Rumen metabolomics – a powerful tool for discovery and understanding of rumen functionality and health
The rumen is a complex ecosystem comprised of microorganisms including bacteria, archaea, fungi and protozoa (de Almeida et al., 2018). In recent years our understanding of the immense impact the gut microbiome has on host performance, health and well-being has come to light. While the majority of studies in the field of microbiome science to date have been conducted in a human context, much of the observed impacts are transferable to animals. Indeed, the gut microbiome has been described as a virtual endocrine organ in the context of both humans (O'Hara and Shanahan, 2006) and domesticated animals (O'Callaghan et al., 2016). This is as a result of its ability to impact the functioning of both local (within the gut) and distal organs and systems throughout the body, such as host metabolism, brain and behaviour, liver function, cardiovascular system, enteric nervous system and immune system (O'Callaghan et al., 2016).
Currently at ~7.5 billion people, the world population is increasing, with that so too does the global demand for food. The rumen provides ruminants with the ability to digest materials that are indigestible for humans, such as cellulose-rich feedstuffs. This digestion produces metabolite substrates for rumen microbes and nutrients for the production of highly nutritious food items such as milk and meat (Saleem et al., 2013). It is therefore apparent that greater
http://dx.doi.org/10.19103/AS.2020.0067.05 © Burleigh Dodds Science Publishing Limited. 2020. All rights reserved.
understanding of the rumen microbiome and its functionality will be key in maximising efficiency, health, sustainability and productivity of domesticated production animals in the future.
The metabolome is formally defined as the collection of all small-molecule metabolites (endogenous or exogenous) that can be found in a living cell or living organism (Wishart, 2005). Metabolomics is an emerging field of study for the analysis, characterisation and quantification of small molecules and metabolites, using technologies such as nuclear magnetic resonance (NMR), liquid/gas chromatography (LC/GC) coupled with mass spectrometry (MS) (Wishart, 2008). The application of metabolomics has traditionally been used in biomedical, nutritional and crop research. However, in recent years, metabolomics has been gaining prominence in the fields of livestock research and livestock monitoring - for animal health assessment, disease diagnosis and bio product characterisation (Goldansaz et al„ 2017). There is also an increased interest in the application of metabolomics for identification of prospective biomarkers of production traits such as weight gain, milk quality and health (do Prado et al., 2018). Biomarkers can also be extremely useful in confirming food authenticity and combatting food fraud increasing food security. Metabolomics offers a variety of benefits over traditional wet chemistry type analysis of samples in that it offers high throughput analysis of a large variety of metabolites at one time. In addition, it can be non-invasive and is often less time-consuming with more cost-effective sample preparation than traditional approaches. However, the instrumentation and expertise required for metabolomic analysis are not yet widely available, are expensive to purchase and maintain, and require large data resources to facilitate and interpret results. Primarily focussing on bovine animals, the objective of this chapter is to provide an overview of the rumen metabolome as we know it today and highlight factors that can affect its composition and resulting functionality.
1.1 Targetted versus untargeted metabolomics
Metabolomic analysis can be carried out using a targeted or non-targeted approach, each with its own advantages and disadvantages. Targeted metabolomics involves analysing samples for a pre-defined list of metabolites with quantitative measurement. As such, prior knowledge of the samples and molecules of interest is required with typical applications including biomarker identification and validation, and analysis of metabolic pathways. Targeted metabolomics can be beneficial in terms of reduced data analysis and spectra interpretation, which can often be very time-consuming. However, by its nature the targeted approach is limiting, as it does not examine global coverage of the metabolome. Untargeted metabolomics, on the other hand, does not work off a predefined list of metabolites but instead aims to identify whatever metabolites are detectable within the sample, restricted only by instrumentation sensitivity and coverage, and extraction methodology. While untargeted metabolomics offers increased coverage and characterisation of the sample metabolome, offering the opportunity for discovery of novel compounds and biomarkers (Cajka and Fiehn, 2016), it can often require more time-consuming data analysis for spectral peak identification and interpretation, and is not as quantitative as the targeted approach.
The rumen metabolome: technologies for analysis and extraction techniques
Efforts to characterise the composition of the rumen metabolome have been carried out in the past. Pioneering work carried out by Saleem et al. (2013) aimed to characterise the rumen metabolome using a range of technologies. The product of this research was the development of the rumen metabolome database (www.rumendb.ca) which is a comprehensive web accessible resource containing >200 positively identified and quantified rumen metabolites, their structures and respective concentrations.
An example process for rumen metabolomics analysis is shown in Fig. 1. The two leading analytical approaches for metabolomic analysis are nuclear magnetic resonance (NMR) and mass spectrometry (MS). Within these a variety of different technologies exist for analysis including 1H-NMR, 13C-NMR, 15N-NMR and 3,P-NMR. Although NMR is a broadly used technique, others including gas chromatography with mass spectrometry (GC-MS), liquid chromatography with mass spectrometry (LC-MS), capillary electrophoresis with mass spectrometry (СЕ-MS) are also used, each with their own advantages and limitations (Markley et al., 2017). While MS techniques offer higher sensitivity, and with that, increased numbers of detectable compounds, NMR is more suited for analysis of more abundant compounds present. NMR has been demonstrated to be highly reproducible and quantitative, often with reduced sample preparation steps involved (Markley et al., 2017). Wishart (2008) reviewed the topic and highlighted the advantages and disadvantages of NMR, GC-MS and LC-MS spectroscopy for metabolomic analysis. As each has its own level of specificity for metabolite detection, a variety of approaches is advocated when trying to comprehensively characterise the metabolome of a sample.
Saleem et al. (2013) examined the rumen metabolome using a combination of NMR spectroscopy, inductively coupled plasma mass-spectroscopy (ICP-MS), GC-MS, direct flow injection mass spectrometry (DFI-MS) and lipidomics. In total 246 metabolites were identified across the various instruments. Using NMR, Saleem et al. (2013) identified and quantified 50 compounds in rumen fluid, and 98% of all visible peaks were assigned to a compound. Using a GC-MS approach 28 polar metabolites were identified and quantified, 8 of
Figure 1 Illustration of methodology for rumen metabolomic analysis.
these compounds could not be detected by NMR; however, NMR detected 30 compounds that GC-MS could not. With that only 60% of visible GC-MS peaks could be positively identified. In the same study, DFI MS-MS was demonstrated to quantify 116 metabolites. In total DFI MS-MS detected 98 compounds or compound species that were undetectable using GC-MS or NMR. Gas chromatography with time of flight MS has also been successfully used to examine the rumen metabolome, identifying 165 metabolites (Sun et al., 2015). The results from the study strengthen the viewpoint that for a more global view of a sample metabolome a multi-technology analysis approach is beneficial.
A study by de Almeida et al. (2018), using liquid chromatography-high resolution MS examined the impact of sample extraction technique on resulting compound identification; including liquid-liquid extraction, solid phase extraction, original QuEChERS, buffered QuEChERS and an acid base QuEChERS technique. In total the study identified 1882 molecular features, only 3.56% of which had a positive match using the Global Natural Product Social Molecular Networking database. Between methods the liquid-liquid extraction resulted in the greatest abundance of molecular features, extracting compounds with moderate polarity, and nonpolar and hydrophobic characteristics. The authors went on to demonstrate the impact of pH on extraction characteristics. Whereas the solid phase extraction technique extracted primarily low to medium polarity compounds, dependent on the choice of solvent, the QuEChERS method was successful for extraction of compounds of low to medium polarity.
Tools for the interpretation of metabolomic data
The collected data then needs to be analysed and interpreted; a powerful tool for the analysis and interpretation of metabolomic data isMetaboAnalyst(http:// www.metaboanalyst.ca). Handling the most common metabolomics data types including MS and NMR, this comprehensive web application also supports a number of data analysis and data visualisation tasks using a range of univariate and multivariate methods such as PCA (principal component analysis), PLSDA (partial least squares discriminant analysis), heatmap clustering and machinelearning methods as well as tools for interpretation analysis (Xia and Wishart, 2016; Chong et al., 2018). Other tools also include SIMCA for multivariate data analysis, Chenomx, which is an NMR analysis software, works with comprehensive Metabolite Reference Libraries to both identify and measure concentrations of compounds visible in the NMR spectra. Bioconductor (https://www.bioconductor.org/) is an open source software for bioinformatics and a variety of packages are available in R such as 'Metab', which is a package for high-throughput processing of metabolomic data with subsequent data analysis from GC-MS (Aggio et al., 2011). There are a variety of tools for the examination of enriched pathways of different metabolites including the Kyoto
Encyclopaedia of Genes and Genomes (KEGG, https://www.genome.jp/kegg/) and the Small Molecular Pathway Database (SMPDB, http://smpdb.ca/). There are also a number of freely available databases to aid in the identification and interpretation of metabolomics, including, but not restricted to, The Bovine Rumen Metabolome Database (http://www.rumendb.ca), The Bovine Metabolome Database (http://www.cowmetdb.ca), The Milk Composition Database (Foroutan et al., 2019), The Human Metabolome Database (Wishart et al., 2013), The Exposome-Explorer (Neveu et al., 2017), Phenol-Explorer (Rothwell et al., 2013), Food DB (http://foodb.ca/). Small Molecule Pathway Database (Jewison et al., 2014), The Toxic Exposome Database (Wishart et al., 2015), The Yeast Metabolome Database (Jewison et al., 2012), The Human Urine Metabolome (Bouatra et al., 2013), The Human Serum Metabolome (Psychogios et al., 2011) and others.