The dissection of GxE interactions using a novel data mining approach uncovers the plasticity of the grapevine berry transcriptome

Grapevine, the most widely-cultivated perennial fruit crop, is also considered a very environmentally sensitive crop. It is characterized by remarkable phenotypic plasticity which in turn is believed to effectively buffer environmental extremes especially through transcriptomic and epigenomic reprogramming. Thus, the final phenotype (P) of a given vine is the result of the close interaction between its genetic composition (G) and the environment (E). Here we analyzed Genotype x Environment (GxE) interactions in two grapevine varieties by characterizing their transcriptome plasticity when cultivated in different environments. Specifically, two genotypes (Sangiovese and Cabernet Sauvignon) were cultivated in three different locations in Italy (Bolgheri -littoral Tuscany-, Montalcino – Appennine Tuscany- and Romagna -foothill area-), trained in an almost identical manner, and sampled at four developmental stages over two grapevine growing seasons, 2011 and 2012, for a total of 144 samples that were analyzed by hybridization to a whole-genome microarray. In order to study the relationships among differential gene expression profiles and environmental cues, we have developed a new statistical data mining tool based on data reduction approaches which allowed a dissection of the transcriptomic data into stage-specific, cultivar-related and GxE important clusters of gene expression. This deep inspection of inner relationships between the different dataset variables allowed the identification of several candidate genes that could represent putative markers of berry quality traits in grapevine GxE interactions. Moreover, the methods used to establish our model provide a framework for the analysis of transcriptome plasticity in other crops as they respond to diverse environments.