Mitochondria are an early target of oxidative modifications in senescing legume nodules


  • Manuel A. Matamoros,

    Corresponding author
    • Departamento de Nutrición Vegetal, Estación Experimental de Aula Dei, Consejo Superior de Investigaciones Científicas (CSIC), Zaragoza, Spain
    Search for more papers by this author
  • Nieves Fernández-García,

    1. Departamento de Biología del Estrés y Patología Vegetal, Centro de Edafología y Biología Aplicada del Segura, CSIC, Espinardo-Murcia, Spain
    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
  • Jorge Loscos,

    1. Departamento de Nutrición Vegetal, Estación Experimental de Aula Dei, Consejo Superior de Investigaciones Científicas (CSIC), Zaragoza, Spain
    Search for more papers by this author
  • Ana Saiz,

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

Author for correspondence:

Manuel A. Matamoros

Tel: +34 976 716064



  • Legume nodule senescence is a poorly understood process involving a decrease in N2 fixation and an increase in proteolytic activity. Some physiological changes during nodule aging have been reported, but scarce information is available at the subcellular level.
  • Biochemical, immunological and proteomic approaches were used to provide insight into the effects of aging on the mitochondria and cytosol of nodule host cells.
  • In the mitochondria, the oxidative modification of lipids and proteins was associated with a marked decline in glutathione, a reduced capacity to regenerate ascorbate, and upregulation of alternative oxidase and manganese superoxide dismutase. In the cytosol, there were consistent reductions in the protein concentrations of carbon metabolism enzymes, inhibition of protein synthesis and increase in serine proteinase activity, disorganization of cytoskeleton, and a sharp reduction of cytosolic proteins, but no detectable accumulation of oxidized molecules.
  • We conclude that nodule mitochondria are an early target of oxidative modifications and a likely source of redox signals. Alternative oxidase and manganese superoxide dismutase may play important roles in controlling ROS concentrations and the redox state of mitochondria. The finding that specific methionine residues of a cytosolic glutamine synthetase isoform are sulfoxidized suggests a regulatory role of this enzyme in senescing nodules.


The symbiotic interaction between rhizobia and legume roots leads to the formation of N2-fixing nodules, allowing legumes to grow under nitrogen-limiting conditions (Oldroyd & Downie, 2008). Nodules produce high amounts of reactive oxygen species (ROS) such as superoxide radicals (inline image) and hydrogen peroxide (H2O2) due to the elevated rates of respiration in bacteroids and mitochondria. These organelles are endowed with an important set of antioxidants that finely regulate ROS concentration (Iturbe-Ormaetxe et al., 2001). Antioxidants prevent the oxidation of mitochondrial components and permit the participation of ROS in intracellular redox signaling (Maxwell et al., 2002; Rhoads et al., 2006; Noctor et al., 2007).

In a previous work, the antioxidant systems of bean (Phaseolus vulgaris) nodule mitochondria were examined in detail (Iturbe-Ormaetxe et al., 2001). It was proposed that inline image generated by the electron transport chain is dismutated to H2O2 and O2 by a manganese superoxide dismutase (MnSOD) present in the matrix (Iturbe-Ormaetxe et al., 2001; Rubio et al., 2004). H2O2 is scavenged by either a membrane bound or by a soluble ascorbate peroxidase (APX) using ascorbate supplied by l-galactono-1,4-lactone dehydrogenase (GalLDH), which is localized in the inner membrane (Matamoros et al., 2006). Oxidized ascorbate can be reduced in the matrix by NADH-dependent monodehydroascorbate reductase (MR) or by glutathione-dependent dehydroascorbate reductase (DR). In bean and some other legumes, homoglutathione may replace glutathione as electron donor in the ascorbate–glutathione pathway and probably in other thiol-related functions. Finally, reduced homoglutathione can be regenerated in the matrix by (homo)glutathione reductase (GR) using NADPH as an electron donor.

Other antioxidant enzymes have been proposed as important players in the redox signaling network of plant cells. Peroxiredoxins (Prxs) catalyze the reduction of H2O2 and alkylhydroperoxides, and in so doing they probably modulate signaling cascades mediated by ROS and reactive nitrogen species such as nitric oxide (NO). In plants, there are four classes of Prxs that differ in their catalytic mechanisms and subcellular location (Dietz et al., 2006). To carry out their functions as antioxidants and redox sensors, the balance between oxidized and reduced Prxs is tightly regulated. Thioredoxins (Trxs) constitute a family of proteins that perform multiple functions related to cellular redox homeostasis and may act as reductants of some Prx isoforms (Dietz et al., 2006). In turn, Trxs can be reduced by NADPH-dependent Trx reductases (NTRs) in the cytosol and mitochondria (Laloi et al., 2001). Recently, we have reported the presence of a NTR-Trx-Prx system in the nodules of the model legume Lotus japonicus (Tovar-Méndez et al., 2011). In this species cytosolic PrxIIB can be regenerated by Trxh1, Trxh4 and NTRs, whereas in mitochondria oxidized PrxIIF is reduced by Trxo and NTRs.

Senescence is a highly complex and regulated developmental process involving the degradation of organelles and macromolecules (Lim et al., 2007). In legume nodules, stress-induced senescence shares several characteristics with natural or developmental senescence (aging), including the decline of N2 fixation, leghemoglobin (Lb) and some antioxidant enzymes and metabolites, as well as the oxidative damage of cell components (Evans et al., 1999; Matamoros et al., 1999a; Hernández-Jiménez et al., 2002; Loscos et al., 2008). By contrast, recent studies with Medicago truncatula nodules suggest partially divergent molecular mechanisms for developmental and stress-induced senescence (Pérez-Guerra et al., 2010).

Ultrastructural studies have provided insight into the events occurring at the subcellular level during nodule senescence (Matamoros et al., 1999a; Hernández-Jiménez et al., 2002; Rubio et al., 2004; Puppo et al., 2005). However, the role of cell organelles in nodule senescence is unclear. In animals, oxidative stress in mitochondria is considered a major hallmark of cellular aging. Consistent with this, transgenic mice that overexpress human catalase in mitochondria show an increased life span and delayed age-related pathologies (Schriner et al., 2005). In plants, Palma et al. (2006) reported an increased activity of the ascorbate–glutathione cycle enzymes and a decreased content of ascorbate in mitochondria from senescent pea (Pisum sativum) leaves. Nevertheless, the sequence of events that lead to plant senescence and the contribution of mitochondria to this process are largely unknown. This work aims to fill this gap by assessing the relative contribution of the mitochondria and cytosol of host cells to nodule aging.

Materials and Methods

Biological material

Common bean (Phaseolus vulgaris L. cv Contender) seeds were surface-sterilized with 70% ethanol, germinated in pots containing a 1 : 1 (v:v) mixture of perlite:vermiculite, inoculated 7 d later with Rhizobium leguminosarum bv phaseoli strain 3622, and grown on a nutrient solution containing 0.25 mM NH4NO3 in a controlled-environment chamber (Loscos et al., 2008). Plants were separated randomly into three groups and nodules were harvested from plants at three different stages of development: young, c. 28-d-old plants without flowers; mature, c. 40-d-old plants at the late flowering–early fruiting stage; and senescent, c. 53-d-old plants with fully developed pods. Nodules were kept on ice and immediately used for purification of mitochondria and cytosol.

Purification of mitochondria and cytosol

All steps were performed at 0–4°C essentially as published (Iturbe-Ormaetxe et al., 2001), with some modifications. Nodules (c. 10 g) were ground very gently in a mortar with 30 ml of a medium containing 0.35 M mannitol, 2 mM EDTA, 10 mM KH2PO4, 2% (w/v) polyvinylpolypyrrolidone and 30 mM morpholinepropanesulfonic acid (MOPS) (pH 7.2). The extract was filtered through four layers of cheesecloth and centrifuged twice at 4000 g for 5 min. The supernatant was centrifuged at 12 000 g for 15 min and the pellet was saved. The soluble fraction was centrifuged at 50 000 g for 15 min and the new supernatant was designated as ‘cytosol’. The 12 000 g pellet was resuspended in 25 ml of washing medium consisting of 0.3 M mannitol, 1 mM EDTA and 20 mM MOPS (pH 7.2). After centrifugation at 12 000 g, the pellet was resuspended in 1.8 ml of washing medium and carefully placed on top of a four-layer Percoll gradient. This was generated by centrifugation at 13 000 g for 35 min with a 70Ti rotor in an Optima XL-100K ultracentrifuge (Beckman Coulter, Fullerton, CA, USA). The mitochondria band, situated between the 15% and 35% Percoll layers, was removed with a syringe. After two washes, mitochondria were resuspended in 500 μl of a medium containing 1 mM EDTA, 0.05% Triton X-100 and 50 mM MOPS (pH 7.2). For the MR, DR and GR assays, the resuspension medium was supplemented with 2 mM β-mercaptoethanol. For the GalLDH assay, the enzyme was extracted in a medium containing 50 mM Tris-HCl (pH 8.0) and 0.15% Triton X-100. Mitochondria were kept on ice for 1 h with occasional vortexing and then centrifuged at 13 000 g to remove intact organelles. Enzyme activities in the supernatant were assayed immediately. For proteomic and immunoblot analyses, the cytosolic and mitochondrial fractions were frozen at −80°C.

Enzyme activity assays

SOD activity was assayed by a method based on the inhibition of cytochrome c reduction by superoxide at 550 nm (Rubio et al., 2002). One SOD unit was defined as the amount of enzyme required to inhibit ferric cytochrome c reduction by 50%. APX and DR activities were determined following ascorbate oxidation at 290 nm (Asada, 1984) and ascorbate formation at 265 nm (Nakano & Asada, 1981), respectively. MR and GR activities were assayed by monitoring the oxidation of NADH (Dalton et al., 1993) and NADPH (Dalton et al., 1986) at 340 nm, respectively. GalLDH activity was determined following the reduction of cytochrome c at 550 nm as described (Matamoros et al., 2006). Glutamine synthetase (GS) activity was determined essentially as described in O'Neal & Joy (1973). Cytosol extracts were incubated for 5 min at 37°C in a buffer containing 20 mM MgSO4, 1 mM diethylenetriaminepentaacetic acid, 80 mM glutamate, 6 mM NH2OH, 8 mM ATP and 100 mM Tricine (pH 7.8). The reaction was terminated by addition of an equal amount of a solution containing 0.37 M FeCl3, 0.67 M HCl and 0.2 M trichloroacetic acid. The formation of γ-glutamyl hydroxamate was quantified spectrophotometrically at 540 nm using commercial γ-glutamyl hydroxamate as standard. Unless otherwise indicated, all activities were assayed at 25°C within the linear range.

Protease activity

Endopeptidase activity was measured in the cytosol using azocasein as a substrate (Pfeiffer et al., 1983). The cytosol fraction was isolated as described above, except that the extraction medium consisted of 0.35 M mannitol, 2 mM EDTA, 2% polyvinylpolypyrrolidone and 100 mM MES (pH 5.0). The cytosol (150 μl) was mixed with 250 μl azocasein (10 mg ml−1 in 50 mM Na-phosphate buffer, pH 7.0) and incubated at 37°C for 1 h. The reaction was stopped with 600 μl of 20% trichloroacetic acid, the samples were centrifuged at 13 000 g for 1 min, and 700 μl of supernatant was added to an equal amount of 1 M NaOH. In some experiments, cytosol extracts were pre-incubated for 15 min at 37°C with 100 μM of trans-epoxysuccinyl-l-leucylamido(4-guanidino)butane (E-64; Sigma–Aldrich) or 1 mM phenylmethylsulfonyl fluoride (PMSF; Sigma–Aldrich), which are specific inhibitors of cysteine proteinases and serine proteinases, respectively. One unit of endopeptidase activity was defined as the amount of enzyme required to produce a change in absorbance at 440 nm of 1.0 after 1 h.

Markers of senescence and oxidative stress

Lb concentration was determined by the pyridine-hemochrome assay (Appleby & Bergersen, 1980). The oxidative damage of lipids was measured by HPLC as malondialdehyde content after its reaction with thiobarbituric acid (Iturbe-Ormaetxe et al., 1998). The oxidative damage of proteins was estimated as the content of total carbonyl groups. Proteins were separated on 12.5% SDS-gels, and carbonyls were quantified by derivatization with 2,4-dinitrophenylhydrazine using the OxyBlot Protein Oxidation Detection kit following the manufacturer's instructions (Chemicon, Temecula, CA, USA).

Gene expression analyses

These were carried out as described (Loscos et al., 2008). In brief, RNA was isolated from bean nodules using the RNAqueous kit (Ambion) and treated with DNaseI (Roche) at 37°C for 30 min. Equal amounts of genomic DNA-free RNA were used for cDNA synthesis with Moloney murine leukemia virus reverse transcriptase (Promega). Quantitative reverse transcription-PCR analyses were performed with the iCycler iQ instrument and iQ SYBR-Green Supermix reagent (Bio-Rad) using specific primers (Supporting Information Table S1). Four reference genes (actin, protein phosphatase 2A subunit, ubiquitin and ubiquitin conjugating enzyme) were initially used to normalize transcript levels. All but the actin gene showed stable expression during nodule aging. Values were used for normalization with ubiquitin, although expression patterns were confirmed with the reference genes protein phosphatase 2A and ubiquitin conjugating enzyme. The relative values of gene expression (R) were then calculated using the inline image method, where CT is the threshold cycle (Livak & Schmittgen, 2001).

Immunoblot analyses

Cytosolic and mitochondrial proteins were separated on 15% SDS gels and transferred onto polyvinylidene fluoride membranes following standard protocols. Equal protein loading and transfer efficiency were assessed by Coomassie blue and Ponceau staining of gels and membranes, respectively. Immunoblot analyses were carried out using rabbit polyclonal antibodies against Arabidopsis thaliana mitochondrial PrxIIF (Finkemeier et al., 2005) and cytosolic PrxIIC (Horling et al., 2003). Primary antibodies were used at a dilution of 1:1000, and the anti-rabbit IgG horseradish peroxidase conjugated secondary antibody at a dilution of 1:20 000. Immunoreactive proteins were visualized using the SuperSignal West Pico (Thermo Scientific, Rockford, IL, USA) chemiluminescent reagent.

Glutathione immunolocalization

Small pieces of nodules (1–2 mm3) were fixed and embedded as described (Fernández-García et al., 2009). Ultrathin sections (80 nm) were obtained using a Leica EM UC6 ultramicrotome (Leica, Vienna, Austria) and collected on formvar-coated nickel grids. The grids were placed in phosphate-buffered saline (PBS) with 5% bovine serum albumin (BSA) for 30 min at room temperature and then incubated for 2 h with the primary antibody, anti-glutathione rabbit polyclonal IgG (Chemicon) diluted 1 : 50 in PBS containing 1% BSA plus 1% goat serum. The sections were washed three times in PBS and incubated for 1.5 h with the secondary antibody (goat anti-rabbit IgG coupled with 10 nm colloidal gold; British BioCell International, Cardiff, UK) diluted 1 : 50 in PBS supplemented with 1% BSA. The grids were washed in PBS and distilled water. Ultrathin sections were stained with uranyl acetate followed by lead citrate. However, for better visualization of differences in gold particle density, statistical evaluation was performed with cells from unstained sections. Negative controls were obtained by pre-adsorbing the anti-glutathione antibody with an excess of free glutathione before its application (Fernández-García et al., 2009). Samples were viewed with a Tecnai 12 electron microscope (Philips, Eindhoven, The Netherlands) equipped with a CCD SIS MegaView III camera.

Morphometrical data were obtained as described by Fernández-García et al. (2009). Gold particle densities in the vacuoles and apoplast were very low (0.1 and 0.2 gold particles μm−2, respectively), similar to the unspecific background labeling (0.2 gold particles μm−2). The data obtained were analyzed statistically using Statistix 8 (NH Analytical Software, Roseville, MN, USA) and presented as the number of gold particles per μm2.

Proteome and methionine sulfoxide analyses

Lyophilized mitochondrial and cytosolic fractions were resuspended in an ice-cold extraction buffer containing 25 mM MES (pH 7.2), 450 mM mannitol, 7 mM EDTA, 7 mM CaCl2, 5 mM MgCl2, 20 mM ascorbic acid and 10 mM dithiothreitol. Mitochondria were further broken and proteins resolubilized by sonication. Proteins were precipitated overnight at −20°C after adding 20 vol of ice-cold acetone containing 0.07% β-mercaptoethanol. Proteins were pelleted by centrifugation at 5000 g for 20 min, washed with cold acetone and β-mercaptoethanol, and air dried at room temperature. Dry pellets were resuspended in 8 M urea and 50 μg of protein was digested with 0.5 μg (1 : 100) of sequencing-grade endoproteinase Lys-C (Roche) for 5 h at 30°C. For trypsin digestion, samples were diluted 1 : 4 in trypsin buffer (10% acetonitrile, 50 mM NH4HCO3, 2 mM CaCl2). Fifty μg of protein was digested with 1.25 μl of trypsin beads (Applied Biosystems, Framingham, MA, USA) at 37°C overnight. Digested proteins were desalted using SPEC C18 columns (Agilent Technologies, Santa Clara, CA, USA), and the resulting peptides were eluted with methanol, vacuum dried and stored until use.

For proteomic analyses, samples were resuspended in 0.1% formic acid and 0.5 μg of protein sample was loaded onto a Peptide ES-18 column (15 cm × 0.1 mm, 2.7 μm; Sigma-Aldrich) in an Ultra HPLC Eksigent system (Axel Semrau, Sprockhövel, Germany) directly coupled to an Orbitrap XL mass spectrometer (Thermo Scientific). Peptides were eluted with a 5–60% acetonitrile gradient for 100 min. Dynamic exclusion settings were as described (Hoehenwarter & Wienkoop, 2010). After mass spectrometry analysis, raw files were searched against the DFCI Phaseolus vulgaris Gene Index (v4.0) using the Sequest algorithm. For identification and spectral count-based data matrix generation, the Proteome Discoverer (v1.2, Thermo Scientific) was used. A decoy database enabled false positive rate analysis. High-confidence peptides (false positive rate <0.1%) with better than 3 ppm precursor mass accuracy and at least two distinct peptides per protein were considered to have passed the criteria. Dynamic modification was set for methionine oxidation. The peptide spectra information was stored in ProMEX (; Wienkoop et al., 2012). Five replicates per developmental stage and subfraction were analyzed. For relative protein quantification, changes in spectral counts of at least two-fold and  0.05 (t-test) were considered significant.

In order to determine the ratios of the methionine vs methionine sulfoxide (MetSO) peptide-containing species, the Mass Accuracy Precursor Alignment (MAPA; Hoehenwarter et al., 2008) was used. In short, ion intensity counts for the precursor masses of the peptides were filtered using the ProtMAX software tool (Hoehenwarter et al., 2008) and a data matrix was generated. The ratios between the oxidized vs the nonoxidized peptide species were calculated using their ion intensity counts (sum of precursor ion intensities for each target peptide).


Oxidative stress occurs in mitochondria but not in the cytosol of senescing nodules

The involvement of mitochondria in nodule aging was assessed by isolating the cytosolic and mitochondrial fractions from bean nodules using differential centrifugation and Percoll density gradients. Nodules were harvested from plants at three different stages of development, termed ‘young’, ‘mature’ and ‘senescent’. To monitor nodule activity with advancing senescence, Lb, soluble protein and protease activity were quantified in the cytosol. Compared to young nodules, the concentrations of Lb and soluble protein in mature nodules did not vary significantly but decreased sharply in senescent nodules (Fig. S1a). Protease activity against azocasein was 4.5-fold greater in senescent nodules than in young or mature nodules (Fig. S1b). This protease activity was partly inhibited by PMSF, an inhibitor of serine proteinases, but not by E64, an inhibitor of cysteine proteinases (Fig. S1b).

The contribution of oxidative stress to the loss of nodule activity was investigated in the mitochondria and cytosol using lipid peroxidation (malondialdehyde) and protein oxidation (carbonyl groups) as markers (Halliwell & Gutteridge, 2007). Compared to young nodules, in mitochondria isolated from mature and senescent nodules, the amount of lipid peroxides increased by 48% and 75%, respectively (Fig. 1a), and the corresponding increases in oxidized proteins were 87% and 3.3-fold (Fig. 1b). By contrast, lipid peroxides in the cytosol of mature nodules decreased by 40% relative to young nodules (Fig. 1a) and the amount of oxidized proteins remained constant (Fig. 1b). Therefore, the mitochondria, but not the cytosol, experience oxidative stress during nodule aging.

Figure 1.

(a) Lipid peroxidation and (b) protein carbonylation in mitochondria and cytosol of young (Y), mature (M) and senescent (S) bean (Phaseolus vulgaris) nodules. Values are means ± SE of 4–8 replicates, each from nodules of 15 plants. Means denoted by the same letter are not significantly different at < 0.05 based on Duncan's multiple range test. MDA, malondialdehyde.

Ascorbate and (homo)glutathione metabolism is impaired in senescing nodule mitochondria

Because oxidative modifications of lipids and proteins were preferentially detected in senescing nodule mitochondria and these alterations might have originated from an inadequate antioxidant protection, we compared some important antioxidant enzymes and metabolites in the mitochondria and cytosol. Mitochondria of leaves (Jiménez et al., 1998) and nodules (Iturbe-Ormaetxe et al., 2001) contain a functional ascorbate-glutathione pathway in which four enzyme activities, acting in concert, keep H2O2 under control using NAD(P)H as reductant. In bean nodule mitochondria, APX and GR activities remained unaltered during senescence (Fig. 2a). However, MR activity was reduced by 49% at the senescent stage, whereas DR activity declined by 15% in mitochondria of mature and senescent nodules (Fig. 2a). By contrast, the activities of MR, DR and GR in the cytosol either increased or remained stable with advancing age, whereas cytosolic APX activity declined by 30% (Fig. 2b). To complete this part of the study, we attempted to estimate indirectly the effect of senescence on ascorbate and (homo)glutathione concentrations in nodule mitochondria. A direct quantification was not reliable because the concentrations are low and the isolation of mitochondria using differential centrifugations and Percoll gradients may cause leakage of water-soluble metabolites. Thus, to get an indication of ascorbate metabolism, we measured GalLDH activity, a mitochondrial enzyme that catalyzes the last step of ascorbate biosynthesis (Wheeler et al., 1998). This activity decreased by 30% in senescing nodules (data not shown), providing support for a limitation of ascorbate synthesis in mitochondria. On the other hand, an attempt was made to immunolocalize and quantify homoglutathione, the thiol tripeptide that replaces glutathione in bean nodules (Matamoros et al., 1999b). The antibody raised against glutathione recognized both the reduced and oxidized (disulfide) forms (Fernández-García et al., 2009) but, unfortunately, did not recognize homoglutathione. We thus decided to quantify total glutathione (reduced + oxidized) in the mitochondria and cytosol of pea (Pisum sativum) nodules (Fig. 3, Table 1), on the assumption that the relative abundance and response of the thiol during aging in the two cellular compartments are similar in bean and pea nodules. This is supported by the observations that in both types of nodules there is a progressive decrease in thiols during senescence induced by nitrate (Escuredo et al., 1996) or prolonged darkness of plants (Gogorcena et al., 1997). Another advantage is that pea nodules contain high concentrations of glutathione, which facilitates immunolocalization. Gold particles, marking the presence of glutathione, were observed (Fig. 3a–e) and counted (Table 1) in the mitochondria, cytosol, nuclei and bacteroids of infected cells. There was also weak labeling in the peroxisomes and plastids (data not shown). In young nodules, glutathione concentration was similar in the mitochondria and cytosol, whereas in mature nodules labeling was five-fold more abundant in the cytosol than in the mitochondria. In senescent nodules, labeling virtually disappeared in the mitochondria and cytosol (Table 1). Labeling in the bacteroids was much more intense than in the other cell compartments, and was highest in mature nodules and drastically reduced in senescent nodules (Fig. 3a,b, Table 1). Interestingly, significant labeling was also observed in the nuclei of infected cells (Fig. 3e), which is in agreement with the recent detection of glutathione in the nuclei of leaf cells (Fernández-García et al., 2009) or cultured cells (Diaz Vivancos et al., 2010) of A. thaliana. As was the case for the cytosol and bacteroids, labeling intensity in the nuclei was maximal in mature nodules and consistently declined in senescent nodules (Table 1).

Table 1. Relative abundance of glutathione in infected cells of pea (Pisum sativum) nodules
Developmental stageaMitochondriaCytosolBacteroidsNucleus
  1. a

    Data (gold particles μm−2) are means ± SE (n = 40 for mitochondria and bacteroids; n = 30 for cytosol; n = 25 for nuclei) and those denoted by different letters within the same columns are significantly different at < 0.05 based on Duncan's multiple range test.

  2. b

    nd, not detectable.

Young6.2 ± 0.9a4.0 ± 0.6a81.8 ± 6.9a2.1 ± 0.3a
Mature2.9 ± 0.3b15.5 ± 1.2b129.1 ± 9.0b3.8 ± 0.3b
Senescentndbnd6.4 ± 0.6cnd
Figure 2.

Specific activities of the enzymes of the ascorbate-(homo)glutathione pathway in (a) mitochondria and (b) cytosol of young (Y), mature (M) and senescent (S) bean (Phaseolus vulgaris) nodules. Values are means ± SE of 4–8 replicates, each from nodules of 15 plants. Means denoted by the same letter are not significantly different at < 0.05 based on Duncan's multiple range test.

Figure 3.

Immunogold localization of glutathione in pea (Pisum sativum) nodules. Images show intense labeling in (a) bacteroids of a mature nodule, (b) very weak or no labeling in the bacteroids of senescent nodules, (c, d) moderate labeling in the mitochondria and cytoplasm (arrows) of infected cells from mature nodules and (e) moderate labeling in the nucleus and nucleolus (arrows) of an infected cell from a mature nodule. (f) A negative control obtained as described in the 'Materials and Methods' section shows absence of labeling in bacteroids and other cellular compartments. b, bacteroid; c, cytoplasm; cw, cell wall; m, mitochondrion; n, nucleus; nu, nucleolus; v, vacuole. Bars: (a, b, e, f) 1 μm; (c, d) 0.5 μm.

Mitochondrial MnSOD and alternative oxidase are upregulated during nodule aging

The effects of nodule aging on two other key antioxidants of plant mitochondria, SOD and alternative oxidase (AOX), were also examined (Fig. 4). Mitochondrial MnSOD is located in the matrix and catalyzes the dismutation of inline image to H2O2, whereas AOX is located in the membrane and may participate in the protection of plant mitochondria from high ROS levels due to its ability to catalyze noncoupled respiratory electron transport (Vanlerberghe et al., 2009). The activity of MnSOD increased by 36% and > 50% in mature and senescent nodules, respectively (Fig. 4a). Likewise, the AOX transcript level was 13-fold higher in senescent nodules than in young nodules (Fig. 4b). These findings may reflect enhanced generation of inline image with age as a result of electron leakage from the respiratory chain (Rhoads et al., 2006).

Figure 4.

(a) Manganese superoxide dismutase (MnSOD) activity in mitochondria and (b) alternative oxidase (AOX) mRNA level in young (Y), mature (M) and senescent (S) bean (Phaseolus vulgaris) nodules. In (a) values are means ± SE of three replicates, each from nodules of 15 plants. Means denoted by the same letter are not significantly different at < 0.05 based on Duncan's multiple range test. In (b) values are means ± SE of four replicates, each corresponding to RNA extractions from nodules of four different sets of plants. Values were routinely normalized with respect to ubiquitin, although expression patterns were confirmed with other reference genes, as outlined in the 'Materials and Methods' section. R values > 2 (upregulation) with respect to the control (= 1 for young nodules) were considered to be significant and marked with an asterisk.

The Trx-Prx redox system may be compromised in aging nodules

The cytosol and mitochondria of legume nodules contain Trx-Prx redox systems that catalyze the reduction of H2O2 and other hydroperoxides (Tovar-Méndez et al., 2011). The mRNA levels of mitochondrial Trxo and PrxIIF declined significantly during nodule aging (Fig. 5). However, immunoblot analysis (Fig. S1c) revealed that the content of PrxIIF protein remained constant. Therefore, the enzyme is apparently quite stable and has a low turnover rate. Immunoblot analyses under nonreducing conditions indicated that the protein is present as high molecular mass aggregates, suggesting that PrxIIF is able to form oligomers or heterocomplexes with other proteins in vivo. The nodule cytosol also contains a NTR-Trx-Prx system comprising several NTR and Trxh isoforms and PrxIIB/C. The most abundant Trx isoforms in L. japonicus nodules are Trxh1 and Trxh4 (Tovar-Méndez et al., 2011). Interestingly, the bean Trxh4 isoform investigated here is probably the functional equivalent of a soybean Trxh essential for ROS scavenging in nodules (Lee et al., 2005). Nodule aging led to significant decreases of PrxIIB mRNA and protein levels and of the Trxh4 mRNA level, whereas no effect was seen on the Trxh1 transcript (Figs 5, S1c).

Figure 5.

Gene expression levels of selected peroxiredoxins and thioredoxins in young (Y), mature (M) and senescent (S) bean (Phaseolus vulgaris) nodules. Mitochondrial enzymes, PrxIIF and Trxo; cytosolic enzymes, PrxIIB, Trxh1 and Trxh4. Values are means ± SE of four replicates, each corresponding to RNA extractions from nodules of four different sets of plants. Values were routinely normalized with respect to ubiquitin, although expression patterns were confirmed with other reference genes as outlined in the 'Materials and Methods' section. R values < 0.5 (downregulation) with respect to the control (= 1 for young nodules) were considered to be significant and marked with an asterisk.

Mitochondrial and cytosolic proteomes

Proteomics is a powerful tool to investigate metabolic pathways in various tissues and stressful conditions. The effect of aging on the mitochondrial proteome, however, has not yet been reported in plants. Here, 143 proteins of mitochondria and 81 proteins of the cytosol were identified with high stringency and functionally classified.

The enzymes involved in the tricarboxylic acid cycle and oxidative phosphorylation accounted for the largest functional group in nodule mitochondria, but proteins related to amino acid metabolism, nucleotide metabolism and ureide biosynthesis were also widely represented (Table 2). In addition, the detection in mitochondrial extracts of catalase and glutamate synthase, marker enzymes of peroxisomes and plastids, respectively, indicates some contamination of mitochondria preparations with these organelles. To estimate protein abundance, a normalization similar to the Protein Abundance Index (Rappsilber et al., 2002) was used. The total spectral counts were normalized for catalase and the mitochondrial ATPase β-subunit, which have similar molecular masses (c. 60 kDa). This comparison revealed that the abundance of the mitochondrial protein is at least 10-fold greater than that of catalase. Also, the ratio of cytocrome c oxidase to uricase activity, which is an indicator of mitochondria purity with respect to peroxisomes, was 37-fold higher in purified mitochondria than in whole nodule extracts (Iturbe-Ormaetxe et al., 2001). On the other hand, glutamate synthase is a large-sized plastidic enzyme (c. 240 kDa) and therefore a direct comparison of the relative abundance of glutamate synthase and the ATPase β-subunit is only possible after normalization. This comparison showed that the abundance of glutamate synthase is c. 600-fold lower than that of ATPase. Overall, our results indicate that mitochondria showed a low contamination with peroxisomes and plastids.

Table 2. Proteins of bean (Phaseolus vulgaris) nodule mitochondria that are induced or repressed with aging
  1. Values (log2 of the number of spectral counts) are means of five biological replicates from five series of plants grown independently. Means denoted by the same letter are not significantly different at P < 0.05 according to Duncan's multiple range test.

  2. a

    Tentative consensus (TC) sequences according to the DFCI Bean Gene Index (4.0).

  3. b

    UniProt accessions (UniRef100).

Amino acid metabolism
Glutamate dehydrogenase 1TC34875Q5F2M92.5a2.5a1.0b
NADH glutamate synthase precursorNP7938775Q93WZ70.3a1.6b0.0a
Serine hydroxymethyltransferase, mitochondrialTC38452P504332.5a2.2ab1.0b
Homoserine kinaseTC37342Q9XEE03.3a3.2a0.4b
d-3-phosphoglycerate dehydrogenaseTC44520O041304.9a5.0a3.7b
Isovaleryl-CoA dehydrogenase precursorGW903826Q9SM622.5a2.7a1.2b
Aspartate aminotransferaseTC32236Q8HQQ02.3a2.3a0.4b
Ascorbate peroxidase, mitochondrial/chloroplasticTC38704Q5QHW73.1a2.0a0.7b
Peroxidase 1FD786965Q9XFL30.4ab1.2a0.0b
Heat shock 70 kDa protein, mitochondrial precursorTC32170Q018994.8a4.5a3.6b
Heat shock 10 kDa proteinTC39508A3FPF33.6a3.5a2.1b
Hypersensitive-induced response proteinTC32777Q6UNT32.1a3.1b3.3b
TCA/oxidative phosphorylation
Pyruvate dehydrogenase E1 component subunit alphaTC41304Q4JIY32.8a3.1a1.0b
Pyruvate dehydrogenase E1 beta subunit isoform 2TC34095Q9ZQY24.2a3.9a3.5b
Malate dehydrogenase, nodule-enhancedTC37821O812783.2a2.7a1.0b
2-Oxoglutarate dehydrogenase, mitochondrial-likeCV532177Q9M2T82.5a2.8a0.5b
2-Oxoglutarate dehydrogenase, mitochondrialTC34429Q3E9W23.0a3.1a1.3b
Fumarate hydratase 1TC32848Q10LR53.1a3.0a1.1b
Complex I (NADH dehydrogenase)TC46103Q438402.1a1.9a0.5b
Complex I (NADH dehydrogenase)TC37286Q5Z7T43.8a3.2b2.4c
Complex I (NADH dehydrogenase [ubiquinone])TC32737Q9SK664.0a3.3a1.5b
Complex I (NADH-ubiquinone oxidoreductase)TC37990Q7ILE42.5a2.4a1.0b
Complex III (ubiquinol-cytochrome c reductase Fe-S subunit)TC44534P497292.1a2.0a0.0b
Cytochrome c oxidase subunit VbTC32892Q0KKQ31.8a0.7ab0.0b
ATPase mitochondrial F1-gamma subunitTC34331Q59I534.2a4.1a3.2b
Nucleotide metabolism/ureide biosynthesis
Aminoimidazole ribonucleotide carboxylaseTC43042Q8W1971.7ab2.7a1.1b
Ribose-phosphate pyrophosphokinase 1TC35588Q425811.0a2.5b0.0c
Ribose-phosphate pyrophosphokinase 2TC35414Q425831.0a2.3b0.0c
Phosphoribosylformylglycinamidine cyclo-ligase, chloroplast/mitochondrialTC35536P524243.7a4.2a2.3b
Phosphoribosylpyrophosphate amidotransferaseTC36958O813580.6a1.5b0.2a
Adenylosuccinate-AMP lyaseTC36068Q8VZX01.3a1.4a0.2b
Formylglycinamide ribonucleotide amidotransferaseTC45362Q8VYU20.6ab1.3a0.0b
Dihydropyrimidine dehydrogenaseTC33820Q6Z7443.2a3.7a2.2b
Voltage-gated anion channel, mitochondriaTC34262Q6W2J23.0a2.6a0.8b
Voltage-dependent anion channelTC45536Q6W2J33.4a3.1a2.3b
Mitochondrial import receptor subunit TOM20, putativeTC36486P927922.5a2.3a0.0b
Nodulin 41TC36046G1FTR54.6a4.5a2.8b
Protein degradation
Mitochondrial processing peptidase beta subunitTC36407Q9AXQ24.7a4.7a4.0b
Subtilisin-like proteaseTC37598Q9ZUF62.8a0.9b0.7b
Serine carboxypeptidase-like proteinTC36390Q8L7B21.2a0.2ab0.0b
Metal homeostasis
Ferritin-2, chloroplast precursorTC32629Q417092.0a1.1b0.0c
CBS domain-containing proteinTC36694Q0IPT33.5a3.3a0.0b
CXE carboxylesteraseTC37468Q0ZPW91.4a2.8b3.1b
Prohibitin-1, mitochondrialTC35529O043613.6a3.4a2.5b
Glutamine cyclotransferase precursorTC35679O812262.5a2.4a1.0b
Enoyl-ACP reductase precursorTC41680O049452.0a2.2a0.0b
Glycine-rich RNA-binding protein PsGRBPTC45102P934862.2a1.2b0.3c

The effect of aging was less pronounced on the mitochondrial proteome than on the cytosolic proteome. Thus, 28% of the mitochondrial proteins (Table 2) and 63% of the cytosolic proteins (Table 3) identified in this work showed decreased concentrations in aged nodules. In mitochondria, proteins that are crucial for nitrogen and carbon metabolism (glutamate dehydrogenase and aspartate aminotransferase) or that are involved in the recognition and translocation of mitochondrial preproteins (TOM20) decreased in senescent mitochondria (Table 2). Only two proteins of mitochondria from senescent nodules were induced with aging, a hypersensitive response-induced protein and a carboxylesterase. The decrease in cytosolic proteins was particularly intense for proteins related to protein synthesis and cytoskeleton organization (Table 3). Notably, some actins and β-tubulins were already reduced in mature nodules, which suggests that the disorganization of cytoskeletal structures is an early event during aging of nodule host cells. The contents of phosphoenolpyruvate carboxylase and CuZnSOD decreased in mature and senescent nodules. Only one protein, heme oxygenase, which is involved in heme catabolism, was induced in mature and senescent nodules with respect to young nodules.

Table 3. Proteins of bean (Phaseolus vulgaris) nodule cytosol that are induced or repressed with aging
  1. Values (log2 of the number of spectral counts) are means of five biological replicates, each from a different series of plants grown independently. Means denoted by the same letter are not significantly different at P < 0.05 according to Duncan's multiple range test.

  2. a

    Tentative consensus (TC) sequences according to the DFCI Bean Gene Index (4.0).

  3. b

    UniProt accessions (UniRef100).

Nitrogen fixation
Amino acid metabolism
Glutamine synthetase PR-1TC35865P047701.9ab2.8a1.5b
Cysteine synthaseTC36772A3RM062.1a1.4ab0.2b
Methionine synthaseTC42435Q71EW82.1a0.9ab0.2b
S-adenosylmethionine synthetaseTC35526A4PU484.2a3.4a1.0b
Carbohydrate metabolism
Fructose-bisphosphate aldolase, cytosolicTC36120O657354.9a3.2ab1.0b
Glyceraldehyde-3-dehydrogenase C subunitTC37464Q07CZ33.1a3.9a1.0b
Glyceraldehyde-3-phosphate dehydrogenase, cytosolicTC33650P349211.6a2.1a0.0b
Triosephosphate isomeraseTC33325Q5JZZ32.2a1.8a0.0b
Fructokinase-like proteinTC43195Q9LNE33.4a3.1a1.0b
Sucrose synthaseTC35849Q8GTA34.9a4.7a3.0b
UDP-glucose pyrophosphorylaseTC36416Q8W5570.9a0.6ab0.0b
Carbon fixation
Phosphoenolpyruvate carboxylaseTC34225Q9ZS532.8a1.4b0.0c
Carbonic anhydraseTC36859Q9AU121.9a1.1a0.0b
Ascorbate peroxidase, cytosolicTC32379Q9XG113.8a3.2a1.1b
Superoxide dismutase [Cu-Zn]FE899597Q417122.4a1.1b0.2b
Heme oxygenase 1TC44951Q06H320.6a2.0b1.7b
Monodehydroascorbate reductaseTC33312Q94FX11.9a2.7a0.4b
Protein synthesis
Ribosomal protein 60S acidic P2TC38965Q66PF92.2a0 b0.0b
Ribosomal protein S8TC41325A8QJ721.7a0.9ab0.0b
Ribosomal protein 40S SATC41632Q3HRZ61.1a0 b0.0b
Ribosomal protein 40S S15-like proteinTC35497O225181.2a0.6ab0.0b
Eukaryotic initiation factor 4A-10-likeTC43412Q9FY640.7ab1.4a0.0b
Translation protein SH3-likeCV531340A9PD171.2a0.5ab0.0b
Putative translation elongation factorTC44261Q2HRD63.1a2.6a0.7b
Predicted elongation factor 1-beta-1 likeTC34901Q9ASR13.0a1.0b0.0b
Pathogenesis-related protein 1FG229903Q9SCX33.0a1.3ab0.7b
Heat shock 70 kDa cognate protein 1TC45292P259852.8a2.0ab0.3b
Heat shock 70 kDa cognate protein 2TC35538Q5QHT42.7a1.8a0.0b
Heat shock 70 kDa cognate protein 3TC39659Q5QHT32.5a2.1a0.0b
Molecular chaperone Hsp90-1TC33556Q5QHT23.3a2.9a1.2b
Peptidyl-prolyl cis-trans isomeraseGW885607Q6UJX42.8a1.3b0.7b
Malate dehydrogenaseTC32699Q411193.4a3.1a1.5b
Nucleotide metabolism
Adenylate kinase BTC32308Q2PYY82.2a2.4a1.0b
Ribonucleoside hydrolaseTC35502Q084801.6ab1.9a0.7b
14-3-3 proteinTC35424Q6ZJ052.5a3.0a0.7b
14-3-3-like protein ATC33838Q93XW11.9ab2.7a0.7b
14-3-3-like protein CTC34878Q964501.0ab1.5a0.0b
Cell wall/Cell organization
Tubulin alpha-3 chainTC33053Q9ZRR52.6a0.9b0.0b
Tubulin beta-1 chainTC43169P373923.5a2.3b0.3c
Tubulin beta-1 chainTC42085A7KQH33.7a2.3b0.3c
Tubulin beta-3 chainTC35683Q406653.4a1.9b0.3c
Tubulin ATC32424Q2TFP21.2a1.7a0.2b
Nodulin 6lTC38826Q8S4Q41.4a2.0a0.0b
Nodulin 30 precursorTC38855Q411212.9a1.2b0.0c
Nodulin 41TC36046G1FTR54.3a3.7a1.1b
Hormone metabolism
CXE carboxylesteraseTC37468Q0ZPW90.0a1.2b0.5ab
Inorganic pyrophosphataseTC34001Q949J11.6a0.8ab0.2b

Post-translational modifications of mitochondrial and cytosolic proteins

The free radical NO is a major signal molecule that participates in many crucial processes of plants (Lamotte et al., 2005). In nodules, NO is a key player in the onset of symbiosis (Baudouin et al., 2006; del Giudice et al., 2011) and in senescence (Cam et al., 2012). Unregulated production of NO may result, however, in formation of peroxynitrite and other nitrating species leading to nitrosative stress (Valderrama et al., 2007; Corpas et al., 2011). We investigated if aging leads to the accumulation of nitrated proteins in the mitochondria and cytosol. To this end, we used an antibody against nitro-tyrosine because increased concentrations of this compound have been found in plant tissues exposed to abiotic stress (Corpas et al., 2011). Contrary to our expectations, nodule aging was accompanied by a decline in cytosolic and mitochondrial proteins containing nitro-tyrosine, which indicates that nitrosative stress was not involved.

On the other hand, the oxidation of methionine to MetSO and the reduction of MetSO back to methionine by peptide MetSO reductases are emerging as a novel mechanism for regulation of cell function (Hardin et al., 2009). Using proteomic analysis combined with the targeted MAPA approach (see the 'Materials and Methods' section), we found that the MetSO content in mitochondrial or cytosolic proteins did not increase with aging, with the exception of nodulin 6l and GS. The effect of methionine oxidation on GS activity was studied in more detail as this enzyme is essential for controlling carbon and nitrogen metabolism in nodules (Vance, 2008) and has been recently found to be also a target of tyrosine nitration (Melo et al., 2011). The GS-N1 isoform contains two methionine residues at positions 31 and 33 that are oxidized to MetSO during aging (Fig. 6). However, methionine oxidation could not be detected in GS-PR1, another cytosolic isoform, in agreement with the observation that in this protein leucine substitutes for the two methionine residues susceptible to oxidation. The effect of methionine sulfoxidation on GS activity was examined in the cytosol of young, mature and senescent nodules. However, no significant differences could be observed in the total GS activity (GS-N1 + GS-PR1), which may be explained either by a genuine lack of effect of sulfoxidation on GS-N1 activity or by a masking effect of GS-PR1 activity, which does not bear MetSO.

Figure 6.

Percentage of methionine sulfoxide (MetSO) to total methionine residues (MetSO/(Met + MetSO) × 100) of the glutamine synthetase isoform GS-N1 in the cytosol. GS-N1 contains two methionine residues at positions 31 and 33 that are prone to sulfoxidation. The peptide containing Met-31 and Met-33 shows differential susceptibility in young (Y), mature (M) and senescent (S) bean (Phaseolus vulgaris) nodules. For quantification, it was possible to distinguish between no oxidation, single MetSO and doubly MetSO. Ratios of the abundance of single MetSO and doubly MetSO peptides with respect to the total abundance of the three peptide species were calculated. Values are means ± SE of five replicates and those denoted by the same letter are not significantly different at < 0.05 based on Duncan's multiple range test.


Legume nodule senescence is a poorly understood process that begins early after the onset of pod filling (Bethlenfalvay & Phillips, 1977) and is characterized by a decrease in N2 fixation, Lb and total protein and by a concomitant increase in protease activity (Pladys & Vance, 1993; Puppo et al., 2005; Groten et al., 2006; Loscos et al., 2008). In this work, most of these changes were monitored as markers of nodule senescence. Two novel observations concerning Lb and proteases are noteworthy. First, the induction of heme oxygenase, an enzyme catalyzing the oxidation of heme to carbon monoxide, free iron and biliverdins (Shekhawat & Verma, 2010). This induction is probably associated with the decrease in Lb (Fig. S1) and the increase in biliverdin-like pigments that are present in the extracts of senescent nodules (data not shown). The release of free heme from Lb, which may be favoured by high proteolytic activity and acidic conditions in senescent nodules, could induce heme oxygenase, although the induction of heme oxygenase genes by heme is still controversial in plants (Baudouin et al., 2004). Second, our experiments with protease inhibitors suggest that the endoprotease activity in the cytosol of senescent bean nodules is mainly due to serine proteinases rather than cysteine proteinases, which is at odds with the situation described in senescing alfalfa (Medicago sativa) nodules (Pladys & Vance, 1993). Compartmentalization of some cysteine proteinases in the vacuoles or symbiosomes, as suggested by Pérez-Guerra et al. (2010), could explain the differences because the experiments in alfalfa were carried out in whole nodule extracts whereas we used purified cytosol. Groten et al. (2006) reported increases in cysteine and serine proteinase activities during aging of pea nodules, pointing to interspecific variations in the major types of proteases associated with senescence.

A proteomic approach was used also to investigate the changes in protein concentrations of the mitochondria and cytosol during the natural senescence of nodules. This type of study is innovative and important because no data are available so far on the proteomes of both nodule fractions. Proteomic analysis revealed that nodule aging entails a marked decline in the concentration of nearly all cytosolic proteins, which can be largely attributed to enhanced proteolytic activity. These data also indicated that some key proteins for nodule function, such as phosphoenolpyruvate carboxylase, CuZnSOD and proteins involved in cytoskeleton dynamics and protein synthesis, were already compromised at the mature stage, before any evident symptom of senescence. However, the number of mitochondrial proteins showing reduced concentrations in senescent nodules was considerably lower than that of the corresponding cytosolic proteins. In mitochondria, the major decreases were observed for the enzymes of de novo purine biosynthetic pathway. These results are consistent with lower rates of N2 fixation in senescent nodules because purines are precursors of ureides, the form in which nitrogen is transported from nodules to the shoot in most tropical legumes (Vance, 2008).

Oxidative stress is frequently associated with plant senescence. Elevated concentrations of randomly oxidized molecules cause cell dysfunction and ultimately cell death. Nevertheless, specific oxidation of key cell components may afford redox signals (‘oxidative signaling’) that can be involved in the orchestration of senescence (Foyer & Noctor, 2005). The occurrence of oxidative stress in aging nodules has been reported in soybean (Evans et al., 1999), lupin (Lupinus albus; Hernández-Jiménez et al., 2002) and bean (Loscos et al., 2008), but not in pea (Groten et al., 2005). In all of these studies whole nodules were analyzed, which provides only partial information and possibly overlooks oxidative events localized in organelles. Here we present new data about the effects of natural senescence on the major antioxidants of nodule mitochondria. Specifically, our results underline the importance of ascorbate and glutathione in nodule senescence at the subcellular level. Changes in the concentrations and redox states of both antioxidants are inherent to nodule senescence (Evans et al., 1999; Matamoros et al., 1999a; Groten et al., 2005; Vanacker et al., 2006; Loscos et al., 2008). We show that aging reduced the capacity of mitochondria, but not of the cytosol, to regenerate ascorbate. Also, the immunological approach used in this work allowed the first direct quantification of glutathione in nodule organelles. The early decrease of glutathione in mitochondria from mature nodules is remarkable and may be due to transport and/or degradation but not to the inhibition of thiol synthesis, which only occurs in the cytosol and plastids (Clemente et al., 2012). Therefore, the redistribution of glutathione and its subsequent effects on the mitochondrial redox state appear to be important features of nodule senescence. Interestingly, Diaz Vivancos et al. (2010) found that cytosolic glutathione is recruited in the nuclei during cell proliferation. However, because in the present work the relative levels of glutathione were quantified in infected cells rather than in meristematic, actively dividing cells of pea nodules, glutathione should perform additional functions in the nuclei, probably related to the redox regulation of transcription factors and the protection of DNA against oxidative damage. In mitochondria, the decrease of glutathione and probably of ascorbate, as a result of diminished MR, DR and GalLDH activities, may trigger an oxidative state during senescence, as evidenced by the accumulation of lipid peroxides and carbonylated proteins. The induction of MnSOD and AOX reflects an increased generation of ROS during electron transport and likely contributes to the oxidative state of senescing mitochondria.

The high concentrations of oxidized proteins in mitochondria of senescing nodules were accompanied only by moderate proteolysis in comparison to the cytosol, according to proteomic analyses. Conversely, the high proteolytic activities in the cytosol were not accompanied by increases in oxidized proteins. These findings are novel and unexpected because oxidized proteins have been reported to be more susceptible to degradation (Palma et al., 2002). Two explanations are that mitochondrial proteases are not fully proficient in removing oxidized proteins and that protein carbonylation is a selective process yielding modified proteins with useful functions. In humans, moderately damaged proteins are prone to attack by proteases, but extensively damaged proteins may be more resistant and tend to form aggregates (Møller & Kristensen, 2004).

On the other hand, peptides derived from mitochondrial oxidized proteins could act as secondary messengers and participate in retrograde ROS signaling (Møller & Sweetlove, 2010). Nevertheless, the mechanism by which ROS derived from mitochondria influences cell metabolism is unknown. The identification of oxidatively modified proteins with altered activities is crucial to understanding how cells respond and adapt to redox changes. Our proteomic analysis, combined with the use of cutting-edge methodology capable of detecting peptides differing only in the presence of MetSO (Hoehenwarter et al., 2008), revealed that specific methionine residues of a cytosolic GS isoform (GS-N1) are sulfoxidized in mature and senescent nodules. The post-translational inactivation of GS by tyrosine nitration has been reported (Melo et al., 2011). These authors proposed that GS inactivation is related to the inhibition of nitrogenase by NO, thus establishing a direct link between N2 fixation and ammonia assimilation in nodules. We report here, for the first time, that methionine sulfoxidation is also important for the regulation of GS in response to ROS. Taking together all these findings, we propose that cytosolic GS is a central element in the cellular signaling network in nodules.

In conclusion, our results provide evidence that changes in the concentration of ascorbate and glutathione and in the capacity to regenerate ascorbate during nodule senescence induce an oxidative shift in mitochondria, which is consistent with the upregulation of MnSOD and AOX. This oxidative state is evidenced by the accumulation of oxidized lipids and proteins, which may be acting also as signaling molecules. Senescence is accompanied by a reduction of key enzymes implicated in carbon metabolism, by the disorganization of the cytoskeleton, and by a sharp and general reduction of cytosolic proteins due to inhibition of protein synthesis and induction of proteolytic activity.


We thank Carmen Pérez-Rontomé for assistance with growing the plants and drawing the figures, Christiana Staudinger for help with proteomic analyses, Enrique Olmos and Euan James for valuable comments on glutathione immunolocalization, and Frank Minchin and two anonymous reviewers for constructive criticism on the manuscript. This work was funded by Ministerio de Ciencia e Innovación-Fondo Europeo de Desarrollo Regional (AGL2011-24524) and Gobierno de Aragón-Fondo Social Europeo (group A53).