Identifying novel treeline biomarkers in lake sediments using an untargeted screening approach

Exploring within-ecodistrict lake organic matter variability and identifying possible environmental contaminant biomarkers using sedimentomics

Sedimentomics methods offer insight into the physiological parameters that influence freshwater sediment organic matter (sedOM). To date, most sedimentomics studies characterized variations across large spatial and environmental gradients; here we examine whether sedimentomics methods capture subtle sedOM variations within a relatively homogeneous study area in southwestern Nova Scotia, Canada. Additionally, we explore the lake sedimentome for candidate biomarkers related to ongoing carnivorous animal farming in the region. Sediment cores were recovered from seven lakes across a trophic (oligo- to eu- trophic) and anthropogenic land use gradient (carnivorous animal farming in catchment, downstream of farming, no farming nearby). Subsamples that dated prior to 1910 (pre-carnivorous animal farming) and later than 2010 (during carnivorous animal farming) were analyzed using UHPLC-HRMS in both negative (ESI-) and positive (ESI+) electrospray ionization modes. Cluster analysis (k-means) showed replicate samples from a given lake clustered distinctly from one another in both ESI modes, indicating sedOM captured subtle variations between lake systems. PCA combined with multiple linear regression indicated carnivorous animal farming and OM source explained most of the observed variation in lake sedOM. Principal component analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS-DA) of ESI- and ESI+ data sets identified 103 unique candidate biomarkers. Ten strong candidate biomarkers were identified using graphical methods; more research is required for biomarker verification and molecular characterization. Our results indicate sedimentomics could be used in environmentally homogeneous areas, offering insight into the controls of sedOM cycling. Additionally, we identified prospective biomarkers related to carnivorous animal farming that could be used to understand relative contributions of farming to ongoing eutrophication issues in southwestern Nova Scotia.

Extraction methods for untargeted metabolomics influence enzymatic activity in diverse soils

Soil organic matter (SOM) is the largest carbon pool in terrestrial ecosystems and underpins the health and productivity of soil. Accurate characterization of its chemical composition will improve our understanding of biotic and abiotic processes regulating its stabilization. Our purpose in this study was to estimate the loss of SOM by microbial and exoenzymatic activity that might occur when soil is extracted for analysis of representative low molecular weight mass features using untargeted metabolomics. Two mined clays (kaolinite, montmorillonite) and three diverse soils (varying in texture, specific surface area and cation exchange capacity) were used to assess the extraction efficiency and loss of three enzymatic activity indicators (2,6-dichloroindophenol sodium salt hydrate [DCIP], 4-methylumbelliferyl phosphate [MUBph] and 3,4-dihydroxy-L-phenylalanine [LDOPA]) during extraction with two different solvents (water and methanol). Losses of the indicators were attributed to extraction method (ultrasonication, shaking, or shaking following chloroform fumigation), physical properties associated with the soil/clay type, and microbial activity. Soil/clay type strongly influenced indicator recovery and hence, SOM recovery. Choice of extraction method strongly influenced the composition and recovery of representative SOM mass features, while the choice of solvent determined whether the soil type or extraction method had a greater influence of compositional differences in the SOM mass features extracted. Extraction following chloroform fumigation had the greatest loss of the indicators, due to enzymatic activity and/or adsorption onto the soil matrix. Minimal variation in composition and loss of SOM mass features occurred during extraction by shaking for the soils tested; we therefore recommend it as the method of choice for untargeted SOM extraction studies.

Lipidomics of Environmental Microbial Communities. I: Visualization of Component Distributions Using Untargeted Analysis of High-Resolution Mass Spectrometry Data

Lipids, as one of the main building blocks of cells, can provide valuable information on microorganisms in the environment. Traditionally, gas or liquid chromatography coupled to mass spectrometry (MS) has been used to analyze environmental lipids. The resulting spectra were then processed through individual peak identification and comparison with previously published mass spectra. Here, we present an untargeted analysis of MS¹ spectral data generated by ultra-high-pressure liquid chromatography coupled with high-resolution mass spectrometry of environmental microbial communities. Rather than attempting to relate each mass spectrum to a specific compound, we have treated each mass spectrum as a component, which can be clustered together with other components based on similarity in their abundance depth profiles through the water column. We present this untargeted data visualization method on lipids of suspended particles from the water column of the Black Sea, which included >14,000 components. These components form clusters that correspond with distinct microbial communities driven by the highly stratified water column. The clusters include both known and unknown compounds, predominantly lipids, demonstrating the value of this rapid approach to visualize component distributions and identify novel lipid biomarkers.

A continental scale spatial investigation of lake sediment organic compositions using sedimentomics

Sedimentomics is a new method used to investigate carbon cycling in sediment organic matter. This untargeted method, based on metabolomics workflows, was used to investigate the molecular composition of sediment organic matter across northern Canada (Nunavut and Northwest Territories). Unique "lake districts" were defined using unsupervised clustering based on changes in sediment organic carbon compositions across space. Supervised machine learning analyses were used to compare the "lake districts" to commonly used regional classification systems like the treeline, ecozones, and/or georegions. Treeline was the best model to explain the compositional variance of sediment organic carbon from lakes across Canada, closely followed by the georegions model. A novel sediment metaphenomics analysis was also applied to determine how well environmental constraints explain the variation of sediment organic matter composition across a continent. We determined that sedimentomics is more informative than traditional measurements (such as total organic carbon) and can be integrated with other "omics" techniques.