Nmr suite 8

Metabolite identification and quantification was carried out using Chenomx NMR Suite 8.12 software (Chenomx Inc., Edmonton, Canada), using the internal reference library (Version 10), and with support of published data for other animals23 (link)24 (link)25 (link).

Each frozen saliva sample was thawed at room temperature and centrifuged at 15,000× g for 10 min at 4 °C to remove eukaryotic and prokaryotic cells, cellular debris, and mucins, according to Gardner et al. [40 (link)]. The supernatants were protein-depleted by ultra-filtration, using Amicon Ultra-4 Centrifugal filters (3000 MWCO, Merck Millipore) at 4000× g for 120 min at 10 °C, and lyophilized.
For ¹H-NMR measurements, each of the lyophilized samples was suspended in potassium phosphate buffer (50 mM, pH 7.4) and 3-trimethylsilyl propanoic acid (TSP) was added as the chemical shift reference (0.00 ppm) and quantitative internal standard.
High-resolution one-dimensional (1D) ¹H-NMR spectra acquisition and processing were carried out according to Pertinhez et al. [41 (link)]. Metabolites identification and quantification were carried out using Chenomx NMR Suite 8.3 software (Chenomx Inc., Edmonton, AL, Canada).
Heatmap analysis was carried out on targeted metabolites, with concentrations higher than 5 μM at least for one saliva subtype. Heatmaps were generated using MetaboAnalystR (https://www.metaboanalyst.ca) [42 (link)], with normalization referenced to TSP and autoscaling.

Urine samples were analyzed using a Bruker AVANCE III 600 MHz 1H NMR spectrometer (Bruker; Rheinstetten, Germany), operating at 600.13 MHz at a temperature of 300° K using a standard NMR pulse sequence (relaxation delay-90°-t-90°-t_m (mixing time)-90°-acquisition) to acquire one-dimensional spectral data. The parameters used are described elsewhere [39 (link)].
¹H NMR spectral data were automatically pre-processed (phasing, baseline correction and calibration to TSP) in Topspin 3.6.0. The spectral data were then imported into MATLAB (version R2018a) for statistical analysis. The regions containing water (δ1H 4.70–4.90), urea (δ1H 5.48–6.28) and TSP (δ1H -0.20–0.20) were removed. Alignment was applied using recursive segment-wise peak alignment method [40 (link)] and aligned data were normalized using the probabilistic quotient method prior to multivariate data analyses. Principal component analysis (PCA) and orthogonal projection to latent structures−discriminant analysis (OPLS-DA) were applied to compare ¹H NMR spectral data between the different groups. Metabolites that contributed to group discrimination were identified based on a previously published study [41 (link), 42 (link)] and confirmed with STOCSY, Chenomx NMR suite 8.3 software (Chenomx Inc. Edmonton, Alberta, Canada) and The Human Metabolome Database (HMDB, http://www.hmdb.ca/).

The resonances of metabolites were assigned by Chenomx NMR Suite 8.3 software (Chenomx Inc, Edmonton, Canada) combined with the Human Metabolome Database and references [59 (link),60 (link)]. Two-dimensional ¹H-¹³C HSQC spectra were required for the confirmation of resonance assignments.

¹H-NMR analysis has been described in detail previously [31 (link)]. In short, spectra were recorded at 800 MHz with a Bruker Advance III HD spectrometer with a 3-mm TCI cryoprobe. NMR data were recoded using the Bruker pulse sequence “zgespe”. A total of 128 scans were collected into 64 k data points. Data processing was performed with TopSpin 3.2p16 (Bruker BioSpin) and MatLab (MathWorks Inc., Natick, MA, USA), using TSP-d4 for referencing. In total 237 peaks were manually aligned and integrated peak-by-peak, and these variables represent ∼70 metabolites. A variable could also reflect more than one metabolite. Only variables of interest were identified.
For annotation Chenomx NMR suite 8.31 (Chenomx Inc., Edmonton, AB, Canada), the Human Metabolome Database [37 (link)] and an in-house implementation of the statistical total correlation spectroscopy (STOCSY) routine [38 (link)] were used.

Tissues samples were flash-frozen in liquid nitrogen and stored at −80°C until analysis. In each patient, approximately 9-mm³ CT and PT was resected and 200-µL serum was used for the metabolomics analysis. Serum and tissue sample preparation was conducted as described previously [27 (link),28 (link)]. Tissuelyser-24 was used for homogenization at 60 Hz for 180 s (Lixin Co., Ltd, Shanghai, China). NMR measurements for ¹H NMR metabolic profiling and analyses were performed as described and using a Bruker Avance III HD 600-MHz NMR spectrometer equipped with a TXI probe head [28 (link),29 (link)]. Chenomx NMR suite 8.4 and reference compounds were used to identify the metabolites in the serum and tissue during the analysis. A Receiver Operating Characteristic (ROC) analysis was done in MetaboAnalyst 5.0 to evaluate the specificity and sensitivity of each metabolite. GraphPad Prism 5.01 (GraphPad Software, La Jolla, CA, USA) was employed to perform a univariate statistical analysis where the data are represented as the mean +/− standard deviation (SD).