Establishment of a quality control mixture for benchmarking LC-MS based dereplication protocols in natural product research

In natural products (NPs) research, recent dereplication workflows are usually based on the treatment of data obtained from generic untargeted metabolite profiling by UHPLC-HRMS/MS of crude extracts [1]. When applied on large collection of complex NPs extracts, the amount of information generated by such approaches is extremely important and requires to finely tune the acquisition parameters. Additionally, various computational solutions, each presenting a great numbers of parametrical options, exists to mine the generated data. In order to assess the overall quality of the NPs profiling and dereplication workflows we propose to use a well-defined Quality Control (QC) sample mixture prepared from the combination of 5 extensively described plants of the European Pharmacopeia for which extensive phytochemical studies have been performed. Plants were selected by mapping the chemical space of their constituents against the space occupied by all NPs known to date for the best coverage and extended polarity range. This mixture was then analyzed in different LC gradient conditions and MS/MS acquisition modes on different MS platforms. By using MS-DIAL [3], for example, it could be observed that the number of identified features increased by a factor ranging from 2 to 15% when the measurement time increased by 2 min steps starting from a 5 min gradient. The quality of automated annotations was also assessed in various chromatographic regions and over a large dynamic range in data dependent or data independent modes.

The results suggested that this plant mixture easily generable in any laboratory could become a universal QC mix for the quality evaluation of NPs profiling and annotation workflows. Evaluation of various dereplication strategies in terms of number of annotations and their quality and acquisition modes will be discussed.

[1] Wolfender J-L, et al. J Chromatogr A 2015; 1382:136 – 164.

[2] Tsugawa H, et al. Nature Methods 2015; 12: 523 – 52.