Efficient constitution of a library of rotenoid analogs active against from a digitalized plant extract collection.

Natural products (NP) have proven to be a rich source of potentially bioactive compounds, and metabolomics is the current method of choice for characterizing natural extracts. To integrate the vast amount of data and information produced by modern metabolomics workflows, we recently developed a sample-centric approach for the semantic enrichment and alignment of metabolomics datasets. The resulting Experimental Natural Products Knowledge Graph (ENPKG) is queryable and integrates both newly acquired digitalized experimental data and information, and previously reported knowledge. It allows the highlighting of putative bioactive compounds at the extract level by comparing, for example, the occurrence of compounds of a given chemical class with bioactivity results. Using this approach, we recently described potent anti- activity of two rotenoids, deguelin and rotenone. These compounds were identified in six active extracts from four plant species: (Connaraceae), , , and (Fabaceae). In this work, we present the results of the phytochemical investigation of four of these extracts and the establishment of a library of structural analogs for bioactivity testing. This work led to the isolation, characterization, and biological evaluation of the anti- potential of 41 compounds, including 11 rotenoids and seven compounds reported for the first time. Thanks to modern metabolite annotation and single-step isolation procedures, this work also demonstrates the possibility of considering natural extract libraries as a reservoir of rapidly accessible pure NPs. This perspective could increase the options for NP research and help accelerate NP drug discovery efforts.