Terpene Patterns Across Analytical Techniques for Advanced Authenticity Control of Natural Products

Abstract

Natural products are secondary metabolites produced by plants to protect against environmental stressors, attract pollinators, and enhance survival. Terpenes, the largest class of these metabolites, are synthesised through the linkage of isoprene units and diversified by specific enzymes. While some terpenes are species-specific, others appear ubiquitous in the plant kingdom. Terpene patterns in plants are not random but are shaped by evolutionary pressures, making them a rich source for drug discovery and development, as they are inherently optimised drug-like molecules. Terpenes exhibit structures with chiral centres, and their composition in natural products is influenced by genetic and environmental factors, such as growth conditions, but also ageing and processing. Detailed analysis of terpene patterns can serve as a fingerprint to verify the identity and authenticity of natural products, as deliberate adulteration can lead to altered compositions. However, due to their structural similarities, terpenes pose analytical challenges requiring high resolution separation and reliable detection methods, which can be time-consuming and difficult to interpret. This thesis focuses on the holistic characterisation of natural products, with a particular emphasis on terpenes, achieved through optimising separation and detection methods, alongside computational analysis. The methods developed further prioritise speed and ease of use. Firstly, this thesis investigates optimised separation methods for terpenes in Cannabis sativa L. and Rosa damascena. Gas chromatography (GC) and liquid chromatography (LC) were evaluated using different column polarities, resulting in five validated methods for both qualitative and quantitative profiling of natural products. These methods enable the authenticity control of R. damascena essential oil (EO) beyond current international standardisation guidelines and an in-depth cultivar determination for C. sativa L., inclusive of minor cannabinoids and terpenes. An enantiomeric excess was found in R. damascena for (-)-cis rose oxide, (-)-linalool, and (-)-citronellol, which differs from adulterations with pelargonium oil. In C. sativa L. an excess of (+)-α-pinene, (+)-β-pinene, (S)-limonene, (+)-linalool, (-)-citronellol, (-)-camphene, (+)-trans-nerolidol, and (-)-cis-menthone were identified. However, indoor-grown C. sativa L. plants lacking citronellol exhibited an excess of (-)-α- and (-)-β-pinene, indicating that cultivation conditions (indoor vs. outdoor) and plant protective agents alter enantiomer production. The entourage effect, a synergistic interplay between secondary metabolites, is also discussed, as contradictive study outcomes may stem from varying enantiomer ratios. Additionally, the validated methods were applied to investigate the effect of storage and ageing on terpene patterns. Stability studies revealed isomerisation, polymerisation, oxidation, and cyclisation reactions, with p-cymene identified as an ageing marker. No enantiomeric conversions were observed. Secondly, different detection methods for terpenes were explored. These ranged from flame ionisation detection (FID) to mass spectrometry (MS) with different ionisation methods. Atmospheric pressure chemical ionisation (APCI) proved suitable for ionising terpenes and is compatible with LC. Dielectric barrier discharge ionisation (DBDI), a low-temperature plasma method, was investigated for GC. The ionisation process in DBDI is not straightforward and remains a subject of ongoing research. In contrast to APCI, which predominantly produces [M+H]+ adducts, DBDI generates a diverse range of adduct species, primarily influenced by functional groups within the analytes. Oxygenated terpenes primarily formed adducts such as [M]+, [M+H]+, [M+2H]+, and [M+NH4]+, while non-oxygenated terpenes formed adducts like [M-H+2O]+, [M+H+3O]+, [M+H+2O]+, and [M+H2O2]+. Furthermore, DBDI-MS spectra enable the distinction of constitutional isomers and diastereomers. Lastly, predictive machine learning models were evaluated, which integrate data from multiple analytical techniques to assess the originality of R. damascena EO. Single-block models using data from one technique were compared to multi-block models, which make use of data fusion across analytical platforms. The highest classification accuracy was achieved with a model based on the quantitative profile with partial least-squares discriminant analysis (PLS-DA). However, a comparable model based on DBDI-MS data achieved similar classification accuracy in a fraction of the amount of time. While multi-block analysis did not exceed the classification accuracy of single-block models, combining FT-IR data with DBDI-MS data improved accuracy compared to using FT-IR data alone in the single-block approach. This thesis introduces a versatile toolbox for the analytical chemist that is suitable for the analysis of terpenes. The methodologies outlined here are neither limited to terpene analysis nor the two model plants studied: they can be extended to a broad range of other secondary plant metabolites. This approach contributes to the safe use of natural products in human applications.