|Dileo, Matthew -|
|Den Bakker, Meghan -|
|Chu, Yiyi -|
Submitted to: The Plant Genome
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: September 4, 2013
Publication Date: N/A
Interpretive Summary: Transgenic modification of crop plants to improve product quality or agronomic performance is a widely used but hotly debated tool. One of the points of contention is concern over unintended effects to food quality or composition in the transgenic crops. The concept of substantial equivalence is often used to describe the degree of difference between the non-transgenic parent and transgenic offspring, that both varieties are substantially equivalent to one another. However, this concept is not a statistical test. Here we utilize a series of statistical tests to help define the meaning of substantial equivalence. We examined chemical composition in diverse, conventional varieties including modern commercial releases, heirloom varieties, and breeding lines to make an estimate the boundaries of consumer-acceptable variation. We next evaluated transgenic tomatoes using the same metrics and saw how they fit into the wider picture. We suggest that this analytical process, both experimental and computational, can help increase our knowledge of food composition and add a new tool to biotechnology risk assessment.
Technical Abstract: Unintended effects to food quality and composition occur no matter the method of plant improvement. The existence of unintended effects is perhaps not the most important point within the discussion, but rather the identity and significance of the compositional changes observed. Here we report the examination of the tomato fruit metabolome by liquid chromatography/mass spectrometry (LC-MS) based methods in diverse conventional tomato germplasm and transgenic varieties with the intended effect of delayed fruit ripening. Field grown plants from two different seasons were analyzed using a statistical pipeline featuring principal components analysis (PCA) and partial least squares discriminant analysis (PLS-DA) to identify key differences between varieties, while weighted correlation network analysis (WGCNA) was used to identify similarities and connections within and between datasets. PLS-DA identified 15 metabolomic markers that differed between the transgenic tomatoes and those varieties capable of fully ripening, but only 5 were significantly different between the non-transgenic parent and transgenic offspring. WGCNA demonstrated the relationships between these 5 metabolomic markers and several others, suggesting that a small suite of highly correlated compounds were significantly different and represented an unintended effect to fruit quality or composition. We assert that metabolomic profiling together with this progression of statistical methods is an efficient and powerful tool to examine food composition and may play a role in the risk assessment process.