Advancing Natural Product Discovery: A Structure-Oriented Fractions Screening Platform for Compound Annotation and Isolation

Natural product discovery is hindered
by the lack of tools that integrate untargeted nuclear magnetic resonance
and mass spectrometry data on a library scale. This article describes
the first application of the innovative NMR/MS-based machine learning
tool, the “Structure-Oriented Fractions Screening Platform
(SFSP)”, enabling functional-group-guided fractionation and
accelerating the discovery and characterization of undescribed natural
products. The concept was applied to the extract of a marine fungus
known to be a prolific producer of diverse natural products. With
the assistance of SFSP, we isolated 24 flavipidin derivatives and
five phenalenone analogues from Aspergillus sp. GE2-6,
revealing 27 undescribed compounds. Compounds 7–22 were proposed as isomeric derivatives featuring a 5/6-ring
fusion, formed by the dimerization of flavipidin E (5). Compounds 23 and 24 were envisaged as
isomeric derivatives with a 6/5/6-ring fusion, generated through the
degradation of two flavipidin E molecules. Furthermore, flavipidin
A (1) and asperphenalenone E (28) exhibited
potent anti-influenza (PR8) activities, with IC50 values
of 21.9 ± 0.2 and 12.9 ± 0.1 μM, respectively. Meanwhile,
asperphenalenone (26) and asperphenalenone P (27) treatments exhibited significant inhibition of HIV pseudovirus
infection in 293FT cells, boasting IC50 values of 6.1 ±
0.9 and 4.6 ± 1.1 μM, respectively. Overall, SFSP streamlines
natural product isolation through NMR and MS data integration, as
showcased by the discovery of numerous undescribed flavipidins and
phenalenones based on NMR olefinic signals and low-field hydroxy signals.