Dereplication of natural products from complex extracts by regression analysis and molecular networking: case study of redox-active compounds from Viola alba subsp. dehnhardtii

Introduction

In natural product research, bioassay-guided fractionation was previously widely employed but is now judged to be inadequate in terms of time and cost, particularly if only known compounds are ultimately isolated. The development of metabolomics, along with improvements in analytical tools, allows comprehensive metabolite profiling. This enables dereplication to target unknown active compounds early in the purification workflow.

Objectives

Starting from an ethanolic extract of violet leaves, this study aims to predict redox active compounds within a complex matrix through an untargeted metabolomics approach and correlation analysis.

Methods

Rapid fractionation of crude extracts was carried out followed by multivariate data analysis (MVA) of liquid chromatography–high resolution mass spectrometry (LC–HRMS) profiles. In parallel, redox active properties were evaluated by the capacity of the molecules to reduce 2,2-diphenyl-1-picrylhydrazyl (DPPH^·) and superoxide (O₂ ^·?) radicals using UV–Vis and electron spin resonance spectroscopies (ESR), respectively. A spectral similarity network (molecular networking) was used to highlight clusters involved in the observed redox activities.

Results

Dereplication on Viola alba subsp. dehnhardtii highlighted a reproducible pool of redox active molecules. Polyphenols, particularly O-glycosylated coumarins and C-glycosylated flavonoids, were identified and de novo dereplicated through molecular networking. Confirmatory analyses were undertaken by thin layer chromatography (TLC)–DPPH–MS assays and nuclear magnetic resonance (NMR) spectra of the most active compounds.

Conclusion

Our dereplication strategy allowed the screening of leaf extracts to highlight new biologically active metabolites in few steps with a limited amount of crude material and reduced time-consuming manipulations. This approach could be applied to any kind of natural extract for the study of various biological activities.