Metabonomics, as discipline to learn more about metabolic fluxes in complex organisms (be it man or plant) and the implications of their perturbations, has gained significant interest in many research areas ranging from clinical diagnostics, drug development to agrochemistry. Despite this variety, the task analytical tools have to fulfil for metabolic profiling are very similar: 1. analyze complex mixtures of partially known compounds and unknowns in varying concentrations; 2. evaluate, visualize and manage the data. Chromatographic techniques (LC, CE, GC) have been applied ever since to reduce sample complexity. The hyphenation of chromatography to mass spectrometry (MS) and/or nuclear magnetic resonance (NMR) has created a really powerful platform for the analysis of complex mixtures yielding both qualitative and quantitative information.

To determine molecular weight and composition with in the fmol/µL range, the technology of choice is LC-MS with high resolution and accuracy within 3ppm based on API-oa-TOF or FT-ICR mass spectrometry. For characterization of an unknown, mass position, isotopic pattern and characteristic fragmentation in MSn are the key information gained by MS. NMR is an inevitable tool for de-novo-structure elucidation when sufficient material is available - nowadays 20-50 µg using a cryoprobe. Besides data acquisition, it is of major importance to intelligently organize data interpretation and visualization. The first step in data interpretation is the reduction of raw data to peak lists being then evaluated with an appropriate statistical technique for classification (e.g. principal component analysis PCA or genetic algorithms).

We are presenting the combination of LC-API-oa-TOF, quadrupole ion trap and NMR for metabolic profiling. The goals of ongoing works are: the search for potential biomarkers, characterization by LC-MS-NMR, generation of libraries for NMR, LC-MS and MSn data including retention time information and an automated screening.

A set of spiked human urine samples has been created for method development and validation; urine samples of children with various indications are subsequently analyzed using reversed phase liquid chromatography and API-oa-TOF MS (ESI, positive and negative). Peak lists of LC-MS data are generated using a fuzzy-logic based peak finder algorithm developed for deconvolution of co-eluting compounds and subsequent de-isotoping. Molecular formulas are generated using the sigma-fit algorithm matching isotope position intensitiy of the measured and a calculated pattern. Statistical analysis is performed using PCA and the results compared to those of NMR. Potential biomarkers are then subjected to LC-MS-NMR for complete characterization. Urine samples are also analyzed with a quadrupole ion trap in the AutoMSn mode. An MSn library is generated from the sample data as well as standards. This MSn library together with an NMR library creates a unique platform for fast automated screening.