Oxidative stress is a consequence, not a cause, of aluminum toxicity in the forage legume Lotus corniculatus

Authors

  • Joaquín Navascués,

    1. Departamento de Nutrición Vegetal, Estación Experimental de Aula Dei, Consejo Superior de Investigaciones Científicas, Apartado 13034, 50080 Zaragoza, Spain
    Search for more papers by this author
  • Carmen Pérez-Rontomé,

    1. Departamento de Nutrición Vegetal, Estación Experimental de Aula Dei, Consejo Superior de Investigaciones Científicas, Apartado 13034, 50080 Zaragoza, Spain
    Search for more papers by this author
  • Diego H. Sánchez,

    1. Max Planck Institute for Molecular Plant Physiology, Wissenschaftspark Golm, Am Mühlenberg 1, Potsdam-Golm, 14476, Germany
    Search for more papers by this author
  • Christiana Staudinger,

    1. Department of Molecular Systems Biology, University of Vienna, 1090 Vienna, Austria
    Search for more papers by this author
  • Stefanie Wienkoop,

    1. Department of Molecular Systems Biology, University of Vienna, 1090 Vienna, Austria
    Search for more papers by this author
  • Rubén Rellán-Álvarez,

    1. Departamento de Nutrición Vegetal, Estación Experimental de Aula Dei, Consejo Superior de Investigaciones Científicas, Apartado 13034, 50080 Zaragoza, Spain
    Search for more papers by this author
  • Manuel Becana

    1. Departamento de Nutrición Vegetal, Estación Experimental de Aula Dei, Consejo Superior de Investigaciones Científicas, Apartado 13034, 50080 Zaragoza, Spain
    Search for more papers by this author

Author for correspondence:
Manuel Becana
Tel: +34 976 716055
Email: becana@eead.csic.es

Summary

  • Aluminum (Al) toxicity is a major limiting factor of crop production on acid soils, but the implication of oxidative stress in this process is controversial. A multidisciplinary approach was used here to address this question in the forage legume Lotus corniculatus.
  • Plants were treated with low Al concentrations in hydroponic culture, and physiological and biochemical parameters, together with semiquantitative metabolic and proteomic profiles, were determined.
  • The exposure of plants to 10 μM Al inhibited root and leaf growth, but had no effect on the production of reactive oxygen species or lipid peroxides. By contrast, exposure to 20 μM Al elicited the production of superoxide radicals, peroxide and malondialdehyde. In response to Al, there was a progressive replacement of the superoxide dismutase isoforms in the cytosol, a loss of ascorbate and consistent changes in amino acids, sugars and associated enzymes.
  • We conclude that oxidative stress is not a causative factor of Al toxicity. The increased contents in roots of two powerful Al chelators, malic and 2-isopropylmalic acids, together with the induction of an Al-activated malate transporter gene, strongly suggest that both organic acids are implicated in Al detoxification. The effects of Al on key proteins involved in cytoskeleton dynamics, protein turnover, transport, methylation reactions, redox control and stress responses underscore a metabolic dysfunction, which affects multiple cellular compartments, particularly in plants exposed to 20 μM Al.

Introduction

Aluminum (Al) toxicity is a major constraint of agricultural production on acid soils (pH < 5.6). In tropical America, acid soils cover nearly 850 million hectares (Rao et al., 1993) and, in Brazil, 32% exhibit Al toxicity (Abreu et al., 2003). In acid soils, Al is solubilized into soil solution from aluminosilicates, inhibiting root growth and function (Ma et al., 2001; Kochian, 2005). At the cellular level, the strong binding affinity of Al with oxygen donor ligands, such as proteins, nucleic acids and phospholipids, results in the inhibition of cell division, cell extension and transport (Mossor-Pietraszewska, 2001). At the molecular level, Al stress causes major changes in the expression patterns of genes, some of which are important in the oxidative stress response (Richards et al., 1998; Watt, 2003; Maron et al., 2008). Indeed, exposure of plants to Al elicits the production of reactive oxygen species (ROS), which may cause oxidative damage to cellular components if antioxidant defenses are overwhelmed (Cakmak & Horst, 1991; Boscolo et al., 2003; Darkóet al., 2004; Sharma & Dubey, 2007). Major antioxidants in plants include catalases, superoxide dismutases (SODs), glutathione peroxidases (GPXs) and the enzymes and metabolites of the ascorbate–glutathione pathway. This pathway ultimately reduces H2O2 to water at the expense of NAD(P)H, and involves four enzymes: ascorbate peroxidase (APX), monodehydroascorbate reductase (MR), dehydroascorbate reductase (DR) and glutathione reductase (GR).

The capacity of plants to overcome Al stress involves diverse mechanisms, one of which is the root exudation of organic acids and phenolic compounds (Pellet et al., 1995; Ma et al., 2001; Barceló & Poschenrieder, 2002). The discovery and characterization of an Al-activated malate transporter (ALMT) provide genetic support for an important role of organic acids in withstanding Al toxicity (Sasaki et al., 2004; Hoekenga et al., 2006). In addition, the use of large-scale (‘omics’) technologies has contributed considerably to our understanding of the effects and mechanisms of Al toxicity. This is exemplified by very recent transcriptomic (Kumari et al., 2008; Maron et al., 2008; Eticha et al., 2010) and proteomic (Yang et al., 2007; Zhen et al., 2007; Zhou et al., 2009) studies. However, to our knowledge, the effects of Al stress have not yet been addressed using metabolic profiling or semiquantitative proteomics. Moreover, the implication of oxidative stress as a primary mechanism of Al toxicity is still controversial. Several authors have associated Al toxicity with the induction of oxidative stress (Richards et al., 1998; Ezaki et al., 2000; Sharma & Dubey, 2007), whereas others have proposed that the oxidation of lipids or proteins (markers of oxidative stress) is not directly responsible for the inhibition of root elongation caused by Al (Cakmak & Horst, 1991; Yamamoto et al., 2001; Boscolo et al., 2003). A complicating factor in this controversy is that the increase in antioxidant enzyme activities and ROS production is often interpreted as being indicative of oxidative stress (e.g. Darkóet al., 2004), although these molecules may be involved in ‘oxidative signaling’ under conditions that do not necessarily imply damage to cellular components, and hence oxidative stress (Foyer & Noctor, 2005).

Forage legumes play an important role in the productivity of cultivated pastures because of their high potential for N2 fixation and growth in soils with low fertility. In particular, Lotus corniculatus has an outstanding agricultural importance and wide distribution in South America (Díaz et al., 2005), and is closely related to L. japonicus, a model species for classical and molecular genetics (Handberg & Stougaard, 1992). Previous work has shown that exposure to high Al concentrations triggers a rapid membrane depolarization in L. corniculatus root cells, suggesting a role of this process in the inhibition of root cell elongation (Pavlovkin et al., 2009). Here, we have investigated the implication of oxidative stress in Al toxicity in L. corniculatus using a multidisciplinary approach. Measurements of physiological and biochemical parameters, in combination with semiquantitative analyses of the metabolome and proteome of roots, were performed to identify the metabolic and cellular processes involved in the long-term response of plants to physiologically relevant Al concentrations.

Materials and Methods

Biological material and plant treatments

Seeds of Lotus corniculatus L. cv Draco were surface disinfected with 70% ethanol, transferred to 0.5% agar plates and stored at 4°C for 2 d. Germinating seeds were then incubated at 28°C for 2 d and placed on 1.5% agar plates (8–10 seedlings per plate; Supporting Information Fig. S1a) containing a complete nutrient medium (modified Fahraeus medium; Boisson-Dernier et al., 2001). After 1 wk, seedlings were transferred to 10-l hydroponic vessels containing deionized water with 200 μM CaCl2 and 0, 10 or 20 μM AlCl3 (adjusted to pH 4.0) in a controlled environment cabinet (ASL, Madrid, Spain) under the following conditions: 23°C : 18°C (day : night), 70% relative humidity, 180 μmol m−2 s−1 and a 16-h photoperiod. Plants were harvested after 14 d (Fig. S1b), and roots and leaves were snap-frozen in liquid nitrogen and stored at − 80°C.

Accumulation of Al and production of ROS

The accumulation of Al in roots was visualized using morin (2′,3,4′,5,7-pentahydroxyflavone; Fluka, Buchs, Switzerland), which forms a highly specific complex with Al at acidic pH. The method of Tice et al. (1992) was followed with minor modifications. Roots were washed six times (30 min each) with desorbing solution (1 mM sodium citrate, 5 mM CaCl2, pH 4.0) and frozen for 6 h. Roots were then thawed, washed four times for 30 min each in desorbing solution, washed in buffer (5 mM ammonium acetate, pH 5.0) for 10 min, stained with 100 μM morin in buffer for 60 min and washed again in buffer for 10 min.

The production of ROS in roots was visualized using specific fluorescent probes (Sandalio et al., 2008). To detect superoxide radical formation, roots were preincubated with 100 μM CaCl2 for 30 min, incubated with 10 μM dihydroethidium (DHE; Sigma-Aldrich) in 100 μM CaCl2 for 30 min, and finally washed with 100 μM CaCl2. DHE is oxidized by superoxide radicals to oxyethidium, which is quite stable and fluoresces with excitation at 488 nm and emission at 520 nm. To detect peroxide production, roots were processed as indicated for superoxide radicals, but replacing DHE by 25 μM of 2′,7′-dichlorofluorescein diacetate (DCF-DA; Calbiochem, Darmstadt, Germany). This compound is able to permeate cells, where it is hydrolyzed by intracellular esterases releasing DCF, which becomes trapped inside the cell. DCF reacts with H2O2 and hydroperoxides forming a fluorescent compound with excitation at 480 nm and emission at 530 nm (Sandalio et al., 2008).

Roots were examined using an M165 FC fluorescence stereomicroscope (Leica, Wetzlar, Germany) with a GFP3 filter (excitation 450–490 nm, emission 500–550 nm) for Al and peroxides, and with a DSR filter (excitation 510–560 nm, emission 590–650 nm) for superoxide radicals.

Physiological parameters and oxidative stress markers

Plant growth was assessed by measurement of the leaf and root fresh weight (FW), leaf area and root length. The root and leaf contents of nitrogen (N) were determined with an NA 2100 Nitrogen Analyzer (ThermoQuest, Milan, Italy). The root and leaf contents of Al were measured by inductively coupled plasma-mass spectrometry (ICP-MS) with an ELAN 6000 instrument (Perkin-Elmer, Waltham, MA, USA) at the Universidad Autónoma de Barcelona (Spain). The root contents of potassium (K), calcium (Ca), magnesium (Mg), phosphorus (P), sulfur (S), iron (Fe), copper (Cu), manganese (Mn), nickel (Ni) and zinc (Zn) were measured by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) with an IRIS Intrepid II XDL instrument (Thermo Electron, Waltham, MA, USA) at CEBAS-CSIC (Murcia, Spain). All metals and other elements were extracted from plant tissues and quantified according to standard protocols.

The oxidative damage of lipids was estimated as the content of malondialdehyde, a cytotoxic aldehyde produced during lipid peroxidation. Briefly, the method involved the extraction of malondialdehyde with 5% metaphosphoric acid containing 0.04% butylhydroxytoluene, and subsequent reaction with thiobarbituric acid at low pH and 95°C to form (thiobarbituric acid)2–malondialdehyde adducts. These were extracted with 1-butanol and quantified by high-performance liquid chromatography (HPLC) with photodiode array detection (Iturbe-Ormaetxe et al., 1998). The identity of the malondialdehyde adduct was verified by scanning of the peak and by co-elution with a standard of 1,1,3,3-tetraethoxypropane (Sigma-Aldrich).

Gene expression

Total RNA was extracted with the RNAqueous isolation kit (Ambion, Austin, TX, USA) and treated with DNaseI (Roche) at 37°C for 30 min. cDNA was synthesized from DNase-treated RNA with (dT)17 and Moloney murine leukemia virus reverse transcriptase (Promega). Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) analysis was performed with an iCycler iQ instrument using iQ SYBR-Green Supermix reagents (Bio-Rad) and gene-specific primers, as indicated previously (Rubio et al., 2007). For the ALMT gene, the following primers were used: 5′-AGGTGCAACACTCAGCAAAAGC-3′ (forward) and 5′-TGACCTCCAACCCCTAAAGCA-3′ (reverse). The PCR program and other details have been described previously (Rubio et al., 2007). The amplification efficiency of primers, calculated using serial dilutions of root cDNAs, was > 75%, except for the primers of the genes encoding peroxisomal APX (APXpx), cytosolic GR (GRc), plastidic GR (GRp) and ALMT, whose efficiencies were > 65%. Expression levels were normalized using ubiquitin as the reference gene. Threshold cycle values were in the range of 17–19 cycles for ubiquitin and 22–29 cycles for the genes of interest. Three additional reference genes were used to confirm the stability of the ubiquitin transcript during Al stress. These genes encode the PP2A regulatory subunit, eukaryotic initiation factor 4A and glycosylphosphatidylinositol (GPI)-anchored protein, and have been selected, together with ubiquitin, from the most stably expressed genes in plants under a variety of stressful conditions (Czechowski et al., 2005; Sánchez et al., 2008). A comparison of their mRNA levels confirmed their stability in roots treated with 10 or 20 μM Al.

Antioxidant enzymes and metabolites

The SOD enzymes were extracted from roots with 50 mM potassium phosphate buffer (pH 7.8), 0.1 mM EDTA, 0.1% Triton X-100 and 1% polyvinylpyrrolidone-10 (PVP-10), and their activities were determined by the ferric cytochrome c method in the absence or presence of the inhibitors KCN (3 mM) and H2O2 (5 mM). These concentrations of KCN and H2O2 inhibit CuZnSOD and CuZnSOD + FeSOD, respectively. Control samples to measure total SOD activity contained 10 μM KCN to inhibit cytochrome oxidase, but not CuZnSOD. The MnSOD, FeSOD and CuZnSOD isoforms were also resolved on 15% acrylamide native gels using the nitroblue tetrazolium method by incubation or not with inhibitors (Beauchamp & Fridovich, 1971).

APX was extracted with 50 mM potassium phosphate buffer (pH 7.0), 0.5% PVP-10 and 5 mM ascorbate, and its activity was measured by following ascorbate oxidation at 290 nm for 2 min (Asada, 1984). GR was extracted with 50 mM potassium phosphate buffer (pH 7.8), 1% PVP-10, 0.2 mM EDTA and 0.1% Triton X-100, and its activity was measured by following NADPH oxidation at 340 nm for 3 min (Dalton et al., 1986). MR and DR were extracted with the same medium as for GR but omitting Triton X-100 and including 10 mM β-mercaptoethanol. MR activity was determined by following NADH oxidation at 340 mm for 90 s (Dalton et al., 1993) and DR activity by following ascorbate formation at 265 nm for 3 min (Nakano & Asada, 1981).

Ascorbate was quantified by MS as indicated below for other organic acids. Glutathione and homoglutathione were quantified by HPLC with fluorescence detection after thiol derivatization with monobromobimane, and the redox state of homoglutathione was determined by an enzymatic recycling method (Matamoros et al., 1999).

Immunoblots

Proteins were extracted from roots at 0°C with 50 mM potassium phosphate buffer (pH 7.8), 0.1% Triton X-100 and 0.1 mM EDTA. Proteins were separated on 12.5% sodium dodecylsulfate gels (Bio-Rad), transferred onto polyvinylidene fluoride membranes and challenged with optimal concentrations of polyclonal antibodies raised against cytosolic DR (DRc) of rice (Oryza sativa; Eltayeb et al., 2006), plastidic CuZnSOD (CuZnSODp) of Spinacia oleracea (Kanematsu & Asada, 1990) and cytosolic FeSOD (FeSODc) of Vigna unguiculata (Moran et al., 2003). The antibody for CuZnSODp also recognizes cytosolic CuZnSOD (CuZnSODc), but both proteins are clearly separated on immunoblots. The secondary antibody for DRc was anti-guinea pig immunoglobulin G conjugated to horseradish peroxidase (Sigma-Aldrich), and was used at a dilution of 1 : 10 000. The secondary antibody for CuZnSODp and FeSODc was anti-rabbit immunoglobulin G conjugated to horseradish peroxidase, and was used at dilutions of 1 : 2000 and 1 : 10 000, respectively. Immunoreactive proteins were visualized using the Supersignal West Pico (Pierce, Rockford, IL, USA) chemiluminescent reagent for peroxidase detection.

Organic acids

Organic acids were analyzed as described elsewhere (Rellán-Álvarez et al., 2011). Briefly, 100 mg of roots were extracted with 2 ml of 4% metaphosphoric acid, 1% PVP-10 and 0.1% formic acid. Samples were centrifuged, filtered and analyzed with a micrOTOF II electrospray ionization mass spectrometer (Bruker Daltonics, Bremen, Germany) coupled to an Alliance 2795 HPLC system (Waters, Milford, MA, USA). Samples were separated isocratically on a Supelcogel H (250 × 4.6 mm, 9 μm; Supelco, Bellefonte, PA, USA) anion exchange column. Internal standards (100 μM of 13C-labeled malic or succinic acids) were used for quantification.

Metabolite profiling

Frozen roots were ground in microvials with stainless steel metal balls using a ball mill grinder, taking care that all material had been precooled with liquid nitrogen. Metabolites were extracted from the frozen powder (60 mg) with methanol/chloroform, and the polar fraction was prepared by liquid partitioning into water and derivatized (Desbrosses et al., 2005). Gas chromatography coupled to electron impact ionization/time-of-flight (TOF) MS was performed using an Agilent (Santa Clara, CA, USA) 6890N24 gas chromatograph with split or splitless injection connected to a Pegasus III TOF mass spectrometer (LECO, St Joseph, MI, USA) as described by Sánchez et al. (2008). Details of the procedures followed for metabolite identification, normalization and quantification have been described previously (Desbrosses et al., 2005; Sánchez et al., 2008).

Proteomic profiling

Proteomic analyses were performed using a gel-free shotgun protocol based on nano-HPLC and MS/MS, as described elsewhere (Larrainzar et al., 2007). In brief, proteins were extracted from roots by acetone precipitation and subjected to in-solution digestion with endoproteinase Lys-C and immobilized trypsin beads. The resulting peptides were desalted, dried and dissolved in formic acid. Protein digests were separated with an Ultra HPLC Eksigent system (Axel Semrau, Sprockhövel, Germany) using a monolithic reversed phase column (Chromolith 150 × 0.1 mm; Merck, Darmstadt, Germany) directly coupled to an Orbitrap XL mass spectrometer (Thermo Scientific, Rockford, IL, USA). Peptides were eluted with a 100-min gradient from 5% to 60% acetonitrile. Dynamic exclusion settings were as described in Hoehenwarter & Wienkoop (2010). After MS analysis, raw files were searched against the DFCI Lotus Gene Index (6.0) using the Sequest algorithm. For identification and spectral count-based data matrix generation, the Proteome Discoverer (v 1.1, Thermo Scientific) was used. A decoy database enabled false positive rate analysis. Only high confidence peptides (false positive rate < 0.1%) with better than 5 ppm precursor mass accuracy and at least two distinct peptides per protein passed criteria.

Results

Plant growth and nutrition

The inhibition of root growth is a typical symptom of Al toxicity (Kochian, 2005) and was used here as a marker to set up treatment conditions of L. corniculatus plants grown in hydroponic cultures. We used simple salt solutions to minimize problems with Al speciation and precipitation (Pellet et al., 1995), and selected two low Al concentrations (10 and 20 μM, equivalent to 6.5 and 13 μM of free Al3+ activity, respectively) and a period of treatment (14 d) long enough to allow for physiologically relevant changes in growth parameters and in the metabolome and proteome of roots. Plants grown in simple salt solution did not show symptoms of nutrient deficiency and were also comparable in size (Fig. S1c). This was confirmed by the similar contents of N in the roots (22 mg g−1 dry weight (DW)) and leaves (28 mg g−1 DW) of plants grown in CaCl2 at pH 4.0 with respect to those found in plants grown in 1 : 4 strength B&D solution at pH 4.0 (Broughton & Dilworth, 1971). By contrast, plant treatment with 10 or 20 μM Al increased the N content of roots by c. 20% (Supporting Information Table S1) and decreased that of leaves by c. 44% (data not shown), which is probably reflective of a differential effect of Al on N assimilation in the two plant organs and/or changes in N allocation between root and shoot. Treatment with 20 μM Al caused significant decreases in K, S, Zn and Ni in the roots (Table S1), but no changes in Ca, Mg, P, Fe, Cu and Mn (data not shown).

Plants supplied with 10 μM Al showed a reduction of 11% in the root length and 39% in the root FW (Fig. 1a). The corresponding decreases with 20 μM Al were 52% and 78%. The shoot growth was also affected by application of 10 and 20 μM Al, with decreases of 45% and 73% in FW and of 36% and 64% in leaf area, respectively (Fig. 1a). These plants accumulated Al in the roots and, albeit at 10-fold lower levels, in the leaves (Fig. 1b).

Figure 1.

Effect of aluminum (Al) concentration on growth parameters (a) and the Al contents of roots and leaves (b) of Lotus corniculatus. Values are means ± SE of 40–100 replicates, and those denoted by the same letter are not significantly different at < 0.05 according to Duncan’s multiple range test. The water contents of roots and leaves of control and Al-treated plants were 91 ± 1% and 82 ± 1%, respectively.

ROS, antioxidant defenses and oxidative damage

Specific fluorescent probes were used to localize Al accumulation and to detect ROS production in roots (Fig. 2). The localization of Al was demonstrated using morin, which strongly binds Al forming a complex that emits green fluorescence. Roots treated with 10 μM Al accumulated this metal along the root, but especially at the tips, which include the cell division and elongation zones. A similar distribution was observed for roots treated with 20 μM Al, although, in this case, fluorescence was more intense. Superoxide radical production was visualized using a method based on the superoxide-mediated oxidation of DHE to oxyethidium, which emits red fluorescence. A low background signal was observed in roots treated with 0 or 10 μM Al, whereas intense red fluorescence was found in roots treated with 20 μM Al, especially at the tips. The production of H2O2 and other hydroperoxides was visualized after intracellular oxidation of DCF-DA to a derivative that emits green fluorescence. As was the case for superoxide formation, a strong fluorescence signal was clearly seen in the tips of roots treated with 20 μM Al.

Figure 2.

Localization of aluminum (Al) accumulation using morin and the detection of superoxide radical and peroxide production employing the fluorescent probes dihydroethidium (DHE) and 2′,7′-dichlorofluorescein diacetate (DCF-DA), respectively, in roots of Lotus corniculatus exposed to 0 (control), 10 or 20 μM Al. The top images correspond to roots viewed with fluorescence excitation, and the bottom images to the same roots examined with white light to mark the position of the roots. Representative images of at least four independent experiments are shown and the size bar is identical for all panels. Note the deformation of the root tip in plants treated with 20 μM Al.

Because plant treatment with the higher Al concentration elicited ROS production and may potentially give rise to oxidative stress, we examined the effects of Al on the expression of key antioxidant enzymes in the roots (Fig. 3). The addition of 10 or 20 μM Al to the rooting medium increased the mRNA level of FeSODc and decreased that of plastidic FeSOD (FeSODp). Moreover, 20 μM Al downregulated the expression of CuZnSODc, GPX1, GPX4, DRc and plastidic DR (DRp). Neither of the two Al concentrations altered significantly the mRNA levels of MnSOD and other GPXs or those of the APX, MR and GR isoforms (Fig. 3). Likewise, the activities of these three enzymes of the ascorbate–glutathione pathway remained unaffected by Al stress (data not shown).

Figure 3.

Expression of antioxidant genes (steady-state mRNA levels) in roots of Lotus corniculatus exposed to 10 (gray bars) or 20 (black bars) μM Al. Data of Al-treated plants are expressed relative to those of control plants, which were given a value of unity, and represent the means ± SE of six biological replicates (RNA extractions) from at least two series of plants grown independently. Asterisks denote significant upregulation (> 2) or downregulation (< 0.5) of the genes. APX, ascorbate peroxidase; c, cytosolic isoform; DR, dehydroascorbate reductase; GPX, glutathione peroxidase; GR, glutathione reductase; MR, monodehydroascorbate reductase; p, plastidic isoform; px, peroxisomal isoform; s, stromal isoform; SOD, superoxide dismutase; t, thylakoidal isoform.

We further investigated whether the changes in the mRNA levels of the cytosolic enzymes, namely CuZnSODc, FeSODc and DRc, were reflected in the protein contents and enzyme activities of root extracts using immunoblots and activity assays (Fig. 4). The downregulation of CuZnSODc and the upregulation of FeSODc were accompanied by similar trends in the proteins and catalytic activities. Interestingly, the total SOD and MnSOD activities of the roots remained constant with Al (data not shown), implying a compensation between the CuZnSODc and FeSODc activities. Likewise, the downregulation of the DRc gene with Al was paralleled by a marked decrease in protein and activity. Although the DR activity assay could not distinguish between the cytosolic and plastidic isoforms, we found, using a specific antibody, that the DRp protein was virtually undetectable in root extracts, and hence the majority of DR activity can be attributed to DRc.

Figure 4.

Specific activities and relative protein abundance of the CuZnSODc, FeSODc and DRc isoforms in roots of Lotus corniculatus exposed to 0 (control), 10 or 20 μM Al. Enzyme activities are means ± SE of six replicates, each corresponding to a different root of two series of plants grown independently. Means denoted by the same letter are not significantly different at < 0.05 according to Duncan’s multiple range test. Immunoblots are representative of four independent experiments and the apparent molecular masses (kDa) of the proteins are indicated on the right. Lanes were loaded with 20 μg (CuZnSOD) or 30 μg (FeSODc and DRc) of protein. c, cytosolic isoform; DR, dehydroascorbate reductase; p, plastidic isoform; SOD, superoxide dismutase.

The effects of Al on the major antioxidant metabolites and on lipid peroxidation were also investigated. However, our first attempts to quantify ascorbate using the ascorbate oxidase method failed, probably because traces of Al in the root extracts interfered with the activity assay. Thus, we had to resort to a highly sensitive HPLC-MS method, which allowed us to quantify ascorbate, but not dehydroascorbate. This oxidized form of ascorbate is broken down during the electrospray process, even at the low voltages used here for organic acid analysis (Rellán-Álvarez et al., 2011). The ascorbate content of roots declined by 25% and 55% with 10 and 20 μM Al, respectively (Fig. 5a). The thiol tripeptides glutathione (γGlu-Cys-Gly) and homoglutathione (γGlu-Cys-βAla) were also quantified in roots. Homoglutathione is only found in certain legume species and tissues, whereas glutathione is ubiquitous in plants and other organisms (Matamoros et al., 1999). The roots and leaves of L. corniculatus contained 3% glutathione and 97% homoglutathione. The content of total homoglutathione (reduced + oxidized forms) in roots increased by c. 35% with 10 or 20 μM Al (Fig. 5a). However, the redox state of homoglutathione (percentage of the reduced form) remained in the range 88–90% for both Al treatments. The oxidative damage of lipids was used as a marker of oxidative stress and assessed by measuring malondialdehyde, a decomposition product of lipid peroxides. The content of malondialdehyde in roots did not change with 10 μM Al, but increased significantly with 20 μM Al (Fig. 5b).

Figure 5.

Contents of antioxidant metabolites (a) and malondialdehyde (b) in roots of Lotus corniculatus exposed to 0 (control), 10 or 20 μM Al. Values are means ± SE of 6–12 replicates, each corresponding to a different root of at least two series of plants grown independently. Means denoted by the same letter are not significantly different at < 0.05 according to Duncan’s multiple range test.

Organic acids and metabolomics

The organic acids most commonly found in plant cells were quantified in roots by HPLC-MS, as some of these compounds constitute a defense mechanism against Al toxicity and their concentrations may be responsive to Al (Pellet et al., 1995; Ma et al., 2001). Moreover, a metabolomic approach was used to study the possible effects of Al on other metabolic pathways in the roots. Both types of analysis were also carried out in the leaves to determine whether the low amounts of Al detected in the shoot interfered with leaf metabolism. Plant treatment with 10 and/or 20 μM Al caused an increase in malate, succinate and fumarate, a decrease in citramalate and no changes in citrate in the roots (Fig. 6). However, the concentrations of these carboxylic acids remained unaffected in the leaves (data not shown).

Figure 6.

Contents of several carboxylic acids in roots of Lotus corniculatus exposed to 0 (control; white bars), 10 (gray bars) or 20 (black bars) μM Al. Values are means ± SE of nine replicates, each corresponding to a different root of three series of plants grown independently. Means denoted by the same letter are not significantly different at < 0.05 according to Duncan’s multiple range test.

Metabolite profiling of roots and leaves of Al-treated plants revealed changes in important amino acids and sugars, as well as in certain organic acids that had not been analyzed by HPLC-MS (Table 1). In roots and leaves, there was an important increase in the asparagine content. This amino acid is a major product of ammonium assimilation in L. corniculatus roots. In addition, Al treatment caused a decline in the root content of glycine and increases in the leaf contents of serine, aspartic acid and glutamic acid, indicating that Al also affected amino acid metabolism and/or protein turnover in the shoot. Likewise, Al stress increased the concentrations of five carboxylic acids in the roots. These included two malic acid derivatives and threonic acid, a product of ascorbic acid metabolism. The largest increases, in the range 80–100%, were found for threonic, 2-isopropylmalic and glyceric acids. The concentrations of threonic and 2-isopropylmalic acids were also augmented by c. 70% in the leaves of plants treated with 10 or 20 μM Al. These plants also showed alterations in the sugar concentrations of roots and leaves. Thus, in roots treated with 10 μM Al, glucose and fructose increased concomitant with a modest decline in sucrose, whereas, in the leaves, glucose, fructose and sucrose remained constant with 10 μM Al, but increased by 22–54% with 20 μM Al (Table 1).

Table 1.   Effects of aluminum (Al) stress on the metabolite contents of roots and leaves of Lotus corniculatus
RootsAl concentration (μM)
01020
  1. Values represent normalized responses of metabolite pool measurements (detector signals in arbitrary units normalized to internal standard and sample fresh weight). Data are means ± SE of 12 biological replicates from two series of plants grown independently. Means denoted by the same letter are not significantly different at < 0.05 according to Duncan’s multiple range test.

Amino acids
 Asparagine8.7 ± 0.7 a10.9 ± 1.3 ab13.1 ± 0.8 b
 Glycine1.0 ± 0.1 a0.5 ± 0.0 b0.7 ± 0.1 a
Organic acids
 Threonic acid3.8 ± 0.4 a6.8 ± 0.4 b6.8 ± 0.8 b
 2-Isopropylmalic acid5.1 ± 1.0 a5.5 ± 0.8 a10.1 ± 0.6 b
 2-Methylmalic acid8.2 ± 0.8 ab6.4 ± 0.8 a9.7 ± 0.7 b
 Pyroglutamic acid9.7 ± 0.7 ab8.1 ± 0.3 a10.0 ± 0.3 b
 Glyceric acid0.6 ± 0.1 a0.8 ± 0.0 a1.1 ± 0.1 b
Sugars
 Glucose9.9 ± 0.9 a14.4 ± 1.3 b9.3 ± 0.9 a
 Fructose10.9 ± 0.7 a14.4 ± 1.1 b9.4 ± 1.0 a
 Sucrose18.7 ± 0.9 a14.6 ± 0.4 b17.4 ± 0.5 a
 Sedoheptulose10.6 ± 0.9 a11.6 ± 1.5 a7.6 ± 1.1 b
Polyols
 Pinitol9.9 ± 0.6 ab8.8 ± 0.2 a10.8 ± 0.3 b
Leaves01020
Amino acids
 Serine12.1 ± 0.9 a18.2 ± 1.6 b14.3 ± 0.9 ab
 Asparagine0.5 ± 0.1 a3.2 ± 1.0 b1.0 ± 0.1 b
 Aspartic acid1.5 ± 0.3 a4.2 ± 0.9 b3.2 ± 0.5 b
 Glutamic acid8.3 ± 0.5 a12.9 ± 1.0 b13.6 ± 0.6 b
Organic acids
 Succinic acid8.6 ± 0.7 a10.7 ± 0.7 a13.9 ± 0.6 b
 Threonic acid7.9 ± 0.5 a11.6 ± 1.1 b14.0 ± 0.5 b
 Threonic acid-1,4-lactone11.6 ± 0.9 a13.8 ± 1.0 ab14.7 ± 0.8 b
 Galactonic acid13.0 ± 0.2 a14.1 ± 0.6 a16.3 ± 0.6 b
 2-Isopropylmalic acid8.2 ± 1.4 a14.2 ± 1.9 b13.2 ± 1.6 ab
Sugars
 Glucose7.9 ± 0.6 a8.4 ± 0.8 a12.2 ± 1.1 b
 Fructose7.7 ± 0.7 a7.8 ± 0.7 a11.1 ± 1.1 b
 Sucrose13.1 ± 0.4 a13.9 ± 0.4 a16.0 ± 0.5 b
Polyols
 Pinitol24.0 ± 1.7 a19.2 ± 0.9 b20.1 ± 0.7 ab

Proteomics

A highly sensitive analysis of the root proteome, entailing nano-HPLC shotgun MS, allowed us to identify proteins that were newly induced or upregulated, as well as those that were suppressed or downregulated, in response to Al stress (Table 2). Proteins were identified based on the sequences available in the L. japonicus databases and were classified into functional groups. For relative quantification, the spectral count number was used as described by Larrainzar et al. (2007). An independent component analysis of the data revealed a progressive separation of the Al-treated samples relative to control samples with increasing Al concentration (Fig. S2). Particularly critical for this separation were the loadings of the first independent component, which accounted for 50% of the total variance.

Table 2.   Effects of aluminum (Al) stress on the Lotus corniculatus root proteome
ProteinTC1UniProt2Al concentration (μM)
01020
  1. Values ((log of the number of spectral counts) × 1000) are means of six biological replicates from two series of plants grown independently. Means denoted by the same letter are not significantly different at < 0.05 according to Duncan’s multiple range test.

  2. 1Tentative consensus (TC) sequence numbers according to the DFCI Lotus Gene Index (6.0).

  3. 2UniProt accessions (UniRef100).

Cell wall/cell organization
 α-TubulinTC62930A9PL191129 a877 a181 b
 α-TubulinTC63835Q2TFP2897 a718 a140 b
 β-TubulinTC61113UPI00015CD56A699 a0 b0 b
 β-TubulinTC63392P295141067 a556 b0 c
 β-TubulinTC62547P373921081 a570 b0 c
 β-TubulinTC57323Q406651073 a556 b0 c
Gene structure and regulation
 Histone H2ABW599450A7P108347 a50 ab0 b
 Histone H2ATC61686A2WQG7656 a116 b0 b
 Histone H4TC70944UPI000050340F556 a351 a0 b
Protein synthesis
 60S ribosomal proteinBW604002Q8H2B9381 a80 b0 b
 60S ribosomal protein L9FS326259P30707754 a317 b0 c
 Elongation factor 1-αTC69520Q3LUM51043 a786 a140 b
 Elongation factor 1-αTC73117Q3LUM21467 a1204 ab701 b
 Elongation factor 1-βFS339508P29545892 a734 a191 b
 Elongation factor 1-γTC60762Q8S3W1708 a413 ab120 b
 Elongation factor EF-2 (putative)TC75757Q9ASR1/Q9SGT41394 a1085 a311 b
Protein degradation
 Cysteine proteinase inhibitorBI418502Q0644580 a463 ab156 b
 Proteasome subunit α typeTC57402A7P6B10 a116 b426 b
 Peptidase C1ATC68381Q2HTQ3296 a660 a1146 b
 PolyubiquitinTC81524A1X1E50 a50 a410 b
 PolyubiquitinTC81113Q0J9W60 a50 a457 b
Transport
 Adenine nucleotide translocator (mitochondrial)TC74603O49875392 a426 a0 b
 ATP synthase subunit γ (mitochondrial)TC57922D7SI12310 a50 ab0 b
 ATP synthase catalytic subunit A (vacuolar)TC75345Q9SM09959 a877 a295 b
Amino acid metabolism
 Methionine synthaseTC70396Q71EW81411 a1157 a402 b
 Methionine synthaseTC65903UPI00015CD060793 a698 a0 b
 S-Adenosylmethionine synthetaseTC69893A4PU481358 a1262 ab874 b
 S-Adenosylmethionine synthetaseTC67258A4ULF81311 a1257 a816 b
 Adenosylhomocysteinase 1TC72761O23255925 a910 a169 b
 Glutamine synthetase (cytosolic isoform)TC72874Q428991490 a1287 a753 b
Organic acid metabolism
 Malate dehydrogenaseTC62158Q9SPB81250 a963 a208 b
 Malate dehydrogenaseTC66662Q6RIB61181 a985 ab499 b
 Malate dehydrogenaseTC59388O81278580 a217 b0 b
 Isocitrate dehydrogenaseTC67164Q06197910 a794 ab426 b
 Carbonic anhydraseTC57320Q5NE211069 a587 b95 c
Carbohydrate metabolism
 UTP-Glucose-1-P uridylyltransferaseTC59881Q9LKG7506 a280 ab0 b
 Sucrose synthaseTC77381P13708965 a658 a169 b
 Sucrose synthaseTC72460Q9AVR8823 a453 ab120 b
 Sucrose synthase (nodule enhanced)TC78224O816101111 a879 ab435 b
 Fructokinase-2 (putative)TC74169Q9LNE31383 a1084 ab605 b
 UDP-Glucose:protein transglucosylase-likeTC76160Q38M711101 a835 ab429 b
Glycolysis
 Pyruvate kinaseTC58669Q5F2M7429 a0 b0 b
 Phosphoglycerate kinaseTC78075A5CAF8734 a522 ab156 b
 Phosphoglycerate kinaseTC57762Q9LKJ21170 a962 ab467 b
 Phosphoglycerate dehydrogenase (putative)TC65829UPI00015C90B8246 ab547 a0 b
 EnolaseTC58226Q6RIB71093 a870 a309 b
Electron transfer/redox/antioxidant
 Ferredoxin-NADP reductaseTC60743Q41014400 a0 b0 b
 CatalaseTC58073A0PG70597 a310 ab140 b
 Pox09TC57306Q9XFL30 a180 b311 b
 Pox13 (precursor)TC60841Q9ZNZ6749 a310 b0 b
 Pox30TC61834A4UN760 a852 b1054 b
 GST15 (tau class)TC57307Q9FQE3151 a239 ab536 b
 GSTin2-1 (lambda class)TC57627Q9FQ9550 a251 a854 b
 Protein disulfide-isomerase A6 precursor (putative)TC72404P38661259 a80 ab0 b
 LipoxygenaseTC57788O24470180 ab458 a0 b
Stress
 Heat shock protein 70GO008419Q409801179 a906 ab511 b
 Heat shock cognate protein 70TC58352Q401511135 a1087 a703 b
 Heat shock cognate protein 70TC77297Q5QHT31189 a940 ab587 b
 Heat shock cognate protein 70TC68669Q410271302 a1140 ab760 b
 Heat shock protein 90TC60546A8WEL7909 a722 a169 b
 BiP-isoform DTC73211Q9ATB81110 a893 ab501 b
 PR protein class 10TC57863Q94IM3680 a1221 b1188 b
Secondary metabolism
 Caffeoyl-CoA O-methyltransferaseTC58984Q40313467 a0 b0 b

The treatment of plants with 10 and/or 20 μM Al led to major decreases in the root contents of proteins implicated in multiple crucial processes, such as cell elongation and division, protein synthesis and degradation, amino acid and organic acid metabolism, glycolysis and carbohydrate metabolism, transport, redox control and stress response (Table 2). Some of these proteins were already undetectable in roots exposed to only 10 μM Al, whereas others were suppressed after application of 20 μM Al. The first group included a β-tubulin chain, pyruvate kinase, ferredoxin-NADP reductase and caffeoyl-CoA O-methyltransferase; the second group included some β-tubulin and ribosomal polypeptides, histones, UTP-glucose-1-P uridyltransferase, phosphoglycerate dehydrogenase, protein disulfide isomerase, a peroxidase precursor and a lipoxygenase isoform. In sharp contrast, a few proteins were newly induced with 10 μM Al and their levels were further enhanced with 20 μM Al. This was the case for the proteasome α-subunit and two peroxidase isoforms. Finally, the contents of other proteins that were constitutively expressed in roots increased in response to Al. Notable examples of this were the cysteine proteinase inhibitor and peptidase C1A, two glutathione transferases (formerly glutathione S-transferases; GSTs) and a pathogenesis-related class 10 (PR-10) protein (Table 2).

The identification of peroxidases and GSTs responsive to Al is presented separately in further detail in Table S2, given the bewildering complexity of these two groups of enzymes that perform multiple roles in plants in addition to those related to their antioxidative properties. There are between 70 and 100 class III or secretory peroxidases (deposited in the PeroxiBase; see Table S2; Cosio & Dunand, 2009) and between 25 and 54 GSTs (McGonigle et al., 2000; Dalton et al., 2009; Dixon et al., 2010) in legumes and other plants. However, only three peroxidases (Pox09, Pox13 and Pox30) and two GSTs (GST15 and GSTin2-1) were affected at the protein level in L. corniculatus roots exposed to Al stress (Table 2).

Discussion

In this work, L. corniculatus plants were exposed to low Al concentrations for a prolonged time to mimic the acidic soil conditions prevailing in some regions of South America, where this forage legume is amply cultivated. In Uruguay, 1 080 000 hectares are sown in mixed legume–grass pastures and 117 000 hectares in pure pastures (DIEA, 2010). In preliminary experiments, two Al concentrations were carefully selected in an attempt to discriminate between the toxic effects of Al and oxidative stress. The long-term application of 10 μM Al to L. corniculatus plants was sufficient to inhibit markedly root and shoot growth. At this stage, there was accumulation of Al, but not of ROS, in the root tip. Moreover, the mRNA levels and activities of antioxidant enzymes, with few exceptions, and the malondialdehyde content were not affected. By contrast, increasing the Al concentration from 10 to 20 μM induced oxidative stress in the roots. The accumulation of malondialdehyde with 20 μM Al can be explained by an exacerbated production of superoxide and H2O2, which may give rise, in the presence of catalytic metal ions, to hydroxyl radicals and other highly oxidizing species necessary to initiate membrane fatty acid peroxidation (Halliwell & Gutteridge, 2007). Other authors have found, using different experimental conditions, an increase in lipid peroxidation in plants treated with Al (Cakmak & Horst, 1991; Yamamoto et al., 2001; Guo et al., 2004; Sharma & Dubey, 2007).

The decrease in ascorbate, which is required for α-tocopherol regeneration, may also contribute to the cumulative peroxidative damage in L. corniculatus roots. Notably, DRc activity, which reduces dehydroascorbate to ascorbate, was transcriptionally downregulated. Dehydroascorbate is quite unstable and, unless rapidly used up by DRc to regenerate ascorbate, is degraded to oxalate and threonate (Green & Fry, 2005). The downregulation of DRc may thus explain the decrease in ascorbate concurrent with the accumulation of threonate in Al-treated roots. Another novel finding related to antioxidant protection was the progressive replacement of CuZnSODc by FeSODc with Al stress. This may be explained by a microRNA-mediated cleavage of the CuZnSODc mRNA. Thus, in A. thaliana plants under low Cu conditions, the miR398 family is involved in the downregulation of CuZnSODc and CuZnSODp, which are replaced by FeSOD (Yamasaki et al., 2007). In L. corniculatus roots, the total contents of Cu or Zn (mainly as constituents of metalloproteins) remained unchanged or decreased with Al, respectively. We cannot rule out the possibility that a lower availability of free Cu2+ and/or Zn2+ ions downregulates the synthesis of functional CuZnSOD in Al-treated plants. Interestingly, the so-called ‘cytosolic’ CuZnSOD and FeSOD isoforms are also present, and at relatively large amounts, in the nuclei (Rubio et al., 2009). We found no apparent functional reason for the change in the prevalent SOD isoform in the cytosol and nuclei of root cells stressed by Al because both types of enzyme are potentially inactivated by H2O2. In any case, this ‘switch’ of SOD isoform seems to be associated with advancing senescence, at least in legume nodules (Moran et al., 2003; Rubio et al., 2007), indicating a compensatory phenomenon between the two enzyme activities.

The combined use of organic acid analysis, metabolomics and proteomics allowed us to unravel some cellular functions and metabolic pathways responsive to Al stress in L. corniculatus. One such pathway is dicarboxylic acid metabolism. Roots exposed to Al have higher concentrations of malic, succinic and fumaric acids. This alteration may be related to the decrease in cytosolic malate dehydrogenase and isocitrate dehydrogenase, observed in our proteomic study, rather than to a specific effect on the citric acid cycle in mitochondria. A detrimental effect of Al on the cytosol of root cells is also substantiated by the strong downregulation of key enzymes involved in sucrose metabolism and glycolysis, as well as by the changes in DRc, CuZnSODc and FeSODc proteins and activities mentioned above. Metabolite profiling led us to identify lesser known organic acids that are also affected by Al stress. Thus, the content of 2-isopropylmalic acid, an intermediate in leucine biosynthesis, increased in roots and leaves. This compound is secreted by budding yeast (Saccharomyces cerevisiae) cells challenged with Al (Kobayashi et al., 2005), and may be involved in its detoxification as it is a powerful chelator of Al3+ (Tashiro et al., 2006). Although the identification of organic acids secreted by L. corniculatus roots is beyond the scope of this study, our results are consistent with a role of malate and 2-isopropylmalate, rather than citrate, in Al detoxification. Thus, in addition to the changes in the root contents of both dicarboxylates and their associated enzymes, we found increases of 5.5-fold with 10 μM Al and nine-fold with 20 μM Al in the ALMT mRNA levels (Fig. S3).

Plant treatment with Al had major effects on cytoskeleton dynamics and protein turnover in the roots. Exposure to 10 and/or 20 μM Al drastically reduced the amounts of α- and β-tubulin and of some ribosomal proteins and elongation factors. These changes are consistent with an inhibitory effect of Al on cell division and protein synthesis. In particular, the root tips were seriously deformed with 20 μM Al as a result of the inhibition of root cell elongation and division. This Al concentration stimulated protein degradation, judging from the increase in the root content of proteases and of the 20S proteasome α-subunit. An induction of the latter protein has been observed in Al-treated tomato (Solanum lycopersicum) roots (Zhou et al., 2009). The application of 20 μM Al to plants also had a strong impact on methionine metabolism. This amino acid is essential not only as a constituent of proteins, but also as a direct precursor of S-adenosylmethionine, which is a major methyl group donor and an intermediate in the biosynthesis of ethylene, polyamines, biotin and nicotianamine (Moffatt & Weretilnyk, 2001; Ravanel et al., 2004). The three enzymes intervening in the activated methyl cycle (methionine synthase, S-adenosylmethionine synthetase and S-adenosylhomocysteine hydrolase) were strongly downregulated with Al stress. This downregulation may result in a restriction of transmethylation reactions and/or alterations in the biosynthesis of hormones, such as ethylene, in the root cells. Recent work has shown that S-adenosylmethionine synthetase and S-adenosylhomocysteine hydrolase are moderately induced by Al in tomato roots (Zhou et al., 2009) and that two S-adenosylmethionine synthetase isoforms are differentially regulated in rice roots (Yang et al., 2007). Overall, these results show that the methyl cycle is a preferential target of Al toxicity.

As anticipated, plant treatment with Al elicited changes in redox and stress proteins. Class III peroxidases and GSTs are multifunctional enzymes encoded by large gene families. However, the response to Al stress was rather specific, as only two isoforms of each family were induced in L. corniculatus roots. To our knowledge, no changes in the content of peroxidase isoforms in Al-treated roots have been reported to date, although the expression of several peroxidase genes was found to be affected at the transcriptional level in A. thaliana (Richards et al., 1998; Kumari et al., 2008). The inducibility of the two GST isoforms strongly suggests that they are efficient at using homoglutathione as substrate, because we found that this glutathione homolog accounts for 97% of the total thiol tripeptides in L. corniculatus roots. A transcriptomic analysis of A. thaliana showed time-dependent changes in the mRNA levels of various GST genes in response to Al (Kumari et al., 2008), whereas proteomic analyses showed that two different GST isoforms were downregulated in soybean (Glycine max; Zhen et al., 2007) and tomato (Zhou et al., 2009). Molecular chaperones play important roles in preventing aggregation and assisting the refolding of non-native proteins, as well as in facilitating proteolytic degradation of unstable proteins (Wang et al., 2004). Interestingly, some heat shock proteins/molecular chaperones of the Hsp70 and Hsp90 families and a protein disulfide isomerase, which may also function as a chaperone, were found to be downregulated. This probably reflects the inability of L. corniculatus to withstand 20 μM Al, a conclusion that is supported by the suppression or consistent downregulation of other proteins, not previously reported in proteomic studies, that are involved in gene regulation, transport, electron transfer and hormone synthesis.

In conclusion, under our experimental conditions, 10 μM Al was sufficient to inhibit root and shoot growth and to affect the contents of some metabolites and proteins of root cells, but did not trigger ROS accumulation or oxidative stress. Therefore, oxidative damage was not the cause of Al toxicity. Increasing the Al concentration to 20 μM elicited ROS accumulation and oxidative stress, inhibited protein synthesis, enhanced proteolysis and intensified the effects on the proteins involved in cytoskeleton organization, organic acid and carbohydrate metabolism, redox regulation and stress responses. These detrimental effects indicate a metabolic dysfunction, which affects the cytosol, mitochondria and other cellular compartments, particularly in plants exposed to 20 μM Al. Finally, a practical consequence derived from this work is that attempts to develop tolerance to oxidative stress will not, by themselves, alleviate the problems of Al toxicity.

Acknowledgements

We are grateful to Professor Joachim Kopka and Professor Javier Abadía for allowing us to use the GC-MS and HPLC-ESI-MS facilities at Golm (Germany) and Zaragoza (Spain), respectively. We also thank Dr Amin Elsadig Eltayeb and Professor Kiyoshi Tanaka for a gift of DRc antibody, Professor Sumio Kanematsu for a gift of CuZnSODp antibody, Dr Manuel A. Matamoros for help with qRT-PCR analyses and Dr Ana Castillo for taking the photographs of Fig. S1. Thanks are also due to three anonymous reviewers and to Professor Marinus Pilon for constructive criticism to improve the manuscript. Joaquín Navascués was the recipient of a postdoctoral contract (JAE-CSIC program). This work was funded by the European Commission (FP6-2003-INCO-DEV2-517617), the Ministry of Science and Innovation-Fondo Europeo de Desarrollo Regional (AGL2008-01298 and AGL2010-16515) and Gobierno de Aragón-Fondo Social Europeo (group A53).

Ancillary