Simple and Reproducible Derivatization with Benzoyl Chloride: Improvement of Sensitivity for Multiple Lipid Classes in RP-UHPLC/MS

Abstract

The chemical derivatization of multiple lipid classes was developed using benzoyl chloride as a nonhazardous derivatization agent at ambient conditions. The derivatization procedure was optimized with standards for 4 nonpolar and 8 polar lipid classes and measured by reversed-phase ultrahighperformance liquid chromatography−tandem mass spectrometry. The derivatization and nonderivatization approaches were compared on the basis of the calibration curves of 22 internal standards from 12 lipid classes. The new method decreased the limit of detection 9-fold for monoacylglycerols (0.9−1.0 nmol/mL), 6.5-fold for sphingoid base (0.2 nmol/mL), and 3-fold for diacylglycerols (0.9 nmol/mL). The sensitivity expressed by the ratio of calibration slopes was increased 2-to 10-fold for almost all investigated lipid classes and even more than 100-fold for monoacylglycerols. Moreover, the benzoylation reaction produces a more stable derivative of cholesterol in comparison to the easily in-source fragmented nonderivatized form and enabled the detection of fatty acids in a positive ion mode, which does not require polarity switching as for the nonderivatized form. The intralaboratory comparison with an additional operator without previous derivatization experiences shows the simplicity, robustness, and reproducibility. The stability of the derivatives was determined by periodical measurements during a one month period and five freeze/thaw cycles. The fully optimized derivatization method was applied to human plasma, which allows the detection of 169 lipid species from 11 lipid classes using the high confidence level of identification in reversed-phase (RP)-ultra high performance liquid chromatography (UHPLC)/ mass spectrometry (MS). Generally, we detected more lipid species for monoacylglycerols, diacylglycerols, and sphingoid bases in comparison with previously reported papers without the derivatization.

Volume-Corrected Free Energy as a New Criterion for Structural Elucidation in Chemical-Tagging-Based Metabolomics

“…6−8 Chemical labeling blocks some active groups and alters the metabolites' retention behavior besides the well-known improvement in selectivity and sensitivity, allowing the strong regularity of the metabolites' retention times. 9,10 Predicting the retention time of an unknown candidate could reduce the false-positive results to some extent from the available experiment information (r.t., accurate mass values, and MS/MS fragmentation). 1,2,4,11,12 Usually, three major advanced directions predict the retention time.…”

“…Retention time (r.t.) and the mass spectrometric information are considered the golden criteria for structural elucidation of unknown compounds in liquid chromatography–mass spectrometry (LC-MS)-based metabolomics. − Chemical-tagging-based metabolomics uses a derivatization reagent to label the subfunctional group in metabolites of interest. − Chemical labeling blocks some active groups and alters the metabolites’ retention behavior besides the well-known improvement in selectivity and sensitivity, allowing the strong regularity of the metabolites’ retention times. , Predicting the retention time of an unknown candidate could reduce the false-positive results to some extent from the available experiment information (r.t., accurate mass values, and MS/MS fragmentation). ,,,, Usually, three major advanced directions predict the retention time. , Deep-learning- or machine-learning-based retention time predictors are developed for small molecules up to 80 000 and peptides. , Transformed retention time prediction is also popular including normalized retention time, retention index, and retention order prediction. , Quantitative structure–retention relationship (QSRR) modeling is the most popular one using one or multiple molecular descriptors of the metabolite or the solid–liquid interface in LC. ,, QSRR usually utilizes multiple steps, such as the chemical structure construction (i.e., SMILES, mol, gjf, and xzy), molecular descriptor calculation (i.e., DRAGON, CODESSA, and RDkit), QSRR modeling establishment, and real situation application. ,− For example, Ovčačíková et al predicted the retention behavior of lipids by polynomial regression with the double bond number and carbon atom number . Some multilevel QSRR models are also developed using analyte-specific and chromatographically specific descriptors (log P, p K a , and functional groups). ,, …”