Nontargeted Identification of Tracer Incorporation in High-Resolution Mass Spectrometry

Abstract

"Fluxomics" refers to the systematic analysis of metabolic fluxes in a biological system and may uncover novel dynamic properties of metabolism that remain undetected in conventional metabolomic approaches. In labeling experiments, tracer molecules are used to track changes in the isotopologue distribution of metabolites, which allows one to estimate fluxes in the metabolic network. Because unidentified compounds cannot be mapped on pathways, they are often neglected in labeling experiments. However, using recent developments in de novo annotation may allow to harvest the information present in these compounds if they can be identified. Here, we present a novel tool (HiResTEC) to detect tracer incorporation in high-resolution mass spectrometry data sets. The software automatically extracts a comprehensive, nonredundant list of all compounds showing more than 1% tracer incorporation in a nontargeted fashion. We explain and show in an example data set how mass precision and other filter heuristics, calculated on the raw data, can efficiently be used to reduce redundancy and noninformative signals by 95%. Ultimately, this allows to quickly investigate any labeling experiment for a complete set of labeled compounds (here 149) with acceptable false positive rates. We further re-evaluate a published data set from liquid chromatography-electrospray ionization (LC-ESI) to demonstrate broad applicability of our tool and emphasize importance of quality control (QC) tests. HiResTEC is provided as a package in the open source software framework R and is freely available on CRAN.

Evaluation of freely available software tools for untargeted quantification of 13C isotopic enrichment in cellular metabolome from HR-LC/MS data

“…Both ion source gases GS1 (gas 1) and GS2 (gas 2) were set at 40 psi while the curtain gas was set at 35 psi. Untargeted analysis was performed using geoRge (Capellades et al., 2016), DynaMet (Kiefer et al., 2015) and HiResTEC (Hoffmann et al., 2018) while a proprietary software, MultiQuant TM 3.0.1 (SCIEX, Framingham, MA), was used for the targeted analysis. Peak areas for the monoisotopic peaks and the isotopologues were integrated for the respective mean m/z values and m/z tolerance of ±0.05 ​Da and retention time window of ±30 ​s.…”

“…Metabolites were extracted from samples quenched at 0, 5, 10, 20, 30, 60, 120, 300, and 600s time points after the introduction of the 13 C tracer. For this dataset X 13 CMS (Patti and CMS, 2014), geoRge (Capellades et al., 2016) and HiResTEC (Hoffmann et al., 2018) were used as the test software for comparative analysis with the reference software, DynaMet (Kiefer et al., 2015).…”

“…Apart from the raw data files, a complete list of features detected in untargeted analysis using mzMatch-ISO (Chokkathukalam et al., 2013) (reference software for this dataset) is available together with the mass isotopic distribution for a few metabolites provided as supplementary information (Srivastava et al., 2017). This information helped to create a benchmarked list for comparison, when the dataset was processed with the test software, X 13 CMS (Patti and CMS, 2014), geoRge (Capellades et al., 2016) and HiResTEC (Hoffmann et al., 2018).…”

“…However, there is a push towards obtaining labeling data for a larger number of intermediate metabolites (Jaiswal et al., 2018). Fluxomics or 13 C-MFA conducted on a global metabolome in a high throughput manner, would therefore be a valuable tool to complement data available from other -omics studies (Hoffmann et al., 2018).…”

“…However, several freely available software tools have recently been developed to provide automation in detection and quantification of isotopologues of cellular metabolites in an untargeted manner in 13 C labeled datasets. These include X 13 CMS (Patti and CMS, 2014), geoRge (Capellades et al., 2016), DynaMet (Kiefer et al., 2015), mzmatch-ISO (Chokkathukalam et al., 2013), MetExtract II (Bueschl et al., 2017), and HiResTEC (Hoffmann et al., 2018). These tools differ in algorithms used for the key steps of analysis such as peak picking, grouping of potential isotopologues, and statistical evaluation of the data.…”