The use of collision cross-section (CCS) values obtained by ion mobility high-resolution mass spectrometry has added a third dimension (alongside retention time and exact mass) to aid in the identification of compounds. However, its utility is limited by the number of experimental CCS values currently available. This work demonstrates the potential of artificial neural networks (ANNs) for the prediction of CCS values of pesticides. The predictor, based on eight software-chosen molecular descriptors, was optimized using CCS values of 205 small molecules and validated using a set of 131 pesticides. The relative error was within 6% for 95% of all CCS values for protonated molecules, resulting in a median relative error less than 2%. In order to demonstrate the potential of CCS prediction, the strategy was applied to spinach samples. It notably improved the confidence in the tentative identification of suspect and nontarget pesticides.

The need for rapid and efficient high throughput metabolic phenotyping (metabotyping) in metabolomic/metabonomic studies often requires compromises to be made between analytical speed and metabolome coverage. Here the effect of column length (150, 75 and 30 mm) and gradient duration (15, 7.5 and 3 min respectively) on the number of features detected when untargeted metabolic profiling of human urine using reversed-phase gradient ultra performance chromatography with, and without, ion mobility spectrometry, has been examined. As would be expected, reducing column length from 150 to 30 mm, and gradient duration, from 15 to 3 min, resulted in a reduction in peak capacity from 311 to 63 and a similar reduction in the number of features detected from over ca. 16,000 to ca. 6500. Under the same chromatographic conditions employing UPLC/IMS/MS to provide an additional orthogonal separation resulted in an increase in the number of MS features detected to nearly 20,000 and ca. 7500 for the 150 mm and the 30 mm columns respectively. Based on this limited study the potential of LC/IMS/MS as a tool for improving throughput and increasing metabolome coverage clearly merits further in depth study.

IntroductionAs large scale metabolic phenotyping is increasingly employed in preclinical studies and in the investigation of human health and disease the current LC–MS/MS profiling methodologies adopted for large sample sets can result in lengthy analysis times, putting strain on available resources. As a result of these pressures rapid methods of untargeted analysis may have value where large numbers of samples require screening.ObjectivesTo develop, characterise and evaluate a rapid UHP-HILIC-MS-based method for the analysis of polar metabolites in rat urine and then extend the capabilities of this approach by the addition of IMS to the system.MethodsA rapid untargeted HILIC LC–MS/MS profiling method for the analysis of small polar molecules has been developed. The 3.3 min separation used a Waters BEH amide (1 mm ID) analytical column on a Waters Synapt G2-Si Q-Tof enabled with ion mobility spectrometry (IMS). The methodology, was applied to the metabolic profiling of a series of rodent urine samples from vehicle-treated control rats and animals administered tienilic acid. The same separation was subsequently linked to IMS and MS to evaluate the benefits that IMS might provide for metabolome characterisation.ResultsThe rapid HILIC–MS method was successfully applied to rapid analysis of rat urine and found, based on the data generated from the data acquired for the pooled quality control samples analysed at regular intervals throughout the analysis, to be robust. Peak area and retention times for the compounds detected in these samples showed good reproducibility across the batch. When used to profile the urine samples obtained from vehicle-dosed control and those administered tienilic acid the HILIC-MS method detected 3007 mass/retention time features. Analysis of the same samples using HILIC–IMS–MS enabled the detection of 6711 features. Provisional metabolite identification for a number of compounds was performed using the high collision energy MS/MS information compared against the Metlin MS/MS database and, in addition, both calculated and measured CCS values from an experimentally derived CCS database.ConclusionA rapid metabolic profiling method for the analysis of polar metabolites has been developed. The method has the advantages of speed and both reducing sample and solvent consumption compared to conventional profiling methods. The addition of IMS added an additional dimension for feature detection and the identification of metabolites.Electronic supplementary materialThe online version of this article (10.1007/s11306-019-1474-9) contains supplementary material, which is available to authorized users.

No abstract

No abstract