Water Deficit Enhances C Export to the Roots in Arabidopsis thaliana Plants with Contribution of Sucrose Transporters in Both Shoot and Roots^,

Mild water deficit enhances C export to the roots and modifies root architecture, with a subset of sucrose transporters involved in both shoot and roots.

Abstract

Root high plasticity is an adaptation to its changing environment. Water deficit impairs growth, leading to sugar accumulation in leaves, part of which could be available to roots via sucrose (Suc) phloem transport. Phloem loading is widely described in Arabidopsis (Arabidopsis thaliana), while unloading in roots is less understood. To gain information on leaf-to-root transport, a soil-based culture system was developed to monitor root system architecture in two dimensions. Under water deficit (50% of soil water-holding capacity), total root length was strongly reduced but the depth of root foraging and the shape of the root system were less affected, likely to improve water uptake. ¹⁴CO₂ pulse-chase experiments confirmed that water deficit enhanced carbon (C) export to the roots, as suggested by the increased root-to-shoot ratio. The transcript levels of AtSWEET11 (for sugar will eventually be exported transporter), AtSWEET12, and AtSUC2 (for Suc carrier) genes, all three involved in Suc phloem loading, were significantly up-regulated in leaves of water deficit plants, in accordance with the increase in C export from the leaves to the roots. Interestingly, the transcript levels of AtSUC2 and AtSWEET11 to AtSWEET15 were also significantly higher in stressed roots, underlying the importance of Suc apoplastic unloading in Arabidopsis roots and a putative role for these Suc transporters in Suc unloading. These data demonstrate that, during water deficit, plants respond to growth limitation by allocating relatively more C to the roots to maintain an efficient root system and that a subset of Suc transporters is potentially involved in the flux of C to and in the roots.

The root system is crucial for the development of sessile plants not only because it anchors them to their substrate but also provides them with water and nutrients. There has been a renewed interest in studying root traits in relation to their foraging capacities both in the context of increased yield (Hammer et al., 2009; Tian and Doerner, 2013) and increased productivity under conditions such as drought (Comas et al., 2013). The depth of rooting is an important parameter for foraging capacities, but the overall distribution in roots in the soil (root system architecture [RSA]) has received attention in different research programs aimed at improving plant resistance to stress (Paez-Garcia et al., 2015).

The model plant Arabidopsis (Arabidopsis thaliana) has been used to unravel the molecular basis of root traits because it can be easily cultured on agar medium. In Arabidopsis, the root system is formed through a reiterative program in which lateral roots are produced along the length of other roots. Despite this apparent simplicity, and even for genetically identical plants, root system development is highly plastic due to variations in its surrounding environment (e.g. nutrient shortage and water deficit; Satbhai et al., 2015).

When water availability is limited, Arabidopsis seedling primary root growth can be occasionally enhanced but usually reduced, depending on the constraint extent and its duration (van der Weele et al., 2000). Conversely, Arabidopsis lateral root growth is mainly inhibited in response to drought (Xiong et al., 2006). Restricting lateral proliferation and promoting depth growth is interpreted as an adaptive response to drought to improve water uptake. This root development pattern is partly associated with local soil water availability, since roots exhibit hydrotropism and hydropatterning (Takahashi et al., 2003; Bao et al., 2014).

Roots are a typical sink and depend exclusively on Suc imported from the source leaves. During the vegetative stage, herbaceous plants are usually composed of one source (mature leaves) and two sinks (developing leaves and roots). For this reason, Suc transport in young seedlings of Arabidopsis has been studied in detail, and the different players involved in this transport have been characterized. In the source leaves, Suc is first released from the cytosol of phloem parenchyma cells to the apoplast surrounding the sieve element/companion cell complex through both AtSWEET11 (for sugar will eventually be exported transporter) and AtSWEET12 Suc effluxers (Chen et al., 2012). Second, Suc is loaded into the companion cell through the activity of the AtSUC2 (for Suc carrier) H⁺:Suc symporter (Gottwald et al., 2000) and then passes across plasmodesmata to reach the sieve element, to be finally translocated to sink organs according to the bulk flow model (Münch, 1930). Suc unloading in sinks can occur along different pathways (Patrick, 1997), some of them involving Suc transporters. Any alteration in root growth due to environmental cues will immediately affect the demand on Suc from the source, leading to a new source/sink balance (Lemoine et al., 2013). For example, water deficit preferentially reduces shoot growth rather than root growth, causing an increased root-to-shoot (R/S) ratio because the growth of root and shoot exhibits different sensitivities to reduced water availability (Chaves et al., 2002; Sharp et al., 2004; Verslues et al., 2006). This supposes a reallocation of carbon (C) to the roots, even though C net assimilation rate is reduced because of the lower CO₂ diffusion caused by the limited stomatal aperture preventing water losses (Cornic, 2000). However, water deficit differentially alters rosette expansion and net photosynthesis, potentially resulting in an increased availability of C for the roots (Hummel et al., 2010).

The aim of this work was to investigate the C source/sink exchanges in Arabidopsis plants subjected to water shortage, with emphasis on the root system and on the Suc transporters involved in these exchanges. The expression of many Suc transporter genes has been reported in the roots of Arabidopsis, such as SUC2 (Truernit and Sauer, 1995; Stadler and Sauer, 1996), SUC1 (Sivitz et al., 2007, 2008), SUC3 (Meyer et al., 2004), SUC4 (Schneider et al., 2012), and SUC5 (Baud et al., 2005). The expression of several SWEET Suc transporter genes has been suggested in roots since the demonstration of the proton-independent transport system of Suc in roots (Chaudhuri et al., 2008). However, the precise role of such Suc transporters in roots has been difficult to evaluate by experiments on mutants, especially for transporters with a strong activity in phloem loading (SUC2, SWEET11, and SWEET12), because the inhibition of phloem loading in leaves potentially masks effects in other organs such as roots.

The initial steps of root development in Arabidopsis are heavily documented (Casimiro et al., 2003; Luijten and Heidstra, 2009; Péret et al., 2009; Petricka et al., 2012), but most of those studies were run on young seedlings grown on agar plates in vitro. To be able to follow root development and architecture at later stages, we developed a rhizobox system where the roots are separated from the soil by a porous membrane with a mesh width that allows an exchange of solutes but does not allow roots to penetrate the soil. The rhizobox system used in this study allowed an almost instant harvest of the root system (for physiological, biochemical, and molecular analyses) and a precise study of the RSA of Arabidopsis in two dimensions.

Taking advantage of this system, we investigated C partitioning between shoot and root in well-watered (WW; 100% soil water-holding capacity) and water deficit (WD; 50% soil water-holding capacity) conditions, two water supplies supposed to lead to discriminating C allocation to the roots (Hummel et al., 2010; Muller et al., 2011). The rhizobox allowed a comparison of the root shape in both conditions, indicating subtle changes in the repartition of roots, and revealed increased C partitioning between shoot and root that are discussed together with variations in Suc transporter gene expression in shoot but also in root.

RESULTS

Choice of the Experimental Procedure

While most studies on root growth and development in Arabidopsis were conducted on seedlings grown in vitro, the data presented in this work were obtained on adult plants (30 d old) grown in rhizoboxes. The rhizobox was first introduced by Kuchenbuch and Jungk (1982) and Youssef and Chino (1988) and consists of a soil culture system where the root is separated from the soil by a porous membrane with a mesh facilitating the exchange of solutes but preventing roots from penetrating the soil. This culture system allows the precise study of RSA over a long period of time through a transparent Plexiglas plate as well as molecular and biochemical measurements on a clean (devoid of soil) and integral (including the finest roots and root hairs) root system. Moreover, this culture system allows the full development of Arabidopsis plants, from germination to seed harvest, and also a quick harvest of roots, without the need for time-consuming washes to clean soil from them, a step that may induce a bias in subsequent measurements. With the rhizobox used in this work (Fig. 1) and adapted from Xiong et al. (2006) and Wenzel et al. (2001), the roots grew between a Plexiglas plate and a nylon membrane (Sefar Nitex 03-7/2; mesh width = 0.7 µm) separating the roots from the soil. For WD experiments, the soil was maintained at two water regimes, WW (0.8 g water g⁻¹ fresh soil) and WD (0.4 g water g⁻¹ fresh soil). It has to be pointed out that 0.8 g water g⁻¹ fresh soil corresponds to the water-holding capacity of the soil. Plants were grown until 30 d after sowing (DAS) to harvest sufficient amounts of material, especially in the case of WD roots. Beyond that date, roots of WW plants reached the bottom of the rhizobox, and this impeded RSA analysis.

Figure 1.

Description of the rhizobox growing system. Two to four plants are cultivated between a Plexiglas plate (A) and a nylon membrane (B) separating roots from the compost (C), itself enclosed in another Plexiglas plate (D). This system allows clean roots to be harvested and analysis of the two-dimensional RSA.

The RSA analysis (main and lateral root length, root system shape, and area colonized by roots) could not be performed directly on photographs of roots due to the lack of contrast preventing a clear separation of roots from the membrane background. Using high-resolution autoradiographs of the labeled root system, this problem was solved (see “Materials and Methods”).

¹⁴CO₂-labeling experiments were also run to study the dynamics of C source/sink allocation. A labeling chamber specific for rhizoboxes was designed (Fig. 2A) and integrated in a standard labeling system (Fig. 2B). Whole rosettes were labeled during a 15-min pulse followed by a 45-min chase period in ambient air, and then ¹⁴C exported from the leaves to the roots was determined by counting radioactivity in the different parts of the plant (rosette and roots; Fig. 2C). Preliminary kinetic experiments demonstrated that a 45-min chase was sufficient to allow radiolabel to reach root tips in WW plants.

Figure 2.

Description of the labeling chamber and protocol used for ¹⁴CO₂ pulse-chase experiments. A, The labeling chamber, built in Plexiglas, can contain two rhizoboxes and is separated into two compartments: one accessible to the radiolabeled air (1) and the other inaccessible (2). B, The system comprises a peristaltic pump, a vial containing the radiolabeled sodium bicarbonate (¹⁴CO₂ generated with sulfuric acid injection; red arrow), the assimilation chamber, and a flask containing potash (10% KOH), which is dispersed in the system by the peristaltic pump (solid arrows for pulse, dashed arrows for trapping). C, Radiolabel measured in the roots corresponds to ¹⁴C exported from the rosette to the roots.

Water Deficit Reduces Biomass, But Leaf Growth Is More Affected Than Root Growth

From 0 to 30 DAS, leaf development was followed by regularly taking photographs (Fig. 3) that clearly indicated that leaf growth was reduced in WD plants, although no sign of wilting was visible.

Figure 3.

Top view of WW and WD whole rosettes. WW and WD whole rosettes are presented at selected DAS (D+14, D+21, D+24, and D+30). Water deficit started from sowing (D+0).

The impact of water deficit was evaluated through a set of parameters related to growth and water status, measured on roots and leaves at 30 DAS. Water deficit reduced final leaf area (Fig. 3) by 4-fold (Table I). This reduction in leaf expansion came with a 30% reduction in leaf number. Rosette biomass, expressed as fresh weight and dry weight, decreased by about 80% in WD plants compared with WW plants (Table I). To check leaf water status, water content and osmotic potential were measured in WW and WD plants. The 50% reduction of water supplied to soil was sufficient to diminish leaf water content in WD plants (0.898) compared with WW plants (0.915). Interestingly, the dispersion of rosette water content values was very weak in WW and WD plants (Supplemental Fig. S1A). The rosette osmotic potential was also lower in WD plants (−1.101 MPa) than in WW plants (−0.898 MPa; Table I). The impact of water deficit was also evaluated at the molecular level in rosette by checking the expression of genes related to photosynthesis (RIBULOSE BISPHOSPHATE CARBOXYLASE SMALL CHAIN1A [AtRBCS] and CHLOROPHYLL A/B-BINDING PROTEIN1 [AtCAB1]), to the establishment of water deficit (RESPONSIVE TO DESICCATION29A [AtRD29a]), and to senescence (SENESCENCE-ASSOCIATED GENE12 [AtSAG12]). Water shortage negatively impacted RBCS expression (0.8-fold, P < 0.01), whereas CAB1 expression was approximately the same in leaves of WW and WD plants (Fig. 4A). The transcript level of RD29a, a gene up-regulated by drought (Yamaguchi-Shinozaki and Shinozaki, 1994), increased in response to water deficit in leaves compared with the control (1.7-fold, P < 0.05; Fig. 4B). SAG12 was poorly expressed in WW leaves, and its expression remained constant in response to water deprivation (Fig. 4C).

Table I. Impact of water deficit on a set of growth and water status parameters measured in roots and leaves.

All measurements were performed at 30 DAS. Projected leaf area (28–46 plants), leaf number (28–46 plants), rosette fresh weight (11–14 plants), rosette dry weight (11–14 plants), and rosette water content (11–14 plants) were measured on whole rosettes. Rosette osmotic potential (eight plants) was measured on a pool of three to five leaves. Measurements could not be made on individual roots because of entanglement but rather were made on roots from one rhizobox (two to three root systems). Root fresh weight (five to six rhizoboxes), root dry weight (five to six rhizoboxes), root water content (five to six rhizoboxes), and R/S ratio (five to six rhizoboxes) were measured. Data are means and (se) of three independent experiments. Asterisks represent significant differences from WW plants as determined by Student’s t test (*, P < 0.005).

Parameter	Unit	Treatment
		WW	WD
Projected leaf area	cm2	12.8 (0.6)	3.3 (0.2)*
Leaf number	–	16.6 (0.3)	11.7 (0.3)*
Rosette fresh weight	mg	268.6 (23.3)	48.4 (5.9)*
Rosette dry weight	mg	22.7 (1.8)	4.9 (0.6)*
Rosette water content	g water g−1 fresh weight	0.915 (0.002)	0.898 (0.002)*
Rosette osmotic potential	MPa	−0.863 (0.025)	−1.101 (0.034)*
Root fresh weight	mg	61.1 (8.5)	13.3 (2.5)*
Root dry weight	mg	4.6 (0.5)	1.6 (0.2)*
Root water content	g water g−1 fresh weight	0.921 (0.004)	0.861 (0.012)*
R/S ratio	–	0.2 (0.01)	0.35 (0.02)*

Figure 4.

Impact of water deficit on a set of marker genes in leaves and roots. All measurements were performed in WW (black bars) and WD (white bars) plants at 30 DAS. Relative expression of AtRBCS and AtCAB1 (A), AtRD29a (B), and AtSAG12 (C) was measured by reverse transcription-quantitative PCR (RT-qPCR) in leaves; AtTIP1.2 (D) expression was measured by RT-qPCR in roots. The expression of AtRBCS and AtCAB1 (A), AtRD29a (B), AtTIP1.2 (D), and AtSAG12 (C) genes was measured to check photosynthesis (A), establishment of water deficit (B and D), and senescence (C). Data are normalized (no unit) to the expression level of the reference gene At5g12240 (Czechowski et al., 2005) and are means ± se obtained from the pool of plants (n ≥ 12) of 11 independent experiments. Asterisks represent significant differences from WW plants as determined by a Two/K sample permutation test (*, P < 0.05; **, P < 0.01; and ***, P < 0.005).

In parallel to rosette biomass reduction, the fresh weight and dry weight of WD roots were diminished by about 80% and 65%, respectively. The smaller difference in root dry weight compared with rosette was the consequence of the strong reduction in water content of stressed root (0.921 [WW] and 0.861 [WD]; Table I). It has to be pointed out that the dispersion of water content values was more important in roots than in shoots, whatever the conditions, although the dispersion was higher under water deficit (Supplemental Fig. S1A). Several correlations between the parameters in roots and shoots were tested, but only decreased root water content was tightly associated with decreased rosette dry weight in WD plants (Supplemental Fig. S1B). Although the decrease in water content was larger in roots than in rosette, the reduction of dry weight was less important in root (65%) than in rosette (80%). This was confirmed by the much higher R/S ratio in WD plants (0.35) compared with WW plants (0.20; Table I). The higher R/S ratio was specific to water shortage, since this ratio was also greater when calculated from WD and WW plants of different ages but exhibiting the same leaf area (data not shown). According to this, the water deficiency applied to plants grown in rhizoboxes impacted leaf growth more than root growth. The impact of water deficit was also evaluated at the molecular level by checking the expression of genes related to the establishment of water deficit in root (TONOPLAST INTRINSIC PROTEIN1;2 [AtTIP1;2]). The transcript level of TIP1.2, a gene negatively regulated by strong water stress (Alexandersson et al., 2005), decreased slightly in WD roots (0.7-fold, P < 0.005; Fig. 4D).

Plants Tend to Optimize the Surface of Soil Colonized by Roots under Water Deficit

Because rhizoboxes allowed the study of root development in two dimensions, the effect of water deficit on RSA, primary and total lateral root length, the soil area colonized by roots, and root system shape were analyzed at 30 DAS on WW and WD plants. Total lateral root length reached almost 2.5 m, whereas primary root length was only 16.5 cm, in WW plants (Fig. 5, A and B). The modeling of root shape allowed us to measure the area colonized by the root system. The root system of an individual plant, which consists essentially of lateral roots (93.5% of the total root length), colonized 13.4% of the available soil area (i.e. 48.3 cm² out of 360 cm² of total surface available [18 × 20 cm]; Fig. 5C).

Figure 5.

Impact of water deficit on RSA. All measurements were performed in WW (black bars) and WD (white bars) plants at 30 DAS. Length and area were measured on photographs of the radiolabeled root system scanned previously. A, Impact of water deficit on primary root length. B, Impact of water deficit on total length of lateral roots. C, Impact of water deficit on area colonized by roots. Data are means ± se obtained from four to six plants of six independent experiments. Asterisks represent significant differences from WW plants as determined by a Two/K sample permutation test (*, P < 0.05 and ***, P < 0.005).

Primary root length was reduced slightly in response to water deprivation (−9%), while total lateral root length was diminished dramatically in WD plants (−62%; Fig. 5, A and B). In contrast, soil area colonized by roots decreased only by about 35%, suggesting a better efficiency to colonize the wet soil of roots from the WD plants (Fig. 5C). Indeed, while 1 cm of WD roots explored on average 0.34 cm² of soil area, 1 cm of WW roots explored only 0.2 cm² of soil area. For each condition (WW and WD), 32 independent root systems were analyzed to model an average shape of the root system. This average shape showed that horizontal expansion of WD roots was reduced significantly down to 6 and 7 cm deep, on the right and left sides, respectively, compared with WW ones (Fig. 6A). This zone corresponds roughly to the shallow soil that alternately dries and wets because water was provided daily from the top of the rhizobox. Below 7 cm deep, no difference in horizontal expansion was observed between WW and WD roots. A statistical analysis of horizontal expansion as a function of depth was performed on shallow soil (to 8 cm deep), where significant differences were observed. Because the root system is approximately symmetric, analysis was averaged to one side by gathering data from the left and right sides of roots. From 0 to 5 cm deep, horizontal expansion showed an increase, to reach 3.1 cm in WW plants. Then, horizontal expansion decreased progressively from 5 to 8 cm deep, to reach 2.5 cm (Fig. 6B). Horizontal expansion of WD plants was characterized by a smaller increase, to reach a maximum of 2.1 cm. Therefore, at 5 cm deep, horizontal expansion was reduced by 32% in WD root systems (Fig. 6B). Interestingly, below 10 cm, the lateral expansion of WD roots occasionally exceeded that of WW roots (Fig. 6A).

Figure 6.

Impact of water deficit on the average shape of the root system. All measurements were performed in WW (black dots) and WD (white dots) plants at 30 DAS. A, Comparison of the average shape of the root system in WW and WD plants. The average shape of the root system is defined with a depth resolution (y coordinate) of 0.5 cm. The origin of the chart (0;0) corresponds to the point separating rosette from root. Data are means obtained from four to six plants of six independent experiments, and arrows accompanied by asterisks delimit the zone where lateral expansion of the WD root system is significantly different from lateral expansion of the WW root system, as determined by a Wilcoxon test (*, P < 0.05). B, Evolution of horizontal expansion of the average shape of the root system in WW and WD plants in shallow soil. Analysis was averaged to one side by gathering data from the left and right sides with a 1-cm depth resolution. The data set used comprises 32 root systems per condition. Different letters represent groups that were significantly different as determined by a Wilcoxon test followed by a multiple comparison test (false discovery rate adjustment; P < 0.05).

Taken together, these results show that differences in lateral root colonization were essentially caused by a lower expansion in WD roots compared with WW ones in the upper part of the soil. However, this resulted in a smaller reduction in colonized area compared with the reduction in root length, indicating a reshaping of root distribution.

Free Sugar Content Is Higher in Roots Than in Rosette and Increases in Source and Sink Organs after Water Deficit

The amounts of the three major soluble sugars (Suc, Glc, and Fru) and starch content were determined in leaves and roots of WW and WD plants. In the rosette, the sole source organ present at 30 DAS, Suc (12.3 µmol g⁻¹ dry weight) was the most abundant sugar, followed by Glc (9.1 µmol g⁻¹ dry weight) and Fru (1.9 µmol g⁻¹ dry weight), in WW plants, and an important starch content (324.8 µmol g⁻¹ dry weight) was detected (Fig. 7). All measurements were made on plants harvested at 1:30 pm, after 4.5 h of light, explaining the high starch content. In the root, an organ that fully depends on reduced C imported from the rosette, Suc (17.8 µmol g⁻¹ dry weight) and Glc (18.5 µmol g⁻¹ dry weight) are the two prominent sugars, followed by Fru (5 µmol g⁻¹ dry weight), in WW plants (Fig. 7). Interestingly, the amounts of total soluble sugars were 1.8-fold higher in WW roots (41.4 µmol g⁻¹ dry weight) than in WW leaves (23.3 µmol g⁻¹ dry weight). In contrast, starch level was very low in the roots (5.7 µmol g⁻¹ dry weight) compared with rosette (Fig. 7), a result expected, as the root is not a storage organ in Arabidopsis plants.

Figure 7.

Impact of water deficit on sugar and starch contents in leaves and roots. All measurements were performed in WW (black bars) and WD (white bars) plants at 30 DAS. Sugar contents are expressed in µmol g⁻¹ dry weight (DW). Starch content is expressed in µmol Glc equivalent g⁻¹ dry weight. Data are means ± se obtained from the pool of plants (n ≥ 12) of three independent experiments. Different letters represent groups that were significantly different as determined by a Wilcoxon test followed by a multiple comparison test (false discovery rate adjustment; P < 0.05).

In WD plants, total soluble sugars increased dramatically in the rosette (62.4 µmol g⁻¹ dry weight; i.e. an increase of 268% compared with WW). This increase was due mainly to Glc (+26.2 µmol g⁻¹ dry weight; i.e. 67% of the increase) and to a lesser extent to Suc (+9 µmol g⁻¹ dry weight; i.e. 23% of the increase; Fig. 7). It was interesting that sugar accumulation in leaves under water deficit had a marginal contribution to the osmotic potential (less than 2%). In parallel to the higher amount of free sugars in leaves, starch level decreased by about 25%. In the roots, free sugar contents increased also in response to water deprivation (87.7 µmol g⁻¹ dry weight; +212%), but to a lesser extent than in the rosette (+268%; Fig. 7). In contrast to the leaves, the increase was mostly caused here by Suc (+23.3 µmol g⁻¹ dry weight; i.e. 50% of the increase), although Glc still represented a significant part of the increase (+17.2 µmol g⁻¹ dry weight; i.e. 37% of the increase). The starch level stayed very low and constant in the roots (Fig. 7).

Water Deficit Promotes the Export of Assimilated ¹⁴C from the Rosette to the Roots

To get a more dynamic view on the impact of water deficit on C partitioning between rosette and root, ¹⁴CO₂ pulse-chase experiments were performed on WW and WD plants. After the chase period, rosette and roots were separated, quickly frozen at −80°C, and lyophilized before autoradiography (Fig. 8A) and radiolabel counting (Fig. 8, B–F). At the end of the pulse-chase experiment, the whole rosette was labeled whatever the condition. The youngest leaves, considered as sinks, were more intensively labeled because they imported ¹⁴C during the chase period (red arrows in Fig 8A). Source leaves did not import ¹⁴C during the chase period: if covered with aluminum foil during the pulse and chase to prevent photosynthesis, no labeling was detected. The root system was also entirely labeled, displaying the RSA (Fig. 8A). The fact that all roots were labeled was a good indication that they were all alive. In addition, a strong accumulation of radiolabel was noted in most root tips of WW plants, but this phenomenon was poorly observed in WD plants. Moreover, although most roots displayed root hairs, those were not detectable on autoradiographs, but this may be due to the resolution of the scanner used. It has to be pointed out that no radiolabel was found in root systems when a pulse-chase experiment was performed on plants whose rosette was cut off, indicating that all radiolabel recovered in roots came from ¹⁴CO₂ assimilation in leaves.

Figure 8.

Impact of water deficit on ¹⁴CO₂ assimilation and ¹⁴C export to the roots. All measurements were performed in WW (black bars) and WD (white bars) plants at 30 DAS. After a ¹⁴CO₂ pulse-chase experiment and lyophilization, plants were exposed on PhosphorImager screens (5 d) and scanned. A, Roots and rosettes of WW and WD plants were radiolabeled after a 45-min chase. Red arrows represent the most labeled leaves. B, Assimilated ¹⁴C, defined as the total amount of ¹⁴C measured in the plant per leaf area unit. C, Percentage of the total amount of ¹⁴C found in rosette. D, Percentage of the total amount of ¹⁴C found in root. E, Amount of ¹⁴C exported per mg of rosette dry weight (DW). F, Amount of ¹⁴C found in root per mg of root dry weight. Data are means ± se obtained from six to seven plants of six independent experiments. Asterisks represent significant differences from WW plants as determined by a Two/K sample permutation test (***, P < 0.005).

Assimilation, which corresponds of the total amount of ¹⁴C recovered in a plant (rosette + roots) per leaf area unit, was reduced significantly by about 40% in response to water shortage (Fig. 8B). While 98% of total ¹⁴C incorporated during the pulse remained in the rosette of WW plants, 94% of label was retained in the rosette of WD plants (Fig. 8C). As a consequence, in percentage, 3 times more ¹⁴C was exported to the root in WD plants compared with WW plants (Fig. 8D). Exported ¹⁴C per unit of rosette dry weight was multiplied by about 1.8 with water deprivation (Fig. 8E). Therefore, the amount of ¹⁴C imported per root dry weight unit increased by about 40% (Fig. 8F).

Relationships between rosette dry weight and C partitioning were established to provide some insights into the fine-tuning of C allocation. The study was performed using Pearson correlations (Fig. 9). ¹⁴C assimilation (Fig. 9A), percentage of ¹⁴C exported to roots (Fig. 9B), and amount of ¹⁴C exported to roots (Fig. 9C) were not correlated with rosette dry weight in WW plants. Thus, whatever the rosette size, ¹⁴C assimilation and ¹⁴C allocation to the root were not altered in WW plants. In contrast, while ¹⁴C assimilation was positively correlated to rosette dry weight under water deficit (Fig. 9A), percentage of ¹⁴C exported to the roots (Fig. 9B) and amount of ¹⁴C exported to the roots (Fig. 9C) were negatively correlated. Thus, the more reduced the size of the rosette, the more ¹⁴C was allocated to roots.

Figure 9.

Relationship between rosette dry weight (DW) and ¹⁴C partitioning parameters in WW and WD plants. A, ¹⁴C assimilation as a function of rosette dry weight. B, Percentage of ¹⁴C found in roots as a function of rosette dry weight. C, Amount of ¹⁴C exported per rosette dry weight as a function of rosette dry weight. All measurements were performed in WW (black dots) and WD (white dots) plants at 30 DAS. Data used to perform correlations result from the ¹⁴CO₂ pulse-chase experiments displayed in Figure 7 (n = 6 or 7; six independent experiments). Asterisks represent significant correlations (*, P < 0.005) after a Pearson correlation test.

Water Deficit Changes Transcript Levels of Suc Transporter Genes

The C source/sink allocation was altered by water deficit, and export of Suc, the long-distance form of C transport, was enhanced between leaf and root. To gain further insights on the effect of water deficit on the C source/sink allocation, the transcript level of Suc transporter genes was studied in roots and leaves of the same samples used for sugar analysis. Preliminary experiments were conducted to check for the absence/presence of expression of the nine AtSUC and seven AtSWEET (from clade III involved in Suc transport) genes in the leaves and roots of WW and WD plants by reverse transcription (RT)-PCR (primers are presented in Supplemental Table S1). Among the nine AtSUC genes, transcripts of SUC6, SUC7, SUC8, and SUC9 were not detected in the roots and leaves in WD and WW conditions. No expression of SWEET9 and SWEET10 was detected in the leaves and roots, while SWEET14 expression was measured in the roots but not in the leaves whatever the condition. In contrast, SWEET11 expression was slightly or not detected in the root of WW plants. Following these results, the transcript levels of SUC1, SUC2, SUC3, SUC4, SUC5, SWEET11, SWEET12, SWEET13, and SWEET15 were monitored by quantitative RT-PCR in both organs, while SWEET14 expression was also assessed in roots.

In leaves, AtSUC2, a companion cell-specific Suc:H⁺ symporter, was the most expressed SUC gene (Fig. 10A), as expected from its essential role for Suc translocation (Gottwald et al., 2000). Water deficit significantly promoted the expression of SUC2 (1.4-fold, P < 0.005) compared with the control. In contrast, transcript levels of the four other SUC genes were clearly lower than SUC2 in WD and WW leaves. Besides, SUC3 and SUC5 transcript levels were low in WW leaves and were repressed slightly in response to water deprivation (0.8-fold, P < 0.01 and 0.7-fold, P < 0.05, respectively). No significant change in SUC1 and SUC4 expression was observed in the leaves in response to water deficit (Fig. 10A).

Figure 10.

Impact of water deficit on specific AtSUC and AtSWEET genes expressed in leaves. All measurements were performed in WW (black bars) and WD (white bars) plants at 30 DAS. Relative expression of AtSUC (A) and AtSWEET (B) was measured by RT-qPCR. Data are normalized (no unit) to the expression level of the reference gene At5g12240 (Czechowski et al., 2005) and are means ± se obtained from the pool of plants (n ≥ 12) of 11 independent experiments. Asterisks represent significant differences from WW plants as determined by a Two/K sample permutation test (*, P < 0.05; **, P < 0.01; and ***, P < 0.005).

In parallel, the transcript levels of the Suc effluxer genes SWEET11 and SWEET12, which catalyze Suc efflux to the apoplast during phloem loading (Chen et al., 2012), were up-regulated in WD leaves (1.4-fold, P < 0.01 and 2-fold, P < 0.01, respectively). SWEET13 transcripts were also more abundant after water deficit in leaves (1.7-fold, P < 0.005). SWEET15, also named SAG29, which was the lowest expressed SWEET gene in control conditions, was overexpressed and became the most expressed SWEET gene in leaves in response to water deficit (Fig. 10B).

In WW roots, SUC1 transcripts were the most abundant of SUC genes, closely followed by SUC2. Since SUC1 and SUC2 expression decreased (0.2-fold, P < 0.005) and increased (2.3-fold, P < 0.005), respectively, in WD roots compared with WW roots, SUC2 became the most expressed SUC gene in roots of WD plants. SUC4 was slightly overexpressed in response to water deprivation (1.4-fold, P < 0.005). SUC5 expression, which was marginal in control conditions, was inhibited in the stressed root compared with the control root (0.5-fold, P < 0.05). No significant change in SUC3 expression was observed in the roots in response to water deficit (Fig. 11A).

Figure 11.

Impact of water deficit on specific AtSUC and AtSWEET genes expressed in roots. All measurements were performed in WW (black bars) and WD (white bars) plants at 30 DAS. Relative expression of AtSUC (A) and AtSWEET (B) was measured by RT-qPCR. Data are normalized (no unit) to the expression level of the reference gene At5g12240 (Czechowski et al., 2005) and are means ± se obtained from the pool of plants (n ≥ 12) of 11 independent experiments. Asterisks represent significant differences from WW plants as determined by a Two/K sample permutation test (*, P < 0.05; **, P < 0.01; and ***, P < 0.005).

Among the five SWEET genes studied in WW roots, SWEET13 was the most expressed, followed by SWEET12 and SWEET14, while the transcript levels of SWEET11 and SWEET15 were low. However, water shortage triggered a strong increase in transcript levels of SWEET11 (27.3-fold, P < 0.005) and SWEET15 (10.3-fold, P < 0.005), and to a lesser extent of SWEET14 (5-fold, P < 0.005), SWEET13 (4.2-fold, P < 0.005), and SWEET12 (1.9-fold, P < 0.01), in the roots. Thus, SWEET13 and SWEET11 became the most expressed SWEET genes in the WD roots (Fig. 11B).

Generally, SWEET gene expression was lower than SUC gene expression in both organs and conditions. In addition, SWEET (SWEET11, SWEET12, SWEET13, SWEET15, and SWEET14 if expressed) gene up-regulation was well coupled with SUC2 gene overexpression in both organs under water deficit (Figs. 10 and 11).

Water Deficit Delays Aging and Promotes Hexose Accumulation Independently from Senescence

Some elements, such as the up-regulation of the SWEET15/SAG29 gene and the hexose accumulation noted in WD 30-d-old (30d) plants, are also reported in response to senescence-associated aging (Quirino et al., 2001; Seo et al., 2011). To distinguish the effects of water deficit from those due to aging, the experiment was extended to 49 DAS. A phenotype comparison between WW and WD plants was performed. The contents of the three major free sugars (Suc, Glc, and Fru) and starch content were determined and compared with corresponding 30d plants. In addition, the transcripts of genes related to photosynthesis (RBCS and CAB1) and senescence (SAG12 and SWEET15/SAG29) as well as Suc transporter genes (SUCs and SWEETs) were monitored.

The phenotype observation on 49-d-old (49d) plants showed that WD plants were smaller than WW plants, as already noted at 30 DAS. A strong reddening was observed at the level of petioles in WW plants at 49 DAS, a phenomenon that was not detected in WD plants. Interestingly, bolting was initiated in some WW plants, indicating a switch from the vegetative to the reproductive stage. Conversely, the absence of emerging inflorescences in stressed plants suggests that they were still in the vegetative stage (Fig. 12A).

Figure 12.

Comparison and impact of water deficit on shoot development, sugar and starch contents, and leaf gene expression at 30 and 49 DAS. A, Top view of WW and WD whole rosettes at 49 DAS. White arrows show inflorescence emergence. In B to D, black bars and circles, white bars and circles, black bars and circles with white dots, and white bars and circles with black dots represent 30d WW, 30d WD, 49d WW, and 49d WD plants, respectively. B, Comparison of sugar and starch contents at 30 and 49 DAS in leaves and roots. Different letters represent groups that were significantly different as determined by a Wilcoxon test followed by a multiple comparison test (false discovery rate adjustment; P < 0.05). C, Comparison of AtSAG12, AtSWEET15, AtRBCS, and AtCAB1 gene expression at 30 and 49 DAS in leaves. D, Multivariate analysis of the impact of water deficit in leaves at 30 and 49 DAS on sugar and starch contents and gene expression related to senescence and photosynthesis in leaves. Principal component analysis shows the sample position in the plane defined by principal component 1 (PC1) and PC2, gathering 90% of the total variance (left). Dotted and solid arrows show the main trends relative to water deficit and aging, respectively. Variables used to perform principal component analysis are presented at right, with their projections in the plane defined by PC1 and PC2. Data at 30 DAS are from two experiments already presented in Figure 7 (sugar content) and Figures 4 and 10 (gene expression). Data are means ± se obtained from the pool of plants (n ≥ 6) of two independent experiments.

While Suc content was still higher in WD 49d leaves than in WW 49d ones, Suc content remained unchanged with aging in WW (WW 30d, 12.7 µmol g⁻¹ dry weight; WW 49d, 13.1 µmol g⁻¹ dry weight) and WD (WD 30d, 22.1 µmol g⁻¹ dry weight; WD 49d, 20.5 µmol g⁻¹ dry weight) leaves (Fig. 12B). The same trend for Suc content was also observed for the roots, suggesting that aging did not impact Suc content during this time span in both organs regardless of the water regime (Fig. 12B).

Glc content increased dramatically (9.6-fold) with aging in WW leaves (30d, 8.6 µmol g⁻¹ dry weight; 49d, 82.7 µmol g⁻¹ dry weight). An increase in Glc content was also observed in WD leaves (30d, 31.5 µmol g⁻¹ dry weight; 49d, 66.1 µmol g⁻¹ dry weight) with aging, but to a lesser extent (2.1-fold). As a result, Glc content was higher in WW leaves than in WD leaves at 49 DAS, but the difference was not statistically significant. In the roots, a slight increase in Glc content was observed in WW plants (30d, 21.6 µmol g⁻¹ dry weight; 49d, 29 µmol g⁻¹ dry weight) and WD plants (30d, 33.9 µmol g⁻¹ dry weight; 49d, 41.24 µmol g⁻¹ dry weight). However, Glc accumulation in response to water deficit was still noted at both stages of development (Fig. 12B). Similarly, Fru content changed with the same pattern in both organs with aging in WW and WD plants, but at lower levels (Fig. 12B).

Starch content increased marginally in WW leaves with development (30d, 340.8 µmol g⁻¹ dry weight; 49d, 410.5 µmol g⁻¹ dry weight; Fig. 12B). In the same time span, starch content in WD leaves (30d, 231.9 µmol g⁻¹ dry weight; 49d, 174.9 µmol g⁻¹ dry weight) decreased slightly (−25%) and remained always lower than in WW leaves. In root, starch content stayed weak and constant with development and water deficit (Fig. 12B).

The expression profile of Suc transporter genes in response to water deficit at 49 DAS was in agreement with the profile observed for 30d plants (Figs. 10 and 11). While RBCS transcript level decreased with aging in WW leaves, it remained stable in this time span in WD leaves. As a result, RBCS gene expression was lower in WW 49d leaves than in WD 49d leaves. In contrast, no change was seen in CAB1 leaf expression, whether in response to water deprivation or aging (Fig. 12C). The SAG12 gene, which was poorly expressed in WW 30d and WD 30d leaves, stayed at the same level in WD 49d leaves (Fig. 12C). However, its expression was clearly higher in WW 49d leaves. SWEET15/SAG29 gene expression was weak and constant with development. In contrast, its expression remained always higher and constant in WD leaves than in WW ones, suggesting that its up-regulation in leaves was only due to water deficit and not aging.

To further distinguish the effects of water deficit and aging on the rosette, a principal component analysis was performed using each condition as an individual and the parameters cited previously as variables (Fig. 12D). Principal components 1 and 2 explained 55.2% and 35.2% of the total variance. While principal component 1 was associated mainly with leaf content in Glc and Fru and the relative expression of SAG12 and RBCS, principal component 2 was linked to leaf content in Suc and starch and SWEET15 relative expression. The effects of aging and water deficit were clearly distinguished in accordance with principal components 1 and 2, respectively (Fig. 12D). Water deficit induced a shift in sample position in accordance with principal component 2 regardless of the development stage (30 or 49 DAS). In contrast, aging provoked a shift in sample position in accordance with principal component 1 for WW plants but not for WD plants.

DISCUSSION

Characterization of the Culture System

The aim of this work was the analysis of the shoot-root C partitioning in Arabidopsis in response to a mild and continuous water deficit, with an emphasis on events occurring in the roots. For this purpose, a culture system providing easy and quick access to the roots was required for molecular and biochemical analyses and for RSA monitoring. Moreover, plants had to be grown close to soil conditions, with the possibility of altering soil characteristics such as water content. Different culture systems, with only some of the desired characteristics, have been developed for Arabidopsis, based on plants grown on a thin layer of soil (Ara-rhizotron; Devienne-Barret et al., 2006) or on nylon membranes (rhizoponics; Mathieu et al., 2015). However, we decided to develop a system adapted from the rhizoboxes described by Xiong et al. (2006), where plants are separated from soil by a nylon membrane (Fig. 1). Plants were able to develop and produced seeds with no sign of mineral deficiency and with a healthy root system. Moreover, compared with plants grown in vitro, the roots were protected from light and were not fed with sugars in the medium.

Characterization of the Effects of Water Deficit Applied to Plants Grown in a Rhizobox

The plants were cultured with two different water regimes (WW and WD) corresponding to maximal and half-maximal water-holding capacity. The impact of water deficit on the physiological status of the roots and shoots was first characterized.

At the end of the experiment, the 30d WD plants displayed some of the typical responses of acclimation (Chaves et al., 2003), such as shoot growth inhibition (Fig. 3; Table I), increased R/S ratio (Table I), metabolic changes (Fig. 7), and variation in specific gene expression (Figs. 4, 10, and 11). The decrease in root water content (Table I) clearly reflected a reduction in soil water availability. Because roots are in direct contact with drying soil, they represent the first organ coping with soil moisture reduction. The water content was more reduced in the roots than in the rosette (Table I). Moreover, the scatter of rosette water content values around the mean was rather low but higher for root water content values in WD plants (Supplemental Fig. S1A). This may reflect an acclimation phenomenon to keep the rosette water content as stable as possible, while the root water content was adjusted to maintain the flux of water, which is linked to the difference between the soil and root water potential (Tardieu and Davies, 1993). Interestingly, leaf biomass decreased concomitantly with root water content under water deficit (Supplemental Fig. S1B). Thus, root dehydration in drying soil altered leaf growth (Supplemental Fig. S1B), likely because less water was delivered to the leaves (Westgate and Boyer, 1985; Davies et al., 1990). However, despite a reduced rosette water content (Table I), no sign of wilting was noticed during the experiments (Fig. 3), regardless of the plants. This could be due to an osmotic adjustment, as suggested by the decrease in leaf osmotic potential (Table I), thus avoiding a strong drop in turgor pressure (Hare et al., 1998). Altogether, these results highlight a fine-tuning in water utilization and loss, presumably by adjusting water uptake, growth, and transpiration to acclimate plants to a reduced supply of water (Chaves et al., 2003).

A strong reduction in leaf area and biomass was induced by water shortage (Fig. 3; Table I), another acclimation phenomenon, to restrict water losses by transpiration (Chaves et al., 2002). The water deficit strongly reduced the rosette area and, to a lesser extent, the leaf number (Table I). Therefore, the reduction in rosette area was due mainly to a decrease in individual leaf expansion, indicating that growth was more affected than development.

Root growth was strongly diminished (Table I; Fig. 5), but less than shoot growth, as indicated by the increased R/S ratio. Differences in the sensitivity of growth for both organs have been reported for several crop species (Sharp et al., 2004) and Arabidopsis plants under osmotic stress (van der Weele et al., 2000) or water deficit (Hummel et al., 2010). Sustained root growth at the expense of leaf is a major acclimation mechanism to maximize the water uptake and minimize water losses (Chaves et al., 2003; Verslues et al., 2006). While root biomass and total length of the lateral roots were strongly reduced by water deprivation (Table I), the length of the primary root was only a little shorter (Fig. 5). Thus, in response to a moderate water deficit, the primary root still grew at an appreciable rate while the elongation rate of the lateral roots was inhibited (Fig. 5). This is in accordance with van der Weele et al. (2000) and could be interpreted as a strategy to improve the depth exploration of the soil to reach wetter zones. Experiments on root patterning allowed further investigation of root growth. The average shapes of both WW and WD roots were rather similar, except in shallow soil (Fig. 5). This indicates that the density of the roots was lower in WD plants but that the soil area explored was more or less maintained between both WD and WW plants. In maize (Zea mays), the study of recombinant inbred lines with contrasting lateral root number and length demonstrated that plants with reduced lateral root density were more tolerant to drought (Zhan et al., 2015). This was related to a reduced metabolic investment in soil exploration, thus favoring the acquisition of limited resources such as water. Remarkably, those results were confirmed in field-grown plants and also in plants grown on mesocosms, a culture system with many similarities to the rhizobox used in this study. The soil area explored by roots was thus less affected than the total length of lateral roots, suggesting a more efficient positioning of roots for WD plants (Fig. 5), presumably to improve the likelihood of finding a wetter soil area. Indeed, it has been demonstrated that roots exhibit hydrotropism (Takahashi et al., 2003) and hydropatterning (Bao et al., 2014) to control root orientation and the positions of lateral roots along the main root, respectively. The reduction in the area explored by WD roots was observed mainly in shallow soil (Fig. 5), since the area colonized was similar deeper in the rhizobox (Fig. 6). In rhizoboxes as in natural conditions, soil dried from the surface in contact with air; thus, shallow soil was subjected to daily variations in water content caused by evaporation-watering cycles. In contrast, bottom soil was characterized by constant soil water content during plant growth. Deak and Malamy (2005) speculated that Arabidopsis plants optimize their root system by repressing root proliferation into regions where less water is available. This is in accordance with our data showing that the extent of lateral proliferation was clearly lower in shallow soil in response to water deficit (Fig. 6). Altogether, these data demonstrate that plants coping with water deficit acclimated by sustaining growth into deeper soil (Fig. 5), while lateral root proliferation was mostly inhibited in shallow soil (Fig. 6), to maintain soil exploration and water uptake in a metabolically efficient way. This confirms that plants exhibit a high degree of phenotypic plasticity by adjusting root growth rate, root growth direction, and root branching to efficiently explore soil (Satbhai et al., 2015).

Although the water deficit applied to plants grown in rhizoboxes was moderate, it generated changes in the expression of water deficit-responsive genes. Indeed, overexpression of the RD29a gene (Chen et al., 2005; Xiong et al., 2006; Ramírez et al., 2009; Sharma and Verslues, 2010; Yamada et al., 2011) in WD leaves (Fig. 4B) and down-regulation of the TIP1.2 gene (Alexandersson et al., 2005; Boursiac et al., 2005) in WD roots (Fig. 4D) confirmed the occurrence of molecular perception of water deficit in both organs (Harb et al., 2010).

Taken together, our results give evidence that a continuous and moderate water deficit applied to plants grown in rhizoboxes produced classical responses to water shortage at the molecular and physiological levels in both shoots and roots and that rhizoboxes are well suited for studying the effect of water deficit, particularly in roots.

In Leaves, Water Deficit Triggered Higher Sugar Content, Which Could Support Enhanced Phloem Loading

An accumulation of sugars was observed in leaves under water deficit (Fig. 7), as reported already in Arabidopsis (Taji et al., 2002; Hummel et al., 2010; Sperdouli and Moustakas, 2012), grapevine (Vitis vinifera; Cramer et al., 2007; Medici et al., 2014), and maize (Kim et al., 2000). In accordance with Hummel et al. (2010), this increase accounted for a marginal part of the decrease in leaf osmotic potential (less than 2%), suggesting another role for the accumulation of sugars in this organ. Starch content, although lower under WD, remained at an appreciable level at midday (Fig. 7). The changes in starch and free sugar contents (Fig. 7) suggest that moderate water deficit did not lead to sugar starvation, as reported in trees (McDowell, 2011), but rather to an increase in free sugars available, despite a reduced ¹⁴C assimilation (Fig. 8B). The reduction in ¹⁴C assimilated (Fig. 8B) on a leaf area basis is certainly caused by a restricted CO₂ diffusion (decreased stomatal conductance plus decreased mesophyll conductance; Flexas et al., 2006a) as a consequence of moderate water deficit. Nevertheless, a slight decrease in RBCS transcripts occurred (Fig. 4A), as in tobacco (Nicotiana tabacum; Kawaguchi et al., 2003) or tomato (Solanum lycopersicum; Bartholomew et al., 1991) subjected to water deficit. This suggests that metabolic impairment also may occur (Flexas et al., 2006a) through mechanisms involving gene repression when stomatal closure develops over several days (Flexas et al., 2006b). The reduction in RBCS transcripts (Fig. 4A) also could be related to higher soluble sugar contents (Fig. 7), since photosynthetic gene expression is inhibited by high carbohydrate contents (Sheen, 1990; Krapp et al., 1993; Van Oosten and Besford, 1994). In this context, higher sugar contents would represent a basic mechanism for the regulation of photosynthesis by sink organs (Krapp et al., 1993) when the production of photoassimilates significantly exceeds the sink capacity to use them. Indeed, leaf growth is reduced earlier and more intensively than photosynthesis (Muller et al., 2011), resulting in an increase in the content of soluble sugars (Fig. 7), despite photosynthesis decline under water deficit (Quick et al., 1992; Hummel et al., 2010). The products of photosynthesis are tightly tuned to the plant demand (Ainsworth and Bush, 2011), and despite a reduced C demand for growth, Suc increased (Fig. 7), although its level is reported to be rather constant (Stitt and Zeeman, 2012). Hummel et al. (2010) proposed that the excess C not used by the leaf for growth, respiration, and osmotic adjustment could be exported to the roots.

Based on this hypothesis, ¹⁴CO₂ pulse-chase experiments were carried out to establish if water deficit enhanced ¹⁴C allocation to the roots. Pulse-chase experiments first demonstrated that ¹⁴CO₂ assimilation was reduced under water deficit (Fig. 8B), agreeing with decreased RBCS level transcripts (Fig. 4A) and leaf growth reduction (Fig. 3; Table I). As expected from the increased R/S ratio, a greater part of ¹⁴C was allocated to roots in WD plants (Fig. 8D). In addition, even though ¹⁴CO₂ assimilation was reduced (Fig. 8B), greater amounts of ¹⁴C per rosette dry weight unit were transferred to the roots under water deficit (Fig. 8E). In accordance with the hypothesis by Hummel et al. (2010), we demonstrate that a moderate water deficit enhanced C export from the leaves to the roots (Fig. 8), at least during the light period. In contrast, the lower proportion of root tips radiolabeled in WD plants might suggest that the velocity of phloem sap was reduced. Wardlaw (1969) underlined the tight relationship existing between the slower growth induced by water deficit and the reduced velocity of photoassimilate movement in wheat (Triticum aestivum). In addition, several studies focusing on the allocation of assimilated C between source and sinks found that the translocation of recently assimilated C from source leaves to sinks was delayed under water deficit (Plaut and Reinhold, 1965; Wardlaw, 1969; Deng et al., 1990; Li et al., 2003; Ruehr et al., 2009). A modeling experiment (Hölttä et al., 2009) suggested that decreasing xylem water potential under water deficit also limits sugar transport in the phloem, due to limited xylem-phloem water exchange (Sevanto, 2014). Our results indicate that C export to the roots was enhanced under water deficit (Fig. 8), even though the velocity of phloem sap was probably reduced, at least during the chase period. Because ¹⁴C allocation to root depends on the rate of phloem loading and the velocity of phloem sap, we propose that phloem loading was enhanced, leading to a higher Suc concentration in the phloem sap in WD plants. The enhanced phloem loading is consistent with the fact that a higher sugar concentration is needed in the phloem to attract water from the xylem when water availability is reduced (Hölttä et al., 2009).

There was also a tight relationship between leaf growth and C export to the roots (Fig. 9). The rosette biomass under water deficit was negatively correlated with the proportion of ¹⁴C recovered in the roots (Fig. 9B) and ¹⁴C exported to the roots per leaf dry weight unit (Fig. 9C). Thus, the more reduced leaf growth was by water deficit, the greater proportion and amount (per g of leaf dry weight) of assimilated C were exported to the roots. This result highlights that the decrease in leaf growth may have resulted in a higher pool of C available for export to the roots, as suggested previously by Hummel et al. (2010).

The enhanced Suc phloem loading could be supported by an increased activity of Suc transporters. In Arabidopsis, Suc is apoplastically loaded into the phloem companion cell in minor and major veins of source leaves and translocated to sinks in the phloem sap. AtSWEET11 and AtSWEE12 (plasma-membrane Suc effluxers; Chen et al., 2012) release Suc from mesophyll cells or phloem parenchyma cells into the apoplast prior to loading. Then, the companion cell Suc-specific Suc:H⁺ symporter AtSUC2 loads Suc into the phloem (Truernit and Sauer, 1995; Stadler and Sauer, 1996; Gottwald et al., 2000). In potato (Solanum tuberosum), tobacco, and sugar beet (Beta vulgaris), transcript levels of SUC2/SUT1 are tightly correlated to [¹⁴C]Suc uptake in plasma membrane vesicles (Lemoine et al., 1996; Vaughn et al., 2002; Leggewie et al., 2003). In Arabidopsis, recent studies demonstrated that the up-regulation of SUC2 results in enhanced [¹⁴C]Suc uptake into the veins of leaf discs, which entails higher phloem loading and a greater part of ¹⁴C being exported to the roots (Dasgupta et al., 2014). Therefore, it is reasonable to conclude that the increase in Suc transporter gene expression relates to an increased Suc transport activity. In accordance with a putative increased phloem loading, our data confirmed that AtSWEET11, AtSWEET12, and AtSUC2 genes were up-regulated in leaves during water deficit (Fig. 10). SWEET13, a Suc effluxer that was up-regulated in the sweet11/sweet12 double mutant and, therefore, could contribute to the first step of phloem loading (Chen et al., 2012), was also induced (Fig. 10), supporting this hypothesis. Several other results provide evidence that leaf Suc transporter gene expression is tuned by water deficit (Fig. 10), as in grapevine (Medici et al., 2014), rice (Oryza sativa; Ibraheem et al., 2011), and Arabidopsis (Gong et al., 2015). As source strength is defined as the ability to export sugar to sinks, our data demonstrate that the leaves of WD plants were a stronger source when the data were weighted by leaf mass. The enhanced C export to roots could be interpreted as a fine-tuned way to maintain phloem transport under reduced water availability.

Roots under Water Deficit: Highlight on the Hidden Half of the Plant

The growth and development of sink organs, such as roots, depend on C supply by sources. The growth and development of roots are related to local hexose concentration, released from unloaded and cleaved Suc, in Arabidopsis (Freixes et al., 2002). Indeed, the elongation rate of primary and secondary root is under the control of C availability when water supply is not limited (Freixes et al., 2002; Muller et al., 2011). However, this tight relationship between C availability and root growth seemed to disappear under water deficit. Reduced WD root growth did not seem to be caused by carbon starvation, as sugars and ¹⁴C accumulated in the roots (Figs. 7 and 8F). Indeed, C export to the roots was even enhanced, resulting in a higher R/S ratio, suggesting that the WD plants invested more C in their roots. Increasing C allocation to roots has been described as a strategy to resist water deficit in a drought-tolerant genotype of wheat (Nicolas et al., 1985).

However, it has to be pointed out that sugar accumulation also could be linked to changes in metabolism. For example, Nicolas et al. (1985) reported that lower root respiration of the drought-tolerant genotype of wheat compared with the sensitive one contributed to more sugar accumulation in roots of the tolerant genotype under drought.

Nevertheless, sugars have to be unloaded in the root cells in order to sustain the hydrostatic pressure gradient between source and sink organs and maintain phloem sap flow. Phloem unloading occurs in a symplastic way on a relatively small proportion of the root (i.e. the growing root tip; Oparka et al., 1994). Indeed, phloem transport seems to be symplastically isolated (Kempers et al., 1998), suggesting apoplastic unloading along the whole phloem path. Since Suc apoplastic unloading requires Suc transporters, we investigated which of those were expressed in roots and responsible for this function.

In 30d WW roots, the analysis of transcript levels of the nine SUCs revealed that five of them were expressed in roots, namely SUC1, SUC2, SUC3, SUC4, and SUC5, even though the latter was poorly expressed. SUC1-induced transport of Suc in the pollen tube was tightly linked to an increased turgor of the pollen tube for growth (Stadler et al., 1999). A similar function in the root cannot be excluded, since SUC1 is expressed in growing areas of the roots (Sivitz et al., 2007). Therefore, decreased SUC1 expression (Fig. 11A) might be associated with the reduced root growth observed (Fig. 6; Table I) and, more particularly, with a reduction in the proportion of growing areas of the root (fewer lateral roots). SUC4, as a Suc:H⁺ symporter, translocates Suc from the vacuole to the cytosol in the natural environment of the Arabidopsis cell (Schulz et al., 2011). Under water deficit, its up-regulation (Fig. 11A) may contribute to regulating the transport and vacuolar storage of Suc into the root cells. However, SUC2 is important for the retrieval of leaked Suc into the phloem of Arabidopsis (Srivastava et al., 2008; Gould et al., 2012). A higher transcript level of SUC2 (Fig. 11A) could be associated with a greater need for Suc retrieval along the phloem pathway, since C export to the root was enhanced (Fig. 8), under water deficit. Because ZmSUT1, an ortholog of AtSUC2 in maize, has been demonstrated to mediate Suc efflux into sinks (Carpaneto et al., 2005), SUC2 also may be involved in phloem unloading. Thus, its up-regulation in roots (Fig. 11A), under water deficit, also may favor Suc unloading in the “release phloem”. As a result, this would allow phloem sap movement by maintaining the hydrostatic pressure gradient between leaves and roots. SWEET11 to SWEET15 were overexpressed in WD roots (Fig. 11B), perhaps to support the increased Suc efflux from the transported phloem to the surrounding tissue, as it has been proposed that SWEETs might be involved in Suc leakage during phloem transport (Ludewig and Flügge, 2013; Chen, 2014). AtSWEET11 and AtSWEET12 were recently demonstrated to transport hexoses and participate in the delivery of sugars to secondary xylem (Le Hir et al., 2015). The root systems of 30d plants contained large secondary structures in their conducting tissues; therefore, we cannot rule out a similar role for AtSWEET11 and AtSWEET12. Interestingly, as observed in the rosette, there was a co-up-regulation of SUC2 and SWEET genes in the roots under water deficit, underlying a similar response pattern in source and sink organs, a quite unexpected result.

These data provide evidence for a transcriptional regulation of root Suc transporter expression in response to water deficit (Fig. 11). The origin of this regulation remains unknown but could be related either to the increase in sugar concentration in the roots or to a not yet demonstrated direct regulation by water status. However, according to the push hypothesis of shoot-root interaction (Farrar and Jones, 2000), the up-regulation of SUC2 and SWEET11 to SWEET15 genes in WD root (Fig. 11) could contribute to draining the C from the shoot. Water deficit seemed to enhance source strength, but it cannot be excluded that root sink strength was also increased to support the increased C export.

Overexpression of the SWEET15/SAG29 Gene in Leaves Was Water Deficit Dependent

SWEET15/SAG29 gene induction in leaves was described previously as a senescence-associated process (Quirino et al., 1999), and induction of leaf senescence by abiotic stress has been widely reported (Buchanan-Wollaston, 1997; Lim et al., 2007; Wingler and Roitsch, 2008). Therefore, SWEET15/SAG29 induction in WD leaves, together with an increase in sugar, could be attributed to either water deficit or accelerated senescence. To clearly distinguish both effects, the water deficit experiment was extended to 49 d to check if senescence-associated events would be induced further. Sugar accumulation is tightly correlated to leaf senescence in Arabidopsis (Quirino et al., 2001) because senescence is responsive to sugars (Pourtau et al., 2006). In WW leaves, a stronger hexose accumulation with aging, coupled with a higher AtSAG12 transcript level, a highly senescence-specific gene (Gan and Amasino, 1995), indicate leaf aging-related senescence at 49 DAS (Fig. 12, B and C). In addition, RBCS expression decreased in WW plants, and many plants displayed bolting. On the contrary, in WD plants, the AtSAG12 transcript level stayed low and constant and RBCS expression remained stable (Ludewig and Sonnewald, 2000). Finally, WD plants also displayed a late flowering time (Fig. 12A), which is in agreement with a delay in senescence, since senescence induction is linked to floral initiation in Arabidopsis (Wingler and Roitsch, 2008). Delaying floral initiation allows more time to accumulate vegetative biomass to be remobilized later toward seeds (Metcalf and Mitchell-Olds, 2009). To conclude, there are numerous clues that SWEET15/SAG29 up-regulation, in WD 49d leaves (Fig. 12C) displaying delayed senescence, was rather specific to moderate water deficit. Its potential role in the redistribution of Suc in plants may have to be reassessed in the light of these results.

CONCLUSION

The design and use of rhizoboxes allowed us to precisely study the response of the Arabidopsis root system to a mild drought and to relate this response to C allocation. Despite a much shorter total root length in WD plants, the surface explored by the roots was less affected than in WW plants. This is in agreement with previous results showing that plants can acclimate by reshaping their root systems. In parallel to this reshaping, a significant amount of C was reallocated to the roots. ¹⁴CO₂ pulse-chase experiments demonstrated a relative increase in phloem transport to the roots in agreement with the higher R/S ratio. This increase in Suc transport confirms, on a quantitative basis, the previous hypothesis suggesting that the excess of C, resulting from the higher effect of water deficit on growth than photosynthesis, could be available for export to the roots (Hummel et al., 2010). The enhanced export measured in our experiments could be linked to the higher transcript levels for SUC2, SWEET11, and SWEET12 in the leaf. Unexpectedly, the same set of Suc transporter genes (SUC2, SWEET11, and SWEET12) was induced in the roots, raising questions about their role (phloem retrieval and/or phloem unloading?). The rhizobox system developed in this study appears perfectly suited for further studies to pinpoint the cellular localization of these transporters and determine their role in Suc fluxes in the roots.

Altogether, these data demonstrate that, when challenged with a mild and continuous soil water deficit, Arabidopsis plants reduced their growth because of the reduced water availability for cell expansion, but more C was allocated to the roots to sustain their development, albeit at a slower rate.

MATERIALS AND METHODS

Rhizobox, Plant Materials, Growth Conditions, and Harvest

The rhizobox was made with two Plexiglas plates (1 cm thick) of 20 cm × 20 cm (width × length) and spaced on both sides with 1-cm × 20-cm (width × length) bars cut from the same Plexiglas plate. The lower part was sealed with tape (pierced to allow water to flow through), and the upper part was left open. The rhizobox, in the open position, was filled with 500 mL of sieved organic compost. A nylon membrane with a 0.7-µm mesh width (Sefar Nitex 03-7/2) was placed inside the rhizobox on the front plate, and the rhizobox was then closed with two pliers. Rhizoboxes were maintained at two water regimes depending on the growth conditions (WW, 0.8 g water g⁻¹ compost; WD, 0.4 g water g⁻¹ compost) from sowing to plant harvest (30 or 49 DAS, depending on the experiment). Arabidopsis (Arabidopsis thaliana Columbia-0 ecotype) seeds were sown in a drop of nutrient agar medium (0.65% (w/v) agar, Murashige and Skoog medium; M0232; Duchefa Biochemie) placed between the front Plexiglas plate and the nylon membrane at the top of the rhizobox (two to four plants per rhizobox). The top of the rhizobox was sealed with cellophane during 5 to 7 DAS to maintain a high hygrometry and prevent seedling dehydration. Rhizoboxes were wrapped with opaque plastic to avoid exposing roots to light and placed in a phytotron at 23°C/18°C, 10-h/14-h day/night, with a 100 µmol m⁻² s⁻¹ light intensity at the level of plants and a 50%/70% day/night relative hygrometry. Rhizobox soil water content was kept constant by daily watering, with the amount of missing water determined by weighing rhizoboxes.

Harvests were performed at 30 and 49 DAS. Root and rosette were harvested separately and frozen in liquid nitrogen for RNA and sugar extractions. Additional plants (rosette and root) were also collected to measure physiological parameters (biomass and water status).

Leaf Area, Biomass, Water Content, Osmotic Potential, and R/S Ratio

The projected leaf area was determined with photographs of 30d plants with the threshold color plugin of ImageJ software (http://imagej.nih.gov/ij/). Leaf number was counted on the same photographs.

Water content was determined as follows in rosette and root. Fresh weight was scored immediately after separation of root and rosette. Dry weight was scored after 24 h at 80°C. Water content was then determined as (fresh weight − dry weight)/fresh weight. For rosette, fresh weight, dry weight, and water content were measured on individual plants. For root, fresh weight, dry weight, and water content were measured on pools of two to three plants grown in the same rhizobox. R/S ratio, defined as root dry weight/rosette dry weight, was determined on pools of two to three plants grown in the same rhizobox.

Rosette osmotic potential was measured on three to five excised leaves per plant pooled in 2-mL syringes. The syringes were frozen successively in liquid nitrogen and thawed at room temperature three times. Sap was then extracted and collected in 15-mL tubes by centrifugation (8,000g, 10 min). Ten microliters of the resulting sap was analyzed using a vapor pressure osmometer (Wescor Vapro 5520). Osmotic potential was calculated from osmolarity using the Van’t Hoff equation at 23°C.

Soluble Sugars and Starch Measurements

Glc, Fru, and Suc were extracted from approximately 10 mg of lyophilized tissue sample by three washings (1.5 mL and twice with 0.5 mL) in methanol:chloroform:water (12:5:3, v/v/v). Supernatants containing soluble sugars were pooled, mixed with 0.6 volume of water, and centrifuged. The upper aqueous phase was collected and evaporated at 50°C with a concentrator (MiVac Quattro; Genevac). Soluble sugars were resuspended in 500 µL of water and quantified using the Suc/Fru/d-Glc Assay Kit (Megazyme).

Starch content was measured from the pellet obtained after methanol:chloroform:water (12:5:3, v/v/v) washings using the Total Starch HK Assay Kit (Megazyme).

RNA Extraction and Complementary DNA Synthesis

Total RNA was extracted from frozen ground tissue (TissueLyser II; Qiagen) as described by Kay et al. (1987). RNA quantity and quality were checked by optical density at 260 nm and agar gel electrophoresis, respectively. The complementary DNA was synthesized from 1 µg of total RNA after DNase treatment (Sigma-Aldrich) using Moloney murine leukemia virus reverse transcriptase (Promega).

Gene Expression Analysis

A set of 16 genes coding for Suc transporters from two distinct families, SUC (AtSUC1, AtSUC2, AtSUC3, AtSUC4, AtSUC5, AtSUC6, AtSUC7, AtSUC8, and AtSUC9) and SWEET from clade III (AtSWEET9, AtSWEET10, AtSWEET11, AtSWEET12, AtSWEET13, AtSWEET14, and AtSWEET15), were studied in rosette and root. An initial screen for these genes was performed by RT-PCR after 40 cycles using GoTaq Flexi DNA Polymerase (Promega; 95°C for 30 s, 60°C for 30 s, and 72°C for 30 s) and was analyzed by 2% agar gel electrophoresis.

For further analyses by RT-qPCR, only Suc transporter genes that showed amplification in RT-PCR were studied. Quantitative PCR was performed on 96-well plates with a MasterCycler Realplex² (Eppendorf) using GoTaq qPCR Master Mix (Promega). Relative expression was determined according to the 2^-ΔCt method. Target gene expression was normalized to the expression of the plant gene At5g12240 (Czechowski et al., 2005). The primers used are listed in Supplemental Table S1.

¹⁴CO₂ Pulse-Chase Labeling

Labeling experiments of whole plants were performed on 30d plants grown in rhizoboxes in a sealed Plexiglas chamber (upper part exposed to the labeled air; 5 L) illuminated with a light intensity of 100 µmol m⁻² s⁻¹. The Plexiglas chamber always contained one rhizobox with two WW plants and one rhizobox with two WD plants. ¹⁴CO₂ with a specific activity of 54 mCi mmol⁻¹ (20 µCi or 74 kBq) was released in the sealed chamber by acidification of sodium bicarbonate (pulse), leading to an increase in the CO₂ level within the chamber lower than 2 μL mL⁻¹. The ambient CO₂ concentration of the air was usually 480 μL mL⁻¹.

After a 15-min pulse, the flux was directed to the trap (10% KOH) to clean the air of ¹⁴CO₂ during 5 min. Then, the chamber was opened and plants were removed from the assimilation chamber to be exposed under light (100 µmol m⁻² s⁻¹) to perform the chase during 45 min. When the chase was over, roots were separated from the rosette. Both were then lyophilized for 4 d, exposed on PhosphorImager screens (Storage Phosphor Screen; Molecular Dynamics), and scanned with a resolution of 25 µm pixel⁻¹ (Scanner Typhoon; GE Healthcare). Tissues were then incubated in a digestion buffer (perchloric acid:30% hydrogen peroxide:0.1% Triton X-100, 56:17:27, v/v/v) at 55°C during 24 h. Scintillation cocktail (Ecolite) was then added, and samples were counted for radioactivity by scintillation spectroscopy (Tri-Carb2910; Perkin Elmer).

RSA Analysis

The analysis of length, area, and shape of the root was carried out on photographs of the radiolabeled root system scanned previously with a resolution of 25 µm pixel⁻¹.

To determine primary and lateral root lengths, the root system was entirely and manually drawn in white on a black background using The Gimp software. From the generated image, primary root and lateral root lengths were estimated using the RootReader2D software (Clark et al., 2013; Supplemental Fig. S2).

The root system shape was first built with multiple points using ImageJ software. (x; y) coordinates of multiple points were used to model a raw root system shape, where point (0; 0) corresponds to the separation point between rosette and root. From the raw root system shape, (x; y) coordinates were extracted with a 0.5-cm specific interval on the y axis using the PlotDigitizer software to produce the net root shape modeled with a 0.5-cm resolution on the y axis. For each root side, x coordinates with the same y coordinates were averaged to generate the mean shape of the root system for WW and WD plants (Supplemental Fig. S3).

From the modeling of the net root shape, the area colonized was determined by calculating integrals with a 0.5-cm-deep resolution on both sides of each root system (Supplemental Fig. S4).

Statistical Analysis

Depending on the sampling size, nonparametric and parametric tests were performed in an R environment. Principal component analysis was performed using XLSTAT software.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Impact of water deficit on water content in both organs and relationship with rosette dry weight.

• Supplemental Figure S2. Method to determine the principal and total lengths of lateral roots.

• Supplemental Figure S3. Method to determine the average shapes of the WW and WD root systems.

• Supplemental Figure S4. Method to determine the area colonized by the root system.

• Supplemental Table S1. Specific primers used in this study.

Glossary

RSA

root system architecture

R/S

root-to-shoot

C

carbon

WW

well-watered

WD

water deficit

DAS

days after sowing

RT

reverse transcription

RT-qPCR

reverse transcription-quantitative PCR

30d

30-day-old

49d

49-day-old