Chemical Science Facility Core [CSFC]

GERG has developed, validated, and applied methods for analyses of known contaminants, including PCBs, PAHs, dioxins/furans, metals, PCE, slightly volatile organic compounds (SVOCs), and VOCs. GERG analyzes thousands of samples each year for these contaminants. Standardized methods for extraction and processing of environmental samples are available for several environmental contaminants of concern. 

Samples will be extracted with an Accelerated Solvent Extractor (ASE) after adding surrogate compounds used to quantitate components in the sample. The extracts will then be concentrated by evaporating solvent to a smaller volume and for sediments treated with copper to remove sulfur. Next, chromatography (silica/alumina, size exclusion, etc.) will be used to purify the extracts before analyses by high-resolution GC/high-resolution MS. Aliquots of component extracts will be split to provide samples for targeted GC-MS and LC-MS analysis.

All samples will be logged in and assigned a unique file number. Samples will be analyzed in batches of 20 samples or less. For each batch of samples, a procedural blank, duplicate matrix spike, and when available, a standard reference material (e.g., National Institute of Technology NIST SRM 1944) will be included. This will provide quality control for the analytical method and validate data acquisition.

For metal analyses, the same quality assurance samples will be analyzed with each sample batch. Samples for metal analyses will be acid digested followed by cold vapor or Inductively Coupled Plasma MS analyses. Routine analyses will include 57 PAHs, 209 PCBs, the 17 toxic dioxins/furans, 122 SVOCs, 70 VOCs, 38 pesticides, and 10 metals. These analytes are routinely measured at GERG in many diverse matrices.

Methods will also be developed for “newly discovered” environmental contaminants as needed by Center investigators. Contaminants currently measured and their detection limits are presented in Table 2. This panel will be expanded as new investigator projects are initiated.

Depending on the need for identification of new contaminants or groups of contaminants om Center investigator projects, new analytical techniques will be developed for routine analyses. GERG’s extensive experience in developing new analytical methods, for example, for the analyses of PAH metabolites, tributyltin (TBT), planar PCB, and PBDE, will ensure responsiveness to project needs for new methods. We will follow an established and successful approach to the development of new analytical techniques for emerging contaminants of concern.³ Physical properties will guide method development and choice of isolation, separation, concentration, and analytical techniques (e.g., GC/MS vs. high-performance LC/MS).

For studies that are in inception stages, CSFC staff will work with the investigators to provide input on statistical design of experiments required for achieving significance in metabolomics readouts. In addition, we will consult with the investigators to help design experiments and incorporate appropriate controls to meet our standard operating procedure requirements. This study design consultation will also help investigators choose the appropriate analytical procedure for the scientific question being asked. For example, short chain fatty acids (SCFAs) are more amenable to GC/MS/MS than LC/MS/MS, whereas other metabolites like bile acids may be analyzed using either method. An important aspect of this stage would be to determine the appropriate analytical method (e.g., hydrophobic or hydrophilic column) and solvents (methanol/chloroform or methanol/water) that should be used for the type of sample and abundance of the metabolites being tested. As required, the DSFC will be involved to account for statistical considerations.    

Two types of metabolomics analysis will be offered through the CSFC: targeted and untargeted.

A discovery workflow for global untargeted metabolomic analysis will be available on an ultra-high performance liquid chromatography electrospray ionization MS platform using a QExactive Plus instrument (Thermo). This workflow can be customized based on whether the user is primarily interested in hydrophilic (e.g., amino acids, amines, cationic metabolites) and/or hydrophobic (e.g., some bile acids, free fatty acids) metabolites by using HILIC (hydrophilic) and/or C18 reverse phase (hydrophobic) columns. This workflow will also leverage several key features of the mass spectrometer, including high-resolution accurate mass (HRAM), and fast acquisition time, to perform data-independent acquisition (DIA) experiments. Initially, a full scan (m/z 67 – 1000) will be carried out at resolution of 70,000, followed by data-dependent MS/MS scans for metabolite identification.

If a Center member wants to utilize a method or workflow that is currently established in the facility and has a small number of samples, the CSFC staff will carry out the sample processing, data acquisition, and routine bioinformatics analysis. If the member wants to analyze a class of molecules not previously studied in the Core, a new method that is specific to the user’s project will be developed by the technical staff. In this mode, it will also be possible for students and postdoctoral fellows to gain hands-on experience after suitable training.

A targeted metabolomics platform on triple quadrupole mass spectrometers (Altis & Quantiva, Thermo) will be used to quantify specific classes of metabolites for Center members. For example, IMAC has developed optimal methods and specific transitions for simultaneous quantification of a panel of ~35 tryptophan-derived microbiota metabolites (e.g., indole, indole-3-aldehyde, tryptamine) using multiple reaction monitoring (MRM). This and other such workflows will be made available to Center members. Metabolite identification will be based on retention time of pure standards and parent/fragment ion mass spectra in databases. As additional metabolites are identified by Center members and become commercially available, they will be added to the methods database. GC/MS/MS workflows will also be provided to identify metabolites such as SCFAs and medium chain fatty acids (MCFAs) that are more amenable to GC-MS analysis than LC-MS. Metabolite extraction and GC-MS analysis will be carried using standard derivatization protocols on a TSQ EVO8000 instrument.

The CSFC will also provide in silico analysis tools for metabolomics. For example, the identification of intermediates and metabolites that can be produced in the body from an environmental toxicant is extremely important for understanding the effects of the toxicant on human health. However, it is not straightforward to identify such metabolites as they can be at very low abundance or be consumed as intermediates in other host or microbiota reactions. Dr. Jayaraman’s group has recently developed a probabilistic prediction algorithm that addresses this problem. The model was used to predict that 58 metabolites could be produced from aromatic amino acids exclusively by the microbiota, and model predictions were validated using conventionally raised and germ-free mice and LC/MS/MS metabolomics. This algorithm and other custom modifications (e.g., metabolites arising from microbiota‒host co-metabolism) will be made available to Center members through the CSFC.

Raw data from untargeted LC/MS projects processed for background subtraction, peak alignment, and deconvolution using the Progenesis QI (Waters) software. Putative compound identification will be carried out using two independent methods. First, mass spectra will be compared against four publicly available metabolite databases – HMDB, KEGG, and NIST. To account for discrepancies in metabolite identification from multiple databases, a positive match in at least two databases will be required for putative identification. If a metabolite is identified only in a single database, secondary in silico validation will be carried out using MetFrag⁴⁴. As an independent measure, putative chemical identities will be assigned to the features based on accurate mass (m/z) and MS/MS data using a recently developed automated annotation tool and implemented in our laboratory. Statistical analysis including normalization, biomarker identification between experimental groups, and pathway analysis will be conducted using Metaboanalyst. As information on additional metabolites with association to exposures become available in the literature, we will use data-independent acquisition to mine the full scan data and identify additional metabolites. For GC/MS data, automated mass spectral deconvolution and identification software (NIST) will be used to process chromatograms for background subtraction and peak integration.⁴⁵ The sample mass spectra will be compared to the NIST library of standards to identify metabolites.⁴⁶ All bioinformatics software will be stored in the cloud with multiple user licenses so that users can access data and carry out analyses from remote locations without blocking access to instruments. 

Statistical analysis of the results will be customized per the needs of the investigator in collaboration with the DSFC using multivariate data analysis and ordination methods.^47-49 The data from untargeted mass spectrometry is usually complex as well as difficult to interpret and the use of multivariate statistical analysis will aid in reducing data dimensions, differentiating similar spectra, and building predictive models.^50,51 Our data analysis workflow will start by looking at the overall structuring of the data using ordination methods like Principal Component analysis and then use linear/non-linear models to identify key markers that are important in the dataset.⁴⁹ 

In collaboration with the DSFC, we also propose to build integrative models utilizing different omics datasets. For example, microbiome composition and metabolome data from the same experiment can be correlated to identify the specific bacterial genera or species responsible for the observed changes. Center members Drs. Chapkin and Ivanov have pioneered such systems biology methods for integrating host gene expression and the gut microbiota metagenome in human subjects. We propose to make this capability available to Center investigators so that different combinations of omics data from the host and microbiota can be integrated to develop novel predictive computational models of exposure. By including the multivariate nature of metagenome, metabolome, and transcriptome data sets, richer information content will be generated compared to analyses focusing on a single data set only (e.g., only metagenomics data) or single variable testing (e.g., gene by gene differential expression).