Grapevine scion gene expression is driven by rootstock and environment interaction

Abstract

Background: Grafting is a horticultural practice used widely across woody perennial crop species to fuse together the root and shoot system of two distinct genotypes, the rootstock and the scion, combining beneficial traits from both. In grapevine, grafting is used in nearly 80% of all commercial vines to optimize fruit quality, regulate vine vigor, and enhance biotic and abiotic stress-tolerance. Rootstocks have been shown to modulate elemental composition, metabolomic profiles, and the shape of leaves in the scion, among other traits. However, it is currently unclear how rootstock genotypes influence shoot system gene expression as previous work has reported complex and often contradictory findings.

Results: In the present study, we examine the influence of grafting on scion gene expression in leaves and reproductive tissues of grapevines growing under field conditions for three years. We show that the influence from the rootstock genotype is highly tissue and time dependent, manifesting only in leaves, primarily during a single year of our three-year study. Further, the degree of rootstock influence on scion gene expression is driven by interactions with the local environment.

Conclusions: Our results demonstrate that the role of rootstock genotype in modulating scion gene expression is not a consistent, unchanging effect, but rather an effect that varies over time in relation to local environmental conditions.

Keywords: Environmental variation; Grafting; Grapevine; Plasticity; Transcriptomics.

Fig. 1

PCA on gene expression colored by tissue, year, phenology, and rootstock. The top two principal components of the quality filtered, normalized, and variance stabilizing transformation (VST) -transformed gene counts, as colored by A tissue, B year of sampling, C phenological stage, and D rootstock genotype

Fig. 2

Self-organizing map captures clusters of genes that vary with rootstock genotype across three years of study. A A principal component analysis on all genes across the samples showing low-dimensional embeddings of variation in scion gene expression. B The principal component plot, colored by assignment to SOM clusters and filtered for proximity to the median gene in the cluster to show the relationship between SOM and PCA. C Examples SOM clusters that showcase commonly occurring patterns. Mean scaled expression for genes assigned to example SOM clusters (numbered) that were significant for rootstock in post-clustering linear modeling are shown

Fig. 3

Differentially expressed gene counts are enriched for a single year of study. A A heat map showing the number of genes identified as differentially expressed across rootstock contrasts, broken down by tissue, year, phenology, and direction of change (17A = 2017 anthesis, 17V = 2017 veraison, etc.). Genes characterized as differentially regulated are presented in reference to the rootstock on the right (in the comparison labeled “Ungrafted - 1103P”, genes designated as ‘Up’ are more highly expressed in 1103P). B Effects size scans showing the number of genes we would retain (y-axis) if we were to filter on various log2 fold-change thresholds (x-axis) within 2018 leaves. C Venn diagrams comparing grafted vines to ungrafted vines in 2018 leaves across phenological stages. Genes upregulated in grafted vines are shown next to an up arrow, where genes down-regulated in grafted vines are shown next to a down arrow. For this analysis, genes were only filtered on adjusted p-values

Fig. 4

The environmental PCA and its relationship to gene expression as mediated by rootstock genotype. A A PCA biplot showing the span of environmental variation over the course of three years and how the features of the environment load onto those PCs. B Gene expression PCs (gPCs) significant for the interaction of rootstock and the first environmental principal component, ePC1. C gPCs significant for the interactions of rootstock and the second environmental principal component, ePC2