Terpene Synthases in Ginger and Turmeric

Ginger (Zingiber officinale Rosc.) and turmeric (Curcuma longa L.) produce important pharmacologically active metabolites at high levels, which include terpenoids and polyketides such as curcumin and gingerols. This dissertation describes the terpenoids produced by ginger and turmeric, candidate ESTs for terpene synthases, and the cloning and expression of several terpene synthases. A comparison of metabolite profiles, microarray results and EST data enable us to predict which terpene synthases are related with the production of specific terpenoids. Analysis of EST data further suggests several genes important for the growth and development of rhizomes. Ginger and turmeric accumulate important pharmacologically active metabolites at high levels in their rhizomes. Comparisons of ginger and turmeric EST data to publicly available sorghum rhizome ESTs revealed a total of 777 contigs common to ginger, turmeric and sorghum rhizomes but absent from other tissues. The list of rhizome-specific contigs was enriched for genes associated with regulation of tissue growth, development, and regulation of transcription. The analysis suggests ethylene response factors, AUX/IAA proteins, and rhizome-enriched MADS box transcription factors may play important roles in defining rhizome growth and development. From ginger and turmeric, 25 mono- and 16 sesquiterpene synthase sequences were cloned and the function of 13 mono- and 11 sesquiterpene synthases were revealed. There are many paralogs in the ginger and turmeric terpene synthase family, some of which have the same or similar function. However some paralogs have diverse functions and this suggests the evolution of terpene synthases in ginger and turmeric. Importantly, α-zingiberene/β-sesquiphellandrene synthase was identified, which makes the substrates for α-turmerone and β-turmerone production in turmeric. Also P450 candidates for α- zingiberene/β-sesquiphellandrene oxidase are proposed. Research involving analysis of metabolite profiles requires the manipulation of a large datasets, such as those produced by GC/MS. We developed an approach to identify compounds that involves deconvolution of peaks obtained using SICs as well as common peak selections between samples even though the peaks may be very small and represent unknown compounds. The limitation of this approach occurs when there are huge peaks in the samples, which distort the SIC of small embedded peaks and sometimes their own SICs