The glutathione peroxidase gene family of Lotus japonicus: characterization of genomic clones, expression analyses and immunolocalization in legumes


Author for correspondence:
Manuel Becana
Tel:+34 976716055
Fax: +34 976716145


  • • Despite the multiple roles played by antioxidants in rhizobia–legume symbioses, little is known about glutathione peroxidases (GPXs) in legumes. Here the characterization of six GPX genes of Lotus japonicus is reported.
  • • Expression of GPX genes was analysed by quantitative reverse transcription–polymerase chain reaction in L. japonicus and Lotus corniculatus plants exposed to various treatments known to generate reactive oxygen and/or nitrogen species.
  • • LjGPX1 and LjGPX3 were the most abundantly expressed genes in leaves, roots and nodules. Compared with roots, LjGPX1 and LjGPX6 were highly expressed in leaves and LjGPX3 and LjGPX6 in nodules. In roots, salinity decreased GPX4 expression, aluminium decreased expression of the six genes, and cadmium caused up-regulation of GPX3, GPX4 and GPX5 after 1 h and down-regulation of GPX1, GPX2, GPX4 and GPX6 after 3–24 h. Exposure of roots to sodium nitroprusside (a nitric oxide donor) for 1 h increased the mRNA levels of GPX4 and GPX6 by 3.3- and 30-fold, respectively. Thereafter, the GPX6 mRNA level remained consistently higher than that of the control. Immunogold labelling revealed the presence of GPX proteins in root and nodule amyloplasts and in leaf chloroplasts of L. japonicus and other legumes. Labelling was associated with starch grains.
  • • These results underscore the differential regulation of GPX expression in response to cadmium, aluminium and nitric oxide, and strongly support a role for GPX6 and possibly other GPX genes in stress and/or metabolic signalling.


In plants, reactive oxygen species (ROS), such as the superoxide radical and H2O2, are formed mainly in the chloroplasts, mitochondria, peroxisomes and apoplast (del Río et al., 2002; Mittler, 2002). These ROS are potentially toxic when produced at high rates and can give rise to highly oxidizing hydroxyl radicals through Fenton reactions (Halliwell & Gutteridge, 2007). Similarly, reactive nitrogen species (RNS), such as nitric oxide (NO) and peroxynitrite (ONOO), are produced in different subcellular compartments of plants (del Río et al., 2002; Lamattina et al., 2003; Valderrama et al., 2007). The reaction between NO and superoxide or between nitrite and H2O2 can generate peroxynitrite, which causes oxidation and nitration of proteins and DNA (Halliwell & Gutteridge, 2007). However, at controlled, low steady-state concentrations, ROS and RNS fulfil essential functions in growth and development and in redox signalling during the stress responses of plants (Mittler, 2002; Lamattina et al., 2003; Foyer & Noctor, 2005). To keep cellular concentrations of ROS and RNS under tight control, plants contain antioxidant enzymes and metabolites at variable amounts in different tissues, cells and organelles.

Glutathione peroxidases (GPXs) catalyse the reduction of lipid peroxides and other organic peroxides to the corresponding alcohols using thioredoxins as the preferred electron donors (Herbette et al., 2002; Navrot et al., 2006). In mammals, there exist five distinct groups of GPXs that differ in structure, substrate specificity and subcellular localization (Ursini et al., 1995). Much less is known about plant GPXs. They are most similar in terms of amino acid sequences to the mammalian GPX4 type of enzymes, which comprises the selenium-dependent phospholipid hydroperoxide GPXs. Recently, a GPX from the green alga Chlamydomonas reinhardtii has been characterized and shown to contain a selenocysteine residue at the catalytic site (Fu et al., 2002), but all GPXs from vascular plants so far identified have a cysteine residue instead (Navrot et al., 2006). The major function of GPXs in plants appears to be the scavenging of phospholipid hydroperoxides and thereby the protection of cell membranes from peroxidative damage (Gueta-Dahan et al., 1997). Recent data showed that some GPXs may also be involved in redox transduction under stressful conditions (Miao et al., 2006). Consistent with these two functions, the expression of many GPXs is enhanced in response to abiotic and biotic stresses, including salinity, heavy metal toxicity and infection with bacterial or viral pathogens (Avsian-Kretchmer et al., 2004; and references therein).

Several cDNA clones that encode GPXs have been isolated from diverse plants of agronomic interest, such as citrus (Citrus sinensis; Holland et al., 1993), pea (Pisum sativum; Mullineaux et al., 1998), barley (Hordeum vulgare; Churin et al., 1999) and tomato (Lycopersicon esculentum; Herbette et al., 2002). However, to date, GPX genes have been studied comprehensively only in two plant species, thale cress (Arabidopsis thaliana; Rodríguez Milla et al., 2003) and poplar (Populus trichocarpa; Navrot et al., 2006). Abundant information in the databases is also available for rice (Oryza sativa), whose genome has been completely sequenced. Comparable studies of GPX genes have not been performed for any leguminous plant, despite the multiple roles that antioxidants play in the rhizobia–legume symbiosis (Dalton, 1995; Matamoros et al., 2003; Puppo et al., 2003). Two legume species, Lotus japonicus and Medicago truncatula, are currently used as models for classical and molecular genetics. In this work, we have identified six GPX genes in L. japonicus, determined their complete structures and promoter sequences, and quantified their expression levels in leaves, roots and nodules. To gain further insights into the regulation of LjGPX genes, plants of L. japonicus and birdsfoot trefoil (Lotus corniculatus), a related species of agronomic interest, were subjected to treatments known to generate ROS (Sugimoto & Sakamoto, 1997; Avsian-Kretchmer et al., 2004; Romero-Puertas et al., 2004; Sharma & Dubey, 2007) and/or RNS (Bethke et al., 2006; Valderrama et al., 2007). Thus, GPX gene expression was investigated in roots of plants treated with salt (NaCl), cadmium (Cd), aluminium (Al) or the NO donor sodium nitroprusside (SNP). Finally, we have immunolocalized the GPX protein(s) in L. japonicus and other legumes.

Materials and Methods

Biological material and plant treatments

For studies with nodulated plants (expression analyses in leaves, roots and nodules), seeds of Lotus japonicus (Regel) Larsen cv. MG20 were scarified, surface disinfected and germinated in petri dishes on filter paper for 2 d at 4°C and 1 d at 22°C in the dark and 2 d at 22°C in the light. Seedlings were then transferred to vermiculite-containing pots, nodulated with Mesorhizobium loti strain R7A and grown under controlled environment conditions (24 : 18°C (day:night), 180 µmol m−2 s−1, 16-h photoperiod). Plants were irrigated twice a week with B&D nutrient solution (Broughton & Dilworth, 1971), supplemented with 0.25 mM NH4NO3, until they were 42 d old (Rubio et al., 2007).

For studies with nonnodulated plants (expression analyses in roots of stressed plants), we found that growth of plants in hydroponic cultures was critical for strict control of the very short treatments. To this end, seeds of L. japonicus cv. MG20 and Lotus corniculatus L. cv. Draco were germinated in petri dishes as indicated above, and seedlings were transferred to 10 × 10 cm plates (eight seedlings per plate) containing 50 ml of a modified Fahraeus medium (Boisson-Dernier et al., 2001) with 15 g l−1 of agar for 7 d (same controlled conditions as before) to allow adequate root growth. Finally, plants were transferred to hydroponic solutions, grown under the same controlled conditions as the nodulated plants (except that the NH4NO3 concentration was 1.25 mM) and treated as follows: (a) L. corniculatus plants were grown on solution A (water plus 200 µM CaCl2, at pH 6.0) for 7 d and were treated with 20 µM CdCl2 for 1–24 h; (b) L. corniculatus plants were grown on solution B ( water plus 200 µM CaCl2, at pH 4.0) for 7 d and were treated with 20 µM AlCl3 for 1–24 h; and (c) L. japonicus plants were grown on solution C (one-fourth strength B&D nutrient solution, at pH 6.0) for 10 d and were treated with 150 mM NaCl, 0.1 mM SNP or 0.1 mM potassium ferricyanide for 1–24 h. Solutions were continuously aerated and replaced every 2 d. We used solutions A and B to avoid interactions of the nutrient salts with Cd and Al, and the pH of solution B was adjusted to 4.0 to avoid Al precipitation (Sugimoto & Sakamoto, 1997). Plants had no apparent symptoms of stress induced by metals or acid pH, with the exception of an increased proliferation of lateral roots at low pH.

For immunolocalization studies, two additional legume species were included. Alfalfa (Medicago sativa cv. Aragón), which produces indeterminate root nodules, and Sesbania rostrata, which produces both root and stem (photosynthetically active) nodules, were grown under controlled conditions as described by Rubio et al. (2004) and James et al. (1996), respectively. Nodules from these two legumes were harvested 30–35 d after inoculation.

Plant material to be used for determination of mRNA or protein levels was flash-frozen in liquid nitrogen and stored at −80°C. Plant material to be used for electron microscopy (EM) studies was harvested fresh and immediately immersed in 2.5% glutaraldehyde (for details see ‘Immunoblot analysis and immunolocalization of GPX proteins’).

Identification and mapping of LjGPX genes and analysis of their promoters

The transformation competent artificial chromosome (TAC) clones LjT04E19 (LjGPX1 and LjGPX2), LjT13O11 (LjGPX3), LjT08L06 (LjGPX4), LjT23J20 (LjGPX5) and LjT10B19 (LjGPX6) were isolated by screening TAC and bacterial artificial chromosome (BAC) genomic libraries using expressed sequence tag (EST) or tentative consensus (TC) sequences. The accession numbers of these sequences, as well as those of the genomic clones, are listed in Table 1. The nucleotide sequences of the candidate TAC clones were determined according to the bridging shotgun method (Sato et al., 2001).

Table 1. Glutathione peroxidase (GPX) genes of Lotus japonicus
GeneClone*Marker*Chr (cM)Position (cM)Number of ESTsESTsORF length (bp)Gene accession number
  • *

    Designation of genomic clones and map markers of the transformation competent artificial chromosome (TAC) libraries.

  • Chromosome (Chr) and map distance in centimorgans.

  • Number of expressed sequence tag (EST) sequences and designation of tentative consensus (TC) or singleton sequences.

  • ORF, open reading frame.

LjGPX4LjT08L06TM1125513.2 1BG662108513AP009590
LjGPX5LjT23J20 1BP048453516AP009591
LjGPX6LjT10B19TM198055.6 2TC12235708AP008152

The LjGPX genes were mapped using the simple sequence repeat markers indicated in Table 1. These markers were used for genotyping of the F2 mapping population of the B-129 × MG-20 cross, as described previously (Sato et al., 2001). The LjGPX6 gene was mapped using a simple sequence repeat marker found in the TAC clone LjT25H19, which overlapped with LjT10B19. A primer set (5′-GCTTTCACTTTTCTAATTGAAAAT-3′ and 5′-AAGCACATATTCTTGCCTTC-3′) that amplified 142-bp and 146-bp products from L. japonicus cvs B-129 and MG-20, respectively, was used for genotyping of the mapping population. The promoter regions of the six LjGPX genes were analysed in silico for the presence of cis-acting regulatory elements using the PlantCARE database algorithm (Lescot et al., 2002).

Predicted properties and subcellular localization of LjGPX proteins

The deduced sequences of the LjGPXs were aligned with the PileUp program (Genetics Computer Group, Madison, WI, USA) to identify conserved domains and were used, together with the sequences of the homologue proteins from other vascular plants, to construct an unrooted phylogenetic tree by the neighbour-joining method with the ClustalW 1.75 (Thompson et al., 1994) and TreeView (Page, 1996) programs. Predictions of subcellular localization were performed with the programs TargetP 1.1 (Emanuelsson et al., 2000), MitoProtII 1.0a4 (Claros & Vincens, 1996) and PSORT (Nakai & Kanehisa, 1992).

Expression analysis of LjGPX genes

Total RNA was isolated from leaves, roots or nodules using the RNAqueous kit (Ambion, Cambridgeshire, UK). The RNA was treated with DNaseI (Roche, Penzberg, Germany) to remove traces of genomic DNA, and reverse transcription was performed using a poly-T17 primer and Moloney murine leukaemia virus reverse transcriptase (Promega, Madison, WI, USA). Real-time quantitative reverse transcription–polymerase chain reaction (qRT-PCR) analysis was carried out with the iCycler iQ System (Bio-Rad, Hercules, CA, USA) using the iQ SYBR-Green Supermix (Bio-Rad) and gene-specific primers (Supporting Information Table S1). The PCR programme consisted of an initial denaturation and Taq activation step of 95°C for 5 min, followed by 50 cycles of 95°C for 15 s and 60°C for 1 min. The specificity of primers was verified using amplicon dissociation curves and gel-electrophoretic analysis. An additional control, consisting of a PCR analysis of RNA samples before reverse transcription, was also included to discount any contamination with genomic DNA. The amplification efficiency of primers, calculated by serial dilutions of root and leaf cDNAs, was > 80%. The mRNA levels were normalized with ubiquitin as the reference gene and were expressed using the method of Livak & Schmittgen (2001). All the reactions were set up in duplicate (two technical replicates) using three to five RNA preparations (biological replicates) from different plants. The threshold cycle (CT) values were in the range of 19–21 cycles for ubiquitin and 21–30 cycles for the LjGPX genes.

Immunoblot analysis and immunolocalization of GPX proteins

For immunoblots, proteins were extracted from roots and leaves at 0°C with 50 mM potassium phosphate buffer (pH 7.8) containing 0.1 mM ethylenediaminetetraacetic acid (EDTA) and 0.1% (v/v) Triton X-100, and were quantified using the Bradford dye-binding microassay (Bio-Rad). Proteins were resolved in 15% sodium dodecyl sulphate gels and were transferred onto polyvinylidene fluoride membranes (Pall Corporation, Ann Arbor, MI, USA) using a transfer buffer consisting of 25 mM Tris-HCl (pH 8.3), 192 mM glycine and 20% methanol. Equal loading of lanes and transfer quality were verified by Ponceau staining of membranes. Immunoblot analyses were performed using an affinity-purified polyclonal antibody raised against poplar GPX3.2 (Navrot et al., 2006) at a dilution of 1 : 1000. Goat anti-rabbit horseradish peroxidase-conjugated antibody (Sigma-Aldrich, St. Louis, MO, USA) was used as the secondary antibody at a dilution of 1 : 20 000, with 5% (w/v) nonfat milk to reduce background signal. Immunoreactive proteins were visualized using the SuperSignal West Pico (Pierce, Rockford, IL, USA) chemiluminescent reagent for peroxidase detection.

For EM studies, nodules of L. japonicus were fixed in 2.5% glutaraldehyde and then high-pressure frozen and embedded in low-temperature resin (Lowicryl HM23; Polysciences, Warrington, PA, USA) according to Moran et al. (2003). Unfixed root nodules, stem nodules and leaves from S. rostrata were prepared in the same way. Roots and leaves of L. japonicus and nodules of alfalfa were fixed in 2.5% glutaraldehyde and then embedded in LR White acrylic resin (Agar Scientific, Stansted, UK) as previously described for nodules of L. corniculatus and Lotus uliginosus by James & Sprent (1999). Immunogold labelling of ultrathin sections was performed according to Moran et al. (2003) using a 1 : 10 dilution of the GPX3.2 antibody and a 1 : 100 dilution of the secondary goat anti-rabbit antibody conjugated to 15-nm gold particles (GE Healthcare, Little Chalfont, UK). Serial sections incubated in nonimmune serum, which was purified with protein A and diluted 1 : 10, were used as a negative control. The sections were viewed and photographed using a JEOL 1200 EX transmission electron microscope (JEOL, Tokyo, Japan).

Results and Discussion

Identification and characterization of LjGPX genes

The first aim of this study was to isolate and characterize the GPX genes of L. japonicus. A search in the gene databases of this model legume allowed us to identify three TC sequences that contained the complete open reading frames (ORFs), as well as some 5′- and 3′-untranslated region (UTR) sequences, of LjGPX1, LjGPX3 and LjGPX6 (Table 1). Database mining also resulted in the identification of two singletons that were assigned to LjGPX4 and LjGPX5. An additional gene, LjGPX2, was identified by screening the TAC libraries using the TC sequence of LjGPX1. The LjGPX2 gene was found to be transcribed by determination of mRNA levels, as there were no ESTs available for this gene. In fact, the number of ESTs for each gene (Table 1), which were obtained by reverse transcription of total RNA from young plants, immature flowers, pods, roots and nodules of L. japonicus, strongly suggests that the expression of LjGPX2, LjGPX4 and LjGPX5 is much lower than that of LjGPX1 and LjGPX3. The EST sequence information was also used to isolate TAC clones and to map five of the genes on chromosomes 4 and 5 (Table 1). Interestingly, LjGPX1 and LjGPX2 are tandemly arranged with the same orientation on the chromosome, and their ORFs are separated by only 1477 bp. However, these two genes show only 71% identity in their ORFs, indicating that they did not originate from a recent duplication event.

To establish the exon–intron organization of the less expressed LjGPX genes, namely, LjGPX2, LjGPX4 and LjGPX5, their ORFs were completely sequenced using information based on the L. japonicus genome project (Sato et al., 2001). All the genes, except LjGPX4, contain six exons interrupted by five introns (Fig. 1). The same exon composition was observed for the eight GPX genes of A. thaliana (Rodríguez Milla et al., 2003), the six GPX genes of poplar (Navrot et al., 2006) and the single GPX gene of citrus (CsGPX1) so far examined (Avsian-Kretchmer et al., 2004). Thus, all the AtGPX and PtGPX genes have six exons, and a previous annotation (Rodríguez Milla et al., 2003) that AtGPX3 and AtGPX7 have seven and five exons, respectively, was found to be incorrect. There is also a high degree of conservation among plant species in the exon lengths. Exons 2 to 5 have identical size for all the LjGPX and AtGPX genes, with the exception of exon 5 in LjGPX4. The length of exon 1 in the LjGPX, AtGPX, OsGPX and PtGPX genes is variable, as expected for genes encoding GPX isoforms lacking or bearing N-terminal signal peptides. Exon 6 is absent in LjGPX4 and has a similar size in the other LjGPX genes (30–45 bp) and in the AtGPX genes (30–39 bp).

Figure 1.

Exon–intron organization of Lotus japonicus glutathione peroxidase (LjGPX) genes. The untranslated regions (UTRs) are depicted in hatched boxes, open reading frames (ORFs) in black boxes and introns in white boxes. Disrupted lines denote intron lengths not drawn to scale. All other lengths are drawn to scale.

The promoter regions of the six LjGPX genes were searched for the presence of putative cis-acting regulatory elements as an indication of the responsiveness of the genes to hormones or environmental cues. Most or all the LjGPX promoters contain elements that are responsive to light, biotic stress and abiotic stress, whereas only some of them contain elements that are involved in the response to anaerobiosis and hormones (Table S2). A similar conclusion was drawn from the in silico analysis of the AtGPX promoters (Rodríguez Milla et al., 2003). The LjGPX1 and LjGPX6 promoters also contain cis-sequences, with one or two mismatches, that may be responsive to oxidative stress. These sequences include the antioxidant responsive element (ARE), which also has been found in the promoters of the three catalase genes of maize (Zea mays; Scandalios et al., 1997).

Predicted properties of LjGPX proteins

The derived amino acid sequences of the six LjGPX proteins contain the three motifs present in most plant and mammalian GPXs (Criqui et al., 1992; Depège et al., 1998; Churin et al., 1999). These motifs contain residues that are proposed to be part of the catalytic site (LjGPX1 numbering): Cys-111, Cys-140, Gln-142 and Trp-200. The presence of Cys-111 indicates that none of the six LjGPX proteins is a selenium-dependent enzyme, as this amino acid residue is replaced by selenocysteine in the homologue proteins of mammals (Ursini et al., 1995) and C. reinhardtii (Fu et al., 2002). From the alignment of the LjGPX sequences shown in Fig. 2 and the data presented in Table 2, three types of protein have been differentiated based on the amino acid sequence length: LjGPX1 and LjGPX6 (c. 235 amino acids; 26 kDa); LjGPX3 (211 amino acids; 24 kDa); and LjGPX2, LjGPX4 and LjGPX5 (c. 170 amino acids; 19 kDa). These differences, and the poor homology among the first 40 or 70 amino acid residues of LjGPX1, LjGPX3 and LjGPX6 (Fig. 2), clearly suggest that these three proteins bear signal peptides for organelle targeting. Prediction programs for subcellular localization suggest that LjGPX6 has a chloroplastic N-terminal transit peptide and that LjGPX1 has an ambiguous N-terminal peptide for targeting to mitochondria and/or plastids. The same programs predicted that LjGPX2, LjGPX4 and LjGPX5 are localized in the cytosol, and LjGPX3 in the cytosol and secretory pathway (Table 2).

Figure 2.

Amino acid sequences of Lotus japonicus glutathione peroxidase (LjGPX) proteins. The three conserved domains found in most animal and plant glutathione peroxidases (GPXs) are marked in green, red and blue, respectively. The amino acid residues that form part of the proposed catalytic site of the GPXs are marked with asterisks. The cysteine residue marked with a green asterisk is replaced by selenocysteine in mammalian phospholipid hydroperoxide GPXs. Amino acid residues that are identical in at least five of the sequences are marked in white lettering on a black or coloured background.

Table 2.  Predicted properties of glutathione peroxidases (GPXs) of Lotus japonicus
ProteinLength (aa)*Molecular mass (kDa)*pI*TP (aa)Subcellular localizationPutative homologues
  • *

    Length (number of amino acid residues), molecular mass, and isoelectric point (pI) of LjGPX proteins deduced from the open reading frames (Table 1). Values for predicted mature proteins are indicated in brackets.

  • Transit peptides (number of amino acid residues) and subcellular localizations (mito, mitochondria; chloro, chloroplasts; cyt, cytosol; sec, secreted) of LjGPX proteins, as predicted by the TargetP, PSORT, and MitoProt programs.

  • Putative homologues of the LjGPX proteins in Arabidopsis thaliana (At), rice (Oryza sativa; Os), and poplar (Populus trichocarpa; Pt).

LjGPX1236 (156)26.2 (17.3)8.78 (5.58)80Mito/chloroAt6, Os1, Os3, Pt3.2
LjGPX217520.05.26CytAt8, Pt5
LjGPX3211 (175)24.2 (19.6)7.61 (7.84)36Sec (cyt)At2, At3, Os5, Pt2
LjGPX417019.09.58CytAt4, At5, Os4, Pt4
LjGPX517119.19.37CytAt4, At5, Os4, Pt4
LjGPX6235 (164)26.3 (18.6)9.01 (7.85)71ChloroAt1, At7, Os2, Pt1

The full-length amino acid sequences of plant GPXs available in the databases were used to build an unrooted phylogenetic tree (Fig. 3). Inclusion of the complete sequences rather than mature sequences was preferred for constructing the tree because there was uncertainty about the cleavage sites for many of the proteins. Alignment of the whole sequences by ClustalW was consistent with the homologies among the predicted mature proteins. The putative subcellular localizations are marked with different colours in the same figure. The tree is composed of five clades and a separated branch for barley GPX3. Clades I and II are hypothesized to contain, respectively, chloroplastic and cytosolic isoforms; clades III and IV, both cytosolic and secreted proteins; and clade V, cytosolic proteins and proteins with N-terminal transit peptides for targeting either to the mitochondria or to both the mitochondria and chloroplasts. This phylogenetic analysis updates a previous version and is fully consistent with the assignments made for clades I and II (Navrot et al., 2006). However, careful inspection of amino acid sequences by us and others (Rodríguez Milla et al., 2003) strongly suggests that at least some proteins of clades III and V may be targeted to multiple subcellular compartments. This hypothesis is based on the presence of ambiguous N-terminal signal peptides in LjGPX1 and of putative alternative translation sites in the mRNAs of LjGPX1, LjGPX3 and AtGPX6 (Table 2; Rodríguez Milla et al., 2003).

Figure 3.

Phylogenetic analysis of glutathione peroxidase (GPX) proteins of vascular plants. Only complete sequences were considered. The tree was constructed using the neighbour-joining method of ClustalW, with 1000 bootstraps, and the bar indicates 0.1 substitution per site. Predicted localizations of proteins are denoted by different colours: green, chloroplasts; red, cytosol; blue, mitochondria; brown, secretory pathway. Abbreviations of plant species: At, Arabidopsis thaliana; Ca, Cicer arietinum; Cs, Citrus sinensis; Gm, Glycine max; Ha, Helianthus annuus; Hv, Hordeum vulgare; Le, Lycopersicon esculentum; Lj, Lotus japonicus; Mc, Mesembryantemum crystallinum; Md, Malus ×domestica; Mt, Medicago truncatula; Ns, Nicotiana sylvestris; Os, Oryza sativa; Pt, Populus trichocarpa; Ps, Pisum sativum; So, Spinacia oleracea; Tm, Triticum monococcum; Vv, Vitis vinifera; Zm, Zea mays. Accession numbers for L. japonicus genes are indicated in Table 1 and those of other plant species are as follows (in brackets): At1 (At2g25080); At2 (At2g31570); At3 (At2g43350); At4 (At2g48150); At5 (At3g63080); At6 (At4g11600); At7 (A4g31870); At8 (At1g63460); Ca (CAD31839); Cs (CAA47018); Gm1 (TC203397); Gm2 (TC203326); Gm3 (TC203271); Ha1 (CAA74775); Ha2 (CAA75009); Hv1 (CAB59895); Hv2 (CAB59893); Hv3 (CAB59894); Hv4 (BAC55016); Le1 (CAA75054); Le2 (AAP59427); Mc (CAB96145); Md (AAQ03092); Ns (CAA42780); Mt1 (TC94412); Mt2 (BM814275); Mt3 (TC106328); Mt4 (TC94581); Os1 (Os04g0556300); Os2 (Os06g0185900); Os3 (Os02g0664000); Os4 (Os03g0358100); Os5 (Os11g0284900); Ps (CAA04142); Pt1 (ABK96776); Pt2 (DT518382); Pt3.1 (ABK96047); Pt3.2 (ABK94488); Pt4 (ABK95195); Pt5 (2P5Q_A); So (BAA22194); Tm (AAQ64633); Vv (CB978870); Zm (AAM88847).

Expression analyses of LjGPX genes in plant organs

The steady-state mRNA levels of the LjGPX genes were determined by qRT-PCR in leaves, roots and nodules of L. japonicus. This study allowed us to compare the abundance of the six LjGPX mRNAs within each plant organ (Fig. 4a) as well as the abundance of a specific LjGPX mRNA among the different plant organs (Fig. 4b). The first comparison was made by normalizing all mRNA levels with respect to those of one gene with significant expression, such as LjGPX6. It can be clearly observed that LjGPX1 and LjGPX3 are the most abundantly expressed genes in all three organs. Relative to LjGPX6, the mRNA level of LjGPX1 was 6-fold greater in the leaves and nodules and 25-fold greater in the roots; also, the mRNA levels of LjGPX3 were 6-, 30- and 43-fold greater, respectively, in the leaves, roots and nodules. In contrast, the less abundant transcripts (< 0.5-fold relative to LjGPX6) were those of LjGPX2 and LjGPX5 in the leaves and those of LjGPX4 in all three plant organs (Fig. 4a). The second comparison was made by normalizing the mRNA levels with respect to those found in the roots (Fig. 4b). It can be seen that there was high expression of LjGPX6 (4-fold) in leaves and of LjGPX3 (2.5-fold) and LjGPX6 (3-fold) in nodules. In contrast, LjGPX4 expression was negligible (< 0.01-fold) in both leaves and nodules (Fig. 4b). Interestingly, a consistent up-regulation (6.8-fold) of the LjGPX3 gene in nodules with respect to uninfected roots was also detected in a transcriptomic study using cDNA arrays (Colebatch et al., 2002).

Figure 4.

Expression of glutathione peroxidase (LjGPX) genes in leaves, roots and nodules of Lotus japonicus. (a) Relative expression of each gene in the three plant organs. Steady-state mRNA levels were normalized with respect to ubiquitin and are expressed relative to the values of the LjGPX6 gene, which were given an arbitrary value of 1. (b) Relative expression of each gene in leaves and nodules with respect to roots. Steady-state mRNA levels were normalized with respect to ubiquitin and are expressed relative to the values found in the roots, which were given an arbitrary value of 1. All data are means ± SE of six replicates.

Expression analyses of GPX genes in plants exposed to stressful conditions

The expression at the transcriptional level of the GPX genes was studied in L. japonicus and L. corniculatus plants exposed to several stress treatments involving production of ROS and/or RNS. We included L. corniculatus in this expression analysis to extend the molecular information gained with L. japonicus to other legume species, as we confirmed by melting curve analysis of the PCR products that the gene-specific primers were also valid for L. corniculatus.

Short-term exposure of L. japonicus plants to salt stress (150 mM NaCl) did not affect the expression of LjGPX genes, except for LjGPX4, which was down-regulated between 1 and 24 h (Fig. 5). This result is at variance with the increase of mRNA levels observed for some AtGPX genes between 3 and 12 h of treatment with 500 mM NaCl (Sugimoto & Sakamoto, 1997; Rodríguez Milla et al., 2003). This discrepancy can be attributed to the different plant species used or, more probably, to variations in the growth and stress conditions used in the experiments. A transient increase of the CsGPX1 mRNA level after 4–7 h of treatment with 200 mM NaCl was also observed in a salt-sensitive line, but not in a salt-tolerant line, of citrus cells (Avsian-Kretchmer et al., 1999). Assuming a similar link between salt sensitivity and up-regulation of GPX genes, the failure of salt stress to induce expression of LjGPX genes suggests that L. japonicus plants are relatively salt tolerant and do not experience oxidative stress. This suggestion is supported by the finding that the same salt stress treatment neither affects expression of many other antioxidant genes nor causes accumulation of lipid peroxides (data not shown).

Figure 5.

Time-course expression analysis of Lotus japonicus glutathione peroxidase (LjGPX) genes in roots of L. japonicus exposed to NaCl (150 mM), sodium nitroprusside (SNP, 100 µM) or potassium ferricyanide (100 µM). Steady-state mRNA levels were normalized with respect to ubiquitin and are expressed relative to the values at time 0 (control), which were given an arbitrary value of 1. Data are means ± SE of three to six replicates. Asterisks denote significant up-regulation (>2.0–fold) or down-regulation (<0.5–fold).

In sharp contrast with salinity, the exposure of plants to SNP, a compound capable of releasing NO for at least 40 h (Bethke et al., 2006), caused significant up-regulation of LjGPX2 after 24 h, LjGPX4 after 1, 3 and 24 h, and LjGPX5 after 3, 6 and 24 h (Fig. 5). However, the most striking effect of SNP was, by far, on the expression of the LjGPX6 gene. After only 1 h of SNP treatment, the steady-state mRNA level of LjGPX6 increased by 30-fold and by 14-, 3- and 8-fold after 3, 6 and 24 h, respectively (Fig. 5). Because decomposition of SNP yields ferricyanide in addition to NO (Bethke et al., 2006), a control treatment with potassium ferricyanide, at the same concentration as SNP, was included in the study. The only changes observed with ferricyanide were a decrease in the LjGPX4 mRNA level after 24 h (Fig. 5), confirming that the effects of SNP were indeed caused by NO. Previously, the effect of NO on GPX expression in plants had not been examined, but NO was found to up-regulate two other enzymes related to thiol metabolism, γ-glutamylcysteine synthetase and glutathione synthetase (Innocenti et al., 2007). Our finding that NO triggers, within 1 h, the expression of specific LjGPX genes, in particular LjGPX6, strongly suggests an important role of NO in the modulation of GPX function and also that at least some GPXs, in turn, may be involved in signalling pathways downstream of NO.

The expression of GPX genes was examined in L. corniculatus plants treated with Cd or Al (Fig. 6), two metals with potent phytotoxic effects. A recent report has shown that, in addition to its general inhibitory effects on plant growth and photosynthesis, Cd induces oxidative stress, as evidenced by an increase in ROS production and in the amounts of oxidized lipids and proteins (Romero-Puertas et al., 2004). In addition, two other pieces of evidence argue in favour of a role of GPX in the response to Cd. First, the protein level of an A. thaliana GPX is increased in response to Cd, and secondly, the resolution of the structure of poplar GPX5, in which 32 Cd atoms are bound to a dimer, suggests that at least some plant GPXs could act as a sink for Cd (Navrot et al., 2006; Koh et al., 2007). We found that Cd caused down-regulation of GPX1, GPX2, GPX4 and GPX6 between 3 and 24 h of treatment, and of GPX5 only after 3 h (Fig. 6). In contrast, there was significant up-regulation (2.7-fold) of GPX3, GPX4 and GPX5 after 1 h of treatment and of GPX3 after 24 h. Thus, Cd triggers a very rapid and transient activation (1 h) of specific GPX genes and a subsequent decrease (within 24 h) of the mRNA levels for all the genes (except GPX3) to control (or below control) values.

Figure 6.

Time-course expression analysis of glutathione peroxidase (GPX) genes in roots of Lotus corniculatus exposed to 20 µM cadmium (Cd) or 20 µM aluminium (Al). Steady-state mRNA levels were normalized with respect to ubiquitin and are expressed relative to the values at time 0 (control), which were given an arbitrary value of 1. Data are means ± SE of three to six replicates. Asterisks denote significant up-regulation (>2.0–fold) or down-regulation (<0.5–fold).

The effect of Al was also studied, for comparison, at the same concentration as Cd. The first symptom of Al toxicity, which develops at acid pH, is the inhibition of root growth (Barceló & Poschenrieder, 2002). Moreover, Al causes oxidative stress, an increase in antioxidant enzyme activities and lipid peroxidation (Sharma & Dubey, 2007; and references therein). We found that Al caused a general decrease of GPX transcripts after 6 and 24 h of treatment (Fig. 6). An earlier report showed that the treatment of A. thaliana plants with a combination of 100 µM Al and 100 µM Fe (pH 4.0) increased the AtGPX6 mRNA level (Sugimoto & Sakamoto, 1997). However, this experiment is not comparable to ours because of major differences in the growth and treatment conditions of plants.

Immunoblot analyses and immunolocalization of GPX proteins

To gain further information on LjGPXs and also as a prerequisite for immunolocalization studies, we performed immunoblot analyses in leaves, roots and nodules of L. japonicus using an antibody raised against poplar GPX3.2 (Navrot et al., 2006). This antibody was not isoform specific and recognized at least two LjGPX isoforms in the three plant organs (Fig. S1). The apparent molecular masses of the immunoreactive proteins were 20.5 and 24 kDa, in the range expected for GPX proteins (Table 2). The antibody also recognized two immunoreactive bands in extracts from L. corniculatus, A. thaliana, and poplar (data not shown) and a single immunoreactive band of c. 20.5 kDa when the antibody was challenged with poplar GPX3.2 (Fig. S1). An antiserum raised against synthetic peptides, based on the sequence of tomato GPXLe-1 (Herbette et al., 2004), recognized the same two immunoreactive bands in leaf, root and nodule extracts of L. japonicus, further confirming that the two immunoreactive bands are genuine GPXs. No major changes were observed in the LjGPX protein levels of L. japonicus or L. corniculatus exposed to the stresses described above for up to 24 h (data not shown), although they were increased in plants treated with 150 mM NaCl for 7 d (Fig. S1).

To our knowledge, there are only two reports on the subcellular localization of GPXs in plants. First, the antiserum against GPXLe-1 was used to detect the protein in the cytoplasm and cell wall of tomato internodes (Herbette et al., 2004). Secondly, a fusion of the N-terminal extension of GPX3.2 with the green fluorescent protein was used to show that GPX3.2 is targeted to the mitochondria and chloroplasts of tobacco (Nicotiana tabacum) cells (Navrot et al., 2006). We have used the GPX3.2 antibody, which was affinity-purified and hence more suitable for EM studies than the GPXLe-1 antiserum, to immunolocalize GPX proteins in L. japonicus. Plant material was high-pressure frozen for optimal preservation of antigenicity (Moran et al., 2003). Most surprisingly, GPX was found exclusively in the chloroplasts or amyloplasts of leaves, roots and nodules (Fig. 7). Furthermore, the labelling was associated with the starch grains in the uninfected cells of the cortex (Fig. 7a) and in the interstitial cells of the infected region (data not shown). The same localization of GPX was found in the proplastids of root tips (Fig. 7c) and in the chloroplasts of leaves (Fig. 7d). For comparison purposes, the localization of GPX was also examined in alfalfa, a crop legume with indeterminate nodules, and in S. rostrata, a tropical legume with determinate nodules on the roots and stems. Thus, abundant labelling was found in the starch grains in the infected cells of alfalfa nodules (Fig. 7e), as well as in the cortical chloroplasts of stem nodules (Fig. 7f) and in the leaf chloroplasts (Fig. 7g) of S. rostrata. Plastids in the nonphotosynthetic root nodules on S. rostrata were also labelled (not shown). Negligible labelling was observed on sections incubated in nonimmune serum, such as those of L. japonicus nodules (Fig. 7b) and S. rostrata leaves (Fig. 7h).

Figure 7.

Immunogold localization of glutathione peroxidase (GPX) proteins within plastids of leaves, roots and nodules of legumes. (a) Plastid (p) within a cortical cell in a nodule of Lotus japonicus. The immunogold labelling (arrow) is specifically localized on the starch grain (s) in the plastid. (b) Serial section to (a) but treated with nonimmune serum substituted for the GPX antibody. There is no significant labelling on the starch grain (s) or elsewhere in the cortical cell. (c) Immunogold-labelled (arrow) starch grain (s) in a plastid (p) within an L. japonicus root tip cell. Note that the adjacent mitochondria (m) are not labelled. (d) Immunogold labelling (arrow) of starch grains (s) in a leaf chloroplast of L. japonicus. (e) Immunogold labelling (arrows) of starch (s) in a plastid within an infected cell of an alfalfa nodule. (f) Chloroplast (c) in a Sesbania rostrata stem nodule with immunogold labelling (arrows) of the starch grains (s) within it. (g) Immunogold labelling of starch grains (s) within chloroplasts (c) in a leaf of S. rostrata. (h) Serial section to (g) but treated with nonimmune serum substituted for the GPX antibody. There is no significant labelling on the starch grain (s) or elsewhere in the leaf cell. b, bacteroid; c, chloroplast; is, intercellular space; p, plastid; v, vacuole. Bars, 1 µm.

The presence of GPX in association with starch grains had not been reported earlier but is consistent with the detection of thioredoxins f and m and of peroxiredoxin BAS1 in the amyloplasts (Balmer et al., 2006; Barajas-López et al., 2007). Thus, thioredoxins are substrates for both GPXs and peroxiredoxins, which are closely related enzymes (Navrot et al., 2006). Our finding of GPX in the amyloplasts suggests that peroxides (their other substrates) are also formed in these organelles and points to a regulatory role of this enzyme, in connection with thioredoxins, in heterotrophic tissues. In this context, at least two questions are worthy of further research. First, are some GPX and peroxiredoxin isoforms involved in the regulation of starch biosynthesis and/or mobilization, perhaps through changes in the thiol redox state mediated by ROS and/or RNS? Secondly, why was the antibody unable to detect any GPX in the cytosol or other subcellular compartments, despite the finding that the LjGPX3 gene, encoding a putative cytosolic or secretory protein (Table 2), is highly expressed at the mRNA level (Fig. 4)? It is possible that LjGPX3 is subjected to post-transcriptional regulation or that the localization of the corresponding protein is confined to certain plant tissues or cells. In any case, the results of the immunolocalization studies along with the surprisingly rapid and strong induction of specific GPX genes by NO underscore the complex regulation of the GPX genes in plants and strongly support the hypothesis that the GPX proteins play an important role not only as antioxidants but also in stress or metabolic signalling.


We thank Patricia Roeckel-Drevet for valuable advice and a gift of tomato GPXLe-1 antibody and Carmen Pérez-Rontomé for excellent technical assistance. This work was supported by grants from the European Commission (FP6-2003-INCO-DEV2-517617), Spanish Ministry of Education and Science (AGL2005-01404) and Gobierno de Aragón (group A53).