Drought-Induced Regulatory Cascades and Their Effects on the Nutritional Quality of Developing Potato Tubers

Competition for scarce water resources and the continued effects of global warming exacerbate current constraints on potato crop production. While plants’ response to drought in above-ground tissues has been well documented, the regulatory cascades and subsequent nutritive changes in developing tubers have been largely unexplored. Using the commercial Canadian cultivar “Vigor”, plants were subjected to a gradual drought treatment under high tunnels causing a 4 °C increase in the canopy temperature. Tubers were sampled for RNAseq and metabolite analysis. Approximately 2600 genes and 3898 transcripts were differentially expressed by at least 4-fold in drought-stressed potato tubers, with 75% and 69% being down-regulated, respectively. A further 229 small RNAs were implicated in gene regulation during drought. Expression of several small RNA clusters negatively correlated with expression of their six target patatin genes, suggesting involvement in the regulation of storage proteins during drought. The comparison of protein homologues between Solanum tuberosum L. and Arabidopsis thaliana L. indicated that down-regulated genes were associated with phenylpropanoid and carotenoid biosynthesis. As is indicative of reduced flow through the phenylpropanoid pathway, phenylalanine accumulated in drought-stressed tubers. This suggests that there may be nutritive implications to drought stress occurring during the potato tuber bulking phase in sensitive cultivars.


Introduction
Potatoes are the fourth most consumed food crop worldwide and are an efficient source of energy, vitamins and minerals in the human diet [1]. High consumption rates and moderate concentrations of dietary antioxidants have led potatoes to be the third-largest source of total phenolics in the American diet [2]. Diets rich in phenolics have been implicated in the prevention of an array of degenerative diseases and concentrations of these compounds vary greatly based on the cultivar, highlighting the potential for the targeted breeding of potato (S. tuberosum L.) to enhance global human health [3]. In potato tubers, the primary polyphenol is chlorogenic acid with the remaining components Saskatchewan. The cultivar "Vigor", a cross between "Agria" and "Wischip" from the Agriculture and Agri-Food Canada Lethbridge Research and Development Centre, was evaluated for its performance under soil moisture stress during the tuber bulking phase. Prominent characteristics of the cultivar are its yellow-fleshed tubers and pigmented (red-violet) flowers. Plants were grown under optimum soil moisture conditions at 70% field capacity (FC) and restricted soil moisture conditions at 35% FC under two high tunnels using drip irrigation. Treatments were imposed at the start of the tuber bulking phase for gradual exposure to drought stress, thereby mimicking natural field conditions (Figure 1). High tunnels were opened from all sides but covered with plastic film on top to mimic the open field condition while preventing rainfall (Supplementary Figure S1A). Plots were laid out in a randomized complete block design containing four replicates with guard rows on either side. Each plot consisted of 12 hills. The two end hills were considered as guard hills for yield estimation purposes. Seed pieces were spaced 1 m between-rows and 20 cm within-rows and were planted on 30 May 2017. The crop was raised using standard management practices (i.e., fertility, irrigation, pest control, etc.). Pre-plant basal fertilizer included urea (46-0-0), mono-ammonium phosphate (11-52-0), and potash (0-0-60). Two applications of ammonium sulphate (21.5-0-0-4) were given at 4 and 7 weeks after planting. Soil moisture was monitored using Watermark sensors (Supplementary Figure S1B). Plots were harvested on 2 October 2017 and graded according to commercial-grade standards.
Genes 2020, 11, x FOR PEER REVIEW 3 of 13 541 m), Saskatchewan. The cultivar "Vigor", a cross between '"Agria" and "Wischip" from the Agriculture and Agri-Food Canada Lethbridge Research and Development Centre, was evaluated for its performance under soil moisture stress during the tuber bulking phase. Prominent characteristics of the cultivar are its yellow-fleshed tubers and pigmented (red-violet) flowers. Plants were grown under optimum soil moisture conditions at 70% field capacity (FC) and restricted soil moisture conditions at 35% FC under two high tunnels using drip irrigation. Treatments were imposed at the start of the tuber bulking phase for gradual exposure to drought stress, thereby mimicking natural field conditions (Figure 1). High tunnels were opened from all sides but covered with plastic film on top to mimic the open field condition while preventing rainfall (Supplementary Figure S1A). Plots were laid out in a randomized complete block design containing four replicates with guard rows on either side. Each plot consisted of 12 hills. The two end hills were considered as guard hills for yield estimation purposes. Seed pieces were spaced 1 m between-rows and 20 cm within-rows and were planted on 30 May 2017. The crop was raised using standard management practices (i.e., fertility, irrigation, pest control, etc.). Pre-plant basal fertilizer included urea (46-0-0), mono-ammonium phosphate (11-52-0), and potash (0-0-60). Two applications of ammonium sulphate (21.5-0-0-4) were given at 4 and 7 weeks after planting. Soil moisture was monitored using Watermark sensors (Supplementary Figure S1B). Plots were harvested on 2 October 2017 and graded according to commercial-grade standards. Figure 1. Visualization of potato growth stages and in-season soil moisture trends when soil moisture was maintained at 35% and 70% field capacity (FC).

Physiological Measurements
Physiological measurements were taken at the end of the tuber bulking phase (90th day after planting) with readings recorded consistently between 11:30 a.m. and 12:30 p.m. Canopy temperature was assessed with an infra-red thermal imaging camera (FLIR T530, FLIR Systems. Wilsonville, OR, USA), and leaflet chlorophyll content (CCI) was recorded using a chlorophyll content meter (CCM-200-Apogee Instruments, Logan, UT, USA). Quantum yield of dark-adapted leaflets (Fv/Fm) was measured using the portable fluorometer FluorPen FP 100 (PSI, DRASO Czech Republic). Detachable clips were used to dark-adapt the leaflets for 20 min, and Fv/Fm was measured on the adaxial surfaces of the top 3rd and 4th leaflet of each sampled plant (three plants per replication).

Physiological Measurements
Physiological measurements were taken at the end of the tuber bulking phase (90th day after planting) with readings recorded consistently between 11:30 a.m. and 12:30 p.m. Canopy temperature was assessed with an infra-red thermal imaging camera (FLIR T530, FLIR Systems. Wilsonville, OR, USA), and leaflet chlorophyll content (CCI) was recorded using a chlorophyll content meter (CCM-200-Apogee Instruments, Logan, UT, USA). Quantum yield of dark-adapted leaflets (Fv/Fm) was measured using the portable fluorometer FluorPen FP 100 (PSI, DRASO Czech Republic). Detachable clips were used to dark-adapt the leaflets for 20 min, and Fv/Fm was measured on the adaxial surfaces of the top 3rd and 4th leaflet of each sampled plant (three plants per replication).

Amino Acid Profiling and Abscisic Acid Content
In both high tunnels from each replication, a tuber was collected from the middle of the plot on 13 September 2017 (106 DAP). The tubers were washed in running water, followed by distilled water, cut into cubes while evading the skin and immediately frozen in liquid nitrogen. Samples were stored at −80 • C until further use.
Amino acids were extracted from 10 mg of ground freeze-dried tissue following Inaba et al. (1994) [19] with some modifications. Briefly, 1 mL of 80% (v/v) ethanol solution (40 • C) was added to each sample, shaken for 30 min at 40 • C and the supernatant was recovered by centrifugation (4000 rpm for 10 min) at 4 • C. The pellets were re-extracted under the same conditions with an additional 500 µL of 80% (v/v) ethanol (40 • C). The supernatants were combined and stored at −20 • C until further use. Amino acids were derivatized following Waters AccQTag Reagent Kit (Waters, Milford, MA, USA; [20]). Briefly, a 10 µL aliquot of sample was mixed with 70 µL borate buffer and 20 µL AccQFluor reagent which was reconstituted in acetonitrile. AccQFluor reagent was reconstituted as follows: 1 mL of AccQFluor reagent diluent was transferred to a vial containing AccQFluor reagent powder and vortexed for 10 s before heating at 55 • C for a maximum of 10 min or until dissolved. The derivatized mixture was transferred to an autosampler vial and incubated at 55 • C for 10 min. High-performance liquid chromatography (HPLC) was conducted, as described in Waters AccQTag's chemistry package instruction manual, with samples separated on a Waters amino acid column −3.9 × 150 mm and quantified at an excitation wavelength of 285 nm and an emission wavelength of 320 nm using a 2475 scanning fluorescence detector (Waters, Milford, MA, USA). The column was set at 37 • C with a 5 µL injection volume. Waters AccQTag buffer (100 mL AccQTag Buffer concentrate +1000 mL deionized water), acetonitrile, and deionized water were used as mobile phases A, B, and C, respectively.
Abscisic acid content was determined following Yan et al., (2016) [21]. Samples were centrifuged to remove debris, and the pellet was washed twice. The supernatant was evaporated in a SpeedVac, and reconstituted in 1 mL of 1% (v/v) acetic acid. Abscisic acid (ABA) was purified by solid-phase extraction using Oasis HLB, MCX, and WAX cartridge columns (Waters, Milford, MA, USA). The solvent was removed under vacuum and subjected to LC-ESI-MS/MS analysis (Agilent 6410 TripleQuad LC/MS system). An LC (Agilent 1200 series) equipped with a 50 × 2.1 mm, 1.8-µm Zorbax SB-Phenyl column (Agilent Technologies, Santa Clara, CA, USA) was used with a binary solvent system comprised of 0.01% (v/v) acetic acid in water (solvent A) and 0.05% (v/v) acetic acid in acetonitrile (solvent B). Separations were performed using a gradient of increasing acetonitrile content at a flow rate of 0.2 mL min −1 . The gradient was increased linearly from 3% B to 50% B over 15 min. The retention time of ABA was 14 min.

Transcriptome and Small RNA Sequencing
Total RNA was extracted from 100 mg of tuber tissue partitioned from the sample taken for metabolite analysis using the RNeasy plant mini kit (Qiagen, Hilden, Germany). RNA quality and concentration were verified using an Agilent Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). TruSeq RNA and small RNA sequencing libraries were constructed following the standard preparation guide (Illumina, San Diego, CA, USA). All eight RNA samples (four replicates of each treatment) were multiplexed in a lane of a flow cell and paired-end sequencing (125 cycles) was performed using an Illumina HiSeq 2500. Similarly, for small RNA sequencing, all 8 samples were multiplexed in a lane of a flow cell and single-end sequencing was carried out on Illumina HiSeq 2500.

RNA and Small RNA Read Mapping and Analysis
Before read mapping and expression quantification, all RNA reads were filtered using Trimmomatic (version 0.36; [22]) by (i) removing adapter sequences, (ii) trimming leading and trailing low-quality sequences, (iii) removing sequences when the average quality per base dropped below 15 within a 4-base wide sliding window and (iv) keeping only those pairs where both reads were longer than 75 bp. Clean reads were aligned to the potato reference genome (SolTub_3.0, EnsemblPlants) with STAR (v2.5.2b) and isoform expression was quantified with the RSEM (v1.3.3) algorithm [23]. The expected read counts generated by the RSEM algorithm were rounded off and fed into DESeq2.

Differential RNA Expression Analysis
Raw read counts obtained from RNAseq were normalized and assessed for differential expression using the Statistical Software "R" version 3.6.0 and the package DESeq2 [28,29]. Log2 fold change threshold of 2 and a 5% false discovery rate (FDR) were used as cut-off values for continuing to the annotation step. R scripts are available on Bioconductor (https://bioconductor.org/) from the package developers and were adapted for the data presented in this paper. The same technique was repeated for the discovery of differentially expressed small RNA, with target gene identification done using psRNATarget [27]. Gene annotations for S. tuberosum and A. thaliana were obtained from the Ensembl Plants database (http://plants.ensembl.org). Arabidopsis homologs with >50% identity to the original potato gene were input into the online DAVID Bioinformatics Resources version 6.8 (https://david.ncifcrf.gov/) for functional annotation clustering and KEGG pathway mapping analyses.

Physiological Response
A high degree of variability existed within the four control and four treatment plots for the agronomic and physiological traits measured during this study (Table 1). Both crop yield and tuber number per plot were not found to be significantly different from one another, although differences in plot averages were observed. Calculated values from spectral measurements, such as CCI and Fv/Fm, showed no significant differences. However, canopy temperatures measured in plots maintained at 35% FC were 3.9 • C higher than control plots with soil moisture maintained at 70% FC (Table 1). ABA concentrations in the well-watered control and water-deficit conditions were 43.9 and 55.4 ng gDW −1 , respectively. Variation was high among the control plants, resulting in no significant differences between treatments.

Tuber Amino Acid Fluctuations in Response to Soil Moisture Deficit
Of the eight essential amino acids, lysine, phenylalanine, isoleucine, and leucine were found to be more abundant in drought-stressed tubers. The largest differences were observed in the concentrations of leucine, phenylalanine, and isoleucine which increased by 3×, 2×, and 1.9×, respectively (Figure 2A). Quantities of branched-chain amino acids, a group that includes leucine, isoleucine, and valine, were therefore significantly higher under drought treatment. Histidine and valine were the most abundant essential amino acids in developing potato tubers (Figure 2A). The majority of non-essential amino acids had similar concentrations in developing tubers regardless of the treatment. Only glutamic acid showed a marked increase of 5.75 µmol g −1 under reduced soil moisture conditions. Concentrations of cysteine, proline, and serine were highest among all amino acids measured ( Figure 2B).
Genes 2020, 11, x FOR PEER REVIEW 6 of 13 the treatment. Only glutamic acid showed a marked increase of 5.75 µmol g −1 under reduced soil moisture conditions. Concentrations of cysteine, proline, and serine were highest among all amino acids measured ( Figure 2B).

Figure 2.
Concentrations of essential (A) and non-essential (B) amino acids in potato tubers (n = 4) sampled during the tuber bulking phase and the associated SEM (p < 0.05) when subjected to 70% and 35% field capacity.

Differential Gene Expression and Regulatory Cascades in Developing Tubers Under Drought Stress
Tubers subjected to soil moisture deficit showed both differential gene and transcript expression, with 75.2% and 68.8% being down-regulated, respectively. One-fifth of genes with differential sampled during the tuber bulking phase and the associated SEM (p < 0.05) when subjected to 70% and 35% field capacity. * significance (p < 0.05) between treatments.

Differential Gene Expression and Regulatory Cascades in Developing Tubers Under Drought Stress
Tubers subjected to soil moisture deficit showed both differential gene and transcript expression, with 75.2% and 68.8% being down-regulated, respectively. One-fifth of genes with differential expression were of unknown function. A full summary of the observed changes is listed in Figure 3A with a list of all differentially expressed genes provided in Supplementary Table S1. Down-regulated genes include those with functions in ABA, auxin and ethylene signaling as well as in auxin, carotenoid, and phenylpropanoid biosynthesis. Up-regulated genes have roles in amino acid biosynthesis, function as molecular chaperones and are involved in ubiquitin-driven proteolysis. Gene names, functional annotation, and the corresponding Arabidopsis homologs used for pathway mapping can be found in Table 2. Focusing on the regulation of the phenylpropanoid pathway, fifteen annotated MYB transcription factors were down-regulated by more than 4-fold under low-soil moisture conditions. Up-regulated transcription factors include two MYB transcription factors and one LIM transcription factor. No other MYB, KNOX or LIM transcription factors were above the cut-off values of 5% FDR and a log-fold change greater than two. Genes with key functions in the phenylpropanoid and carotenoid pathways that were down-regulated in potato tubers at 35% FC are highlighted in Figure 4. Raw read counts for gene expression analysis can be found in Supplementary Table S2. with a list of all differentially expressed genes provided in Supplementary Table S1. Down-regulated genes include those with functions in ABA, auxin and ethylene signaling as well as in auxin, carotenoid, and phenylpropanoid biosynthesis. Up-regulated genes have roles in amino acid biosynthesis, function as molecular chaperones and are involved in ubiquitin-driven proteolysis. Gene names, functional annotation, and the corresponding Arabidopsis homologs used for pathway mapping can be found in Table 2. Focusing on the regulation of the phenylpropanoid pathway, fifteen annotated MYB transcription factors were down-regulated by more than 4-fold under low-soil moisture conditions. Up-regulated transcription factors include two MYB transcription factors and one LIM transcription factor. No other MYB, KNOX or LIM transcription factors were above the cutoff values of 5% FDR and a log-fold change greater than two. Genes with key functions in the phenylpropanoid and carotenoid pathways that were down-regulated in potato tubers at 35% FC are highlighted in Figure 4. Raw read counts for gene expression analysis can be found in Supplementary  Table S2.
The isolated small RNA were grouped into 87,213 clusters with an additional 10,209 unassigned sequences. Of these, 103 clusters and 126 unassigned sequences were differentially expressed. Additional summary statistics are listed in Figure 3B. Differentially expressed small RNA clusters with identified gene targets are listed in Supplementary Table S3. None of the small RNA clusters with target genes listed in Table 2 showed differential expression between the two treatments. Expression of target MYB transcription factor genes was also not correlated to small RNA cluster expression (Supplementary Table S4). Interestingly, the expression of small RNA clusters primarily targeting patatin genes was significantly up-regulated and negatively correlated to target gene expression (r = -0.61). These clusters and their targets can be found in Table 3.  thaliana were considered if identity was greater than 50%. Functional annotation clustering to KEGG pathways was based on Arabidopsis gene IDs using the DAVID Bioinformatics online resource 6.8 (https://david.ncifcrf.gov/). Table 2. List of differentially expressed genes in drought-stressed potato tubers, the corresponding homologs in A. thaliana, and the pathway in which they participate. Pathway mapping was done based on the Arabidopsis gene names and similarities between the original S. tuberosum gene and its homolog are expressed as the percentage of identical base pairs in the gene sequences (% ID).   significantly suppressed under drought (Supplementary Table S1). Unlike previous findings in potato leaves [10], expressions of small RNA and their target MYB transcription factors were not correlated in the drought-stressed tubers. There was therefore no evidence to suggest that small RNA played a role in regulating the phenylpropanoid or carotenoid pathways under drought conditions in potato tubers (Supplementary Table S4).

Conclusions
Potato is among one of the most important food crops, yet maintaining plant productivity in this drought-sensitive crop has become a challenge. From a nutrition perspective, decreasing soil water availability during tuber filling, as a function of a warming climate or as a production practice to induce senescence for an earlier harvest, may lead to a reduction in tuber quality. While mild drought increases the proportion of essential amino acids, potential losses in protein and phenolic content would outweigh the benefit. While MYB transcription factors may be targeted to reduce effects on the phenylpropanoid pathway, identification of small RNA as the regulator of patatin gene expression suggests it may be difficult to maintain patatin expression in drought-susceptible cultivars using current breeding techniques.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1; Figure S1: Experimental conditions;   The isolated small RNA were grouped into 87,213 clusters with an additional 10,209 unassigned sequences. Of these, 103 clusters and 126 unassigned sequences were differentially expressed. Additional summary statistics are listed in Figure 3B. Differentially expressed small RNA clusters with identified gene targets are listed in Supplementary Table S3. None of the small RNA clusters with target genes listed in Table 2 showed differential expression between the two treatments. Expression of target MYB transcription factor genes was also not correlated to small RNA cluster expression (Supplementary Table S4). Interestingly, the expression of small RNA clusters primarily targeting patatin genes was significantly up-regulated and negatively correlated to target gene expression (r = −0.61). These clusters and their targets can be found in Table 3.

Discussion
Optimal potato tuber growth occurs around 20 • C and plants are susceptible to losses in productivity under hot, arid conditions. Such conditions are expected to increase in the coming decade, therefore functional indicators of plant stress and the cascading effects on the developing tubers were evaluated. In this study, the Canadian potato cultivar "Vigor" was gradually exposed to increasing water deficit to a level of 35% FC beginning at the start of the tuber bulking phase ( Figure 1). As seen previously in the literature, fluorescent measurements were not distinguishable between treatments [15] and thus were not dependable indicators of drought stress in potato plants (Table 1). At the plot level, there were no significant differences in yield or tuber number between the treatments (Table 1), however, discrepancies could become more prominent in commercial field production. Canopy temperature was considerably elevated in the drought treatments and has evidence supporting its use for drought stress assessments [30]. Average concentrations of ABA trended upwards in potato tubers exposed to water deficit compared to well-watered controls, suggesting drought responses had been initiated. Gene expression data further corroborated that drought signaling pathways had been activated as there was marked down-regulation of an ABA receptor PYR1, down-regulation of a series of small auxin up-regulated RNA (SAUR) genes involved in cell expansion and organ elongation in response to the environment [31] and the up-regulation of heat shock factor proteins [32] (Table 2). Notably, outside of common changes to the regulation of heat shock proteins, several genes previously identified as differentially expressed in severely drought-stressed potato leaves [32] were inversely regulated in the mildly stressed potato tubers collected in this study. These include the WRKY transcription factor (PGSC0003DMG400001434) and the developmental gene UPA16 (PGSC0003DMG400031742) [32] which had 2-fold and 78-fold increases in expression compared to well-watered control tubers. Lists of genes implicated in drought stress response in potato leaf tissue [33] and potato stolons [34] have been compiled and here we provide those in developing tubers (Table 2; Supplementary Table S1).
Metabolic effects of drought on parameters such as free amino acids, soluble protein, and phenolics were assessed. The gradual drought stress to which the potato tubers were exposed resulted in no significant differences in total free amino acid concentrations, although treatment averages appeared to be divergent with 154.1 and 207.2 µmol g −1 in the control and drought treatments, respectively. These differences could be associated with the up-regulation of genes involved in proteolysis (Table 2). Elevated concentrations of proline have been shown to indicate stress in potato leaves [35], however, similar concentrations were observed in tubers irrespective of treatment. The largest changes occurred in the amino acid profile, where concentrations of branched-chain amino acids leucine and isoleucine increased ( Figure 2). This indicated a greater proportion of dietary essential amino acids. Increases in branched-chain amino acids are likely attributable to the up-regulation of acetolactate synthase ( Table 2), which is the first enzyme in the branched-chain amino acid synthesis pathway [36].
A major fraction (up to 40%) of the soluble protein in potato tubers consists of glycoproteins, known as patatins, that act both as storage proteins and show activity as non-specific lipid acyl hydrolases (LAH) with potential roles in plant defense against biotic stressors [37,38]. In the case of abiotic stress, it was observed that five patatin genes were down-regulated by at least 16-fold with the regulation of gene expression likely occurring via an increased presence of small RNA (Table 3). A possible consequence is reduced protein content in the resulting potato tubers.
As one of the major sources of plant phenolics in the human diet, potatoes have been targeted in breeding for greater total phenolics and antioxidant capacity [3,39]. Phenolic content is known to show a high degree of environmental plasticity with cooler temperatures during the growing period and storage, attributed to higher average accumulation [4,5]. Under drought conditions, expressions of key enzymes required for the biosynthesis of anthocyanins (DFR), flavonoids (FLS) and chlorogenic acid (HCT) were drastically reduced ( Figure 4A). Initial flow into the phenylpropanoid pathway through PAL was also reduced, leading to the accumulation of phenylalanine observed in Figure 2. Similar results have been previously observed in the literature [5]. Key enzymes of the carotenoid pathway were also downregulated ( Figure 4B). Environmental conditions leading to the repression of phenolic biosynthesis could minimize gains achieved in breeding programs. Regulation of the phenylpropanoid pathway can occur via MYB transcription factors [8], fifteen of which were significantly suppressed under drought (Supplementary Table S1). Unlike previous findings in potato leaves [10], expressions of small RNA and their target MYB transcription factors were not correlated in the drought-stressed tubers. There was therefore no evidence to suggest that small RNA played a role in regulating the phenylpropanoid or carotenoid pathways under drought conditions in potato tubers (Supplementary  Table S4).

Conclusions
Potato is among one of the most important food crops, yet maintaining plant productivity in this drought-sensitive crop has become a challenge. From a nutrition perspective, decreasing soil water availability during tuber filling, as a function of a warming climate or as a production practice to induce senescence for an earlier harvest, may lead to a reduction in tuber quality. While mild drought increases the proportion of essential amino acids, potential losses in protein and phenolic content would outweigh the benefit. While MYB transcription factors may be targeted to reduce effects on the phenylpropanoid pathway, identification of small RNA as the regulator of patatin gene expression suggests it may be difficult to maintain patatin expression in drought-susceptible cultivars using current breeding techniques.