Adam Rosebrock (Presenter)
Department of Pathology, Stony Brook School of Medicine
Bio: Dr. Adam Rosebrock is an assistant professor in the Department of Pathology at Stony Brook School of Medicine and the Stony Brook University Cancer Center. He has had longstanding interest in using “big data” to address fundamental biological questions. The focus of his research is on understanding the regulation of gene expression and biochemical activities that underlie cell division, growth, and survival across diverse external states. The Rosebrock lab actively develops new experimental and analytical methods and builds genetic, hardware, and computational tools to enable high-throughput and high-content biology, with particular emphasis on quantitative mass-spectrometry metabolomics.
Authorship: Olga Zaslaver (2), Amy A. Caudy (2), Adam P. Rosebrock (1)
(1) Department of Pathology, Stony Brook School of Medicine, Stony Brook NY, (2) Department of Molecular Genetics, University of Toronto, Toronto, ON Canada
| Short Abstract Full-scan LC-MS analysis is gaining widespread use in metabolomic applications, but is often viewed as secondary platform for routine quantitation in favour of inherently targeted triple-quadrupole approaches. We have developed a workflow using a biologically-derived, isotopically labeled reference and an integrated software platform for robust quantitation on time-of-flight instruments. In addition to quantitation of known compounds, our approach enables simultaneous data mining for discovery and analysis of new mass spectral features. We have applied our tools to analysis of central carbon metabolites and other organic acids from patient derived cell lines and biofluids in the context of large, multi-site glioblastoma multiforme project. In addition to greatly improved quantitation and reproducibility, our approach has enabled identification of previously unknown metabolites. |
Long Abstract
Introduction
Quantitative LC-MS holds the promise of enabling direct measurement of metabolites and metabolic fluxes in biological samples. As basic science discoveries uncover new reactions and reactants, a critical early step is the identification of biological contexts in which these reactions occur. As part of a large multi-site project analyzing patient-derived xenograft and cell lines from glioblastoma multiforme tumours, we have been tasked with measurement of a number of central carbon intermediates, including the TCA-cycle intermediate alpha-ketoglutarate and the oncometabolite D-2-hydroxyglutarate, in a range of sample matrices. Using a biologically-derived, isotopically labeled reference approach, we are able to reproducibly identify a range of organic acids and central carbon metabolites at endogenous levels in a matrix-insensitive fashion. In addition to quantitation of known metabolites, we have an active interest in identification of new compounds as potential biomarkers and to provide mechanistic insight into tumour biology. Our LC-MS analytical workflow thus incorporates both targeted quantitation and novel feature identification from a single software package that has been purpose-built for TOF/qTOF metabolomic data analysis.
Methods
Polar metabolites from cultured cells, tissue sections, and biological fluids were extracted as previously described. Briefly, a -20C ternary mixture of acetonitrile, methanol, and water was added to cells, frozen-pulverized tissue, or biofluids to quench enzyme activity and precipitate proteins. Extraction liquor was clarified and dried down under nitrogen prior to resuspension in buffered aqueous media. Where relevant, samples were reconstituted with a 2x concentrated biological reference material prior to final filtration through a 0.45µm hydrophilic polypropylene membrane before loading into autosampler vials. Reconstituted samples were stored at -80C prior to analysis.
Aliquots of each sample were injected onto an aqueous-compatible reverse phase UPLC column (HSS-T3, Waters), separated using a low pH aqueous solvent system: acetonitrile, and analyzed by TOF or qTOF mass spectrometry in MS1 mode.
For known compound analysis, mass spectral features (a unique coordinate of accurate mass:charge in retention time window) were extracted from raw, profile-mode data files using the ChromXtractorPRO toolkit. Where applicable, mass spectral features for 12C14N biological features and their cognate 13C15N internal controls were extracted in parallel. Extracted features were locally aligned and consensus-bound integrated prior to statistical analysis.
For unbiased feature discovery, samples were analyzed using MassHunter Profinder (Agilent Technologies) in a recursive approach. Mass spectral features were exported for subsequent analysis, as above, in ChromXtractorPRO. Statistical analysis was performed on integrated signal intensities or reference-corrected ratios of intensities using R/Bioconductor.
Preparative LC was performed by scale-up of the UHPLC-MS method for analysis. Fractions were collected from separation using a semi-preparative reverse-phase column (Atlantis T3, Waters) and analyzed by UHPLC-MS as described above. Fractions containing the metabolite of interest were pooled and analyzed further as described.
13C flux experiments were performed by rapid switch of cells from labeled to unlabeled nutrient (or vice versa) one hour after a pre-experiment media change. At the time-points indicated, metabolites were extracted as described and analyzed by UHPLC-MS. ChromXtractorPRO was used to extract the entire isotopologue group for each analyte and perform consensus-bound integration.
Results
Central carbon metabolites, including the oncometabolite 2HG, were routinely visible in cells and biological fluids analyzed. Hierarchical and k-means clustering of reference-corrected intensities revealed strong similarities in subsets of patient samples and stratified by site of origin, clinical stage, and in vitro response to therapeutic challenge. Samples were analyzed in parallel with and without internal reference. Using our biological reference design, change of less than 1.5x could be routinely and reproducibly measured. This consistency extended to re-analysis of multiple samplings of the same biological material, repeat injections of a single sample, and analysis across instrument platforms.
In addition to identifying new metabolic signatures associated with individual patient samples, we have identified a previously uncharacterized organic acid accumulating in a subset of samples and also present in fetal tissue. We have begun to characterize this compound using our previously-described approach including isotopic labeling, MS fragment analysis, and 1D NMR spectroscopy.
We have also leveraged our data-mining tools to identify the subset of data files that reflect a significant accumulation of this compound from our internal data archive. The speed and batch-ready nature of our analytical software tools enable rapid analysis of tens-of-thousands of archived data files.
Conclusions
LC-MS quantitation of a range of central carbon metabolites can be performed by a robust, rapid full-scan analysis using TOF and qTOF instrumentation. For routine applications, the vast majority of the full-scan data are collected and ignored during subsequent analysis, a process that is streamlined by tools designed specifically for this "targeted full-scan" approach. The presence of full-scan data in archive storage provides a powerful tool for re-analysis and data mining of existing files for newly discovered compounds. In stark contrast to QqQ-based methods, full scan approaches provide a combination of analytical performance for known compounds, the potential to troubleshoot poor sample preparation or instrument performance issues, and an unparalleled ability to engage in post-acquisition data mining. Despite a strong foothold of triple-quadrupole instruments in the research-facing clinical laboratory, advances in the workflows for full-scan mass spectrometry make a compelling case for rapid and widespread deployment of this complementary instrument platform for current and future metabolomics applications.
References & Acknowledgements:
| Description | Y/N | Source |
| Grants | no | |
| Salary | no | |
| Board Member | no | |
| Stock | no | |
| Expenses | yes | Agilent Technologies |
IP Royalty: no
| Planning to mention or discuss specific products or technology of the company(ies) listed above: | no |