Conventional isolation methods in natural products chemistry are time-consuming and costly and often result in the isolation of moderately active compounds or the detection of already known natural products (NPs). A fast and cost-effective way to identify bioactive metabolites in plant extracts prior to isolation has been developed based on the nuclear magnetic resonance (NMR)-heterocovariance approach (NMR-HetCA). In order to evaluate in depth the application of this chemometrics-based drug discovery methodology, simple mixtures of 10 standard NPs simulating a fast centrifugal partition chromatography (FCPC) fractionation (artificial fractions, ArtFrcts), as well as a more complex mixture of 59 natural standard substances simulating a crude plant extract (artificial extract, ArtExtr), were prepared. FCPC was employed for the fractionation of the ArtExtr, while the inhibitory activity of all fractions against DPPH was evaluated, and their chemical profile was recorded using NMR spectroscopy. Spectral information was processed in the MATLAB environment, and statistical approaches, including HetCA and statistical total correlation spectroscopy (STOCSY), were applied to identify bioactive compounds. Total heterocovariance plots (pseudospectra) facilitated the detection of highly correlated metabolites and led to the direct identification of 52.6% of the active compounds. The success in identifying the ArtExtr bioactive substances increased to 63.2% when spectral alignment was implemented. HetCA incorporates chromatographic (fractionation), spectroscopic (NMR profiling), and bioactivity results along with advanced chemometrics and could be established as a method of choice for the rapid and effective identification of bioactive NPs in plant extracts prior to isolation.

The assignment of bioactivity to compounds within complex natural product (NPs) mixtures remains a significant challenge in NPs research. The present research introduces a comprehensive protocol, named "PLANTA (PhytochemicaL Analysis for NaTural bioActives)" protocol, for the detection and identification of bioactive compounds in complex natural extracts prior to isolation combining the NMR-HeteroCovariance Approach (NMR-HetCA), high-performance thin-layer chromatography (HPTLC), and chemometric techniques. This study emphasizes two novel components: STOCSY-guided targeted spectral depletion, adapted to resolve overlapping NMR signals in complex matrices, improve minor component detection, and facilitate identification through NMR databases, as well as a new SHY variant termed SH-SCY (Statistical Heterocovariance - SpectroChromatographY), a new cross-correlation method linking orthogonal datasets by identifying the corresponding HPTLC spot from a single NMR peak and reconstructing of the ¹H NMR spectrum from a specific HPTLC spot, enhancing dereplication confidence. In this proof-of-concept study, an artificial extract (ArtExtr) composed of 59 standard compounds was evaluated for the detection of compounds active against the free radical 2,2-diphenyl-1-picrylhydrazyl (DPPH). Statistical approaches were applied to the spectral, chromatographic, and bioactivity data to identify the highly correlated bioactive compounds. The PLANTA protocol achieved an 89.5% detection rate of active metabolites and 73.7% correct identification of them. The integration of NMR and HPTLC with HetCA provides a robust and sensitive strategy for preisolation identification of bioactive constituents. This methodology addresses core challenges in metabolite profiling of complex mixtures and offers a streamlined, reproducible workflow for natural product dereplication and discovery.

High-Performance Thin Layer Chromatography (HPTLC) is widely utilized in natural products research due to its simplicity, low cost, and short total analysis time, including data treatment. While bioautography can be used for rapid detection of bioactive compounds in extracts, the number of available bioautographic methods is limited mainly due to the high cost and difficulty in developing protocols that lead to accurate and reproducible results. For this reason, an alternative method for the detection of bioactive compounds in plant extracts prior to their isolation using HPTLC, combined with multivariate chemometrics, was previously explored by our lab. To evaluate this method and compare it to other chemometrics-based methods, an artificial mixture (ArtExtr) of 59 standard compounds was used as a case study. The ArtExtr was fractionated by FCPC and the inhibitory activity of all fractions against DPPH was evaluated, while their chemical profiles were recorded using HPTLC. Multivariate statistics and the heterocovariance approach (HetCA) were employed and compared, with the success rate in detecting the ArtExtr bioactive substances being 85.7% via sparse heterocovariance (sHetCA). HPTLC combined with sHetCA can serve as a valuable tool for the detection of bioactive compounds in complex mixtures when bioautography is not feasible.