Figure 1. Metabolomics workflow.

(a) Profiling. Mice were infected intraperitoneally with LCMV Armstrong, and plasma was collected at days 1, 3, 7 and 14. Metabolites from plasma were extracted with methanol and applied to a capillary reversed-phase column run with a gradient. Electrospray ionization in positive mode was used, with TOF data collected from m/z 100 to 1,000. Each sample produced a three dimensional data set. Automatic peak finding, followed by nonlinear alignment in the time domain and peak integration, was performed using the XCMS program. A feature table was produced consisting of m/z, retention time, and integrated intensity for each sample. An extracted ion chromatogram was automatically generated for each feature. (b) Compound identification. The compound identification workflow begins (left) with the routine identification of compounds using a QTOF. The LC conditions used in profiling are replicated, and ions of interest (eg those found to be significantly changing over time) For more difficult compounds, Fourier Transform Ion Cyclotron Resonance Mass Spectrometer (FT-ICR) can be used for identification (right). Three aspects of FT-ICR are important: high mass accuracy, high resolution, and MS/MS. When there is no match of the ion to one of the metabolite databases, it is necessary to determine the identity of the compound de novo, the starting point for which is determination of the molecular formula. For this purpose, having high resolution and mass accuracy is essential, and both can be achieved using an FT-ICR instrument. Using LC in combination with a 7 Tesla FT-ICR with internal calibration coming from a dual electrospray configuration, it was possible to obtain a mass accuracy of less than 1 ppm. This accuracy often allows for the calculation of a unique molecular formula, which can substantially narrow the task of compound identification (Table III). This molecular formula can then be used to search chemical databases, such as PUBCHEM. The FT-ICR resolving power was just over 110,000 for the largest metabolites, and exceeded 200,000 for the smaller molecules. The high resolution allows for the measurement of isotopic fine structure (Figure 2), which when compared with a simulated mass spectrum, can be used to confirm the elemental composition that was obtained from the accurate mass measurement, and eliminate candidates if there is more than one possible molecular formula.