Elevated Glutathione Biosynthetic Capacity in the Chloroplasts of Transgenic Tobacco Plants Paradoxically Causes Increased Oxidative Stress

Correspondence to: Gary Creissen, creissen{at}bbsrc.ac.uk (E-mail), 44-1603-456844 (fax)

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Glutathione (GSH), a major antioxidant in most aerobic organisms, is perceived to be particularly important in plant chloroplasts because it helps to protect the photosynthetic apparatus from oxidative damage. In transgenic tobacco plants overexpressing a chloroplast-targeted -glutamylcysteine synthetase (-ECS), foliar levels of GSH were raised threefold. Paradoxically, increased GSH biosynthetic capacity in the chloroplast resulted in greatly enhanced oxidative stress, which was manifested as light intensity–dependent chlorosis or necrosis. This phenotype was associated with foliar pools of both GSH and -glutamylcysteine (the immediate precursor to GSH) being in a more oxidized state. Further manipulations of both the content and redox state of the foliar thiol pools were achieved using hybrid transgenic plants with enhanced glutathione synthetase or glutathione reductase activity in addition to elevated levels of -ECS. Given the results of these experiments, we suggest that -ECS–transformed plants suffered continuous oxidative damage caused by a failure of the redox-sensing process in the chloroplast.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Plants, like all aerobic organisms, possess an array of hydrophilic and lipophilic antioxidants, such as glutathione (-L-glutamyl-L-cysteinyl-L-glycine [GSH]), ascorbic acid (vitamin C), phenolic isoflavanoid compounds, -tocopherol (vitamin E), and the carotenoids, including the xanthophylls (Fryer 1993 ; Mullineaux and Creissen 1996 ). The reduced forms of these compounds, together with antioxidant enzymes, scavenge reactive oxygen species (ROS) and other products of oxidative reactions. These enzymes include subcellular compartment–specific isoforms of superoxide dismutase (SOD), catalase, ascorbate peroxidase (APX), glutathione S-transferase/glutathione peroxidase (GST/GPX), dehydroascorbate reductase, monodehydroascorbate free radical reductase, and glutathione reductase (GR). Several reduction–oxidation (redox) cycles that scavenge ROS in different subcellular compartments and that involve these enzymes and antioxidants have been proposed (e.g., the ascorbate–GSH cycle). The reducing equivalents for these reactions are derived ultimately from photosynthetic electron transport (Foyer and Halliwell 1976 ; Mullineaux and Creissen 1997 ). Thus, the degree of reduction of major antioxidant pools is generally considered to reflect the redox status of the tissue in question and is consequently an indicator of oxidative stress.

Glutathione, either as GSH or as GSSG (glutathione disulfide; oxidized glutathione), is regarded as a key component of antioxidant defenses in most aerobic organisms, including plants (Foyer et al. 1997 ). However, the high (i.e., millimolar) concentration of GSH in the chloroplast (Foyer and Halliwell 1976 ; Law et al. 1983 ; Bielawski and Joy 1986 ) is in apparent conflict with its proposed roles: the regeneration of ascorbate (Foyer and Halliwell 1976 ), reduction of lipid hydroperoxides (Mullineaux et al. 1998 ), and regulation of chloroplast gene expression by thiol-mediated modulation of RNA polymerase subunits and protein kinases (Link 1996 ).

GSH is synthesized from its constituent amino acids in an ATP-dependent two-step reaction catalyzed by the enzymes -glutamylcysteine (-EC) synthetase (-ECS; EC 6.3.2.2) and glutathione synthetase (GS; EC 6.3.2.3; Figure 1A). In plants, GSH biosynthesis occurs in the cytosol and the chloroplast, with at least one control point being the regulation of activity of -ECS (Hell and Bergmann 1990 ; Ruegsegger and Brunold 1993 ; Noctor et al. 1996 , Noctor et al. 1997 ). Additional regulation of GSH biosynthesis may be achieved by the supply of its constituent amino acids (Strohm et al. 1995 ; Noctor et al. 1997 ). Foliar GSH levels have been successfully raised by three- to fourfold in poplar transformed with the coding sequence of the -ECS gene (gshI) from Escherichia coli under the control of the cauliflower mosaic virus (CaMV) 35S promoter and overexpressing -ECS activity by at least 24-fold (Noctor et al. 1996 ). Conversely, poplar transformants overexpressing a transgene encoding E. coli GS did not show any increase in foliar GSH content (Strohm et al. 1995 ). In both sets of transgenic poplar, the products of the transgene were restricted to the cytosol.

View larger version (16K): [in this window] [in a new window]   Figure 1. Synthesis and Cycling of GSH. (A) Biosynthesis and cycling of GSH in higher plant chloroplasts (Foyer et al. 1997 ; Mullineaux and Creissen 1997 ). ASA, ascorbic acid; DHA, dehydroascorbate; DHAR, dehydroascorbate reductase; -EC, -glutamylcysteine; GR, glutathione reductase; GSH, reduced glutathione; GSHI, -glutamylcysteine synthetase; GSHII, glutathione synthetase; GSSG, oxidized glutathione. (B) Schematic diagrams of the T-DNAs of pGSH106 and pGSH209 harboring chimeric genes encoding E. coli gshI and gshII, respectively. TP is the sequence encoding the chloroplast transit peptide from the pea ribulose biphosphate carboxylase small subunit gene; 2x35S is the CaMV 35S promoter with doubled enhancer region; and polyA (for poly[A]) is the CaMV 19S polyadenylation sequence (Guerineau et al. 1988 ; see Methods). The chimeric genes were inserted into the T-DNA of pBIN19 (Bevan 1984 ), and nosNPTII is the kanamycin resistance gene used to select tobacco transformants. LB and RB refer to the left and right borders of the pBIN19 T-DNA, respectively. Transformants were designated as cpGSHI or cpGSHII lines.

The cellular levels of ROS must be tightly regulated in plants, as they are in mammalian cells (Meyer et al. 1993 ). At least part of this regulation is achieved by transcriptional control: mRNAs of several components of the ROS scavenging system are induced in response to many biotic and abiotic stresses (Criqui et al. 1992 ; Conklin and Last 1995 ; Grantz et al. 1995 ; Willekens et al. 1995 ; Karpinski et al. 1997 ). ROS in plants, particularly H₂O₂ and O₂^{• -}, are also implicated in programmed cell death as part of developmental processes and in the hypersensitive response to pathogens. This response is accompanied by local and systemic changes in the redox state of the foliar GSH pool and the induction of pathogenesis-associated genes, some of which encode antioxidant enzymes (Greenberg et al. 1994 ; Levine et al. 1994 ; Baker and Orlandi 1995 ; Jabs et al. 1996 ; May et al. 1996 ; Mittler et al. 1996 ; Pennell and Lamb 1997 ; Allen and Fluhr 1997 ).

Treatments of protoplasts, suspension cultures, or leaves of several different plant species with GSH induce marked changes in expression of genes encoding components of the ROS scavenging system (Herouart et al. 1993 ; Wingsle and Karpinski 1996 ; Karpinski et al. 1997 ). However, the effect of GSH is not uniform in different experimental systems. For example, treatment of bean suspension cultures with GSH strongly induces transcripts encoding chalcone synthase (CHS) and phenylalanine ammonia-lyase (PAL; Wingate et al. 1989 ), whereas pretreatment of Arabidopsis leaves with GSH before exposure to photoinhibitory light intensities results in a failure to induce APX transcripts that normally increase greatly in response to this stress (Karpinski et al. 1997 ).

Taken collectively, these data provide circumstantial evidence that in plants, the regulation of nuclear gene expression is influenced by ROS and the redox states of antioxidant pools, although to our knowledge, no regulatory factor has been isolated from plants that is specifically redox modulated. Furthermore, considerations concerning ROS homeostasis in cells are complicated by the membrane-permeable nature of some ROS, such as H₂O₂. For example, this means that increased oxidative stress in the chloroplast could have a regulatory impact on the maintenance of ROS levels in other subcellular compartments (Asada 1994 ; Karpinski et al. 1997 ).

In this study, we report the outcome of a set of experiments with transgenic tobacco plants in which the transgene-expressed E. coli -ECS and GS were fused to the chloroplast transit peptide sequence from the small subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase from pea, which ensures efficient targeting to the chloroplast stroma (Anderson and Smith 1986 ; Guerineau et al. 1988 ; Creissen et al. 1995 ). In these experiments, our goals were to raise the level of GSH in chloroplasts and to determine whether doing so raised the tolerance of the plants to photooxidative stress and decreased their susceptibility to photoinhibition, thus providing a direct link among plastid redox status, ROS content, and the efficiency of photosynthesis under adverse conditions. Paradoxically, however, increased GSH biosynthetic capacity in the chloroplast resulted in greatly enhanced sensitivity to oxidative stress. Our data indicate that a cycle of oxidative damage is established that is caused by the effects of ROS accumulation on photosynthesis. This cycle is possibly initiated by a failure of the chloroplast homeostatic mechanism that balances ROS scavenging with ROS production during photosynthesis.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Phenotypes of Primary (T₁) Transgenic Tobacco Plants Expressing Chloroplast-Targeted GSHI or GSHII
Plants overexpressing chloroplast GSHI (cpGSHI) and cpGSHII (Figure 1B) were recovered as kanamycin-resistant plantlets from tissue culture after leaf disc transformation with pGSH106 and pGSH209, respectively, and placed under greenhouse conditions (see Methods). Immunodetection of GSHI and GSHII in purified chloroplast fractions from transformed plants confirmed that the E. coli GSHI and GSHII proteins were targeted to the chloroplast (Figure 2A and Figure 2B). Whereas cpGSHII lines were phenotypically indistinguishable from wild-type tobacco, the five independent cpGSHI lines that were recovered developed severe necrotic lesions on their leaves, a mild example of which is shown in Figure 3A (cpGSHI-24). In all cases, the necrosis developed rapidly as the leaf aged (Figure 3A); by the time the leaf in a wild-type plant would have been fully expanded, the necrotic patches on the cpGSHI-overexpressing leaves had fused and the leaves had withered. All of the cpGSHI lines were sterile under greenhouse conditions. At an earlier stage in their growth, leaf discs were cut from young prenecrotic leaves of all the cpGSHI lines and placed back in tissue culture to recover clonal propagants. These clonally propagated plants, once rooted, were placed in a controlled-environment room (see Methods). Under these conditions, the cpGSHI lines were less severely affected than were those grown under greenhouse conditions (Figure 3B). Nevertheless, as the leaves developed, they displayed spreading chlorosis, which eventually covered the leaf. Necrotic patches subsequently appeared. However, under the controlled-environment conditions, all of the lines were fertile, and thus, most subsequent work described here was performed with controlled-environment-grown plants.

View larger version (57K): [in this window] [in a new window]   Figure 2. Localization of Transgene-Encoded -ECS and GS in the Chloroplast. (A) Immunodetection of -ECS in purified chloroplast fractions from progeny of cpGSHI-24 x cpGSHII-1. (B) Immunodetection of GS in purified chloroplast fractions from progeny of cpGSHI-24 x cpGSHII-1. Chloroplast fractions (20 µg of protein per lane) free of any detectable cytosol contamination were resolved by SDS-PAGE, and -ECS and GS were detected on gel blots by using antisera raised against recombinant E. coli GSHI and GSHII (Arisi et al. 1997 ). Lane 1, wild type; lane 2, cpGSHI-24 x cpGSHII-1; and lane 3, cpGSHII-1. Numbers at left indicate apparent molecular weight in kilodaltons.

View larger version (41K): [in this window] [in a new window]   Figure 3. Phenotypes of Primary cpGSHI Transformants. (A) Necrotic lesions on leaves from the primary transgenic line cpGSHI-24 grown under greenhouse conditions (700 to 900 µmol m-2 sec-1). The leaves are from successive nodes (oldest at top left and youngest at bottom right). (B) A tissue culture–derived wild-type regenerant and primary transgenic tobacco lines cpGSHI-8, cpGSHI-22, cpGSHI-26, cpGSHI-24, and cpGSHI-23 (clockwise from top left), grown under controlled-environment conditions (see Methods). The plants were regenerated from leaf explants from greenhouse-grown primary transformants, rerooted, and placed in the controlled-environment room for 3 weeks (see Results).

Leaves of controlled-environment-grown cpGSHI lines had up to twofold higher levels of foliar H₂O₂ when compared with control material (Table 1), which confirmed that these plants were subjected to increased oxidative stress. In the greenhouse in July 1995, when the first transfer of T₁ cpGSHI lines from tissue culture occurred, the light intensity (photosynthetically active photon flux density [PPFD]) fluctuated between 700 and 900 µmol m^-2 sec^-1, whereas in the controlled environment, the PPFD was a constant 250 µmol m^-2 sec^-1 (1 m from the lights). Therefore, we hypothesized that the cpGSHI lines were sensitive to high light intensities and that this sensitivity was manifested as oxidative damage to their chloroplasts, which induced extreme necrosis in the greenhouse and less severe chlorosis in the controlled environment.

View this table: [in this window] [in a new window]   Table 1. GSH, -EC, and H2O2 Levels and -ECS Activity in Leaves of Primary cpGSH Transgenic Linesa

Expression of cpGSHI and Measurement of Foliar GSH Levels
All five controlled-environment-grown cpGSHI lines had readily detectable levels of cpGSHI transcripts, immunodetectable levels of E. coli -ECS protein in their chloroplasts (data not shown), and up to 85-fold increased levels of -ECS activity (Table 1). GSH and -EC levels in leaves were elevated by up to five- and 25-fold, respectively (Table 1). Foliar cysteine levels were unchanged compared with control plants (data not shown). Transgenic plants expressing cpGSHII showed no changes in foliar GSH or -EC levels when compared with wild-type controls (data not shown).

The oxidative damage in the cpGSHI lines was associated with an increased capacity to synthesize GSH and was the opposite of what was anticipated (see Introduction). We hypothesized further that the greatly increased foliar levels of -EC, a molecule that normally does not accumulate to any degree in unstressed plant tissue (Noctor et al. 1996 ; Polle 1996 ), may have been the cause of this phenotype.

Generation of cpGSHI x cpGSHII Hybrid Transgenic Tobacco Lines
We reasoned that the high levels of -EC in leaves of cpGSHI transgenic lines could perhaps be lowered and that GSH levels could be increased concomitantly by crossing these lines with a tobacco line overexpressing GS in the chloroplast (cpGSHII-1). Such crosses were performed with cpGSHI-8 and cpGSHI-24, two cpGSHI lines that represented extremes in the severity of the symptoms (see Figure 3B). Progeny were screened using protein gel blotting (see Methods) for those individuals containing cpGSHI, cpGSHII-1, both transgenes (cpGSHI x cpGSHII-1 hybrids, hereafter termed cpGSHI/II hybrids), or no transgene (the wild type). Both crosses produced cpGSHI/II hybrid progeny that showed a distinct amelioration of symptoms compared with their cpGSHI siblings, although a wild-type phenotype was not achieved in the cpGSHI/II hybrid groups (Figure 4A and Figure 4B).

View larger version (64K): [in this window] [in a new window]   Figure 4. Progeny Resulting from cpGSHI-8 x cpGSHII-1 and cpGSHI-24 x cpGSHII-1 Crosses. (A) cpGSHI-8 x cpGSHII-1. (B) cpGSHI-24 x cpGSHII-1. All plants were 4 weeks old and were grown under controlled-environment conditions. Genotypes clockwise from bottom left in (A) and (B) are double transgene recessive (wild type), cpGSHII-1, cpGSHI, and cpGSHI/cpGSHII-1 hybrid. The genotypes of individual progeny from these crosses were identified by immunoblot screening by using antisera specific to the E. coli -ECS (in cpGSHI-lines) and GS (in cpGSHII-1) (see Methods; Arisi et al. 1997 ).

GSH Determinations in the cpGSHI/II F₂ Hybrid Progeny
The cpGSHI/II hybrid progeny from crosses with the two different cpGSHI lines had up to sixfold elevated levels of foliar glutathione (GSH and GSSG) than their wild-type recessive siblings and up to threefold higher levels than those siblings expressing only the extra chloroplast -ECS (cpGSHI alone; Figure 5). Similarly, the levels of total -EC (-EC plus di--glutamylcystine [ESSE]) were higher than those in cpGSHI siblings. However, there was also a marked increase in the redox states of these two thiol pools in both sets of cpGSHI/II hybrids compared with their cpGSHI siblings. Thus, the hybrids contained more GSH and -EC and less GSSG and ESSE than did their cpGSHI siblings (Figure 5). Therefore, amelioration of the oxidative stress symptoms in cpGSHI/II hybrids compared with their cpGSHI counterparts was associated with changes in the redox status of the GSH and -EC pools in favor of their reduced forms. Cysteine pools were not affected by any combination of transgenes in progeny arising from these crosses (data not shown).

View larger version (31K): [in this window] [in a new window]   Figure 5. Foliar GSH (GSH plus GSSG) and -EC (-EC plus ESSE) in cpGSHI x cpGSHII-1 Hybrids. (A) Total GSH and -EC in progeny of cpGSHI-8 x cpGSHII-1. (B) Total GSH and -EC in progeny of cpGSHI-24 x cpGSHII-1. Values are in nanomoles per gram fresh weight (fwt) for GSH plus GSSG (left) and for -EC plus ESSE (right). The redox states of these compounds (GSH/[GSH + GSSG]; -EC/[-EC + ESSE]) and the genotype of the progeny are shown beneath the histograms. The histograms are the means from at least three leaf samples from four individual plants (a minimum of n = 12) of each progeny genotype, and the vertical lines are the standard error of their means. GSH and -EC were determined from neutralized acid extracts by derivitization with monobromobimane (Newton et al. 1981 ), separation of the products by reversed-phase HPLC, and quantification using an on-line fluorometer. Aliquots of the neutralized extracts were treated with DTT to reduce oxidized thiols and then treated as described. For more details, see Methods. WT, wild type.

Generation of Hybrids between GR Transformants and cpGSHI Plants
To determine whether it was the change in the redox state of both or only one of the thiol pools in cpGSHI/II hybrids compared with their cpGSHI siblings that was responsible for the amelioration of the chlorotic symptoms, we crossed, as a pollen donor, a transgenic tobacco plant harboring up to 10-fold higher levels of GR in its chloroplasts (tobacco line GR46-27; Creissen et al. 1995 ) with homozygous cpGSHI-8 and cpGSHI-24. In GR46-27 x cpGSHI progeny from the two crosses, there was no amelioration of chlorotic symptoms, although there was an elevation of the redox state of the GSH pool by a minimum of 40% compared with cpGSHI controls (Table 2). In vitro, GR was only inhibited <10% by -EC or ESSE at concentrations 10-fold in excess of those encountered in the cpGSHI leaves; also, GR was not able to use ESSE as a substrate (data not shown). Furthermore, extractable GR activity from leaves of cpGSHI lines was at least equivalent to that from their wild-type siblings (data not shown). Therefore, GR was unlikely to be inhibited in vivo by the enhanced levels of -EC or ESSE.

Photosynthesis and Chloroplast Integrity
The sensitivity of the cpGSHI-containing transgenic lines to light intensity was confirmed when the performance of photosynthesis (measured as light-dependent CO₂ fixation) was measured as a function of PPFD (Figure 6). For plants grown in controlled-environment conditions, the saturation of CO₂ fixation occurred at ~800 µmol m^-2 sec^-1 for both wild-type (transgene recessive) and cpGSHII-1 progeny (Figure 6). These light saturation curves are typical of plants grown under low-light conditions (Baker 1994 ). cpGSHI progeny showed a much-depressed capacity for photosynthesis as revealed by the saturation of CO₂ fixation at a PPFD of ~250 µmol m^-2 sec^-1, and saturation of CO₂ fixation in the cpGSHI/II hybrids occurred at ~500 µmol m^-2 sec^-1 (Figure 6). Without exception, progeny from the cpGSHI-8 x cpGSHII-1 cross showed an identical pattern of response in these experiments to that of the cpGSHI-24 x cpGSHII-1 hybrid group shown in Figure 6. It is important to note that the leaf disc samples for the photosynthesis experiments were taken from just fully expanded leaf sections that did not display any chlorosis. Thus, the poor photosynthetic performance of the cpGSHI lines was not an immediate consequence of loss of chlorophyll, although subsequent measurements of photosynthetic parameters in chlorotic tissues displayed a yet further worsening of chloroplast function (data not shown).

View larger version (18K): [in this window] [in a new window]   Figure 6. CO2 Assimilation in Response to Changes in Incident PPFD. Response of CO2 assimilation (A), corrected for dark respiration, to PPFD was measured in the third leaves of wild-type, cpGSHI, cpGSHII, and cpGSHI/II plants. Data points are the means of five replicates, and standard errors were below 10% of these means in all cases. WT, wild type.

Ultrastructural analysis of chloroplasts in green sectors of cpGSHI lines (Figure 7A and Figure 7B show cpGSHI-8) revealed the presence of spherical bodies sandwiched between the stroma thylakoids of otherwise normal-looking chloroplasts (cf. Figure 7A and Figure 7C). The nature of the spherical bodies has not been investigated further. Similar images in chlorotic and necrotic tissues revealed swelling and then deterioration in chloroplast grana, and under these conditions the spherical bodies were no longer evident (Figure 7B).

View larger version (63K): [in this window] [in a new window]   Figure 7. Transmission Electron Microscopy of Chloroplasts from Leaf Tissue of cpGSHI-8 and Wild-Type Tobacco. (A) Chloroplast from a cpGSHI-8 green leaf showing presence of spherical vesicles (v) in the stroma between stromal thylakoids. (B) Chloroplast from chlorotic tissue of cpGSHI-8 showing swollen granum thylakoids (g). (C) Chloroplast from healthy leaf tissue of tobacco cultivar Samsun. Bars = 0.5 µm.

Expression of Genes Encoding Components of the ROS Scavenging System and Pathogenesis-Associated Genes in the cpGSHI Lines
The oxidative stress phenotype and elevated foliar H₂O₂ levels suggest that the foliar transcript levels of several stress-responsive genes encoding proteins important in pathogen responses and parts of the plant's antioxidant defenses should be elevated (see Discussion). Therefore, the cpGSHI lines grown in controlled-environment conditions were screened for the levels of the transcripts that encode Fe–SOD, Mn–SOD, Cu/Zn SOD (plastidial and cytosolic isoforms), GR (plastidial and cytosolic isoforms), peroxisomal catalase, phospholipidhydroperoxide (PH)–GPX, GST/GPX, APX (cytosolic isoform), monodehydroascorbate free radical reductase, CHS, PAL, and pathogenesis-related protein 1 (PR-1a). Surprisingly, the levels of none of these transcripts were elevated (data not shown). However, in four independent experiments in which plants were dark adapted for 72 hr and subsequently exposed to the controlled-environment light levels, Fe–SOD transcript accumulation was slower in cpGSHI-24 than in the wild type (Figure 8). In a more limited experiment with greenhouse-grown material displaying severe necrotic symptoms, the PR-1a transcript was substantially induced (Figure 9).

View larger version (19K): [in this window] [in a new window]   Figure 8. Induction of Fe–SOD Transcripts in Wild-Type Tobacco and cpGSHI-24. Six-week-old plants were maintained in darkness for 72 hr before reexposure to the light under controlled-environment conditions. Samples were collected for RNA isolation at 0, 6, 24, and 30 hr. Slot blots (15 µg RNA per slot) were probed with an Fe–SOD fragment from N. plumbaginifolia, and hybridization signals were quantified by using a PhosphorImager. Equal RNA loading was confirmed by probing for rRNA. Data are shown for four independent experiments and are expressed relative to wild-type transcript levels at the start of the light period. Open bars are wild type, and solid bars are cpGSHI-24.

View larger version (51K): [in this window] [in a new window]   Figure 9. PR-1a Transcript Is Induced in Necrotic Tissue from the Leaves of T3 Progeny of cpGSHI Lines. RNA gel blot of 15 µg of total RNA from leaves of each of the lines cpGSHI-8, cpGSHI-24, and cpGSHI-26 (lanes 2, 4, and 6, respectively) or their recessive siblings (lanes 1, 3, and 5), respectively. RNA was subjected to electrophoresis through a denaturing agarose gel, blotted onto a nylon membrane, and probed with a 32P-labeled PR-1a DNA fragment generated by polymerase chain reaction (PCR) amplification of tobacco genomic DNA. The blot was washed at high stringency, and the bands were visualized by autoradiography.

Effects on Catalase Expression
Tobacco plants that have much reduced peroxisomal catalase activity display a light intensity–dependent necrosis and induction of PR-1a very similar to the symptoms described here for the plants overexpressing -ECS in their chloroplasts (see Discussion; Willekens et al. 1997 ). Therefore, we investigated whether the phenotype in the cpGSHI-transformed plants could have been caused by an impairment of catalase activity. Foliar catalase activity was approximately twofold higher in the cpGSHI lines compared with their recessive siblings (Table 3). Therefore, the light intensity–dependent oxidative stress described here was not caused by an inhibition of catalase activity, in keeping with the lack of perturbation of catalase mRNA levels (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

The photosynthetic apparatus is particularly susceptible to oxidative damage, despite an apparently well-developed plastidial ROS scavenging system (see Introduction). Therefore, we reasoned that any increased capacity of the chloroplast to synthesize a key antioxidant such as GSH would make leaves and, therefore, the plant less susceptible to oxidative stress. From the data presented in this study, clearly an increase in total foliar GSH and -EC can be achieved by targeting to the chloroplast extra -ECS activity alone (in cpGSHI lines) and even more by enhancing this activity with that of GS (in cpGSHI/II hybrid lines; Table 1 and Figure 5). Paradoxically, enhanced capacity for GSH biosynthesis in these transgenic plants was associated with a light-dependent chlorotic/necrotic phenotype (Figure 3 and Figure 4). The characteristics of these transgenic plants and considerations of how such properties came about are summarized in the following sections.

The Chlorotic/Necrotic Phenotypes Are a Consequence of Oxidative Stress
The most severe chlorotic/necrotic phenotypes were observed in those lines expressing only -ECS (i.e., cpGSHI lines), and the severity of the phenotype was ameliorated when extra GS activity was present in the cpGSHI/II hybrids (Figure 3 and Figure 4). The cpGSHI lines possessed enhanced foliar thiol pools (i.e., both GSH and -EC), which were primarily in the oxidized state (Figure 5); therefore, the severity of the symptoms was associated with the low redox state of the GSH and -EC pools.

Because the cpGSHI/II hybrids had more foliar GSH (GSH plus GSSG) and because this pool was in a more reduced state compared with their cpGSHI siblings (Figure 5), the possibility that the high levels of GSH were the cause of the chlorosis could be eliminated. This ruled out the inhibition of thiol-sensitive antioxidant enzymes such as plastidial APX and violaxanthin de-epoxidase (Asada 1994 ; Foyer and Harbinson 1994 ) as causes of the observed phenotype.

Foliar H₂O₂ levels in the cpGSHI lines were approximately double those in the wild-type material (Table 1). Furthermore, H₂O₂ levels in the cpGSHI/II hybrid progeny were intermediate between those of their wild-type and cpGSHI siblings (data not shown). The considerations listed above and these data indicate that these plants were suffering increased levels of oxidative stress. This could be brought about by either an increase in ROS production or a defect in ROS scavenging. These possibilities are not mutually exclusive.

Possible Cause of the Lowered Redox State of the GSH Pool in cpGSHI Transformants
The cpGSHI/II hybrid leaves contained up to twice the level of total GSH as did their cpGSHI siblings and yet this GSH was in a more reduced state (Figure 5). Thus, the low redox state of the GSH pool in cpGSHI plants cannot be due entirely to an inability to reduce any GSSG that is formed during normal cellular oxygen metabolism. We could find no evidence that GR activity was inhibited in the cpGSHI lines or that -EC or ESSE could be a substrate for or act as an inhibitor of the enzyme in vitro (see Results). The oxidative stress phenotype was not ameliorated in cpGSHI x GR46 hybrid progeny that contained up to 10-fold elevated levels of plastidial GR activity from a CaMV 35S–pea GR transgene (see Results; Broadbent et al. 1995 ; Creissen et al. 1995 ). This was despite a minimum of a 40% increase in the redox state of the GSH pool in these hybrids compared with cpGSHI controls (Table 2). We conclude that the lowered redox state of the GSH pool was not the direct initiator per se of the chlorotic/necrotic phenotype in the cpGSHI lines.

Possible Causes for the Accumulation of ESSE
-EC and ESSE also accumulated in the leaves of cpGSHI lines and in their cpGSHI/II siblings. Wild-type plants under nonstress conditions in the light do not accumulate foliar -EC to high levels compared with GSH and cysteine (Table 1 and Figure 5; Noctor et al. 1996 , Noctor et al. 1997 ; Polle 1996 ). To our knowledge, there are no reports that wild-type plants have detectable levels of ESSE, and we are not aware of any enzyme in plants that could reduce this compound back to -EC. Therefore, -EC may not participate in any redox cycle in the plant, and because it is a biosynthetic intermediate on the pathway to GSH (Figure 1A), the flux through this small pool must ensure that its oxidized form does not accumulate to any significant degree. Therefore, the only route open to removal of ESSE is possibly by degradation in a way described for -EC (Noctor et al. 1998 ).

The cpGSHI/II hybrids had greatly increased GSH levels compared with their cpGSHI siblings (Figure 5). This presumably reflects the fact that both transgene-encoded enzymes were at high enough concentrations to escape limitations of feedback regulation or that the E. coli–derived enzymes are not subject to the same type of control as the native tobacco plastid enzymes. This point, plus the higher redox state of the -EC pool in the cpGSHI/II hybrids compared with the cpGSHI plants, strongly suggests that the -EC pool in the cpGSHI/II hybrids is turned over more rapidly (as a consequence of increased GS activity), which may slow the rate at which this pool is oxidized.

Photosynthetic Performance of cpGSHI Dysfunctional Chloroplasts
The severity of the chlorotic/necrotic symptoms depended on the prevailing light intensity (growth PPFD; Figure 3) and was associated with a degeneration of chloroplasts as the symptoms developed (Figure 7). Chloroplasts from young, non-symptomatic leaves in the cpGSHI lines, and to a lesser extent in the cpGSHI/II hybrids, were less efficient at photosynthetic carbon fixation at growth PPFD, and this inefficiency increased as leaf discs were exposed to higher light intensities (Figure 6). Furthermore, chloroplasts from the young green leaves of such plants showed the accumulation of spherical bodies sandwiched between the stroma thylakoids, which disappeared as the chloroplasts degenerated (Figure 7). Although the nature of these spherical bodies is not known, they indicate that the chloroplasts of nonchlorotic leaves were different from those of wild-type plants even at this pre-symptomatic stage.

Decreased abilities for photosynthetic carbon fixation in the cpGSHI lines, and to a lesser extent in the cpGSHI/II hybrids, could have enhanced photosynthetic reduction of O₂ (the Mehler reaction; see Introduction). This would in turn cause more ROS production and further oxidation of the GSH pool, the combination of which would lead to still greater inhibition of carbon fixation. Thus, a downward spiral of oxidative destruction would be set in train, leading to premature senescence of the chloroplast and eventual death of the cell. Clearly, the higher the prevailing light conditions, the faster this destructive process would ensue—a situation that was readily observed in our experiments.

The Initiation of Oxidative Destruction of the Chloroplast: A Failure in Redox Sensing?
A striking feature of the cpGSHI transgenic plants grown under the standard controlled-environment conditions was that despite the clear oxidative stress, none of the transcripts encoding components of the antioxidant system in tobacco was elevated. These transcripts include those encoding catalase and SOD and APX isoforms as well as those encoding PH–GPX and GR isoforms. In various environmental conditions known to promote oxidative stress in tobacco or related Nicotiana spp, the levels of at least some of these transcripts have been shown to be increased (Bowler et al. 1989 ; Tsang et al. 1991 ; Criqui et al. 1992 ; Willekens et al. 1995 ; Mittler et al. 1996 ). Furthermore, the induction of Fe–SOD transcripts under conditions of oxidative stress, induced by transferring dark-adapted leaves to the light, was delayed in the cpGSHI lines compared with the wild type, but the Fe–SOD transcripts continued to accumulate in the cpGSHI lines to higher levels than did those in the wild-type control (Figure 8). It has previously been established that increases in Fe–SOD mRNA in such a transition is a consequence of increased internally generated oxidative stress in the chloroplast brought about by the onset of photosynthesis (Tsang et al. 1991 ). Thus, the altered pattern of accumulation of Fe–SOD mRNA in these plants undergoing a dark–light transition indicates that they may be disrupted in redox sensing. Such a mechanism would normally monitor ROS production throughout the life of the leaf and adjust the activity of the antioxidant system as appropriate.

These plants also display increased sensitivity to abiotic stresses such as ozone fumigation (Wellburn et al. 1998 ) and chilling (M. Fryer, G. Creissen, P. Mullineaux, and N. Baker, manuscript in preparation), which is consistent with this hypothesis. That GSH may play a role in redox sensing as well as the response to oxidative stress is supported by the observation that pretreatment of Arabidopsis leaves with GSH before the application of a photoinhibitory light stress renders photosynthesis more sensitive to the stress than control treatments and is associated with a marked failure to increase the levels of APX transcripts normally induced under this stress in this species (Karpinski et al. 1997 ).

From these experiments, it was proposed that changes in the redox state of the foliar GSH pool may be a factor in sensing oxidative stress. It has been suggested (Foyer et al. 1997 ) that the balance between ROS synthesis and destruction allows for some controlled oxidation, which may be important in signaling detrimental changes in the environment, thus permitting redirection of resources to minimize damage. Disruption of this balance in either direction may have similar consequences for the plant.

Comparison of cpGSHI Plants with Catalase-Suppressed Transgenic Plants
Transgenic tobacco plants with lowered levels of peroxisomal catalase activity show a light intensity–dependent necrotic phenotype that can be completely ameliorated by growing such plants under nonphotorespiratory conditions. Furthermore, these plants have higher foliar GSH levels that are in a more oxidized state than their wild-type counterparts, and in severely necrotic tissues the level of the PR-1a transcript is markedly induced (Chamnongpol et al. 1996 ; Willekens et al. 1997 ). These properties are clearly analogous to those described for the cpGSHI lines (Figure 3 Figure 4 Figure 5 and Figure 9, and Table 1). However, there are also fundamental differences in the two sets of transgenic plants that make it unlikely that a failure in peroxisome function is the cause of the oxidative stress phenotype displayed by the cpGSHI transgenic lines. These differences are as follows: (1) prenecrotic leaves of catalase-suppressed plants appear to photosynthesize normally in contrast to green sectors of cpGSHI lines (Figure 6; Willekens et al. 1997 ); (2) the chlorotic phenotype in the cpGSHI transgenic lines is associated with an increase in foliar H₂O₂ levels in contrast to the catalase-suppressed plants; and (3) there was no inhibition of catalase activity or reduction in the level of catalase transcripts in the cpGSHI-containing lines (Table 3).

The properties of the catalase-suppressed plants have led to the hypothesis that the peroxisome is a major sink for H₂O₂ in the leaves of C₃ plants, irrespective of its source (Willekens et al. 1997 ). If this is the case, then in the cpGSHI plants this sink failed to deal with the H₂O₂ under photorespiratory (i.e., ambient CO₂ concentrations) conditions.

Comparison of cpGSHI Plants with Lesion Mimic Mutants
The observed symptoms of the cpGSHI transgenic lines superficially show parallels to lesion mimic mutants, in particular the lsd (for lesion simulating disease) and acd (for accelerated cell death) mutants of Arabidopsis (Deitrich et al. 1994 ; Greenberg et al. 1994 ; Jabs et al. 1996 ). The lsd mutants grown under extended photoperiods and the transgenic tobacco grown under controlled-environment conditions show a similar spreading chlorotic phenotype, and both are clearly perturbed in ROS metabolism (Deitrich et al. 1994 ; Jabs et al. 1996 ). However, in the cpGSHI lines, none of the transcripts normally associated with responses to pathogens (e.g., those encoding PR-1a, CHS, and PAL) has higher levels in these chlorotic tissues (see Results). Only under conditions of severe necrosis in greenhouse-grown material did we observe an increase in PR-1a transcript levels (Figure 9). Although in the cpGSHI lines we have not investigated other possible properties of lesion mimic mutants, we consider the failure to observe any increase in pathogenesis-associated mRNAs in chlorotic tissue as an indication that this comparison is at best limited. Whereas a role for oxidative stress in the chloroplast has been ruled out as initiating the phenotype of the lsd mutants in Arabidopsis (Jabs et al. 1996 ), the recent demonstration that H₂O₂ levels in tobacco chloroplasts transiently increase during initial contact with a pathogen (Allen and Fluhr 1997 ) suggests that the role of the chloroplast in the early stages of the hypersensitive response might be more prominent than hitherto acknowledged.

	METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Polymerase Chain Reaction Amplification and Cloning of gshI and gshII Coding Sequences from Escherichia coli B DNA
Cloning of the genes encoding -glutamylcysteine synthetase (-ECS) and glutathione synthetase (GS), gshI and gshII, respectively, was achieved from polymerase chain reaction (PCR)–amplified DNA by using the following specific primers. For gshI, the primers were 5'-CATGATGTGGTGGCACTAATTGTAG-3' (nucleotide positions 247 to 271; Watanabe et al. 1986 ) and 5'-CTGTCAGGCGTGTTTTTCCAGCCAC-3' (nucleotide positions 1897 to 1873). For gshII, the primers were 5'-TGATTGGCCCGGAAGGCGGTTTATC-3' (nucleotide positions 196 to 220; Gushima et al. 1984 ) and 5'-TCAGAGTCTCAACGAGATCCTTCTC-3' (nucleotide positions 1352 to 1328). The PCR conditions were 40 cycles of 30 sec at 92°C, 30 sec at 55°C, and 90 sec at 72°C, using 1 ng of E. coli B DNA (Sigma) and 15 pmol of each primer in a 100-µL reaction. At the end of the cycles, the reactions were incubated at 72°C for 20 min. PCR fragments of the predicted lengths (1.65 kb for gshI and 1.15 kb for gshII) were eluted from a 1% (w/v) agarose gel and ligated into the dideoxy T-tailed EcoRV site of pBluescript II KS+ (Stratagene, La Jolla, CA; Alting-Mees and Short 1989 ) by using standard in vitro recombination techniques (Sambrook et al. 1989 ). The plasmid-borne cloned gshI and gshII genes were confirmed to be functional by their ability to complement the corresponding gshI- or gshII-defective E. coli mutants (Apontoweil and Berends 1975 ), permitting their growth on minimal medium containing 100 µg mL^-1 tetramethylthiuramdisulfide (TMTD). TMTD is detoxified in bacteria by its conjugation with glutathione (GSH; Apontoweil and Berends 1975 ). To facilitate further manipulations, an SphI site (GCATGC) was introduced by site-directed mutagenesis such that the underlined sequence provided the translation initiation codon of both genes (Gushima et al. 1984 ; Watanabe et al. 1986 ). The mutants were verified by sequencing, and a repeat of the complementation tests was performed, as described above. The plasmids harboring the SphI-containing gshI and gshII genes were designated pGSH1-S and pGSH2-S, respectively.

Gene Constructs, Plant Material, and Transformation
Two binary vectors were constructed. The coding sequence of gshI of pGSH1-S was fused to the rbcS sequence encoding the chloroplast transit peptide from the small subunit of ribulose-1,5-bisphosphate carboxylase/oxygenase from pea and placed under the control of the cauliflower mosaic virus (CaMV) 35S promoter with duplicated enhancer region and CaMV polyadenylation sequences (pJIT117; Guerineau et al. 1988 ) to create chloroplast GSHI (cpGSHI). The transit peptide and GSHI coding sequences were fused via their respective SphI sites. This chimeric gene construct was ligated into the polylinker of pBIN19 (Bevan 1984 ) to produce pGSH106 (Figure 1B). A similar construct (pGSH209) containing the gshII coding sequence fused to the rbcS transit peptide sequence (cpGSHII) and placed in pBIN19 was made (Figure 1B). Both constructs were mobilized into Agrobacterium tumefaciens LBA4404 by the triparental mating procedure (Ditta et al. 1980 ).

Plants (Nicotiana tabacum cv Samsun NN) were maintained in a greenhouse with supplementary lighting to provide an 18-hr day length. Production of transgenic tobacco was achieved by leaf disc cocultivation, as described elsewhere (Creissen et al. 1995 ). Transgenic plants were grown either in the greenhouse or in a controlled-environment room (16-hr day at 25°C and at 80% relative humidity, with a 250 µmol m^-2 sec^-1 photosynthetically active photon flux density [PPFD]).

RNA Extraction and Analysis
RNA was purified from deproteinized leaf extracts by lithium chloride precipitation, as described elsewhere (Creissen and Mullineaux 1995 ). For RNA gel blot analysis, total RNA (15 µg) was resolved on 1% (w/v) agarose denaturing (formaldehyde) gels. RNA was transferred to nylon membranes (Hybond N; Amersham, Little Chalfont, UK) by capillary blotting, and hybridizations were performed using standard procedures (Sambrook et al. 1989 ). Filters were washed at high stringency (0.1 x SSC [1 x SSC is 0.15 M NaCl and 0.015 M sodium citrate] and 0.1% SDS at 65°C) for homologous probes and at reduced stringency (2 x SSC and 0.1% SDS at 65°C) for heterologous probes before visualization by autoradiography or by use of a PhosphorImager (Fuji Photofilm, Kanagawa, Japan).

GSH, -Glutamylcysteine, and H₂O₂ Determinations
Thiols were determined as their monobromobimane (MB)-derivatized products (Newton et al. 1981 ). Leaf discs for determination of GSH and -glutamylcysteine (-EC) were cut from 6-week-old plants (nodes 4 to 6 counted from the top of the plant), immediately frozen in liquid nitrogen, and stored at -80°C until analysis was performed. Leaf material (50 mg) was extracted with 1 mL of 0.1 M HCl and centrifuged in a microcentrifuge (5 min at 4°C). Cleared supernatants were neutralized by adding an equal volume of 0.1 M NaOH, and aliquots (50 µL) were derivatized by adding to 1.5-mL tubes containing 35 µL of water, 10 µL of Tris-HCl, pH 8.0, and 5 µL of 10 mM MB (Calbiochem). Samples were incubated at room temperature for 15 min before the addition of 900 µL 5% acetic acid. For determining total (reduced and oxidized) thiols, extracts were first reduced by the addition of DTT (2.5 mM final concentration), and subsequently the MB final concentration used was 5 mM.

Separation of MB-derivatized thiols was achieved by C18 reversed-phase HPLC, and their detection using an on-line fluorometer was according to the method of Newton et al. 1981 . Identification and quantification of thiol compounds were performed by comparison of retention times of known amounts of standard compounds derivatized with MB. GSH was obtained from Boehringer Mannheim, and -EC was a kind gift from C. Foyer (IGER, Aberystwyth, UK). Di--glutamylcystine (ESSE) was prepared by incubation of a 15-mM solution of -EC with 200 mM H₂O₂ for 48 hr at 37°C, at the end of which time there was no detectable thiol by reaction with 5,5'-dithiobis(2-nitrobenzoic acid) (Smith et al. 1989 ). The ESSE was freeze-dried and used to prepare standard solutions.

We also used amperometric detection of GSH and GSSG after separation on a Dionex (Sunnyvale, CA) HPLC, primarily to determine the redox state of the GSH pool in cpGSHI x GR46 hybrid samples (see Results and Table 2). The method is derived from application note 110 from Dionex and is as follows. Leaf samples (0.05 g) were harvested and plunged into liquid nitrogen and then ground in 1.5% (v/v) acetonitrile/0.1 N perchloric acid, left on ice for 30 min, and centrifuged in a chilled microcentrifuge at top speed for 10 min. The supernatant was filtered through a 0.2-µm Anotop 10/1C inorganic membrane filter (Whatman). Filtered supernatant (0.1 mL) was recovered, and 0.05 mL was injected onto the HPLC column. Separation of GSH and GSSG was achieved by reversed-phase HPLC by using a Dionex DX500 BioLC machine fitted with a Zorbax (Dionex, Camberley, UK) RP300-C18 column (150 x 4.6 mm) and a mobile phase of 1.5% (v/v) acetonitrile in 0.1 N perchloric acid at a flow rate of 0.8 mL/min. Integrated amperometric detection was used with a gold working electrode and sodium reference electrode. Integrated peak areas were compared with those from 30 to 300 pmol GSH and GSSG standards to calculate amounts of these compounds in leaf extracts. This method could not be used to separate -EC from cysteine and thus was not applied for these purposes.

H₂O₂ was determined by a modified fluorescence procedure (Guilbault et al. 1967 ). Leaf discs (19-mm diameter) were cut and immediately frozen in liquid nitrogen. The tissue was extracted in 500 µL of 25 mM HCl in a mortar and centrifuged at 5000g for 5 min at 4°C. Pigments were removed by vortexing in the presence of activated charcoal and centrifuging to remove most of the charcoal, and then 200 µL of the supernatant was centrifuged through a Wizard mini-column (Promega, Madison, WI) to remove any remaining particulate matter. For the assay, 50 µL of the cleared extract was mixed in a 3-mL fluorescence cuvette with 2.89 mL of 50 mM Hepes, pH 7.5, and 30 µL of 50 mM homovanillic acid in 50 mM Hepes, pH 7.5. The reaction was started by the addition of 30 µL of 4 µM horseradish peroxidase. The H₂O₂ concentration was obtained by measuring relative fluorescence (excitation of 315 nm; emission of 425 nm) against a standard curve.

Measurement of the Activities of -ECS, GSH Reductase, and Catalase
Measurements of extractable -ECS activity were conducted in an anaerobic chamber on samples collected from controlled-environment-grown plants and frozen in liquid nitrogen. Preweighed leaf discs were ground in a mortar in 0.1 M Tris-HCl, pH 7.5, 5 mM EDTA (10 mL g^-1) with 1 g g^-1 polyvinylpolypyrrolidone and acid-washed sand. Cleared supernatant was obtained by centrifuging twice for 15 min at 15,000g. The -ECS assay was performed in a 500-µL reaction containing 0.1 M Tris-HCl, pH 7.5, 0.1 M MgCl₂, 10 mM ATP, 50 mM Na-L-glutamate, 8 mM phosphocreatine, 2 units of creatine phosphokinase, and 200 µL of cleared supernatant. The mixture was incubated under anaerobic conditions at 37°C for 10 min before the reaction was started by adding 50 µL of 20 mM L-cysteine. L-cysteine was dissolved in N₂-sparged water just before use. Samples (50 µL) were collected at 10-min intervals, the reaction was stopped, and the products were derivatized with MB and separated and quantified by HPLC, as described above. -ECS activity was quantified as nanomoles of -EC formed per minute per gram fresh weight leaf material.

GSH reductase activity was measured as described by Creissen et al. 1995 , and catalase activity was determined by the method of Aebi 1984 .

Chloroplast Isolation
Chloroplasts were isolated as previously described (Creissen et al. 1995 ). The absence of cytosol contamination of chloroplast fractions was confirmed by assaying for the cytosol-specific markers pyrophosphate-dependent phosphofructokinase (Journet and Douce 1985 ) and glucose-6-phosphate dehydrogenase (Doehlert et al. 1988 ). Chloroplast fractions that were free of detectable cytosol marker enzyme activity were used for immunodetection of the transgene-encoded proteins, as described below.

Immunodetection of cpGSHI and cpGSHII in Foliar Extracts
Cell-free protein extracts were prepared as described above but without using an anaerobic chamber. Total protein (10 to 30 µg) was subjected to SDS-PAGE and electroblotted to nitrocellulose; the use of polyclonal antibodies to immunodetect specific polypeptides has been described previously (Stevens et al. 1997 ). Polyclonal antibodies raised against the purified E. coli–derived -ECS and GS polypeptides were used at 1:1000 dilution to detect the corresponding proteins in the plant extracts and did not cross-react against any plant-derived polypeptide (Arisi et al. 1997 ). The antisera were kind gifts from A.-C.M. Arisi (INRA, Versailles, France) and C.H. Foyer (IGER, Aberystwyth, UK).

Photosynthetic Gas Exchange Measurements
Net photosynthetic CO₂ uptake was measured on third leaves, essentially as described previously by Bongi and Long 1987 , in an assimilation chamber on a small area of the leaf surface (2.0 cm²), illuminated by a quartz-iodide source. The light incident on the leaf was varied by interposing neutral density filters. The PPFD at the leaf surface was determined by placing a quantum sensor below the chamber window in the position normally occupied by the leaf (LI-190 SR; Li-Cor, Lincoln, NE). Measurements were started after net CO₂ uptake reached a steady state in air containing 360 ± 10 µmol mol^-1 CO₂ at a temperature of 24°C. Leaf gas exchange parameters were calculated according to the equations of Von Caemmerer and Farquar 1981 , and light dose–response curves were constructed for intensities between 100 and 1200 µmol m^-2 sec^-1 incident PPFD.

Ultrastructural Studies
Leaf tissue fixation and preparation for electron microscopy were performed as previously described (Wellburn and Wellburn 1994 ).

	FOOTNOTES

² It is with deep regret that we report that Alan Wellburn died on May 8, 1999.

	ACKNOWLEDGMENTS

We are grateful to Drs. Ana Arisi and Christine Foyer for the gifts of -ECS and GS antisera and to the latter for the gift of purified -EC. We thank Drs. Rod Casey, Carolina Escobar, and Ana Jiménez for critical reading of the manuscript. This work was supported by the Biotechnology and Biological Sciences Research Council (BBSRC) Global Environmental Change Initiative, the Commission of the European Communities (CEC) Agro-Industrial Research (AIR) Programme (Contract No. AIR1-CT92-0205), and the BBSRC Core Strategic Grant to the John Innes Centre. G.P. is supported by a CEC AIR Fellowship.

Received January 11, 1999; accepted April 6, 1999.

	REFERENCES

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Aebi, H. (1984) Catalase in vitro. Methods Enzymol. 105:121-126[ISI][Medline].

Allen, A.C., and Fluhr, R. (1997) Two distinct sources of elicited reactive oxygen species in tobacco epidermal cells. Plant Cell 9:1559-1572[Abstract].

Alting-Mees, M.A., and Short, J.M. (1989) pBluescript II, gene mapping vectors. Nucleic Acids Res. 17:9494[Free Full Text].

Anderson, S., and Smith, S.M. (1986) Synthesis of the small subunit of ribulose-bisphosphate carboxylase from genes cloned into plasmids containing the SP6 promoter. Biochem. J. 240:709-715[ISI][Medline].

Apontoweil, P., and Berends, W. (1975) Isolation and initial characterization of glutathione-deficient mutants of Escherichia coli K12. Biochim. Biophys. Acta 399:10-22[Medline].

Arisi, A.-C.M., Noctor, G., Foyer, C.H., and Jouanin, L. (1997) Modification of thiol contents in poplars (Populus tremula x P. alba) overexpressing enzymes involved in glutathione synthesis. Planta 203:362-372[CrossRef][ISI][Medline].

Asada, K. (1994). Production and action of active oxygen species in photosynthetic tissues. In Causes of Photooxidative Stress and Amelioration of Defense Systems in Plants, C.H. Foyer and P.M. Mullineaux, eds (Boca Raton, FL: CRC Press), pp. 77–104.

Baker, C.J., and Orlandi, E.W. (1995) Active oxygen in plant pathogenesis. Annu. Rev. Phytopathol. 33:299-321[CrossRef][ISI].

Baker, N.R. (1994). Chilling stress and photosynthesis. In Causes of Photooxidative Stress and Amelioration of Defense Systems in Plants, C.H. Foyer and P.M. Mullineaux, eds (Boca Raton, FL: CRC Press), pp. 127–154.

Bevan, M. (1984) Binary Agrobacterium vectors for plant transformation. Nucleic Acids Res. 12:8711-8721[Abstract/Free Full Text].

Bielawski, W., and Joy, K.W. (1986) Reduced and oxidised glutathione and glutathione-reductase activity in tissues of Pisum sativum. Planta 169:267-272[CrossRef].

Bongi, G., and Long, S.P. (1987) Light dependent damage to photosynthesis in olive leaves during chilling and high-temperature stress. Plant Cell Environ. 10:241-249.

Bowler, C., Alliotte, T., De Loose, M., Van Montagu, M., and Inzé, D. (1989) The induction of manganese superoxide dismutase in response to stress in Nicotiana plumbaginifolia. EMBO J. 8:31-38[ISI][Medline].

Broadbent, P., Creissen, G.P., Kular, B., Wellburn, A.R., and Mullineaux, P.M. (1995) Oxidative stress responses in transgenic tobacco containing altered levels of glutathione reductase activity. Plant J. 8:247-255[CrossRef].

Chamnongpol, S., Willekens, H., Langebartels, C., Van Montagu, M., Inzé, D., and Van Camp, W. (1996) Transgenic tobacco with a reduced catalase activity develops necrotic lesions and induces pathogenesis-related expression under high light. Plant J. 10:491-503[CrossRef].

Conklin, P.L., and Last, R.L. (1995) Differential accumulation of antioxidant mRNAs in Arabidopsis thaliana exposed to ozone. Plant Physiol. 109:203-212[Abstract].

Creissen, G.P., and Mullineaux, P.M. (1995) Cloning and characterisation of glutathione reductase cDNAs and identification of two genes encoding the tobacco enzyme. Planta 197:422-425[ISI][Medline].

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