Aldehyde quantification by ultra-performance liquid chromatography tandem mass spectrometry

Quantification of aldehydes from tissue samples used the dinitrophenylhydrazine method⁴⁵ with numerous updates. Fastidious measures were taken to minimize environmental contamination, including working in clean air environment, extracting DNPH (0.2 M in 70% ethanol in phosphoric acid, Sigma) four times with hexane (1:20 v/v) immediately prior to use, scrupulously clean materials which were baked at 70 °C overnight, and fresh bottles of UPLC-grade solvents were used for every run. Blank samples involving every element of the sample preparation process were regularly tested to ensure background was controlled and that carryover did not occur. Alternating low and high calibration points were injected regularly to ensure consistent instrument response.

Forty-two aldehydes and ketones were selected for study (as indicated in Supplementary Data ¹). The rationale for analyte selection was any of: (i) known genotoxin, (ii) physiological metabolite, (iii) present in exhaled breath, (iv) a specific target for ALDH isoenzymes of interest. Analytical variation was controlled with six isotope-labeled standards (ISTDs)—C2:0-d₄ (Sigma, for short-chain alkanals), C6:0-d₁₂ (Sigma, for medium-chain alkanals), C16:0-d₅ (Santa Cruz, for fatty alkanals), HNE-d₃ (Cambridge Bioscience, for dienals), ONE-d₃ (Cambridge Bioscience, for dialdehydes), and MDA-d₂ (prepared by diluting 10 μL 1,1,3,3-tetraethoxypropane-1,3-d₂ (Santa Cruz) in 10 mL of 0.1 M HCl and hydrolyzing at 100 °C for 10 min)⁴⁶).

For tissue, three endoscopic biopsies (9–12 mg) were homogenized under liquid nitrogen in a clean pestle and mortar and weighed. To this 180 μL of dry-ice-cold UPLC acetonitrile/water (50:50 v/v) containing 62.5 ng/mL internal standards was added with a baked ceramic bead. The slurry was extracted in a Reitsch oscillator for 60 s at −40 °C, and protein cleared by centrifugation (20,000g, 3 min). The supernatant was derivatized with 40 μL DNPH; the pellet re-extracted with the same volume of extraction solvent. The two extracts were combined and derivatized for 1 h at 25 °C. The phases were separated by adding 20 mg of baked sodium chloride, followed by centrifugation at 20,000g. The organic phase was transferred to a clean glass UPLC vial and sealed with Teflon. In vitro aldehydes were assayed by mixing 180 μL of media with 180 μL acetonitrile containing 62.5 ng/μL ISTDs, which was then centrifuged, and derivatized as above. Protein content was measured per well by the bicinchoninic acid method (Sigma). Sample preparation for each experiment was conducted in a continuous run on a single day.

This was undertaken on a Acquity UPLC and Waters TQS MS/MS system. A C18 Cortecs column (particle size 1.6 μm, internal pore 2.1 μm, Waters) was used, with UPLC-grade water (A) and UPLC-grade acetonitrile (B) (both Sigma) mobile phases. Column temperature was 40 °C, sample temperature 4 °C, and the flow rate 0.5 mL/min. Injection volume was 5 μL, under initial conditions of 30% B. The ratio was changed as follows: 0.20–8 min to 40%, then 8–16 min to 95%, 19–19.1 min to 30% and held until 20 min. The source settings were as follows: source temperature 150 °C, capillary voltage 2.5 kV, cone voltage 10 V, cone gas flow rate 200 L/h, desolvation gas temperature 400 °C, desolvation gas flow rate 650 L/h. A scheduled multi-reaction monitoring MS/MS method was established (Supplementary Data ²). Each sample was prepared in 2–3 biological replicates depending on availability, with the final result being averaged (one result per tissue for each patient in the final analysis).

Acquired peaks were integrated using Targetlynx (Waters, SCN855) and manually checked. For quantitative analysis, internal standard calibration curves were calculated by dividing a serially diluted mix of all unlabeled standards by the respective ISTD (concentration 50 ng/μL, see Supplementary Data ¹). The intensities of the unknown concentrations were also divided by the respective ISTDs, and concentrations calculated by comparing to the ISTD calibration. Tissue aldehydes were normalized to sample mass (milligrams). In vitro aldehydes were normalized to protein concentration in a parallel well as quantified by the bicinchoninic acid method. Isomer/isobar ketones, alkanals, and dialdehydes (e.g. acetone, propanal, glyoxal, molecular weight = 58) were unambiguously determined on the basis of retention time, parent ion mass-to-charge (e.g. dialdehydes reacting to form a di-DNPH hydrazine) and characteristic abundant fragment ions (e.g. 163, 152, 182 for alkanals, ketones, and dialdehydes, respectively). A lower limit of quantification (LLOQ) of 20 pg/mg was set, given a minimum tissue input of 10 mg, an extraction volume of 0.2 mL, and that 1 ng/mL was well above the limit of detection for all targets. The manually checked integrations were exported to R to reformat the matrix (RStudio version 1.1.456, RStudio, Inc., code provided in Supplementary Data ⁷) and complete the processing as above. All data processing was blinded.