Water-stress induced downsizing of light-harvesting antenna complex protects developing rice seedlings from photo-oxidative damage

The impact of water-stress on chloroplast development was studied by applying polyethylene glycol 6000 to the roots of 5-day-old etiolated rice (Oryza sativa) seedlings that were subsequently illuminated up to 72 h. Chloroplast development in drought environment led to down-regulation of light-harvesting Chl-proteins. Photosynthetic proteins of Photosystem II (PSII) and oxygen evolving complex i.e., Cytb559, OEC16, OEC23 and OEC33 as well as those of PSI such as PSI-III, PSI-V, and PSI-VI, decreased in abundance. Consequently, due to reduced light absorption by antennae, the electron transport rates of PSII and PSI decreased by 55% and 25% respectively. Further, seedling development in stress condition led to a decline in the ratio of variable (Fv) to maximum (Fm) Chl a fluorescence, as well in the quantum yield of PSII photochemistry. Addition of Mg2+ to the thylakoid membranes suggested that Mg2+-induced grana stacking was not affected by water deficit. Proteomic analysis revealed the down-regulation of proteins involved in electron transport and in carbon reduction reactions, and up-regulation of antioxidative enzymes. Our results demonstrate that developing seedlings under water deficit could downsize their light-harvesting capacity and components of photosynthetic apparatus to prevent photo-oxidative stress, excess ROS generation and membrane lipid peroxidation.


Results
To understand the impact of water-stress on chloroplast biogenesis during the early photomorphogenesis, 5-d-old etiolated rice seedlings were treated with polyethylene glycol (PEG 6000) for 16 hours prior to their transfer to continuous cool-white fluorescent (plus incandescent) light (100 µmol photons m −2 s −1 ). In this paper, we have analyzed the structure and function of chloroplasts, ROS production, and associated processes to assess the role of water-stress on early seedling development.

Impact of Water-Stress on Chloroplast Development. Chl a fluorescence measurements. Fo, Fm and
Fv/Fm ratio were monitored after water-stress treatment of etiolated rice seedlings for 72 h with 40 mM and 50 mM PEG, in the presence of light (100 μmol photons m −2 s −1 ). Seedlings were kept for 20 min in darkness (see Materials and Methods) before the initial (Fo), and the maximum (Fm), fluorescence was measured. The minimum fluorescence Fo decreased by 11% and 21% in the seedlings treated with 40 and 50 mM PEG compared to the controls, while Fm decreased by 32% and 63% respectively (see Fig. 1A).
The ratio Fv/Fm = (Fm − Fo)/Fm, a "proxy" of the maximum quantum yield of PSII photochemistry 37 , declined in water-stressed samples; Fv/Fm calculated after 72 h of greening decreased by 10% and 22% in 40 and 50 mM PEG-treated seedlings (Fig. 1A).
Quantum yield of PSII (φPSII) and non-photochemical quenching (NPQ) during 300 s illumination with actinic light. As expected, φPSII declined at higher light intensity. Compared to the controls, φPSII, at the highest light intensity (~275 µmol photons m −2 s −1 ), was reduced by 33% and 58% in samples treated with 40 mM and 50 mM PEG (Fig. 1B). On the other hand, the NPQ ((Fm − Fm′)/Fm′) increased in response to increase in light intensity. It was higher in 40 mM and 50 mM PEG-treated seedlings than in controls by 21% and 29% under the highest (~275 µmol photons m −2 s −1 ) light intensity used (Fig. 1C).
Room temperature (298K) Chl a fluorescence spectra. Chl a fluorescence emission spectrum of thylakoid membranes isolated (after 72 h greening) had a peak at 684 nm; this is mostly, from PSII antenna 38 . Upon the addition of Mg 2+ in the suspension medium, this peak increased in the control as well as in the water-stressed samples ( Fig. 2A). In water-stressed samples, the fluorescence intensity at 684 nm was substantially lower both in the presence and the absence of Mg 2+ , compared to the controls.
Low temperature Chl a fluorescence spectra. Low temperature (77 K) fluorescence emission spectra monitored after 72 h of light exposure are shown in Fig. 2B. Thylakoid membranes of control samples, suspended in a low salt medium (0 mM Mg 2+ ), had a peak at 686 nm due to PSII and at 740 nm due to PSI 38 ; usually, PSII shows two emission bands at ~684 nm (from light-harvesting complex) and at 695 nm (from CP47), although a single peak has been observed in some cases (e.g. in Gonyaulax polyedra) 39 . Fluorescence emission spectra, shown here, were normalized at 686 nm. In water-stressed samples, the emission peak of PSI shifted from 740 nm to 738 nm. Upon addition of Mg 2+ (4mM), the PSI fluorescence emission at 740 nm (normalized at 686 nm; PSII) decreased in both the control and the water-stressed samples.   Fig. 1. Fluorescence emission spectra were recorded in ratio mode in a photon counting SLM-AMINCO 8000 spectrofluorometer. For 77 K spectral measurements, excitation and emission slit widths were set at 4 nm. For room temperature spectra, the excitation and emission slit widths were set at 8 nm and 4 nm, respectively. The room temperature spectra were corrected for photomultiplier tube response. Rhodamine B was used in the reference channel as a quantum counter. A tetraphenylbutadiene (TPD) block was used to adjust the voltage to 20000 counts per second in the sample as well as in the reference channels at excitation and emission wavelengths of 348 nm and 422 nm, respectively. Fluorescence spectra were measured three times and identical results were obtained. PSII, PSI and the whole chain electron transport in isolated thylakoid membranes. To probe further results obtained from Chl a fluorescence measurements, we monitored partial PSII and PSI electron transport reactions, as well as the whole photosynthetic electron transport chain in thylakoid membranes isolated from control and water-stressed seedlings.
PSII activity. The partial PSII electron transport, which was measured polarographically as light-driven electron transport from H 2 O to phenylenediamine, increased as the greening process of etiolated control seedlings progressed (Fig. 3A). Further, in 50 mM PEG-treated seedlings, PSII activity was reduced by 35% and 55% after 48 h and 72 h of greening compared to that in the controls.
PSI activity. The partial PSI electron transport, which was measured polarographically as light-driven electron transport from ascorbate/DCIP to methylviologen, increased in response to chloroplast development during light exposure. Similar to PSII, PSI electron transport decreased in water-stressed seedlings, compared to controls, but to a lower extent than that of PSII: i.e., by 20% and 25% after 48 h and 72 h of chloroplast biogenesis (Fig. 3B).
The whole (photosynthetic) electron transport chain. The whole chain electron transport through PSII and PSI, which was measured polarographically as light-driven electron transport from H 2 O to methylviologen, was reduced by 35% after 48 h and 65% after 72 h of greening in stressed seedlings compared to control (Fig. 3C).
Light saturation curves of PSII and PSI electron transport reactions. PSII. To further ascertain if the inhibition of PSII reaction was due to reduction in the quantum yield of PSII photochemistry measured in limiting light intensities or it was in light saturated electron transport, we measured the rate of PSII reaction as a function of different light intensities, using thylakoid membranes isolated from the control as well as from water-stressed seedlings, after 72 h of greening. The dependence of PSII activity on light intensity showed typical saturation kinetics (Fig. 3D). Both the initial slope at limiting light intensities as well as light-saturated electron transport, were affected in PEG-treated seedlings. As compared to the control thylakoids, the percent inhibition of PSII reaction in water-stressed thylakoids was almost constant (nearly 50%) at all the light intensities used  3D). The Eadie plot (i.e., the rate of oxygen evolution vs rate of oxygen evolution/light intensity in terms of % saturation 40 showed a straight line characterized by the equation y = −28.73x + 159.1 (R 2 = 0.948) for the control seedlings and y = −33.05x + 80.65 (R 2 = 0.917) for the water-stressed seedlings (insets in Fig. 3D). Both the intercepts on the abscissa and on the ordinate were reduced by nearly 50%.
PSI. In the PSI case, both the initial slope of electron transport rate at limiting light intensity as well as high light intensity i.e., for saturated electron transport rate, were reduced almost equally by ~30% (Fig. 3E). The Eadie plot showed a straight line with an equation y = −27.62x + 1028 (R 2 = 0.983) for the control seedlings and y = −31.03x + 739.8 (R 2 = 0.979) for the water-stressed seedlings (inset Fig. 3E). Intercepts on the abscissa and on the ordinates were reduced by nearly 30%.

Immunoblot analysis of photosynthetic proteins.
To understand the mechanism of water-stress induced down-regulation of photosynthesis during the development of the photosynthetic apparatus in the rice seedlings, we performed immunoblot analysis of proteins involved in PSII, PSI and the inter-system electron transfer.
PSII. Most of proteins/pigment-protein complexes of PSII increased upon illumination of the seedlings. The abundance of cyt b559, an intrinsic membrane protein intimately associated with PSII reaction center, declined by 58% and 52% from that in the controls, in the developing rice seedlings after 24 and 72 h of water-stress treatment (Fig. 4). Further, light-harvesting pigment-protein complexes associated with PSII i.e., Lhcb2 and Lhcb1, were reduced by 30% and 63.4% after 24 h of water-stress treatment, and by 30% and 64% after 72 h treatment (Fig. 4).
Oxygen Evolving Complex, OEC. Most of the oxygen evolving complex proteins were severely reduced in the rice seedlings subjected to water-stress during chloroplast development. The OEC16 and OEC23 were reduced by ~75% and 66% respectively in water-stressed seedlings (Fig. 4) after 24 h of stress. Similarly, at this time point, other major OEC protein i.e., OEC33 was reduced by ~50% (Fig. 4). Cytochrome b 6 /f. The protein abundance of cyt f, as well as the subunit IV of cyt b 6 /f complex declined by 4% and 15%, and by 40% and 7% after 24 h and 72 h of water-stress respectively during chloroplast biogenesis (Fig. 4).
PSI. In water-stressed seedlings, the protein abundance of the PSI subunit III (PsaF, 22 kD) was reduced by ~5%, while protein expression of the subunit V (PsaK, 17 kD) was severely reduced by ~80% after 24 h/72 h treatment (Fig. 4). Further, the PSI subunit VI (PsaH, 11 kD) was reduced by 88% and 67%, due to 24 h and 72 h water-stress. In contrast, PSI subunit IV (PsaE, 11 kD) had increased by 4 and 5 fold in water-stressed seedlings (Fig. 4). The PSI antenna proteins Lhca1 and Lhca4 were reduced by 75% and 46% after 24 h of water-stress treatment, and by 42% and 55% in seedlings water-stressed for 72 h.
Ultrastructure of chloroplast thylakoids. After 24-72 h of greening, both thylakoids and grana were well developed and contained starch granules in control samples. However, chloroplasts in the water-stressed seedlings had swollen thylakoids, disintegrated granal organization (mostly after 72 h) and fewer starch granules   41 . Since there was similar amount of H 2 O 2 production in both the control and water-stressed seedlings (after 24 h of greening), we did not observe any significant increase in MDA content either. After 72 h of greening under water-stress, the MDA content increased by 50% above the control value, as expected (Fig. 6B).   Differentially expressed proteins. Soluble and peripheral membrane proteome was analyzed by 2-D gel electrophoresis and MALDI-TOF/TOF or ESI-MS/MS analysis of spots, after 72 h of greening. Silver-staining as well as colloidal Coomassie Brilliant Blue (CBB)-staining was used to visualize the spots (Fig. 7A). In silver-stained gel, we identified 31 differentially expressed proteins; 10 were up-regulated and 21 were down-regulated. In CBB-stained gels, we further identified 15 up-regulated and 13 down-regulated proteins. In total, 34 down-regulated proteins and 25 up-regulated proteins were identified (Table 1): Silver-stained proteins are named starting with 'SS' , and CBB-stained spots are named staring with 'CB'; C3 and W3 denote spots from the 'control' and the 'water-stressed' samples, respectively, after 72 h/3 d of greening. Gene Annotations, Gene Index (GI) number, locus names, gene names and localization (wherever known or predicted) for identified proteins are provided in the Supplementary Table S1. Out of 59 differentially expressed identified proteins, for 18 Uniprot IDs, no gene models were retrieved from the rice database (http://ricedb.plantenergy.uwa.edu. au/). For the remaining 41 (59-18) Uniprot IDs, 56 corresponding gene-model/loci were retrieved as shown in the Supplementary Table S1. GO enrichment analysis performed using loci given in Supplementary Table S3, is  provided in the Supplementary Table S2. Enriched GO-terms (Biological Process) (p < 0.05) were glycolysis, tricarboxlic acid cycle, cellular carbohydrate metabolic process, oxidation-reduction, cysteine biosynthetic process, photosystem II stabilization, carbon fixation, hydrogen peroxide catabolic process, cysteine biosynthetic process from serine, phosphatidyl inositol metabolic process and phosphate metabolic process.
Category-wise distribution of down-regulated and up-regulated identified proteins is presented in Fig. 7A and B respectively. Differentially-regulated proteins were functionally distributed into 12 groups (Table 1).

Discussion
During photomorphogenesis, Chl, proteins, and lipids are synthesized and assembled to form the functional photosynthetic apparatus. Earlier, Dalal and Tripathy 33 showed that water-stress during transition from skotomorphogenesis to photomorphogenesis led to a 42% reduction in Chl content, due to down-regulation of expression of genes and proteins involved in Chl biosynthesis 33 . This must be, at least partly, responsible for the lower Fo and Fm (Fig. 1A), as well as the lower Chl a fluorescence in water-stressed rice seedlings (Fig. 4A). Shibata shift (blue shift of absorption maximum of chlorophyllide from 684 nm (aggregated form) to 672 nm (disaggregated form), discovered by K. Shibata 42 that takes place within a few minutes of light exposure of etiolated seedlings was also impaired in stressed seedlings 33 . In higher plants, both Chl a and Chl b are bound to light-harvesting pigment-protein complexes, LHCs. The availability of Chl b is essential for the assembly and functioning of most LHC proteins 43 . Binding of Chls to the LHC proteins stabilizes the latter in the thylakoid membranes 44 . Further, ROS produced in chloroplasts of water-stressed plants could down-regulate the expression of genes involved in Chl biosynthesis and photosynthesis via retrograde signaling 45,46 . In water-stressed rice seedlings, there was not only a downregulation of Chl biosynthesis, but also a reduction in pigment-protein complexes; this, obviously, led to lower Chl a fluorescence (Fig. 2). Moreover, biosynthesis of PSII reaction center protein D1 and its repair were also inhibited by H 2 O 2 , and other photosynthetic inhibitors and uncouplers 47,48 . In water-stressed developing rice seedlings components of light-harvesting complexes of both PSII and PSI i.e., Lhcb1, Lhcb2, Lhca1 and Lhca4 decreased (Fig. 4). Lack of Chl b in ch1 mutant of Arabidopsis resulted in the absence of accumulation of LHCs, even though they had normal mRNA expression for the LHCs 49 . Conversely, an increase of Chl b and total Chl content in tobacco upregulated gene expression and protein abundance of different pigment-protein complexes 50 .
In contrast to our present observation in developing rice seedlings, water-stress imposed on four-week-old well developed green wheat plants had no effect on the leaf chlorophyll content, and on the abundance of ATP synthase, PSII, PSI, and light-harvesting complexes 51 . However, in these plants, thylakoid membranes and other proteins were oxidized, due to the generation of ROS as a consequence of excess light absorption by the LHCs 51 . Decreased LHCII is an essential protection mechanism that allows plants to survive under unfavourable conditions 52 . Our results demonstrate that germinating seedlings protect themselves from water-stress-induced oxidative stress and photo-damage by downsizing their light-harvesting antenna and photosynthetic reaction centers (Figs 4 and 6).
The decreased Fo and Fm (Fig. 1A) in water-stressed rice seedlings is most likely due to a reduced Chl content 33 , and/or stress-induced alteration in xanthophyll-cycle dependent non-radiative energy dissipation 53  decrease in Fv/Fm, which is a proxy of PSII photochemistry efficiency 37 , in water-stressed developing seedlings (Fig. 1A), might have resulted from the down-regulation of synthesis and assembly of PSII reaction centers. This would have reduced not only the concentration of PSII reaction centers, but also of the oxygen evolving complex proteins, such as OEC33 (PsbO), OEC23 (PsbP) and OEC 16 (PsbQ) (Fig. 4). The observed changes in Chl a fluorescence spectra (Fig. 2) are in agreement with this observation. Furthermore, the "operating" quantum yield of PSII (φPSII) during actinic illumination also sharply declined in water-stressed developing seedlings (Fig. 1B), OEC33 is important for stabilization of Mn and Cl co-factors in water oxidizing complexes 54 ; its reduction via RNAi results in a decrease in quantum yield of photosynthesis 55 . Beside providing appropriate amounts of calcium and chloride ions for water splitting reactions, OEC23 and OEC16 play an additional role in the assembly of PSII supercomplexes, and their decreased abundance was clearly correlated with reduced PSII-dependent O 2 evolution 54 ; therefore, down-regulation of OEC proteins also explains the impaired PSII-dependent water oxidation and oxygen evolution that we found in thylakoid membranes isolated from water-stressed developing seedlings (Fig. 3).
Higher NPQ in water-stressed seedlings compared to the controls at all measured PAR values, especially at low actinic light intensities (Fig. 1C) suggest an adaptive response of developing seedlings for increased dissipation of absorbed energy as heat to protect their thylakoid membranes from photodamage 29 . Down-regulation of δ-subunit of ATP synthase was observed in our proteome profile of water-stressed seedlings ( Table 1). Impaired ATP synthase would have resulted in reduced transport of protons across the thylakoid membrane; the lowered dissipation of ΔpH and increased acidification of the lumen 56 could lead to higher NPQ in (water) stressed seedlings even at low light intensities.
In water-stressed samples, reduced amount of Chl, LHCs and reaction center proteins result in reduced light absorption, and decreased amount of oxygen evolving complex proteins result in reduced utilization of the absorbed energy. Consequently, reduction of PSII activity was almost similar (nearly 50%) under limiting as well as under saturating light intensities. In the same vein, the downsizing of the light-harvesting Chl-binding proteins of PSI i.e., LHCI (Lhca1 and Lhca4) and PSI core complex protein subunits III (PsaF), V (PsaK) and VI (PsaH) in water-stressed seedlings, resulted in diminished light absorption as well as light utilization by PSI, as was evident from the reduction of PSI reaction by 30% at limiting and higher light intensities (Fig. 3E). Relatively lower reduction of PSI activity than that of PSII (Fig. 3) suggests that development of PSI of developing rice seedlings is more tolerant to water-stress. The Eadie plot 31,40 for control and treated seedlings revealed a marked drop in the quantum efficiency of PSI and PSII reactions and a substantial reduction of Vmax in stressed developing  Table 1. Differentially expressed proteins in water-stressed rice seedlings categorized into different subgroups according to their function. a Spot No. generated by UMAX PowerLook 2100XL Image Scanner (v 6.0, GE Healthcare); Numbered spots correspond to the identified proteins,.refer the spots shown on the representative image in Fig. 7. b Accession number from NCBI database. SC, sequence coverage; Mr, theoretical Molecular weight; pI, theoretical Isoelectric point; Fold change refers to increased (Up) or decreased levels (down) of protein content from control to water-stressed seedlings. Fold change ± 2 was used as threshold for protein identification. Orthologues were obtained by using Uniprot protein sequence and BLAST.
Scientific RePoRTs | (2018) 8:5955 | DOI:10.1038/s41598-017-14419-4 seedlings (Fig. 3); these suggest a coordinated developmental down-regulation of light-harvesting and reaction center proteins of PSI as well as PSII in response to osmotic stress 31,40 . Room temperature fluorescence emission spectrum of thylakoid membranes had a peak at 684 nm due to PSII, when they were excited at 440 nm ( Fig. 2A). Water-stressed samples had a reduced fluorescence emission, probably due to a block in electron donation from water side of PSII; decreased abundance of oxygen evolving complex proteins (33 kDa, 23 kDa and 18 kDa) supports this hypothesis. Gross perturbation of structural organization of thylakoid membranes usually induces differences in fluorescence spectra at low temperature (77 K); these spectra usually have emission peaks at 685 nm (F 685 ), at 695 nm (F 695 ), which mostly originates from PSII CP43 and CP47 respectively 38 , and a F 735-740 peak that originates mostly from PSI 57 . If a part of LHCI antenna is removed from PSI by detergent treatment, the inner antenna of the PSI reaction center fluoresces at 722 nm 58 , while isolated LHCI complexes fluoresce around 735-740 nm; this is consistent with the assignment by Briantais et al. 59 : the inner PSI antenna for F 722 and for LHC I for F 735-740 59 . A shift to F 738 from F 740 in our water-stressed samples are due to a partial loss of components of LHCI.
In the absence of Mg 2+ , the F 686 /F 740 ratio in thylakoid membranes isolated from control seedlings was ~1.54 (Fig. 2B). In experiments by others, LHCII has been shown to migrate closer to PSII 60 in presence of Mg +2 , leading to stacking of thylakoids into granum 61 (see a review on role of ions by Kaňa and Govindjee) 62 . This results in efficient energy transfer from LHCII to the reaction center of PSII, and consequently, an increase in PSII fluorescence at 686 nm and, thus, higher F 686 /F 740 [see Results] ratio (an increase of F 686 /F 740 ratio to 1.73 in the control samples with Mg 2+ , compared to 1.54 without Mg 2+ (Fig. 2B)). In water-stressed seedlings, Mg +2 -induced increase in F 686 /F 740 was almost similar to control. Although a reduced grana stacking was observed in electron micrographs of the plastids of water-stressed seedlings, they retained the ability of Mg 2+ -induced migration of LHCPII to PSII core complex.
As revealed from our proteomics studies, water-stressed rice seedlings downsized their carbon reduction cycle enzymes i.e., Rubisco LSU, fructose bisphosphate aldolase (FBA), triosephosphate isomerase (TPI) and phosphoglycerate kinase (3PGK) ( Table 1). In contrast, in water-stressed well-developed plants, Rubisco content was often not affected, although its initial and total activity is known to decrease due to blocking of catalytic sites by 2-carboxyarabinitol-1-phosphate and other sugar phosphates 63 . Sugar phosphates can be removed from Rubisco by ATP-dependent enzyme Rubisco activase 17 , leading to its reactivation, hence underlining the role of Rubisco activase and ATP production in regulating carbon assimilation under drought. Drought-induced stomatal closure, and consequent decreased carboxylation, also result in down-regulation of protein abundance of Calvin-Benson cycle enzymes [13][14][15][16] .
Down-regulation of peptidyl prolyl isomerase (PPIase), in our water-stressed rice seedlings, suggest that there was a decrease in the ability to repair photosynthetic proteins, which, in turn, results in reduced PSI and PSII activities.
In mature plants, only minimal amounts of Chl and proteins are synthesized, mostly to replace those photodestroyed in PSI and PSII and LHCs, due to light. Therefore, it appears that mature plants do not downsize the components of the photosynthetic apparatus in response to drought. Several studies have demonstrated that abiotic stresses, including water-stress, cause ROS production due to an over-reduction of photosynthetic electron transport chain 25,29,64,65 . Abiotic stresses also impair reaction centers of well-developed seedlings or mature plants, leading to reduced transfer of absorbed light energy from light harvesting complexes to photodamaged reaction centers and consequently generation of higher amounts of singlet oxygen via photosensitization reactions of Chl 27,28,31 . Well developed rice plants treated for 48 h with 30% PEG 6000, a concentration similar to that used in our current experimental protocol, accumulated 2.5-4 fold higher H 2 O 2 , which impaired Rubisco activity and promoted stomatal closure 25 . However, we showed that germinating seedlings protect themselves by downsizing their light-harvesting antenna and photosynthetic reaction centers to reduce ROS production (Figs 4 and 6). In addition, we found that antioxidative enzymes i.e., ascorbate peroxidase 1 (APx1) and dehydroascorbate reductase (DHAR), were up-regulated in developing stressed seedlings (Table 1) to neutralize comparatively smaller amount of ROS generated. Therefore, we did not observe any increase of H 2 O 2 in developing seedlings subjected to water-stress for 24 h, and there was only a small increase (23%) after 72 h. Consequently, membrane lipid peroxidation, monitored as MDA production did not increase after 24 h of stress treatment, but it increased by 50% only after 72 h.
In conclusion, we can clearly state that unlike mature plants, seedlings exposed to water-stress during early photomorphogenesis, protect themselves from photo-oxidative stress by downsizing their photosynthetic apparatus. Therefore, we recommend that attempts should be made to down-regulate the light absorption during stress condition by reducing Chl biosynthesis. This can be achieved by either using molecular marker assisted plant breeding methods or by using transgenic RNAi approaches, silencing the expression of gene encoding for glutamyl-tRNA reductase or other early enzymes responsible for the synthesis of 5-aminolevulinic acid, a precursor of Chl biosynthesis. The RNAi expression could be modulated by promoters that sense increased ROS (e.g., H 2 O 2 ) produced during stress conditions.

Materials and Methods
Plants growth conditions and stress treatment. Seeds  Polarographic measurement of photosynthetic electron transport. Approximately 5 g of leaves were homogenized in 40 ml of isolation buffer containing 0.4 M sorbitol, 0.05 M Hepes/KOH (pH 7.3), 1mM MgCl 2 , and 1mM EDTA at 4 °C, under green safe light (28 Chakraborty and Tripathy, 1992). The homogenate was passed through 8 layers of cheese cloth and 1 layer of Mira cloth, and centrifuged at 5,000 rpm for 7 min. The pellet was suspended in a buffer solution containing 0.4 M sorbitol, 0.05 M Tris (pH 7.6), 1 mM MgCl 2 and 1 mM EDTA; Chl concentration was estimated according to Porra et al. 67 .
Photosynthetic electron transport was measured at 25 ± 1 °C with a Clark-type oxygen electrode (Hansatech, Kings Lynn, UK), by using 1 ml suspension of thylakoid membranes containing 20 µg Chl ml −1 ; illumination was with white-light from a tungsten source of 1,500 μmol photons m −2 s −1 . The whole electron transport chain, from H 2 O to methylviologen (MV) (1 mM), was monitored as O 2 uptake 68,69 . PSII activity was monitored as O 2 evolution resulting from electron transport from H 2 O to p-phenylenediamine (PD) (0.5 mM). The partial electron transport through PSI was measured as oxygen consumption. Ascorbate (1 mM)/DCIP (0.2 mM) couple was used as electron donor to PSI, and MV (1 mM) as electron acceptor 68,69 ; the electron flow from PSII was blocked by 3-(3, 4-dichlorophenyl) 1,1-dimethyl urea (DCMU) (20 μM), and 1 mM sodiumazide was used as an inhibitor of catalase.
Light saturation curves were measured at several light intensities obtained by using neutral density filters (Balzers, Neugrüt, Lichtenstein).
Transmission electron microscopy. Leaves (in triplicate) were fixed with 2.5% glutaraldehyde 70 and 1% OsO 4 for 2-4 h, and then washed, dehydrated with acetone, cleared with epoxy propane/xylene, and infiltrated with resin containing araldite and toluene (up to 75% araldite). After embedding, araldite was polymerized at 50 °C, and then at 60 °C. Ultra-thin sections were cut, stained in saturated uranyl acetate in 50% ethanol, washed in 50% ethanol and water, and viewed in a Transmission Electron Microscope (TEM) (Phillips-CM-10, Eindhoven, Netherlands).
Chlorophyll a fluorescence measurements. Chl a fluorescence signal is a highly sensitive signature of diverse aspects of photosynthesis, and thus, it is widely used in photosynthesis research 38,71 (see chapters in Papageorgiou and Govindjee) 72 .
Fluorescence emission spectra of isolated thylakoids. Room temperature (RT; 298 K) and low temperature (77 K) fluorescence emission spectra of isolated thylakoids were recorded with a SLM-AMINCO-8000 spectrofluorometer; for 77 K-spectra, the excitation and emission slit widths were 4 nm, while for RT-spectra, the excitation and emission slit widths were 8 nm and 4 nm respectively. Rhodamine-B was used in the reference channel as quantum counter. A tetraphenylbutadiene block was used to adjust the voltage in sample, as well as in reference channels to 20000 counts s −1 at excitation and emission wavelengths of 348 nm and 422 nm, respectively. RT-spectra were corrected for the instrument response.
Chl a fluorescence induction curves. Chl a fluorescence induction curves at 25 °C were recorded on attached leaves, with a portable PAM-2100 fluorometer (Walz, Effeltrich, Germany); all measurements were repeated 3 times. The initial (Fo) and maximum (Fm) fluorescence, was measured on seedlings that were dark-adapted for 20 min; the measuring light was red (650 nm), and of very low intensity (<0.1 μmol photons m −2 s −1 ); its frequency was 0.6 KHz. Then, a 0.8 s saturation light pulse of approximately 8,000 μ mol photons m −2 s −1 was applied to measure the maximum fluorescence, Fm. The Fo and the Fm values were used to calculate the ratio Fv/Fm = (Fm − Fo)/Fm, where Fv is the maximum variable fluorescence; this ratio represents the maximum quantum yield of PSII photochemistry. Light response curves were obtained by measuring fluorescence as a function of increasing actinic light intensity (4 to 270 μmol photons m −2 s −1 ; wavelength: 665 nm). The quantum yield of PSII (φPSII) was calculated as (Fm′ − Ft)/Fm′, where Fm′ is the maximum fluorescence at steady state in pre-illuminated samples, and Ft is fluorescence immediately before the application of the saturation pulse 37 . Non-photochemical quenching (NPQ) was calculated as (Fm − Fm′)/Fm′. Thylakoid isolation and western blotting. 20 μg of plastid protein 28 , estimated as described by Bradford 73 , was separated on SDS-PAGE three times for each photosynthetic target protein, transferred to nitrocellulose (NC) membranes and blocked with BSA solution. Subsequently, blots were incubated with polyclonal and heterologous primary antibodies (Supplementary Table S4) and alkaline phosphatase-conjugated antibodies and developed for color. Image J (NIH, USA) was used for quantification.