Gene regulation in parthenocarpic tomato fruit

Abstract

Parthenocarpy is potentially a desirable trait for many commercially grown fruits if undesirable changes to structure, flavour, or nutrition can be avoided. Parthenocarpic transgenic tomato plants (cv MicroTom) were obtained by the regulation of genes for auxin synthesis (iaaM) or responsiveness (rolB) driven by DefH9 or the INNER NO OUTER (INO) promoter from Arabidopsis thaliana. Fruits at a breaker stage were analysed at a transcriptomic and metabolomic level using microarrays, real-time reverse transcription-polymerase chain reaction (RT-PCR) and a Pegasus III TOF (time of flight) mass spectrometer. Although differences were observed in the shape of fully ripe fruits, no clear correlation could be made between the number of seeds, transgene, and fruit size. Expression of auxin synthesis or responsiveness genes by both of these promoters produced seedless parthenocarpic fruits. Eighty-three percent of the genes measured showed no significant differences in expression due to parthenocarpy. The remaining 17% with significant variation (P <0.05) (1748 genes) were studied by assigning a predicted function (when known) based on BLAST to the TAIR database. Among them several genes belong to cell wall, hormone metabolism and response (auxin in particular), and metabolism of sugars and lipids. Up-regulation of lipid transfer proteins and differential expression of several indole-3-acetic acid (IAA)- and ethylene-associated genes were observed in transgenic parthenocarpic fruits. Despite differences in several fatty acids, amino acids, and other metabolites, the fundamental metabolic profile remains unchanged. This work showed that parthenocarpy with ovule-specific alteration of auxin synthesis or response driven by the INO promoter could be effectively applied where such changes are commercially desirable.

Fig. 1.

The Agrobacterium binary vectors pDU04.1004 and pDU04.1602 control ovule-specific expression of the iaaM gene from Agrobacterium tumefaciens while pDU04.4522 and pDU04.4001 regulate ovule-specific expression of the rolB gene from Agrobacterium rhizogenes. The vectors pDU04.1004 and pDU04.4522 contain the novel INO ovule-specific promoter from Arabidopsis thaliana and vectors pDU04.1602 and pDU04.4001 contain the reference DefH9 promoter from Antirrhinum majus previously shown to display ovule-specific expression and pathenocarpy with iaaM from Pseudomonas syringae. Other common components present on all vectors include an nptII-selectable marker gene driven by the mannopine synthase 2 promoter (mas5) and a uidA scorable marker gene driven by the ubi3 promoter (ubi3). Arrows indicate the direction of transcription. LB and RB indicate the left and right T-DNA border sequences.

Fig. 2.

Wild-type and parthenocarpic transgenic tomato fruit generated with four different constructs (INO-iaaM, DefH9-iaaM, INO-rolB, and DefH9-rolB). No seeds are visible in the parthenocarpic lines.

Fig. 3.

Pairwise comparison of each transgenic sample with control fruits without seeds. (A) and (B) the number of down-regulated and up-regulated genes (adjusted P-value <10⁻⁴) of samples from DefH9-iaaM and INO-iaaM lines. (C) and (D) Number of down-regulated and up-regulated genes (adjusted P-value <10⁻⁴) of samples from DefH9-rolB and INO-rolB lines. Much overlap exists between down-regulated and up-regulated genes in DefH9-iaaM and INO-iaaM samples. This is also the case in DefH9-rolB and INO-rolB samples.

Fig. 4.

Principal component analysis of 1748 genes with significant differences at P <0.05 in expression among transgenic and control fruits. The dots represent each of the 17 individual microarrays used in this study, and the ellipses represent the 95% confidence limit for the replicates in each group.

Fig. 5.

Cluster analysis of gene expression in control fruit with or without seeds and transgenic fruit with INO-iaaM, DefH9-iaaM, INO-rolB, or DefH9-rolB. Genes were clustered based on differential expression using the made4 package of R statistical software. Three biological replicates were used for each genotype, except for DefH9-iaaM, from which only two biological replicates were available. (A) Expression data for 1748 (18%) target genes with P <0.05 could be divided into five groups based upon expression patterns. (B) Hierarchical clustering and heat map for 62 (0.6%) target genes with P <10⁻⁴. Three of the five groups appeared in the latter.

Fig. 6.

Functional categorization of the six expression pattern groupings obtained from pairwise comparison of each transgenic sample with control fruits without seeds. MapMan display of pathway assignments for gene groups created in previous cluster analysis: (A) Metabolism overview, (B) Regulation overview, (C) Large enzyme families. Colours in small squares were assigned based on positions of target genes in the cluster analysis, indicated on the right margin of heat map figures. White squares indicate target genes whose expression did not pass the P <0.05 cut-off.

Fig. 7.

Gene set enrichment analysis (GSEA) of microarray expression data. (A) Functional categories up-regulated in at least one of the transgenic constructs, compared to controls with seeds. (B) Functional categories down-regulated in at least one of the transgenic constructs, compared to control with seeds. Affymetrix tomato GeneChip targets matched Arabidopsis genes in >800 categories in the MapMan knowledge base. The 51 categories listed contained significant numbers of differentially expressed genes with a false discovery rate (FDR) of <0.27. Numbers in parentheses indicate the number of genes in each set.

Fig. 8.

Principal component analysis of the relative abundance of significantly regulated metabolites obtained from a profile of 400 metabolites sampled for in control and transgenic tomato fruit. The dots represent the biological replicates of the different lines and ellipses define the 95% confidence limits of the metabolite data.