Impact of Drought on Soluble Sugars and Free Proline Content in Selected Arabidopsis Mutants

Simple Summary Drought has severe effects on plants, negatively impacting economic, agricultural and environmental processes. Depending on the duration and the strength of water stress conditions, plants adjust a series of physiological, cellular, and molecular mechanisms aimed at providing a correct stress response and, if possible, at establishing stress tolerance. The model plant Arabidopsis thaliana is one of the best tools for analyzing the involvement of specific genes in the response to drought. Thanks to this tool, the role of two genes encoding enzymes involved in sugars metabolism, and one gene encoding an enzyme involved in proline synthesis, have been investigated. In addition to suggesting an interaction between the metabolism of proline and that of soluble sugars, results broaden our understanding of the predominant role played by the accumulation of soluble sugars in counteracting mild osmotic stress. Abstract Water shortage is an increasing problem affecting crop yield. Accumulation of compatible osmolytes is a typical plant response to overcome water stress. Sucrose synthase 1 (SUS1), and glucan, water dikinase 2 (GWD2) and δ1-pyrroline-5-carboxylate synthetase 1 (P5CS1) are members of small protein families whose role in the response of Arabidopsis thaliana plants to mild osmotic stress has been studied in this work. Comparative analysis between wild-type and single loss-of-function T-DNA plants at increasing times following exposure to drought showed no differences in the content of water-insoluble carbohydrate (i.e., transitory starch and cell wall carbohydrates) and in the total amount of amino acids. On the contrary, water-soluble sugars and proline contents were significantly reduced compared to wild-type plants regardless of the metabolic pathway affected by the mutation. The present results contribute to assigning a physiological role to GWD2, the least studied member of the GWD family; strengthening the involvement of SUS1 in the response to osmotic stress; showing a greater contribution of soluble sugars than proline in osmotic adjustment of Arabidopsis in response to drought. Finally, an interaction between proline and soluble sugars emerged, albeit its nature remains speculative and further investigations will be required for complete comprehension.


Introduction
Drought is one of the main factors affecting agricultural yields in both irrigated and rainfed systems [1]. In recent decades, much effort has been made to study plant responses to drought in order to address the present and future risks associated with climate change [2,3]. In plants, drought rapidly causes an osmotic imbalance. Due to low water content, soil water potential decreases counteracting the osmotic potential generated by plants, with consequent reduction in water uptake. responsible for the de novo accumulation of proline that occurs primarily in chloroplasts in response to stress or ABA treatments [27,29].
In the present study, the role of SUS1, GWD2 and P5CS1 has been analyzed in hydroponically grown Arabidopsis plants exposed to mild osmotic stress imposed through the addition of 150 mM mannitol. The aim was to study the fluctuations of the two metabolic pools of carbohydrates and amino acids, known to be strongly involved in the response of plants exposed to drought [30][31][32]. Several lines of evidence suggest an interaction between the two pools in response to osmotic stress.

Plant Material
Stock seeds of sus1 line (At5g20830, SALK_014303C) were purchased from the European Arabidopsis Stock Center (NASC, Nottingham, U.K.). The homozygous T-DNA insertion was confirmed by two independent PCR amplifications on 400 ng of genomic DNA extracted from leaves of T3 plants grown on soil. The two pairs of primers were sus1 fw 5 -CTCAAGAGTGCAAGGATCAGG-3 plus sus1 rev 5 -ACGCTGAACGTATGATAACGC-3 and sus1 rev 5 -ACGCTGAACGTATGATAACGC-3 plus LBb1.3 5 -ATTTTGCCGATTTCGGAAC-3 , the first pair specific for the gene of interest and second pair specific for the gene of interest and the left border of T-DNA. The following condition was used for PCR amplifications in a Biometra T-gradient thermocycler: 5 min at 94 • C and 35 cycles of 30 s at 93 • C, 30 s at 58 • C and 1 min at 72 • C. PCR products were analyzed on a 0.8% agarose gel in 40 mM Tris-acetate buffer, pH 8.0, containing 10 mM EDTA and visualized with gel red ( Figure S1). Homozygous T-DNA lines gwd2 (At4g24450, SALK_080260C) and p5cs1 (At2g39800, SALK_063517) have already been characterized in [24] and [33], respectively.

Plant Growth Conditions and Plant Harvests
If not differently specified, plants were hydroponically grown in a growth chamber not equipped with humidity control, at constant temperature of 22 • C, under a 12/12 h light/dark cycle with a photosynthetic photon flux density of approximately 110 µmoL m −2 ·s −1 measured at the level of the leaf surface, as described [34]. Osmotic stress was provided at 12 h dark by transferring 35- Rosettes were harvested at 12 h light under control and after 0.5 days after treatment (DAT), 4.5 DAT and 6.5 DAT. Plant materials were immediately ground with liquid nitrogen and the powders stored at −80 • C until use.

Determination of Water Content
For each time point, the water content was determined on 20-30 plants for each genotype harvested at 12 h light. The fresh weight (FW) of rosettes was recorded immediately after excision and the dry weight (DW) was recorded after 24 h of drying at 80 • C. Water content was calculated as the difference between the FW and the DW of each plant and expressed as a percentage of the FW.

Determination of Lipid Peroxidation
About 100 mg of frozen-ground material was used to measure lipid peroxidation by the 2-thiobarbituric acid (TBA) assay [35]. Three volumes of 0.1% (w/v) trichloroacetic acid (TCA) were mixed with frozen leaf powder. After centrifugation, 200 µL of supernatant was mixed with 800 µL of 20% (w/v) TCA and 5 µL of 0.5% (w/v) TBA. Samples were incubated for 30 min at 90 • C before the reaction was stopped by rapid cooling in ice water. Samples were spun, and the absorbance of supernatants was measured at 532 nm and 600 nm in a UV-Vis Cary60 (Agilent Technologies Inc., Santa Clara, CA, USA) spectrophotometer. The concentration of malondialdehyde (MDA) was calculated by subtracting non-specific absorption at 600 nm and using an extinction coefficient at 532 nm of 155 mM −1 ·cm −1 . For each sample two technical repeats were measured. On average, five independent samples of 5-8 whole rosettes per sample were collected at each time point.

Leaf Starch Quantification
Leaf starch content was quantified as described in [36]. Briefly, in order to remove chlorophylls, 100-200 mg of frozen-ground material was washed in 80% (v/v) ethanol and boiled for 3 min. The ethanol wash was repeated until the samples turned white. Once the samples were bleached, the pellets were dried by evaporation and resuspended in 10 volumes of MilliQ ® water (Milli-Q Integral, Burlington, MA, USA). Suspension aliquots of 0.4-0.5 mL, depending on the weight of the powders, were transferred in clean 1.5 mL tubes and heated at 100 • C for 10 min. After cooling, an equal volume of 200 mM Na-acetate, pH 5.5, was added. Samples were mixed and split in two clean tubes. In one tube, 6 U of amyloglucosidase (Roche Diagnostics, Indianapolis, IN, USA) and 0.5 U of α-amylase (Roche) were added, while in the second tube, corresponding to the control sample, an equal volume of water was added. After overnight incubation at 37 • C, samples were centrifuged and the supernatants were assayed for glucose content by adding 1 U of hexokinase (Roche) and 1 U of glucose 6-phosphate dehydrogenase (Roche) in a reaction mixture containing 100 mM HEPES, pH 7.5; 0.5 mM ATP; 1 mM NAD + and 4 mM MgCl 2 . For each sample a minimum of two technical repeats were measured at 340 nm in a UV-Vis Cary60 (Agilent Technologies) spectrophotometer. On average, five independent samples of 5-8 whole rosettes per sample were collected at each time point and analyzed.

Cell Wall Carbohydrates Quantification
Cell wall carbohydrates content was quantified as described in [37], with some modifications. Briefly, Arabidopsis frozen-ground material (about 100 mg) was washed 3 times with 1.5 mL 80% (v/v) ethanol and centrifuged to remove chlorophylls. Pellets were then washed 2 times with 1.5 mL 80% (v/v) acetone for 10 min and centrifuged again. To remove starch, dried pellets were resuspended in 0.5 mL of MilliQ ® water and gelled at 100 • C for 10 min. Samples were centrifuged and supernatants were discarded. Pellets were resuspended in 0.5 mL of 100 mM Na-acetate, pH 5.5, and the residual starch was degraded overnight by the addition of 6 U of amyloglucosidase (Roche) and 0.5 U of α-amylase (Roche) at 37 • C. Enzymes were inactivated at 100 • C for 10 min and samples spun down. Supernatants were discarded and pellets dried at 70 • C before adding 1 mL of Updegraff solution (glacial acetic acid:nitric acid (65% v/v):water solution in a 8:1:2 ratio; [38]). Resuspended samples were incubated at 100 • C for 30 min. After cooling, samples were spun down and supernatants discarded. Then pellets were washed once with 1.5 mL of MilliQ ® water, 3 times with acetone and dried overnight at room temperature.
Cell wall sugars were released by acid hydrolysis. The pellets were resuspended in 150 µL of 72% (v/v) H 2 SO 4 and the samples were incubated for 45 min at room temperature, vortexing after 30 min of incubation. The acid hydrolysis was stopped with the addition of 600 µL of MilliQ ® water. Samples were centrifuged and the released sugars were quantified in the supernatants by Anthrone assay, mixing 100 µL of sample with 400 µL of Anthrone solution (1 mg·mL −1 Anthrone in 95% H 2 SO 4 ). Samples were incubated at 100 • C for 5 min and then read at 630 nm using a plate reader (EnSpire Multimode Plate Reader, Perkin-Elmer, Beaconsfield, Buckinghamshire, UK). Glucose (0-50 µg) was used as a standard. On average, four independent samples of 5-8 whole rosettes per sample were collected at each time point and analyzed. Two technical repeats for each sample were measured.

Quantification of Total Water-Soluble Sugars
Total water-soluble sugars were extracted from Arabidopsis powders in a 1:2 ratio with cold MilliQ ® water. Extraction was performed three times by vortexing (1 min), centrifuging (15 min at 13,000 rpm) and collecting the supernatants in a single, clean tube. At the end of the extraction, total water-soluble sugars content was quantified by means of the Anthrone assay. Briefly, the supernatants were properly diluted in water and 350 µL of each sample was mixed with 700 µL of Anthrone solution. Samples were incubated 5 min at 100 • C and after cooling the absorbance at 630 nm was measured spectrophotometrically in 1 mL cuvettes (UV-Vis Cary60, Agilent Technologies). The standard curve was made with sucrose in a range of 0-50 µg. Two technical repeats for each sample were measured. On average, five independent samples of 5-8 whole rosettes per sample were collected at each time point and analyzed.

Glucose and Fructose Quantification
Glucose and fructose were extracted from 100 mg of frozen-ground material by three subsequent washes performed with an equal volume of cold MilliQ ® water. At each extraction, the samples were shaken vigorously for 1 min with a vortex, then centrifuged at 4 • C for 15 min at 13,000 rpm and the supernatants transferred to a clean tube. At the end of the extraction, the samples were incubated at 70 • C for 45 min and centrifuged for 10 min at 13,000 rpm in order to remove any debris. Glucose and fructose contents were evaluated using the Sucrose/D-Fructose/D-Glucose Assay Kit (Megazyme International Ireland, Co. Wicklow, Ireland) in a 96-well plate (EnSpire Multimode Plate Reader, Perkin-Elmer) following the manufacturer's instructions. Two technical repeats for each sample were measured. On average, five independent samples of 5-8 whole rosettes per sample were collected every time point and analyzed

Proline and Amino Acids Quantification
Free amino acids and proline contents were quantified by the acid ninhydrin assay [39]. Leaf powders (about 100 mg) were mixed with 2 volumes of 3% (w/v) 5-sulfosalicylic acid and extracted by vortexing for 5 min at 1 min intervals. After extraction, the samples were centrifuged at 13,000 rpm for 10 min and the supernatants were transferred in clean tubes and spun again. Four technical repeats were assembled in a 96-well plate using from 5 to 15 µL of each sample. When necessary, the samples were brought to 15 µL with the addition of 3% 5-sulfosalicylic acid. After the addition of 15 µL of 3.0 M Na-acetate and 200 µL ninhydrin solution (0.15% (w/v) ninhydrin in glacial acetic acid), the background absorbance was read at 352 nm for proline and at 540 nm for total amino acid pool (T0) using a plate reader (EnSpire Multimode Plate Reader, Perkin-Elmer). The plate was then incubated at 50 • C for 12 min and allowed to cool for 3 min. Measures were repeated just after cooling (T15 ) and after further 2.5 and 5 min (T17.5 and T20 ). T0 was subtracted from T15 , T17.5 and T20 , and the average value was compared with standard curves to calculate proline and amino acids content. As standard, samples from 2.5 to 15 µL of a 25 mM mixture of amino acids solution similar to their abundance in plant extracts (5 mM glutamine, 2 mM glutamate, 2 mM aspartate, 1 mM the other proteinogenic amino acids, proline included- [40]) were measured. Analyses were performed on 3-4 independent samples of 5-8 whole rosettes per sample collected at each time point.

Water Content under Osmotic Stress
To assess possible differences in osmotic compensation, the plant water content (WC) of each genotype under study was followed for 6.5 days by collecting samples at 12 h light over different days of treatment ( Figure 1).  Table S1.

Drought Leads to Different Oxidative Stress
Drought promotes the formation of reactive oxygen species (ROS) that interact directly with different macromolecules causing oxidative stress. In particular, the increase in ROS levels leads to lipid membranes oxidation with the subsequent formation of lipid hydroperoxides and their toxic aldehyde degradation products. The levels of lipid peroxidation, expressed as malondialdehyde (MDA) content, were measured in mannitol-treated plants. As shown in Figure 2, no difference was observed among the various genotypes under control conditions and at the onset of the osmotic stress (0.5 DAT). However, as the stress proceeded, the behavior of wild-type plants and T-DNA lines started to differ. At 4.5 DAT, the concentrations of MDA in gwd2 and sus1 plants were about double compared to wild-type plants. At 6.5 DAT, the difference was maintained only in gwd2 plants, while the sus1 line did not show significantly different levels with respect to wild-type plants (Table S2).
Unexpectedly, the degree of lipid peroxidation in the p5cs1 line was similar to wild-type plants, probably due to the mild stress applied in the present study.  Table S1.
Mild osmotic stress was imposed to 35-day-old plants by adding 150 mM mannitol to the hydroponic medium. With the sole exception of p5cs1, which under control condition and at 4.5 DAT showed a slightly but significantly lower WC in comparison to wild-type plants (Table S1), after 6.5 days of treatment the WC in wild-type and all T-DNA plants was comparable, decreasing from 90% to about 65% of the FW (Figure 1).
Considering that all genotypes behaved similarly, in the subsequent experiments the metabolite content was normalized based on the FW.

Drought Leads to Different Oxidative Stress
Drought promotes the formation of reactive oxygen species (ROS) that interact directly with different macromolecules causing oxidative stress. In particular, the increase in ROS levels leads to lipid membranes oxidation with the subsequent formation of lipid hydroperoxides and their toxic aldehyde degradation products. The levels of lipid peroxidation, expressed as malondialdehyde (MDA) content, were measured in mannitol-treated plants. As shown in Figure 2, no difference was observed among the various genotypes under control conditions and at the onset of the osmotic stress (0.5 DAT). However, as the stress proceeded, the behavior of wild-type plants and T-DNA lines started to differ. At 4.5 DAT, the concentrations of MDA in gwd2 and sus1 plants were about double compared to wild-type plants. At 6.5 DAT, the difference was maintained only in gwd2 plants, while the sus1 line did not show significantly different levels with respect to wild-type plants (Table S2).
Unexpectedly, the degree of lipid peroxidation in the p5cs1 line was similar to wild-type plants, probably due to the mild stress applied in the present study. Biology 2020, 9, x 7 of 15  Table S2.

Effect of Drought on Carbohydrate Pools
In the metabolic effort to overcome the stress, primary starch mobilization is emerging as an important factor within plant responses [31,41]. Thus, primary starch accumulation was measured in Arabidopsis rosettes at the end of the day. As shown in Figure 3a, in all T-DNA lines the storage capacity of transitory starch at 12 h light was similar to wild-type plants and not significantly affected by the stress. On the contrary, the concentration of water-soluble sugars was strongly affected. In response to stress, levels in wild-type plants increased up to 15-fold at the end of the treatment. An increase was evident also in mutants, but much less pronounced (Figure 3b).  Table S2.

Effect of Drought on Carbohydrate Pools
In the metabolic effort to overcome the stress, primary starch mobilization is emerging as an important factor within plant responses [31,41]. Thus, primary starch accumulation was measured in Arabidopsis rosettes at the end of the day. As shown in Figure 3a, in all T-DNA lines the storage capacity of transitory starch at 12 h light was similar to wild-type plants and not significantly affected by the stress. On the contrary, the concentration of water-soluble sugars was strongly affected. In response to stress, levels in wild-type plants increased up to 15-fold at the end of the treatment. An increase was evident also in mutants, but much less pronounced (Figure 3b).
Focusing on the two main hexoses of the central metabolism, i.e., glucose and fructose, already under the control condition, all T-DNA lines had a lower concentration of both hexoses (Figure 3c,d). In this case also, a dramatic increase was evident in wild-type plants following the exposure to osmotic stress conditions, whereas the increase in the mutant lines was significantly lower (Table S3), except for glucose in sus1 plants (Figure 3c). On the whole, the difference was higher for fructose than glucose. Interestingly, the most severe phenotype was observed in p5cs1 mutant, the only mutant line not impaired in carbohydrate metabolism.
Finally, considering that cell wall rearrangements have been reported to take part in adaptation to diverse abiotic and biotic stresses [42,43], the glucose deriving from the acid hydrolysis of the cell wall was quantified during the treatment (Figure 4).
Compared to wild-type plants, no differences were evident in the T-DNA lines, suggesting that GWD2, SUS1 and P5CS1 have no influence on cell wall metabolism, despite SUS1 can potentially produce UDP-glucose and therefore be involved in cellulose biosynthesis [44].  Table S3.
Focusing on the two main hexoses of the central metabolism, i.e., glucose and fructose, already under the control condition, all T-DNA lines had a lower concentration of both hexoses (Figure 3c,d). In this case also, a dramatic increase was evident in wild-type plants following the exposure to osmotic stress conditions, whereas the increase in the mutant lines was significantly lower (Table S3), except for glucose in sus1 plants (Figure 3c). On the whole, the difference was higher for fructose than glucose. Interestingly, the most severe phenotype was observed in p5cs1 mutant, the only mutant line not impaired in carbohydrate metabolism.
Finally, considering that cell wall rearrangements have been reported to take part in adaptation to diverse abiotic and biotic stresses [42,43], the glucose deriving from the acid hydrolysis of the cell wall was quantified during the treatment (Figure 4).  Table S3.   Table S4.
Compared to wild-type plants, no differences were evident in the T-DNA lines, suggesting that GWD2, SUS1 and P5CS1 have no influence on cell wall metabolism, despite SUS1 can potentially produce UDP-glucose and therefore be involved in cellulose biosynthesis [44].

Effect of Drought on the Amino Acid Pools
The concentration of free amino acids was measured in control and stressed plants. The free  Table S4.

Effect of Drought on the Amino Acid Pools
The concentration of free amino acids was measured in control and stressed plants. The free amino acids levels in all lines increased along the stress without showing significant differences compared to wild-type plants (Figure 5a). the central line represents the median and error bars show the highest and the lowest data of the set. The t-test (wild-type vs. T-DNA lines) was used for statistics. p-values are reported in Table S4.
Compared to wild-type plants, no differences were evident in the T-DNA lines, suggesting that GWD2, SUS1 and P5CS1 have no influence on cell wall metabolism, despite SUS1 can potentially produce UDP-glucose and therefore be involved in cellulose biosynthesis [44].

Effect of Drought on the Amino Acid Pools
The concentration of free amino acids was measured in control and stressed plants. The free amino acids levels in all lines increased along the stress without showing significant differences compared to wild-type plants (Figure 5a).   Table S5.
Among the proteinogenic amino acids, accumulation of proline is a well-documented response to several adverse conditions [27,45,46]. For this reason, proline accumulation can be assumed as an indication of proper response to stress. All genotypes, except p5cs1, accumulated proline in response to 150 mM mannitol treatment (Figure 5b). However, at 6.5 DAT, the increase was much less pronounced in gwd2 and sus1 lines, which accumulated only 44% and 52% of wild-type proline content, respectively ( Figure 5b).
As expected, the concentration of free proline in the p5cs1 mutant was extremely low already at the onset of the stress (0.5 DAT) (Figure 5b). These data are not surprising considering that the P5CS1 gene is known to respond to water-limiting condition [27]. However, the data not only underline the role of P5CS1 in proline accumulation in response to drought but confirm that the homologous protein P5CS2 cannot replace P5CS1.

Discussion
Drought, high soil salinity and elevated temperatures are among the major causes of plant mortality affecting both crop yield and the maintenance of the natural heritage [47][48][49]. A recent hypothesis suggested that prolonged drought stress causes plant mortality through a combination of different factors, substantially attributable to carbon limitation and hydraulic failure, which together, rather than individually, lead to plant death [50]. Many different post-translational modifications eventually lead to cell death [51][52][53].
In consideration of the increase in the soluble sugars content observed in plants exposed to drought [54][55][56], the role played by the sole carbon starvation as the main metabolic determinant that triggers the transduction chains ending with the plant death appears less relevant. A possible explanation for such an increase is based on the fact that under moderate drought, plant growth declines before photosynthesis [55,56], resulting in an excess of carbon skeletons, which can be directed to osmolytes production [57][58][59].
In the present study, the analysis focused on metabolites acting as osmolytes or precursors for osmolyte production. Relying on the induction of the expression under drought or osmotic stress verified through eFP browser [60], three Arabidopsis single mutants, all members of gene families composed by two to six genes, were selected.
Under normal growth conditions, the pools of starch, total soluble sugars, cell wall carbohydrates and total amino acids were similar in all genotypes, although differences were specifically observed in glucose and fructose levels which were reduced in all mutants in respect to wild-type plants (Figure 3c,d). In response to stress, the non-soluble fraction of carbohydrates, here analyzed as transitory starch (Figure 3a) and cell wall carbohydrates (Figure 4), was unaffected by the mutations. The unchanged ability to accumulate leaf starch during the day is in agreement with the cytosolic localization of GWD2 and SUS1, and with the biological function of P5CS1, and suggests that none of the above enzymes causes a negative feedback loop, which acts on the ability of the chloroplast to accumulate starch. Similarly, the absence of GWD2, SUS1 and P5CS1 had no effect on the cell wall, although cell wall rearrangements have been reported to take part in the adaptation mechanisms that occur in both abiotic and biotic stress responses [42,43]. While this result was expected for gwd2 and p5cs1 mutants, the same was not true for sus1 in view of the potential ability of SUS1 to produce UDP-glucose, which may be directed towards cellulose biosynthesis [44].
On the contrary, the pool of soluble sugars was largely affected by mannitol treatment. Soluble sugars are central components of energetic and biosynthetic metabolism, but they may act as well as both compatible osmolytes, re-establishing the osmotic balance, and protective macromolecules or scavengers of reactive oxygen species [61][62][63][64]. In comparison to wild-type plants, all mutants under study showed a lower capacity in soluble sugars accumulation (Figure 3b), with glucose and fructose as the most relevant components (Figure 3c,d). These findings further support the suggested role of SUS1 in the response to osmotic stress [15,16], and provide for the first time a role to GWD2 in stress response. However, an exhaustive comprehension of the phenotype shown by the gwd2 mutant is complicated by the still unknown function of GWD2. It has been demonstrated that, in vitro, GWD2 phosphorylates glucans, likewise the plastidial isoform GWD1, which is fundamental for transitory starch degradation in chloroplasts [22,23]. However, in vivo evidence supporting a role for GWD2 in the mobilization of sugar from soluble and cytosolic substrates, such as phytoglycogen and soluble heteroglucans [65][66][67], remains scarce [24].
Considering the catalytic function of P5CS1 [45], the reduced capacity to accumulate soluble sugars could depend on the reduced ability of the p5cs1 mutant to generate proline and therefore to regenerate NADP + [26]. The reduction in NADP + regeneration could overload the photosynthetic electron transport chain and compromise the production of soluble sugars.
Accumulation of both soluble sugars and proline takes part in the drought-response system [6,7,25,27] and it has been already suggested that the two signaling pathways interact [30,46]. This interaction has little influence on the total amino acids pool, rather it is specific for proline, whose strong increase under stress conditions is impaired in mutants (Figure 5a,b). However, we cannot exclude the simple explanation that the reduced capacity to accumulate proline observed in the gwd2 and sus1 mutants (Figure 5b) can be a mere consequence of the lower accumulation of glucose and fructose observed in these genotypes (Figure 3c,d). In fact, the lower concentration of hexoses could slow down the carbon flow through glycolysis diminishing the supply of pyruvate to tricarboxylic acids cycle with consequent reduction in glutamate production and thus proline synthesis. As a possible alternative, a reduced hexose availability could reduce NADPH production in the cytosol by the oxidative pentose phosphate pathway (which is early induced under osmotic stress [68]), affecting in turn proline production [69].
Considering that under stress conditions plants must produce osmolytes to protect the photosynthetic apparatus, maintain cell turgor and avoid a hydraulic failure, the results herein presented show that soluble sugars reach a concentration (about 260 nmol/mg FW) much higher than proline (concentration of 32 nmol/mg FW). In this respect, consistently with recent data on two halophytes [40], we can conclude that sugar metabolism gives a more substantial contribution to osmotic adjustment under drought stress. This notwithstanding, proline accumulation can benefit the cell through several other protective mechanisms [70].

Conclusions
Besides shedding light on the physiological role of GWD2 in response to drought, our results support a predominant role of soluble sugars in the stress response to mild osmotic stress in Arabidopsis. At least under the tested conditions, the accumulation of soluble sugars can be used to counteract osmotic stress more than proline. Although a putative interaction between proline and soluble sugars has emerged, the nature of this interaction remains to be investigated.