2.2. Fecal Metabolomics Analysis

The preprocessed pipeline of fecal samples was established. First, the samples were mixed with methanol (3 mL/g) and homogenized (8 m/s, 15 s) with ceramic beads (1 mm) (Omni International, Bedford, NH, USA) [22]. Then, the supernatants after centrifugation were filtered using syringe filters (0.22 μm, Millipore Corp, Billerica, MA, USA). To monitor potential contamination of sample processing, a procedural blank sample of water (Thermo Fisher Scientific, Waltham, MA, USA) was used. Quality control (QC) samples were prepared by mixing equal volumes of all fecal extraction, and processed with the real samples simultaneously.

The metabolomics profiling was performed using a Dionex UltiMate 3000 RS ultraperformance liquid chromatography (UPLC) system coupled with Orbitrap Q Exactive mass spectrometry (MS) with a Hypersil Gold C-18 column (2.1 × 100 mm, 1.9 µm) (Thermo Fisher Scientific, Waltham, MA, USA). Both ESI+ and ESI– ionization modes were executed with a m/z range 70–1050 and a full MS scan at 60,000 resolution.

Raw data were acquired by Xcalibur 4.1 software and processed by Compound Discoverer 3.1 software (Thermo Fisher Scientific, Waltham, MA, USA) for peak detection, peak alignment, and peak integration. The processing procedures were modified from a previous method [23]. The multivariate statistical analysis including principle component analysis (PCA) and partial least-squares-latent structure discriminate analysis (PLS-DA) were performed in SIMCA-P 13.0 (Umetrics AB, Umea, Vasterbotten, Sweden) [24].

The biologically metabolites were picked out using Compound Discoverer 3.1. Briefly, the local library and online database including ChemSpider and mzCloud, of which the Human Metabolome Database (HMDB), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Biocyc were used. The identity of the candidate biomarkers were confirmed by the standard compound libraries of both our lab and the public database.