The microarray study described herein primarily focused on the leaf transcriptomes of the potato cultivar White Lady (WT) and the TPS1 transgenic derivative (T2) exposed to drought stress in the form of 30% soil moisture content. By analysing the microarray data, more than 5,000 genes, which had statistically significant changes in their expression, were identified in the WT plants under drought versus WT plants under irrigation (WTd-WTw), T2 plants under drought versus T2 plants under irrigation (T2d-T2w), and T2 plants under drought versus WT plants under drought (T2d-WTd) comparisons. Although the stress treatment resulted in higher water loss in the drought-sensitive WT plants relative to the drought-tolerant TPS1 transgenic plants, many more genes showed altered expression in response to stress in T2 than in WT leaves (3,658 versus 930 genes, respectively, Figure 1).
One major challenge in microarray analysis is to give biological sense to statistically significant data. In our work, we used three approaches to achieve this. First, we analysed our data by two-way ANOVA to identify genes whose expression depend on two factors, plant genotype (wild-type and transgenic) or treatment (drought stress and irrigation), or on the interaction of these two factors. By this analysis, we reduced the number of genes from 5,446 to 1,496. All genes returned by ANOVA had an expression ratio larger than two. Second, this reduced set of genes was annotated into functional categories using the MapMan software. The annotation returned 1,009 genes with known association with biochemical pathways or regulatory functions. Third, we identified the common genes between the ANOVA- and MapMan-returned sets. This resulted in 379 genes, which fulfilled all of the following criteria: (i) gene expression depends on either genotype or treatment or on the interaction of the two, (ii) they have a known function and (iii) the expression ratio is larger than two and is statistically significant at the P ≤ 0.05 level. We consider these 379 genes to have biological importance in drought physiology of potato.
Out of the 379 genes with altered expression, 112 were regulated in the same direction in response to drought in WT and T2 plants. Because the relative water content (RWC) of the T2 plants was only reduced from 85% to 81%, it appears that the expression of these genes is particularly sensitive to water loss. An alternative explanation is that the plant senses the water content of the soil and regulates the transcription of these genes accordingly. The 112 commonly regulated genes included nine down-regulated and three up-regulated genes involved in photosynthesis and carbohydrate metabolism, including chlorophyll a-b binding proteins, fructose-1,6-bisphosphatase, trehalose-6-phosphate synthase (all down-regulated), and sucrose synthase (up-regulated). Recently, Evers et al.  compared two potato clones of the Andean cultivar group with different drought tolerance phenotypes. Although the RWC of leaves exposed to prolonged drought stress was reduced by only 2–3%, repression of chlorophyll a–b binding proteins, fructose-1,6-bisphosphatase, and trehalose-6-phophate synthase and induction of sucrose synthase genes occurred in both Andean cultivars as well.
Induction of sucrose synthase 3 (SUS3) occurred not only under stress conditions in the WT plants (this study) but also in well-watered TPS1 transgenic plants . Furthermore, an ATP-dependent caseinolytic protease (an essential housekeeping enzyme in plant chloroplasts ), actin 7 (a structural constituent of the cytoskeleton ), and a V-type proton ATPase gene (an enzyme that transforms the energy from ATP hydrolysis to electrochemical potential differences in proton concentrations across diverse biological membranes ) were up-regulated in irrigated TPS1 transgenic plants  and induced by stress in WT plants. There might therefore be a common signal generated by the expression of TPS1 and drought stress that leads to the up-regulation of these four genes.
Evers et al.  reported that while biochemical changes did not clearly reflect gene expression changes in Andean cultivars, galactose, inositol and galactinol contents were higher in the drought-stressed tolerant cultivar relative to the more sensitive strain. Although we were also unable to directly correlate transcriptional changes with biochemical differences, we found an increase in galactose content in the sensitive WT plants and elevated inositol contents in both WT and TPS1 transgenic plants. We also observed a 65% reduction in the starch content of WT leaves but no dramatic changes in sucrose in either line (Figure 5). The starch content of the TPS1 transgenic leaves was not reduced in the drought-stressed plants but remained at the same low level as observed under well-watered conditions. We therefore speculate that a constant sucrose level may be very important for potato plants. Since stress reduces the rate of photosynthesis maintenance of a constant sucrose level under drought stress conditions may require the plants to reduce starch synthesis and channel the carbohydrates to sucrose synthesis.
Inositol is a versatile compound that generates diversified derivatives upon phosphorylation. These compounds have dual functions as signalling molecules as well as key metabolites under stress . We previously found a 1.4- to 1.6-fold increase in the inositol level of leaves of TPS1 transgenic plants grown under well-watered conditions. This elevation was further increased by drought to 2.6- to 3.2-fold higher than the well-watered WT control. In WT plants, a 4.4-fold increase in inositol content was detected in response to drought. Because the high level of inositol correlates with the low level of starch, we assume that inositol serves as a signal for the reduction of starch synthesis. Besides phosphatidylinositol, inositol-derived galactinol and associated raffinose family oligosaccharides are emerging as antioxidants and putative signalling compounds . In a comparison of the carbohydrate metabolism of a drought-tolerant advanced potato clone and a sensitive commercial variety, the tolerant clone presented an increase in galactinol and raffinose contents, especially in the leaves . We also found a very robust increase (5.5- to 11-fold) in raffinose content that was more pronounced in WT than TPS1 transgenic plants. Unlike inositol, the raffinose level was not elevated under well-watered conditions in the TPS1 transgenic lines compared to WT plants (data not shown). The regulatory mechanisms that underlie these increases in inositol and raffinose contents are likely quite different. While inositol synthesis is influenced by the transcriptional and/or biochemical changes triggered not only by drought but also by the expression of yeast TPS1 in potato, raffinose synthesis is induced by water loss and is negatively correlated with leaf RWC.
Drought stress induced the accumulation of proline in both WT and TPS1 transgenic leaves. Plant proline concentrations are regulated by an interplay of biosynthesis, degradation and intra- as well as intercellular transport processes. Proline is synthesised from glutamate or ornithine, and the first pathway initiated by Δ1-pyrroline-carboxylate reductase (P5CR) is considered to be dominant under stress conditions . In Andean potato cultivars, the increase in proline was linked to the up-regulation of P5CS and the down-regulation of proline dehydrogenase ( PDH), which is involved in proline catabolism . In our experiments, P5CS and PDH were not among the 379 selected genes, suggesting that other processes than transcriptional regulation might also influence the accumulation of proline in leaf cells.
We identified 57 genes with differential expression in T2 but not WT leaves. This difference in expression might be explained by the different RWCs of drought-stressed WT (65 ± 7%) and T2 (81 ± 1%) leaves but could also be attributed to transcriptional and metabolic changes induced by the transgene under well-watered  as well as drought stress conditions. We found four TFs uniquely up-regulated in T2 leaves: two different proteins involved in chromatin modification, one involved in plastid genome transcription, and one involved in jasmonate responses. Because TFs generally influence the transcription of a set of genes, it is possible that the four TFs alter the expression of several target genes and trigger a cascade of downstream signalling events.
Several different sets of cis- and trans-acting factors are known to be involved in stress-responsive transcription. Some are controlled by the phytohormone abscisic acid (ABA), but others are not, indicating the involvement of both ABA-dependent and ABA-independent regulatory systems for stress-responsive gene expression . Expression of StDS2 is highly drought-specific and independent of ABA . In this study, induction of StDS2 expression was detected in both WT and TPS1 transgenic leaves. Surprisingly, however, ABA-responsive marker genes such as RD22
ERD15 and ERD3 did not appear in our selected list of genes. Instead, we found that abscisic aldehyde oxidase, which catalyses the last step of ABA biosynthesis, and an ABA-mediated dehydration-responsive protein transcript were down-regulated in both WT and T2 and only WT plants, respectively (Figure 4). Together, these correlative changes suggest that the ABA level after prolonged drought stress is not as high as observed in short-term responses to osmotic stress, although this has yet to be directly verified.