4 DISCUSSION
Drought is one of the major abiotic stresses affecting agronomic productivity worldwide. Water restriction exerts its negative effect at various physiological and metabolic levels, and the plant protective responses exhibit a comparable complexity. Stomatal closure is among the earliest, aimed at preventing water loss through transpiration (Sperryet al. 2017). An unwanted consequence of this defensive mechanism is the inhibition of gas exchange and CO2 assimilation, which in turn leads to NADPH build-up and blockade of photosynthetic electron transport due to limitation of electron acceptors (oxidized Fd and NADP+). The excess of reducing equivalents accumulated in the PETC might increase adventitious O2reduction and ROS propagation, triggering redox-based signaling pathways and eventually oxidative damage (Gómez et al. 2019). Plant responses to this particular aspect of the drought syndrome include a suite of alternative electron transport pathways that dissipate the surplus of excitation energy from the PETC and/or export reducing equivalents to other cellular compartments. They comprise cyclic electron transport, photorespiration, chlororespiration, flavin-diiron proteins, the malate valve and the Mehler-Asada cycle (reviewed in Gómezet al. 2019). Scavenging enzymes and metabolites complement the activity of the dissipative systems, limiting ROS accumulation and toxicity. However, drought also causes down-regulation of many photosynthetic components including Fd (Evers et al. 2010; Kondrák et al. 2012; Iovieno et al. 2016). While this response effectively alleviates the hazardous combination of light absorption and highly reduced redox intermediates in a context of elevated oxygen levels, the final outcome is further inactivation of photosynthesis.
In many phototrophic microorganisms, environmental stresses such as salinity, heat and iron starvation cause Fd repression (as in plants), accompanied by induction of the equivalent electron shuttle Fld (Pierella Karlusich et al. 2014). This functional substitution allows algae and cyanobacteria to survive and reproduce in hostile environments (Pierella Karlusich et al. 2015). Introduction of Fld in plants via genetic engineering increased tolerance to multiple stresses including drought (Tognetti et al. 2006; Li et al. 2017), complemented Fd-deficient plants (Blanco et al. 2011) and affected the expression of hundreds of genes even at physiological growth conditions and Fd levels (Pierella Karlusich et al. 2017). Moreover, leaves from Fld-expressing tobacco plants displayed a high-light acclimation phenotype (Gómez et al. 2020). These promising results encouraged application of the Fld approach to crops and to the most agronomically relevant stress, drought.
In the present study, we demonstrate that potato plants expressing a plastid-targeted Fld were more tolerant to a short-term drought regime, as reflected by significant preservation of photosynthetic activity (Figure 1) and lower ROS build-up in plastids of Stpfld plants (Figure 2). Drought stress activates a wide array of responses in plants, which together with constitutive traits determine whether the plant will be able to cope with the adverse situation (André et al. 2009). To gain further knowledge on the mechanism of tolerance conferred by Fld, a transcriptomic analysis was carried out on watered and drought-stressed WT and Stpfld 252 potato plants. Since chloroplast Fld could prevent early effects of drought treatment (Figures 1, 2), we chose a 3-day water restriction regime to monitor initial stress responses.
Expression of the plastid-targeted flavoprotein, per se , resulted in significant changes in the expression levels of 1675 genes under control conditions, representing pathways that presumably respond to chloroplast-dependent redox-based regulation (Figure 3; Supplementary Table S2). A similar observation has been previously reported for Fld-expressing tobacco plants in terms of the fraction of DE transcripts (5-6% of total leaf-expressed genes) and, as revealed by functional enrichment analysis, in the pathways over-represented among them (Pierella Karlusich et al. 2017).
Protein degradation via the proteasome is the most conspicuous enriched category. A total of 67 proteasomal genes were induced in Stpfldplants, 48 of which showed a similar behavior in tobacco (Supplementary Figure S8; Pierella Karlusich et al. 2017), suggesting that up-regulation of this pathway is a general plant response to changes in chloroplast redox poise. Proteasome activity has been shown to modulate several key plant responses to both developmental and environmental stimuli (Stone 2014; Bahmani, Kim, Lee & Hwang 2017; Kovács et al. 2017), indicating that this particular effect of Fld might have profound consequences for plant growth and welfare. Research is currently underway to evaluate these possibilities. In tobacco, most of the proteasomal genes induced by Fld were up-regulated by biotic stress in the wild type (Pierella Karlusich et al. 2017), suggesting that Fld presence primed the plant against subsequent pathogen challenge. Instead, water limitation had little or no effect on proteasomal expression in WT potato (Supplementary Figure S6). Then, induction of this pathway was not involved in the differential drought tolerance displayed by Stpfld lines.
Fld-dependent repression of genes involved in ethylene metabolism was also common to both species, including transcription factors of the apetala2/ERF family, 13 in potato (Figure 3; Supplementary Table S2) and 11 in tobacco (Pierella Karlusich et al. 2017). Common transcriptional responses to Fld presence were observed for 10 of these regulatory factors in the two species, including members of the ERF1, ERF4 and ERF5 sub-families (Supplementary Table S2).
After 3 days of water deprivation, 5701 transcripts were differentially expressed in WT plants, but only 4097 in Fld-expressing siblings (Figure 4). Then, the overall effect of Fld presence was to mitigate the changes in gene expression driven by the drought treatment; either induction or repression, suggesting that plants expressing chloroplast Fld suffered less stress than their WT counterparts, and accordingly displayed an attenuated response that affected many functional categories, as illustrated in Supplementary Figure S9. It is likely that protection of photosynthesis and other metabolic routes contributes to this regulatory effect, presumably by limiting ROS build-up (Figure 2) that might signal subsequent stress responses (Foyer, Ruban & Noctor 2017; Gómez et al. 2019). There is no strict correlation between ROS levels and the magnitude of gene expression reprogramming, but it is expected that decreasing ROS propagation will result in a weaker response as observed in our microarray assay.
Down-regulation of genes encoding photosynthetic components is an universal feature of water limitation (Evers et al. 2010; Kondráket al. 2012; Zhang et al. 2014; Iovieno et al.2016), which aggravates the direct inhibition of photosynthesis through stomatal closure and acceptor side limitation. Under our conditions, 3 days of treatment were sufficient to cause significant repression of transcripts coding for members of the light-harvesting complexes, the PETC and the Calvin-Benson cycle (Figures 4-6), even though decrease of photosynthetic activity was only moderate at this stage (Figure 1b), presumably reflecting slow turnover of the corresponding proteins. Fld expression provided partial or complete protection against repression. It is worth noting, within this context, that photosynthesis was over-represented as a functional category in clusters 7, 8 and 12 (Figures 5, 6), but not in cluster 2, which harbored those genes whose drought-dependent repression was not affected by Fld presence (Supplementary Figure S5), underscoring the relevance of the flavoprotein for the preservation of photosynthetic activity in the stressed plants.
A most conspicuous functional category among induced genes was, not unexpectedly, stress, enriched in clusters 1, 5, 6 and 11 (Figures 5, 6; Supplementary Figure S5), indicating that Fld mitigated the stress response of the plant in most cases. While abiotic stress concentrated specifically in clusters 1 and 6, biotic stress was over-represented in clusters 5 and 11. They included transcripts encoding PR proteins, which have been defined as gene products induced during pathogen infection or wounding. Members of the PR-6, PR-2 (endo-β-1,3-glucanases) and PR-12 (defensins) families were present in cluster 11, and chitinases in cluster 5.
Genes found in clusters 14-17 responded to Fld presence under control conditions in the same direction as they did in the wild type under drought (Figure 6c). We coined the term priming to describe genetic traits displaying this behavior (Pierella Karlusich et al. 2017). While mitigation of drought responses by Fld might be a consequence of diminished stress sensed by the plant due to productive interaction of the flavoprotein with photosynthesis and other metabolic routes, priming might actually help the plant to better cope with the environmental challenge, thus representing a cause of the increased drought tolerance. Indeed, priming has been reported to substantially contribute to the phenotypes of drought-tolerant barley varieties, which exhibited stressed-like expression patterns in the absence of stress (Janiaket al. 2018). As in our case, drought caused stronger transcriptional changes in sensitive barley genotypes compared to tolerant varieties (Janiak et al. 2018; You et al. 2019; Zenda et al. 2019). Enriched functional categories among primed potato genes included induction of myo-inositol synthesis (Figure 6c). Overexpression of a MIPS-encoding gene enhanced inositol levels and salt stress tolerance in Arabidopsis, tobacco and rice (Tan, Wang, Xiang, Han & Guo 2013; Joshi, Ramanarao & Baisakh 2013; Kusuda et al.2015).
The effects of drought on cell wall metabolism and architecture are complex and depend on the plant species, genotype and age (Le Gallet al. 2015). Figure 4 shows that under the conditions employed, the overall effect of water limitation was repressive, with only marginal protection by Fld. Cluster analysis, in turn, confirmed that most functional categories related to cell wall metabolism grouped in cluster 2 (Supplementary Figure S5), although they were also found in clusters 8, 12 and 13 (Figures 5, 6). In all cases, inclusion in these clusters implied that the relevant genes were down-regulated by drought in the WT, the transformant or both. Genes encoding pectin methylesterases, xyloglucan endotransglucosylases/hydrolases and expansins were extensively represented in these categories (Supplementary Tables S4, S6, S7), indicating that both matrix properties and hemicellulose deposition were compromised by the drought treatment (Tenhaken 2015; Le Gall et al. 2015). The results also suggest that the stress tolerance conferred by Fld expression was not related to protection of cell wall metabolism.
Transcriptional profiles of plants exposed to drought stress have been reported for various crop species including barley, maize, cotton, rice, tomato and potato (Zhang et al. 2014; Iovieno et al. 2016; Janiak et al. 2018; Hasan et al. 2019; Zenda et al.2019; Yang et al. 2019). In those studies, RNA was usually collected at stages of the treatments in which plants already exhibited stress symptoms. Comparison of the functional categories over-represented among DE genes in drought-stressed potatoes with those found in our analysis revealed many similarities such as repression of photosynthetic genes and transcripts associated to signaling by receptor kinases (Evers et al. 2010; Kondrák et al. 2012). In addition, stress-repressed genes involved in cell wall metabolism in our transcriptional profiling were the same as those reported by Evers et al. (2010), and some genes involved in abiotic stress such as heat-shock proteins were also shared in several studies (Evers et al. 2010; Zhang et al. 2014). The results suggest that modulation of these stress-responsive genes was initiated early and maintained as the adverse condition proceeds. Direct comparison of early vs. late drought responses in Arabidopsis further supports this contention (Harb, Krishnan, Ambavaram & Pereira 2010).
Phenotypic effects exerted by chloroplast-located Fld were also evident in the levels of leaf carbohydrates and amino acids. Increased accumulation of transient starch in the absence of stress might result from the higher photosynthetic activity displayed by pfld leaves (Tognetti et al. 2006; Ceccoli et al. 2012; Rossi et al. 2017); its preservation under drought from a lower repression of photosynthetic genes (Figures 5, 6), including starch synthase (Supplementary Table S6; Figure 7). Water limitation led to a strong increase in amino acid levels, most conspicuously Pro, in WT leaves, an effect that was significantly attenuated by Fld presence (Figure 8b). Amino acids synthesized from oxaloacetate as Asn, Asp, Lys, Ile and Thr were induced in WT leaves under drought. A similar response was observed in the 2-oxoglutarate-driven pathway, with increased levels of Pro and Gln only in the WT and Glu in both genotypes under water deprivation. Up-regulation of nitrogen mobilization has been associated with drought and other abiotic stresses (Zhang, Meng, Li & Zhao 2018), with Gln/Glu and Asn/Asp ratios being customarily used as markers of nitrogen cycling (Goel & Singh 2015). In our case, they increased significantly in WT leaves under drought (4.3- and 1.9-fold, respectively); but remained nearly the same in Stpfld 252 plants (Supplementary Figure S10). The results suggest that nitrogen mobilization was involved in this short-term water deficit condition, in agreement with the up-regulation of genes involved in nitrogen uptake and assimilation observed in similar drought assays (Evers et al. 2010; Goel & Singh 2015).
Finally, improvement of physiological and molecular stress responses by Fld presence in chloroplasts resulted in increased tuber yield under a long-term non-lethal water restriction protocol (Figure 9). While most drought- and Fld-dependent effects occurred and were monitored in leaves, it is likely that preservation of photosynthesis and other central metabolic pathways (e .g. , glycolysis and starch synthesis) under water limitation favored production and transport of photosynthates from source to sink tissues. Further research will be necessary to identify the mechanisms by which the better biochemical performance of Stpfld leaves translated into improved tuber yield, and to evaluate the quality of the resulting tubers in terms of nutritional value and organoleptic properties.
Taken together, our results indicate that the Fld technology constitutes a remarkable tool to improve potato production under less-than-optimal conditions. Field trials are required to properly evaluate this possibility and its agronomic relevance.
LEGENDS TO FIGURES
Figure 1. Phenotypes and photosynthetic activities of WT and Fld-expressing potato plants under short-term drought treatment.Thirty-days old plants were exposed to hydric stress under growth chamber conditions by interrupting irrigation. (a) Plants were photographed after 14 days of water withdrawal. (b-d) Photosynthetic parameters Fv/Fm (b), ϕPSII (c) and NPQt (d) were determined at the indicated days of treatment, as described in Materials and Methods. Values are means ± SE of 4-6 biological replicates, and asterisks indicate statistically significant differences with the wild type using ANOVA and Tukey’s Multiple Comparison Test at P ≤ 0.1 (*) or P ≤ 0.05 (**).
Figure 2. ROS accumulation in WT and Fld-expressing potato leaves under short-term drought treatment. ROS were visualized by fluorescence microscopy after leaf infiltration with DCFDA. (a) Confocal microscopy analysis of subcellular ROS accumulation in leaves of 30-days old WT and Stpfld252 plants after 0, 3 and 14 days of water withdrawal. Images show ROS fluorescence (left, green), chlorophyll auto-fluorescence (middle, magenta), and the merge of the two channels (right). Scale bar: 40 μm. Quantification of ROS levels in whole-leaf tissue (b) and chloroplasts (c) of WT and Stpfld252plants. Results are means ± SE of 5 replicates, and asterisks indicate statistically significant differences between lines using ANOVA and Tukey’s Multiple Comparison Test (P ≤ 0.05).
Figure 3. Effect of Fld expression on the potato transcriptome.Upper part: pie charts showing the number of genes that were induced (FC > 2 and FDR < 0.05) or repressed (FC < 0.5 and FDR < 0.05) by Fld in leaves of 30-days oldStpfld252 plants under growth chamber conditions. Lower part: over-representation analysis of Mapman functional among Fld-responsive genes. The analysis was carried out separately for induced and repressed genes (Fisher’s exact test with Bonferroni correction and FDR < 0.05). The list of genes and their corresponding descriptions, functional assignments and FC values are described in Supplementary Table S2.
Figure 4. Drought stress led to extensive transcriptional reprogramming in both WT and Fld-expressing plants.Venn diagrams of genes differentially expressed (DE) in response to drought are shown in the right upper corner. Genes were defined as induced when FC > 2 and FDR < 0.05, and as repressed when FC < 0.5 and FDR < 0.05. On the left, list of functional categories with over-represented DE genes during the response of WT and Stpfld252 plants to drought. The analysis was carried out separately for induced and repressed genes (Fisher’s exact test with Bonferroni correction and FDR < 0.05). The list of genes and their corresponding descriptions, assignments and FC values are described in Supplementary Table S3.
Figure 5. Clusters containing genes in which the drought response was abolished or ameliorated by Fld expression. Each gray line of the charts corresponds to a particular gene, and the dark line represents the average behavior of all the genes contained in each cluster. Labelings in the abscissa correspond to the WT line under control (WT.C) and drought conditions (WT.D), and the Stpfld252line under control (Fld.C) and drought conditions (Fld.D); those in the ordinates correspond to the FC values in log2 scale. For each pairwise comparison between lines or treatments, genes were defined as induced when FC > 2 and FDR < 0.05, and repressed when FC < 0.5 and FDR < 0.05. The total number of genes in each cluster is indicated above the corresponding panel, and theover-represented functional categories are shown below (analyzed using Fisher’s exact test with Bonferroni correction and FDR < 0.05). The list of genes belonging to these clusters and their corresponding descriptions, functional assignments and FC values in the multiple comparison tests are described in Supplementary Table S6.
Figure 6. Clusters containing DE genes with various interactions between treatment and Fld. (a) Genes affected by drought exclusively in Fld-expressing plants. (b) Genes regulated by Fld and drought in opposite directions. (c) Drought-responsive genes which were already primed by Fld in the absence of stress. Each gray line of the charts corresponds to a particular gene, and the dark line represents the average behavior of all cluster genes. Labelings of the abscissa and ordinates, total number of genes in each cluster and list of over-represented pathways are indicated as in Figure 5. Further details are provided in Supplementary Tables S7 (a), S8 (b) and S9 (c).
Figure 7. Comparison of expression patterns of selected DE genes as determined by microarray analysis (dark circles) and qRT-PCR (open circles). Labelings in the abscissa correspond to the WT line under control (WT.C) and drought conditions (WT.D), and theStpfld 252 line under control (Fld.C) and drought conditions (Fld.D); those in the ordinates correspond to the FC values in log2 scale relative to those of WT siblings under control conditions (WT.C). Each data point of qRT-PCR determinations represents the mean and standard deviation of 4-5 biological replicates. Within each line, a statistically significant effect relative to WT.C is indicated as a gray asterisk for microarray data (FDR < 0.05 and FC > 2 or FC < 0.5) and a black asterisk for qPCR data (P values < 0.05 according to ANOVA test). psaK, photosystem I reaction center subunit; Fd1, Fd isoform 1; psbY, photosystem II core complex protein; GBSSI, granule-bound starch synthase; NR3, nitrate reductase 3; ATHB7, Arabidopsis homeobox-leucine zipper protein HB7; ACO1, aminocyclopropane-1-carboxylate oxidase 1; GAST1, gibberellin-regulated protein GAST1; PSA2B, proteasome subunit PSα2β; RPN9b, 26S proteasome subunit RPN9b.
Figure 8. Fld presence attenuates build-up of soluble sugars and amino acids under water limitation. Extracts were prepared from leaves of 30-days old plants at 3 days of water deprivation and from their watered controls (8 h within the light period), and the levels of the indicated sugars and amino acids were determined as described in Materials and Methods. (a) Carbohydrate contents are given as means ± SE of 5-8 independent plants. Statistically significant differences between lines are shown by asterisks and were determined using ANOVA and Tukey’s Multiple Comparison Test (P < 0.05). FW, fresh weight. (b) Heat map of amino acids assayed for the different lines and treatments. Color scale corresponds to the standardized scores (dark blue for low values, yellow for high values). Quantitative data of amino acid determinations are shown in Supplementary Figure S7. C, control conditions; D, drought. Heat maps were produced in R language using the heatmap.2 function of the gplots package.
Figure 9. Phenotypes of WT and Fld-expressing potato plants under long-term drought treatment. Plants were grown in soil for 30 days at 100% FiC. Water irrigation was then interrupted until soil reached 40% FiC, rehydrated to 70% FiC and this protocol repeated for a total treatment of 90 days (Supplementary Figure S2). (a) Plants were photographed after 90 days of drought treatment. (b) Fresh weight of aerial parts. (c) Water potential. (d) Representative photograph of tubers. (e) Tuber yield. Results are means ± SE of 5 replicate plants, and asterisks indicate statistically significant differences between lines using non-parametric Mann-Whitney test (P ≤ 0.1).