Natural vitamin B6 consists of six interconvertible compounds, pyridoxine, pyridoxal, pyridoxamine and their phosphorylated derivatives, pyridoxine 5'-phosphate, pyridoxal 5'-phosphate and pyridoxamine 5'-phosphate [1–3]. Most bacteria, fungi and plants possess vitamin B6 biosynthesis pathways, but mammals must acquire the vitamin in their diet. In plants, thede novo pathway of vitamin B6 biosynthesis relies on two proteins, PDX1 and PDX2, which function as a glutamine amidotransferase and produce pyridoxal-phosphate from intermediates of glycolysis and the pentose phosphate pathway [4,5]. PDX1 and PDX2 work together, with the latter protein as the glutaminase and the former as the synthase domain.

Vitamin B6 plays essential roles as a cofactor in a wide range of biochemical reactions, predominantly in amino acid metabolism [6,7]. Recently, besides their classical role as coenzymes, a new function has emerged for the various vitamin B6 compounds in cellular antioxidant defense. A link between vitamin B6 and oxidative stress was originally established in the phytopathogenic fungusCercospora nicotianae. Mutant strains were identified that were particularly vulnerable to their own toxin cercosporin, a photosensitizer that produces singlet oxygen (¹O₂) in the light [8]. Unexpectedly, cloning of the mutant genes inC. nicotianae revealed that the mutated fungi were affected in a gene of the vitamin B6 biosynthesis pathway [9]. Subsequently, it was shownin vitro that vitamin B6 is able to quench¹O₂ with a high efficiency [9,10]. Additional analyses revealed that vitamin B6 is also able to quench superoxide [11]. The antioxidant capacities of vitamin B6 were confirmed in yeast or animal cell cultures supplied with exogenous vitamin B6 compounds and exposed to different oxidative treatments [12–16]. Similarly, exogenously applied vitamin B6 was found to protect plant protoplasts against¹O₂-induced cell death [17]. Thesein vitro results indicate that vitamin B6 is a potential antioxidant and raise the question as to whether plants employ vitamin B6 to protect themselves against reactive oxygen species (ROS), particularly¹O₂. Several mutants ofArabidopsis thaliana defective in vitamin B6 biosynthesis have been recently isolated which could help answering this question. A knock out of the singlePDX2 gene is lethal forArabidopsis [4]. There are 3 homologues ofPDX1 inArabidopsis,PDX1.1,PDX1.2 andPDX1.3. Two of these (PDX1.1 andPDX1.3) have been shown to be functional in vitamin B6 synthesis [4]. While disruption of both genes causes lethality, the single mutantspdx1.1 andpdx1.3 are viable, indicating that one gene can compensate, at least partially, for the lack of the other. However,PDX1.3 is more highly expressed thanPDX1.1, and aPDX1.3 knockout accumulates less vitamin B6 --about 30-40% of the wild type (WT) level) and has a more severe mutant phenotype in sterile medium [18–20]. Thus,PDX1.3 appears to be more important for vitamin B6 synthesis thanPDX1.1.

When grown in sterile medium in the absence of vitamin B6, seedlings of thepdx1.3 mutant are strongly reduced in shoot growth and primary root growth [18,19,21,22]. Under these conditions, mutant seedlings were also found to be more sensitive to the¹O₂-generating dye Rose Bengal, to salt stress and to UV radiation relative to WT seedlings [21]. Although this is consistent with the idea that vitamin B6 could play a rolein planta as an antioxidant, it is difficult to draw a definite conclusion because of the rather severe phenotype of the mutant in sterile culture. Interestingly, when grown on soil, the mutant phenotype of thepdx1.3 mutant was much less pronounced. The reason for the less severe phenotype in soil is unknown. It has been suggested that there is a source of the vitamin in the soil [18]. However, the vitamin B6 concentration in the leaves ofpdx1.3 mutant plants grown on soil remains very low compared to WT [19,20]. Alternatively, it is possible that growth in sterile medium in a Petri dish represents a form of stress to which plants with low levels of vitamin B6 are more sensitive. In this study, we took advantage of the nearly normal development of the vitamin B6-deficientpdx1.3 Arabidopsis mutant grown on soil to explore in detail the possibility that this vitamin functions as a photoprotector and an antioxidant in plants. We show that vitamin B6 acts as a new class of¹O₂ quencher, thereby protecting plants against photooxidative stress.

Growth and leaf chlorophyll content ofpdx1plants

Vitamin B6-deficientpdx1.3 plants grown on soil (abbreviated aspdx1 hereafter) looked similar to WT plants, except that young leaves in the center of the rosette were paler (Fig.1A) as previously reported [18,21]. This was due to a decrease in photosynthetic pigments (Fig.1B): both chlorophylls (Chl) and carotenoids were reduced by about 15-20%, and this was accompanied by a significant increase in the Chla/b ratio. This reduction of the pigment content tended to disappear in mature, well developed mutant leaves. We also measured the concentration of various Chl precursors in young leaves (Fig.1C). No significant change was observed in protochlorophyllide (PChlide) and chlorophyllide (Chlide) levels between WT and mutant leaves. In contrast, a decrease in the geranylgeranylated forms of Chl, namely geranylgeranyl Chl (GG-Chl), dihydrogeranylgeranyl Chl (DHGG-Chl) and tetrahydrogeranylgeranyl Chl (THGG-Chl) was found in young leaves of thepdx1 mutant. It is known from studies of etiolated seedlings that GG-Chl is formed through a preferential esterification of Chlide by geranylgeranyl disphosphate catalyzed by the enzyme Chl synthase [23–25]. GG-Chl is then reduced stepwise to Chl via DHGG-Chl and THGG-Chl by geranylgeranyl reductase [26]. Therefore, the marked decrease in GG-Chl and other geranylgeranylated intermediates in leaves of thepdx1 mutant suggests that the Chl synthase activity is somehow affected by thepdx1 mutation, ultimately leading to a reduction in Chl concentration in the leaves. Therefore, it is likely that either the catalytic activity of Chl synthase itself is inhibited or that levels of the substrate geranylgeranyl diphosphate are more limiting. However, the unchanged level of tocopherols in thepdx1 mutant (see below) would suggest that levels of geranylgeranyl phosphate are not limiting. Moreover, a rice mutant with impaired Chlide esterification by Chl synthase has a phenotype that strongly resemblespdx1 mutants: decreased Chl levels were associated with an increased Chla/b ratio in young plants, and these effects progressively disappeared as leaves matured [27]. We also found that the change in Chl content of leaves of thepdx1 mutant relative to WT leaves was strongly dependent on light intensity (Fig.2): the difference in Chl concentration and in the Chla/b ratio between WT andpdx1 was strongly attenuated when plants were grown in low light (80-100 μmol photons m^-2 s^-1) and was enhanced when plants were grown in high light (1000 μmol m^-2 s^-1).

Pigment content of young leaves of WTArabidopsisand of thepdx1mutant. A) Plants aged 4 weeks. B) Chlorophyll and carotenoid content of young leaves. Chl, total chlorophyll; Xanth, xanthophylls; β-car, β-carotene. C) Level of various chlorophyll precursors in young leaves: Pchlide, protochlorophyllide; Chlide, chlorophyllide; GG-, DHGG- and THGG-Chl, geranylgeranyl-chlorophyll, dihydrogeranylgeranyl-chlorophyll and tetrahydrogeranylgeranyl-chlorophyll, respectively. Data are mean values of 4 measurements + SD. *, significantly different from the WT value withP < 0.01 (t test).

Full size image

A) Chlorophyll content and B) chlorophylla/bratio in leaves of WT andpdx1plants grown at different PFDs. Data are mean values of 3 measurements ± SD.

Full size image

The decrease in photosynthetic pigments in leaves of thepdx1 mutant was not associated with substantial changes in photosynthetic electron transport. The quantum yield of linear electron transport measured by Chl fluorometry was comparable in WT andpdx1 leaves (Fig.3A). Similarly, the rate of O₂ evolution measured with a Clark electrode did not appear to be affected by thepdx1 mutation (Fig.3B). Also, neither shoot growth or root growth were significantly affected by inactivation of thePDX1.3 gene (Additional File1). Normal development of vitamin B6-deficient shoot grown on soil was previously reported [18,21]. Clearly this was also the case for root development in soil.

Photosynthetic parameters of WTArabidopsisleaves and leaves of thepdx1mutant grown under control conditions (150-200 μmol m^-2s^-1, 25°C). A) Quantum yield of PSII photochemistry (ΔF/Fm'), B) oxygen exchange and C) NPQ measured at different PFDs. Data are mean values of 3 or 4 measurements ± SD. D) Light-induced conversion of violaxanthin (V) into zeaxanthin (Z) and antheraxanthin (A), as calculated by the equation (A+Z)/(V+A+Z). Zeaxanthin synthesis was induced by white light of PFD 1000 μmol m^-2 s^-1. Each point corresponds to a different leaf (1 measurement per point).

Full size image

We observed a difference in nonphotochemical energy quenching (NPQ) between WT leaves and leaves of thepdx1 mutant, with NPQ being enhanced in the latter leaves, particularly at high photon flux densities (PFDs) above 500 μmol photons m^-2 s^-1 (Fig.3C). NPQ is a photoprotective mechanism that requires a transthylakoid pH gradient and the synthesis of zeaxanthin from violaxanthin in the light-harvesting antennae of PSII [28,29]. The increased NPQ in thepdx1 mutant is thus consistent with the increased rate of photoconversion of violaxanthin to zeaxanthin: zeaxanthin synthesis in high light was faster, and the final extent of conversion was increased in thepdx1 mutant relative to WT (Fig.3D).

In vitrosensitivity of vitamin B6-deficient leaves to ROS

Leaf discs were exposed to eosin, a xanthene dye that generates¹O₂ in the light [30]. Illuminating leaf discs floating on a solution (0.5%) of eosin has been previously shown to cause leaf photooxidation and lipid peroxidation [30,31]. We visualized the effect of eosin by autoluminescence imaging. This technique measures the faint light emitted by triplet carbonyls and¹O₂, the by-products of the slow and spontaneous decomposition of lipid hydroperoxides and endoperoxides [32–34]. Deactivation of excited carbonyls and¹O₂ produces photons (in the blue and red spectral regions, respectively) which can be recorded with a high-sensitivity, cooled CCD (charge coupled device) camera [34]. This technique has been used to map lipid peroxidation and oxidative stress in various biological materials including detached leaves [35], whole plants [36,37], animals [38] and humans [39]. As shown in Fig.4A,¹O₂-induced lipid peroxidation was associated with a marked enhancement of leaf disc autoluminescence, as expected. Interestingly, the increase in autoluminescence was more pronounced in discs punched out frompdx1 leaves than in WT discs (Fig.4A). We quantified the autoluminescence intensity, and we found a 50%-increase in thepdx1 mutant relative to WT (Fig.4B). Thus, thepdx1 mutant appeared to be more sensitive to¹O₂ toxicity than WT. This was confirmed by thermoluminescence analyses of lipid peroxidation (Fig.4C). Thermal decomposition of lipid hydroperoxides is associated with photon emission in the 120-140°C range [33,40]. The amplitude of the thermoluminescence band peaking at ~135°C has been correlated in previous studies with the extent of lipid peroxidation as measured biochemically [33,36,41]. The 135°C band amplitude was noticeably higher in eosin treated leaf discs taken frompdx1 than from the WT. Using HPLC, we also found that the level of malondialdehyde, a 3-carbon aldehyde produced during lipid peroxidation, was 29% higher inpdx1 leaf discs than in WT discs after the eosin treatment (3 repetitions, data not shown). Together these results show that eosin treatment results in significantly increased lipid peroxidation in the mutant.

Oxidative stress inArabidopsisleaf discs (WT andpdx1) exposed to the¹O₂generator eosin (0.5%). A) Autoluminescence imaging of leaf discs exposed for 3.5 h or 5 h to eosin in the light (400 μmol photons m^-2 s^-1). 'Dark' corresponds to eosin-infiltrated leaf discs kept in the dark for 5 h. B) Autoluminescence intensity in leaf discs exposed for 0 or 5 h to eosin in the light. Data are mean values of 10 measurements + SD. *, significantly different from the WT value withP < 0.001 (t test). C) Thermoluminescence band at high temperature (ca. 135°C) in leaf discs exposed for 5 h to eosin in the light. Control, leaf discs frompdx1 kept in eosin in the dark. Control WT disks (not shown) was in the same thermoluminescence intensity range. The band peaking at ca. 60°C in the control is typical ofArabidopsis. Its origin is unknown; it is not related to lipid peroxidation and could be due to thermolysis of a (yet unidentified) volatile compound [84].

Full size image

In contrast to¹O₂, other ROS such as hydrogen peroxide and superoxide did not induce different amounts of photooxidation between mutant and WT leaf discs (Additional File2). Although exposure of leaf discs to both ROS enhanced autoluminescence, this effect was similar in WT andpdx1. Similarly, the 135°C thermoluminescence band ofpdx1 and WT leaf discs after H₂O₂ and superoxide treatment were indistinguishable (data not shown).

Vitamin B6-deficient plants are more sensitive to¹O₂-mediated lipid peroxidation than WT leaves

¹O₂ was recently shown to be the major ROS involved in photooxidative damage to leaves [42]. A combination of low temperature and high light is known to be particularly favorable for inducing photooxidative stress in higher-plant leaves [43]. Therefore, we exposed leaf discs to a high photon flux density (PFD) of 1000 μmol photons m^-2 s^-1 at low temperature (10°C). This treatment induced lipid peroxidation, as measured by autoluminescence (Fig.5A) and thermoluminescence (Fig.5B). Leaf discs from thepdx1 mutant were clearly more sensitive to the high light treatment than WT discs: both signals were enhanced in the mutant compared to WT. When leaf discs taken from thepdx1 mutant were infiltrated with vitamin B6 before the light treatment, the increased thermoluminescence relative to WT was lost, confirming that exogenous vitamin B6 can function as an antioxidant [17].

Photooxidative stress in leaf discs (WT andpdx1). A) Autoluminescence of leaf discs exposed for 6 h to 1500 μmol m^-2 s^-1 at 10°C. B) Thermoluminescence band at high temperature (ca. 135°C) in leaf discs exposed to high light stress for 0, 5, 6 or 20 h. The thermoluminescence signal of discs taken from leaves of thepdx1 mutant and preinfiltrated with vitamin B6 (2 mM) is also shown (5 h + vitamin B6).

Full size image

The high photosensitivity of vitamin B6-deficient leaf discs prompted us to investigate the responses of whole plants to photooxidative stress conditions. Figure6 shows the effect of 2-d exposure ofArabidopsis plants to photooxidative stress induced by very high light (1500 μmol photons m^-2 s^-1) at low temperature (6°C) on lipid peroxidation. Again, autoluminescence emission was much higher inpdx1 than in WT after this treatment (Fig.6A). This was particularly visible in the external leaves, in agreement with previous studies that have emphasized the higher sensitivity of mature leaves to oxidative stress relative to young, developing leaves [e.g. [31,44]]. This observation indicates that the increased sensitivity ofpdx1 to photooxidative stress is not directly attributable to the low-Chl phenotype ofpdx1 which was visible mainly in the young leaves.

Photooxidative stress of wholeArabidopsisplants (WT andpdx1). A) Autoluminescence imaging of lipid peroxidation after high light stress (2d, 6°C, 1500 μmol m^-2 s^-1). B) Thermoluminescence signal of WT leaves and leaves of thepdx1 mutant before and after high light stress (LL and HL, respectively). C) Lipid hydroperoxide level (HOTE) in leaves of control and high light-stressed WT andpdx1 plants. *, significantly different from the WT value withP < 0.015 (t test). D) Distribution of lipid hydroperoxide (HOTE) isomers in leaves of control and high-light stressed WT andpdx1 plants. Data are mean values of 3 to 5 measurements + SD.

Full size image

The differential sensitivity of thepdx1 mutant and WT to light stress was confirmed by thermoluminescence measurements (Fig.6B) and also by HPLC analyses of lipid hydroperoxide concentrations (Fig.6C). The level of HOTE (hydroxyl octadecatrienoic acid), the product of the oxidation of linolenic acid (the major fatty acid in plant leaves) doubled in WT plants after light stress. Inpdx1 the HOTE concentration increased by a factor of 5. Figure6D shows the relative proportions of the different HOTE isomers during lipid peroxidation induced by high light stress. Isomers specific to¹O₂ (10-HOTE and 15-HOTE, [45]) were present in high amounts, and their level relative to the isomers 9-HOTE and 16-HOTE, which are produced by all ROS (free radicals and¹O₂) was typical of¹O₂ attack on polyunsatured fatty acids (see [42]). Thus, one can conclude thatpdx1 plants are more sensitive to endogenous¹O₂ production than WT plants.

¹O₂ levels during illumination are enhanced in thepdx1mutant

Singlet oxygen sensor green (SOSG) reagent is a fluorescein derivative compound that is selective to¹O₂ with no appreciable response to superoxide and hydroxyl radical [46]. In the presence of¹O₂, it emits a green fluorescence that peaks at 525 nm. However, this fluorescent probe has a relatively low stability in the light, so that the use of this probe to measure¹O₂ production should be restricted to short illumination only. Figure7A shows the fluorescence spectrum ofArabidopsis leaves infiltrated under pressure with SOSG and illuminated for 40 min at a PFD of 400 μmol photons m^-2 s^-1. SOSG fluorescence at 525 nm was well visible in the fluorescence emission spectrum of the illuminated leaves. This fluorescence was enhanced inpdx1 relative to WT, indicating an increased level of¹O₂ in the former plants. Figure7B shows the fluorescence emission at 525 nm (F525) normalized to the fluorescence of chlorophylls at 680 nm (F680) in leaves infiltrated with SOSG, with vitamin B6 or with both. The only condition that caused a significant increase in the F525/F680 ratio, indicative of an increased production of¹O₂, was the illumination of SOSG-infiltrated leaves of thepdx1 mutant. Interestingly, the photoinduced increase in the F525/F680 ratio ofpdx1 leaves was lost when leaves were infiltrated with vitamin B6 in addition to SOSG. This loss of SOSG fluorescence indicates that exogenous vitamin B6 can quench¹O₂in vivo, thus confirmingin vitro data [10].

Fluorescence of SOGS in WT and mutant (pdx1) leaves exposed to high light. A) Fluorescence of leaves infiltrated with SOGS after exposure to white light (HL = 450 μmol photon m^-2 s^-1 for 40 min). Controls (= c) were kept in dim light before fluorescence measurements. B) Fluorescence ratio F525/F680 of WT leaves and mutant leaves infiltrated with SOGS and/or vitamin B6 before or after illumination. Data are mean values of 3 measurements + SD. *, significantly different from the WT value withP < 0.025 (t test).

Full size image

Thepdx1 mutation enhances the photosensitivity of thevte1 npq1mutant

Thevte1 npq1 double mutant is deficient in two major¹O₂ quenchers, vitamin E (tocopherols) and the carotenoid zeaxanthin [47].Vte1 npq1 is photosensitive, exhibiting oxidative stress and lipid peroxidation in high light [42,47]. This is illustrated in Fig.8 wherevte1 npq1 plants were exposed to a rather moderate light stress (white light of PFD 1000 μmol m^-2 s^-1 at 10°C). This treatment brought about leaf bleaching (Fig.8A) and increased autoluminescence (Fig.8B). On the contrary, both WT andpdx1 plants appeared to be resistant to this treatment. Similarly, the single mutantsvte1 andnpq1 did not display symptoms of photooxidative damage under these conditions (data not shown). Thevte1 npq1 mutant was crossed with thepdx1 single mutant to generate a triple mutant (vte1 npq1 pdx1) deficient in vitamins E and B6 and in zeaxanthin. The triple mutant exhibited an extreme sensitivity to high light: most leaves bleached (Fig.8A) and leaf autoluminescence increased markedly (Fig.8B). We also measured the HOTE concentration in leaves (Fig.8C), which was higher in the triple mutant than in the double or single mutants. Thus, removing vitamin B6 in thevte1 npq1 background led to a highly photosensitive phenotype. Analysis of the lipid peroxidation signature indicated that lipid peroxidation in the triple mutant was mediated by¹O₂ (Fig.8D). The high photosensitivity of leaves of thevte1 npq1 pdx1 triple mutant compared to leaves of thevte1 npq1 andpdx1 mutants suggests that there is some overlap in the functions of vitamin B6 and the zeaxanthin-vitamin E duo.

Effects of high light stress (1000 μmol photons m^-2s^-1at 10°C for 2 d) on WT plants and onpdx1,vte1 npq1andvte1 npq1 pdx1mutant plants. A) Plants after the high light treatment. B) Autoluminescence imaging of lipid peroxidation. C) HOTE level. a, significantly different withP < 0.03 (t test). D) Distribution of HOTE isomers in leaves of thevte1 npq1 pdx1 triple mutant exposed to the high light treatment. Data are mean values of 3 or 4 measurements + SD.

Full size image

Protective mechanisms against¹O₂ in leaves of thepdx1mutant

Figure8 shows thatpdx1 plants are able to tolerate high light, provided the stress is not too severe. We analyzed the level of various antioxidant compounds inpdx1 and WT plants during acclimation for 7 days to a PFD of 1000 μmol m^-2 s^-1. Carotenoids and tocopherols are major quenchers of¹O₂ in plant leaves while ascorbate is one of the most efficient scavengers of¹O₂ [48]. Under control growth conditions, the ascorbate and tocopherol content ofpdx1 and WT plants was similar. Light acclimation led to a comparable increase in ascorbate, in WT andpdx1 (Fig.9A). Tocopherol was increased as well, but this change was less pronounced inpdx1 (Fig.9B). This could be due to the consumption of tocopherol by increased oxidative stress in the mutant. Although the total Chl level (on a leaf area basis) did not change during photoacclimation (Fig.9C), the Chla/b ratio increased, especially inpdx1 (Fig.9D). The most obvious change in carotenoid composition was an accumulation of antheraxanthin and zeaxanthin, which was more pronounced in thepdx1 mutant than in WT (Fig.9E). β-Carotene also increased, but by a similar amount inpdx1 and WT. Lutein and neoxanthin did not change significantly during photoacclimation although they were slightly reduced in the mutant compared with WT. This reduction reflects a decrease in the PSII antenna size in the mutant (see below). The Chl-to-carotenoid ratio differed noticeably between WT andpdx1, falling from 4.17 to 3.89 and from 3.93 to 2.76 respectively during high light acclimation. Accumulation of carotenoids, especially zeaxanthin, and the putative consumption of α-tocopherol by oxidation suggests that thepdx1 mutant senses a higher level of photostress than WT.

Levels of chlorophyll and various antioxidants in WT leaves and leaves ofpdx1after long-term exposure to high light (1000 μmol m^-2s^-1, 10°C, 7d). A) Ascorbate, B) α-Tocopherol, C) Total chlorophyll, D) Chlorophylla/b ratio, E) β-carotene (car) and xanthophylls (lutein (lut), violaxanthin (vio), antheraxanthin (ant), zeaxanthin (zea), neoxanthin (neo)). Data are mean values of 3 measurements + SD. C = control plants; S = plants exposed to the high light treatment. *, ** and ***, significantly different from the WT value withP < 0.001, 0.035 and 0.01, respectively (t test). White bars, WT; black bars,pdx1 mutant.

Full size image

PSII antenna size is decreased in leaves of thepdx1mutant

The decreased Chl levels and increased Chla/b ratio ofpxd1 mutants (particularly at high PFD, Fig.2) suggest that there is a differential adjustment of the photosynthetic complexes to the light environment in mutant compared to WT plants. Therefore, we analyzed the relative abundance of Chl-containing photosynthetic complexes in thylakoids prepared from WT andpdx1. The pigmented protein complexes of thylakoids were solubilized in 0.8% α-dodecylmaltoside and were separated by ultracentrifugation on sucrose gradient (Fig.10A). As expected, acclimation of WT leaves to high light (1000 μmol m^-2 s^-1) brought about a substantial decrease in the PSII antenna system (monomeric Lhcb and trimeric LHCII; B2 and B3 bands in Fig.10A, respectively) relative to the PSII reaction center (B5 band). The PSI-LHCI supercomplex (B6 band) was also reduced during photoacclimation. Rather surprisingly the profile of thylakoids isolated from young leaves of low light-grownpdx1 plants was very similar to the profile of high light-grown WT plants. High light-grownpdx1 leaves showed a rather extreme situation: the PSII antennae were strongly reduced compared to the PSII core and the abundance of PSI-LHCI supercomplexes was extremely low. Long-term acclimation ofpdx1 to high light was also associated with an increased level of free carotenoids (B1 band). Thus, the enhancement of the carotenoid/Chl ratio in leaves of thepdx1 mutant seems to be largely due to unbound carotenoids. However, the quality of the separation of the photosynthetic complexes of thylakoids prepared from high light-acclimated leaves ofpdx1 was poor in 0.8% α-dodecylmaltoside, presumably because of a high lipid/protein ratio. Consequently, a higher α-dodecylmaltoside concentration (1.2%) was used to improve solubilization of thylakoids prepared frompdx1 leaves after acclimation to high light (Fig.10B). By comparison with low light-grownpdx1 plants, the profile obtained with high-light treatedpdx1 at this detergent concentration confirmed that the effects of high light were drastic in the mutant, with a strong decrease in PSI-LHCI and PSII antenna size and an increase in the level of free pigments (Fig.10B).

A) Separation of pigmented photosynthetic complexes of thylakoids prepared from leaves of WT andpdx1by solubilization in 0.8% dodecylmaltoside and ultracentrifugation on sucrose gradient. Thylakoids were prepared from leaves of WT andpdx1 grown in low light (c, 200 μmol photons m^-2 s^-1) or acclimated for 7 d to high light (hl, 1000 μmol m^-2 s^-1). B1, free pigments; B2, monomeric Lhcb antennae; B3, LHCII trimers; B5, PSII core (monomeric), B6, PSI-LHCI supercomplex. The B4 band (LHCII-CP29-CP24 supercomplex, see [85]) is not visible in this gradient. B) Ultracentrifugation gradient of thylakoids (pdx1, c and hl) solubilized in 1.2% dodecylmaltoside. In the controlpdx1 sample, an additional band appeared in the bottom of the gradient, which was hardly visible at 0.8% dodecylmaltoside and which corresponded to dimeric PSI-LHCI. This is presumably due to an artificial aggregation the high detergent concentration used in this preparation as previously found [86]; the same phenomenon was observed with WT thylakoids (data not shown). C and D) SDS-PAGE separation of the B2, B3 and B6 bands using two different buffer systems: tricine (C) and urea (D). See ref. [87] for identification of the bands. BBY = PSII-enriched membranes used as a reference for the PSII proteins.

Full size image

A global reduction of the PSII antenna system (gradient fractions B2 and B3 vs. B5) could explain the increase in the Chla/b ratio in thepdx1 mutant. However, the absorption spectra of the B2 and B3 bands showed that the light-harvesting complexes of PSII themselves contain less Chlb (Additional File3), suggesting that the composition of these bands was modified. This prompted us to analyze the protein composition of the different bands by SDS-PAGE. Two different buffer systems were used: Tricine (Fig.10C) and Laemmli-urea (Fig.10D). The former system allows a good separation of the Lhcb polypeptides whereas the latter system is more appropriate for separating the Lhca proteins. In WT, acclimation to high light resulted in the decreased relative abundance of several PSII antennae (Lhcb1-2 and CP24, also named Lhcb6) and the increased relative abundance of CP26 (Lhcb5) with respect to control conditions. The abundance of CP29 (Lhcb4) was little affected (Fig.10C). Low-light grownpdx1 plants showed similar changes in the relative abundances of Lhcb1-2, CP24 and CP26 indicating that even under low light this mutant suffers light stress comparable to that of the WT at a PFD of 1000 μmol m^-2 s^-1. These changes were strongly amplified whenpdx1 was exposed to high light. Since CP26 and CP29 have a higher Chla/b ratio than other Lhcb antennae [49], the relative enhancement of these antennae might help contribute to the increased Chla/b ratio inpdx1. The Chla/b ratio of band B2 was particularly high (2.9) inpdx1 plants grown under high light. Band B2 consists of a mixture of different monomeric antennae that usually have Chla/b ratios between 1.2 and 3.0 [49]. Therefore the high Chla/b ratio of the B2 bandpdx1 plants cannot simply be explained by a decrease in the abundance of the Chlb-rich monomers. Instead there must be an increased Chla/b ratio within the Lhcb complex itself. This could be explained by either a reduced Chlb availability as a result of stress that results in Chla-rich folding of the Lhc complexes, or else by the preferential accumulation of specific Lhcb isoforms that are rich in Chla, as previously suggested for maize [50]. We also observed a higher abundance of ATPase relative to antenna proteins under high light (Fig.10C and10D). However a precise quantification is not possible from these gels since ATPase fragments into several subcomplexes during gradient centrifugation, with the most intact complex migrating in B6 together with PSI. However, we were able to further confirm the higher abundance of ATPase relative to Chl-binding complexes by SDS-PAGE separation of total thylakoid proteins (data not shown).

Changes in the relative proportions of the Lhca proteins in response to high light and/or inpdx1 were much less pronounced than those occurring in the PSII antenna system (Fig.10D). Nevertheless, a relative increase in PsaD and possibly Lhca4 abundance seemed to occur inpdx1 plants that had been acclimatised to high light (Fig.10D).

Together, the data of Figs.9 and10 suggest that vitamin B6-deficient leaves sensed a higher level of light stress at a given PFD and over-reacted to increasing PFD compared to WT leaves. Incidentally, the smaller antenna system ofpdx1 was not associated with substantial changes in photosynthetic electron transport efficiency (Fig.3). This is consistent with previous studies of PSII antenna mutants of Arabidopsis which have shown that rather strong reductions of the antenna system do not necessarily affect the photochemical activity of leaves [e.g. [51]].

Vitamin B6 accumulation during high light acclimation

The expression of thePDX1 andPDX2 genes is up-regulated by several stress conditions, including high light [11,18,52]. However, so far the vitamin B6 concentration in plant tissues has not been measured under those conditions. Using HPLC, we were able to measure the non-phosphorylated forms of vitamin B6. Figure11 shows the effect of high light (1000 μmol photons m^-2 s^-1 at 10°C for 7 d) on the concentration of nonphosphorylated vitamin B6 components ofArabidopsis leaves. Pyridoxine and pyridoxamine were the major vitamin B6 constituents measured in leaves, with pyridoxal being present in low amounts only. Pyridoxine and pyridoxal noticeably increased in high light while pyridoxamine did not change, so that the total (non-phosphorylated) vitamin B6 level increased by about 70%.

Vitamin B6 components (expressed in μg/g fresh weight) in leaves ofArabidopsisplants grown in low light (LL) or acclimated for 7 d to high light (HL, 1000 μmol photons m^-2s^-1at 10°C). F. W. = fresh weight. PM, pyridoxamine; PN, pyridoxine; PL, pyridoxal. Data are mean values of 2 or 3 measurements + SD.

Full size image

Vitamin B6 deficiency leads to¹O₂-mediated photodamage

Vitamin B6-deficientArabidopsis leaves were more sensitive to treatments with the¹O₂ generator eosin than WT leaves, and exogenous application of vitamin B6 reduced¹O₂ level and mitigated lipid peroxidation in leaf discs exposed to high light. The protective role of vitamin B6 observedin vitro was confirmedin vivo inArabidopsis plants challenged with endogenous¹O₂ production induced by high light stress. Exposure ofArabidopsis plants to high light led to a rise in¹O₂concentration and an accumulation of oxidized lipids, which were higher inpdx1 than in WT. The increased level of lipid peroxidation in mutant leaves was attributable to a¹O₂ mediated attack on lipids. Those results show that vitamin B6 has a function in the protection of plants against¹O₂ toxicity and photooxidative stress. This confirmsin vivo the antioxidant capacity of vitamin B6 previously inferred fromin vitro studies [9–17]. The role of vitamin B6 in the response of plants to light stress was further supported by our observation that the concentration of this vitamin is increased inArabidopsis leaves exposed to high light intensity. This finding is in line with previous studies that have shown an increased expression of genes of the vitamin B6 biosynthesis pathway (PDX1 andPDX2) by abiotic stresses [11,18,52]. Illumination ofpdx1 seedlings grown under sterile conditions has been reported to provoke degradation of the D1 protein of the PSII reaction center and to exacerbate the associated photoinhibition of PSII [18]. The latter phenomenon is attributed to¹O₂ attack on the D1 protein itself, triggering structural changes in the PSII centre that initiate proteolytic degradation of the protein [53]. These data add further support to our conclusions that reduced levels of vitamin B6 inpdx1 leads to enhanced accumulation of¹O₂.

Direct versus indirect effect of vitamin B6 in photoprotection

The photoprotective role of vitamin B6 could be direct or indirect. A direct role would mean that vitamin B6 quenches¹O₂ produced by light in the chloroplasts. This is plausible because this vitamin is able to quench¹O₂in vitro with a rather high efficiency [10]. The¹O₂quenching rate constant of vitamin B6 is comparable to that of ascorbate and tocopherol [9]. However, because of the high reactivity of¹O₂, this supposes that vitamin B6 is presentin planta in the vicinity of the¹O₂ production sites, namely the PSII reaction center and the chlorophyll antenna system in the chloroplasts [53]. Vitamin B6 levels inArabidopsis leaves are relatively high ([20], this study), in the same range of concentrations as glutathione [48], but its sub-cellular distribution is unknown. To check if chloroplasts constitute a site of vitamin B6 accumulation in plant leaves, we prepared intact chloroplasts and we titrated vitamin B6 by HPLC (Additional File4). Because our HPLC method requires large amounts of material (> 10 g of fresh weight), it was difficult to prepare sufficient amounts of intact chloroplasts fromArabidopsis leaves, and consequently we measured vitamin B6 in another plant species, tobacco, that is more suitable for purifying intact chloroplasts by ultracentrifugation on Percoll gradient. Both pyridoxine and pyridoxamine were detected in intact tobacco chloroplasts (Additional File4). When normalized to the Chl content, the (nonphosphorylated) vitamin B6 content of chloroplasts (~0.16 μg/mg Chl) was approximately 3 times lower than the concentration in leaves. Considering that the chloroplast volume represents about 25% of the total cellular volume [54] and that Chl is localized exclusively in the chloroplasts, this suggests that there is a uniform distribution of vitamin B6 between the chloroplast and the rest of the cell. However, one cannot exclude that the level of vitamin B6 in chloroplasts was underestimated due to vitamin export during the chloroplast isolation. The occurrence of vitamin B6 in chloroplasts, as reported here, is consistent with a number of previous observations. First, the N-terminal amino acids of one of the enzymes of the vitamin B6 pathway, pyridoxine (pyridoxamine) 5'-phosphate oxidase, have been identified as a chloroplast transit peptide [55], suggesting a chloroplastic localization for this protein. Both components of the pyridoxal synthase complex, PDX1 and PDX2, have been shown to be attached to membranes, including chloroplastic membranes [11,21]. Furthermore, the present study has shown that vitamin B6 deficiency impacts the activity of Chl synthase, a plastid-localized protein. Since vitamin B6 is an efficient quencher of¹O₂in vitro, it is easy to speculate that the presence of a vitamin B6 pool in the chloroplast would reduce¹O₂ levels. However, under conditions of severe light stress,¹O₂ has been reported to leave thylakoid membranes and to migrate to the cytoplasm [56]. Therefore, since the light stress conditions used in this work to induce photooxidative damage were rather drastic (1500 μmol photons m^-2 s^-1 at 6°C), a leakage of¹O₂ from the chloroplast to the cytosol cannot be excluded and therefore an action of vitamin B6 within the cytosol is also possible.

Hydroperoxides and endoperoxides generated in lipid peroxidation are known to undergo fragmentation to produce a broad range of reactive intermediates called reactive electrophile species [57,58]. Reactive electrophiles are harmful to macromolecules by reacting with nucleophilic groups, resulting in a variety of adducts and irreversible modifications. Compared to ROS, reactive electrophile species are stable and, due to their non-charged structure, some of them can migrate through hydrophobic membranes and hydrophilic media, so that they are able to propagate oxidative stress far from their site of formation [59]. Interestingly, pyridoxamine has been shown to trap lipid-derived carbonyl intermediatesin vitro [60,61], and pyridoxamine adducts to lipid peroxidation products have been detected in the urine of pyridoxamine--treated animals [60]. In humans, pyridoxamine and pyridoxine are considered to be promising drug candidates for treatment of chronic conditions in which carbonyl compounds confer pathogenecity, such as diabetes [62,63]. A similar function as scavenger of intermediates in lipid peroxidation could be envisaged for vitamin B6 in plant cells. However, this mechanism does not explain the selective sensitivity of leaf discs to¹O₂ (Fig.4 vs. Additional File2) since free radical-induced lipid peroxidation also generates reactive carbonyl species. Moreover, we administrated 4-hydroxynonenal, one of the most toxic carbonyl compounds produced from lipid peroxides [58], to detachedArabidopsis leaves, using the procedure described by Mano et al. [64]. As expected, necrosis developed concentrically from the application site of the hydroxynonenal solution on the leaf, but the extent of necrosis was similar in WT leaves and leaves of thepdx1 mutant (data not shown). Thus, vitamin B6 deficiency does not seem to enhance the sensitivity to reactive carbonyls, and an indirect function of vitamin B6 as scavenger of oxidized lipid derivatives seems unlikely.

One can also exclude the possibility that the increased level of¹O₂ in leaves of thepdx1 mutant relative to WT leaves after illumination was due to an increased production of¹O₂by the photosystems rather than a decreased quenching activity. In plants,¹O₂ is produced mainly from chlorophyll triplet states, which are formed when the balance between light absorption by the photosystems and light utilization by the photosynthetic processes is upset in favor of the former process. This can be excluded in leaves of thepdx1 mutant since photosynthetic electron transport was not affected significantly relative to WT. Moreover, the total Chl concentration inpdx1 was lowered by ca. 20%, at least in young leaves, and this would be expected to reduce¹O₂ production [65,66].¹O₂ can also be produced by Chl precursors such as Pchlide, as it is the case in theflu Arabidopsis mutant [67]. Based on our analyses of Chl biosynthesis intermediates, we can exclude this phenomenon inpdx1. The fact that exogenously applied¹O₂ was more toxic topdx1 than to WT is another indication that a change in¹O₂ production by the photosystems cannot be the sole factor involved in the increased sensibility to¹O₂ damage inpdx1. In this context, it is important to mention a recent work of Lytovchenko et al. [68] who showed that the profile of lipophilic compounds was not substantially affected in shoots of vitamin B6-deficientArabidopsis plants. Therefore, we consider that the management of¹O₂was less efficient inArabidopsis leaves when vitamin B6 concentration was abnormally low.

The most efficient biological quenchers of¹O₂ are thought to be the carotenoids and the vitamins C and E. Neither vitamin C (ascorbate) nor vitamin E (tocopherol) levels were reduced inpdx1. Although the total carotenoid content (on a leaf area basis) was lowered, the carotenoid concentration normalized to the Chl content was enhanced inpdx1. Among carotenoids, the xanthophyll zeaxanthin is known to play a crucial role in photoprotection [28,29,36,69]. Zeaxanthin synthesis and the associated NPQ were found to be stimulated inpdx1, and during long-term exposure to high light, the steady-state level of zeaxanthin was higher inpdx1 than in WT. Thus, the major antioxidant mechanisms involved in¹O₂ elimination in leaves did not appear to be reduced inpdx1, supporting the notion that the reduced capacity of¹O₂ quenching was directly related to the low concentration of vitamin B6, rather than to a secondary effect of vitamin B6 deficiency on the level of other antioxidant mechanisms. In sterile growth conditions, roots ofArabidopsis seedlings deficient in vitamin B6 displayed significant changes in lipid constituent content, such as a strong increase in α-tocopherol, supporting the idea that oxidative stress is involved in the inhibition of root growth [68].

Vitamin B6 deficiency induces chronic light stress in leaves

Acclimation of WTArabidopsis to high light induced marked changes in the protein composition of thylakoids. As previously reported [e.g. [70,71]], the most obvious modification was a decrease in the PSII antenna size, leading to a higher Chla/Chlb ratio. The abundance of all Lhcb proteins, except CP26 and to a lesser extent CP29, was decreased in high light. CP26 is supposed to constitute with CP29 an inner part of the antenna system that undergoes limited modifications with environmental conditions [71]. Interestingly, the loss of PSII antennae was observed in low light when thylakoids prepared from leaves ofpdx1 were compared with WT thylakoids and was strongly exacerbated whenpdx1 was exposed to high light (1000 μmol m^-2 s^-1). The gradient profile and characteristics of the photosynthetic complexes from low-light-grownpdx1 were very similar to that of high-light-acclimated WT thylakoids. Consistent with these observations the decreased Chl levels inpdx1 versus WT was strongly dependent on light intensity: in very low light (~100 μmol photons m^-2 s^-1), WT leaves and mutant leaves had very similar Chla/b ratio and total chlorophyll content whereas the Chla/b value differed drastically in high light. Thus, comparison of the photosynthetic complexes betweenpdx1 and WT suggests that, for a given PFD, the mutant senses a higher level of light stress than WT. Since the¹O₂ level induced by light inpdx1 was enhanced relative to WT, it is possible that the loss of Chl antennae represents a response to¹O₂ stress in the mutant. Although long-term acclimation of vascular plants to¹O₂has not yet been investigated,¹O₂ is known to induce changes in gene expression. Particularly, the gene coding for the PSII antenna Lhcb2 has been shown to be strongly and specifically downregulated by¹O₂[67]. In the green algaChlamydomonas, the early phases of¹O₂-mediated photooxidative stress were associated with the repression of the Lhcbm1 and Lhcbm2 genes at the RNA level [72]. UV-B radiation, which is know to induce the production of ROS including¹O₂, has been shown to downregulate expression of several Lhcb genes [52]. Interestingly, these conditions also up-regulated the expression of aPDX1 homologue,PYROA [52]. Alternatively, the loss of PSII antennae could also result from the inhibition of Chl synthesis in thepdx1 mutant. However, previous work on different transgenic plants have shown that a decreased availability of Chl induces a decrease in the amount of photosynthetic complexes embedded in the thylakoid membranes, but it does not change the PSII antenna size [73,74].

The potential function of the vitamin B6 constituents as antioxidants has been reported in severalin vitro studies in which yeast or animal cells were treated with different ROS [12–16]. There are also a few preliminary studies performedin vitro, that support the idea that vitamin B6 could fulfill a similar role in plant cells [11,17,21]. The present study of wholeArabidopsis plants provides the first evidences for an active and specific antioxidant role of vitamin B6in planta. Vitamin B6 deficiency was associated with a marked decrease in the tolerance to photooxidative stress, which manifested itself as an increase in the¹O₂ level in high light and a marked enhancement in¹O₂-mediated lipid peroxidation. On the other hand, it is known that there are some redundancies between the antioxidant systems in chloroplasts, so that removing one antioxidant mechanism is generally compensated, at least partially, by an increase in other protections. This has been established inArabidopsis and cyanobacteria for two classes of¹O₂ quenchers, the carotenoids and the tocopherols [47,75]. Similarly, removal of vitamin B6 from anArabidopsis mutant deficient in both carotenoids and tocopherols resulted in an extreme sensitivity to high light stress. These result indicate that vitamin B6 may play a specific role in antioxidant defense that is not completey fulfilled by carotenoids or tocopherols. Consequently, vitamin B6 can be considered as a new member of the network of protective compounds involved in the management of¹O₂ in plants.

Plant material, growth conditions and treatments

Wild-typeArabidopsis thaliana (ecotype Col-0) and thepdx1.3 (At5g01410) T-DNA line were grown in a phytotron under controlled conditions: PFD was 150-200 μmol photons m^-2 s^-1, photoperiod 8 h, air temperature 23/28°C (day/night) and relative air humidity 75%. Most of the experiments were performed on plants aged 5 weeks. Light stress was imposed by transferring plants to a growth chamber at 6/12°C (day/night) under a PFD of 1500 μmol photons m^-2 s^-1 and a photoperiod of 8 h. In preliminary experiments where we checked a number of light/temperature conditions, we selected this stress condition that appeared to be the most suitable to discriminate between WT andpdx1 in terms of photosensitivity. Thepdx1 mutant was crossed with thevte1 npq1 double mutant (see [47]) to generate the triple mutantvte1 npq1 pdx1 deficient in vitamin E, zeaxanthin and vitamin B6. The triple mutant and the double/single mutants were exposed to light stress by transferring them to a PFD of 1000 μmol photons m^-2 s^-1 at 10°C.

Leaf discs of 1 cm in a diameter were treated with a solution of 3.5% H₂O₂, 50 μM methylviologen or 0.5% eosin Y, as previously described [31]. The infiltrated discs were exposed to white light of PFD 400 μmol photons m^-2 s^-1 (for the eosin or methylviologen treatment) or 100 μmol m^-2 s^-1 (for the H₂O₂ treatment). Attached leaves were slowly infiltrated with 100 μM SOSG (Singlet Oxygen Sensor Green, Invitrogen) and/or vitamin B6 (1 mM pyridoxal) under pressure with a syringe. A 1-ml syringe, without needle and filled with the solution to be infiltrated, was pushed against the lower surface of the leaf, and the solution (200 μl) was forced to enter inside of the leaf under pressure. Plants with SOSG-infiltrated leaves were kept in darkness for 1 h and then exposed for 40 min to white light of PFD 450 μmol photons m^-2 s^-1. For high light treatment of leaf discs, the discs (diameter, 1 cm) were exposed at constant temperature (10°C) to white light (PFD, 1000 μmol photons m^-2 s^-1), as previously described [47]. In some cases, leaf discs were preinfiltrated with 2 mM vitamin B6 (pyridoxal) for 1 h. PFDs were measured with a Li-Cor quantum meter (Li-185B/Li-190SB).

Chlorophylls, carotenoids and vitamin E

One leaf disc (diameter, 1 cm) was ground in 400 μl of cold methanol. After filtration through a 0.45-μm PTFE filter (Iso-Disc, SUPELCO), 80 μl of the extract was immediately analyzed by HPLC, as previously described [47]. Pigments were detected at 445 nm and α-tocopherol was detected by fluorescence (λ_ex = 295 nm, λ_em = 340 nm). Running time was 22 min, flow rate was 1.5 ml.min^-1.

Chlorophyll precursors

Chlorophyll esters and (proto)chlorophyllide were quantitated using reverse phase HPLC analysis according to [76], except that detection was performed by absorbance at 430 nm.

Ascorbic Acid

Ascorbate was analyzed by HPLC as described elsewhere [47]. Total ascorbate was measured by reducing dehydroascorbic acid to ascorbic acid with TCEP (Tris-carboxyethylphosphine). Three leaf discs of 1 cm in diameter (about 100 mg) were ground in 750 μL of 0.1 M metaphosphoric acid. Samples were filtered through a 0.2 μm nylon membrane (Spin-X Costar). A 6 μL sample was immediately injected, and 6 μL were treated for 4 h with 10 mM TCEP in darkness at 25°C. Ascorbate was detected at 245 nm in sulphuric acid-acidified water (pH 2.5) with a retention time of 5 min under a flow of 0.65 mL min^-1.

Lipid peroxidation analyses

Lipids were extracted from 0.5 g frozen leaves by grinding with 2 × 1 mL chloroform containing 1 mg/mL triphenyl phosphine and 0.05% (w/v) butylated hydroxytoluene, with 15-hydroxy-11,13(Z, E)-eicosadienoic acid as internal standard. The organic phase was evaporated under a stream of N₂. The residue was recovered in 1.25 mL ethanol and 1.25 mL 3.5 M NaOH and hydrolyzed at 80°C for 15 min. After addition of 2.2 mL 1 M citric acid, hydroxyl fatty acids were extracted with 2 × 1 mL hexane/ether (50/50). An aliquot of the organic phase (50 μl) was submitted to straight phase HPLC (Waters, Millipore, St Quentin-Yvelines, France) using a Zorbax rx-SIL column (4.6·250 mm, 5 μm particle size, Hewlett Packard, Les Ullis, France), isocratic elution with 70/30/0.25 (v/v/v) hexane/diethyl ether/acetic acid at a flow rate of 1.5 ml min^-1, and UV detection at 234 nm. ROS-induced lipid peroxidation was evaluated from the levels of the different hydroxyoctadecatrienoic acid (HOTE) isomers as previously described using 15-hydroxy-11,13(Z, E) eicosadienoic acid as internal standard [77]. LOX-induced lipid peroxidation was estimated from the level of 13-HOTE after substraction of racemic 13-HOTE (attributable to ROS-mediated lipid peroxidation), as explained in [77].

The distribution of hydroxy fatty acid isomers was analyzed by HPLC-electrospray ionization-MS/MS as detailed previously [42]. Aliquots from the hydroxyl fatty acid extracts were evaporated and recovered in aqueous 1 mM ammonium acetate/acetronitrile (60/40, v/v) with [¹⁸O₂]13-HOTE used as internal standard. Hydroxy fatty acids were separated by HPLC and analyzed using a Waters Micromass Quatro premier triple quatrupole mass spectrometer in the negative electrospray ionization mode.

Thermoluminescence and autoluminescence imaging

Lipid peroxidation was measured in leaf discs by thermoluminescence using a custom-made apparatus that has been described previously [40]. The amplitude of the thermoluminescence band peaking at ca. 135°C was used as an index of lipid peroxidation [40,78]. The samples (2 leaf discs of 8 mm in diameter) were slowly heated from 25°C to 150°C at a rate of 6°C min^-1. Photon emission associated with lipid peroxidation was also imaged at room temperature using a highly sensitive charge coupled device (CCD) camera (VersArray LN/CCD 1340-1300B, Roper Scientific), with a liquid N₂ cooled sensor to enable measurement of faint light by signal integration [34]. Treated plants were dark-adapted for 2 h before imaging, to allow chlorophyll luminescence to fade away. Acquisition time was 20 min. Full resolution of the CCD is 1300 × 1340 pixels. On-CCD binning of 2 × 2 pixels was used to increase detection sensitivity, so that the resulting resolution was 650 × 670 pixels.

Photosynthetic electron transport

Chl fluorescence from attached leaves was measured in air at room temperature with a PAM-2000 fluorometer (Walz) [47]. The quantum yield of PSII photochemistry was calculated in white light as ΔF/Fm', where ΔF is the difference (Fm'-Fs) between the maximal fluorescence level Fm' (measured with a 800-ms pulse of saturating light) and Fs, the steady-state fluorescence level. White light was produced by a Schott KL1500 light source. NPQ was calculated as (Fm/Fm')-1 where Fm is the maximal fluorescence level in the dark [47].

O₂ exchange by leaf discs was measured in a Clark-type O₂ electrode (Hansatech LD2/2) under CO₂ saturating conditions. CO₂ was generated in the cell with a carbonate/bicarbonate buffer. White light was produced by a Hansatech LS2 light source combined with neutral density filters.

Membrane preparation and solubilisation

Arabidopsis leaves were shortly grinded in a solution containing 20 mM Tricine KOH pH 7.8, 0.4 M NaCl, 2 mM MgCl₂ and the protease inhibitors 0.2 mM benzamidine, 1 mM є-aminocaproic acid. The solution was filtered through miracloth tissue and centrifuged 10 min at 1400 g. The pellet was resuspended in a solution containing 20 mM Tricine KOH pH 7.8, 0.15 M NaCl, 5 mM MgCl₂ and protease inhibitors as before and then centrifuged 10 min at 4000 g. The pellet was resuspended in 20 mM Hepes 7.5, 15 mM NaCl, 5 mM MgCl₂ and centrifuged again 10 min at 6000 g and stocked in 20 mM Hepes 7.5, 0.4 M Sorbitol, 15 mM NaCl, 5 mM MgCl₂.

Membranes corresponding to 150 μg Chls were washed once with 5 mM EDTA, 10 mM Hepes pH 7.5, resuspended at 1 mg/ml Chls in 10 mM Hepes pH 7.5 and then solubilized at 0.5 mg/ml Chls by adding an equal volume of dodecyl-α-D-maltoside solution to have at a final detergent concentration of 0.8% or 1.2% and vortexing for a few seconds. The solubilised samples were centrifuged at 15.000 × g for 10 min to eliminate unsolubilised material and then fractionated by ultracentrifugation in a sucrose gradient (20 h, 288.000 × g, 4°C). The gradient was formed directly in the tube by freezing at -80°C and thawing at 4°C a 0.5 M sucrose solution containing 0.06% α-DM and 10 mM Hepes pH 7.5.

Chlorophylls and carotenoids were extracted in acetone (80% final concentration buffered with Na₂CO₃) and measured by fitting of the absorption spectrum of acetone extracts [79].

SDS-Page

Electrophoresis were performed using the Tris-Tricine system at 14% acrylamide concentration [80] or the Laemmli system [81] with the modification as in [82].

Vitamin B6

HPLC measurements of nonphosphorylated vitamin B6 components were carried out on leaves or isolated chloroplasts as described elsewhere [19,20]. Vitamin B6 was extracted from approximately 10 g of leaves (fresh weight). Intact chloroplasts were prepared from about 100 g of tobacco leaves, as described previously [83].