSe analysis on the microbiota community of milk, rumen fluid, and feces. At the genus level, only Stenotrophomonas and Sphingomonas were enriched in milk from SARA cows compared to the control cows (Fig. 8A and B). Additionally, the relative abundance of Stenotrophomonas in rumen fluid and feces was significantly increased in SARA cows compared to control cows (Fig. 8C–F). These results suggested that a large amount of Stenotrophomonas derived from the rumen may be associated with the development of mastitis.

To explore the relationship of elevated Stenotrophomonas in the rumen of SARA cows and the incidence of mastitis, we detected the effect of S. maltophilia (the only species of Stenotrophomonas) on mastitis in mice. The results showed that oral infection of S. maltophilia resulted in damage to mammary gland tissues (Fig. 9A and B), and the production of the proinflammatory cytokines TNF-α, IL-1β, and IL-6 was significantly increased in the mammary glands (Fig. 9C–E). In addition, to detect whether S. maltophilia on the mammary gland surface can induce mastitis, the feces from mice that were gavaged with S. maltophilia were collected and smeared onto the surface of healthy mice. H&E staining and ELISA analysis showed that no inflammation was present in the mammary glands of the mice smeared with feces from the mice that were gavaged with S. maltophilia for the duration of the experiment (Fig. 9A–E). These results suggested that gut-derived S. maltophilia is an important agent to induce the development of mastitis.

Sixteen late-lactating Holstein Frisian cows (average body weight, 547 ± 51 kg, lactation days and similar weight) were used in this experiment. For 15 days before the start of the experiment, eight cows were offered free access to a diet containing a forage-to-concentrate ratio (F:C) of 40:60 to ensure adaptation to the diet. After 8 weeks, eight cows received a high-concentrate diet (HCD) comprising 30% forage and 70% mixed concentrate. Rumen fluid, milk, feces, and blood were harvested before feeding a HCD and were used as a control group for comparison following treatment with HCD cows. Throughout the experiment, the cows were fed daily at 5:00 and 18:00, and they had free access to water during the experimental period. The full proposal was reviewed by the Institutional Animal Care and Use Committee of the Jilin University ethics committee, which approved the animal care and use permit license. Experimental protocols for obtaining bovine clinical samples used in this study were carried out in strict accordance with the Animal Ethics Procedures and Guidelines of China. The sample collection work was approved with signed informed consent from the cattle farm owner.

Rumen fluid from each sample was collected by rumenocentesis using nonpyrogenic needles (2.0 × 120 mm) and 50 mL syringes at the beginning and after 8 weeks of feeding a HCD. The samples were centrifuged at 10,000 × g for 45 min at 4°C, and the supernatant was aspirated gently and passed through a disposable 0.22-μm LPS-free filter. The filtrates were transferred into sterile, depyrogenated glass tubes (Chinese Horseshoe Crab Reagent Manufactory Co., Ltd., Xiamen, China) and partly stored at −20°C for LPS detection. The others were frozen in liquid nitrogen and kept at −80°C for 16S rRNA rDNA gene amplicon pyrosequencing. Blood was collected from the jugular vein and lacteal vein at 0 and 8 weeks after feeding a HCD. Each sample was separated into two portions. One portion was collected with vacuum tubes containing heparin lithium and centrifuged at 3000 × g at 4°C for 15 min, and the plasma was harvested to detect the levels of ALT, AST, ALB, TP, and GLU using a biochemical analyzer (Vet Test 8008, IDEXX, USA). The second portion was collected with vacutainer tubes with EDTA. The plasma was separated from the blood by centrifugation at 3000 × g at 4°C for 15 min; transferred into a sterile, depyrogenated glass tube, and then kept at −20°C for LPS detection. Fecal samples were collected at the same time as blood and divided into two portions. One portion was immediately distilled with sterile PBS (1:5, w:v) to detect pH using a pH meter, and the second portion was collected and kept at −80°C for 16S rRNA gene amplicon pyrosequencing. The teats were disinfected using cotton wool dipped in 70% ethanol, and the first three handfuls of milk were discarded. Then, milk samples were collected from four quarters of each cow, mixed and divided into three portions from control and SARA cows. One portion was collected and transferred into sterile 50 mL vials with potassium dichromate. The samples were used to detect the milk composition (MilkoscanTM FT1, FOSS, Denmark) and SCC (Fossomatic 5000, FOSS). The second portion of the milk sample was collected and centrifuged at 14000 × g for 30 min at 4°C. The supernatants were transferred into a sterile, depyrogenated glass tube and kept at −20°C for LPS detection. The third portion was collected and kept at −80°C for microbiota analysis. Liver, rumen, and intestine tissues were collected from healthy and SARA animals after euthanasia. All samples were divided into two portions. One portion was fixed in 4% formaldehyde (Solarbio, Shanghai, China) for H&E staining (Solarbio, Shanghai, China), and another portion was kept at −80°C until use. The mammary gland tissues were collected and divided into three portions. The specimens were stained with H&E, kept at −80°C, and subjected to LPS detection. For LPS detection, 1 g of mammary gland tissues was weighed and homogenized with 0.5 mL of sterile PBS on ice, and the samples were centrifuged at 2000 r for 40 min at 4°C. The supernatant was collected and used for LPS detection.

The levels of LPS from samples of rumen fluid, plasma, milk, and mammary glands were detected by a chromogenic endpoint assay (Chinese Horseshoe Crab Reagent Manufactory Co., Ltd., Xiamen, China) with a minimum detection limit of 0.1 EU/mL (rumen fluid and feces) or 0.01 EU/mL (plasma, milk, and mammary gland) according to the manufacturer’s instructions.

Mammary gland tissues from control and SARA cows were weighed and homogenized with PBS (1:10, wt/vol) and centrifuged at 2000 × g for 40 min at 4°C. The lipids on the surface were removed, and the supernatants were collected. Milk was centrifuged at 1000 × g for 20 min at 4°C, and the supernatants were collected. Centrifugation was repeated, and the supernatant was kept at −20°C until use. The levels of TNF-α, IL-1β, IL-6, LDH, and IgG in plasma and milk were tested by ELISA kits (Lanpai BIO, Shanghai, China) according to the manufacturer’s instructions. SAA in the milk and plasma was detected using a biochemical analyzer (Diano, China).

The total proteins from the rumen, mammary gland, and intestine tissues were extracted with tissue protein extracted reagent (Thermo Scientific, MA, USA), and the concentration of proteins was detected using a BCA protein assay kit (Thermo Scientific). The proteins were fractionated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto PVDF membranes (Bio Trace, Pall Co., USA). The membranes were blocked with 3% BSA at room temperature at shaking for 2 h, and then incubated with anti-IκBα, anti-phospho-NF-κB p65, anti-NF-κB p65, and anti-TLR4 at 4°C overnight. Following tris-buffered saline with tween-20 (TBS-T) washing, all membranes were incubated with secondary antibody for 1 h. After washing with TBS-T. The targeted proteins were detected by Supersignal West Pico Chemiluminescent Substrate (Thermo Scientific), and the bands were imaged by a Protein Simple imager (ProteinSimple, Santa Clara CA, USA).

The tissues of the mammary gland, liver, rumen, and intestine of cows were histopathologically analyzed. Pathological injury scoring of liver tissue was conducted using a system that included three categories: glycogenated nuclei (graded 0–1, from none or rare to many continuous patches), liver cell ballooning injury (graded 0–2, from absent to severe ballooning injury), and inflammatory cell filtrate (graded 0–3, from none or rare to transmural) (38, 50). Rumen pathological injury was scored using a scoring system that included three categories: severity of epithelial injury (graded 0–3, from absent to mild) and the extent of inflammatory cell infiltrate (graded 0–3, from none or rare to transmural) (51). Each mammary gland index was evaluated through three categories, including hyperemia/edema (graded on 0 to 3, from normal to severe), milk stasis/acinar necrosis (graded on 0 to 3, from normal to severe), and infiltration with neutrophils (graded on 0 to 5, none or rate to transmural) (52). The intestinal injury score includes the severity of epithelial damage (scored 0–3, from none or rare to severe), the extent of inflammatory cell infiltrate (scored 0–3, from none or rare to transmural), and goblet cell depletion (sored 0–1) (53, 54). Three tissue sections from each sample were used to assess the numerical score.

Total RNA of 0.1 g mammary gland, rumen, and intestine was extracted using TRLzol (Invitrogen, USA) according to the manufacturer’s instructions. RNA concentrations and quality were analyzed by an RNA/DNA calculator (Cambridge, UK). The total RNA of each sample was reverse transcribed into cDNA using random primers from a TransScript One-Step gDNA Removal kit and cDNA Synthesis SuperMix (TransGen Biotech, China) according to the manufacturer’s instructions. Quantification of relative mRNA concentration was conducted using a SYBR green Plus reagent kit (Roche, Swiss) and a 7500 Fast real-time PCR system (Applied Biosystems, USA). The relative expression of each gene was normalized to GAPDH, and the primers for each gene was shown in online Table S2.

Total genome DNA from rumen fluid, milk, feces, and blood was extracted by a CTAB/SDS method. The DNA concentration and purity of each sample were detected by 1% agarose gels and the concentration of DNA was diluted to 1 ng/μL by sterile water. The 16S rRNA was amplified by specific primer (16S V4:515F-806R) with the barcode targeting the V4 region. All PCRs were carried out with Phusion High-Fidelity PCR Master Mix (New England Biolabs). PCR products were mixed with an equal volume of 1×loading buffer (containing SYB green) and subjected to electrophoresis on a 2% agarose gel for detection. PCR products were mixed in equidensity ratios, and the products were purified with Gene JET Gel Extraction Kit (Thermo Scientific). Sequencing libraries were generated using Iso Plus Fragment Library Kit 48 rxns (Thermo Scientific) following manufacturer’s instructions. The library quality was evaluated by a Qubit@ 2.0 Fluorometer (Thermo Scientific). Finally, the library was sequence on an Ion S5 XL platform and 400 bp/600 bp single-end reads were generated.

Quality filtering of the raw reads was performed under specific filtering conditions to obtain high-quality clean reads according to the Cutadapt (V1.9.1) quality control process. The reads were compared with the reference database using the UCHIME algorithm (UCHIME Algorithm) to detect chimeric sequences, and the chimeric sequences were removed to obtain clean reads. Sequence analysis was performed by Uparse software (Uparse v7.0.1001). Sequences with ≥97% similarity were assigned to the same OTUs. Measures of alpha diversity, including Chao1, ACE, Shannon, and Simpson, were used to evaluate the complexity of species richness and diversity for each sample. All of the indices were calculated with QIIME. In addition, beta diversity and weighted and unweighted UniFrac analyses were selected to evaluate differences in samples in species complexity and were calculated using QIIME software. Principal coordinate analysis (PCoA) of weighted UniFrac analysis was used to evaluate the distance among the rumen fluid, feces, and milk in control and SARA cows. LEfSe was conducted to identify bacterial taxa differentially represented between rumen fluid, feces, and milk microbiota from SARA and control cows. Venn diagrams were created to analyze the number of core genera in milk, rumen fluid, and feces in SARA and control cows using the principles of bioinformatics and evolutionary genomics.

In the study of the effect of Stenotrophomonas on mastitis, Stenotrophomonas maltophilia (S. maltophilia), the only species in the genus of Stenotrophomonas, were administered to mice (1 × 10⁸ CFU/200 μL PBS) after delivery by oral gavage for 1 d, 3 d, 5 d, and 7 d consecutive days (once a day). In addition, 1 g fecal samples from each of the S. maltophilia was mixed together and then suspended in sterile saline solution. After thorough mixing and resting (to minimize the number of bacteria lost), 2 mL of the supernatant was collected and gently smeared on the mammary gland surface by swabs for 7 consecutive days (14). The control group mice were smeared with the same volume of normal saline. Each group contained 8 mice after delivery. Then, the mice were sacrificed, and mammary glands were collected and stored at −80°C until use.

Statistical analysis was conducted using GraphPad Prism 6.01 (GraphPad Software, Inc., San Diego, CA). All data are expressed as the means ± SEM. Differences between date means were determined using one-way ANOVA (Dunnett’s t test) and the two-tailed t test. A P < 0.05 was considered to be statistically significant.

16S rRNA raw data from the study were deposited in the NCBI Sequence Read Archive (SRA) under accession no. PRJNA786003↗.