Dietary energy drives the dynamic response of bovine rumen viral communities

Abstract

Background: Rumen microbes play a greater role in host energy acquisition than that of gut-associated microbes in monogastric animals. Although genome-enabled advancements are providing access to the vast diversity of uncultivated microbes, our understanding of variables shaping rumen microbial communities is in its infancy. Viruses have been shown to impact microbial populations through a myriad of processes, including cell lysis and reprogramming of host metabolism. However, little is known about the processes shaping the distribution of rumen viruses or how viruses may modulate microbial-driven processes in the rumen. To this end, we investigated how rumen bacterial and viral community structure and function responded in five steers fed four randomized dietary treatments in a crossover design.

Results: Total digestible nutrients (TDN), a measure of dietary energy, best explained the variation in bacterial and viral communities. Additional ecological drivers of viral communities included dietary zinc content and microbial functional diversity. Using partial least squares regression, we demonstrate significant associations between the abundances of 267 viral populations and variables driving the variation in rumen viral communities. While rumen viruses were dynamic, 14 near ubiquitous viral populations were identified, suggesting the presence of a core rumen virome largely comprised of novel viruses. Moreover, analysis of virally encoded auxiliary metabolic genes (AMGs) indicates rumen viruses have glycosidic hydrolases to potentially augment the breakdown of complex carbohydrates to increase energy production. Other AMGs identified have a role in redirecting carbon to the pentose phosphate pathway and one carbon pools by folate to boost viral replication.

Conclusions: We demonstrate that rumen bacteria and viruses have differing responses and ecological drivers to dietary perturbation. Our results show that rumen viruses have implications for understanding the structuring of the previously identified core rumen microbiota and impacting microbial metabolism through a vast array of AMGs. AMGs in the rumen appear to have consequences for microbial metabolism that are largely in congruence with the current paradigm established in marine systems. This study provides a foundation for future hypotheses regarding the dynamics of viral-mediated processes in the rumen.

Keywords: Auxiliary metabolic genes; Phage ecology; Rumen; Viral diversity; Viral metagenome.

Fig. 1

Normalized abundances (log₂ counts per kilobase per million reads) of viral families identified in rumen viral metagenomes. Genome-binning of contigs resulted in 2243 viral populations (see the “Methods” section), of which 118 displayed homology to viruses in the RefSeq database. Viral populations with homology to known viruses accounted for 2.4% of all viral metagenome reads. Viral annotations were assigned based on nucleotide similarity of contigs to the viral RefSeq database (BLASTN, E-value threshold of 10⁻⁵, bit-score threshold of 50). 27CDS—27% condensed distillers solubles; 40MDGS—40% modified distillers grains plus solubles; 55CS—55% corn silage

Fig. 2

Phylogenetic relationships and normalized median diet abundances (log₂ counts per million reads) of identified OTUs. Families with a maximum abundance greater than 3% in a sample are denoted on the phylogenetic tree. The most abundant bacterial families included Rumminococcaceae, Lachnospiraceae, and Prevotellaceae, in agreement with previous studies of the rumen microbiome [3]

Fig. 3

Alpha diversity comparisons of rumen bacterial OTUs and viral populations across diets. Species richness (Chao1 index) and diversity (Shannon’s diversity index) were found to significantly differ by diet in bacterial (a) and viral (b) communities (P < 0.05, three-way ANOVA considering the effects of diet, period, and steer). The combinations of diets that differed in richness and diversity in the bacterial and viral datasets were unique (P < 0.05, post-hoc pairwise t tests). No differences were observed in bacterial or viral alpha diversity metrics based on host animal or period. Letters denote differences in richness and diversity observed when comparing the alpha diversity metrics using pairwise comparisons of diets. 27CDS—27% condensed distillers solubles; 40MDGS—40% modified distillers grains plus solubles; 55CS—55% corn silage

Fig. 4

Unconstrained NMDS ordination analysis was used to visualize beta diversity based on OTU (a) and viral population (b) abundances across samples. PERMANOVA results indicate that structuring of bacterial communities is best explained by diet (diet: P = 0.001, R ² = 0.365; period: P = 0.188, R ² = 0.140; steer: P = 0.312, R ² = 0.166). Both dietary and host animal effects were found to significantly explain the variation in viral populations (diet: P = 0.002, R ² = 0.266; period: P = 0.096, R ² = 0.145; steer: P = 0.048, R ² = 0.290). 27CDS—27% condensed distillers solubles; 40MDGS—40% modified distillers grains plus solubles; 55CS—55% corn silage. 222, 259, 346, 3244, and 3257 represent animal identifiers

Fig. 5

Analysis of intra-diet and intra-steer distances and dissimilarities reveals a larger influence of host-associated factors on viral populations compared to bacterial OTUs. a) Within diet weighted UniFrac distances were significantly lower than within animal weighted UniFrac distances in the rumen bacterial communities (P = 0.001, t = − 3.366, t test). b) No statistically significant differences were observed between intra-diet and intra-steer Bray-Curtis dissimilarities in viral populations (P = 0.067, t = − 1.896, t test)

Fig. 6

Ordination of CAP analysis displaying the factors influencing rumen bacterial and viral communities. The CAP model with TDN, protein, and zinc as independent variables explained a significant portion of the variation in bacterial OTUs (P = 0.001, F = 2.736, variation explained = 33.9%, ANOVA permutation test for CAP) (a). Subsequent testing of the marginal effects of each term found TDN (P = 0.001, F = 4.342), zinc (P = 0.002, F = 3.587), and protein (P = 0.012, F = 2.379) to vary significantly with the microbial communities. Backward selection identified TDN, protein, zinc, and microbial functional diversity (Shannon’s diversity index) as potential drivers of rumen viral populations (b). The CAP model including these independent variables explained a significant portion of the variation in viral metagenomes (P = 0.002, F = 1.383, variation explained = 35.6%, ANOVA permutation test for CAP). Investigation of the marginal effects of each variable revealed TDN (P = 0.001, F = 1.936), zinc (P = 0.025, F = 1.482), and microbial functional diversity (P = 0.022, F = 1.402) to be the predominant factors influencing rumen viral populations. 27CDS—27% condensed distillers solubles; 40MDGS—40% modified distillers grains plus solubles; 55CS—55% corn silage

Fig. 7

Normalized abundances (log₂ counts per million reads) of virally encoded AMGs in deep viral metagenomes. The identification of AMGs was restricted to deep viral metagenome contigs from low (55CS) and high (27CDS) TDN diets that were longer than 1.5 kbp and identified as viral by VirSorter [61]. 27CDS—27% condensed distillers solubles; 40MDGS—40% modified distillers grains plus solubles; 55CS—55% corn silage