Low-glutathione mutants are impaired in growth but do not show an increased sensitivity to moderate water deficit

Summary

Glutathione is considered a key metabolite for stress defense and elevated levels have frequently been proposed to positively influence stress tolerance. To investigate whether glutathione affects plant performance and the drought tolerance of plants, wild-type Arabidopsis plants and an allelic series of five mutants (rax1, pad2, cad2, nrc1, and zir1) with reduced glutathione contents between 21 and 63% compared to wild-type glutathione content were phenotypically characterized for their shoot growth under control and water-limiting conditions using a shoot phenotyping platform. Under non-stress conditions the zir1 mutant with only 21% glutathione showed a pronounced dwarf phenotype. All other mutants with intermediate glutathione contents up to 62% in contrast showed consistently slightly smaller shoots than the wild-type. Moderate drought stress imposed through water withdrawal until shoot growth ceased showed that wild-type plants and all mutants responded similarly in terms of chlorophyll fluorescence and growth retardation. These results lead to the conclusion that glutathione is important for general plant performance but that the glutathione content does not affect tolerance to moderate drought conditions typically experienced by crops in the field.

Keywords:

Phenotypes – Plant resistance to abiotic stress – Leaves – Water resources – Seedlings – Drought adaptation – Glutathione – Arabidopsis thaliana

Introduction

Crop yield is severely constrained by environmental stress factors resulting in a gap between the yield potential and the actual yield. The yield gap is predicted to increase in the future due to climate change and due to increasing temperature and extended phases of moderate to severe drought in particular [1–4]. Understanding the tolerance and protection mechanisms of plants is mandatory to breed crops that are able to ensure high yield even under intermittent phases of stress. Plant growth and stress defense are controlled by a multitude of different factors building tight regulatory networks that provide the plasticity that is required to ensure survival and yield of plants even under adverse conditions. The tripeptide glutathione (GSH) is considered a key metabolite in plant defense reactions against biotic and abiotic stress factors [5]. One important function of GSH is the detoxification of reactive oxygen species and organic peroxides that are frequently formed in access in stress situations [6,7]. H₂O₂ is in part detoxified via the glutathione-ascorbate cycle in which ascorbate peroxidases reduce H₂O₂ at the expense of ascorbate [8]. Dehydroascorbate ultimately resulting from ascorbate oxidation is subsequently reduced by GSH resulting in the formation of glutathione disulfide (GSSG), which is then again reduced by glutathione-disulfide reductases (GRs). The glutathione-ascorbate cycle and dehydroascorbate reductase in particular was recently shown to play a central role in minimizing drought-induced grain yield loss in rice [9]. Arabidopsis mutants lacking GR in the cytosol have been shown to have a slightly less negative glutathione redox potential (E_GSH) [10]. It has also been shown that cytosolic GR plays a crucial role in leaf responses to intracellular H₂O₂ and in regulation of gene expression through salicylic acid and jasmonic acid signaling pathways [11].

Oxidation of two GSH molecules results in one GSSG molecule, which makes E_GSH dependent on both the degree of oxidation and the total amount of glutathione [12]. The glutathione pool in the cytosol, plastids, mitochondria and peroxisomes is, however, in a reduced state with almost 100% GSH (in the low mM range) and only nanomolar concentrations of GSSG [13,14]. GSH is synthesized in two enzyme-dependent steps catalyzed by glutamate-cysteine ligase (GSH1) and glutathione synthase (GSH2). The biosynthetic pathway is controlled by negative feedback control of GSH on the first enzyme GSH1 [15]. With GSH as a key metabolite in stress defense reactions it is not surprising that several independent genetic screens for mutants sensitive to abiotic or biotic stresses resulted in isolation of mutants with diminished GSH content caused by mutations in GSH1 [16–21]. Mutants with intermediate GSH levels of 20 to 40% are frequently described with a wild-type like phenotype under control conditions [18], although some reports also indicate slightly retarded growth [22]. A causal link between GSH and plant growth is particularly emphasized by the Arabidopsis mutants zinc tolerance induced by iron 1 (zir1), which develops as a dwarf and contains only 15% of wild-type GSH, and root meristemless 1 (rml1) in which postembryonic development is largely abolished due to an almost complete lack of GSH [20,21].

As a key metabolite in the glutathione-ascorbate cycle and a putative cofactor for other ROS scavenging systems like glutathione S-transferases (GSTs), glutathione has frequently been linked to general stress responses in plants [5]. During drought stress responses guard cells are known to produce H₂O₂ that may be exploited for signaling but also needs to be detoxified to avoid serious damage [23,24]. Large-scale changes of cellular redox homeostasis and particularly of E_GSH have been considered to link primary stress responses to downstream targets [12,25]. Stress-dependent changes in E_GSH can be visualized in live cells with redox-sensitive GFP (roGFP) [13,26,27] and the question has been raised whether severe water stress might trigger an oxidative response that might be involved in drought signaling. While harsh hypo-osmotic treatments (1 M mannitol) did not cause any acute effects as measured by the roGFP2 biosensor [28]. Jubany-Mari and colleagues reported a gradual drought-induced oxidative shift of about 10 mV in the cytosol over a period of several days [29].

The latter finding raised the question whether changes in E_GSH participate in the water deficit response and whether mutants with altered E_GSH are impaired in their response. First experiments to answer this question have provided contradicting results. Overexpression of the first and regulatory enzyme in GSH biosynthesis, GSH1, in tobacco has been reported to confer tolerance to drought stress [30]. Consistent with this an increased GSH level was reported to confer tolerance to drought and salt stress while the partially GSH-deficient mutant pad2 displays a significantly lower survival rate than wild-type plants after a two-week drought treatment [31]. In contrast, it has been reported recently that GSH deficiency in pad2 plants does not affect the water deficit response during a 9-day drought period [32]. Similarly, it has also been reported that partial depletion of GSH in the gsh1 mutant alleles cad2, pad2, and rax1, did not adversely affect the leaf area of seedlings exposed to short-term abiotic stress [22]. Surprisingly, the negative effects of long-term exposure to oxidative stress and high salt concentrations on leaf area were less marked in the GSH synthesis mutants than in wild-type plants. The apparent contradiction between these studies may result from different stress treatments and different scoring systems recording either survival [31] or biomass increase and leaf area [22,32]. Furthermore, the informative value of multiple studies on the role of GSH in stress tolerance is limited by the fact that frequently only individual mutants are compared to wild-type plants. Schnaubelt et al. considered this point by testing three different mutants, which, however, all contained intermediate levels of GSH with little phenotypic variation under non-stress conditions [22,33]. In addition, the severe stress regimes used in this study with high salt (75 mM NaCl) and osmotic stress (100 mM sorbitol) and the evaluation of clearly visible macroscopic markers is unlikely to provide a refined picture of stress sensitivity in Arabidopsis. Such a strategy can be particularly problematic when used to assess whether the growth of mutant or transgenic lines is impacted by changes in stress signaling pathways because difficulties with experimentation including possible physical damage of roots and uptake of osmotica during transfer and possible uptake and breakdown of osmotica [2,34]. Similarly, harsh drought treatment ultimately leading to wilting and death of soil-grown plants is not suitable to compare the performance of different genotypes with different growth characteristics, such as smaller plants [35].

To investigate the potential role of GSH in the response of Arabidopsis plants to moderate water deficit in more detail, we extended the allelic series of GSH-deficient mutants used by Schnaubelt et al. [22,33] by the latest additions, namely the mutants nrc1 [19] and zir1 [20] of which the latter is particularly interesting given its reported dwarf phenotype. Wild-type plants and all mutants were compared side-by-side for their growth and drought tolerance by using advanced non-invasive high-throughput shoot phenotyping enabling continuous recording of growth responses. We demonstrate that GSH-deficient mutants display diminished growth that is more severe in low GSH mutants, while even in mutants with pronounced growth deficits the decrease in GSH does not negatively impact on tolerance to moderate drought treatment.

Materials and methods

Plant material and growth conditions

Arabidopsis thaliana L. (Heyn.) ecotype Columbia-0 (Col-0) was used for all experiments as a wild-type control. Mutants with defects in GSH1 (At4g23100) were provided by the colleagues who reported their isolation and initial characterization. Arabidopsis plants were grown on a mixture of soil (www.floragard.de), sand and perlite in 10:1:1 ratio and kept in controlled growth chambers under long day conditions with 16 h light at 19 °C and 8 h dark at 17 °C. Light intensity was kept between 50 and 75 μE m^-2 s^-1 and relative humidity at 50%. Seeds of wild-type and all mutants were harvested at the same time and used at a similar age for all further experiments. For initial phenotypic and physiological characterization seedlings were grown on vertically oriented agar plates under sterile conditions. Seeds were surface-sterilized with 70% (v/v) ethanol for 5 min and plated on half-strength standard MS medium (M0222.0050; Duchefa, www.duchefa-biochemie.nl), supplemented with 0.5% (w/v) sucrose and solidified with 0.8% (w/v) agar. Seeds were then stratified for 2 d in the dark at 4 °C and germinated under long day conditions with 16 h light at 22 °C and 8 h dark at 18 °C. Light intensity was 75 μE m^-2 s^-1 and relative humidity at 50%.

Details on growth conditions applied in large-scale non-invasive phenotyping experiments are provided in the respective method descriptions.

Analysis of low-molecular-weight thiols

Approximately 20 mg plant material from 5-day-old seedlings grown on plates under sterile conditions was homogenized and extracted in a 10-fold volume of 0.1 N HCl. Samples were centrifuged for 10 min at 4 °C. 25 μL of the supernatant were mixed with 20 μL of 0.1 M NaOH and 1 μL of 100 mM freshly prepared dithiothreitol (DTT) to quantitatively reduce disulfides. Samples were vortexed, spun down and kept for 15 min at 37 °C in the dark. Afterwards, 10 μL 1 M Tris/HCl pH 8.0, 35 μL water and 5 μL 100 mM monobromobimane in acetonitrile (Thiolyte^® MB, Calbiochem, www.merckmillipore.com) were mixed and added to the samples. The samples were vortexed, spun down and kept for 15 min at 37 °C in dark. 100 μL of 9% (v/v) acetic acid were added, samples were vortexed and centrifuged at 13,000 g for 15 min at 4 °C. 180 μL of the supernatant were filled in HPLC vials. Thiol conjugates were separated by HPLC (SpherisorbTM ODS2, 250 × 4.6 mm, 5 μm, Waters, http://www.waters.com) using buffer C (10% (v/v) methanol, 0.25% (v/v) acetic acid, pH 3.7) and D (90% (v/v) methanol, 0.25% (v/v) acetic acid, pH 3.9). The elution protocol was employed with a linear gradient from 4 to 20% D in C within 20 min, with the flow rate set to 1 mL/min. Bimane adducts were detected fluorimetrically (UltiMate^™ 3000, Thermo Fisher, http://www.thermofisher.com) with excitation at 390 nm and emission at 480 nm.

Phenotypic characterization of shoot growth and drought stress response

Shoot growth was analyzed automatically by using the GROWSCREEN FLUORO setup described earlier [36,37]. Seeds of WT and all gsh1 mutant lines were stratified for 3 days at 4 °C in the dark and then placed in pots individually. Subsequently seeds were germinated and on day seven after germination seedlings with similar size were transferred into larger pots (7 x 7 x 8 cm) and randomized on trays with 30 plants on each tray. In exceptional cases plants died after this transfer and were removed from further analysis. The plants were then grown in growth chambers under fully controlled conditions at 22/18 °C, 170 μmol m^–2 s^–1 photosynthetically active radiation, and 8/16 h day/night regime. The soil water content (SWC) was recorded gravimetrically. After initial soaking the SWC was allowed to decrease until a value of approx. 40% was reached. Subsequently this SWC was held through intermittent addition of water. Starting from day 15 after sowing all plants were initially documented for the projected leaf area (PLA) and chlorophyll fluorescence every second day. During the exponential growth phase of shoots in weeks 5 and 6 after sowing all growth parameters were collected on a daily basis. All readings were taken around midday to ensure that the rosettes are oriented almost horizontally above the soil. The PLAs A₁ and A₂ of two consecutive days were used to calculate the relative shoot growth rate (RGR_Shoot) (% d^-1) according to the equation RGR_Shoot = 100 × 1/t × ln(A₂/A₁). Chlorophyll fluorescence was recorded after dark adaptation for at least 30 min with a camera-based system to calculate color coded images of F_v/F_m as a measure of the potential quantum yield of photosystem II. For further analysis, average values for F_v/F_m for whole rosettes were calculated.

For drought stress experiments the plants were split into two subpopulations of which the first population was well-watered throughout the experiment while the second population was exposed to drought from day 24 onwards until growth ceased on day 37. Subsequently plants were again watered to a soil water content of about 40% and allowed to recover. All plants were harvested after 44 days to determine fresh and dry weight.

Determination of rosette morphology

Leaf circumference, rosette compactness, rosette stockiness, and eccentricity were calculated from the PLA as described earlier [36].

Ratiometric roGFP2 imaging

5-day-old Arabidopsis seedlings stably expressing Grx1-roGFP2 in the cytosol were imaged by CLSM. Seedlings were mounted on a glass slide in a drop of distilled water and immediately transferred to a Zeiss LSM780 confocal microscope. Images of root tips and cotyledons were collected with a 40x lens (Zeiss C-Apochromat 40x/1.2 NA water immersion) or a 25x lens (Zeiss Plan-Apochromat 25x/0.8 Imm Korr), respectively. RoGFP2 was excited in multi-track mode with line switching between the 405 and 488 nm laser lines. For roots images were collected with a pixel size of 0.415 μm in x and y and 1.58 μs pixel dwell time and for cotyledons with a pixel size of 0.664 μm in x and y and 2.55 μs pixel dwell time. The pinhole was set to 73 μm (2.0–2.3 Airy units dependent on the wavelengths). Emission was collected from 508 to 535 nm. Autofluorescence excited at 405 nm was collected from 430 to 470 nm. Images were processed using a custom built MATLAB tool with x,y noise filtering, fluorescence background subtraction and autofluorescence correction as described previously [38].

Statistical analysis

Statistical analysis was performed using the software GraphPad Prism 7. Different data sets were analyzed for statistical significance with the One- and two-way Analysis of Variance (One- and two-way ANOVA) followed by Tukey's multiple comparisons test.

Results

Severe osmotic stress imposed by mannitol causes partial oxidation of cytosolic glutathione

To investigate the impact of osmotic challenge on Arabidopsis plants, we initially grew seedlings expressing the E_GSH-sensor Grx1-roGFP2 in the cytosol on 300 mM mannitol to mimic severe drought stress. In addition to wild-type seedlings, the effect of mannitol on E_GSH was also studied in gr1 mutants deficient in cytosolic GR because impaired reduction capacity for GSSG renders the glutathione pool more sensitive to stress-induced oxidation [10]. Consistent with our earlier reports [10] the 405/488 nm fluorescence ratio in wild-type cotyledons grown under control conditions indicated almost complete reduction of roGFP2 while the probe was partially oxidized in the cytosol of gr1 cotyledons (S1A Fig). While growth on 300 mM mannitol for 5 days did not cause a visible oxidation in wild-type cotyledons a very pronounced oxidation was observed in gr1 cotyledons (S1A Fig). Similarly, the 405/488-nm fluorescence ratio for Grx1-roGFP2 in wild-type root tips was close to values for full reduction measured after incubation with DTT, the readout for the cytosol of gr1 roots was significantly higher with intermediate ratios between full reduction and full oxidation of the sensor (S1B Fig). From these measurements, cytosolic redox potentials of about -310 mV in the wild-type and close to the midpoint of the sensor at -280 mV for gr1 can be deduced. In both cases, for wild-type and gr1 seedlings, treatment with 300 mM mannitol did not cause significant changes in the fluorescence ratio during the first 16 h. This indicates that within this time window the seedlings were still fully capable of maintaining their thiol redox homeostasis in the cytosol. In contrast to this observation, continuous growth on 300 mM mannitol for 5 days caused a pronounced ratio shift in wild-type and gr1 root tips. While the ratio values in wild-type root tips reached intermediate values, the ratios in gr1 root tips approached full oxidation in the root cap while the meristematic region appeared still partially reduced (S1C Fig).

Based on our observation that long-term drought stress impacts on E_GSH and causes oxidation in the cytosol, as well as on published reports on decreased drought tolerance of GSH-deficient mutants [31], we set out to test the hypothesis that the GSH content of Arabidopsis plants correlates with drought sensitivity.

Compilation of an allelic series of gsh1 mutants with different glutathione levels

Several genetic screens have led to identification of different gsh1 mutant alleles that have all been reported to be GSH-deficient albeit at different degrees. For the degree of GSH depletion, however, values between 20 and 40% GSH compared to wild-type plants are frequently cited, which precludes unambiguous ranking of the mutants according to their GSH content. While GSH levels may indeed vary between growth conditions, the annotation of mutants with concentration range of GSH makes selection of the most appropriate alleles for comparative experimental work difficult. To rank all five available mutants of the allelic series that can be maintained in their homozygous state according to decreasing GSH content, seedlings were grown under controlled conditions on agar plates and analyzed for their GSH content. The HPLC analysis revealed a separation of wild-type and the different gsh1 mutants into 3 distinct classes (Fig 1A). Wild-type seedlings contained 273 ± 39 nmol g^-1 FW glutathione and rax1 seedlings 142 ± 46 nmol g^-1 FW. The third class included pad2, cad2, nrc1 and zir1 which all contained less than 73 nmol g^-1 FW glutathione. Although not statistically different the measured mean content of glutathione in zir1 was only 45 ± 20 nmol g^-1 FW and thus lower than in the other three mutants of this class. Depletion of GSH due to mutations in GSH1 was accompanied by a concomitant increase of cysteine as one of the GSH1 substrates (Fig 1B).

Soil-grown GSH-deficient mutants display growth phenotypes but are insensitive to moderate water deficit

To study the relationship between glutathione content and biomass gain under drought, we analyzed the growth of gsh1 mutants and wild-type in a sub-lethal drought stress assay. A large population of soil-grown plants was separated into two sub-populations of which one was used as a well-watered control group while the second was exposed to drought stress until growth ceased (Figs 2A and 3). At the end of the drought period plants did not yet show obvious signs of wilting (Fig 3). After the drought period, all plants were watered again reaching a SWC of about 40–50% for another week before the plants were harvested (Fig 2A). Within the population of plants grown with sufficient water supply throughout the entire growth period no distinct growth phenotype could be observed for cad2 (Figs 2B and 4C). The glutathione-deficient mutants rax1, pad2 and nrc1 were consistently smaller than wild-type plants by up to 30% at the end of the 6-week growth period (Figs 2B and 4C). Growth retardation was apparent already on day 22 after sowing (Fig 4A). In contrast to mutants with intermediate levels of GSH, zir1 showed a distinct dwarf phenotype with only 19% of the wild-type biomass.

Lack of water supply for an interim period of 13 days during the entire growth phase of 44 days caused a pronounced growth retardation in wild-type plants by about 50% as compared to the well-watered controls recorded as a decrease in shoot fresh and dry weight as well as PLA (Figs 2B, 2C and 4A–4D and S2A and S2C Fig). The shoot water content at the end of the experiment was similar in control and drought-stressed plants indicating that the applied drought stress did not cause serious damage, such as permanent lesions (S2B and S2D Fig). Direct comparison of drought-stressed gsh1 mutants and the respective wild-type revealed that for the intermediate GSH-deficient mutants the relative retardation was less pronounced than under control conditions or could not even be detected statistically (Figs 2C and 4D and S2C Fig). The major exception with the most pronounced difference to the wild-type is zir1 for which the retardation as clearly visible in all recorded parameters. The severe dwarf phenotype of zir1 enabled further analysis of the PLA at the end of the experiment for all mutants despite the lack of consistent phenotypic differences for the intermediate mutants. Comparison of the PLA with the dry weight (DW) for well-watered control plants and drought-stressed plants resulted in linear relationships indicating that the specific leaf area (PLA g^-1 DW) in each group of plants is not affected by glutathione deficiency (Fig 4E and 4F). To further test whether mutants with very low GSH are more seriously affected by drought than mutants with intermediate levels of GSH and wild-type plants, we plotted the PLA of drought-stressed plants against the PLA of control plants. The linear relationship strongly indicates that even low GSH mutants that grow as dwarfs are not severely affected in their ability to withstand moderate water deficit (Fig 4G).

Drought-induced growth retardation was also apparent from the continuous recording of the PLA (Fig 4A–4D) and the RGR_Shoot calculated from these measurements (Fig 5A and 5B). Regular measurement of the PLA over the entire growth period also enabled the calculation of the relative shoot growth rate (RGR_Shoot) on a daily basis. RGR_Shoot in the control population gradually decreased from 28% d^-1 at the start of the recordings on day 16 after sowing to 12% d^-1 at the end of the growth period (Fig 5A). Over the entire growth period no obvious deviations in RGR_Shoot were apparent between the wild-type and the different gsh1 mutants with intermediated GSH content. For zir1 RGR_Shoot showed a similar decrease over time, but there were a few intermittent days with significantly lower RGR_Shoot than in all other plants (Fig 5A). In the drought-stressed population RGR_Shoot decreased significantly faster during the growth period approaching zero after 13 days of water withdrawal (Fig 5B and 5C). The drought stress period started by chance during a phase when zir1 already showed particularly low values for RGR_Shoot in the control population (Fig 5A and 5B). From this point onwards RGR_Shoot was consistently lower in zir1 compared to all other lines for almost the entire drought period (Fig 6B). Immediately after re-watering on day 37 the RGR_Shoot increased to peak values of about 20% d^-1 on day 41 and subsequent decline to values of about 15% d^-1 on day 43 for all lines including zir1. After re-watering, the glutathione-deficient mutants even showed a tendency of faster growth compared to wild-type plants (Fig 5B and 5C).

With each measurement of all individual plants the potential quantum yield of photosynthesis (F_v/F_m) was recorded after dark adaptation for at least 30 min. From measurements of the whole rosette single average values for the respective rosettes were calculated. In plants kept under well-watered control conditions throughout the entire growth phase, F_v/F_m reached values between 0.71 and 0.72 for wild-type plants and for the mutants with intermediate levels of GSH (rax1, pad2, cad2 and nrc1). Slightly but significantly lower values for F_v/F_m of 0.70 to 0.71 were only found in the more severely GSH depleted zir1. While the differences between control plants and drought stressed plants were very small in all lines under investigation there was a tendency towards slightly increased values for F_v/F_m in drought-stressed plants both at the last day of the drought period and just before harvest at the end of the experiment after full recovery of drought stressed plants (Fig 6A and 6B). This difference reached significance only for pad2, cad2 and nrc1 on the last day of the drought phase and in cad2 at the time of harvest.

Recording of PLA and subsequent image analysis enabled the extraction of additional morphological characteristics of the rosette shape. These factors, included the area by circumference, the compactness of rosettes as a measure for the length of petioles, the rosette stockiness as a measure of how indented the rosette is, and the eccentricity as a parameter describing the shape of the rosette compared to a circle. Generally, all parameters reflected the measurements of PLA (S3 Fig). The only exception was zir1 which showed an increased stockiness compared to all other lines. Due to the earlier decrease in RGR_Shoot zir1 also developed a more compact rosette, a plateau in rosette stockiness and a more pronounced eccentricity after onset of drought.

Discussion

Glutathione affects growth under non-stress conditions

Glutathione-deficient mutants have been identified in genetic screens for heavy metal sensitivity (cad2) [16], stress defense gene expression (rax1) [17], pathogen sensitivity (pad2) [18], lack of cadmium-induced gene expression (nrc1) [19], and iron-mediated zinc tolerance (zir1) [20]. Side-by-side comparison of all available viable gsh1 mutants for their glutathione content revealed different levels in the different mutants. The observation that rax1 contains more glutathione than cad2 and pad2 confirms an earlier report by Parisy et al. [18] and the lowest glutathione content in zir1 is in line with the original report of this mutant [20]. nrc1 has very similar glutathione levels as cad2 which is in line with their similar Cd²⁺-sensitivity [19]. Phenotypic comparison showed that growth of all mutants except cad2 was impaired albeit to different degrees. zir1 mutants were significantly smaller than all others. Comparison of our results with findings from earlier reports on individual mutants or side-by-side comparison of a subset of mutants revealed some pronounced differences. While Schnaubelt et al. [22] reported 14-day-old shoots of rax1, pad2 and cad2 to be significantly smaller than wild-type shoots, the original reports on the initial characterization of cad2, rax1, and pad2 all found no phenotypic difference between mutant and wild-type seedlings [16–18]. While Ball et al. [17] found no effect of light on the phenotypic appearance of rax1 at all developmental stages, a slight retardation of pad2 seedlings observed under low light conditions got reverted under high light [22]. Here, the shoot biomass and PLA of cad2 was found to be higher than in other mutants with intermediate glutathione content. A similar pattern was observed by Schnaubelt et al. albeit not consistently in all data sets [22]. These observations collectively emphasize that it is critical to apply perfectly constant growth conditions and use all mutants together with the wild-type side-by-side. Indeed, GSH levels may vary significantly between experiments due to some non-controlled experimental differences [18]. The use of a complete allelic series of mutants grown under exactly the same conditions in an automated phenotyping setup should avoid such limitations with lab-based phenotyping and provide more robust data. Comparison of shoot phenotypes further supports the conclusion that phenotypes are not linearly correlated with the GSH content as long as the GSH content can be kept above a certain threshold [20]. A possible explanation for the lack of a linear correlation between glutathione content and growth may be that the available mutants originate from different EMS-treated populations that were screened for sensitivities to different stress factors or stress-dependent gene induction. This implies that, despite several rounds of backcrossing, the mutants may still contain additional cryptic mutations that are not linked to glutathione homeostasis. Such mutations may affect growth properties in a pleiotropic way similar to recent findings on ascorbate-deficient mutants [39]. In any case, this further emphasizes the added value of investigating several allelic mutants with slight deviations in GSH particularly in the low GSH range side-by-side.

The lower threshold for maintaining a wild-type-like phenotype is clearly passed in the zir1 mutant resulting in pronounced dwarfism [20]. Glutathione is a key metabolite with essential functions in detoxification, cellular redox homeostasis and as a co-factor [5,40]. GSH is a co-factor in detoxification of methylglyoxal [41] and Fe-S cluster transfer by glutaredoxins [42]. In addition, GSH acts as a S-donor in glucosinolate biosynthesis as well as a co-substrate of ascorbate peroxidases and glutathione S-transferases (GSTs) for peroxide detoxification and of GSTs for conjugation of electrophilic xenobiotics [43–46]. Beyond this glutathione is the most abundant low molecular thiol with low millimolar concentrations in the cell [47] and as such together with glutathione reductases important for maintaining highly reducing redox potentials in the cytosol, plastids, mitochondria and peroxisomes [10,13,14]. With this multitude of functions, it is not surprising that a decrease in GSH levels eventually impair some processes, even though this may be indirectly. Although the experiments reported here were not designed to answer questions on which molecular processes are impaired, the results nevertheless indicate that a certain threshold of GSH depletion needs to be reached before growth is impaired. Below this threshold the developmental phenotype is strongly dependent on the GSH concentrations as evidenced by the zir1 mutant but also the rml1 mutant, which has less than 5% GSH and arrests in growth after germination [21]. The availability of at least one viable mutant with low GSH, which shows already strong and consistent growth impairment, appears particularly useful to further test hypotheses regarding a causal link between GSH content and stress sensitivity.

Glutathione content and drought tolerance

Plants reprogram their metabolism and growth when exposed to water limitation [2] and these changes in growth and physiological status can often be monitored non-invasively at the level of whole plants [48]. The potential quantum yield of photosynthesis F_v/F_m is a parameter for the functional status of photosystem II and it is well known that severe drought stress causes a decline in F_v/F_m [49,50]. At the same time, it is also established that F_v/F_m during the growth process increases with leaf size due to a relative increase of the lamina which typically shows high F_v/F_m values compared to leaf borders with low F_v/F_m value [36]. The slightly increased F_v/F_m in plants exposed to moderate water deficit might thus correspond to reduced growth and concomitant earlier leaf differentiation and hence a higher fraction of leaves with F_v/F_m values typical for full-grown leaves [36]. Similarly, the lower F_v/F_m values for zir1 mutants at the end of the drought period and still after a recovery phase may thus be a consequence of delayed development and less matured leaves compared to wild-type and all other mutants. Alternatively, the lower F_v/F_m with values around 0.70–0.71 may also be indicative of partial photoinhibition in zir1. Photoinhibition and concomitant production of reactive oxygen species (ROS) is a common feature of stress responses. Given that GSH is involved in detoxification of ROS via the glutathione-ascorbate cycle the growth phenotype appears to be consistent with reduced growth in the ascorbate-deficient mutant vtc2-1 [51]. This apparent link between ascorbate content and plant growth, however, has been questioned recently after identification and characterization of a true null allele vtc2-4 that despite low ascorbate content has a wild-type-like phenotype [39].

External supply of 400 μM GSH has been reported to render Arabidopsis plants more drought tolerant [52]. Vice versa pad2 mutants are reported to be more sensitive to drought than wild-type plants [31]. Our results, however, show no such drought sensitivity. Instead, our systematic approach rather indicates that soil-grown Arabidopsis plants exposed to moderate water deficit generally respond with a gradual decline in growth but no signs of wilting at the point of complete growth arrest. This response is not affected by the GSH content. This is even true when GSH is down to 21% of wild-type levels, which causes severe growth retardation in zir1 under non-stress conditions. Direct comparison of wild-type plants and mutants with GSH levels of 62 to 21% compared to wild-type all have similar specific leaf areas, i.e. PLA/DW ratios, irrespective of whether plants had been grown under control conditions or exposed to drought stress (Fig 5). A temporary phase of moderate water deficit caused by lack of water supply for 13 days led to a pronounced decrease in growth, but even a slight increase in the specific leaf area. This result indicates that plants that experienced temporary water deficit subsequently grew slightly faster than well-watered control plants. Re-growth after rewatering has been accounted for by cell expansion resulting in actual growth and not only a reversible change due to increased turgor [53]. Interestingly, plants exposed to mild drought stress have been observed to accumulate sugars and starch [54]. This may contribute to a quick restart of growth after re-watering in all lines including zir1. Important in this context is the observation that the linearity for the specific leaf area across all mutants and the wild-type is maintained, which indicates that even the very low glutathione mutant zir1 did not get severely damaged during the stress treatment.

Conclusion

Our analysis provides a systematic, quantitative foundation to the observation that glutathione deficiency causes retarded growth of both roots and shoots. Direct side-by-side comparison of mutants with different degrees of glutathione depletion show no gradual decrease in growth, but a minor retardation, in mutants with 25–62% GSH and more severe only for mutants with only 21% GSH. This suggests that under non-stress conditions partial depletion of the cellular glutathione pool is tolerable while passing a threshold below about 25% GSH leads to gradual impairment of plant growth.

Our systematic analysis did not show any indication that glutathione content of Arabidopsis plants correlates with drought resistance. This corroborates a recent meta-analysis of published data lead to the conclusion that drought tolerance is more likely linked to up-regulated ascorbate-dependent antioxidant capacity than to the response of the thiol-redox network [55]. Moderate water deficit applied through water withdrawal until shoot growth ceased showed that wild-type plants and all mutants responded similarly in terms of all morphological parameters analyzed, as well as the photosynthetic machinery as analyzed by chlorophyll fluorescence. Taken together the results indicate that glutathione is important for general plant performance, but does not affect tolerance to moderate drought conditions typically experienced by crops in the field. However, drought conditions in the field are typically accompanied by other stress factors, like high light intensities. Future work will thus also need to investigate the response of glutathione-deficient mutants to combinatorial stresses.

Supporting information

S1 Fig [a]
Long-term osmotic stress causes partial oxidation of cytosolic Grx1-roGFP2 in Arabidopsis seedlings.

S2 Fig [a]
Shoot dry weight and shoot water content for wild-type and mutants.

S3 Fig [pdf]
Morphological characteristics of rosettes grown under control and drought-stress conditions.