All PRX proteins contain a conserved Cys in their N-terminal part and some of them possess a second conserved Cys residue. The strictly conserved Cys residue is oxidized during the mechanism of ROS scavenging (Schroder and Ponting, 1998; Seo et al., 2000). To obtain regenerated PRX, the oxidized Cys residue forms a disulfide bridge, which is next reduced by disulfide oxidoreductases. On the basis of the number and position of conserved Cys residues, the PRX proteins are classified in several sub-families (Choi et al., 1999). Phylogenetic analysis using a wide range of PRX amino acid sequences and including related sequences of unidentified function (Verdoucq et al., 1999), allows the division of PRX into four main groups. The 2Cys-PRX group corresponds to PRX containing two conserved Cys residues, the 1Cys-PRX group includes PRX with only one conserved Cys residue, and the type II PRX family is composed of another type of PRX with two conserved Cys residues. In addition, a fourth group of PRX, called type Q PRX, has also been described and contains homologs of the E. coli bacterioferritin-associated protein (Jeong et al., 2000; Kong et al., 2000).

In plants, the four types of PRX have been described (Baier and Dietz, 1996; Horling et al., 2002; Konig et al., 2002). Two Arabidopsis genes encode chloroplastic 2Cys-PRX (Dietz et al., 2002). All plant homologs present a transit peptide suggesting a chloroplastic localization of the mature proteins, and several publications have reported the characterization and involvement of 2Cys-PRX from barley (Hordeum vulgare) and spinach (Spinacia oleracea; Baier and Dietz, 1996, 1997, 1999), Chinese cabbage (Brassica campestris L. subsp. pekinensis; Cheong et al., 1999), or Arabidopsis (Baier et al., 2000; Konig et al., 2002) in the protection of chloroplasts against ROS. 1Cys-PRX has also been studied in diverse plant species such as rice (Oryza sativa), Arabidopsis, buckwheat (Fagopyrum esculentum), or barley (Stacy et al., 1996, 1999; Haslekas et al., 1998; Lee et al., 2000; Lewis et al., 2000) and are described as being related to seed dormancy (Lewis et al., 2000) or at least to antioxidant activity in seeds (Lee et al., 2000). A type Q PRX described in Sedum lineare is also proposed to act in vivo as an antioxidant (Kong et al., 2000). One Arabidopsis gene encodes a PRX-Q protein homologous to the S. lineare PRX-Q (Horling et al., 2002).

In this paper, we describe the most recently discovered PRXII family. In Arabidopsis, it is composed of six members that define three different subgroups. Using RT-PCR and plant reporter gene studies, we present the differential expression patterns of AtPRXII. We also report the detection of AtPRXII proteins in different organs and in three distinct cellular compartments. We discuss the in vitro activity characteristics of AtPRXII and demonstrate that at least one of them is reduced by the GRX system.

Figure 1. Phylogenetic tree of proteins homologous to PRX. The DARWIN software (Gonnet et al., 1992) was used to generate this tree. The first letters correspond to the initials of the organism genus and species names. GenBank accession numbers are indicated: (more ...)

Homologs of the Brewer's yeast AHP1 protein, the last discovered yeast PRX (Jeong et al., 1999; Lee et al., 1999; Verdoucq et al., 1999), form the type II PRX group, which is the focus of this study. It includes human and plant sequences. At least in plants, the type II PRX group is more complex than the three other groups. The Arabidopsis genome contains six different genes belonging to this group (Fig. 1). AtPRXII-F was detected during the analysis of the Arabidopsis mitochondrial proteome (Kruft et al., 2001). The ORF encodes 199 amino acids that includes an N-terminal extension of 28 residues (Kruft et al., 2001). AtPRXII-E has an ORF encoding 234 amino acids with an N-terminal extension of 70 amino acids that could code for a transit peptide predicted to address the protein to mitochondria or plastids depending on the prediction programs used. AtPRXII-B, -C, and -D are closely related and will be analyzed in detail in this study. Each gene has an ORF of 162 amino acids, which is exactly the same length as the PRXII described in poplar (Populus spp.; Rouhier et al., 2001) and B. rapa (Choi et al., 1999).

AtPRXII-B and AtPRXII-C are present as a tandem repeat. AtPRXII-A is on the same BAC F12P19 at a distance of approximately 10 kb, whereas AtPRXII-D is also on chromosome I, but further away. AtPRXII-E and AtPRXII-F are located on chromosome III. The coding sequences of AtPRXII-A, -B, -C, and -D are interrupted by two introns at the same positions, whereas AtPRXII-E has no intron. The AtPRXII-F coding sequence is interrupted by three introns at positions unrelated to those of AtPRXII-A, -B, -C, and -D. Thus, based on the exon/intron structure, the AtPRXII family can be divided into three subgroups.

Due to an error in the Arabidopsis genome annotation, AtPRXII-D was previously described as a pseudogene (Horling et al., 2002), but the isolation of a partial cDNA in our laboratory allowed us to correct the annotation. AtPRXII-A shares homologies to these sequences, but the predicted protein is far larger (560 amino acids) due to a C-terminal domain homologous to numerous Arabidopsis sequences without known function. An alignment of the deduced amino acid sequences of these five Arabidopsis type II PRXs is shown in Figure 2. Each of these type II PRXs have the characteristic feature of all of the PRXII sequences analyzed so far. They contain one strictly conserved Cys residue (Cys-51 in AtPRXII-B, -C, and -D) with the surrounding consensus sequence: P[G/L][A/D]FT[P/F][T/V]C[S/P/T] (Rouhier et al., 2001). In addition, all of them possess a second Cys residue separated from the first one by the same number of residues.

Figure 2. Multiple sequence alignment of Arabidopsis AtPRXII proteins. The alignment of the protein sequences of AtPRXII-B, AtPRXII-C, AtPRXII-D, AtPRXII-E, AtPRXII-F, and of the N-terminal part of AtPRXII-A was obtained using the ClustalW software. The active (more ...)

It is not known if the AtPRXII-A protein is correctly predicted because no EST or complete cDNA are available at the present time. Therefore, this protein is not further analyzed in this study. AtPRXII-B, -C, and -D are highly similar, showing between 93% to 99% identity. They share about 60% identity with the amino acid sequence of AtPRXII-E without the N-terminal extension, whereas they only share 50% similarity with the predicted mature form of AtPRXII-F. Thus, the amino acid sequence of AtPRXII allowed us to define three subgroups corresponding to those already defined by the exon/intron structures of the genes. We were able to find rice homologs of each sub-group, suggesting that this organization is common to all higher plants (Fig. 1).

To test the in vitro peroxidase activity of these different AtPRXII, we used the classical assay of plasmid DNA protection against ROS generated by the Fenton reaction induced by this metal-catalyzed system. Dithiothreitol (DTT) and FeCl₃ were incubated with 1 μg of plasmid DNA at room temperature in the presence or absence of 20 μm of recombinant proteins. After 5 h of incubation, the DNA mixed without PRX was completely degraded, as was DNA incubated with bovine serum albumin (BSA). In contrast, DNA incubated with AtPRXII-B or Δ70AtPRXII-E was efficiently protected (Fig. 3). Nevertheless, we observed a difference in the pattern of DNA protection between AtPRXII-B and Δ70AtPRXII-E, because the plasmid super-coiled form is much less abundant using Δ70AtPRXII-E than with AtPRXII-B. These results are in good agreement with the DTT-dependent H₂O₂-reducing activities obtained by Horling et al. (2003). They showed that AtPRXII-E has a DTT-dependent activity three times less efficient than that obtained for AtPRXII-B. Thus, a low protection of DNA by Δ70AtPRXII-E could be attributed to its lower peroxidase activity, as compared with that of AtPRXII-B.

Figure 3. Type II AtPRX-dependent inactivation of DNA degradation by metal catalyzed oxidants. One microgram of pBluescript plasmid was incubated with a mixture of 10 μm FeCl3 and 1 mm DTT. The addition of AtPRXII-B or Δ70AtPRXII-E proteins led (more ...)

The reaction mechanism of 2Cys-PRX includes homodimer formation. In contrast, the activity of type II PRXs analyzed so far is associated with monomers. We tested the ability of the recombinant AtPRXII to form dimers. Recombinant proteins were analyzed directly after purification under nonreducing conditions or were incubated under reducing conditions with 5% (v/v) β-mercaptoethanol and then submitted to SDS-PAGE (Fig. 4). Under nonreducing conditions, a dimer of about 44 kD and a monomeric form of about 22 kD were observed with His-tagged AtPRXII-B. The dimeric form completely disappeared in the presence of β-mercaptoethanol. ΔAtPRXII-E protein also formed a dimer in nonreducing conditions. Thus the recombinant Arabidopsis type II PRXs are able to form disulfide bonds between two molecules. Nevertheless, the amount of dimers is not increased when PRXIIs are incubated with H₂O₂ at 0.5 to 50 mm (data not shown). In addition, no dimers were detected in crude plant extracts under nonreducing conditions by western blot using sera raised against AtPRXII-B or AtPRXII-E (data not shown). Although we were not able to demonstrate the existence of these dimers in planta, we cannot exclude that this property of the AtPRXII proteins to form dimers may not play a regulating role under certain conditions.

Figure 4. Effect of 2-mercaptoethanol on type II AtPRX dimerization. Recombinant AtPRXII-B and truncated Δ70AtPRXII-E proteins were boiled for 5 min in a loading buffer in the presence (+) or absence (-) of the reducing agent 2-mercaptoethanol. About (more ...)

To further investigate the thiol dependence of the type II AtPRX peroxidase activity, we followed the oxidation of NADPH by the TRX or GRX system. The TRX system consists of Arabidopsis TRX reductase (Jacquot et al., 1994) and Arabidopsis TRX h2 (GenBank accession no. S58123). With NADPH as the primary electron donor, it reduces insulin efficiently (data not shown) and acts as an efficient reducer of AHP1, the PRXII of Brewer's yeast (Verdoucq et al., 1999). We were not able to detect any peroxidase activity in the presence of this TRX system for either AtPRXII-B or Δ70AtPRXII-E. However, in the presence of a GRX system, composed of a commercial glutathione reductase associated with reduced glutathione (GSH) and a recombinant Arabidopsis GRX, the initial rate of NADPH oxidation increased as a function of the AtPRXII-B concentration (Fig. 5B). The turnover number of the peroxidase activity of AtPRXII-B was estimated to be 2.86 10^-2 s^-1 under the conditions of the experiment. The reduction system without GSH did not show any activity. The absence of either GRX or PRX led to a residual NADPH consumption (Fig. 5A), which is most probably due to GSH oxidation by hydrogen peroxide. No significant activity was observed with the recombinant truncated Δ70AtPRXII-E. These results suggest that GRX is the main electron donor to AtPRXII-B and that the AtPRXII-E enzyme has a thiol-dependent activity but its electron donor has still to be identified.

Figure 5. Thiol peroxidase activity of AtPRXII-B followed by oxidation rate of NADPH. NADPH oxidation was monitored at 340 nm in a 500-μL reaction mixture containing 0.1 m potassium-phosphate buffer, pH 7, 2 mm EDTA, 1 mm H2O2, 0.5 mm NADPH, 0.5 unit (more ...)

To get insights into the subcellular localizations of the AtPRXII, we performed immunoblots against subcellular fractions of Arabidopsis protoplasts. The purity of each subcellular fraction was tested using antibodies raised against proteins specifically located in the different organelles: the large subunit of the light-harvesting complex II from C. reinhardtii for the chloroplast, TRX h (AtTRXh3) for the cytosol, pyruvate dehydrogenase for the mitochondrion, and catalase for the peroxisome. No contamination was detected in chloroplast and cytosol fractions (Fig. 6, lines C and E). However, a very weak contamination by chloroplasts in the mitochondrial fraction can be observed (Fig. 6, line D). Anti-AtPRXII-E serum reveals a 19-kD band in total protoplast and chloroplast fractions (Fig. 6, line B). A very faint signal is also observed in the mitochondrial fraction, which is most probably due to the weak contamination of this fraction by chloroplasts. The chloroplast localization of AtPRXII-E was confirmed by import experiments of the radiolabeled protein into purified organelles. No import could be detected into potato (Solanum tuberosum) mitochondria, whereas processing of the major 29-kD protein, obtained by in vitro coupled transcription/translation, was observed after its incubation under light with pea chloroplasts (data not shown). A signal of 19 kD resistant to added proteinase K appeared at the same position as the Arabidopsis protein immunodetected in chloroplast extracts. Thus we concluded that AtPRXII-E is only located in the chloroplasts.

Figure 6. AtPRXII sub-cellular localization. Arabidopsis total protoplast (P), cytosol (Cy), chloroplast (Ch), and mitochondrial (M) fractions were probed with antibodies directed against AtTRXh3, pyruvate dehydrogenase (PDH), catalase, and the light-harvesting (more ...)

Using the serum raised against AtPRXII-B, signals were obtained in protoplast and cytosol fractions but not in those of mitochondria or chloroplasts (Fig. 6, line A). As shown using the serum raised against catalase (Fig. 6, line F), peroxisomes copurify with the mitochondrial fraction, whereas soluble peroxisomal proteins contaminate the cytosolic fraction. Because no signal is detected by the AtPRXII-B serum in the mitochondrial fraction, we conclude that AtPRXII-B and probably AtPRXII-C and -D are cytosolic proteins and are not located in the peroxisomes.

The expression of each gene was analyzed by reverse transcription followed by radioactive PCR (Fig. 7). This allowed us to amplify the sequence of interest and the control gene, actin 2 (GenBank accession no. U41998), which is supposed to be uniformly expressed in all tissues, in the same tube. Because the Act2 gene is not expressed in dry seeds, we used primers designed to amplify EM1 mRNA (Carles et al., 2002) as the internal control for seeds. Signals were then quantified using a phosphorimager, and the intensities were related to the intensity of the actin signal, enabling us to compare levels of expression between the different tissues (Fig. 7B). Because of the different efficiency of each couple of primers, it is not possible to compare the level of expression of the different genes. Nevertheless, the number of ESTs in the database gives some estimation of the relative expression, at least in the tissues used for the construction of the cDNA libraries: AtPRXII-A, no EST; AtPRXII-B, 30 ESTs; AtPRXII-C, 10 ESTs; AtPRXII-D, no EST; AtPRXII-E, 3 ESTs; and AtPRXII-F, 14 ESTs.

Figure 7. Expression of AtPRXII-B to -F in different plant organs. A, Semiquantitative RT-PCR was performed using gene-specific (AtPRXII) and reference gene (Act2 or EM1) primers. B, Signal intensities were measured using a phosphorimager and relative abundance (more ...)

AtPRXII-B mRNA is detected in all tissues but mostly in reproductive tissues such as buds, flowers, siliques, and seeds. AtPRXII-B is also strongly expressed in callus and cell suspensions. Concerning the AtPRXII-C transcripts, the strongest signals were obtained with buds and flowers, whereas a very faint signal is detected in all other tissues, except in roots. AtPRXII-D mRNA is present exclusively in buds and flowers. A signal corresponding to the AtPRXII-E transcripts was obtained with all tissues, but with a high predominance in buds, siliques, seeds, cell suspensions, and plantlets. AtPRXII-F transcripts are detected in all tissues without any significant difference of intensity (Fig. 7). Thus, the genes encoding PRXs have distinct tissue expression patterns.

To get more insight into the expression patterns of AtPRXII genes, we analyzed their promoter activities using transgenic Arabidopsis plants carrying AtPRXII-B to -E promoter-β-glucuronidase (GUS) reporter gene fusion. 5′-Upstream regions of 1.465, 1.139, 1.141, and 1.831 kb from the gene initiation codon of AtPRXII-B, -C, -D, and -E, respectively, were used to generate transcriptional fusions with the GUS reporter gene. The constructs were introduced into wild-type Arabidopsis plants via Agrobacterium tumefaciens-mediated transformation. Independent primary transformants were obtained, and the T₂ or T₃ generations were analyzed for GUS expression. Five lines of T₂ generation for AtPRXII-B, -C, and -E, respectively, and three homozygous lines of T₃ generation for AtPRXII-D were analyzed. Concerning the expression of AtPRXII-C and -D, similar GUS staining for the different reporter lines, consistent with the RT-PCR data, was observed: AtPRXII-C is almost exclusively expressed in flower buds, and the GUS staining was restricted to mature pollen (Fig. 8C). In pAtPRXII-D::GUS plants, GUS activity is detected in mature pollen but also in germinating pollen, pollinic tubes, and fertilized ovules (Fig. 8A). To verify whether the staining observed in fertilized ovules in siliques was from pollinic tubes or from maternal material, we did reciprocal fertilizations using pollen of wild-type plants to fecundate transgenic pAtPRXII-D::GUS pistils, or pollen from transgenic pAtPRXII-D::GUS flowers to fertilize wild-type pistils. GUS staining in ovules was observed exclusively when fertilizing wild-type pistils with pAtPRXII-D::GUS pollen, leading to the conclusion that only the male material expressed the GUS gene.

Figure 8. Histochemical localization of GUS activity in Arabidopsis plants. GUS activity was revealed by blue staining after incubation of 6- to 8-week-old plants in 5-bromo-4-chloro-3-indolyl-β-glucuronic acid.

Concerning pAtPRXII-B::GUS plants, we did not obtain consistent results with the different lines. One line exhibited blue staining in vascular tissues of stem and roots, whereas another line was stained in blue exclusively in young anthers. Some lines were not stained at all. No line gave a staining pattern coherent with the RT-PCR data. Most probably the 1,465 bp of the promoter did not contain all of the information responsible for in vivo expression of AtPRXII-B gene.

Looking at pAtPRXII-E::GUS plants, we observed blue staining in stamen of young flowers, the embryo sac of young seeds, and the albumen of older seeds from green or yellow siliques (Fig. 8B). These results are consistent with the expression pattern we obtained by RT-PCR. However, we did not observe blue staining in 10-d-germinating plantlets, whereas transcripts were detected by RT-PCR (Fig. 7). This may be due to a difference of stability of the AtPRXII-E transcripts in plantlets. The AtPRXII-E is weakly expressed, but the corresponding stable transcripts accumulate to a high level and thus are detectable by RT-PCR.

Table I. pl and molecular masses (kD) of AtPRXII predicted mature forms

We first performed a one-dimensional western blot from an SDS-PAGE loaded with exactly 10 μg of proteins from the different tissues. Using the serum raised against AtPRXII-B, a stronger signal was observed with extracts from buds, flowers, and seeds than with proteins from other tissues (Fig. 9B). Probing the same membrane with the serum raised against AtPRXII-E (Fig. 9D), we observed the presence of AtPRXII-E in an approximately equal quantity in each tissue that was tested, except for roots where a weaker signal was seen. The serum also reacts with proteins of higher M_r that fit exactly the signals showed by anti-AtPRXII-B serum with the same membrane (Fig. 9D).

Figure 9. Type II AtPRX protein accumulation patterns. The serum raised against AtPRXII-B (A and B) and AtPRXII-E (C and D) detected the type II PRX proteins in planta. A and C, Total protein extracts from different plant tissues were separated by two-dimensional (more ...)

With all tissue extracts, one or several signals were obtained on two-dimensional gels with the serum raised against AtPRXII-B (Fig. 9A). These different signals may correspond to the AtPRXII-B, -C, and -D proteins or to modified isoforms of AtPRXII-B, -C and/or -D.

Using serum raised against AtPRXII-E, two types of signal were obtained with two-dimensional blots (Fig. 9C): strong signals most probably corresponding to the mature AtPRXII-E, and weaker signals (indicated in Fig. 9C by a B) present at exactly the same position as those obtained with serum raised against AtPRXII-B and then corresponding to AtPRXII-B to -D. AtPRXII-E signals were observed with all tissues, leading to the conclusion that AtPRXII-E as well as AtPRXII-B and/or AtPRXII-C and/or AtPRXII-D are present in all plant tissues. The position of the AtPRXII-E spots, compared with the ones of AtPRXII-B to -D, coincides with the molecular mass and pI of the predicted mature AtPRXII-E (Table I; Fig. 9).

More striking is the almost exclusive expression in pollen of the two AtPRXII-C and -D genes. To our knowledge, it is the first report of PRX expression localized in flowers. Until now plant PRXs were mainly characterized in leaves and seeds (Baier and Dietz, 1996; Haslekas et al., 1998; Lewis et al., 2000; Rouhier et al., 2001; Horling et al., 2002, 2003). This particular expression pattern is reminiscent of the TRX regulation of the S-locus receptor kinase, implicated in cauliflower (Brassica oleracea) pollen auto-incompatibility (Cabrillac et al., 2001), and suggests a developmental function for these two proteins. In addition, this expression pattern may mean that the PRXII-C and/or -D proteins are implicated in the protection of pollen components. Pollen grains are exposed to an increase of free radical production during desiccation (Van Bilsen and Hoekstra, 1993) and PRXs constitute one of the molecular antioxidants able to alleviate oxidative stress in dried tissues (Hoekstra et al., 2001).

The gene encoding plastidial PRX AtPRXII-E is expressed constitutively but is overexpressed in anther tapetum and during seed formation, suggesting that the function of this protein is not limited to chloroplasts, but is also very active in other types of plastids. In particular, elaioplats are plastids present in anther tapetum and contain mainly globuli of neutral esters (Wu et al., 1997; Ting et al., 1998; Hernandez-Pinzon et al., 1999). AtPRXII-E could thus be involved in a protective mechanism preventing a high level of lipid peroxidation in elaioplasts. The gene coding for the mitochondrial PRX AtPRXII-F (Kruft et al., 2001) is ubiquitously expressed. The protein is probably implicated in the protection of mitochondria against ROS damage and may be reduced by the mitochondrial TRX system described by Laloi et al. (2001), as suggested by Horling et al. (2003).

The plastidial AtPRXII-E cannot be efficiently reduced by any of these systems. This is not surprising because, on the basis of the Arabidopsis genome, no gene was found to encode a GRX with a transit peptide. The lack of activity in the presence of the cytosolic TRX system is more surprising, because the substrate specificity of TRXs is rather low in vitro. For example, a C. reinhardtii cytosolic TRX was reported to be an excellent electron donor for a chloroplastic 2-Cys PRX from Arabidopsis (Goyer et al., 2002). Nevertheless, a systematic test with chloroplastic TRXs should be performed in the future.

Table II. Sequences of primers used for RT-PCR experiments

As a control for genomic DNA contamination, PCR was carried out on RNA not treated by reverse transcriptase, with each couple of primers. No amplification was observed. Furthermore, different couples of primers we used are able to amplify genomic DNA leading to the production of DNA fragments of higher M_r (612, 1,390, 770, 1,010, and 1,503 bp for Act2, AtPRXII-B, AtPRXII-D, AtPRXII-E, and AtPRXII-F, respectively). But such an amplification was never observed.

To overproduce AtPRXII in Escherichia coli, BL21(DE3) strains were cotransformed with pSBET (Schenk et al., 1995) and pET16b containing the AtPRXII constructs, and grown in 500 mL of liquid B medium up to A₆₀₀ nm = 0.5, when production was induced by addition of 0.4 mm isopropyl-1-thio-β-d-galactopyranoside for 3 h at 37°C. Cells were then pelleted and stored at -80°C. Total proteins from bacteria were extracted using a hydraulic press (model 3968, Carver, Wabash, IN) and purified on a nickel affinity column as described by Verdoucq et al. (1999). The productions of antibodies from rabbit were done by Eurogentec using 400 μg of each purified AtPRXII-B and Δ70AtPRXII-E.

Immunodetections were done as described by Mouaheb et al. (1998) with the following modifications: Wet membranes were incubated at room temperature for 1 h in Tris-buffered saline plus Tween 20 (TBS-T) buffer (20 mm Tris-HCl, 500 mm NaCl, and 0.05% [v/v] Tween 20, pH 7.6) containing 5% (w/v) lyophilized skimmed milk, washed once in 50 mL of TBS-T buffer one time for 15 min and twice for 5 min. The reaction with the first antibody was carried out for 2.5 h at room temperature in 20 mL of TBS-T buffer containing 1% (w/v) lyophilized skimmed milk and 20,000-fold diluted rabbit immunoserum directed against the different PRXs. The membranes were washed and incubated for 1 h at room temperature in 20 mL of TBS-T buffer containing 20,000-fold diluted goat anti-rabbit horseradish peroxidase conjugate (Bio-Rad Laboratories, Hercules, CA). Finally, antibodies were visualized with ECL+ Western Blotting detection reagent (Amersham Biosciences, Uppsala).

Concerning in vitro import into isolated chloroplast and mitochondria, the precursor proteins were synthesized from the corresponding cDNA clones by coupled in vitro transcription/translation (TNT kit, Promega, Madison, WI) in the presence of [³⁵S]Met. Chloroplasts were isolated from leaves of 10-d-old pea seedlings and import assays were carried out as described previously (Bruce et al., 1994). Mitochondria were isolated from cauliflower with a juice extractor as described (Fey et al., 1999). Import assays into mitochondria were carried out as described (Wischmann and Schuster, 1995).

AtPRXII GRX-dependent or TRX-dependent peroxidase activity was assayed as follows: After 5 min of preincubation, the reaction was initiated by the addition of 1 mm H₂O₂ to 0.5 mL of 0.1 m potassium phosphate buffer at pH 7 containing 2 mm EDTA, 0.5 mm NADPH (Sigma-Aldrich), different concentrations of AtPRXII, and the GRX system composed of 0.5 unit of glutathione reductase (Roche Diagnostics), 0.5 mm GSH (Sigma-Aldrich), and 10 μm Arabidopsis GRX AtGRX1, or the TRX system composed of 10 μm Arabidopsis TRX AtTRXh2 and 0.2 μm Arabidopsis TRX reductase AtNTRB. The consumption of NADPH was followed spectrophotometrically at 340 nm at 22°C. The TRX insulin disulfide reduction assays were performed as described (Laloi et al., 2001). These experiments were done two to five times, and comparable results were obtained.

We thank Dr. C. Laloi for the gift of recombinant NTRB, Dr. L. Verdoucq for pET-AtPRXII-B construct, Y. Chartier for technical help, Dr C. de Vitry for the gift of anti-light-harvesting complex II serum, and Dr. J.-P. Reichheld and Dr. R. Cooke for critical reading of the manuscript.