Investigating the use of untargeted metabolomics for herbal product applications
Restricted (Penn State Only)
- Author:
- Abraham, Ellie
- Graduate Program:
- Plant Biology
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- April 08, 2024
- Committee Members:
- Teh-Hui Kao, Program Head/Chair
Joshua Kellogg, Chair & Dissertation Advisor
Joshua Lambert, Outside Field Member
Jesse Lasky, Major Field Member
Justin Silverman, Outside Unit Member - Keywords:
- Metabolomics
predictive modeling
herbal products
Herbal prodects - Abstract:
- Herbal products are classified as dietary supplements under 21CFR111, and thus must meet specific requirements for identity, purity, potency, and limits of contaminants. These analyses are complicated for botanical ingredients, especially considering their large, dynamic chemical profiles. This dissertation focuses on understanding how advanced mass spectrometry (MS) instruments can improve herbal identity and potency studies, as well as the implications of instrument advancement for defining proper specifications in a regulatory context. Modern mass spectral instruments can produce fully untargeted metabolite profiles. This shifts data analysis from quantifying one to a few metabolite markers specific to a plant species or therapeutic effect to analyzing the relative abundance of thousands of compounds simultaneously. Throughout this dissertation, I used Ocimum, or basil/Tulsi, to investigate how various multivariate statistical techniques can aid in the interpretation of this data, and how the developed models can improve botanical regulatory testing. I discovered that the liquid chromatography-MS metabolite profiles of three species of greenhouse-grown Ocimum reference materials are significantly different from each other but are not useful for constructing multivariate models to predict the identity of consumer-available products. Additionally, I demonstrate that the same materials (greenhouse and consumer Ocimum) have variations in cytotoxicity against HT-29 human colorectal cells in an MTT assay. The range of activities provides enough biological variation to construct models for identifying compounds associated with increases in biological activity. Specifically, I used a series of filtering and preselection steps to improve a Partial Least Squares model to identify gallic acid as a key driver of Ocimum cytotoxicity variations. In the final chapter, I compared the metabolite profiles of two Ocimum species’ essential oils resulting from high- and low- resolution gas chromatography-MS. This study highlighted that high resolution instrumentation improves metabolite coverage and dynamic range but lacks the supporting libraries for improved compound identification at this time. Additionally, I determined that the two instruments result in different key compounds distinguishing the two species, and that targeted compound specifications must be instrument specific. Taken together, these studies suggest that untargeted metabolomics and multivariate analysis has a place in herbal product evaluations, but there are still many remaining steps before it becomes a mainstream technique in regulatory testing environments.