Follow:

Metabolite measurement of GC MS

https://doi.org/10.1016/j.isci.2022.105738

We collected blood samples (5 mL) of participants who fasted overnight from forearm veins into tubes containing ethylenediaminetetraacetic acid (EDTA; Termo, Tokyo, Japan). We performed sample preparation and GC-MS analysis in the following steps, as described in our previous study.42 The internal standard solution (2-isopropylmalic acid, 0.1 mg/mL in purified water) and extraction solvent (methanol: water: chloroform = 2.5:1:1) were mixed at a ratio of 6:250, and added to 50 μL of each plasma sample. The resulting solution was mixed using a shaker at 1,200 rpm for 30 min at 37°C. After centrifugation at 16,000 × g for 5 min at 4°C, 150 μL of the supernatant was collected and mixed with 140 μL of purified water. The solution was thoroughly mixed and centrifuged at 16,000 ×g for 5 min at 4°C. Finally, 180 μL of the supernatant was collected and lyophilized. The lyophilized sample was dissolved in 80 μL of methoxyamine solution (20 mg/mL in pyridine) and agitated at 1,200 rpm for 30 min at 37°C. We added 40 μL of N-methyl-N-trimethylsilyltrifluoroacetamide solution (GL science, Tokyo, Japan) for trimethylsilyl derivatization, followed by agitation at 1,200 rpm for 30 min at 37°C. After centrifugation at 16,000 × g for 5 min at room temperature, 50 μL of the supernatant was transferred to a glass vial. We performed GC-MS analysis using a GCMS-QP2010 Ultra (Shimadzu Corp.). The derivatized metabolites were separated on a DB-5 column (30 m × 0.25 mm id, film thickness 1.0 mm) (Agilent Technologies, Palo Alto, CA). Helium was used as the carrier gas at a flow rate of 39 cm/s. The inlet temperature was 280°C. The column temperature was first held at 80°C for 2 min, then raised at a rate of 15°C/min to 330°C, and held for 6 min. One microliter of the sample was injected into the GC-MS in split mode (split ratio 1:3). The mass conditions were as follows: electron ionization mode with an ionization voltage of 70 eV, ion source temperature of 200°C, interface temperature of 250°C, full scan mode in the range of m/z 85–500, scan rate: 0.3 s/scan. Data acquisition and peak processing were performed using GCMS solution software version 2.71 (Shimadzu, Kyoto, Japan).

We identified low-molecular-weight metabolites as described previously.42 Chromatographic peaks were identified by comparing their mass spectral patterns to those in the NIST library or Shimadzu GC/MS Metabolite Database Ver. 1. The identification of metabolites was further confirmed through the coincidence of retention indices in samples with those in the corresponding authentic standards. Retention indices were determined and calibrated daily by measuring the n-alkane mixture (C8-40) (Restek, Tokyo, Japan), which was run at the beginning of the batch analysis. We quantified each metabolite peak using the area under the curve and then normalized using an internal standard.

We checked the linearity of the internal standard (IS; 2-isopropylmalic acid) in the concentration range of 0.03 to 300 μg/mL and confirmed a high correlation (Pearson’s r = 0.9997, Figure S52A). Based on the AUC value at the lowest concentration of IS (0.03 μg/mL), for which linearity was confirmed in the experiment described above, we set the detection limit at AUC = 1,000. We have not evaluated concentration dependence. We demonstrated the high correlation of uric acid and glucose concentrations (Pearson’s r = 0.94, 0.94, respectively, Figures S52B and S52C) measured by clinical laboratory test using 4,888 samples, validating the accuracy of our measurement.

We identified 127 metabolites with known chemical structures in 8,270 samples. Among them, four metabolites were excluded because they were detected in water samples, and two were excluded due to the high relative standard deviation (>1). The median call rate of the remaining 121 metabolites was 99.99 (ranged from 48.0 to 100.0) %. Detailed information for each metabolite, including the biochemical name and class based on its chemical structure, is provided in Table S1. We then conducted a principal component (PC) analysis using the 121 quantified metabolite data of 8,270 samples. We removed four samples as outliers (from −10×IQR (interquartile range) below the 25th percentile to 10×IQR above the 75th percentile in one of the top two inferred axes of variation). To normalize measurement variations caused by inter-day instrument tuning differences, the medians of each run were aligned to 1.0 4,43, and the proportion of other values was taken. Normalization effects on the machines were visually confirmed (Figure S53). Moreover, PC was significantly associated with measurement dates and instruments before normalization (PC1: p < 1.1 × 10−16, PC2: p < 1.1 × 10−16, one-way ANOVA) but not after (PC1: p = 1.0, PC2: p = 1.0).

Leave a Comment