An UltiMate 3000 quaternary UHPLC equipped with a refrigerated autosampler (6 °C) and a column heater was used. Both gradients used a Phenomenex Synergy Polar-RP column (150 × 2 mm i.d., 4 µm; 80 Å, Phenomenex, Torrance, CA) at a flow rate of 0.2 mL/min. Solvent A was 5 mM DIPEA and 200 mM HFIP and solvent B was methanol with 5 mM DIPEA 200 mM HFIP. The linear gradient 1 was as follows: 100 % A for 6 min, 98 % A at 8 min, 86 % A at 12 min, 50 % A at 14 min and 10 % A at 15 min. 10 % A was held for 4 min, back to 100 % A over 1 min prior to a 5 min equilibration. Linear gradient 2 was shorter, in order to run the unstable redox cycling metabolites (NAD+, NADH, NADP+ and NADPH) as quickly as possible after the extractions. 100 % A for 2 min, 80 % A at 4 min, 10 % A at 6 min, and 10 % A at 8 min. 10 % A was held for 2 min, back to 100 % A over 1 min prior to a 4 min equilibration. The separations were performed at 55 °C. MS analysis was conducted on a Thermo Scientific TSQ Quantum Ultra AM mass spectrometer (Thermo Scientific, San Jose, CA, USA) equipped with a HESI II source operating in negative mode. The TSQ Quantum operating conditions were as follows: spray voltage 4000 V; vaporizer temperature 200 °C; capillary temperature 350 °C; tube lens 90 V. The sheath gas (nitrogen) and auxiliary gas (nitrogen) pressures were 45 and 10 (arbitrary units), respectively. Both Q1 and Q3 resolutions were set at 0.7 amu. Scan width was 0.002 and the dwell time was 20 msec. Collision-induced dissociation used argon as the collision gas at 1.5 mTorr. The collision energy was optimized for each metabolite and the values are reported in Table 1. Data analysis was done with Xcalibur software 2.6. Statistical analysis was performed in Microsoft Excel or Prism v6 (GraphPad Software Inc. La Jolla, CA).