If you have any problems related to the accessibility of any content (or if you want to request that a specific publication be accessible), please contact us at email@example.com.
A 1H NMR Network Approach for the Study of Specialized Metabolites: Development and Applications
AdvisorJeffrey, Christopher C
AltmetricsView Usage Statistics
The history of the World could not be told without mentioning specialized metabolites. Long before we had the notion of organic molecules, we were already sailing across the oceans in the pursue of their valuable sources, and raising whole empires based on the economy of spices, tobacco and coffee. Our own survival and adaptability as a species can be credited to discoveries of nature-sourced drugs through the centuries, such as penicillin, paclitaxel and artemisinin. As our awareness towards the environment evolved, so did our perspective on these molecules, which are now recognized as important products and intermediates of the interaction between organisms and their ecosystems. It is not an overstatement thus, to affirm that in the light of an imminent environmental crisis, the study of specialized metabolites will be of fundamental importance for the preservation and recovery of biodiversity. The development of untargeted analytical techniques has greatly advanced our ability to investigate these compounds in their natural contexts. Among those, the application of Proton Nuclear Magnetic Resonance (1H NMR) spectroscopy for chemical profiling enabled the recognition of compound structural features and facilitated the identification of unknown compounds even prior to the laborious work of isolation. However, because multiple resonance signals arise from a single compound, a considerable amount of overlap is observed in biological samples which can limit our ability to detect those key structural markers. There are certain NMR pulse experiments that can aid in deconvoluting these signals, but a more practical approach resides in the statistical treatment of the 1H NMR spectrum, in which regular variations across the spectrum are directly mapped to variables pertinent to the system in study, such as biological activity, biogeographical data, or phylogenetic classification. This document presents an innovative strategy in which gene co-expression network analysis is adapted for the statistical treatment of 1H NMR spectral data, resulting in the deconvolution of metabolite signals and simplification of the spectrum into a few variables. These variables represent statistically measurable chemical patterns that in conjunction with other measurements can support addressing a multitude of topics in Chemical Ecology. In Chapter 1, we describe the method development and validation in a controlled experiment with prepared mixtures of known compounds and demonstrate how it recognizes metabolite identity at different structural levels. We then demonstrate its applicability in the study of natural mixtures with the investigation of ontogenetic changes of metabolism in Piper kelleyi. In Chapter 2, we applied the method to the identification of biologically active compounds from a chemically heterogeneous set of 29 Piper extracts. By quantifying the association between chemical patterns and measurements of antifungal activity, we accurately identified specific targets for the isolation of antifungal compounds, while also establishing a framework to evaluate the effect of specific structural features in modulating the toxicity of different plant species. Finally, in Chapter 3, we adopted an untargeted approach to investigate the phylogenetic signal of specialized metabolites in a broader collection of Piper species. Based upon measurements of chemical similarity, we identified the chemical traits most strongly associated with the clade Schilleria, ultimately leading us to the characterization of novel lignans. Altogether, this 1H NMR network approach represents a powerful tool for the study of specialized metabolites in a myriad of contexts relevant to the field of Chemical Ecology.