Comparative Analysis of Cabernet Sauvignon (Vitis vinifera L.) and Kober 5BB (V. berlandieri × V. riparia) Root Transcriptomes Reveals Multiple Processes Associated with Drought Tolerance in Grapevines

Abstract

Grapevine cultivars (Vitis vinifera L.) generally exhibit relatively high tolerance to drought stress. In contrast, the drought tolerance of other Vitis species and their hybrids used as rootstocks differs considerably. In order to attain a better understanding of the molecular basis of drought tolerance in grapevine, we conducted a comparative transcriptomic analysis of roots of drought-stressed Cabernet Sauvignon (CS, Vitis vinifera L.) and rootstock of Kober 5BB (V. berlandieri × V. riparia) using the Affymetrix Vitis Genome Array Version 2.0. We identified 1279 (745 upregulated and 534 downregulated) and 1925 (807 upregulated and 118 downregulated) differentially expressed genes in 5BB and CS. Numerous genes that are putatively involved in aquaporins, proline-rich protein, reactive oxygen species, osmoprotection, and lipid transfer were differentially expressed in response to drought stress in both genotypes. When gene ontology analyses were examined, it was observed that CS and 5BB genotypes were associated with the highest number of similar genes in both the molecular function (protein binding, catalytic activity, and DNA binding) and the biological process (metabolic process and translation) categories. The identification of different regulated genes between 5BB and CS roots is expected to help advance our understanding of molecular mechanisms operating during drought stress in grapevine roots.

1. Introduction

Among abiotic stressors, drought is considered one of the most critical factors restricting plant growth and agricultural output worldwide [1,2]. In recent years, the combination of global warming and drought has had notable effects on plant production. In general, drought leads to a decrease in water content within plants. However, its impact on plant development varies depending on several factors, including the duration of the drought, the structure of the plant root system, and the plant’s tolerance to water stress [3].

As a direct consequence of drought stress, the plant experiences turgor loss and a decrease in osmotic potential, leading to increased osmolyte accumulation in cells [4,5]. During the process known as osmotic adjustment, a number of physiological events, including reductions in cytokinin and gibberellic acid levels and an increase in abscisic acid (ABA) levels, are often observed. As a result of turgor loss, the stomata are closed during drought adaptation, reducing transpiration-mediated water loss and preventing CO₂ emission [6]. Additionally, photosynthesis is inhibited, leading to an increase in reactive oxygen species (ROS), which triggers the activation of antioxidant defences [7].

In plants, roots are particularly important for absorbing soil water and nutrients, synthesizing organic compounds, and interacting with the soil environment, and are the primary response to adapt to drought stress [8,9]. Root system architecture (RSA) is based on root morphology and topology and covers the spatial distribution of the root system during growth [10]. RSA has an important role in a plant’s adaptation to drought stress. Recent advances in multi-omics technologies have notably enabled the identification of molecular mechanisms and associated genes in the regulation of drought stress and RSA [11]. As is well known, roots exhibit alterations in length and depth in response to drought stress [12]. In recent years, the potential roles of various stress-associated root proteins have been investigated in detail [13,14,15,16]. Nevertheless, the effect of drought stress on root molecular mechanisms remains less understood compared to those in aboveground tissues. This is even though drought tolerance is influenced by both below- and aboveground components. Thus, the knowledge of perception of drought signals in roots and the understanding of molecular mechanisms and morphological changes associated with them is still lacking [12].

The grapevine (Vitis vinifera L.) is an economically important plant species adapted to diverse environmental conditions, and in recent years, due to climate change, grapevine-berry quality and yield have been affected by drought stress [17,18]. However, in vineyards, mild drought stress can positively affect berry and wine quality. Due to their extreme susceptibility to grapevine phylloxera (Daktulosphaira vitifoliae, Fitch), most grapevine cultivars are grafted onto phylloxera-resistant American rootstocks [19,20]. This puts grapevine in a unique position where both cultivar and rootstock can influence drought tolerance. Certain rootstocks such as 110R, 140Ru, and 99R are more drought-tolerant, whereas others such as 5BB, 5C, and SO4 are less drought-tolerant than V. vinifera L. Similarly, Vitis berlandieri × Vitis riparia hybrids demonstrate higher drought tolerance than Vitis berlandieri × Vitis rupestris hybrids [21]. 5BB, which is a hybrid of V. riparia × V. berlandieri and a widely used rootstock, is known to be sensitive to drought [21,22,23]. Numerous studies have demonstrated that, compared to grapevine rootstocks such as 110R, 1103P, and 140 Ruggeri, 5BB exhibits higher sensitivity, particularly in terms of root growth, photosynthetic activity, and water potential [13,24]. Moreover, previous studies have evaluated the mechanisms by which rootstocks increase the drought tolerance of grapevines [13,23,25,26].

Cabernet Sauvignon (CS) is the most popular red wine grapevine cultivar and is moderately drought-tolerant [27]. Under drought stress, CS demonstrates yield loss and a decrease in water potential. However, CS can develop adaptive mechanisms when re-exposed to drought stress [27]. In recent years, it has been reported that CS mitigates water loss under drought by exhibiting isohydric behavior, primarily through effective stomatal regulation and closure mechanisms [28].

This study aimed to evaluate the differences in gene expression within root tissues of the drought-sensitive 5BB rootstock and the drought-tolerant CS cultivar under drought stress, considering their functional roles in drought tolerance. For this purpose, we comparatively analyzed root transcriptomes of 5BB and CS using GeneChip™ Vitis vinifera Genome Array. To the best of our knowledge, differences in root gene expression profiles of 5BB and CS have not been comparatively examined before.

Determination of the drought-stressed root genes will help improve drought tolerance at the whole grapevine level. Among grapevine rootstocks, the number of genotypes tolerant to drought stress is very low; therefore, the development of tolerant rootstocks that can affect the drought tolerance of cultivars is a breeding priority. On the other hand, as quantitative trait loci (QTL) associated with drought tolerance in Vitis vinifera cultivars and rootstocks are very limited, our findings may aid in the development of gene-biomarker-based pools used in varietal comparisons and germplasm selections in classical grapevine breeding programs.

2. Materials and Methods

2.1. Plant Material

Plant material was obtained from the Tekirdağ Agricultural Research Institute of the Ministry of Food, Agriculture and Animal Husbandry, Türkiye. The cuttings of CS and 5BB were grown in growth chambers (Biotechnology Institute, Ankara University) at 24 °C with a 16 h photoperiod to produce 1-year rooted plants, which were then used to produce new in vitro grown shoots.

2.2. Tissue Culture

Rooting studies were carried out using Roubelakis rooting medium [29] modified from MS (Murashige and Skoog, Sigma, Burlington, MA, USA, M5519-50L) as in the other studies conducted by the same authors [30,31]. Cultures were incubated in a growth chamber for 5–6 weeks under conditions of a 16 h light period at 28 °C and an 8 h dark period at 23 °C.

2.3. Drought Stress Treatment and Leaf Stem Water Potential (MPa) Measurements

Plants with fully developed roots and shoots were transplanted into sterilized sand (120 °C, 20 min) in a growth chamber. To reduce the risk of contamination due to sugar in the medium, plant roots were carefully washed with sterile distilled water following their removal from the magenta during transplantation. The plants were gradually acclimatized and grown for another 5–6 weeks. During this time, plants were watered with Hoagland solution (10%) [32] every day at the same time.

For drought stress treatment, watering was withdrawn for a period of 10 days, and MPa values were measured on fully expanded leaves using a pressure meter (Model 600, PMS Instrument Company, Albany, NY, USA) [31]. Control plants were grown under identical growth conditions without exposure to drought stress. Three biological and three technical replicates for each stress and control plant were performed.

The MPa values were compared to control values using a t-test at a significance level of p < 0.05. All measurements were performed in four replicates, and the data were analyzed using ANOVA [30,31].

2.4. RNA Isolation and Microarray Hybridization

Total RNA from the root tissues of each genotype was isolated using the grapevine RNA isolation protocol [33]. Each genotype was pooled from three samples, and care was taken to include three biological replicates. RNA concentration was quantified using a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), while RNA quality was evaluated via 1% agarose gel electrophoresis. Additionally, RNA integrity was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). This study included 18 Vitis vinifera Genome Array Chips, one treatment (drought), two grapevine genotypes (5BB and CS), one tissue type (root), two time points (0th day and 7th day), and three biological replicates for the drought stress treatment. Vitis vinifera genome array chip experimental designs are provided in Table S1.

The protocols for first-strand cDNA synthesis, second-strand cDNA synthesis, in vitro transcription of labelled aRNA, and aRNA purification were performed using the Affymetrix platform (http://www.affymetrix.com/support/technical/index.affx (accessed on 1 February 2015)) as described in Çakır Aydemir et al. [30] and Yüksel et al. [31]. Fragmentation of the labelled aRNA was achieved by incubation in a thermal cycler (Bio-Rad, USA) at 94 °C for 35 min, after which the fragmented aRNA samples were collected in suitable amounts for the array format and hybridized. Probe arrays were hybridized in a hybridization oven at 60 rpm at 45 °C for 16 h (GeneChip™ Vitis vinifera Genome Arrays, Affymetrix, Santa Clara, CA, USA). Post-hybridization washing was carried out using the GeneChip^® Fluidics Station 450 (Affymetrix), and array scanning was performed with the GeneChip^® Scanner 3000 (Affymetrix).

2.5. Microarray Data Analysis

Gene identities, accession numbers, sequence information of probe sets exhibiting altered expression, and homologous genes in other plant species were determined using the NetAffyx™ analysis centre (www.Affymetrix.com (accessed on 1 February 2015)) and plant expression database (PLEXdb, http://plexdb.org (accessed on 1 February 2015)). Raw chip intensity values were analyzed using the R programming environment, and preprocessing was performed with the robust multi-array average (RMA) normalization method [31,34]. Following the expression difference (fold change) analysis conducted with the R-Affy Express package (Affy, version 1.48.0), 10.569 out of 16.602 probes were filtered out based on the criterion that at least two out of three biological replicates exhibited an expression value below 6. Further analyses were carried out using the remaining 6033 probes.

For gene expression analysis across different time points, 6033 probes with a fold change larger than 2 and a p-value less than 0.01 were selected using linear models for microarray data (LIMMA) and false discovery rate (FDR) correction methods implemented in the R-Affy-Express package [31,34], considering the interaction between drought stress treatment and two time points. As a result, it was determined that 1925 probes in 5BB and 1279 probes in CS (with 818 probes commonly regulated in both genotypes) exhibited more than a two-fold difference in gene expression analysis.

Gene ontology (GO) categories and the Kyoto Encyclopedia of Genes and Genomes (KEGG) database pathway analyses were performed similarly to other transcriptome studies conducted by the same authors [30,31]. Accordingly, GO categories were conducted using the tool provided by the Gene Ontology Consortium (http://geneontology.org/ (accessed on 1 February 2020)) to classify probes into the domains of molecular function and biological process [35,36]. In these analyses, it was considered that a single probe might be annotated to multiple ontology domains. For pathway analysis, probe IDs corresponding to differentially expressed genes (DEGs) were used to retrieve the relevant cDNA sequences from the Ensembl Plants database [37]. Subsequently, the KEGG database (https://www.genome.jp/kegg/pathway.html (accessed on 1 February 2020)) was utilized to elucidate the functional roles and possible biological pathways associated with drought stress [38].

2.6. qRT-PCR Analysis

Induction ratios of 10 drought-responsive genes selected by microarray analysis were independently validated by qRT-PCR analysis. This revealed similar induction values for these genes in stress-exposed plants. In selecting genes for qRT-PCR validation, those showing significantly high fold changes between two genotypes (up or downregulated, with log₂ ≥ 2 or ≤–2, p < 0.05 ANOVA-BH) were focused on. To more effectively control the validation of microarray results, care was taken in gene selection to include genes involved in different metabolic pathways (sugar and amino acid metabolism) and functional tasks (including osmotic regulation, transport, signal transduction, and structural proteins) (Table S2). For each gene, specific primer pairs were designed using the Primer-BLAST web tool provided by NCBI (National Center for Biotechnology Information, https://www.ncbi.nlm.nih.gov/tools/primer-blast (accessed on 5 February 2015)).

All qRT-PCR reactions were performed using a Light Cycler 480 (Roche, Basel, Switzerland). The qRT-PCR protocol, standard curves optimization, and related PCR programs for the reactions were conducted in accordance with the qRT-PCR methodology described by [31,39]. The Vitis vinifera eukaryotic initiation factor 4A-8 gene (EIF4α; LOC100261822, XM_002277667.3) was used as a housekeeping gene to normalize qRT-PCR expression data [31,39]. All reactions were repeated three times using 3 biological replicates.

As reported in the other gene expression studies [30,31,39] conducted by the same authors, gene expression levels were calculated using the REST 2009 software (http://www.gene-quantification.de/rest-2009.html (accessed on 5 February 2015)), employing the 2^−ΔΔCT method [40]. For each genotype, the possible correlations between microarray-derived fold change values and qRT-PCR-derived fold change values for selected genes were evaluated using Microsoft Excel software (Excel 16.0).

3. Results

3.1. Leaf Stem Water Potential Measurements (-MPa)

The leaf water potential of CS changed from −0.2 MPa on the first day of stress application to −0.8 MPa on the seventh day of stress. There was no significant change in leaf water potential after the seventh day of stress. This value remained constant over the next three days (Figure 1A).

The leaf water potential of 5BB plants changed from −0.3 MPa on the first day of drought treatment to −1.20 MPa on the seventh day of stress. On the tenth day, the stress samples had a value of about −1.40 MPa. (Figure 1B). As was expected, the leaf water potential values of control plants did not undergo significant alterations (−0.2 MPa to −0.3 MPa) during this period (Figure 1A,B). Overall, these measurements suggested that the drought stress treatment we applied was effective in inducing physiological changes in both genotypes and that, as was expected, 5BB plants were more sensitive to drought than CS based on a greater reduction observed in their leaf water potential.

3.2. Microarray Analyses of CS and 5BB in Drought-Stressed Roots

A total of 1925 DEGs were identified in the roots of 5BB, of which 807 were upregulated and 1118 downregulated. Similarly, drought-stressed roots of CS exhibited 1279 DEGs, 745 were upregulated, and 534 were downregulated (Table S3). The number of DEGs differentially expressed in 5BB and CS roots was 1107 and 461, respectively (Figure 2A). Additionally, 818 DEGs were commonly expressed differentially in both genotypes (Figure 2A). Among these, 16 DEGs exhibited antagonistic regulation, being upregulated in 5BB roots while downregulated in CS roots. 369 DEGs were upregulated in both genotypes, while 326 DEGs were downregulated in both genotypes (Figure 2B).

3.3. Gene Ontology Analyses

The most significant molecular function GO categories for the upregulated DEGs of CS were protein binding (36 genes), catalytic activity (18 genes), and DNA binding (17 genes) (Figure 3a). Conversely, CS DEGs associated with structural constituents of the ribosome (26 genes) were distinctly downregulated (Figure 3b). Within the same category in 5BB, genes related to nucleotide binding (44 genes) and protein binding (33 genes) were differentially upregulated (Figure 3c), while genes involved in nucleotide binding (66 genes), and catalytic activity (41 genes) were differentially downregulated (Figure 3d).

Analysis of biological process categories revealed that the upregulated genes in CS and 5BB were significantly enriched in metabolic processes, comprising 59 and 47 genes, respectively (Figure 4a,c). Moreover, the same categories with the greatest number of downregulated genes were translation, comprising 76 genes in CS and 105 genes in 5BB (Figure 4b,d).

As a result of KEGG pathway analysis of genes with altered expression, it was revealed that amino sugar and nucleotide sugar metabolism and glycolysis/gluconeogenesis pathways in both genotypes (Figure 5A,B) were affected.

3.4. DEGs Under Drought Stress

The most strongly up- and downregulated genes in 5BB and CS are presented in

Table 1. The highest upregulated and lowest downregulated genes in the roots of 5BB rootstock (p < 0.05).

Upregulated Genes
Probe Set ID	Annotation	Fold Change
1621592_s_at	Dehydrin 1a	361.74
1611682_at	ACT domain-containing protein	239.17
1609875_at	protease inhibitor	206.65
1616317_at	alpha-amylase/subtilisin inhibitor-like	204.50
1620305_at	polygalacturonase	137.82
1611875_at	methionine γ-lyase	125.89
1617466_at	L-allo-threonine aldolase-like	123.16
1614441_at	flavonol synthase/flavanone 3-hydroxylase	121.62
1614008_at	polygalacturonase	118.83
1618333_at	homocysteine S-methyltransferase 3	115.21
1614862_at	tropinone reductase homolog At1g07440	108.81
1616116_at	probable galactinol-sucrose galactosyltransferase 6	104.43
Downregulated Genes
Probe Set ID	Annotation	Fold Change
1617745_at	putative lipid-transfer protein DIR1	−180.96
1622607_at	aquaporin TIP1-1	−170.83
1609652_s_at	glucomannan 4-beta-mannosyltransferase 2	−116.47
1612244_s_at	aquaporin PIP2	−99.13
1609063_at	BURP domain-containing protein 3-like	−98.56
1619613_at	14 kDa proline-rich protein DC2.15	−95.64
1614387_s_at	aquaporin TIP1-1	−92.23
1615722_s_at	aquaporin-like	−83.81
1607541_at	glycine-rich cell wall structural protein 2-like	−83.40
1606530_s_at	14 kDa proline-rich protein DC2.15	−81.10
1621879_at	putative lipid-transfer protein DIR1	−76.10
1607766_at	14 kDa proline-rich protein DC2.15	−73.84
1619687_at	subtilisin-like protease	−73.38
1619703_at	aquaporin-like	−70.61
1612873_at	subtilisin-like protease	−62.84
1613467_at	pistil-specific extensin-like protein	−62.20
1606669_s_at	aquaporin-like	−60.40

and

Table 2. The highest upregulated and lowest downregulated genes in the roots of CS (p < 0.05).

Upregulated Genes
Probe Set ID	Annotation	Fold Change
1622651_at	polyphenol oxidase chloroplastic-like	42.31
1616698_at	aconitase 2 mitochondrial	31.25
1612465_at	beta-galactosidase-like	31.10
1617466_at	L-allo-threonine aldolase-like	29.66
1606794_at	thaumatin-like protein	29.16
1619573_at	zinc finger protein ZAT10-like	28.97
1616116_at	probable galactinol-sucrose galactosyltransferase 6-like	25.83
1620063_at	beta-1.3-glucanase	25.12
1613461_s_at	class IV chitinase	22.20
1607133_at	R2R3 transcription factor MYB108-like protein 1	19.90
1620390_s_at	thaumatin-like protein///thaumatin-like protein	19.90
1616045_a_at	proline-rich cell wall protein-like	19.44
1621592_s_at	Dehydrin 1a	19.18
1622147_at	aconitase 2, mitochondrial	19.03
1607620_at	NAC domain-containing protein 29-like	18.62
1610880_s_at	probable indole-3-acetic acid-amido synthetase GH3.1-like	18.39
1620065_at	probable sulfate transporter 3.5-like	18.18
1615375_at	bidirectional sugar transporter SWEET10-like	17.63
1621397_at	aspartate aminotransferase, chloroplastic-like	16.84
1619916_s_at	glucan endo-1,3-beta-glucosidase-like	16.44
1614820_at	tonoplast dicarboxylate transporter-like	15.53
Downregulated Genes
Probe Set ID	Annotation	Fold Change
1617745_at	putative lipid-transfer protein DIR1-like	−81.79
1612030_at	organ-specific protein S2-like	−59.83
1607766_at	14 kDa proline-rich protein DC2.15-like	−56.96
1608800_s_at	14 kDa proline-rich protein DC2.15-like	−37.05
1621879_at	putative lipid-transfer protein DIR1-like	−32.11
1612462_at	snakin-1-like	−28.08
1610096_at	histone H4-like	−25.10
1607732_at	chalcone synthase	−23.88
1617019_at	chalcone synthase	−23.72
1620332_at	histone H3.2-like	−23.69
1616409_at	delta(24)-sterol reductase-like	−20.12
1619662_at	36.4 kDa proline-rich protein-like	−19.06
1610607_at	gibberellin-regulated protein 4-like	−16.98
1622656_at	glucan endo-1.3-beta-glucosidase-like	−16.69
1613827_s_at	fasciclin-like arabinogalactan protein 2-like	−16.19

. In 5BB, highly upregulated genes included dehydrin 1a, an ACT domain-containing protein, and a protease inhibitor, whereas the most downregulated genes encoded glucomannan 4-beta-mannosyltransferase 2, aquaporin TIP1-1, and a putative lipid transfer protein DIR1 (Table 1). In CS, the most upregulated genes included polyphenol oxidase chloroplastic-like and a aconitase 2 mitochondrial, while the most downregulated genes encoded a putative lipid-transfer protein DIR1-like and organ-specific protein S2-like (Table 2).

3.5. qRT-PCR Analysis

Correlation graphics related to the results of microarray expression (fold change) and qRT-PCR (fold change) in 10 selected genes (Table S2) are shown in Figure 6. For both genotypes, correlation levels were found to be at highly significant (5BB: R² = 0.92 and CS: R² = 0.89).

4. Discussion

4.1. Leaf Water Potentials of CS and 5BB Under Drought Stress

In most studies, water potential measurement is considered the standard approach for assessing the status of plant water [41]. However, several studies have indicated that osmotic adjustment under drought stress serves as an important drought-tolerance mechanism, whereby osmotic potential decreases due to solute accumulation [42]. Depending on the plant species, when leaf water potential decreases to low rates (e.g., −0.3 MPa and −0.8 MPa), the turgor pressure of leaf cells declines rapidly, leading to the synthesis of the stress hormone of ABA and subsequent stomatal closure to minimize water loss [43].

During prolonged drought stress, the leaf water potential may decrease to −1.0 and −2.0 MPa. Leaf water potential measurements have largely been used to assess the water status of grapevines [44,45,46]. According to the studies conducted on grapevines, an average leaf water potential of approximately –0.8 MPa is considered indicative of a non-stressed physiological state, whereas values around –1.2 and −1.5 MPa reflect mild to severe drought stress, respectively [43,47].

In our study, CS and 5BB depicted average values of −0.7 and −0.95 MPa, respectively, suggesting that the drought-tolerant CS can better maintain existing water content than the drought-sensitive 5BB. Additionally, these findings are consistent with the reported MPa values ranging from −0.75 to −1.00 MPa after eight days of drought treatment [47].

4.2. Comparing the Response Genes of 5BB and CS to Drought Stress

In this study, we have identified differences in the numbers of DEGs in the roots between 5BB and CS. We have also identified a total of 1925 DEGs in 5BB. Of these, 807 DEGs demonstrated upregulation while 1118 DEGs were downregulated by drought stress. In a study regarding drought-related RSA regulation in the roots of tree-grafted grapevine rootstocks (110R, 5BB, and 41B), 1196 DEGs were found on 5BB rootstock, which is parallel to this study [13]. In CS, out of 1279 DEGs, 745 were upregulated, while 534 DEGs were downregulated. Overall, this suggested that the extent of transcriptomic responses to drought in drought-sensitive 5BB rootstock was stronger than that in drought-tolerant CS genotype.

Generally, tolerant genotypes exhibit a lower number of DEGs in response to stress [48,49]. These genotypes often display more efficient and specific targeted responses due to the presence of pre-activated genes even before the onset of stress (pre-adapted) [50]. Therefore, excessive gene activation may not be required to maintain homeostasis in these tolerant genotypes. Although it has been suggested that plants with higher drought tolerance may respond to water limitation with less pronounced changes in gene expression [51], this may not always be the case [52]. It can be seen from our experience with transcriptome studies under stress conditions that the quality of the response, including which genes are expressed and the dynamics of their regulation, interacts with the number of DEGs.

Moreover, another of the most important results of this study in terms of DEGs analyses was that fold-induction values observed in CS were overall lower than those observed in the drought-sensitive 5BB. Additionally, the magnitude of fold changes expressed was larger in 5BB than in CS. Overall, these findings suggest that 5BB exhibits a stronger transcriptional response to stress compared to CS, and the larger foldchange and differential expression values observed in 5BB further indicate that this genotype may be more severely impacted by drought stress.

It was also previously reported that the number of DEGs is lower in tolerant genotypes during different abiotic stress responses, and that fold change values are generally lower. This proves that tolerant genotypes manage the stress response in a controlled and energy-efficient manner [53,54].

In this study, the most responsive gene to drought stress in 5BB was found to be a dehydrin 1a (DHN1a, 1621592_s_at, 361 fold change). Transcripts of DHN1 genes (DHN1a, b) were determined in V. vinifera and V. riparia and these genes were found to be upregulated after exposure to low temperatures, drought and ABA [55]. Yang et al. [56] investigated the gene expression of four types of dehydrin genes (DHN1- DHN4) in Pinot Noir clone (PN40024) and V. yeshanensis (Chinese) under various abiotic and biotic stresses. Similarly to our study, only the DHN1 gene was reported to be upregulated by drought stress in grapevines among four genes. This circumstance is based on the hypothesis that the DHN1 gene is necessary for drought tolerance but not for defense in sufficient quantity [56]. In another similar study, it was reported that the DHN1a and DHN1b genes in the Korean grapevine V. flexuosa demonstrated a continuous increase throughout 14 days of drought and that these genes play various roles in maintaining intracellular homeostasis under abiotic stress [57]. Interestingly, in contrast to 5BB, only a small change in the expression of the DHN1a encoding gene was observed in CS.

In Arabidopsis, polygalacturonases (PGs) play an important role in cell wall events [58], while in rice, overexpression of the PG gene decreases pectin content and cell adhesion and increases sensitivity to abiotic stress [59]. Moreover, PGs are mainly expressed in root tissues, suggesting that they may play crucial roles in vegetative growth [60]. In this study, two polygalacturonase genes (1620305_at, 1614008_at) were significantly upregulated in 5BB rootstock roots. Recent studies have demonstrated that this gene is not associated with drought tolerance; rather, its expression is increased under drought stress in sensitive genotypes. Accordingly, these findings are consistent with observations made regarding the drought-sensitive 5BB rootstock. It has been suggested that the upregulation of such genes contributes to reduced drought tolerance due to the degradation of cell wall integrity [61].

Protease inhibitors are responsible for regulating the function of proteases by inhibiting their catalytic activity, and they are also involved in plant defense against abiotic stresses. Additionally, protease inhibitors may play a role in the upregulation of ABA-related genes necessary for adapting to drought stress [62]. Based on the findings of the study, it is hypothesized that drought-tolerant genotypes exhibit lower levels of proteolysis due to higher expression of protease inhibitors [62]. An unexpected finding in this study was the upregulation of a protease inhibitor (1609875_at) gene by approximately 206fold change in the drought-sensitive 5BB rootstock. The upregulation of protease inhibitor genes in sensitive genotypes is associated with an adaptation process to limit cellular damage [62].

As observed in 5BB (Table 1), some genes (e.g., DHN1, protease inhibitor) that are expressed at low or normal levels during the early stages of stress can become highly upregulated when the stress persists over time. This suggests that these genes may be delayed response genes and/or dysfunctional genes whose contribution to stress tolerance is limited or potentially detrimental. This issue is addressed in the Section 4 of this paper.

The highest upregulated DEGs in CS include those encoding a polyphenol oxidase (1622651_at), a chloroplastic-like protein (1622651_at) aconitase 2, mitochondrial protein (1616698_at), and beta-galactosidase-like (1612465_at) genes. Among these genes, polyphenol oxidase was found to be induced by drought stress, and similar to our study, it was upregulated in the tolerant wheat genotype [63]. Studies on the overexpression or silencing of polyphenol oxidase genes have shown the key role of these genes in plant defense responses. However, increased polyphenol oxidase activity has been reported to contribute to drought tolerance through ABA signaling and regulation of stress genes [64]. The mitochondrial TCA (tricarboxylic acid) cycle enzyme aconitase genes play a central role in oxidative stress-induced organelle signaling [65]. While RNA-seq data indicate that these genes are differentially regulated under drought stress, the upregulation of some aconitase isoforms during stress suggests their involvement in stress tolerance [66]. Although a direct correlation between the aconitase 2 gene and drought tolerance has not been definitively established yet, its upregulation profile in tolerant CS supports its association with stress tolerance.

In the CS genotype, we also observed that beta-galactosidase-like gene (1612465_at) was others of the most upregulated genes. Consistent with our findings, in both the root and leaf of a drought-tolerant sweet potato [67], the expression of beta-galactosidase genes was upregulated, but only a few genes were downregulated. In the same study, the same genes were significantly downregulated in the sensitive genotype, suggesting that beta-galactosidase genes may be used as drought tolerance biomarkers [67]. In this study, no clear expression difference was observed in this gene in the 5BB rootstock and CS genotype.

According to studies, the chalcone synthase (CHS) gene is crucial for adapting to drought stress [68,69]. In this study, two CHS genes (1607732_at, 1617019_at) were significantly downregulated in the CS genotype. In a proteomic study conducted by Gu et al. [68] in tea plants, CHS genes were downregulated during drought stress, which is in accordance with our study. Similarly, ZmCHS25 knockout mutants of this gene in maize plants significantly reduced drought and salt stress tolerance by increasing oxidative damage parameters [69]. The upregulation of the CHS gene in drought-tolerant genotypes is primarily due to its function as the initial and key enzyme in flavonoid biosynthesis. Flavonoids are well known for their crucial role in protecting plants against oxidative stress, and their accumulation is enhanced in tolerant genotypes [70].

4.3. DEGs Involved Aquaporins (AQPs) and Proline-Rich Protein (PRP) Genes Under Drought Stress

To understand the role of aquaporins in drought tolerance, numerous studies have focused on naturally drought-tolerant plant species, which provide insights into the mechanisms of water transport and stress tolerance and adaptation [71,72]. In grapevine, the majority of genes encoding permeability of plasma membrane aquaporins (PIP) and tonoplast intrinsic proteins (TIP) type aquaporins were also differentially expressed in the leaves of the grapevine rootstock 100 Richter under drought stress [73]. On the other hand, in recent research under drought stress, the impact of many aquaporins [74] on water homeostasis of champinii cv. Ramsey leaves and rising drought tolerance has been revealed. In our work, a total of 11 aquaporins (PIP, TIP type) have been found to show obvious expression changes in 5BB and CS roots. Interestingly, all of these aquaporins were downregulated in 5BB, while only 2 (1622607_at, 1614387_s_at: aquaporin TIP1-1) were found to be downregulated in CS. In some studies, it was demonstrated that the expression levels of PIP and TIP aquaporin gene families significantly increased in drought-tolerant genotypes. This upregulation facilitates cellular water transport and maintains cellular water homeostasis, thereby enhancing the plant’s tolerance to stress [73,75]. In contrast, the significant induction of AQPs genes does not necessarily confer drought tolerance uniformly across all plant species [76]. In our study, consistent with the findings observed in the 5BB rootstock, the expression levels of VvPIP2;1, VvPIP2;2, and VvTIP2;1 were significantly downregulated in the drought-sensitive grape genotype, ‘Kabarcık’, following seven days of drought stress. This downregulation was highlighted as a factor that adversely impacts drought tolerance by limiting water transportation [77]. This situation can be considered a strategy to reduce water loss. Under drought stress, plant cells can reduce the expression of certain AQPs (PIP, TIP) at the transcriptional and/or post-translational levels to minimize water loss in roots, thereby ensuring short-term survival [78,79]. On the other hand, increased ABA levels under drought stress can downregulate the expression of some aquaporins [78]. In addition, irregular stress responses in sensitive genotypes can also downregulate the expression of some PIP and TIP aquaporins [80].

Moreover, the downregulation of the AQPs gene expression in drought-sensitive rootstocks, contrasted with their upregulation in tolerant genotypes, has been consistently corroborated by recent studies on grapevine rootstocks [77,81]. In this respect, the role and function of AQPs genes in molecular breeding come to the forefront. In this study, no significant change in the expression of jasmonate and phenylpropanoid metabolism genes was found.

Our study revealed that three PRP genes (1607766_at, 1606530_s_at, 1619613_at, and 1608800_s_at) were significantly downregulated by drought stress in 5BB and CS. Plant PRPs are regulated by various external abiotic stress factors. In a study conducted on tomato plants under drought stress, the expression of the SlPRP gene correlated with proline levels in different tissues, and a simultaneous increase in cellular proline concentration was detected in all tissues under stress except the root. The study explained that this may support the maintenance of available cellular proline to function as an osmoprotectant during stress [82]. Our study is also supported by the identification of downregulated responses to drought stress of PRP genes in chickpea plants [83]. On the other hand, all PRP genes were identified as the ‘14 kDa proline-rich protein DC2.15-like’ gene, which was isolated in soybean plants by Choi et al. [84], and the expression of this gene was reported to be root specific.

4.4. DEGs Involved in ROS Under Drought Stress

Our study identified several DEGs encoding putative antioxidant enzymes in roots.

A total of 15 DEGs encoding glutathione S-transferases (GSTs) were identified, of these, 12 genes were only present in 5BB, most of which were downregulated. GSTs are thought to increase cell division in meristematic root regions, promoting root elongation under drought stress [85,86]. In tomato plants, it has been reported that the expression patterns of different SlGST genes under osmotic stress are genotype specific and that GSTs play a role in drought responses and their expression may be downregulated in drought-sensitive genotypes [86]. In oats, the significant downregulation of the AsGSTU37 gene under drought suggests that some of these genes may be related to drought stress tolerance [87]. Within the framework of these findings, the downregulation of GSTs under drought stress may be related to a strategy of redirecting the plant’s energy and metabolic resources to more important stress response pathways. As with the 5BB rootstock in this study, downregulation of GSTs is particularly observed in the sensitive genotype. Further molecular research is needed to make a clearer distinction in this regard.

Previous studies examining the expression of genes encoding antioxidant enzymes in roots have reported the induction of these genes in the roots [88]. However, in 5BB, we observed downregulation of genes encoding antioxidant enzymes such as peroxidases, catalases, and reductases. Under severe and prolonged stress, with a significant increase in ROS, some transcription factors and gene regulation systems may be damaged, and energy depletion-induced signaling disorders may occur. In this case, the activity of antioxidant genes may be downregulated, and the cellular defense mechanism may be weakened as a result. For example, some genes involved in the thiolredox pathway, such as antioxidant-related glutathione-dependent enzymes and peroxiredoxin, can be downregulated, which can decrease the adaptive capacity of the plant and increase oxidative damage [88]. Additionally, it has already been described in the studies that under severe abiotic stress conditions, especially drought, ROS scavenging gene expression may be downregulated, and this may vary irregularly depending on the sensitive/tolerant genotype [89,90,91].

4.5. DEGs Involved in Transcription Factor Genes (TFs)

Under drought stress, differentially expressed TFs were identified in both CS and 5BB. Unlike CS, 5BB showed a pronounced presence of TFs such as MYB (myeloblastosis transcription factors), bHLH (basic Helix–Loop–Helix transcription factor), WRKY, and HAT (histone acetyltransferase). Similarly, different RNA-Seq analyses have also reported strong activation of MYB, NAC, and BHLH TFs in grapevine under drought stress [92]. Among these transcription factors, the R2R3 transcription factor MYB108-like protein 1 (1607133_at) and the WRKY-type DNA-binding protein 1 (1622778_at) emerged as the most highly upregulated TFs in 5BB rootstock.

Recent studies have shown that MYB108 TFs are actively involved in gene regulation in plants under abiotic stress conditions [93,94,95]. When MYB108 is overexpressed in herbaceous peony, a significant increase in flavonoid accumulation and antioxidant enzyme activities was observed, along with increased drought tolerance in the plants [95]. The highest expression level of this gene has been observed under severe drought stress [96], and its contribution to stress response through indirect mechanisms such as antioxidant systems, flavonoid, and metabolite accumulation has been demonstrated in various plant species [93,96,97,98]. The findings related to this gene are consistent with the results of this study.

WRKY1 has been identified as a key transcription factor in plant adaptation to environmental stresses such as drought, salinity, and high temperature [99,100]. Recent studies indicate that the WRKY1 gene can regulate drought and salinity tolerance in different plant species by influencing the ABA signaling pathway through various mechanisms (including the involvement of AtNCED2, AtNCED5, AtNCED6, and AtNCED9 in enhancing ABA biosynthesis) [101,102]. Within this context, the observed upregulation of WRKY1 in drought-sensitive 5BB rootstock suggests activation of ABA-related stress signaling mechanisms; however, further functional studies are required to clarify the role of this upregulation in stress tolerance.

Recent research has demonstrated that bHLH transcription factors play a crucial role in the response mechanisms of plant drought stress. Differential regulation of these factors at both transcriptional and post-translational levels enables plants to adapt to drought stress [103]. Notably, among these TFs, bHLH transcription factors (1609442_at, 1610882_s_at, 1622116_at, and 1606543_at) were downregulated in 5BB.

In tomato plants, suppression of the SlbHLH96 gene reduced drought tolerance. This was associated with ROS metabolism, and it was reported that bHLH96 modulates the drought response by regulating the ABA signalling pathway [104]. However, in populus, downregulation of the PebHLH35 under drought stress led to reduced stomatal development and photosynthesis, confirming the role of the bHLHs in stomatal control and water use efficiency [105]. Similarly to the drought-sensitive 5BB rootstock in this study, the downregulation of bHLH transcription factors in drought-sensitive plants has been associated with limiting water loss, and this is considered an important strategy for ensuring plant survival [106]. From this perspective, a deeper understanding of the molecular mechanisms of the bHLH genes could contribute to the development of plant varieties with high drought tolerance.

In response to drought stress, an ethylene-responsive transcription factor (ERF), a zinc finger protein, and a zinc finger CCCH domain-containing protein genes were strongly upregulated in CS roots. Among these, the transcript encoding the zinc finger protein ZAT 10-like (1619573_at) showed the highest level of upregulation. Upregulation of the AtSTZ1 (arabidopsis ZAT10 homologous) gene has been shown to increase drought tolerance in cotton plants [107]. Also, in arabidopsis, GmZFP3 expression enhances tolerance to PEG while reducing sensitivity to drought stress [108]. Li et al. [109] confirmed that the MhZAT10 expression is directly regulated and activated by the MhDREB2A in response to drought stress. In this study, no significant DREB2A-associated gene expression was observed in the CS genotype. Therefore, we can not suggest that the expression of the zinc finger protein ZAT 10 could be directly activated by DREB2A in response to drought stress.

4.6. DEGs Involved in Osmoprotection Under Drought Stress

Osmotic stress induced by drought leads to cellular imbalances, and osmoprotectants such as amino acids, other amines, and sugar alcohols play a key role in maintaining cellular function [110]. In this study, the polyamine oxidase (1622587_at) gene was upregulated only in 5BB. This gene oxidizes polyamines to produce amino-aldehyde, thereby functioning as a molecule that triggers ROS signaling and stress responses [111]. From this perspective, in 5BB, lower stress tolerance may result in higher ROS accumulation, which could enhance the expression of the polyamine oxidase gene and increase ROS signaling. In summary, the upregulation of the polyamine oxidase gene in 5BB under drought stress can be considered as one of the plant’s defense strategies to protect from oxidative stress.

As another osmoprotectant, the expression of expansins, which play a role in plant cell wall expansion, was observed to increase in the roots during the initial stages of drought stress in various transcriptome studies [112]. Interestingly, in this study, a transcript encoding a putative expansin was upregulated in the roots of drought-tolerant CS while it was downregulated in the roots of drought-sensitive 5BB, suggesting that this putative expansin plays a role in the regulation of the turgor pressure and ionic strength in the roots of CS.

4.7. DEGs Involved in Plant Lipid Transfer Proteins (LTPs) Under Drought Stress

LTPs are often differentially expressed during the biotic and abiotic stress [113]. In this study, LTPs (e.g., putative lipid-transfer protein DIR1-like) were found to be downregulated under drought stress in the roots of both genotypes. However, the fold change in the downregulation of the lipid-transfer protein DIR1-like gene in 5BB was approximately 2.5 times higher than in CS. It has been suggested that the downregulation of LTP genes in response to salt and drought stress may be associated with insufficient response or dysfunction in sensitive genotypes [113]. The high fold change in downregulation in 5BB supports the findings of this study. Similarly, in populus, the expression of LTPs was downregulated under drought stress, and this reduction has been emphasized as potentially leading to impaired cell membrane stability and increased oxidative stress [114]. Another study conducted on Nicotiana tabacum demonstrated that overexpression of the NtLTP4 gene enhances tolerance to salt and drought stress [115].

In recent years, various approaches such as the investigation of specific gene families, transcriptome analyses, and linkage mapping have been employed to identify biomarkers or key genes associated with drought tolerance [116]. For instance, transcriptome analyses in coconuts have revealed that DEGs involved in hormonal pathways, including ABA, jasmonate, auxin, ethylene, and gibberellin signaling, may serve as potential candidate genes conferring drought tolerance [117]. Similarly, in historical olive genotypes, genes encoding lipid transfer proteins, glycosyl transferases, and UDP-glucosyl transferases have been highlighted as promising new biomarker candidates for drought stress tolerance [118]. This is shown in this study as the significant fold change difference in the expression of the DIR1-like gene (1617745_at) in 5BB and CS suggests this gene has high potential as a candidate biomarker gene for selection of tolerant genotypes. Genes involved in ABA biosynthesis have also been suggested to be a critical drought tolerance biomarker in grapevines [119]. In our study, no direct ABA-related gene with potential to be a biomarker was found. However, the protease inhibitor gene (1609875_at), which plays a role in ABA-related gene regulation and showed significantly high gene expression in 5BB rootstock, can be considered as a candidate biomarker gene in this direction. In another transcriptome study conducted by the authors of this study, the expression patterns of 10 genes exhibiting significant differential expression were analyzed across 18 different grapevine rootstocks. This study highlighted that genes such as glycine-rich protein (1607606_at), proline-rich protein (1621384_at), glutamate dehydrogenase (1612389_at), and glutamate decarboxylase (1607457_at) may serve as potential biomarker candidates, particularly for drought tolerance in different grapevine rootstocks [39].

DEGs were selected based on different metabolic pathways (e.g., osmotic regulation, transport, signal transduction, sugars, amino acid metabolism) for qRT-PCR validation, following approaches similar to previous studies [120,121]. Among these genes, the tonoplast dicarboxylate transporter-like (1614820_at) and leucine-rich repeat receptor-like serine/threonine protein kinase BAM1 (1607524_at) genes, which are carrier signal proteins, play distinct yet complementary roles in the response to drought stress. Tonoplast dicarboxylate transporter-like gene expression is altered under drought, indicating changes in vacuolar organic acid partitioning that may contribute to osmotic adjustment and redox balance under water deficiency [122]. Furthermore, the upregulation of an LRR-RLK (BAM1-like) gene under drought indicates its role in long-distance dehydration signalling (e.g., the CLE25–BAM1/BAM3 pathway) and potentially contributes to stoma regulation and systemic responses to water deficiency [123]. In this study, leucine-rich repeat receptor-like serine/threonine protein kinase BAM1 was downregulated in 5BB. Transcriptome studies explained that some of these genes are only induced by specific abiotic stress applications and can be downregulated [124]. This induction may enable the plant to use its energy and resources more efficiently.

Galactinol-sucrose galactosyltransferase 6-like (1616116_at) and Galactinol synthase 2 (1614207_at) genes are critical genes that regulate plants’ metabolic responses to drought stress and are involved in the biosynthesis of raffinose family oligosaccharides. Increasing the expression of these genes may enhance plants’ resilience to drought stress. Determining the expression levels of these genes using techniques such as RNA-seq, Microarray and qRT-PCRwill contribute to a better understanding of their effects on drought tolerance [125].

Despite these findings, certain limitations should be considered. Plants attempt to protect themselves by activating their internal defense systems; however, severe drought can cause dysfunction in some genes [126]. Similarly, in this study, the high fold change expression observed in certain genes in 5BB suggests that these genes may be dysfunctional. Although in different plants [127], some genes (e.g., DHN1a) show high upregulation in response to stress, these genes should be further evaluated using multi-omics approaches for a more detailed understanding. Another important factor is that the lack of morphological parameters related to the root system constitutes a limitation in terms of directly linking gene expression results to the physical adaptations of plants to drought stress. In studies, a detailed examination of root system morphology parameters will enable us to more comprehensively understand the physiological effects of the obtained gene expression changes.

5. Conclusions

In transcriptome studies on abiotic stress in grapevine, studies on root tissue remain few compared to studies on shoots, leaves and berries. In this study, we found a higher number of DEGs under drought stress in the roots of 5BB than in CS. There were differences in the expression of genes encoding TFs, phytohormones, antioxidants, PRPs, LTPs, and AQPs between the two genotypes. On the other hand, the number of common genes between CS and 5BB root tissues reflects the common pathways in abiotic stress response mechanisms between the two genotypes.

Many response genes were regulated by drought stress in this study, indicating that much remains to be discovered regarding the mechanism of drought tolerance in grapevine rootstocks and cultivars. In our opinion, obtaining transcriptomic data including all gene information is crucial to determining new tolerance gene candidates.