- Delgadillo, David;
- Burch, Jessica;
- Kim, Lee Joon;
- de Moraes, Lygia;
- Niwa, Kanji;
- Williams, Jason;
- Tang, Melody;
- Lavallo, Vincent;
- Khatri Chhetri, Bhuwan;
- Jones, Christopher;
- Rodriguez, Isabel;
- Signore, Joshua;
- Marquez, Lewis;
- Bhanushali, Riya;
- Woo, Sunmin;
- Kubanek, Julia;
- Quave, Cassandra;
- Tang, Yi;
- Nelson, Hosea
The structural determination of natural products (NPs) can be arduous because of sample heterogeneity. This often demands iterative purification processes and characterization of complex molecules that may be available only in miniscule quantities. Microcrystal electron diffraction (microED) has recently shown promise as a method to solve crystal structures of NPs from nanogram quantities of analyte. However, its implementation in NP discovery remains hampered by sample throughput and purity requirements, akin to traditional NP-discovery workflows. In the methods described herein, we leverage the resolving power of transmission electron microscopy (TEM) and the miniaturization capabilities of deoxyribonucleic acid (DNA) microarray technology to address these challenges through the establishment of an NP screening platform, array electron diffraction (ArrayED). In this workflow, an array of high-performance liquid chromatography (HPLC) fractions taken from crude extracts was deposited onto TEM grids in picoliter-sized droplets. This multiplexing of analytes on TEM grids enables 1200 or more unique samples to be simultaneously inserted into a TEM instrument equipped with an autoloader. Selected area electron diffraction analysis of these microarrayed grids allows for the rapid identification of crystalline metabolites. In this study, ArrayED enabled structural characterization of 14 natural products, including four novel crystal structures and two novel polymorphs, from 20 crude extracts. Moreover, we identify several chemical species that would not be detected by standard mass spectrometry (MS) or ultraviolet-visible (UV/vis) spectroscopy and crystal forms that would not be characterized using traditional methods.