Update on Plant Antioxidants

Biochemistry and Molecular Biology of Antioxidants in theRhizobia-Legume Symbiosis1,2

Manuel A. Matamoros, David A. Dalton, Javier Ramos, Maria R. Clemente, Maria C. Rubio, andManuel Becana*

Departamento de Nutricion Vegetal, Estacion Experimental de Aula Dei, Consejo Superior deInvestigaciones Cientıficas, Apartado 202, 50080 Zaragoza, Spain (M.A.M., J.R., M.R.C., M.C.R., M.B.); andBiology Department, Reed College, Portland, Oregon 97202 (D.A.D.)

The complete reduction of molecular oxygen towater requires four electrons and is catalyzed bycytochrome oxidase in aerobic bacteria and mito-chondria. However, 1% to 3% of all oxygen con-sumed by respiration is inevitably reduced to super-oxide radicals and hydrogen peroxide (H2O2). Theseand other oxygen-derived molecules with moderateto very high reactivity are known as reactive oxygenspecies (ROS). The term includes free radicals (mol-ecules with one or more unpaired electrons, such asthe superoxide and hydroxyl radicals) and non-freeradicals (molecules with no unpaired electrons, suchas H2O2 and singlet oxygen). The main sources ofROS in plants under physiological conditions arerespiration, photosynthesis, and N2 fixation (Table I).In addition, ROS are produced at high rates whenplants are exposed to abiotic, biotic, or xenobioticstress. Similarly, the term reactive nitrogen species(RNS) refers to nitrogen-derived molecules with vari-able reactivity and includes free radicals (nitric ox-ide) and non-free radicals (peroxynitrite). Nitric ox-ide is involved in many key physiological processesin animals and, as shown in recent years, also inplants (Table I). It reacts with the superoxide radicalsto form peroxynitrite and probably with thiol com-pounds to form nitrosothiols. The investigation ofRNS is at present a truly novel and important field inplant biology.

The superoxide radical, H2O2, and nitric oxidehave moderate reactivity toward biomolecules and,thus, may have some direct detrimental effects inplants. The superoxide radical inactivates dihydroxy-acid dehydratase (required for the synthesis ofbranched chain amino acids) and aconitase (required

for the operation of the Krebs cycle) by oxidizing theiron-sulfur clusters at the active site, and ribonucle-otide reductase (required for DNA synthesis) by ox-idizing an essential Tyr radical. Also, H2O2 can inac-tivate Calvin cycle enzymes, metalloproteins such assuperoxide dismutases (SODs), and hemoproteinssuch as nodule leghemoglobin (Dalton, 1995; Scan-dalios et al., 1997). However, the real threat of thesuperoxide radical and H2O2 is their potential to actas precursors of the hydroxyl radical. The hydroyxlradical can readily oxidize amino acid residues ofproteins, fatty acids of phospholipids, and deoxy-Riband bases in DNA (Halliwell and Gutteridge, 1999).Nitric oxide can directly inhibit iron-containing pro-teins (Neill et al., 2002), but its toxicity stems mainlyfrom its ability to react with the superoxide radical toform peroxynitrite. This compound can induce lipidperoxidation, nitration of Tyr residues of proteins,oxidation of thiols, and nitration or deamination ofDNA bases (Halliwell and Gutteridge, 1999).

However, the same three ROS or RNS mentionedabove may perform useful roles in plants. This islargely because they show moderate reactivity andare mainly generated by enzymes; hence, their ratesand subcellular sites of production may be undermetabolic control. The superoxide radical and H2O2are involved in lignification of cell walls, defenseagainst pathogen attack, and sensing of, and subse-quent adaptation to, stressful conditions. H2O2 canalso induce programmed cell death during theplant’s hypersensitive response to infection by mod-ulating gene expression (Neill et al., 2002). Nitricoxide also acts as a signal molecule and is involved inthe control of gene expression, hypersensitive re-sponse, antioxidant defense, organogenesis, and sto-matal closure (Neill et al., 2002; Lamattina et al.,2003).

Plant cells contain an impressive array of antioxi-dant metabolites and enzymes that scavenge or pre-vent the formation of the most aggressive ROS andRNS, thus protecting cells from oxidative damage. Inaddition, antioxidant enzymes control the steady-state levels of the moderately reactive ROS and RNS,allowing them to perform important roles at specificsites, environmental conditions, or developmentalstages of plants. Although antioxidants have multi-

1 This work was supported by the National Science Foundation(grant nos. IBN–9507491 and IBN–9816583 to D.A.D.) and by theDireccion General de Investigacion Cientıfica (Spain; grant no.AGL–2002– 02876 to M.B.).

2 This paper is dedicated to Robert V. Klucas, our friend andmentor, whose wisdom and kind spirit will long remain as aninspiration to those who knew him.

* Corresponding author; e-mail becana@eead.csic.es; fax 34–976–716145.

Article, publication date, and citation information can be foundat www.plantphysiol.org/cgi/doi/10.1104/pp.103.025619.

Plant Physiology, October 2003, Vol. 133, pp. 499–509, www.plantphysiol.org © 2003 American Society of Plant Biologists 499

ple roles in diverse physiological processes in plants,we present here a restricted overview of the role ofantioxidants in the rhizobia-legume symbiosis. Read-ers are referred elsewhere for a more general cover-age of antioxidants in plants, in particular the excel-lent reviews by May et al. (1998) and Mittler (2002).

As a result of the complex and continuous molec-ular interplay between the bacteria and the plant,large amounts of ROS and possibly RNS are gener-ated during the lifetime of nodules; hence, an impor-tant asset of antioxidant enzymes is expected in bothsymbiotic partners. These and other molecular stud-ies of the symbiosis are greatly facilitated by selectingMedicago truncatula or Lotus japonicus as model le-gumes, respectively, for indeterminate or determi-nate nodulation (Udvardi, 2001). Both legume specieshave a small diploid genome, are autogamous, havea short generation time and large seed production,and are amenable to transformation and mutantscreening. In addition, the chloroplastic genome of L.japonicus and the genomes of Sinorhizobium melilotiand Mesorhizobium loti (the bacterial components ofthe symbioses) have been entirely sequenced, and thenuclear genomes of M. truncatula and L. japonicus arebeing sequenced at a fast pace.

ASCORBATE, GLUTATHIONE, ANDHOMOGLUTATHIONE ARE MAJORANTIOXIDANTS OF NODULES

Ascorbate (vitamin C) is a water-soluble reductantthat can be found in nodules at concentrations of 1 to2 mm. Ascorbate is required for the progression ofthe cell cycle and for cell elongation. The latter effecthas been attributed to its participation as cofactor ofprolyl hydroxylase (required for the synthesis ofHyp-rich proteins of the cell wall) and to the abilityof apoplastic ascorbate to alter the properties of theplasma membrane or to inhibit the cross-linking ofHyp-rich proteins by phenols (Horemans et al., 2000).However, the best known functions of ascorbate arebased on its properties as an antioxidant. Ascorbate

regenerates the �-tocopherol oxidized by ROS at themembrane-cytosol interface, is a direct scavenger ofmost ROS, and is the substrate of ascorbate peroxi-dase (APX). The major pathway for ascorbate synthe-sis has been elucidated (Wheeler et al., 1998). The laststep, catalyzed by l-galactono-�-lactone dehydroge-nase, occurs in the inner membrane of mitochondria(Horemans et al., 2000).

The thiol tripeptide GSH (�Glu-Cys-Gly) is also anabundant metabolite of plants, where it performsmultiple functions, including transport and storageof sulfur, control of cell redox status, progression ofthe cell cycle, protection of protein thiol groups, anddetoxification of heavy metals and xenobiotics (Mayet al., 1998). GSH is an important antioxidant in itsown right but also as a substrate for glutathionereductase and glutathione peroxidase (GSH-PX).However, in some legumes, homoglutathione (�Glu-Cys-�Ala) may partially or completely replace GSH.Homoglutathione is the major tripeptide in nodulesof soybean (Glycine max), common bean (Phaseolusvulgaris), and mung bean (Vigna radiata), whereasGSH is predominant in nodules of pea (Pisum sati-vum), alfalfa (Medicago sativa), and cowpea (Vignaunguiculata). In each case, the major thiol is present atconcentrations of 0.5 to 1 mm. The synthesis ofGSH and homoglutathione proceeds through twoATP-dependent steps catalyzed, respectively, by�-glutamylcysteine synthetase and a specific gluta-thione or homoglutathione synthetase (Fig. 1). Theenzymes from pea, mung bean, and tobacco (Nicoti-ana tabacum) leaves have been partially purified andlocalized to the cytosol and plastids (Rennenberg,1997). The biochemical properties of such enzymes,along with the information gained for the noduleenzymes using molecular approaches, are summa-rized in Table II. Genomic and cDNA clones for allthree enzymes have been isolated, and gene struc-tures have been determined. The �ecs gene of L.japonicus contains 15 exons with identical size andhigh sequence homology (78% identity) to that ofArabidopsis (Matamoros et al., 2003). In both M.

Table I. Production of ROS and RNS in plants

ROS or RNS Cellular Source

Superoxide radical Electron transport chains of mitochondria, bacteroids, chloroplasts, endoplasmic reticulum,peroxisomes, and plasma membrane. NADPH oxidase in membranes. Oxidation ofleghemoglobin in cytosol. Xanthine oxidase and membrane polypeptides in peroxisomes.Oxidation of nitrogenase and ferredoxin in bacteroids.

H2O2 Electron transport chains of mitochondria, bacteroids, chloroplasts, endoplasmic reticulum,and plasma membrane. CuZnSOD in cytosol and plastids, MnSOD in mitochondria andbacteroids, FeSOD in plastids. Photorespiration, fatty acid �-oxidation, urate oxidase,and MnSOD in peroxisomes.

Organic and lipid peroxides Nonenzymatic lipid peroxidation. Lipoxygenase.Hydroxyl radical Reaction of superoxide radical with H2O2 catalyzed by trace amounts of Fe or Cu.Singlet oxygen Photoinhibition in chloroplasts.Nitric oxide Nitrate reductase in leaves and other plant organs. P protein of the Gly decarboxylase

complex in infected tobacco leaves. Nitric oxide synthase (?) in peroxisomes.Peroxynitrite Reaction of nitric oxide with the superoxide radical.

Matamoros et al.

500 Plant Physiol. Vol. 133, 2003

truncatula and L. japonicus, the gshs and hgshs geneshave 12 exons of identical size (except for the firstones, which are very close in size), show high se-quence homology (83% identity between the codingsequences of the two genes), and are tandemly ar-ranged (and with the same orientation) in the ge-nome. These observations indicate that the two genesoriginated by duplication (Frendo et al., 2001; Mata-moros et al., 2003). In L. japonicus, the genes areseparated by only 8 kb, appear to be present as singlecopies, and encode proteins with putative plastidsignal peptides. The expression patterns of gshs andhgshs are clearly different in the two model legumes.In M. truncatula, hgshs is preferentially expressed inthe roots and nodules and gshs in the leaves (Frendoet al., 2001), whereas in L. japonicus, hgshs is ex-pressed in the roots and leaves and gshs in the nod-ules (Matamoros et al., 2003). Why the hgshs gene wasrecruited during evolution exclusively in the legumefamily and is only expressed in some species or or-gans remain unsolved questions but the differentialexpression of gshs and hgshs do suggest specific rolesfor their enzymatic products.

Bacteroids also have high GSH concentrations dueto their own �-glutamylcysteine synthetase and glu-tathione synthetase (Moran et al., 2000). They lackhomoglutathione synthetase, but significant amountsof homoglutathione are found in bean nodule bacte-

roids as a result of uptake from the host infected cells(Moran et al., 2000). The GSH may be internallyconsumed by bacteroids in metabolic reactions andmaintenance of cellular redox status rather than be-ing exported to the plant (Iturbe-Ormaetxe et al.,2001). Recently, a mutant of Rhizobium tropici hasbeen isolated which is deficient in glutathione syn-thetase and contains only 3% of the GSH present inthe wild-type strain (Riccillo et al., 2000). This mutantis sensitive to weak organic acids and to osmotic andoxidative stress, and the addition of GSH restores theresponses to these stresses to wild-type levels. Inter-estingly, the mutant can form effective nodules onbean, but it is out competed by the wild-type strain,indicating that GSH is important for stress toleranceand the symbiotic process (Riccillo et al., 2000).

NODULES CONTAIN THREE TYPES OFPEROXIDASES WITH DISTINCT FUNCTIONS

APXs belong to the class I of hemoperoxidases(intracellular enzymes) and catalyze the reduction ofH2O2 to water by ascorbate. In nodules, APX activityhas been found in the cytosol and mitochondria (Dal-ton et al., 1993; Iturbe-Ormaetxe et al., 2001), butadditional isoforms probably exist in peroxisomesand plastids, as occurs in leaves (Jimenez et al., 1997).Cytosolic APX has been purified from soybean nod-

Figure 1. Three types of peroxidases that can be found in legume nodules. The scheme depicts gene structures, proteins, andactivities catalyzed by representative enzymes of each type: APX of pea leaf cytosol, GPX of horseradish roots, and GSH-PXof L. japonicus nodules. Gene diagrams show exons (except untranslated regions [UTRs]) in red, introns in yellow, and UTRsin blue. Numbers are length in base pairs. Protein diagrams show: a, in APX and GPX, the distal and proximal His residues(H) that bind the heme groups (in red); b, in GPX, the N- and C-terminal signal peptides, the four conserved disulfide bridges,and one of the eight glycosylated Asn residues (N*); and c, in GSH-PX, the plastid signal peptide and some importantresidues of the three typical domains (“signatures”). Numbers are length in amino acid residues. ASC, ascorbate; MDHA,monodehydroascorbate; RH2, artificial reductant.

Antioxidants in the Rhizobia-Legume Symbiosis

Plant Physiol. Vol. 133, 2003 501

ules, cDNA clones isolated, and recombinant enzymeand antibody produced. The most important proper-ties of APX are compiled in Table II. The cytosolicenzyme is stable in the absence of ascorbate, contraryto the chloroplastic isoforms, and is also immunolog-ically distinct from them. Its amino acid sequence haslittle homology with guaiacol peroxidases (GPXs) butsignificant homology with yeast (Saccharomyces cer-evisiae) cytochrome c peroxidase (Mittler and Zilins-kas, 1992). The genes for cytosolic APX (apx1) of peaand Arabidopsis have 10 exons and nine introns (Fig.1). The first intron is located in the 5�-UTR, and thismay have an effect on transcription, perhaps enhanc-ing expression level (Mittler and Zilinskas, 1992).This does not occur with the gene (apx2) encoding thetwo isoforms (stromal and thylakoidal) of chloroplas-tic APX (Shigeoka et al., 2002).

GPXs (also called “nonspecific” or “classical” per-oxidases) are class III peroxidases (secretory en-zymes) found in the extracellular spaces and vacu-oles. They have been implicated in a wide range ofprocesses, including lignification, suberization, auxincatabolism, defense against pathogens, salt tolerance,and oxidative stress. GPXs use phenolic compounds

as substrates and are typically assayed with artificialelectron donors. In nodules, they exist as multipleisoforms, but none of them have been characterized.APXs and leghemoglobins also display “GPX” activ-ity but at much lower rates. However, APXs areinactivated by the thiol reagent, p-chloromercuri-benzoate, because they contain free Cys residues,whereas archetypal GPXs (e.g. horseradish peroxi-dases) contain four conserved disulfide bridges. Theuse of such inhibitors is the basis for an assay todiscriminate between GPXs and APXs (Amako et al.,1994). The two types of peroxidases share little ho-mology, with the exception of the heme-binding do-main, and the corresponding antibodies do not cross-react. The differences are also important at the genelevel. The number and position of introns of GPXs(three or less) are very different from those of APXs(Fig. 1).

GSH-PXs are class I peroxidases that catalyze thereduction of H2O2, organic hydroperoxides, and lipidhydroperoxides to water by GSH. Once thought to bepresent only in animals and bacteria, it now seemsthat this enzyme is also present in plants. The firstevidence came from citrus plants, which were found

Table II. Antioxidant proteins of legume nodules

Enzyme Biochemical Properties

CuZnSOD In cytosol and plastids. Dimer (32 kD, 2 Cu, 2 Zn). Inhibited by KCN and H2O2.MnSOD In mitochondria and bacteroids. The plant enzyme is a tetramer (82 kD, 4 Mn). The bacterial

enzyme is a dimer (43 kD, 2 Mn) and may be cambialistic. Resistant to KCN and H2O2.FeSOD In plastids and cytosol. Dimer (56–58 kD, 2 Fe). Structurally related to MnSODs. Inhibited by

H2O2 but resistant to KCN.Catalase In peroxisomes and bacteroids. The plant enzyme is a tetramer (220 kD, 4 heme). Inhibited by

KCN and aminotriazole. The bacterial enzymes have a subunit size of 63 kD.APX Mainly in cytosol (0.9% of total soluble nodule protein). Dimer (subunits of 27 kD, 2 heme). They

are inactivated by p-chloromercuribenzoate and strongly inhibited by KCN. The cytosolic isoformis distinguished from chloroplastic isoforms by its insensitivity to ascorbate depletion. All iso-forms use ascorbate effectively as a reductant, in contrast to GPXs that do not. Km � 300 �M forascorbate and 20 �M for H2O2. Membrane-bound isoforms exist inmitochondria and possibly in peroxisomes.

GPX In vacuoles and cell walls. They have four disulfide bridges and structural Ca2� ions, and most ofthem are glycosylated. They use phenolics as substrates and are not inactivated byp-chloromercuribenzoate. Strongly inhibited by KCN.

GSH-PXa In plastids and cytosol. Probably, monomers (20–25 kD). Non-selenium enzymes.Monodehydroascorbate reductase Mainly in cell wall, also in mitochondria and possibly in peroxisomes and plasma membrane.

Monomer (40 kD, 1 FAD, active thiol groups). Two isoforms. Km � 6 to 7 �M for NADH andmonodehydroascorbate.

Dehydroascorbate reductaseb In cytosol. Monomer (23 kD, active thiol groups). Km � 390 �M for dehydroascorbate and 3.5 mM

for GSH. Labile in the absence of thiol protectant.Glutathione reductase In cytosol, plastids, and mitochondria. Probably, a tetramer (135–190 kD; subunits of 32–60 kD).

Km � 23 �M for GSSG and NADPH.�-Glutamylcysteine synthetasea In plastids. Biochemical data suggest it is a dimer (58–60 kD). Km � 70 to 190 �M for Cys and 4

to 10 mM for Glu. Very labile enzyme. Strict requirement for ATP, Mg2� and K�. Inhibited invitro and in vivo by buthionine sulfoximine. Feedback inhibited in vitro by GSH.

(Homo)glutathione synthetasesc In cytosol and plastids. Dimer (113–120 kD; subunits of 56–61 kD). Strict requirement for ATP andMg2�. Km � 20 to 70 �M for �Glu-Cys and 0.2 to 1 mM for Gly (glutathione synthetase). Km �1.9 mM for �Ala (homoglutathione synthetase).

Ferritin In plastids. Multimeric protein (550–600 kD; subunits of 23–28 kD). At least three isoproteins.a Molecular mass predicted from cDNA sequences. b No data available for the nodule enzyme. Data for the cytosolic enzyme of potato

tubers (Dipierro and Borraccino, 1991). c Dimer molecular mass for higher plant enzymes (Rennenberg, 1997). Subunit molecular masspredicted from cDNA sequences of the pea and bean nodule enzymes.

Matamoros et al.

502 Plant Physiol. Vol. 133, 2003

to contain phospholipid hydroperoxide GSH-PX ac-tivity (Beeor-Tzahar et al., 1995). This activity wasclearly different from glutathione S-transferases(some isoforms of which may also exhibit GSH-PXactivity) and was induced by salt stress. Since then,several cDNA clones encoding homologous enzymeshave been isolated in pea and other higher plants(Mullineaux et al., 1998). All of them are predicted tocontain Cys (UGU or UGC codons) at their active siteinstead of the rare seleno-Cys residue (UGA codon)found in mammalian phospholipid hydroperoxideGSH-PXs. Many plant GSH-PXs reported so far arepredicted to be located in the plastids; however, pu-tative cytosolic and peroxisomal isoforms werefound in barley (Hordeum vulgare; Churin et al., 1999).We also have isolated several cDNA and genomicclones encoding GSH-PXs that are expressed in nod-ules of L. japonicus (Fig. 1). The deduced proteinscontain the three conserved signatures (domains)found in animal and plant GSH-PXs and lack seleno-Cys. Prediction programs of subcellular localizationindicate that there are cytosol and plastid isoforms.Two genes have been completely sequenced and

found to comprise six exons of almost identical sizes(except for the first exon) but different intron sizes.Therefore, gene structures allow for a clear separa-tion among the three types of nodule peroxidases(Fig. 1).

THE ASCORBATE-GLUTATHIONE PATHWAY ISCRITICAL FOR NODULE FUNCTIONING

The initial product of APX is monodehydroascor-bate (ascorbate free radical), which then dispropor-tionates to ascorbate and dehydroascorbate. Mono-dehydroascorbate and dehydroascorbate are reducedback to ascorbate by specific reductases using NADHand GSH, respectively. Finally, the GSSG formed bydehydroascorbate reductase is reduced to GSH byglutathione reductase using NADPH. Therefore, theascorbate-GSH pathway involves four enzymes op-erating in concert to remove H2O2 at the expense,ultimately, of the reducing power of NADH orNADPH (Fig. 2).

APX was described above; hence, we will focusnow on the other enzymes of the pathway (Fig. 2).

Figure 2. Antioxidant enzymes of legume nodules. ASC, Ascorbate; CAT, catalase; CuZnSODc, cytosol CuZnSOD;CuZnSODp, plastid CuZnSOD; DHA, dehydroascorbate; DR, dehydroascorbate reductase; �EC, �Glu-Cys; �ECS, �-glu-tamylcysteine synthetase; ETC, electron transport chain; GL, L-galactono-�-lactone; GLDH, L-galactono-��lactonedehydrogenase; GR, glutathione reductase; (h)GSH, (homo)glutathione, reduced form; (h)GSHS, (homo)glutathionesynthetase; (h)GSSG, (homo)glutathione, oxidized form; Lb, leghemoglobin; MDHA, monodehydroascorbate; MR, mono-dehydroascorbate reductase; Ox met, oxidative metabolism.

Antioxidants in the Rhizobia-Legume Symbiosis

Plant Physiol. Vol. 133, 2003 503

Glutathione reductases are found in the cytosol, mi-tochondria, and bacteroids, but are probably presentalso in nodule plastids because the enzyme is abun-dant in chloroplasts and root plastids (Bielawski andJoy, 1986) and because cDNA clones encoding a pu-tative plastid glutathione reductase have been iso-lated (Tang and Webb, 1994). It is likely that theenzymes of nodule mitochondria and plastids arecoded for by a single gene, as occurs with the en-zymes of pea leaves (Creissen et al., 1995). Somenodules, such as those of bean and soybean, synthe-size homoglutathione rather than GSH; therefore, theenzyme is functionally a homoglutathione reductase.Monodehydroascorbate reductases are ubiquitousflavoproteins of plants that occur in soluble andmembrane-bound isoforms. In nodules, two isoformshave been found, and at least one of them is associ-ated with the cell wall. Only low enzyme levels aredetected in the cytosol (Dalton et al., 1993). Thus, itappears that recycling of ascorbate through theascorbate-GSH pathway in the cytosol is mainly ac-complished by dehydroascorbate reductase, whereasmonodehydroascorbate reductase may be involvedin regeneration of apoplastic ascorbate, synthesis of

Hyp-rich proteins, and lignification of cell walls.Very little is known about this enzyme in plants and,in particular, in nodules (Table II). Dehydroascorbatereductase is a monomeric protein with active thiolgroups and has been localized to the cytosol andmitochondria of nodule host cells (Dalton et al., 1993;Iturbe-Ormaetxe et al., 2001). The ascorbate-GSHpathway seems to be operative in nodule compart-ments other than the cytosol. The four enzymes of thepathway have been detected in nodule mitochondria(Dalton et al., 1993; Iturbe-Ormaetxe et al., 2001). Amodel has been proposed (Iturbe-Ormaetxe et al.,2001) for bean nodule mitochondria, in which H2O2generated in the inner membrane is removed bymembrane-bound APX, and the resulting ascorbateoxidation products are regenerated to ascorbate bymonodehydroascorbate, dehydroascorbate, and ho-moglutathione reductases in the matrix or the cytosol(Fig. 2). The enzymes of the ascorbate-GSH pathwayhave also been found in pea leaf peroxisomes(Jimenez et al., 1997); thus, the pathway is probablyfunctional in nodule peroxisomes.

Several lines of evidence show that the ascorbate-GSH pathway is critical for nodule functioning (Dal-

Figure 3. Localization of APX, SODs, and H2O2

in alfalfa nodules. A, Immunofluorescence local-ization of cytosolic APX. High levels are evidentin the central, infected region (arrowhead) and ina ring of cells in the nodule parenchyma (arrow;from Dalton et al., 1998). B, Tetrazolium stainingof respiratory dehydrogenase activity. Activity isenhanced in the nodule parenchyma (arrow), in-dicating increased respiration associated with theoxygen diffusion barrier and probably with theenhanced level of APX protein shown in A (fromDalton et al., 1998). C, In situ hybridization ofcytosolic CuZnSOD mRNA. Transcript is mostabundant in the nodule apex (arrow), which in-clude the meristem and invasion zones. D, In situhybridization of MnSOD mRNA. Transcript ismost abundant in the infected region, and espe-cially in the infected cells (arrow). E and F, Lo-calization of H2O2. Fresh nodule tissue was per-fused with cerium chloride and processed forelectron microscopy. The presence of H2O2 ismarked by the deposition of cerium perhydroxideprecipitates, which can be seen in the walls andmatrix of infection threads (arrows in E) and in thecell walls and intercellular spaces of the cortex(arrows in F). Note that H2O2 can be also ob-served surrounding the bacteria within thethreads (arrowhead in E).

Matamoros et al.

504 Plant Physiol. Vol. 133, 2003

ton, 1995; Dalton et al., 1998). The activity, protein,and transcript of APX, the key enzyme of the pathway,are very abundant in nodules, particularly in the in-fected and parenchyma cells (Fig. 3A). In the infectedcells, APX protects leghemoglobin and other redox-sensitive proteins from H2O2, whereas, in the noduleparenchyma (a few cell layers outside the infectedzone), the enzyme may participate in the operation ofthe oxygen diffusion barrier. This barrier has beenproposed to be located, for the most part, in the nod-ule parenchyma and controls oxygen entry into theinfected zone. The parenchyma cells have not onlyhigh levels of APX but also of ascorbate and respira-tory dehydrogenases (Fig. 3B). Thus, we proposed thatthe parenchyma cells would regulate oxygen access tothe infected region by adjusting their respiratory ac-tivity (Dalton et al., 1998). The concentration of theresulting H2O2 then would be finely tuned by APX,allowing for H2O2 to act as a signal molecule for the“opening” or “closure” of the oxygen diffusion barrier(Minchin, 1997; Dalton et al., 1998).

There are further indications of the importance ofthe ascorbate-GSH pathway for N2 fixation. The ac-tivities of all four enzymes are much higher (2- to36-fold) in nodules than in uninfected roots. Theenzyme activities and thiol contents are also substan-tially higher (1.5- to 5.5-fold) in effective than inineffective nodules. Also, treatment of plants withfixed nitrogen (urea) inhibits N2 fixation concomi-tantly with three enzyme activities of the pathway,indicating that there is a link between N2 fixation andantioxidant defenses. The most compelling evidencefor the connection between antioxidants and N2 fix-ation comes from observations that direct infusion ofascorbate into stems of soybean plants leads to anincrease in leghemoglobin content, a 4-fold increasein rates of N2 fixation, and a substantial delay innodule senescence (Bashor and Dalton, 1999). Inclu-sion of ascorbate and purified recombinant APX inan in vitro reconstitution system containing leghemo-globin and bacteroids results in improved oxygen-ation of leghemoglobin and up to a 4.5-fold increasein N2 fixation (Ross et al., 1999). Collectively, theseand other observations have confirmed that antioxi-dants play an important role in protecting and en-hancing N2 fixation.

SODs AND CATALASES ARE CRITICAL FORPROTECTION OF NITROGEN FIXATION ANDOCCUR IN BOTH SYMBIOTIC PARTNERS

SODs are a family of metalloenzymes that catalyzethe dismutation of superoxide radicals into molecu-lar oxygen and H2O2. Three classes of SODs, differ-ing in their metals at the active site, may coexist inplants, and all of them have been found in the noduleplant fraction. The subcellular localizations and bio-chemical properties of the CuZnSOD, FeSOD, andMnSOD of nodules are presented in Table II. Re-

cently, the proteins and transcripts of cytosolicCuZnSOD and mitochondrial MnSOD have been lo-calized in alfalfa and pea nodules (M.C. Rubio, E.K.James, M.R. Clemente, B. Bucciarelli, C.P. Vance, andM. Becana, unpublished data). The CuZnSOD is pre-dominant in the nodule apex (Fig. 3C), especially inthe infection threads, cytosol adjacent to cell walls,and apoplast; the MnSOD is abundant in the infectedzone, especially in the infected cells (Fig. 3D). Anadditional CuZnSOD isozyme, the plastid CuZnSOD,is localized to the amyloplasts, whereas MnSOD isalso found in the bacteroids and bacteria within in-fection threads. The distinct tissue localizations of“cytosolic” CuZnSOD and MnSOD suggest specificfunctions for the two enzymes. The CuZnSOD maybe associated with cell wall growth in the meristems,infection threads, and apoplast, and with the plant’sresponse to bacterial infection. The MnSOD wouldplay a role related to the protection and functioningof symbiotic tissue in mature nodules.

The structures of the genes encoding cytosolicCuZnSOD (sodCc) and mitochondrial MnSOD (sodA)of L. japonicus have been determined. The sodCc geneconsists of eight exons; interestingly, the first intronis in the 5�-UTR, as occurs for the pea apx1 gene. ThesodA gene has six exons with no apparent specialfeatures. The FeSODs are the most enigmatic class ofSODs; in fact, the corresponding gene (sodB) wasonce thought to be present, or expressed, only in afew families of higher plants. The FeSODs, whenpresent, appear to be localized exclusively in thechloroplast stroma. We have found FeSODs in nod-ules of most legumes examined and isolated cDNAsfor some species. Two types of FeSOD were clearlyrecognized: the typical FeSOD localized in the plas-tids of alfalfa and pea nodules and an unusualFeSOD localized in the cytosol of cowpea nodules(Moran et al., 2003).

Bacteroids possess a MnSOD in the cytoplasm anda CuZnSOD in the periplasmic space. These enzymesare encoded by the respective bacterial sodA and sodCgenes. The MnSOD of S. meliloti shows high aminoacid sequence similarity with bacterial FeSODs and isa “cambialistic” enzyme; in other words, it remainsactive (though less so) when the Mn is replaced by Fe(Santos et al., 1999). Interestingly, the enzyme is re-sistant to H2O2 regardless of the metal at the activesite. The sodA� mutant of S. meliloti fails to differen-tiate into bacteroids and nodulates poorly (Santos etal., 2000). The CuZnSOD of S. meliloti is expressedduring infection (Ampe et al., 2003), perhaps as aresponse of the bacteria to the superoxide radicalsproduced by the plant (see below).

Catalases decompose H2O2 to water and molecularoxygen without consuming reductants and, thus,may provide plant cells with an energy-efficientmechanism to remove H2O2 (Scandalios et al., 1997).However, catalases have a much lower affinity forH2O2 than APXs and are expected to be active only at

Antioxidants in the Rhizobia-Legume Symbiosis

Plant Physiol. Vol. 133, 2003 505

subcellular sites where H2O2 or catalase concentra-tions are very high, such as the peroxisomes. Catalasefrom nodule peroxisomes was purified and appearsto be similar to other higher plant enzymes. It isunclear if a specific catalase isoform occurs in nodulemitochondria, as was observed for maize (Zea mays)leaves (Scandalios et al., 1997). The production ofoxygen by catalase would appear to be unfavorablein the case of bacteroids because nitrogenase andother proteins with high redox potential are readilyinactivated by excess oxygen. Therefore, it is inter-esting that bacteroids do not contain peroxidases buthave catalases instead. Free-living S. meliloti andother rhizobia have three catalases: two monofunc-tional catalases (KatA and KatC) and one bifunc-tional catalase-peroxidase (KatB). Only KatA is in-duced by H2O2 and highly expressed in bacteroids,whereas KatB and KatC are expressed by the bacteriawithin the infection threads (Jamet et al., 2003). Dou-ble mutants (katAkatC or katBkatC) are severely im-paired in N2 fixation, whereas the single mutantsdisplay no effect. Overall, these results demonstratethe essentiality of catalases during the infection pro-cess (Jamet et al., 2003).

NODULES ALSO HAVE IMPORTANTANTIOXIDANT DEFENSES AGAINST MEMBRANEDAMAGE AND IRON TOXICITY

Several antioxidant enzymes are bound to plantmembranes, such as the APX, monodehydroascor-bate reductase, and MnSOD of peroxisomes (Corpaset al., 2001) and the APX of mitochondria (Iturbe-Ormaetxe et al., 2001). These enzymes protect mem-branes from ROS but may have additional usefulroles. Peroxisomal monodehydroascorbate reductasecan generate superoxide (a common characteristicwith the chloroplastic enzyme), which may then beused by the plant as a signal molecule. Plant mem-branes are mainly protected against lipid peroxida-tion and other types of oxidative damage by smalllipophilic molecules such as tocopherols, ubiquinol,lipoic acid, and flavonoids. �-Tocopherol is found atconcentrations of 15 �g g�1 in both young and oldsoybean nodules (Evans et al., 1999). Ubiquinol (thereduced form of ubiquinone) and lipoic acid areabundant in membranes of mitochondria and otherorganelles, where they act as potent inhibitors oflipid peroxidation. However, they have not beenquantified in nodules. Flavonoids and other polyphe-nols are found in nodules at concentrations of 0.4 to4 mm. Some of these compounds have importantantioxidant properties, protecting membranes byneutralizing lipid radicals (Moran et al., 1997). Poly-amines are also abundant in nodules and inhibit lipidperoxidation in vitro, probably by their ability toassociate with phospholipids and stabilize mem-branes (Fujihara et al., 1994).

Plant cells also have an adequate protection againstiron-mediated toxicity. Iron in the free form or bound

to small chelators is potentially toxic because it cancatalyze formation of hydroxyl radicals. The excep-tions seem to be phytic acid and certain phenoliccompounds that are able to chelate iron in a catalyt-ically inactive form and may inhibit oxidative dam-age of lipids and proteins in vitro (Moran et al., 1997).The supply of free iron must be tightly regulatedbecause plants require a steady, low iron supply forthe synthesis of iron-proteins, DNA, and some hor-mones (Briat and Lobreaux, 1997). This need must becarefully balanced against the potential toxicity ofexcess iron. The protein ferritin stores up to 4,500atoms of iron in a form that avoids the deleteriouseffects of iron while controlling its availability formetabolic purposes. Plant ferritins are composed of acentral iron-filled cavity surrounded by a shell of 24identical subunits. Nodules have an active iron me-tabolism and abundant ferritin, which is localized inplastids and amyloplasts, much like the ferritin ofleaves that is localized exclusively in the chloroplasts(Lucas et al., 1998; Matamoros et al., 1999). The fer-ritin protein and transcript increase very early innodulation. Later in nodule development, the ferritinprotein (but not its transcript) declines concomitantlywith the increase in nitrogenase and leghemoglobinproteins. This suggests that ferritin is a reservoir ofiron and supplies it for nitrogenase and leghemoglo-bin synthesis. Because changes in ferritin protein andtranscript are not coordinated, ferritin expressionmay be posttranscriptionally regulated (Ragland andTheil, 1993). The ferritin content increases duringnatural and stress-induced senescence of nodules,although variations can be found depending on le-gume species and nodule tissue. This is most proba-bly due to induction of ferritin expression by the ironreleased during degradation of leghemoglobin andother iron proteins (Lucas et al., 1998; Matamoros etal., 1999).

ROS AND RNS ARE INVOLVED IN NODULEFORMATION AND SENESCENCE

Plants respond defensively to pathogen infectionwith a hypersensitive reaction, an early feature ofwhich is the rapid and transient production of ROS(“oxidative burst”) (Lamb and Dixon, 1997). Infectionof legume roots by rhizobia also elicits a hypersensi-tive reaction. After the first nodule primordia havebeen induced, an increasing proportion of infectionthreads abort in a few cortical cells in which bothrhizobia and host cells undergo necrosis. The hyper-sensitive reaction may be part of a mechanismwhereby the plant controls infection and, thus, reg-ulates nodulation (Vasse et al., 1993). As in the case ofattack by pathogens, root cells respond to rhizobialinfection with an enhanced production of superoxideand H2O2 (Santos et al., 2001; D’Haeze et al., 2002;Ramu et al., 2002). It has not been definitively proventhat this is a genuine oxidative burst, but the finding

Matamoros et al.

506 Plant Physiol. Vol. 133, 2003

that H2O2 accumulation is restricted to the very earlystages of nodule formation in Sesbania rostrata (E.K.James, unpublished data) supports this hypothesis.Interestingly, one of the genes more rapidly inducedby compatible rhizobia or Nod factors (rip1) seems toencode a peroxidase and has cis-elements in its pro-moter region that may be responsive to ROS. Becauseexogenous H2O2 is sufficient to activate rip1 tran-scription in the absence of Nod factors, ROS may actdownstream in the signal transduction pathway(Ramu et al., 2002). In this respect, Ramu et al. (2002)and D’Haeze et al. (2002) have concluded that Nodfactor-induced nodulation requires H2O2.

Most likely, the “early” production of H2O2 is partof an oxidative burst, but, in later stages of noduleformation, H2O2 accumulation may be more relatedto cell wall formation and cross-linking of glycopro-teins, both of which are required for successful infec-tion. However, an as yet unsolved question is whysome rhizobia have success during infection andform functional nodules. It is thought that, duringinfection, rhizobia may escape or inhibit the defen-sive response. This inhibition has been attributed tothe bacterial exopolysaccarides (Gonzalez et al.,1996). The enzymes responsible for enhanced ROSformation during infection and nodule organogene-sis have not been identified definitively. The super-oxide radicals are formed in the infection threads(Santos et al., 2001), possibly by a membrane-boundNADPH-oxidase, much like the superoxide genera-tion during the oxidative burst in activated neutro-phils. Possible sources for H2O2 are cell wall peroxi-dases, germin-like oxalate oxidases, and diamineoxidases (Wisniewski et al., 2000). We have foundthat H2O2 accumulates in the walls and lumen ofinfection threads, surrounding bacteria within thethreads, and in the apoplast of the nodule cortex (Fig.3D). Based mainly on colocalization studies, we pro-pose that CuZnSOD is a potential source of H2O2.This peroxide may be important for the cross-linkingof cell wall proteins in the apoplast and of the matrixglycoprotein in the infection threads (Wisniewski etal., 2000).

Additional signal molecules may be important fornodule formation. Salicylic acid may be implicated inthe early stages of infection because compatible Nodfactors inhibit the accumulation of salicylic acid (adefensive response) in the root (Martınez-Abarca etal., 1998). Nitric oxide could be another signal mole-cule because both nitric oxide synthase-like activity(Cueto et al., 1996) and nitric oxide (Mathieu et al.,1998) have been detected in nodules. Recently, Cor-pas et al. (2001) have found nitric oxide synthaseactivity and its product in pea leaf peroxisomes.However, Klessig’s group has identified the induc-ible nitric oxide-producing enzyme as a variant of theP protein of the Gly decarboxylase complex (Chan-dok et al., 2003). It will be of great interest to deter-mine if nitric oxide is produced in the specialized

nodule peroxisomes and to identify its origin. In anycase, it is clear that the topic of RNS in nodules is anemerging area of study.

There is a second period in the lifetime of nodulescharacterized by an enhanced production of ROS andprobably RNS. Large amounts of H2O2 accumulate inthe cells and apoplast in the central zone of senescingsoybean nodules (Alesandrini et al., 2003) and sur-rounding bacteroids in the senescent zone of alfalfaand pea nodules (E.K. James, M.C. Rubio, and M.Becana, unpublished). In the senescing nodule tissue,there is a major decrease in antioxidant defenses,oxidative degradation of leghemoglobin to nonfunc-tional green pigments, and enhanced autolytic pro-cesses (Mellor, 1989; Matamoros et al., 1999). Theseare all situations conducive to uncontrolled ROS andRNS production. As a consequence, oxidative dam-age of lipids, proteins, and DNA has been observedin nodules during natural (Evans et al., 1999) andstress-induced (Becana and Klucas, 1992; Matamoroset al., 1999) senescence. Similarly, the structuralbreakdown of organelles, symbiosomes, and bacte-roids in the host cells usually accompanies senes-cence. All these structural and biochemical alter-ations may be interpreted in terms of a switch from areductive to an oxidative state, which may be a gen-eral characteristic of plant senescence (Swaraj andBishnoi, 1996).

ACKNOWLEDGMENTS

We are most grateful to Carmen Perez-Rontome for her excellent job withfigures and Euan James for helpful comments. Thanks are also due to allcollaborators and colleagues who have contributed, directly or indirectly, tothe information provided in this Update. We apologize to those colleagueswhose work we were unable to cite because of space limitations.

Received April 16, 2003; returned for revision June 9, 2003; accepted July 15,2003.

LITERATURE CITED

Alesandrini F, Mathis R, Van de Sype G, Herouart D, Puppo A (2003)Possible roles for a cysteine protease and hydrogen peroxide in soybeannodule development and senescence. New Phytol 158: 131–138

Amako K, Chen GX, Asada K (1994) Separate assays specific for ascorbateperoxidase and guaiacol peroxidase and for the chloroplastic and cyto-solic isozymes of ascorbate peroxidase in plants. Plant Cell Physiol 35:497–504

Ampe F, Kiss E, Sabourdy F, Batut J (2003) Transcriptome analysis ofSinorhizobium meliloti during symbiosis. Genome Biol 4: R15

Bashor CJ, Dalton DA (1999) Effects of exogenous application and steminfusion of ascorbate on soybean (Glycine max) root nodules. New Phytol142: 19–26

Becana M, Klucas RV (1992) Transition metals in legume root nodules:iron-dependent free radical production increases during nodule senes-cence. Proc Natl Acad Sci USA 89: 8958–8962

Beeor-Tzahar T, Ben-Hayyim G, Holland D, Faltin Z, Eshdat Y (1995) Astress-associated citrus protein is a distinct plant phospholipid hydroper-oxide glutathione peroxidase. FEBS Lett 366: 151–155

Bielawski W, Joy KW (1986) Reduced and oxidised glutathione andglutathione-reductase activity in tissues of Pisum sativum. Planta 169:267–272

Briat JF, Lobreaux S (1997) Iron transport and storage in plants. TrendsPlant Sci 2: 187–193

Antioxidants in the Rhizobia-Legume Symbiosis

Plant Physiol. Vol. 133, 2003 507

Chandok MR, Ytterberg AJ, van Wijk KJ, Klessig DF (2003) The pathogen-inducible nitric oxide synthase (iNOS) in plants is a variant of the Pprotein of the glycine decarboxylase complex. Cell 113: 469–482

Churin Y, Schilling S, Borner T (1999) A gene family encoding glutathioneperoxidase homologues in Hordeum vulgare (barley). FEBS Lett 459: 33–38

Corpas FJ, Barroso JB, del Rıo LA (2001) Peroxisomes as a source of reactiveoxygen species and nitric oxide signal molecules in plant cells. TrendsPlant Sci 6: 145–150

Creissen G, Reynolds H, Xue Y, Mullineaux P (1995) Simultaneous target-ing of pea glutathione reductase and of a bacterial fusion protein tochloroplasts and mitochondria in transgenic tobacco. Plant J 8: 167–175

Cueto M, Hernandez-Perera O, Martın R, Bentura ML, Rodrigo J, Lamas S,Golvano MP (1996) Presence of nitric oxide synthase activity in roots andnodules of Lupinus albus. FEBS Lett 398: 159–164

Dalton DA (1995) Antioxidant defenses of plants and fungi. In S Ahmad, ed,Oxidative Stress and Antioxidant Defenses in Biology. Chapman andHall, New York, pp 298–355

Dalton DA, Baird LM, Langeberg L, Taugher CY, Anyan WR, Vance CP,Sarath G (1993) Subcellular localization of oxygen defense enzymes insoybean (Glycine max [L.] Merr.) root nodules. Plant Physiol 102: 481–489

Dalton DA, Joyner SL, Becana M, Iturbe-Ormaetxe I, Chatfield JM (1998)Antioxidant defenses in the peripheral cell layers of legume root nodules.Plant Physiol 116: 37–43

D’Haeze W, Chaparro C, De Keyser A, Deleu S, De Rycke R, GoormachtigS, Lievens S, Mathis R, Schroeyers K, Van de Velde W et al. (2002) Roleof reactive oxygen species and ethylene in programmed cell death duringnodule initiation on Sesbania rostrata. In T Finan, M O’Brian, D Layzell, KVessey, W Newton, eds, Nitrogen Fixation: Global Perspectives. CABInternational, Oxon, UK, pp 137–141

Dipierro S, Borraccino G (1991) Dehydroascorbate reductase from potatotubers. Phytochemistry 30: 427–429

Evans PJ, Gallesi D, Mathieu C, Hernandez MJ, de Felipe M, Halliwell B,Puppo A (1999) Oxidative stress occurs during soybean nodule senes-cence. Planta 208: 73–79

Frendo P, Hernandez MJ, Mathieu C, Duret L, Gallesi D, Can de Sype G,Herouart D, Puppo A (2001) A Medicago truncatula homoglutathionesynthetase is derived from glutathione synthetase by gene duplication.Plant Physiol 126: 1706–1715

Fujihara S, Abe H, Minakawa Y, Akao S, Yoneyama T (1994) Polyaminesin nodules from various plant-microbe interactions. Plant Cell Physiol 35:1127–1134

Gonzalez JE, Reuhs BL, Walker GC (1996) Low molecular weight EPSII ofRhizobium meliloti allows nodule invasion in Medicago sativa. Proc NatlAcad Sci USA 93: 8636–8641

Halliwell B, Gutteridge JMC (1999) Free Radicals in Biology and Medicine,Ed 3. Oxford University Press, Oxford

Horemans N, Foyer CH, Potters G, Asard H (2000) Ascorbate function andassociated transport systems in plants. Plant Physiol Biochem 38: 531–540

Iturbe-Ormaetxe I, Matamoros MA, Rubio MC, Dalton DA, Becana M(2001) The antioxidants of legume nodule mitochondria. Mol Plant-Microbe Interact 14: 1189–1196

Jamet A, Sigaud S, Van de Sype G, Puppo A, Herouart D (2003) Expressionof the bacterial catalase genes during Sinorhizobium meliloti-Medicagosativa symbiosis and their crucial role during the infection process. MolPlant-Microbe Interact 16: 217–225

Jimenez A, Hernandez J, del Rıo LA, Sevilla F (1997) Evidence for thepresence of the ascorbate-glutathione cycle in mitochondria and peroxi-somes of pea leaves. Plant Physiol 114: 275–284

Lamattina L, Garcıa-Mata C, Graziano M, Pagnussat G (2003) Nitric oxide:the versatility of an extensive signal molecule. Annu Rev Plant Biol 54:109–136

Lamb C, Dixon RA (1997) The oxidative burst in plant disease resistance.Annu Rev Plant Physiol Plant Mol Biol 48: 251–275

Lucas MM, Van de Sype G, Herouart D, Hernandez MJ, Puppo A, deFelipe MR (1998) Immunolocalization of ferritin in determinate andindeterminate legume root nodules. Protoplasma 204: 61–70

Martınez-Abarca F, Herrera-Cervera JA, Bueno P, Sanjuan J, Bisseling T,Olivares J (1998) Involvement of salicylic acid in the establishment of theRhizobium meliloti-alfalfa symbiosis. Mol Plant-Microbe Interact 11:153–155

Matamoros MA, Baird LM, Escuredo PR, Dalton DA, Minchin FR, Iturbe-Ormaetxe I, Rubio MC, Moran JF, Gordon AJ, Becana M (1999) Stress-

induced legume root nodule senescence: physiological, biochemical, andstructural alterations. Plant Physiol 121: 97–111

Matamoros MA, Clemente MR, Sato S, Asamizu E, Tabata S, Ramos J,Moran JF, Stiller J, Gresshoff PM, Becana M (2003) Molecular analysisof the pathway for the synthesis of thiol tripeptides in the model legume,Lotus japonicus. Mol Plant-Microbe Interact 16: 1039–1046

Mathieu C, Moreau S, Frendo P, Puppo A, Davies MJ (1998) Direct detec-tion of radicals in intact soybean nodules: presence of nitric oxide-leghemoglobin complexes. Free Rad Biol Med 24: 1242–1249

May MJ, Vernoux T, Leaver C, Van Montagu M, Inze D (1998) Glutathionehomeostasis in plants: implications for environmental sensing and plantdevelopment. J Exp Bot 49: 649–667

Mellor RB (1989) Bacteroids in the Rhizobium-legume symbiosis inhabit aplant internal lytic compartment: implications for other microbial endo-symbioses. J Exp Bot 40: 831–839

Minchin FR (1997) Regulation of oxygen diffusion in legume nodules. SoilBiol Biochem 29: 881–888

Mittler R (2002) Oxidative stress, antioxidants and stress tolerance. TrendsPlant Sci 7: 405–410

Mittler R, Zilinskas BA (1992) Molecular cloning and characterization of agene encoding pea cytosolic ascorbate peroxidase. J Biol Chem 267:21802–21807

Moran JF, Iturbe-Ormaetxe I, Matamoros MA, Rubio MC, Clemente MR,Brewin NJ, Becana M (2000) Glutathione and homoglutathione syntheta-ses of legume nodules: cloning, expression, and subcellular localization.Plant Physiol 124: 1381–1392

Moran JF, James EK, Rubio MC, Sarath G, Klucas RV, Becana M (2003)Functional characterization and expression of a cytosolic iron-superoxidedismutase from cowpea (Vigna unguiculata) root nodules. Plant Physiol133: 773–782

Moran JF, Klucas RV, Grayer RJ, Abian J, Becana M (1997) Complexes ofiron with phenolic compounds from soybean nodules and other legumetissues: prooxidant and antioxidant properties. Free Radic Biol Med 22:861–870

Mullineaux PM, Karpinski S, Jimenez A, Cleary SP, Robinson C, CreissenGP (1998) Identification of cDNAs encoding plastid-targeted glutathioneperoxidase. Plant J 13: 375–379

Neill SJ, Desikan R, Clarke A, Hurst RD, Hancock JT (2002) Hydrogenperoxide and nitric oxide as signalling molecules in plants. J Exp Bot 53:1237–1247

Ragland M, Theil EC (1993) Ferritin (mRNA, protein) and iron concentra-tions during soybean nodule development. Plant Mol Biol 21: 555–560

Ramu SK, Peng HM, Cook DR (2002) Nod factor induction of reactiveoxygen species production is correlated with expression of the earlynodulin gene rip1 in Medicago truncatula. Mol Plant-Microbe Interact 15:522–528

Rennenberg H (1997) Molecular approaches to glutathione biosynthesis. InWJ Cram, LJ DeKok, C Brunold, H Rennenberg, eds, Sulphur Metabolismin Higher Plants. Backhuys Publishers, Leiden, The Netherlands, pp59–70

Riccillo PM, Muglia CI, de Bruijn FJ, Roe AJ, Booth IR, Aguilar OM (2000)Glutathione is involved in environmental stress responses in Rhizobiumtropici, including acid tolerance. J Bacteriol 182: 1748–1753

Ross EJH, Kramer SB, Dalton DA (1999) Effectiveness of ascorbate andascorbate peroxidase in promoting nitrogen fixation in model systems.Phytochemistry 52: 1203–1210

Santos R, Bocquet S, Puppo A, Touati D (1999) Characterization of anatypical superoxide dismutase from Sinorhizobium meliloti. J Bacteriol 181:4509–4516

Santos R, Herouart D, Puppo A, Touati D (2000) Critical protective role ofbacterial superoxide dismutase in Rhizobium-legume symbiosis. MolMicrobiol 38: 750–759

Santos R, Herouart D, Sigaud S, Touati D, Puppo A (2001) Oxidative burstin alfalfa-Sinorhizobium meliloti symbiotic interaction. Mol Plant-MicrobeInteract 14: 86–89

Scandalios JG, Guan L, Polidoros AN (1997) Catalases in plants: genestructure, properties, regulation, and expression. In JG Scandalios, ed,Oxidative Stress and the Molecular Biology of Antioxidant Defenses.Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp343–406

Matamoros et al.

508 Plant Physiol. Vol. 133, 2003

Shigeoka S, Ishikawa T, Tamoi M, Miyagawa Y, Takeda T, Yabuta Y,Yoshimura K (2002) Regulation and function of ascorbate peroxidaseisoenzymes. J Exp Bot 53: 1305–1319

Swaraj K, Bishnoi NR (1996) Physiological and biochemical basis of nodulesenescence in legumes: a review. Plant Physiol Biochem 23: 105–116

Tang X, Webb MA (1994) Soybean root nodule cDNA encoding glutathionereductase. Plant Physiol 104: 1081–1082

Udvardi MK (2001) Legume models strut their stuff. Mol Plant-MicrobeInteract 14: 6–9

Vasse J, de Billy F, Truchet G (1993) Abortion of infection during theRhizobium meliloti-alfalfa symbiotic interaction is accompanied by a hy-persensitive reaction. Plant J 4: 555–566

Wheeler GL, Jones MA, Smirnoff N (1998) The biosynthesis pathway ofvitamin C in higher plants. Nature 393: 365–369

Wisniewski JP, Rathbun EA, Knox JP, Brewin NJ (2000) Involvement ofdiamine oxidase and peroxidase in insolubilization of the extracellularmatrix: implications for pea nodule initiation by Rhizobium leguminosa-rum. Mol Plant-Microbe Interact 13: 413–420

Antioxidants in the Rhizobia-Legume Symbiosis

Plant Physiol. Vol. 133, 2003 509