Hydroxy and keto acids are important intermediates in central carbon metabolism and redox regulation, yet their comprehensive quantitative analysis remains challenging because of high polarity, poor chromatographic retention, and substantial concentration variation in biological matrices. Here, we develop an orthogonal chemical encoding strategy coupled with LC-MS/MS for simultaneous and quantitative profiling of hydroxy and keto acids. Keto acids are selectively modified at carbonyl groups using O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine (PFBHA), followed by carboxyl-directed derivatization of both keto and hydroxy acids with 3-nitrophenylhydrazine (3-NPH). This sequential labeling approach introduces complementary chemical signatures for distinct metabolite classes, improving keto acid detectability while preserving broad coverage of hydroxy acids. The analytical performance of the method was systematically evaluated, demonstrating wide linear dynamic ranges, subnanomolar detection limits, high precision, and minimal matrix effects for 20 hydroxy and keto acids. Furthermore, isotope-coded derivatization using 13C6-3-NPH enabled accurate isotope-based internal standardization without requiring individual isotope-labeled standards for each analyte. The applicability of the method was demonstrated by profiling serum samples from diabetic mice, revealing altered ketone body metabolism, branched-chain amino acid-derived keto acid pathways, and hydroxy/keto acid interconversion. This orthogonal chemical encoding approach provides a robust analytical strategy for quantitative characterization of structurally related polar metabolites in complex biological matrices.
Chromatographic retention provides experimentally measured information on molecular interaction propensity, but the extent to which different biomimetic stationary phases provide distinct rather than redundant information remains incompletely defined. We used pesticide soil sorption as an external partitioning system to compare immobilized artificial membrane (IAM), human serum albumin (HSA), alpha-1-acid glycoprotein (AGP), and reversed-phase chromatographic hydrophobicity index-derived log P (CHI log P) descriptors across 96 pesticides and four independently treated sorption endpoints (log Kd, log Koc, log Kf, and log Kfoc). The experimental chromatographic descriptors were benchmarked against nine calculated log P estimates and conventional physicochemical descriptors. Predictive performance and robustness were assessed using repeated and nested cross-validation together with complementary validation and sensitivity analyses, while a forced paired augmentation analysis tested whether each experimental descriptor added information beyond conventional calculated log P on identical cross-validation splits. The chromatographic descriptors shared information but showed distinguishable endpoint-dependent behavior. Across four conventional log P baselines non-collinear with all experimental descriptors, AGP gave the largest median incremental gain for log Kd and log Kfoc, whereas IAM gave the largest gain for log Koc and log Kf. The broader model-selection framework favored simple single-descriptor consensus equations: IAM for log Koc and AGP for log Kd, log Kf, and log Kfoc. Together, the paired and nested analyses support complementary, phase-specific chromatographic information within the investigated pesticide set. Absolute predictive performance was moderate; transfer to independent compounds remains to be assessed.
