Human Faecal H-1 NMR Metabolomics: Evaluation of Solvent and Sample Processing on Coverage and Reproducibility of Signature Metabolites

摘要

The human faecal metabolome is complex, but rich in information and allows investigation of the host metabolism as a function of diet and health. The faecal metabolome is still much less explored than the plasma and urine metabolome, and in order to generate comparable data across laboratories and cohorts, standard operating procedures are required. This study evaluates 10 protocols, using different extraction solvents and sample processing methods for measuring the human faecal metabolome using proton nuclear magnetic resonance (H-1 NMR) spectroscopy. Three solvents: water, methanol, and dimethyl sulfoxide (DMSO) were investigated at varying concentrations for their ability to extract metabolites directly from faecal slurry or after freeze-drying. The protocols were evaluated on four different pools of human feces. The study also demonstrates a novel signature mapping (SigMa) method for rapid and unbiased processing of complex NMR spectra applied for the first time to human faecal metabolomics. The method is provided with a library containing the chemical shift ranges of 81 common faecal metabolites for future unambiguous and rapid faecal metabolite annotations. The result from the 10 faecal extraction protocols were investigated in terms of reproducibility, coverage, and ability to extract low concentration metabolites. The solvent type was shown to induce the highest variation in the data (45.7%) and the water based extractions allowed detection of the greatest number of metabolites and resulted in the highest reproducibility. Direct extraction of faecal slurry was proved to be more reproducible than freeze-drying. In addition, freeze-drying caused a relative loss of short chain fatty acids (SCFA). DMSO was used for the first time to extract faecal metabolites and enabled the detection of certain bile acids. Some derivatives of SCFA were only detected using methanol as solvent.