Non-targeted multimodal metabolomics data from ovine rumen fluid fractions

ABSTRACT From an animal health perspective, our understanding of the metabolites in rumen fluid across different host species is poorly understood. Here, we present a metabolomic data set generated using hydrophilic interaction liquid chromatography and semi-polar (C18) chromatography methods coupled to high-resolution mass spectrometry of fractionated ovine rumen samples.

R uminant livestock are an important component of feeding the growing human population while also being sources of global greenhouse gas emissions. The rumen is a strictly anaerobic environment enriched with a complex community of bacteria, protozoa, fungi, archaea, and bacteriophages. Rumen microbiota breakdown and convert plant proteins and polysaccharides from feed into energy sources but also result in methane formation that affects ruminant productivity. Metabolomics is a powerful and sensitive approach for investigating low-molecular-weight metabolite profiles present in rumen biofluids. It can be used to identify potential roles of metab olites in the rumen microbiome and provide understanding of host-level regulatory mechanisms associated with animal production. While rapid developments in genomics have accelerated our knowledge of rumen molecular biology (1), there has been less work focusing on the low-molecular-weight molecules that stem from rumen fermenta tion of feed and a complete absence of metabolomics studies on ovine rumen samples (2)(3)(4)(5)(6)(7)(8).
Whole rumen content samples were collected post-mortem and pooled from five sheep (Fig. 1A) grazing ad libitum on a ryegrass and clover pasture diet in Palmerston North, New Zealand (40°18′ S, 175°45′ E). A method was developed to acquire dia lyzed rumen fluid (DRF) fractions that enrich for different sized components (Fig. 1B). DRF fractions based on three molecular weight cutoffs (MWCO) were obtained using Spectra-Por Float-A-Lyzer G2 dialysis systems with MWCOs of 20 kDa (Z726931, Sigma-Aldrich), 8-10 kDa (Z726605, Sigma-Aldrich), and 100 Da (Z727253, Sigma-Aldrich). Approximately 5 L of rumen contents was collected from each animal and filtered through four layers of cheesecloth (335 µm mesh) to account for the macro components of rumen fluid and transferred into Schott gas washing bottles fitted with Drechsel type head connections (GL 14, DURAN). To obtain each DRF fraction, replicates of each individual MWCO apparatus (n = 5 for each) were dialyzed against 10 mL of autoclaved phosphate buffered saline buffer overnight at 39°C in a water bath under anaerobic conditions obtained by insufflating a stream of O 2 -free CO 2 inside the container with constant mixing. DRF samples were snap-frozen in liquid nitrogen, transferred to glass vials, and stored at −80°C until further use.
To comprehensively survey metabolites associated with DRF fractions, we used hydrophilic interaction liquid chromatography (HILIC) to separate polar compounds and C18 chromatography for separation of semi-polar compounds (Fig. 1C). Analyses were performed in both positive and negative electrospray ionization mode at the resolving power setting of 25,000 with a maximum trap fill time of 100 ms using the Xcalibur v4.3 software. The LC-MS raw data files were converted to mzXML files using MSConvert function of ProteoWizard v3 software (9). Quality control and peak deisotoping analysis were based on our previously published procedures (10). The details of extraction Announcement procedures, chromatographic gradients, and instrument settings have been previously described by Palevich et al. (11). For each metabolomic data set, principal component analysis (PCA) was performed on the log 10 intensities for raw chromatographic peaks to assess similarity and separation of DRF fractions. The mixomics package v6.16.3 in R v4.1.1 was used to perform the PCA and generate plots.
Overall, according to the PCA scores plots (Fig. 1C), each of the three DRF frac tions had considerably different profiles for both types of metabolomic analyses and regardless of ionization mode. The three DRF fractions separated completely either on the first or second component or a combination thereof within the 95% CI ellipse. This study highlights the potential of HILIC and C18 chromatography combined with non-targeted mass spectrometric methods to detect the polar and semi-polar metabo lite species of the ruminal fluid metabolome. The presented untargeted metabolomics data provide a detailed snapshot of the ovine ruminal fluid metabolome that can be used as a reference for future studies of the rumen metabolome or as a comparator for other ruminant species. Special thanks go to Linley Schofield in our Rumen Microbiology lab for providing the dialysis equipment, her guidance, and advice on using the apparatus. We thank Hailey Gillespie and Trevor Holloway for being always willing to collaborate and especially for their timely assistance with the collection of rumen fluid. We also wish to acknowledge Arvind Subbaraj for assistance with the LCMS analysis for this study and Alastair Ross and Karl Fraser for their feedback on early versions of the manuscript.

DATA AVAILABILITY
The data sets presented in this study and supporting the conclusions of this article have been made available in the MetaboLights database (MTBLS1717) online repository.