Droplet digital PCR quantification of selected microRNAs in raw mastitic cow’s milk from the west of Poland

Inflammation of the mammary gland, referred to as mastitis, is the most common disease in dairy cattle worldwide. It has been recognised as a major cause of economic losses in the dairy industry because it decreases milk production, impairs fertility, necessitates culling of affected animals, and imposes treatment cost (13, 30, 32). Mastitis is classified as clinical (CM) or subclinical (SCM), depending on the effects of the inflammation on the mammary gland. Although SCM has no visible effects on the udder or milk, it can affect milk composition and consequently product quality (28). Subclinical mastitis is approximately forty times more common than the clinical variety and is more difficult to detect (11). A well-known indicator of mastitis is somatic cell count (SCC) in milk, which reflects the health status of the udder (38). Usually, SCC values of over 200,000 cells/mL are considered to indicate SCM (12, 29). However, SCC in the milk can vary with many other factors, such as breed, age, parity, stage of lactation, body condition score, stress, and methods of transportation and storage of milk samples (31). A further demerit of SCC as a mastitis indicator is that the rise in the number of somatic cells appears after a significant delay in relation to the progress of inflammation (4). There are also other methods of determining the severity of inflammation (electrical conductivity of milk, lactose level and acute phase protein level); however, the results obtained by these methods are subject to significant error and so their usefulness is low. The search is in progress for alternative methods of assessing the severity of the inflammation of the mammary gland which will allow the severity of mastitis to be identified with greater accuracy (7). Novel, sensitive, and accurate methods for the early detection of affected animals, especially those with mastitis in the subclinical stage, are therefore urgently needed (5).

MicroRNAs (miRNAs) are a class of small noncoding RNA molecules present in all body tissues and fluids; they have been recognised as useful biomarkers in the detection of many diseases (10, 33). In milk they can be present both as free molecules and packaged in vesicles, such as milk exosomes or fat globules (2). MicroRNAs encapsulated in vesicles or conjugated to RNA-binding lipoproteins are more stable and can be measured by a range of techniques, including quantitative PCR, digital PCR and high-throughput assays (25). A number of studies have identified miRNAs associated with mastitis in cattle (18, 19, 34, 37); however, different miRNAs have been proposed as mastitis biomarkers in different studies (24). The low repeatability of the results supporting certain miRNAs as viable biomarkers may be associated with the type of microorganism responsible for infection. The most common causes of mastitis are Gram-positive and Gram-negative bacteria, although viral, fungal and protothecal infections can also cause mastitis to develop (39). It has been shown that prevalence of individual pathogens isolated from milk samples can change over the years. For example, Staphylococcus aureus was the most commonly isolated bacterium in northeastern Poland in 2013, while from 2014 to 2019 Streptococcus spp. were the most common (15). Therefore, studies identifying miRNAs that correlate well with mastitis stage in groups in different locations are needed. To the best of our knowledge, studies using miRNAs from milk as predictors of mammary gland inflammation in cattle have not been conducted yet in Poland.

The aim of this study was to use the highly sensitive droplet digital PCR (ddPCR) method to quantify selected miRNAs (miR-21a, miR-92a, miR-146a and miR-383) in bovine milk with numbers of somatic cells characteristic of different microbiological profiles. These miRNAs were selected based on literature results that point to their frequent association with bovine mastitis.

Milk quarter samples were collected from 100 cows, but 10 cows with clinical mastitis were excluded because clots in the raw milk made it impossible to obtain an SCC. Cows were described in terms of their SCC and CMT parameters. During CMT testing, 38 cows were determined to be healthy and 52 cows’ mastitis was confirmed. After verification by SCC, a difference between the mastitic status indicated by the CMT and the actual inflammation state (Low, Medium or High) based on the SCC was noted. There was only 1 positive CMT test among the Low cases, but there were 14 negative CMT tests among the Medium and High cases.

Analysis of miRNA in milk samples

The expression of all miRNAs was successfully detected using ddPCR, but the level of miR-383 was very low; therefore, this miRNA was excluded from further analysis.

Since ddPCR allows absolute quantification of miRNAs, in the first step, the miR-21a, miR-92a and miR-146a concentrations were examined. It has been shown that all these molecules are upregulated in mastitic milk. In the next step, the level of miRNA concentration (directly measured) was analysed in relation to the SCCs. In the case of miR-21a, the correlation between concentration and SSC was quite low in all analysed samples (R = 0.32, P-value < 0.01) (Fig. 1), and was probably driven by two outliers from the High class.

Fig. 1.

The correlation was not found to be statistically significant when the milk samples were divided into the three Low, Medium and High SCC groups (Fig. 1). In the case of miR-92a, a negative correlation (R = -0.34, P-value < 0.01) was observed when all the samples were analysed together (Fig. 1), and no significant results were found when the samples were divided into the three groups (Fig. 1). The most promising results were obtained for the concentration of miR-146a, where the correlation with SSC was high (R = 0.71, P-value < 0.01) for all samples (Fig. 1). When these were divided into the three SSC groups, a statistically significant correlation was observed for the Low (R = 0.44, P-value < 0.05) and High groups (R = 0.7, P-value < 0.01) (Fig. 1).

Although ddPCR allows absolute quantification of nucleic acids, various normalisation methods are recommended to avoid technical errors. In acknowledgement of this, it was decided to show results not only as concentrations but also as relative expression levels. One of the examined miRNAs, miR-92a, has been proposed as a housekeeping gene in the analysis of bovine mastitis–related microRNA in milk (17); it was used to normalise miR-21a and miR-146a expression (Fig. 2).

Fig. 2.

The correlation between SCC and miRNA relative expression level for the samples analysed together was as follows: R = 0.53 (P-value < 0.01) for miR-21a and R = 0.79 (P-value < 0.01) for miR-146a (Fig. 2). When the samples were divided into the three SSC groups, a correlation was observed for miR-21a in the High group (R = 0.44; P-value < 0.05) (Fig. 2) and for miR-146a in the Medium (R = 0.59, P-value < 0.01) and High groups (R = 0.7, P-value < 0.01).

Taking the correlations all together, miR-146a proved to be the most promising potential biomarker for the diagnosis of mastitis in the samples, since a positive correlation was observed between concentration and relative expression of this miRNA and SSC milk samples for all the samples, including when the samples were grouped. The strongest correlation was observed in the High group.

Correlation between the isolated mastitis pathogens and the examined miRNAs

Comparison of the miR-21a relative expression levels across the Control, Negative and Positive groups showed no significant differences (P-value = 0.08). For miR-146a, the relative expression levels between the groups differed significantly (P-value < 0.01; Fig. 3). Multiple comparisons revealed that the expression level did not differ between the Positive and Negative groups (P-value = 0.9), but that the level in the Control group was significantly different to those of the other two (P-value < 0.01).

Fig. 3.

The CMT is currently the most common method of assessing the severity of bovine mastitis. It suffers from low sensitivity (<50%) but has the advantage that it can be performed under field conditions (9). Another common method of assessing the severity of inflammation of the mammary gland is through determination of the SCC of raw milk. This method is much more precise, but requires laboratory conditions. It also has the disadvantage that the increase in SCC follows the development of inflammation with a significant delay (1). Alternative methods of diagnosing mastitis in dairy cattle are thus needed. MicroRNA has recently been recognised as a potential diagnostic tool in the detection of bovine mastitis (24). Although many studies of miRNA expression levels in bovine milk have been carried out, most of them have evaluated mastitis status using CMTs, whereas our study employed both CMTs and SCC. In our experiment, 14 negative CMT results proved false, as SCC showed mastitis in these animals (SCC ≥ 200,000 cells/mL). Comparing the expression level of miRNA to the CMT results may thus introduce a significant error. We therefore decided to use only SCC data in the final analysis. We examined the potential role of miRNA for detecting mastitis on Polish cattle farms, and to the best of our knowledge investigated this as the first to do so. We also examined the milk samples bacteriologically, a step which may be highly productive when there is a relationship between a specific pathogen and a specific miRNA’s expression. However, in this case, bacteriological examination did not allow us to find a significant difference between bacterial and nonbacterial mastitis in the expression level of miR-21a or miR-146a.

The results of the study by Lai et al. (18) and of other studies led us to initially select four miRNA molecules (miR-21a, miR-92a, miR-146a and miR-383) for the research. It has been shown that miR-21 expression in blood is upregulated in cows with mastitis caused by Staphylococcus aureus (6). Moyes et al. (21) suggested that the inflammatory response in the mammary gland involves crosstalk with the liver to coordinate the inflammatory response. Sensory neuron–derived exosomes containing miR-21a can increase the expression of proinflammatory genes and proteins when phagocytosed by macrophages. This result suggests that miR-21a plays an important role in communication between systemic organs in fighting infections of the mammary gland. Our results showed low levels of correlation between miR-21a and SCC. Similar conclusions were made by Naeem et al. (22), who found no significant difference in miR-21a expression levels between cows with mastitis caused by Streptococcus uberis and those with a healthy udder. On the other hand, Li et al. (20) found that mastitis caused by Staphylococcus aureus resulted in a 1.7-fold increase in miRNA-21a expression levels.

The ribonucleic acid miR-92a has been recognised as pro-inflammatory miRNA that is involved in the upregulation of several proinflammatory genes in macrophages (8). Our study did not find a correlation between the concentration of this molecule and SCC. Thus, following the recommendation of Lai et al. (17), the level of miR-92a was treated as a housekeeping gene level for the analysis of bovine mastitis-related microRNA in milk.

Our preliminary studies did not confirm that miR-383 is upregulated in mastitic milk; therefore, this molecule was excluded from the experiment. However, other studies found that miR-383 was significantly upregulated in CMT-positive milk (18) and in mammary epithelial cells in Streptococcus agalactiae-induced mastitis (26).

Many studies have examined the use of miRNA-146a in diagnosing bovine mastitis in milk and serum samples. As mentioned, Lai et al. (18) indicated that there were changes in the expression levels of this miRNA in mastitic cows confirmed by CMT. However, the number of animals in that experiment (18 CMT−animals and 17 CMT+ animals) was quite small. The same research group performed a genome-wide miRNA analysis of normal milk samples and milk samples from cows with clinical mastitis, also confirmed by CMT. They found 25 differentially expressed miRNAs during bovine mastitis, only three of which (miR-21a, miR-146a and miR-222) were the same as in the report published previously (19). In another study, that of Srikok et al. (34), eight a priori selected miRNAs were evaluated; three of them – miR-29b-2, miR-146a and miR-155 – were found to be good candidates to determine the infected status of cows using milk samples. Promise has been attributed to miR29b-2 as a biomarker for bovine mastitis. The recent study of Tzelos et al. (37) examined four miRNAs (miR-26a, miR-142, miR-146a and miR-223) in a large number of individual quarter milk samples (n = 236) collected from dairy cows at different lactations. They found that levels of miR-142, miR-146a and miR-223 could be used to identify the early stages of inflammation in individual quarter milk samples with high accuracy.

The examples presented above indicate that miR-146a may indeed be a good biomarker of bovine mastitis. It is known that this molecule regulates various biological pathways, including immunity, the development and function of immune components (14), and the response to microbial infection (35). For example, the expression level of miR-146a in B and T lymphocytes is an induced response to microbial infection (36). In our study, miR-146a showed differential expression between the control and experimental groups. High sensitivity (0.98) and specificity (0.95) were noted only when the cut-off value between mastitic and non-mastitic milk samples was an SSC of 1,000,000/mL or higher. The ability to correctly predict positive and negative results was also high and was reflected in sensitivity of 0.95 and specificity of 0.98. Our results allow us to conclude that the level of expression of miR-146a is a good indicator of the severity of inflammation of the udder, especially for more severe inflammations, where the SCC in the tested milk is 1,000,000/mL or above. Using miR-146a to detect the severity of inflammations where the SCC is between 200,000 and 999,999/mL is much less sensitive (0.67) and specific (0.85). Our results indicate a need for further studies to identify new inflammatory markers for detecting inflammation in cases where the SCC is greater than 200,000/mL. In this study, we performed the first analysis of selected miRNAs in milk samples from healthy Polish cows and Polish cows suffering from clinical or subclinical mastitis confirmed using SCC. We showed that ddPCR is a useful method of quantifying the tested miRNAs in milk samples. As markers for bovine mastitis, miRNAs and miR-146a proved promising, but it is necessary to seek new markers that would allow identification of subclinical bovine mastitis.