Eight Chinese Holstein cows intended for slaughter were provided by the Heilongjiang Yuda Animal Husbandry, China. Fifty-milliliters milk samples were collected from the four milking areas of each cow into sterile centrifuge tubes, and the animals were then euthanized. The udders were disinfected thrice with 75% alcohol, and the skin was peeled using pre-cooled sterile surgical scissors. Pre-cooled sterile scalpels were then used to cut 40 pieces of mammary tissues measuring 0.3 × 0.3 × 0.5 cm. After rinsing thrice with PBS, 10 tissue pieces per animal were immersed in 4% paraformaldehyde, and the remaining pieces were snap frozen in liquid nitrogen and stored at −80°C. The fixed tissues were dehydrated, clarified, embedded in paraffin, and stained with hematoxylin and eosin as per standard protocols. The stained mammary gland sections were then observed for the morphological changes associated with mastitis.

The somatic cell counts (SCC) of the 32 milk samples were determined using a cell counter (Thermo Fisher Scientific, USA). Briefly, SCC of 100,000/mL indicated healthy udder, 200,000–500,000/mL indicated subclinical mastitis, and >1,000,000/mL indicated clinical mastitis (19, 20). The milk samples were then inoculated into LB broth and cultured overnight at 37°C. Ten microliters of each suspension was plated on LB agar and cultured at 37°C for 24 h. The individual colonies were picked and subjected to Gram staining to distinguish between gram positive/negative cocci or bacilli. The colonies initially identified as cocci were further sub-cultured into LB broth and blood agar, and finally differentiated as streptococci or S. aureus on the basis of their colony morphology, and the catalase and coagulase tests. The colonies identified as S. aureus were inoculated into LB broth and incubated overnight at 37°C.

Total RNA was extracted from 50 mg frozen udder tissue per animal using the 5 min rapid RNA extraction kit (HiGene, China) as per the manufacturer's instructions. The miRNA libraries were constructed using the small RNA sample premix kit (Illumina, USA), and sequenced using HiSeq2500. The miRNA profiles of three healthy cows (accession numbers are SRR8185411, SRR8185412, and SRR8185413) and three mastitic cows (accession numbers are SRR8185414, SRR8185415, and SRR8185416) were deposited in the SRA database of NCBI. The known and novel miRNAs were identified using the Bowtie (http://bowtie-bio.sourceforge.net/index.shtml) and miRDeep2 (https://www.osc.edu/book/export/html/4389) programs. EBseq (21) was then used to screen for the DEMs using corrected p-value (Benjamini-Hochberg method) < 0.05, false discovery rate (FDR) ≤ 0.05 and | log₂ ^(FC) | ≥ 1 as the screening criteria. The target genes of the DEMs were predicted using TargetScan (http://www.targetscan.org/vert_72) and miRWalk (http://mirwalk.umm.uni-heidelberg.de/), and functionally annotated to GO terms and KEGG pathways using DAVID (https://david.ncifcrf.gov/).

Mac-T cells (Laboratory of Animal Stress Regulation Mechanism of Heilongjiang Bayi Agricultural University) were stimulated with 1, 10, 20, and 40 μg LTA for 3, 6, 12, and 24 h to induce inflammation. The bta-miR-223 mimic (5′-UGUCAGUUUGUCAAAUACCCCA-3′), mimic negative control (NC; 5′-CAGUACUUUUGUGUAGUACAA-3′), inhibitor (5′-UGGGGUAUUUGACAAACUGACA-3′), and inhibitor NC (5′-CAGUACUUUUGUGUAGUACAA-3′) sequences were purchased from Shanghai Sangon Biotech, China. The cells were transfected with 30 pmol/ml of the above oligonucleotides using RFect transfection reagent (Changzhou Biogenerating Biotechnologies, China), and the fluorescence intensity was checked at 0, 12, 24, 36, and 48 h post-transfection. In addition, the Mac-T cells were also transfected with The 4 μg/mL CBLB overexpression and interference vectors (Hedgehogbio Biotechnology, China) using Lipofectamine 3000 (Thermo Fisher Scientific, USA), and the fluorescence intensity was checked after 0, 24, 48, and 72 h of transfection. The expression/silencing efficiency was assessed by RT-PCR and Western blotting.

After 1-week of adaptive feeding, 108 8-weeks-old BALB/c mice (81 females and 27 males; Changsheng Biotechnology, China) were set up for mating with three females and one male per cage. The pregnant mice were divided into the PBS, and the 6, 12, 24, and 48 h S. aureus-infected groups (n = 6 each). The mice were separated from their pups 7 days postpartum and weaned for 6 h. The standard S. aureus CMCC(B)26003 strain (Biobw Biotechnology, China) suspension was prepared by inoculating 5 μL of the glycerol stock in 200 ml LB broth and culturing overnight at 37°C till the bacterial load was 10⁹CFU/mL. After anesthetizing the mice with ether, the fourth pair of nipples were disinfected with 75% alcohol and gently lifted with sterilized surgical forceps. Around 100 μL S. aureus suspension or PBS was slowly injected into the mammary gland along the milk duct. The mice were put back into the cages and euthanized after 6, 12, 24, and 48 h. The tissue under the fourth nipple pair was dissected, and mammary glands measuring ~0.2 × 0.2 × 0.2 cm were fixed and embedded in paraffin for routine histological staining. In addition, 20 mg tissue per mouse was snap frozen in liquid nitrogen for RNA and protein extraction.

The miRNAs were reverse transcribed using a first-strand cDNA synthesis (stem-loop method) kit (Sangon, China) and specific primers (Table S1). RT-qPCR was performed using TB green® premix Ex Taq™ (TAKARA, Japan) and the primer sequences are listed in Table S2. The primer sequences specific for bovine TNFA, IL1B, and IL6, bovine CBLB, sp1 transcription factor (SP1), conserved helix-loop-helix ubiquitous kinase (CHUK) and interleukin 6 signal transducer (IL6ST), and murine TNFA, IL6, and IL1B are listed in Tables S3–S5, respectively.

Total protein was extracted from the suitably treated cells and murine tissues using the RIPA lysis buffer. Western blotting was performed as per standard protocols using antibodies against CBLB, PI3K, AKT, NF-κB, and p-NF-κB p65 (Proteintech, USA).

HEK-293T cells (Procell Life Technology, China) were co-transfected with 4 μg/ml wild-type or mutant CBLB-3'-UTR along with bta-miR-223 mimic or mimic NC using Roche X-tremeGENE HP DNA transfection reagent (Promega, USA). The dual-glo luciferase assay system kit (Promega, USA) was used to detect the fluorescence activity.

The relative miRNA and mRNA expression levels were measured using the 2^−ΔΔCt method. Image J software was used to analyze the gray value of protein bands. One-way ANOVA analysis was performed using SPSS 19.0, and the data was shown as mean ± standard deviation. P < 0.05 was considered statistically significant.

Based on the SCCs, the cows No. 1, No. 4, and No. 8 were initially identified as healthy, and No. 2, No. 6, and No. 7 as mastitic (Table S6). No pathogenic bacteria were isolated from the milk samples of the healthy cows, whereas that from mastitic cows were positive for common mastitis-causing pathogens like Escherichia coli, Streptococcus, and S. aureus (Figure 1). Furthermore, the mammary glands of healthy cows had closely arranged epithelial cells, and did show any inflammatory cell infiltration in the acinar cavity. However, the mastitic mammary glands showed numerous inflammatory cells and epithelial loss in the acinus cavity (Figure 2). Taken together, No. 1, No. 4, and No. 8 were verified as healthy, and No. 2, No. 6, and No. 7 as mastitic cows.

A total of 1,245 miRNAs were identified in the udder tissues of six cows, of which 793 are known (Table S7) and 452 are novel (Table S8). In addition, 465 miRNAs were differentially expressed in the mastitic vs. healthy mammary glands, including 231 up-regulated (143 known and 88 novel) and 234 down-regulated (185 known and 49 novel) miRNAs (Table S9). With |log2(FC)| ≥ 1 and FDR ≤ 0.01 as the thresholds, 48 DEMs were statistically significant, including 10 up-regulated and 19 down-regulated known miRNAs (Table 1), and 19 up-regulated novel-miRNAs (Table S10). The DEMs potentially involved in mastitis resistance were screened by functionally annotating their predicted target genes. Inflammation-related pathways were significantly enriched among the target genes (Figure 3A), including pathways in cancer, MAPK signaling pathway, neuroactive ligand-receptor interaction etc. (Table S11). Accordingly, bta-miR-223, bta-miR-205, and bta-miR-21-5p were identified as mastitis resistance-related miRNAs, and also validated by RT-qPCR (Figure 3B).

To further determine the role of bta-miR-223 in mastitis, the LTA-stimulated Mac-T cells were transfected with the mimics or inhibitors, and bta-miR-223 expression was successfully enhanced or silenced (Figure 4). The levels of TNFA, IL6, and IL1B mRNAs in Mac-T cells were unaffected after 3–6 h of LTA exposure, and while TNFA and IL6 were significantly upregulated following 12 h treatment with 40 μg/ml LTA, IL1B remained unchanged. However, a 24 h LTA stimulation led to a marked increase in the levels of all three (Figure 5), which was decreased significantly by the bta-miR-223 mimic (Figure 6). Thus, overexpression of bta-miR-223 can mitigate the inflammatory response induced by S. aureus-derived LTA in vitro, indicating that it likely plays a role in mastitis resistance.

According to bioinformatic analyses, CHUK, SP1, CBLB, and IL6ST were predicted as the target genes of bta-miR-223. Consistent with this, bta-miR-223 overexpression significantly downregulated SP1 and CBLB mRNA in the Mac-T cells whereas bta-miR-223 silencing had the opposite effect. Furthermore, the effect of bta-miR-223 was stronger on CBLB compared to SP1 mRNA. Since CBLB protein was also significantly down-regulated by bta-miR-223 mimic (Figure 7), we next determined whether the latter also bound to the CBLB promoter through the dual luciferase assay. Briefly, HEK-293T cells were transfected with bta-miR-223 mimic or control with luciferase reporter gene under the wild-type or mutant CBLB-3′-UTR, and the relative luciferase activity was measured. As shown in Figure 8, bta-miR-223 mimic sharply decreased the relative fluorescence intensity of wild type but not of mutant CBLB-3′-UTR, indicating that bta-miR-223 directly binds to and represses the activity of CBLB promoter. The role of CBLB in LTA-stimulated inflammation was likewise determined by transfecting the Mac-T cells with the knockdown and overexpression vectors (Figure 9). CBLB overexpression in the LTA-stimulated Mac-T cells significantly up-regulated PI3K, AKT, and p-NF-κB p65 (Figure 10), while CBLB silencing had an inhibitory effect on their expression levels (Figure 11). Taken together, CBLB drives the S. aureus LTA-stimulated inflammatory damage in Mac-T cells via the PI3K/AKT/NF-κB pathway, and is targeted by bta-miR-223.

To further explore the above hypothesis, we established a mouse model of S. aureus-induced mastitis by injecting a high bacterial titer directly into the mammary glands of postpartum mice. Compared to intact tissue structure, closely arranged epithelial cells, and lack of inflammatory cells in the acinar cavity of the mammary glands of PBS-injected controls, the S. aureus-infected glands showed a time-dependent increase in inflammation and tissue damage. The inflammatory cells started to infiltrate into the acinar cavity within 6 h of S. aureus inoculation, and increased over the course of infection. The mammary tissue structure was largely intact at the 6 h time point and started to collapse by 12 h with epithelial cell depletion and massive infiltration of the inflammatory cells. These pathological changes steadily worsened till 48 after the infection. Consistent with the inflammatory damage, TNFA, IL6, and IL1B mRNA levels started to spike 12 h after S. aureus infection, and increased steadily till 48 h (Figure 12). Furthermore, the infected glands showed significantly higher levels of p-NF-κB p65 after 12 h compared to the uninfected control, while PI3K, AKT, and p-NF-κB p65 were upregulated after 24 h. Interestingly, CBLB protein levels increased significantly only after 48 h of infection, wherein all the aforementioned proteins were also upregulated (Figure 13). These results clearly indicate that CBLB plays a key role in driving S. aureus-induced inflammatory damage in mammary glands by activating the PI3K/AKT/NF-κB pathway.