Comparative analytical work-up for reproducible metabolite derivatisation for gas chromatography-mass spectrometry

Abstract

Gas chromatography-mass spectrometry (GC-MS) is one of the most popular analytical instruments used for metabolomics analyses. However, chemical derivatisation of the sample is generally required to increase the volatility of non-volatile and semi-volatile metabolites prior to GC-MS analyses. Various methods of chemical derivatisation exist, but only a few, well standardised techniques such as silylation, acylation, and alkylation are routinely employed. The method of choice is highly dependent on the nature of compounds analysed making it important to familiarise oneself with the different derivatisation methods to improve or develop analytical methods suitable for the samples at hand. The current investigation sets out to compare five different derivatisation techniques, utilising the reagents N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA), BSTFA with methoxyamine hydrochloride, N-methyl-(trimethylsilyl)trifluoroacetamide (MSTFA), MSTFA with methoxyamine hydrochloride, and methyl chloroformate (MCF) on a set of key metabolites prior to GC-MS analyses. The samples of interest included a standard solution of 19 selected compounds and four internal standards (2-amino-3-bromo-5-methylbenzoic acid, 3phenylbutyric acid, norleucine and nonadecanoic acid methyl ester), as well as biological quality control serum samples. In both instances, the samples were freeze-dried, derivatised and analysed using a GC-MS approach. The data obtained were processed followed by statistical comparisons. Data obtained from analyses of the standard stock solution showed clear separation when subjected to principal component analyses, distinguishing between silylated and alkylated samples, suggesting different profiles for the same set of compounds. In terms of compound response and repeatability of each derivatisation method, MCF derivatisation resulted in the lowest average relative standard deviation (RSD) values for the majority of the analysed compounds. Additional comparisons showed that the use of silylation (BSTFA and MSTFA), without the methoximation step, resulted in the same number of compounds detected, the formation of multiple derivatives for the same compound and displayed similar repeatability amongst tested methods. Likewise, the use of methoximation prior to silylation showed no differences when using BSTFA or MSTFA. When alkylation (MCF) was used as a derivatisation method, mostly a single derivative was produced per compound, suggesting that alkylation prevents the formation of multiple derivates per compound. Stability of derivatives produced per derivatisation method over time, showed similar metabolite profiles when using MSTFA with methoxyamine hydrochloride solution (MeOX) or MCF derivatisation methods. Analyses of the same batch of samples 84 hours post-preparation confirmed that the derivatisation method influenced the relative abundance of the compounds of interest, highlighting the need for method optimisation for the samples at hand. Analyses of two sets (groups A and B) of biological samples were used to evaluate and compare the derivatisation methods as proof of concept for metabolomics studies. In this comparison, BSTFA with MeOX was identified as superior compared to the other methods, as this method resulted in the highest number of statistically significant features detected between groups A and B. This study confirms that no universal derivatising agent can produce satisfactory results for every compound in a sample, independent of its chemical class. Derivatisation reagents should be standardised for the compounds of interest within a specific metabolomics study, to find the method that ensures repeatability for the majority of the compounds present in a sample, while also considering compounds that cannot be reliably detected and using the outcome of that with caution.