Identification of bovine respiratory disease through the nasal microbiome

Abstract

Background: Bovine respiratory disease (BRD) is an ongoing health and economic challenge in the dairy and beef cattle industries. Multiple risk factors make an animal susceptible to BRD. The presence of Mannheimia haemolytica, Pasteurella multocida, Histophilus somni, and Mycoplasma bovis in lung tissues have been associated with BRD mortalities, but they are also commonly present in the upper respiratory tract of healthy animals. This study aims to compare the cattle nasal microbiome (diversity, composition and community interaction) and the abundance of BRD pathogens (by qPCR) in the nasal microbiome of Holstein steers that are apparently healthy (Healthy group, n = 75) or with BRD clinical signs (BRD group, n = 58). We then used random forest models based on nasal microbial community and qPCR results to classify healthy and BRD-affected animals and determined the agreement with the visual clinical signs. Additionally, co-occurring species pairs were identified in visually BRD or healthy animal groups.

Results: Cattle in the BRD group had lower alpha diversity than pen-mates in the healthy group. Amplicon sequence variants (ASVs) from Trueperella pyogenes, Bibersteinia and Mycoplasma spp. were increased in relative abundance in the BRD group, while ASVs from Mycoplasma bovirhinis and Clostridium sensu stricto were increased in the healthy group. Prevalence of H. somni (98%) and P. multocida (97%) was high regardless of BRD clinical signs whereas M. haemolytica (81 and 61%, respectively) and M. bovis (74 and 51%, respectively) were more prevalent in the BRD group than the healthy group. In the BRD group, the abundance of M. haemolytica and M. bovis was increased, while H. somni abundance was decreased. Visual observation of clinical signs agreed with classification by the nasal microbial community (misclassification rate of 32%) and qPCR results (misclassification rate 34%). Co-occurrence analysis demonstrated that the nasal microbiome of BRD-affected cattle presented fewer bacterial associations than healthy cattle.

Conclusions: This study offers insight into the prevalence and abundance of BRD pathogens and the differences in the nasal microbiome between healthy and BRD animals. This suggests that nasal bacterial communities provide a potential platform for future studies and potential pen-side diagnostic testing.

Keywords: 16S rRNA gene; Bovine respiratory disease; Cattle nasal microbiome; qPCR.

Fig. 1

Alpha diversity of the nasal microbiome in cattle that are apparently healthy or display BRD clinical signs (BRD). Observed ASVs (a) and Chao 1 (b) measure the richness of the microbiome community. Evenness was measured with Pielou (c), and the phylogenetic relationship was measured with Faith’s PD (d). An asterisk (*) and horizontal line represent a statistical difference (p ≤ 0.05) between the two groups. Colored circles and lines represent the means and standard error of the BRD and healthy groups, respectively, and the gray dots represent the raw data of each group

Fig. 2

Principal coordinate analysis (PCoA) of Weighted UniFrac distances (a) and Bray–Curtis dissimilarity (b) between BRD and healthy animals. Ellipses indicate a 95% confidence interval of individuals belonging to each health status group. Axis 1 represents the axis that explains the greatest amount of the variation followed by Axis 2. Larger points indicate the centroids of the ellipses. Distances of the centroids between the two groups are indicated in the caption below each plot

Fig. 3

Differentially abundant taxa (ASVs) between animals with BRD clinical signs (BRD) and healthy animals. Bar plot shows the taxa with a log₂ fold change greater than 2 or less than − 2 and p ≤ 0.05. Those with a log₂ fold change > 2 were those enriched in BRD animals, while a log₂ fold change < − 2 were those decreased in the BRD animals. Taxa names contain numbers in parenthesis if multiple ASVs were assigned the same taxonomy

Fig. 4

Prevalence of BRD pathobionts in the nasal microbiota of Holstein steers (n = 133) and between healthy (n = 75) and BRD (n = 58) Holstein steer pen-mates according to clinical signs. Prevalence of Pasteurella multocida (a), Mannheimina haemolytica (b), Histophilus somni (c) and Mycoplasma bovis (d)

Fig. 5

Difference in bacterial abundance per sample (200 µl of extracted DNA) for Mycoplasma bovis (a), Mannheimia haemolytica (b) and Histophilus somni (c) between animals with (BRD) and without (Healthy) BRD clinical signs (n = 133). An asterisk (*) and horizontal line represent a statistical difference (p ≤ 0.05) between the two groups. Colored circles represent the means of the BRD, and healthy group, vertical lines indicate the standard error of the means, and the gray dots represent individual samples of each group

Fig. 6

Probability of classifying animals as BRD or healthy (< 0.5 = healthy, > 0.5 = BRD) using Random Forest analysis. Classification of the animals was based on the microbial community composition (ASV table) and quantification of BRD pathobionts, 16S rRNA gene abundance and age (qPCR). The color indicates the initial animal classification based on the BRD clinical signs. Shape indicates if the animal classification agreed between the three methods: visual classification based on BRD clinical signs (V), microbial community composition (ASV Table) and quantification of BRD pathobionts and 16S rRNA gene abundance (qPCR)