Rumen-centric assembly of the cattle holobiont

Abstract

The holobiont perspective takes into account the complete system of an animal including its microbiomes and to some extent the factors that form its external environment. This view has a scientific motivation: In the case that the microbiome affects the host, studying the host in isolation severely limits access to the complete information needed to understand the biology of the host. As a model for such a “holobiont” system, a starting point was taken in cattle and focused on the rumen and the interface that connects its inherent microbiome to the metabolism of the host. As the rumen microbiome produces a vital component of metabolites that the host budgets for energy and nutrient assimilation, it has a wide potential to impact the host. The association between the host and its rumen microbiome has made it a focal point for modulation strategies to improve health, nutrition and sustainable production of ruminants. However, the complexity of the rumen microbiome and its interactions with the host represent major challenges that must be overcome before microbiome-based approaches can be used in practice. To improve reconstruction of the rumen microbiome, a high-resolution dataset was generated for deep analysis from 80 cattle subjected to a feedlot trial. Here, the rumen microbiome was sampled over time, and host tissue (rumen wall and liver) samples were collected upon sacrifice, after rigorous measurement of the cattle’s key performance traits (KPTs) and methane emissions. To study the ruminant holobiont, molecular layers in both the host and its rumen microbiome were reconstructed. Multiple molecular layers (DNA, RNA, protein, metabolites, and glycans) as well as the host phenotype were analyzed, in order to track how potential interactions affected metabolism in 24 individual animals that exhibited the highest natural variation in measured methane yield. As most biological variation of an organism is encoded in the genome, DNA sequencing is central to forming a foundation upon which to assemble the holobiont. To further enhance our DNA analyses, long-read and high accuracy short-read shotgun metagenome sequencing was applied to reconstruct the microbial genomes of the rumen microbiome. To track which genes were expressed, transcriptomics was applied, and to analyze the presence of translated proteins and their derived metabolites, also proteomics and metabolomics. For complex eukaryotic populations that are unamenable to shotgun sequencing approaches, such as the protozoa and fungi, genomes were sourced from collaborative projects. Analyzing a single molecular layer requires a specific set of technical tools. For this purpose, it is described how microbial genomes can be reconstructed, and how their relevant metabolic functions can be identified. Specifically in relation to the carbohydrate-active enzymes (CAZymes) that enable ruminants to assimilate carbon and energy from plant fibers, representing the basis for the energy budget of the host. As it was not possible to identify a suitable pipeline with the tools necessary to analyze and compare the metabolic function of the archaeal and bacterial genomes generated in our datasets, an easy to use platform for analyzing metagenome-assembled genomes (MAGs) was developed. This pipeline is now distributed on Bioconda as CompareM2. To apply the wide tool set that was put together and attempt to improve resolution and general understanding of how the ruminant host and its microbiome function as an integrated unit, an experimental cattle holobiont dataset was analyzed. The sampled molecular layers were refined to become biologically relevant representations on which integrative holo-omic methods could be applied to identify and investigate possible host-microbiome interactions. In practice, simpler computational dimensionality reduction methods may offer greater interpretability and allow more direct biological interpretation than more complex computational methods. Applying these methods to our experimental data led to the discovery that the protozoal fraction of the rumen seemingly drives two exclusive community types across the individual animals that were sampled, which have previously been described from micrography and 18S studies. These are referred to as RCT-A and -B (rumen community type). RCT-B is dominated by protozoa affiliated to Epidinium spp. that were observed to employ a plethora of fiber-degrading enzymes, which most likely provide favorable conditions for saccharolytic bacteria such as Prevotella spp. Conversely RCT-A is dominated by Isotricha and Entodinium protozoal species and harbors a wider representation of fiber, protein and amino acid fermenters. While no clear host effect for these rumen community types is found, there are signs in the more complex network analysis based computational methods that certain microbial populations of Acutalibacteraceae prevalent in RCT-A affect methionine metabolism in the rumen wall. This calls for further refinement of the holo-omic analyses and biological characterization. Finally, our work highlights the need for de facto standards to refine individual molecular layers, and to find common methods for data integration across these molecular layers that represent the host-microbiome axis.

Holobiont-perspektivet tar hensyn til det komplette systemet til et dyr, inkludert dets mikrobiomer og, til en viss grad, faktorene som danner dets ytre miljø. Dette perspektivet har en vitenskapelig motivasjon: I tilfeller der mikrobiomene påvirker verten, begrenser det å studere verten i isolasjon tilgangen til den komplette informasjonen som er nødvendig for å forstå vertens biologi. Som en modell for et slikt holobiont-system har vi tatt utgangspunkt i storfe, med fokus på vomma og grensesnittet som forbinder dets iboende mikrobiom med vertens metabolisme, spesifikt vomveggen og leveren. Siden mikrobiomet i vomma produserer en viktig komponent av metabolitter som verten bruker til energibudsjettering og næringsopptak, har det et stort potensial til å påvirke verten. Sammenhengen mellom verten og dens iboende vommikrobiom er derfor et fokusområde for moduleringsstrategier for å forbedre helse, ernæring og bærekraftig storfeproduksjon. Vommikrobiomets kompleksitet og dets interaksjoner med verten representerer imidlertid store utfordringer som må takles før mikrobiom-baserte tilnærminger kan brukes i praksis. For å forbedre rekonstruksjonen av vommikrobiomet har vi generert et høyoppløselig datasett for dyp analyse fra 80 storfe som ble eksponert for en fôringsprøve. Her ble vommikrobiomet tatt prøver av over tid, og prøver fra vomveggen og leveren ble samlet ved avliving, kort tid etter grundige målinger av storfeets nøkkelprestasjonsegenskaper, inkludert metanutslipp.