Plant material and growth conditions

A total of 20 clonally propagated one-year-old plants from the drought tolerant almond-peach hybrid [P. amygdalus Batsch, syn P. dulcis (Mill.) x P. persica (L.) Batsch], ‘Garnem’, were used for the experiment. The plants were acquired from Agromillora Iberia S.L. nursery (Barcelona, Spain). Prior to the drought experiment, the plants were placed in 5-cm diameter pots with a mix of peat, 30% coconut fiber and 20% sand and maintained in a greenhouse at CITA facilities in Zaragoza, Spain (41°43’28.6”N, 0°48’31.1”W). Plants were watered three times a week and fertilized monthly with 15:9:10 N:P:K + 0.2% MgO (Nitricol). Greenhouse temperatures during the growth period ranged from 28 °C to 18 °C (day / night), with a 12 h day / 8 h night photoperiod (S1 Fig).

Stress conditions and treatment

The plants were divided into control (n = 12 plants) and treatment groups (n = 8 plants). The control plants were maintained under optimal watering conditions, until field capacity, in their 5-cm diameter pots with the initial substrate during the experiment. In this study, drought stress conditions were simulated by providing osmotic stress to the treatment group of plants. The roots of the plants to be treated were placed in a dialysis membrane containing a peat moss substrate and then were submerged in a Polyethylene glycol, PEG-6000, solution (Sigma-Aldrich, Co. St. Luis, MO, USA) (500 g l^-1), corresponding to an osmotic pressure of -2,68 MPa as per the Michel and Kaufmann equation [28] (S1 Fig). As the plants acclimatized to the PEG solution, the plant osmotic potential was controlled using a Scholander-type pressure chamber (Soil Moisture Equipment Corp. Santa Barbara, CA, USA) [29] until day 7 (S1 Table). The 0-hour time point for the experiment began at the termination of the 7-day acclimation period. Root samples were harvested at 0 h (four control plants), 2 h (four each of control and treatment plants) and 24 h (four each of control and treatment plants) and flash frozen in liquid nitrogen prior to being transferred to storage at -80 °C for subsequent RNA extraction.

Plant water status

Leaf Water Potential (LWP) was measured in duplicate for each plant (two leaves, each leaf as one biological replicate) using a Scholander-type pressure chamber (Soil Moisture Equipment Corp. Santa Barbara, CA, USA) [29]. Stomatal conductance (gs) was measured in duplicate for each plant with a Leaf Porometer (Decagon Devices Inc. Pullman, WA, USA). Relative Water Content (RWC) was measured in duplicate (using the same leaves were previously used for LWP measurement) as per previously published method [30]. Briefly, three 1-cm diameter leaf discs were weighed (W) and rehydrated to their turgid weight (TW) by floating them in petri plates containing deionized water for 4 h at room temperature. The dry weight (DW) was obtained after 24 h at 80 °C in an oven. RWC was calculated following the Eq (1): (1)

Electrolyte Leakage (EL) was calculated from Cell Membrane Stability (CMS) rate. CMS was evaluated in duplicate (two leaves, each leaf as one biological replicate) following previously published protocol [31]. Briefly, three 1-cm diameter leaf discs, previously cleaned twice with deionized water to remove surface-bound electrolytes, were submerged in a 50 ml vial containing 10 ml of deionized water and incubated in the dark for 24 h at room temperature. Conductance was then measured with a conductivity meter (CRISON micro CM 2201, Barcelona, Spain). This measurement was taken for control (C1) and treated (T1) samples. After the measurement, the sample vials were autoclaved for 15 min at 121 °C. When the samples reached room temperature, a second reading was recorded for control (C2) and treated (T2) samples. CMS and EL were calculated according the following formulas (2) and (3): (2) (3)

Each of the parameters described above were measured and recorded at 0, 2 and 24 h for both treatments.

RNA isolation, cDNA library construction and sequencing

Total RNA was extracted from 0.5 g of root tissue for each time point using the CTAB method described previously [32] with minor modifications [33–35]. Extracted RNA was quantified using a NanoDrop ND-1000 UV-vis spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). RNA integrity was verified by electrophoresis on a 1% agarose gel. Contaminating genomic DNA was removed using DNase I (TURBO DNA-free Ambion, Life Technologies) per the manufacturer’s instructions. Samples were submitted to Lifesequencing S.L. (Paterna, Valencia, Spain) for RNAseq library preparation and sequencing. A total of 1 μg of cellular RNA (RNA integrity number (RIN) > 7.6) was used for TruSeq RNA library construction (Illumina Inc. San Diego, CA, USA). The mRNA was purified using Oligo(dT) cellulose, and was subsequently fragmented into short pieces. First and second-strand cDNA were synthesized using the fragmented RNA as template. Following purification with the QiaQuick PCR extraction kit (Qiagen, Venlo, The Netherlands), sequencing adapters with identification barcodes were ligated to the fragments in order to distinguish different samples. Fragments with lengths of 200–300 bp were purified by Ampure XP beads (Beckman Coulter, Brea, CA, USA), and selectively amplified via PCR in the final step of the library preparation. A total of 10 libraries were sequenced using an Illumina HiSeq 2000 configuration 100 PE (Illumina Inc. San Diego, CA, USA). The libraries represented the following samples: 0 h control (2 biological replicates), 2 h control and 2 h stressed (2 biological replicates for each treatment), and 24 h control and 24 h stressed (2 biological replicates for each treatment) (S2 Table).

RNAseq data processing

The Illumina HiSeq generated DNA sequence reads in the 2x100 paired format. The resulting fastq files were imported into the CLC Bio Genomics Workbench 6.0.1(CLC Bio, Aarhus, Denmark) for quality assessment, pre-processing, and assembly. Contigs with less than 2x coverage and / or less than 200 bp in length were filtered out. The original, non-trimmed reads from each individual dataset, were mapped back to the master transcriptome assembly in order to count the number of individual sample reads per contig (S2 Table). The master transcriptome was then exported as a fasta file for downstream functional annotation, and the read counts for each dataset were exported and normalized via Reads Per Kilobase per Million reads (RPKM) method [36]. Finally, RPKM values were compared between drought and stressed treatments, using the 0-hour control as a baseline. Thereafter, the RPKM values used for differential expression analysis were derived from the total read count in a pairwise comparison of treatments (drought and control conditions, and 2 h and 24 h of treatment condition). Only genes with a log10 fold change > 5 and p-value < 0.05 were selected for further analysis (S3 Table).

Functional annotation, pathway analysis and GO enrichment

Gene Ontology (GO) annotation was conducted using the Blast2GO v. 3.3 [37]. Differentially expressed genes in each of the treatment comparisons were functionally annotated using the Blast2GO functional genomics suite using the default parameters. The ontology annotations were refined using InterPro Scan and expanded using ANNEX. GoSlim was used as an additional annotation step to summarize the resulting information. Furthermore, the KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis was performed to map differentially expressed, annotated transcripts to respective metabolic pathways. GO enrichment was performed using the two-tailed Fisher’s exact test (FDR < 0.05) in order to reveal the over- and under-represented functions during PEG-induced drought stress.

Quantitative real time PCR validation of DEGs

RNA samples (2,500 ng) from root tissue were reverse transcribed with SuperScript III First-Strand Synthesis System (Invitrogen, Life Technologies, Carlsbad, CA, USA) in a total volume of 21 μl according to the manufacturer’s instructions. Primers were designed for 18 randomly selected differentially expressed genes (DEGs) (S4 Table) using Primer3Plus software [38] with the corresponding transcriptome contig as the query sequence for each primer pair. The amplification of the target regions was evaluated using genomic DNA from ‘Garnem’ genotype as template for quality assurance of the primer pairs [39]. Two microliters of 40X dilution of the synthesized cDNA was used for each amplification reaction in a final volume of 10 μl. qRT-PCR was performed in triplicate for each of the two biological replicates on an Applied Biosystems 7900HT Fast PCR System using iTAQ Universal SYBR Green Supermix (Bio-Rad, Hercules, CA, USA). The amplification conditions consisted of an initial denaturation at 95 °C for 10 min, followed by 40 cycles of 15 s at 95 °C for denaturation, and 1 min at 60 °C for annealing and extension. Amplification was preceded by melting curve analysis. Primers for a translocation elongation factor gene (TEF2), designed from the available P. persica TEF2 DNA sequence (Gene Bank accession number TC3544), were used as an internal reference control reaction for the qRT-PCR experiments. Relative expression was quantified using the -2ΔΔCt method [40].

Statistical analysis

Availability of supporting data

The raw sequencing data from RNAseq analysis were deposited in the NCBI Sequence Read Archive (SRA) under the accession number SRP134116.

Physiological responses to PEG-induced drought

The LWP values at the 2 h and 24 h time points reached -0.76 and -0.47 MPa in control plants and -1.30 MPa and -1.15 MPa in PEG-treated ‘Garnem’ plants, respectively (Fig 1A). The LWP was significantly different between the control and treated plants at both 2 h and 24 h time points. The PEG-induced symptoms produced a significant difference in LWP compared to the 0 h time point (Fig 1A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Leaf water potential (LWP) (A), stomatal conductance (gs) (B), relative water content (RWC) (C) and electrolyte leakage (EL) (D) during the drought experiment for control and treated plants of ‘Garnem’.

Continuous lines indicate well-watered plants, while dash lines indicate stressed plants. (h = hours. Error bars represent the standard error of the mean. Asterisks represent significant differences (p ≤ 0.05) between treatments (control and stressed) for each time point of the experiment.

https://doi.org/10.1371/journal.pone.0205493.g001

No significant differences were observed in the stomatal conductance, gs, between control and treated plants at any of the time points (Fig 1B); however, there was a notable decrease in the gs values from 60.57 mmol m^-2 s^-1 to 29.85 mmol m^-2 s^-1 between 2 h and 24 h time points in the PEG-treated plants (Fig 1B), indicating that the osmotic stress induced stomatal closure. Changes in stomatal conductance from 2 h in the PEG-treated plants are consistent with published literature that reports stomatal closure in drought stress conditions [41,42].

The RWC decreased in the PEG-treated plants throughout the course of the experiment. Significant differences were observed between the control and the treated samples at 24 h, at which the PEG-treated plants reached a minimum RWC value of 84.71% (Fig 1C). As RWC is a good parameter by which cellular water deficit can be extrapolated, these results indicate that the treated ‘Garnem’ samples are in fact experiencing drought stress with respect to the control, particularly at the 24 h timepoint, where the difference is significant. Although RWC decreased at the 24 h time point following PEG treatment, the electrolyte leakage (EL) rate was not significantly affected (1D). Furthermore, EL rates of the PEG-treated plants remained similar to control plants even at low LWP values. While no direct measurement of osmotic potential was conducted, the lack of change in EL indirectly indicates that PEG-mediated osmotic stress may have induced a stress avoidance strategy in the plants mediated by accumulation of solutes leading to an osmotic adjustment under stress conditions [9,42].

The significant changes in physiological parameters associated with stress lend support to the first part of our hypothesis that PEG treatment will induce drought stress in plants. In order to assess the second part of our hypothesis, tissues from plants harvested at two time points that represent the early stages of response to water limiting conditions were subjected to a time-course RNAseq analysis followed by biochemical pathway analysis to identify the underlying metabolic and genetic components.

Assembly of the time-course RNAseq data and identification of differentially expressed contigs (DECs)

As result of RNAseq analysis, 10 sequenced libraries were obtained. These libraries represented the 0 h, 2 h and 24 h time points of both control and drought-stressed conditions for each one of the two biological replicates of Garnem’ roots studied. An approximate mean Q score of 36 for each library validated the quality of the assay (S2 Table).

In total, approximately 0.42 billion reads, each 100 nucleotides long, were generated, of which 96% (0.4 billion reads) were retained after trimming and filtering low quality reads. Mapping of the original, untrimmed reads from each individual condition and time point back to the master assembly generated 117,356 (79.4%); 140,041 (94.8%); 121,596 (82.3%); 131,251 (88.8%); 138,682 (93.9%) contigs for the 0 h control, 2 h control, 2 h stress, 24 h control and 24 h stress time points, respectively (S2 Table) with a mean contig size of 522 bp. The RPKM values were calculated and used to identify the contigs that were differentially expressed by logFC > 5 in each pairwise comparison (S3 Table).

Four pairwise comparisons of expression values from control and stressed samples at different time points were performed:

1. 2 h Stressed vs. 24 h Stressed (2hS-24hS), which compares changes in transcript expression between PEG treatment time points in order to identify drought-responsive contigs;
2. 2 h Control vs. 2 h Stressed Normalized to 0 h control (2hC-2hSN), which allowed for identification of contigs that were differentially expressed during the first 2 h of drought;
3. 24 h Control vs. 24 h Stressed Normalized to 0 h control (24hC-24hSN), which allowed for identification of genes that were differentially expressed after one day of stress; and
4. 2 h Stressed vs. 24 h Stressed Normalized to 0 h control (2hS-24hSN), which enabled the identification of contigs that changed in expression as a result of PEG addition.

There was increased transcriptional activity at the 2 h stressed time point in comparison with that of the 24 h stressed time point, with the highest number of differentially expressed contigs (DECs) present exclusively within the 2hC-2hSN pairwise comparison (Fig 2). Notably, only 0.3% of DECs exhibited differential expression across all four pairwise comparisons (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Venn diagrams.

Number of DECs (Differentially Expressed Contigs) for the four pairwise comparisons between control and stressed samples collected at different time points: 2 h stressed vs. 24 h stressed (2hS-24hS); 2 h control vs. 2 h stressed normalized to 0 h control (2hC-2hS N0C); 24 h control vs. 24 h stressed normalized to 0 h control (24hC-24hS N0C); and 2 h Stressed vs. 24 h stressed normalized to 0 h control (2hS-24hS N0C).

https://doi.org/10.1371/journal.pone.0205493.g002

Interestingly, at the 2 h time point of PEG-induced stress, a larger proportion of the genes were upregulated than downregulated. This trend reversed itself at the 24 h time point (Fig 2), perhaps as a result of feedback inhibition of the amplified stress response. These observations indicate that drought-induced transcription occurs primarily in the first few hours of stress as a consequence of the observed physiological adjustments, which may include activation of stress resistance or avoidance mechanisms [9,42]. These observations are consistent with previously published reports on the analysis of transcriptomic response to drought in wheat [43], Brassica juncea [44], and in the roots of Prunus rootstock under hypoxia conditions [25].

Functional annotation of the differentially expressed genes (DEGs) and GO term enrichment

Of the 83,110 DECs, 49,512 returned positive Blast hits when aligned to the NCBI database with the BLASTX algorithm. The species distribution for the top Blast hits indicated that ‘Garnem’ transcripts had the highest similarity with P. persica and P. mume, with 21.6% and 11.1% correspondence, respectively (S2 Fig). Besides peach and Japanese apricot, other Prunus species, including P. dulcis, P. salicina, P. armeniaca and P. dulcis × P. persica, were identified as the top hits (S2 Fig). These observations are in agreement with previous studies where a strong homology between various Prunus species was reported [26,27]. In total, 26,700 DEGs were annotated and categorized by biological process (BP) (15,870 DEGs), molecular function (MF) (22,595 DEGs) and cellular component (CC) (13,883 DEGs) sets (Fig 3). Due to a lack of sufficient homology with any annotated gene in the nr nucleotide database, 1.15% of the DECs were not annotated. They were either classified as proteins of unknown function or as hypothetical proteins. These DEGs may represent important proteins that play a role in drought acclimation, however their function remains to be determined [45,46].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Annotated gene ontology (GO) term distribution at 2-level for the three GO categories after GO-slim analysis.

https://doi.org/10.1371/journal.pone.0205493.g003

GO enrichment analysis revealed significantly d enriched GO terms at each time point of the PEG-induced drought (see S5 Table). The GO term “growth” was significantly underrepresented at the 2 h stressed treatment (S5 Table). Genes with this ontology are likely associated with numerous cell wall related processes that are activated as a plant adjusts to drought stress. One such process that was observed in ‘Garnem’ was down regulation of 2-dehydro-3-deoxyphosphooctonate aldolase 1 (KdsA), and consequent decrease of 3-Deoxy-D-manno-oct-2-ulosonic acid (kdo) at the 2 h time point. This observation indicates that there is an inhibition of cell wall formation during the early stages of drought response, which is one of the numerous adaptive mechanisms that plants utilize to adjust to new stress conditions [47]. Additional genes enriched under the “growth” ontology term have been implicated in mediating alterations in root growth under stress. Auxin-related genes, such as auxin-binding proteins (ABP4, ABP-T85) and the enzyme AVP1-pyrophosphatase 1 [48] facilitate auxin transport from shoots to roots. Auxin promotes elongation of root cells, thereby enhancing the root system and facilitating increased water uptake from deeper soil layers [49].

Confirmation of RPKM trends using qRT-PCR

To verify the expression trends determined by RNAseq, qRT -PCR was performed on 18 genes, which represented genes whose RPKM values increased, decreased or remained the same across different time points. The qRT-PCR expression pattern of 16 of the 18 genes (89%) correlated with the RNAseq RPKM values, indicating robustness of the differential gene expression analysis (Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. qRT-PCR validation of select genes in control and treated plants.

The grey-scale bars represent relative gene expression in control (dark grey) and treated plants (light grey) by qRT-PCR analysis (right y-axis). qRT-PCR data show the average relative expression of two biological samples with three technical replicates each one. Lines represent RPKM values of the transcripts in control (black line) and treated plants (dotted line) by RNAseq (right y-axis). The error bars represent the standard error between replicates in qRT-PCR analysis.

https://doi.org/10.1371/journal.pone.0205493.g004

PEG-induced drought activates genes related to stress perception and downstream signalling cascades

Several prior studies have demonstrated that responses to drought are mediated by complex signalling networks that are activated as the plant experiences stress. It is generally considered that a hierarchical sequence of molecular events enables the plants to adapt to stress, which involve stress perception by cell membrane receptors that in turn activate the secondary messengers. As a result, phosphorylation cascades are triggered, resulting in downstream activation of regulatory genes that modulate expression of effector genes of drought stress tolerance [13,49,50].

Based on this paradigm, the annotated DEGs identified in the present study were classified into three major groups: (i) Genes involved in signalling cascades and transcriptional control; (ii) Genes that act as cellular protectors against dehydration-related damage; and (iii) Genes implicated in water and ion uptake and transport [51]. Changes in expression patterns and related discussion for some of the key DEGs annotated under the first category, and important additional mediators of stress signalling is provided in S2 and S3 Appendices sections.

In this study, DEGs annotated as drought-related genes are described in S7 Table. The identity of the drought-related DEGs in ‘Garnem’ rootstocks agrees with previously published studies [26,27, 52].

Biochemical pathways involved during PEG-induced drought stress

Biochemical pathway analysis was performed to gain a better understanding of the response and adaptation of ‘Garnem’ rootstock to PEG-induced drought conditions by mapping the annotated DEGs to their respective KEGG pathways. It was observed that 718 DEGs identified from the 2hS-24hS comparison mapped to 106 pathways, 2,327 DEGs from the 2hC-2hSN comparison mapped to124 pathways, 2,630 DEGs from the 24hC-24hSN and 3,992 DEGs from the 2hS-24hSN comparison mapped to139 pathways each. Of all the DEGs that mapped to KEGG pathways, 655 were annotated as enzymes, which represented 3 major classes namely Hydrolases (55% of DEGs), Transferases (17%) and Oxidoreductases (16%) (Fig 5, S6 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Annotated enzyme distribution of the six main enzyme families in the four DEG pools.

https://doi.org/10.1371/journal.pone.0205493.g005

PEG-induced drought in ‘Garnem’ roots induced several enzymes in purine metabolism (56.42% of DEGs– 51 annotated enzymes), thiamine metabolism (29.56% of DEGs– 6 annotated enzymes), biosynthesis of antibiotics (19.11% of DEGs– 156 annotated enzymes), aminobenzoate degradation (7.62% of DEGs– 6 annotated enzymes), starch and sucrose metabolism (6.44% of DEGs– 32 annotated enzymes) and glycolysis/gluconeogenesis (6.22% of DEGs– 25 annotated enzymes) (S6 Table).

As the most enzyme represented pathway purine metabolism highlights. Metabolites related to purine catabolism contribute to enhance drought tolerance by sensitivity of photosystem reduction, antioxidant protection and ABA metabolism activation [53–55]. It has been demonstrated that Xanthine Dehydrogenase (XDH) acts as a key metabolite in purine catabolism. The suppression and intensification of the XDH activity in rice lines confirmed its role in drought tolerance response by regulating photosystem and reactive oxygen metabolism [55]. In ‘Garnem’ roots, one DEG encoding a Xanthine Hydroxydase (XDH) homolog to AtXDH1 was upregulated at 2 h and downregulated at 24 h time points (S3, S6 and S7 Tables). Then, the XDH enzyme activity might give support antioxidant machinery in ‘Garnem’ rootstock during the first response to drought. On the other side, the accumulation of the intermediary metabolite allantoin functions in drought stress tolerance by influencing in ABA production in aln Arabidopsis mutant lines [lines disrupted in allantoinase (ALN) activity], as well as aah Arabidopsis mutant lines [disrupted in allantoate amidohydrolase (AAH) activity] submitted to drought conditions [53]. In our results, a DEG encoding the ALN enzyme was upregulated, and a DEG encoding the AAH enzyme was downregulated both at 2 h time point (S3, S6 and S7 Tables). In purine metabolism pathway, ALN acts in a reverse reaction synthetizing allantoate from allantoin, which was catabolized by AAH to obtain ureidoglycine. In ‘Garnem’ roots, these two enzyme activities might have a key impact in the first response to drought. The inhibition activity of the AAH might lead to the synthesis of allantoate which might be catabolized by ALN and then, allowing the accumulation of allantoin metabolite. A deeper study about changes of metabolic activity could confirm the crucial role of these enzymes in ‘Garnem’ rootstock.

Drought-responsive enzymes also mapped to pyruvate metabolism, amino sugar and nucleotide sugar metabolism, pentose phosphate pathway, carbon fixation in photosynthetic organisms, fructose and mannose metabolism, pentose and glucoronate interconversions, carbon fixation pathways in prokaryotes, glyoxylate and dicarboxylate metabolism, galactose metabolism, inositol phosphate metabolism and ascorbate and aldarate metabolism. The involvement of these biochemical pathways has also been observed in previous reports [9,18,56] (S6 and S7 Tables). Additional pathways included lipid metabolism such as glicerolipid metabolism, glicerophospholipid metabolism, fatty acid degradation, fatty acid biosynthesis, sphingolipid metabolism, arachidonic acid metabolism, fatty acid elongation, α-linolenic acid metabolism, steroid hormone biosynthesis of unsaturated fatty acids, ether lipid metabolism, steroid degradation, linoleic acid metabolism, steroid biosynthesis cutin, suberin and wax biosynthesis. These findings agree with previously published studies [12,56–58] (S6 and S7 Tables).

Previous reports have demonstrated that carbohydrate metabolism mediated osmotic adjustment and energy production and preservation are crucial for plant adaptation to water stress [57,59]. Under drought conditions, lipids undergo various changes in their metabolism, which help to maintain cellular homeostasis [12]. Activation of pathways related to carbohydrate and lipid metabolism in ‘Garnem’ transcriptome when exposed to PEG-induced drought indicates that they may be involved in achieving a homeostatic state as has been observed in several previous studies in sweet potato [60], peach [27] and Lulium multiflorum [56].

PEG-induced drought activates genes related to protective mechanisms against dehydration-related damage

Adaptation and tolerance to drought is facilitated by the effector genes that play an important role in regulation of processes involved in cell protective functions. These include heat shock proteins (HSPs) or chaperones, dehydration responsive genes including Late Embryogenesis Abundant (LEA) proteins and dehydrins, osmoprotectants, ROS-responsive genes, transporters, and cell wall modifying enzymes [14].

Stomatal movement and water use efficiency (WUE) modulate PEG-induced drought response

ABA-induced stomatal closure reduces water loss, stimulates leaf senescence, downregulates plant growth, and induces biosynthesis of protective substances [81]. Regulation of stomatal closure via ABA accumulation results in reduced transpiration, thereby improving water use efficiency (WUE) of the plant [82]. WUE has been recognized as the most important indicator of plant drought adaptation and tolerance [83,84].

Under drought conditions, ABA regulates changes in turgor of guard cells, thereby modulating stomatal movements and flux of CO₂ and water in plants [8]. The accumulation of ABA is sensed by the PYR1/PYL/CAR receptors, which inhibit the PP2C phosphatase-mediated dephosphorylation of SnRK2 kinases, such as SNF1 [85]. As a result, phosphorylated SnRK2 kinases activate ABRE-binding transcription factors (ABF), which in turn result in the induction of ABA-responsive genes, resulting in stomatal closure [69,86,87].

In PEG-stressed ‘Garnem’ roots, genes representing PYL2-like and PYL8-like ABA receptors were observed to be upregulated (S3 and S7 Tables). However, PYL4-like receptors, which are recognized by Jasmonic Acid [88], were downregulated (S3 and S7 Tables). Most of the DEGs coding for PP2C phosphatases were seen to be downregulated at the 24 h time point, indicating inhibition by PYR1/PYL/RCAR receptors. In this dataset, SNF1-related protein kinases were observed to be upregulated (S3 and S7 Tables). It has been suggested that some members of this gene family may be involved in the regulation of ABA-induced stomatal movement [89,90].

Several contigs were annotated as SnRK2 substrates, which were predicted to be localized in the guard cell membranes. These included the K⁺ channel, KAT1-like and the S-type anion channel SLAH-2-like (homologous to SLAH3). The former was observed to be downregulated at the 24 h time point, while the latter was upregulated at the 2 h time point (S3 and S7 Tables). These observations, which is the first time they have been observed in ‘Garnem’ roots, are similar to a previous report in Arabidopsis, where SLAH3 impairs the inward-rectifying K⁺ channel KAT1 in guard cells, thereby keeping the stomata closed during drought stress conditions [91]. In addition to the above-mentioned genes, ABI5, an ARM repeat protein interacting with ABF2 (ARIA), was induced during drought. This protein positively regulates ABA response in Arabidopsis by interacting with the ABF2 protein [92].

Furthermore, in the ‘Garnem’ transcriptome data, other DEGs implicated in ABA-mediated stomatal closure were identified. These include WRKY TFs and NAC TFs, which were represented by a number of DEGs that were overexpressed during drought in ‘Garnem’ roots (S7 Table). These TFs may act as positive or negative regulators of stomatal movements via ABA signalling [93,94,95]. There were several additional ABA-related DEGs were indicated to be involved in mediating PEG-induced drought response, which are described and discussed in S5 Appendix section. It is important to note that the pathway modelling analysis was conducted on transcriptome data obtained from root tissues. The differential expression of genes only represents transcript behavior and the correlative change on physiological processes is expected in the correct spatial (tissue) context only. The observed differential expression of genes related to stomatal closure or photosynthetic processes is correlated to the physiological measurements at the 24 h time point, at which stomatal conductance decreased significantly in the PEG-mediated osmotically stressed plants. This suggests that ‘Garnem’ most likely adapts to stress conditions by reducing transpiration via stomatal closure, thereby reducing water loss and modulating the photosynthetic processes.

Additional photosynthesis-related genes that were found to be differentially expressed are listed in S3 and S7 Tables. Previous studies have suggested that the observed changes in photosynthetic proteins could be due in part to compensation of photosynthetic electron transport or enzyme activity, which would maintain a partially open state of stomata during drought, leading in turn to maintenance of normal root growth [52].

In addition, contigs annotated as hexokinase I, which are involved in sugar signalling and metabolism, were found differentially expressed in the current dataset (S7 Table). It has been shown that Hexokinase I regulates stomatal closure [96], thereby reducing stomatal conductance and transpiration, resulting in improved WUE. These proteins may have a similar role in the regulation of stomatal movement during drought adaptation in ‘Garnem’.

Three key DEGs that play a role in enhancing WUE were found to be upregulated at the 2 h time point. These included, contig_78795, annotated as ERF023 TF, contig_134330 annotated as LRR receptor-like serine/threonine-kinase ERECTA, and contig_128543 annotated as NF-YB3 TF (S7 Table).

PpERF023 (ppa026139m) is homologous to the AtHARDY gene (At2g36450), and AP2/ERF-like TF. A previous study in rice demonstrated that AtHARDY improves WUE by enhancing assimilation of photosynthates and decreasing transpiration, thereby resulting in improved drought response [97]. In ‘Garnem’, the HARDY gene may play a role in maintenance of root growth processes that are required for drought adaptation.

The second DEG, contig_134330, is homologous to ppa00847m (LRR receptor-like serine/threonine-kinase ERECTA isoform X2). Overexpression of PdERECTA in transgenic Arabidopsis enhanced WUE by eliciting changes in leaf epidermal and mesophyll differentiation, which in turn positively affected growth and accumulation of biomass [98].

The third DEG, contig_128543 annotated as NF-YB3-like TF, which is homologous to PdNF-YB7, is a TF that is induced by osmotic stress and ABA. The overexpression of this gene has been shown to promote primary root elongation and increased photosynthesis, thereby conferring increased WUE and drought tolerance in transgenic Arabidopsis lines [99]. This finding suggests that NF-YB3-like may have a similar function in ‘Garnem’ to those in Arabidopsis and could, therefore, potentially increase WUE in Prunus.

Based on the previous identification of genes directly related to WUE improvement in rice and Arabidopsis [100–102], these three DEGs (ERF023 TF, ERECTA gene and NF-YB3-like TF) identified in ‘Garnem’ transcriptome represent promising targets for further characterization in the context of improving drought tolerance in Prunus.