Ethylene Regulates Energy-Dependent Non-Photochemical Quenching in Arabidopsis through Repression of the Xanthophyll Cycle

Energy-dependent (qE) non-photochemical quenching (NPQ) thermally dissipates excess absorbed light energy as a protective mechanism to prevent the over reduction of photosystem II and the generation of reactive oxygen species (ROS). The xanthophyll cycle, induced when the level of absorbed light energy exceeds the capacity of photochemistry, contributes to qE. In this work, we show that ethylene regulates the xanthophyll cycle in Arabidopsis. Analysis of eto1-1, exhibiting increased ethylene production, and ctr1-3, exhibiting constitutive ethylene response, revealed defects in NPQ resulting from impaired de-epoxidation of violaxanthin by violaxanthin de-epoxidase (VDE) encoded by NPQ1. Elevated ethylene signaling reduced the level of active VDE through decreased NPQ1 promoter activity and impaired VDE activation resulting from a lower transthylakoid membrane pH gradient. Increasing the concentration of CO2 partially corrected the ethylene-mediated defects in NPQ and photosynthesis, indicating that changes in ethylene signaling affect stromal CO2 solubility. Increasing VDE expression in eto1-1 and ctr1-3 restored light-activated de-epoxidation and qE, reduced superoxide production and reduced photoinhibition. Restoring VDE activity significantly reversed the small growth phenotype of eto1-1 and ctr1-3 without altering ethylene production or ethylene responses. Our results demonstrate that ethylene increases ROS production and photosensitivity in response to high light and the associated reduced plant stature is partially reversed by increasing VDE activity.

In order to investigate how ethylene might affect photosynthetic processes, we employed mutants altered in ethylene production or signaling in Arabidopsis. ethylene overproducer 1-1 (eto1-1) produces elevated levels of ethylene [14,80,81]. ETO1 interacts with type 2 ACC synthases and reduces their activity either through direct inhibition or through promoting their degradation in a proteasome-dependent manner [82]. Loss of ETO1 results in constitutively active type 2 ACS and increased ethylene evolution. When grown in light, eto1-1 plants are significantly smaller than wild-type plants [80]. The loss of CTR1 expression in the constitutive triple response 1-3 (ctr1-3) mutant disrupts the ability of ethylene receptors to repress the activity of the downstream components of the ethylene response pathway, resulting in a constitutive ethylene response [14]. The ctr1-3 mutant is characterized by a substantial reduction in cell size and plant stature, greater than that exhibited by eto1-1 plants [14]. In contrast, the ethylene insensitive 2-5 (ein2-5) mutant is insensitive to ethylene due to the lack of EIN2 expression which is epistatic to CTR1 [19] and is required to activate ethylene responses [83].
In this study, we demonstrate that increased ethylene signaling affects the xanthophyll cycle directly and indirectly. eto1-1 and ctr1-3 exhibit aberrant induction of NPQ following exposure to light which can be corrected by inhibiting ethylene perception, e.g., in eto1-1. The defects in qE and qI observed in these mutants were associated with an impaired functioning of the xanthophyll cycle resulting from reduced expression and activation of VDE, a reduction in PsbS expression, and a lower transthylakoid membrane pH gradient, and an increase in superoxide production. eto1-1 did not exhibit the compensating increase in α-tocopherol and ascorbic acid content that occurs in npq1 which is a VDE null mutant. Restoring VDE activity in eto1-1 and ctr1-3 reduced superoxide production and photosensitivity upon exposure to high light. Restoring VDE activity also reversed the small stature of eto1-1 and ctr1-3 specifically under high light growth conditions without affecting ethylene production or responsiveness. Restoring VDE activity in eto1-1 and ctr1-3 did not improve growth under low light, implicating ROS as contributing to the small stature of eto1-1 and ctr1-3 under high light conditions. These results demonstrate that ethylene represses functioning of the xanthophyll cycle while increasing ROS production and reversing these effects improves growth under high light.
containing the Aglo1 strain of Agrobacterium containing the transgene. Infiltrated plants were kept on their side for one day and allowed to continue to flower in an upright position in the same growth room. Seeds of infiltrated plants were collected and screened on 0.25 x MS plates containing 50 μg/ml kanamycin and 500 μg/ml vancomycin.
For high light experiments, leaves that had been dark-adapted for 16 hours were floated on ice-containing water and exposed to high light (1300 PFD as supplied from high output sodium light) or sunlight (1900 PFD) for the times indicated. F o and F m were measured immediately before the high light exposure and at time points during recovery. For the analysis, leaves with similar initial F v /F m values were used.

Imaging NPQ
The induction of NPQ and its relaxation were performed using an IMAGING-PAM M-Series Chlorophyll Fluorometer (Heinz Walz GmbH, Effeltrich, Germany). Fluorescence measurements of plants dark-adapted overnight were taken using a relative humidity of 50% and an ambient level of CO 2 . At the start of each experiment, the leaf was exposed to 2 min of far-red illumination (1 PFD) for the determination of F o (minimum fluorescence in the dark-adapted state). False color was applied to the images in a range from black (representing a value of 0) to purple (representing a value of 1). For inhibition of PSII quantum yield measurements, F v /F m was measured immediately after the actinic light was turned off and continued for 25 min at 1 min intervals. Inhibition of PSII quantum yield was calculated from (F v /F m of reference − F v / F m of sample)/(F v /F m of reference) where an area selected from a non-stressed WT plant was used as the reference.

Gas exchange and fluorescence measurements
Gas exchange and fluorescence measurements were performed using a LI-COR Li-6400 portable photosynthesis system (LI-COR, Lincoln, NE) with LI-6400-40 leaf chamber, a relative humidity of 50%, and ambient level of CO 2 . Fluorescence measurements were taken using overnight dark-adapted leaves. At the start of each experiment, the leaf was exposed to 2 min of far-red illumination (1 PFD) for the determination of F o (minimum fluorescence in the dark-adapted state). Saturating pulses (0.8 s of 5000 PFD) were applied to determine the F m or F m ' values. Actinic light, consisting of 90% of red light (λ = 630 ± 20 nm) and 10% blue light (λ = 470 ± 20 nm) was provided by LED (light emission diode) sources. F s is the steady fluorescence yield during actinic illumination. F o ' (minimum fluorescence in the light-adapted state) was determined in the presence of far-red (λ = 740 nm) light after switching off the actinic light. A total of four to six samples were measured in each experiment. All data presented were calculated from at least three independent measurements. Conventional fluorescence nomenclature was used [84]. NPQ was calculated from (F m -F m ')/F m ', φPSII from (F m '-F s ')/F m ', qP from (F m '-F s )/(F m '-F o '), and the electron transport rate (ETR) from φPSII Ã f Ã α leaf , where f is the fraction of absorbed quanta that is used by PSII and is typically assumed to be 0.5 for C3 plants; α leaf is leaf absorbance. NPQ f and NPQ s were determined as described [85].
Chlorophyll, α-tocopherol, and xanthophyll pigment measurements Chlorophyll a and b were measured spectrophotometrically as described [86]. Leaf samples were ground in liquid nitrogen and extracted with 90% (v/v) acetone. The absorbance at 664 and 647 nm was determined and used to calculated chlorophyll a and b content by the equations: Chl a = 11.93A 664 -1.93A 647 and Chl b = 20.36A 647 -5.50A 664 , respectively. Each experiment was repeated 2-3 times and representative results presented.
For α-tocopherol, young Arabidopsis leaves were extracted with 1 ml methanol, dried, and re-dissolved with 0.2 mL ethanol. Samples were filtered (0.2 μm Whatman GD/X) and loaded onto a Spherisorb ODS-1 column (Alltech) which was eluted with a gradient of hexane (solvant A) methyl-t-butyl ether (solvent B) as described [88]. α-Tocopherol content was calculated using a standard curve of α-tocopherol.

qPCR analysis
Plant material was frozen in liquid nitrogen, ground to a fine powder, and 100 mg was resuspended in 1 ml TRIZOL 1 Reagent (Invitrogen, Carlsbad CA, USA). Following centrifugation, the supernatant was extracted with 200 μl chloroform and centrifuged to separate the phases. RNA was precipitated from the aqueous phase using isopropyl alcohol, the RNA pellet washed with 75% ethanol and resuspended in RNase-free H 2 O. 1 μg RNA was used to obtain the firststrand cDNA by Omniscript RT Kit (Qiagen, Valencia CA, USA) in a 20 μl reaction. The qPCR analysis was performed using a iQ5 Real-Time PCR Detection System (Bio-Rad, Hercules CA, USA) in 25 μl reactions containing 1x SYBR Green SuperMix 500 nM forward and reverse primers and 10 ng cDNA. ROX was used as the passive reference dye. Reactions were carried out using the following conditions: 95°C/5 min (1 cycle); 95°C/30 sec, 55°C/30 sec, 72°C/30 sec (35 cycles). To detect the presence of NPQ1, a forward primer, NPQ1-F1, 5'-ATGACTGGTATATCCTGTCATC-3', and a reverse primer, NPQ1-R1, 5'-CGTTCTAAT GAATGTGCTGAAG-3' were used. To detect the presence of NPQ4, a forward primer, NPQ4-F1, 5'-TATGATCGGTTTCGCTGCATC3', and a reverse primer, NPQ4-R1, 5'-CAA CAGAGTGAACAAGATGAAG-3' were used. Protein phosphatase PP2A (At1g13320) was used as the reference gene for the quantitation of NPQ1 and NPQ4 expression in Arabidopsis leaves. To detect the expression of PP2A, a forward primer, PP2A-FW, 5'-AGTATCGCTT CTCGCTCCAG-3' and a reverse primer, PP2A-RV, 5'-GTTCTCCACAACCGCTTGGT-3' were used. The efficiency of PCR was determined by five 10-fold serial dilutions of the template DNAs in triplicate. Three biological replicates were used for each target gene.

Ascorbate measurements
HPLC analysis of ascorbate acid was performed as previously described [90]. Samples were ground with liquid nitrogen and approximately 0.1 g of leaf powder was extracted with 0.5 ml of 0.2% metaphosphoric acid contaning 0.54 mM Na 2 EDTA and 0.01% polyvinylpolypyrrolidone. Extracts were centrifuged at 12,000 x g for 10 min at 4°C to remove debris. The supernatant was filtered (0.25 μm) and 20 μl was loaded onto a C-18 ODS column and eluted with isocratic mobile phase (20 mM NaAc, 0.54 mM Na 2 EDTA, 1.5 mM N-octylamine) at a flow rate of 0.5 ml/min. For dehydroascorbic acid analysis, the extract was neutralized with 0.5 M NaOH to pH 6.5, reduced with 2.5 mM glutathione and 2 ng of recombinant dehydroascorbate reductase for 20 min at room temperature. Ascorbate was detected by a Waters 996 photodiode array detector at 265 nm using L-ascorbate as a standard to generate a standard curve.

VDE, ZE, and CA activity assays
To measure VDE activity, the thylakoid membrane fraction was isolated from Arabidopsis leaves essentially as described [89]. VDE activity was measured in a reaction containing 10 μL of 1 μm violaxanthin in methanol, 25 μL of 300 μM monogalactosyldiacylglycerol (Lipid Products, South Nutfield, UK) in methanol, 550 μL of 0.2 M sodium citrate pH 5.1, and 20-50 μL of the thylakoid membrane fraction (equivalent to 25-50 μg Chl a). The reaction mixture was vortexed, incubated for 5 min at 30°C, and initiated following the addition of 6 μL of 3M sodium ascorbate. After 4-10 min, the reaction was stopped by the addition of 1N NaOH, centrifuged at 20,000 x g for 2 min, and the pellet containing the lipids and pigments analyzed by HPLC as described [91].
To measure ZE activity, the chloroplast fraction was prepared and ZE activity measured as described [92]. Lysed chloroplast extract equivalent to 18 μg Chl a was added to a reaction containing 0.5 μm zeaxanthin, 400 mM sorbitol, 50 mM HEPES-NaOH pH 7.2, 16 mM sodium ascorbate, 0.5 mM NADPH, and 0.3 mg/ml BSA. The reaction was incubated at room temperature in the dark for 40 min. Total pigments were extracted with methanol, dried, and re-dissolved with acetone. The pigments were determined by HPLC.
To measure carbonic anhydrase (CA) activity, total leaf soluble extract and the chloroplast fraction was prepared and the CA assay performed essentially as described [93]. Extract was added to a reaction containing 100 mM potassium phosphate pH 7.7, and 10 mM DTT. The reaction was initiated following the addition of 3 ml of CO 2 -saturated water. While stirring, the decrease in pH over time was monitored. Units of CA activity were calculated as (Tb/Te)-1, where Tb and Te represent the time (in sec) for the pH to drop from 7.5 to 7.0 in the control and sample reactions, respectively.
Three to four biological replications were assayed for each of these assays. The average and standard derivation was reported.

Luciferase assay
Cell extract representing the soluble protein fraction prepared from seedlings in luciferase assay buffer [20 mM Tricine, pH 7.8, 1.07 mM (MgCO 3 ) 4 Mg(OH) 2 -5H 2 O, 2.67 mM MgSO 4 , 0.1 mM EDTA, 33.3 mM DTT, 270 μM CoA, and 500 μM ATP (Promega)], and the reaction was initiated with the injection of 100 μL of 0.5 mM luciferin in luciferase assay buffer. Photons were counted using a Monolight 2010 Luminometer (Analytical Luminescence Laboratory, San Diego, CA). Each mRNA construct was assayed in duplicate and the average is reported. Protein concentration was determined as described [94].

Determination of the transthylakoid membrane pH gradient
Intact chloroplasts from WT or mutant Arabidopsis were isolated as described [97]. Chloroplasts equivalent to 15 μg chlorophyll were added to 50 mM Hepes-KOH, pH 7.8, 5 mM MgCl 2 , 330 mM sorbitol, 0.4 μM 9-AA to a final volume of 1 ml. Fluorescence from 9-AA was monitored with a modular spectrofluorimeter (Fluorolog-3, Horiba Scientific, Edison, NJ) using an excitation wavelength of 365 nm and fluorescence of 9-AA detected by scanning from 450 to 460 nm at 0.02 sec intervals. Actinic light was supplied with either high-power LED lights or with a 150W reflector spot lamp. Only the uncharged form fluoresces and moves freely across the thylakoid membrane whereas its protonated form neither fluoresces nor moves across membranes. At physiological pH, the protonated form of 9-AA (K a = 1.26 x 10 -10 ) predominates. Following the generation of a transthylakoid membrane pH gradient, the 9-AA accumulates in the thylakoid lumen where it is protonated and quenched. The amount of 9-AA that moves into thylakoid lumen is determined from the quenching of 9-AA fluorescence from which the ΔpH is calculated using the equation: [H + ] in = K a x [AH + ]/[A] in , where [H + ] in is the proton concentration of the thylakoid lumen; K a = 1.26x10 -10 ; [AH + ] is the concentration of protonated 9-AA; and [A] in is the concentration of uncharged 9-AA in the thylakoid lumen [98]. The ΔpH is determined from the difference between the pH of the thylakoid lumen and the buffer.

Determination of the quantum requirement for photosynthesis
The quantum requirement was determined using a LI-COR Li-6400 portable photosynthesis system in which chlorophyll fluorescence and the rate of CO 2 assimilation in leaves from adult plants was measured simultaneously under a series of actinic light levels (from 0 up to 1200 PFD), under low oxygen air (2% oxygen, 98% nitrogen) supplied with 400 ppm CO 2 . The quantum yield of PSII is calculated by φPSII = (F m '-F s )/F m ', where F m ' is the maximal fluorescence during a saturating light flash, and Fs is steady-state fluorescence. The quantum yield of CO 2 assimilation is calculated by φCO 2 = (A-A dark )/Iα leaf where A is the assimilation rate, A dark is dark assimilation rate (both with units of μmol CO 2 •m -2 •s -1 , but A dark is a negative value). I is the incident photon flux density (μmol •m -2 •s -1 ), and α is leaf absorbance (a value of 0.85 is used). The values of φPSII were then plotted against φCO 2 values, and the quantum requirement calculated from the slope. Each data point was the average of four to six individual measurements.

Ethylene determination
Ethylene was measured from whole seedlings which were placed in glass vials and capped with a rubber septum. Following a 2 hour incubation, 0.9 ml of headspace was sampled from each vial and the ethylene content measured using a 6850 series gas chromatography system (Hewlett-Packard, Palo Alto, CA) equipped with a HP Plot alumina-based capillary column (Agilent Technologies, Palo Alto, CA), which can detect as little as 10 nl/l (10 ppb) ethylene. The ethylene peak was identified as that which had the same retention time as pure ethylene. Tissue fresh weight was measured for each sample. Three to four replicates were measured and the average and standard deviation reported.

Results
Increased ethylene signaling results in an aberrant induction of nonphotochemical quenching NPQ is a light-inducible process that competes for and diverts absorbed energy from photochemistry in order to regulate light harvesting and photosynthetic performance during conditions of excess light [48]. Following exposure of dark-adapted plants to 336 μmol photons m -2 s -1 (PFD), an aberrant NPQ induction profile was observed in leaves of eto1-1, in which ethylene production is elevated, and in ctr1-3, in which the ethylene response is constitutive (Fig 1). Although eto1-1 and ctr1-3 exhibit reduced plant stature resulting from an ethylene-mediated reduction in cell size [14,80,81], NPQ is a ratio of fluorescence values, and therefore, is independent of leaf area or cell size. Thus, comparisons of NPQ among mutant and wild-type (WT) plants can be made without the need for normalization for cell size. During the initial exposure to light, NPQ was lower in eto1-1 and ctr1-3 relative to WT and the ethylene insensitive ein2-5 mutant (Fig 1A and 1B) but its level in eto1-1 eventually exceeded the WT level under these conditions (Fig 1D and 1E). During recovery in the dark, measurements of the inhibition of PSII quantum yield indicated that eto1-1 was slower to recover ( Fig 1F). Using endpoint measurements in plants with a similar dark-adapted F v /F m , NPQ was significantly elevated in ctr1-3 (p < 0.05) than in WT leaves following 60 min of exposure to 400 PFD ( Fig  1G). Because elevated ethylene production or signaling results in a complex and dynamic alteration of total NPQ accumulation in response to light exposure, whether NPQ in eto1-1 or ctr1-3 is observed to be lower or higher than WT depends on when it is measured during its induction (see discussion of S1A Fig below).
NPQ is composed of qE, which serves to dissipate excess absorbed excitation energy as heat, qT, which is quenching associated with state transition, qI or photoinhibitory processes, and qZ, which is independent of PsbS and distinct from state transitions (qT) [47]. The processes that contribute to NPQ can be grouped into those that relax quickly following the transfer of light-treated plants to the dark and those that relax slowly. These can be measured as the fast and slow relaxation components of NPQ (i.e., NPQ f and NPQ s , respectively) during the recovery of plants from exposure to high light. NPQ f largely represents qE whereas NPQ s represents slower components of NPQ, including qI. To examine whether the aberrant induction characteristics of NPQ in eto1-1 and ctr1-3 are due to a change in the fast or slow relaxation components of NPQ, we measured NPQ f and NPQ s . For dark-adapted plants grown at 250 PFD and exposed to 1800 PFD, NPQ f was significantly lower in eto1-1 (p < 0.001) and ctr1-3 (p < 0.0001) relative to WT (Fig 1G), and NPQ s significantly higher in ctr1-3 (p < 0.001) relative to WT (Fig 1G), suggesting an impaired induction of qE that is accompanied by an increase in the slow relaxation component of NPQ.
To determine the kinetics of its induction, we measured NPQ in dark-adapted eto1-1 and ctr1-3 plants grown at 250 PFD following exposure of leaves to 400 PFD. NPQ was induced in WT leaves to a high level within 5 min following the imposition of light after which point, NPQ decreased to a steady-state value (S1A Fig). The rapid increase in NPQ immediately following exposure to light is a result of excess light conditions prior to the activation of Calvin cycle reactions. The subsequent decrease in NPQ represents full activation of photochemistry, a process that competes with NPQ for light energy. Following its initial induction and transient decline in eto1-1 and ctr1-3, however, NPQ rose again to levels exceeding the steady-state WT level (S1A Fig). Measuring the induction of NPQ at shorter time intervals to obtain information regarding initial rates revealed a slower initial induction in eto1-1 and ctr1-3 which, over time, exceeded the WT level (S1B Fig). The slower initial induction followed by a higher level of NPQ attained in eto1-1 and ctr1-3 suggested an impaired induction of qE and an elevated qI, consistent with the observed reduction in NPQ f and increase in NPQ s , respectively, in these mutants ( Fig 1G). These data reveal that elevated ethylene production or signaling alters total NPQ accumulation in a complex and dynamic manner that results in lower or higher total NPQ accumulation depending on when NPQ is measured during its induction.
A second type of qE quenching, qE TR , occurs transiently following the transfer of darkadapted plants to nonsaturating light [99,100]. The generation of a transthylakoid membrane proton gradient and PsbS are essential for qE TR , but the magnitude of qE TR is also determined by the extent of xanthophyll de-epoxidation [99,100,101]. When dark-adapted mutant and WT leaves were exposed to 100 PFD, representing a nonsaturating level of light, NPQ was induced transiently (S1C Fig), representing the light-dependent generation of a transthylakoid membrane pH gradient followed by its breakdown with the activation of Calvin cycle reactions [99,101,102]. The maximum initial level of NPQ achieved in eto1-1 was lower than in WT and was lower still in ctr1-3, suggesting a possible reduction in xanthophyll cycle activity and/ or PsbS expression.
The elevated ethylene signaling present in ctr1-3 cannot be reversed using standard inhibitors of ethylene biosynthesis or perception. Therefore, to determine whether inhibiting ethylene signaling prevents the aberrant induction of NPQ in eto1-1, we examined the effect of 1-MCP, which blocks ethylene binding to receptors and thus inhibits induction of ethylene responses [103]. Exposure of dark-adapted eto1-1 plants to 400 PFD resulted in the eventual elevated induction of NPQ relative to WT (S1D Fig) as observed above. Treatment with 1-MCP for 20 hr largely restored a wild-type NPQ induction profile to eto1-1 and reduced the final level of NPQ achieved in eto1-1 to a level lower than WT (S1D Fig). Treatment of WT plants with 1-MCP did not significantly alter the kinetics of NPQ induction relative to untreated WT plants. These results support the conclusion that increased ethylene signaling alters the induction of NPQ in response to light.

Violaxanthin de-epoxidase is repressed by increased ethylene signaling
The xanthophyll cycle contributes to qE through the light-mediated generation of Z [70,104]. Exposure to light results in the activation of violaxanthin de-epoxidase (VDE) which catalyzes the de-epoxidation of V to A and Z [105]. As a result, V predominates in dark-adapted plants. Consistent with this, we observed that the predominant xanthophyll cycle pigment in darkadapted eto1-1, ctr1-3, and WT leaves was V (Fig 2A). In order to determine whether the aberrant induction of NPQ in eto1-1 and ctr1-3 resulted from the reduced de-epoxidation of V, we measured the extent of de-epoxidation in each during exposure to light. Following exposure to 500 PFD, significant de-epoxidation of V to Z was observed in WT leaves within 5 min with additional de-epoxidation occurring upon longer exposure (Fig 2A and 2B). In contrast, the rate of de-epoxidation was significantly lower in eto1-1, resulting in a de-epoxidation state that was substantially lower than that in WT (p < 0.05, p < 0.005, and p < 0.05 at 5, 10, and 30 min, respectively) (Fig 2A and 2B). De-epoxidation of V was lower still in ctr1-3 (p < 0.005, p < 0.001, and p < 0.001 at 5, 10, and 30 min, respectively), resulting in an initial de-epoxidation state that was reduced more than three fold relative to WT (Fig 2A and 2B). The pool size of lutein and neoxanthin was not substantially different among the lines (data not shown). These data suggest that de-epoxidation activity is reduced in eto1-1 and ctr1-3, consistent with their lower initial induction of NPQ (S1B Fig) and lower NPQ f ( Fig 1G).
VDE, which catalyzes the de-epoxidation reaction, requires ascorbic acid (Asc) as a cofactor ( Fig 3A) [106]. To determine whether a reduction in Asc may account for the observed reduction in de-epoxidation activity in eto1-1 and ctr1-3, we measured the levels of Asc and its oxidized form, dehydroascorbate (DHA). The levels of Asc, DHA, and the Asc redox state in eto1-1 and ctr1-3 were similar to WT (Fig 3B), suggesting that their reduced de-epoxidation activity was not a result of limited Asc availability.
The reduction in de-epoxidation activity in eto1-1 and ctr1-3 could result from a reduction in the expression and/or activation of VDE ( Fig 3A). To determine whether the transcript level of VDE was affected in eto1-1 and ctr1-3, we performed qPCR analysis of NPQ1 mRNA. No expression was detected in the npq1 mutant (Fig 3C), which lacks VDE expression [49], confirming the specificity of the qPCR analysis. The npq1 mutant also lacked VDE protein as determined by Western analysis (Fig 3E), confirming the specificity of the antiserum. The level of NPQ1 mRNA in ctr1-3 was 52.1% of WT (p < 0.01) (Fig 3C), consistent with the lower level of VDE protein in this mutant ( Fig 3E). The level of NPQ1 mRNA ( Fig 3C) and VDE protein ( Fig 3E) in eto1-1 was similar to WT, suggesting a high level of ethylene signaling is required for the repression of VDE expression. The level of NPQ1 mRNA in ein2-5 was largely unchanged relative to WT (Fig 3C), correlating with a level of VDE protein that was similar to WT ( Fig 3E). VDE expression was also unchanged in the npq4 mutant (Fig 3C), which is null for PsbS [107], demonstrating that VDE expression is not affected by the absence of PsbS expression. We also performed qPCR analysis of PsbS, which is required for qE. The level of PsbS mRNA in ctr1-3 was 42.0% of WT (p < 0.05) (Fig 3D), consistent with the lower level of PsbS protein in the mutant (Fig 3E). A more moderate reduction in PsbS transcript was observed in eto1-1 but this was not significantly lower than WT ( Fig 3D). These results suggest that a high level of ethylene signaling represses VDE and PsbS expression.
To determine whether a reduction in VDE enzyme activity correlated with the transcript and protein data for eto1-1 and ctr1-3, we measured its activity. The assay for VDE involves an in vitro pH-mediated activation of its activity, mimicking the light-mediated acidification of the thylakoid lumen where VDE resides. As a consequence, the VDE assay measures the maximum capacity of VDE activity, not necessarily the level of activation achieved in vivo, as this depends on the magnitude of the transthylakoid membrane pH gradient established. The maximum level of VDE activity measured in ctr1-3 was just 43.6% of the WT level (Fig 3F), correlating with the reductions in transcript and protein levels observed for this mutant. The maximum level of VDE activity measured in eto1-1 was similar to the WT level (Fig 3F), which also correlated with a level of transcript and protein that was similar to WT. The reduced level of de-epoxidation in light-treated eto1-1 and ctr1-3 could also have been a result of an increase in zeaxanthin epoxidase (ZE) activity, which catalyzes the epoxidation of Z to A and then to V, i.e., the reverse of VDE activity ( Fig 3A). ZE enzyme activity in eto1-1 and ctr1-3, however, was little changed relative to the WT level ( Fig 3F), suggesting that the slower rate of de-epoxidation observed in eto1-1 and ctr1-3 was not due to altered ZE activity.

NPQ1 promoter activity is regulated by ethylene
The reduction in NPQ1 transcript level in ctr1-3 may result from reduced NPQ1 promoter activity or increased NPQ1 mRNA turnover. To distinguish between these two possibilities, we fused the NPQ1 promoter to the firefly luciferase reporter and, following transformation of wild-type Arabidopsis, we isolated lines homozygous for each construct. We used Arabidopsis containing the 35S promoter driving expression of luciferase as a control. We grew seedlings homozygous for each construct in the presence or absence of ACC, which is converted by the seedlings to ethylene, thus elevating the endogenous production of ethylene ( Fig 4A). The level of luciferase activity in 35S::Luc seedlings grown in the presence or absence of ACC was similar (Fig 4B), demonstrating that a 10.6-fold increase in ethylene production does not substantially affect luciferase expression from the 35S promoter. In contrast, an approximate 2-fold reduction in expression in NPQ1::Luc seedlings was observed when grown in the presence of ACC relative to growth in the absence of ACC (Fig 4B), suggesting NPQ1 promoter activity is repressed by elevated ethylene.
We also examined genetically the regulation by ethylene of NPQ1 promoter activity by introducing the NPQ1::Luc construct or the 35S::Luc construct into eto1-1, ctr1-3, and ein2-5 through crosses with wild-type lines containing these constructs. We then isolated progeny homozygous for each Luc construct and the respective ethylene mutation. NPQ1 promoter activity as measured by luciferase activity was reduced in eto1-1 and ctr1-3 to 88.7% and 53.3%, respectively, of the level observed in WT seedlings (Fig 4C). NPQ1 promoter activity was slightly greater in ein2-5 than in WT seedlings but this difference was not significant (Fig 4C). Expression from the control 35S::Luc construct in eto1-1, ctr1-3 and ein2-5 was 104%, 80.9%, and 104% respectively, of that in WT seedlings ( Fig 4C). These data indicate that NPQ1 promoter activity is repressed by an increase in ethylene signaling resulting from increased endogenous ethylene production (e.g., WT seedlings grown in the presence of ACC) or genetically through an increase in ethylene responses (e.g., eto1-1 and ctr1-3). As the reduction in NPQ1 promoter activity observed in eto1-1 was modest at best, a significant increase in ethylene signaling is required (e.g., ctr1-3) for the repression of NPQ1 promoter activity to be reflected at the protein ( Fig 3E) and enzyme activity levels ( Fig 3E). The extent of the repression of NPQ1 promoter activity observed in ctr1-3 is consistent with the approximate 2-fold reduction in NPQ1 transcript (Fig 3C), protein (Fig 3E), and in vitro enzyme activity (Fig 3F). The even greater reduction in the rate of de-epoxidation observed in vivo for eto1-1 and ctr1-3 (Fig 2B), however, raised the possibility that the light-mediated activation of VDE enzyme activity may also be repressed following an increase in ethylene signaling.

Increased ethylene signaling reduces the light-mediated activation of VDE
Although a modest reduction in NPQ1 promoter activity was observed in eto1-1, the lack of change in NPQ1 transcript (Fig 3C) or protein (Fig 3E) suggested that the reduction in deepoxidation observed for this mutant (Fig 2) may be due to a lower degree of activation of VDE in response to light. VDE activity is regulated by the pH of the lumen whereas ZE is constitutively active. VDE undergoes activation during the acidification of thylakoid lumen which is generated by proton pumping from the stroma during photosynthesis, resulting in a lumen that can approach pH 4-5 and a stroma that can approach pH 8 [108]. Stromal pH determines the concentration of total carbonic CO 2 which is mostly present as HCO 3 in the basic stroma during photosynthesis [108]. As CO 2 is the ultimate sink for electrons transported through the photosystems, a reduction in total carbonic CO 2 in the stroma can cause the overreduction of PSII, leading to an increase in ROS generation and photoinhibition [109].
If the reduction in de-epoxidation observed in eto1-1 resulted from a reduced transthylakoid membrane pH gradient, this may be accompanied by a lower rate of CO 2 assimilation and a higher intercellular partial pressure of CO 2 (C i ) needed to support a maximum rate of CO 2 assimilation. At ambient CO 2 , eto1-1 has a lower rate of CO 2 assimilation than WT or ein2-5 ( Fig 5A). As neither the stomatal conductance nor the stomatal index, i.e., the density of stomata relative to the epidermal cell number, in eto1-1 was altered, the C i of eto1-1 was not reduced relative to WT (data not shown), indicating that CO 2 diffusion was not limiting. The intercellular partial pressure of CO 2 required to support the maximum rate of CO 2 assimilation (V max ) can be determined from an A-C i or CO 2 response curve in which the rate of CO 2 assimilation is measured as a function of C i . The rate of CO 2 assimilation increased in WT and eto1-1 as the intercellular partial pressure of CO 2 increased (Fig 5A). The rate of CO 2 assimilation in eto1-1 remained substantially lower than WT even as the C i reached 740 μbar, a level at which the maximum rate of CO 2 assimilation was attained in WT and ein2-5 ( Fig 5A). Only as C i was elevated to approximately 1170 μbar did the rate of CO 2 assimilation in eto1-1 begin to plateau ( Fig 5A). The difference in the intercellular partial pressure of CO 2 required to support a ½V max of CO 2 assimilation in eto1-1 and WT supports the notion of a lower solubility of CO 2 in the eto1-1 stroma due to a less basic stroma.
The relative difference in stromal pH in eto1-1 versus WT leaves can be calculated from the difference in the intercellular partial pressure of CO 2 required to support a ½V max rate of CO 2 assimilation. The concentration of CO 2 dissolved in the stroma, i.e., CO 2 (aq), is determined by CO 2 (aq) = CO 2 (i) x K H , where K H is the Henry's constant for CO 2 (at 25°C, K H = 3.4x10 -2 M/ atm). Thus, as C i at ½V max is 234 ppm in WT and 326 ppm in eto1-1, the concentration of dissolved CO 2 needed to reach the K m(CO2) for Rubisco is 7.91 x 10 -6 M in WT and 11.0 x 10 -6 M in the eto1-1 stroma. Once dissolved in water, CO 2 is rapidly converted into carbonic acid (H 2 CO 3 ) by carbonate anhydrase and dissociates into HCO 3 and CO 3 ] eto1-1 % CO 2 (aq) wt / CO 2 (aq) eto1-1 or pH wt -pH eto1-1 % lg(CO 2 (aq) eto1-1 )-lg(CO 2 (aq) wt ) = lg(11.0x10 -6 ) − lg(7.91x10 -6 ) = 0.142. Therefore, if the pH in the WT stroma is 7.8 during photosynthesis, it would be approximately 7.658 in the eto1-1 stroma following light exposure. The observation that the level of carbonic anhydrase activity was actually elevated in eto1-1 (and ctr1-3), particularly in chloroplasts,  (Table 1) demonstrated that a decrease in carbonic anhydrase activity in eto1-1 was not responsible for a reduction in stromal carbonic CO 2 . The increase in carbonic anhydrase activity in eto1-1, however, may represent an attempt to compensate for its lower stromal pH.
The lower stromal pH in eto1-1 suggests a reduced ability to establish and/or maintain a proton gradient across the thylakoid membrane. To confirm directly if the transthylakoid membrane pH gradient (ΔpH) in eto1-1 and ctr1-3 is smaller than in WT, we measured the ΔpH in intact chloroplasts by measuring the fluorescence of 9-aminoacridine (9-AA) which is quenched as a function of the light-mediated decrease in pH in the thylakoid lumen [110,111,112,113,114]. The relative difference in the ΔpH was significantly smaller in eto1-1 and ctr1-3 than in WT following light exposure ( Table 2). In contrast, the relative difference in the ΔpH in npq1 or npq4 was not significantly different from WT. Addition of 2,6-dichloro-p-benzoquinone (DCBQ), an exogenous PSII electron acceptor that enables maximum electron flow from PSII reaction centers, increased the ΔpH established upon exposure to light as expected but eto1-1 and ctr1-3 continued to exhibit a smaller relative increase in the ΔpH than WT ( Table 2). The inability to fully acidify the thylakoid lumen in eto1-1 would limit the lightmediated activation of its VDE. This conclusion is consistent with the approximate 2-fold reduction in de-epoxidation activity in eto1-1 within the initial 10 min of light exposure ( Fig  2B) and the observation that increasing C i disproportionately increases the level of photochemistry in eto1-1 relative to WT (Fig 5A). Although VDE expression in ctr1-3 was reduced approximately 2-fold (Fig 3), the 4-fold reduction in its de-epoxidation activity within the initial 10 min of light exposure (Fig 2B) and the substantially reduced transthylakoid membrane pH gradient during light exposure (Table 2) also indicates a reduced activation of VDE in this mutant. Although the reduced VDE activity in eto1-1 can account for the slower initial rate of induction of NPQ (S1B Fig), it does not account for the gradual increase in NPQ that eventually overtakes the WT steady-state level (S1 Fig). Reductions in the transthylakoid membrane pH gradient and stromal CO 2 levels could result in increased ROS and photoinhibitory processes that would be measured as the qI component of NPQ as indicated by the increase in NPQ s ( Fig  1G). If so, increasing the level of CO 2 dissolved in the eto1-1 stroma by increasing the C i should reduce qI which would be indicated by the absence of the gradual increase in NPQ that overtakes the WT steady-state level. To test this, we measured the induction of NPQ in eto1-1 leaves exposed to 1800 ppm CO 2 in order to saturate the eto1-1 stroma. The induction of NPQ in eto1-1 at ambient CO 2 exhibited the same increase in NPQ that overtakes the WT steady-state level (Fig 5B) as observed above (S1A Fig). At 1800 ppm CO 2 , however, the elevated induction of NPQ in eto1-1 was reversed (Fig 5B). These results suggest that increasing the level of soluble CO 2 in the eto1-1 stroma is sufficient to prevent the aberrant accumulation of NPQ following prolonged exposure to high light, consistent with the conclusion that the elevated increase of NPQ in eto1-1 represents photoinhibition.
Increasing VDE expression reverses the defect in NPQ in eto1-1 and ctr1-3 The above results suggest that elevated ethylene signaling represses VDE activation and, at very high levels, represses VDE expression as well. To investigate whether restoring VDE activity in eto1-1 or ctr1-3 would correct the impaired functioning of the xanthophyll cycle and the induction of NPQ, we used a transgenic approach to increase the expression of VDE in these mutants. For this, the Arabidopsis NPQ1 coding region was placed under the control of the 35S promoter, which is largely unaffected by ethylene (Fig 4A), and the construct introduced into wild-type Arabidopsis. We screened transformants containing the construct for expression using Western analysis and identified a VDE overexpressing line (WT T::NPQ1) exhibiting elevated qE [115]. We introduced the 35S::NPQ1 transgene into eto1-1 or ctr1-3 through crosses and isolated the homozygous eto1-1 T::NPQ1 and ctr1-3 T::NPQ1 lines. WT, eto1-1, and ctr1-3 containing the NPQ1 transgene exhibited a greater induction of NPQ (Fig 6A) which correlated with a higher level of in vitro VDE activity (Fig 6B).
To examine whether increasing VDE activity in eto1-1 and ctr1-3 corrected the defect in the functioning of the xanthophyll cycle, we measured the generation of A and Z in response to high light (i.e., 1000 PFD) in the mutants containing the 35S::NPQ1 transgene. As was observed in Fig 2, the generation of A and Z was substantially lower in eto1-1 and ctr1-3 than WT ( Fig  6C and 6D), despite the higher level of light used relative to that in Fig 2. Increasing VDE expression in eto1-1 and ctr1-3 increased the generation of A and Z (Fig 6C) and increased the rate of de-epoxidation to a level similar to WT plants containing the 35S::NPQ1 transgene ( Fig  6D). Increasing VDE activity did not alter zeaxanthin epoxidase activity or substantially alter the pool sizes of neoxanthin or lutein (data not shown).
To determine whether increasing VDE activity in eto1-1 or ctr1-3 corrected their aberrant NPQ induction profile, we measured the kinetics of NPQ induction. During exposure to saturating light, i.e., 1000 PFD, the initial rate of NPQ induction was lower in eto1-1 and ctr1-

generated.
As eto1-1 and ctr1-3 exhibited a lower transthylakoid membrane pH gradient during exposure to light, we examined whether restoring VDE activity would correct the impaired transthylakoid membrane pH gradient in eto1-1 and ctr1-3 by measuring the quenching of 9-AA in intact chloroplasts. The relative difference in the transthylakoid membrane ΔpH was significantly lower in eto1-1 and ctr1-3 than in WT following light exposure (Table 4) in good agreement with the measurements in Table 2. Restoration of VDE activity in eto1-1 and ctr1-3 increased the relative difference in the transthylakoid membrane ΔpH generated in response to light exposure. In contrast, overexpression of VDE in the WT background did not increase its relative difference in the transthylakoid membrane ΔpH. When exposed to a subsaturating level of light (i.e., 150 PFD for 3 min), the relative difference in the ΔpH in eto1-1 was not significantly different from WT but it was lower in ctr1-3, although to a smaller extent than following exposure to 800 PFD ( Table 4). Restoration of VDE activity in ctr1-3 restored the relative difference in the ΔpH to the WT level following exposure to low light whereas overexpression of VDE in WT had the opposite effect and reduced the relative difference in the ΔpH (Table 4). These results support the conclusion that eto1-1 and ctr1-3 are impaired in their ability to establish or maintain a full ΔpH in high light which can be partially corrected by increasing VDE activity.
The impairment in establishing a full transthylakoid membrane pH gradient in eto1-1 and ctr1-3 following exposure to high light could be a result of fewer electrons entering PSII reaction centers while the elevated production of O 2 .in these mutants may indicate an increase in the fraction of those electrons that do enter PSII reaction being used in ROS generation rather than for CO 2 fixation. To examine whether an increase in ethylene signaling alters electron flow, we measured the electron transport rate (ETR) as a function of increasing light intensity. As would be expected, the ETR increased in WT plants with increasing light (S3A Fig). Interestingly, ETR levels in eto1-1 and ctr1-3 were actually higher than in WT. In contrast, the level of ETR in npq1 and npq4 was not higher than the WT level. As qE is impaired in npq1 and npq4, this suggests that the increase in ETR in eto1-1 and ctr1-3 is not a result of their reduced qE alone. Restoring VDE activity in eto1-1 and ctr1-3, however, decreased the ETR in these mutants as it did in WT (S3B Fig),  reduce the ETR. These data show that, rather than a reduction in the rate of electron transport, eto1-1 and ctr1-3 exhibit an increase in electron transport relative to wild-type plants. We next measured the quantum requirement of eto1-1 and ctr1-3, i.e., the number of electrons required to fix one molecule of CO 2 , from the relationship between the quantum yield of PSII electron transport (φPSII) and the quantum yield of CO 2 assimilation (φCO 2 ) [117]. A quantum requirement of approximately eight electrons per CO 2 molecule fixed was observed in WT plants and overexpression of VDE did not alter this significantly (Table 5). In contrast, a quantum requirement of approximately 10.31 and 12.85 electrons per CO 2 molecule fixed was observed in eto1-1 and ctr1-3, respectively. The restoration of VDE activity in eto1-1 decreased this elevated quantum requirement significantly (P<0.05 relative to eto1-1) as it did in ctr1-3 (P<0.01 relative to ctr1-3) ( Table 5) although not to WT levels. These data suggest that the reduced VDE activity of eto1-1 and ctr1-3 results in increases in ETR and the quantum requirement and correlates with an increase in O 2 .production, supporting the notion that a greater number of electrons entering PSII are used to generate ROS. The data also show that restoring VDE activity in eto1-1 and ctr1-3 partially reverses their elevated ETR and quantum requirement.
As restoring VDE activity in eto1-1 and ctr1-3 partially reverses the elevated O 2 .production in these mutants, we examined whether it would also reduce the extent of photoinhibition when exposed to high light. Therefore, the induction of NPQ was measured in the mutants with or without the T::NPQ1 transgene under ambient and elevated CO 2 (i.e., 1800 ppm). The induction of NPQ in eto1-1 and ctr1-3 under ambient CO 2 was induced initially to a lower level than in WT (within 2-4 min of light exposure) but eventually overtook the WT steadystate level (Fig 7A) as observed above (Fig 5B). Exposure to 1800 ppm CO 2 , however, substantially reduced this gradual increase in NPQ in eto1-1 and ctr1-3 such that following its initial induction, the level NPQ continued to rise only slightly (Fig 7B). Increasing VDE expression in eto1-1 and ctr1-3 resulted in a NPQ induction profile under ambient and elevated CO 2 that was largely similar to WT in that, following its initial induction, NPQ partially relaxed and did not exhibit a subsequent increase above the level observed for WT overexpressing VDE ( Fig  7A and 7B, respectively). These results suggest that the qI generated in eto1-1 and ctr1-3 can be prevented by restoring VDE activity or by increasing CO 2 . As shown above, eto1-1 and ctr1-3 exhibit a smaller NPQ f and larger NPQ s , the fast and slow relaxation components of NPQ, respectively (Fig 1G). To examine whether restoring VDE activity in eto1-1 and ctr1-3 corrects their defects in NPQ f and NPQ s , we measured the fast and slow relaxation components of NPQ in eto1-1 and ctr1-3 overexpressing VDE. In dark-adapted plants exposed to 1800 PFD, NPQ f was substantially lower and NPQ s substantially higher in eto1-1 and ctr1-3 than in WT (Fig 8A), confirming the photosensitivity of eto1-1 and ctr1-3. A decrease in NPQ f and slight increase in NPQ s was observed in leaves of WT overexpressing VDE (Fig 8A) as previously reported [115]. In contrast, NPQ f was significantly increased and NPQ s significantly reduced in eto1-1 T::NPQ1 and ctr1-3 T::NPQ1 relative to eto1-1 and ctr1-3 (Fig 8A), consistent with the effect that overexpressing VDE had on reducing the qI component of NPQ in eto1-1 and ctr1-3 (Fig 7). As increasing VDE expression in WT plants did not have the same effect on NPQ f and NPQ s , these results suggest its effect was specific to these mutants.
To examine whether the restoration of VDE activity in eto1-1 and ctr1-3 protects against damage to the PSII reaction center, we measured the recovery of eto1-1 and eto1-1 T:NPQ1 plants following a 2 hr exposure to sunlight (i.e., 1900 PFD) by measuring the quantum yield of PSII (i.e., F v /F m ), upon their transfer to darkness. The data was expressed relative to the darkadapted F v /F m value (set to a value of 1) in order to make direct comparisons between lines. High light treatment of dark-adapted plants resulted in a substantial drop in their relative F v /F m from which they largely recovered during the subsequent 20 hr in dark (Fig 8B and 8C). eto1-1 and ctr1-3 exhibited a greater reduction in the relative F v /F m following high light treatment and a slower rate of recovery than did WT plants (Fig 8B and 8C). Overexpression of VDE did not improve the rate of recovery in WT plants but did improve the rate of recovery in eto1-1 and ctr1-3 (Fig 8B and 8C). The observation that increasing VDE activity in WT plants does not increase phototolerance under these conditions supports the conclusion that the ability of restoring VDE activity to increase phototolerance in eto1-1 and ctr1-3 is specific to these mutants. Recovery from photoinhibition requires new protein synthesis to repair damaged PSII reaction centers [118,119,120,121,122]. Inhibiting this repair with inhibitors of chloroplast protein synthesis results in a greater reduction in the quantum efficiency (i.e., F v /F m ) during exposure to light [123]. To examine whether the rate of photoinhibition in eto1-1 and ctr1-3 was greater than in WT in the absence of repair activity, we infiltrated adult leaves with either 1 mM chloramphenicol/0.1% ethanol, which inhibits chloroplast protein synthesis, or 0.1% ethanol only and exposed to 1000 PFD. We then measured the quantum yield of PSII (i.e., F v /F m ) during light exposure. In the absence of chloramphenicol, the quantum yield of eto1-1 and ctr1-3 decreased at a greater rate than in WT (Fig 8D). The reduction in the quantum yield of all lines was greater in the presence of chloramphenicol than it was in its absence but the decrease in eto1-1 and ctr1-3 was even greater than the WT rate (Fig 8D), suggesting that eto1-1 and ctr1-3 experience a greater level of photodamage particularly when the repair of PSII reaction centers is inhibited.  As α-tocopherol also protects against photodamage, we examined whether the greater photosensitivity of eto1-1 and ctr1-3 may be due to reduced level of α-tocopherol. We measured α-tocopherol content in plants grown at 250 PFD and exposed to sunlight up to 6 hr. The level of α-tocopherol was not significantly different in eto1-1 relative to the WT prior to and following sunlight exposure (Table 6). α-Tocopherol content in ctr1-3 was actually higher than in WT plants both before and after exposure to sunlight. Restoring VDE expression in eto1-1 and ctr1-3 did not alter the level of α-tocopherol in WT or mutants plants, indicating that the greater photosensitivity of eto1-1 and ctr1-3 is not a result of a reduction in α-tocopherol levels. α-Tocopherol content in npq1 was also higher than in WT plants ( Table 6) and, as previously reported, the higher levels of α-tocopherol and ascorbic acid in npq1 may have protected against ROS-mediated lipid peroxidation during exposure to high light [124]. As eto1-1 exhibited no increase in α-tocopherol or ascorbic acid content (Table 6 and Fig 3), its reduced VDE activity occurs in the absence of any increase in these compensating antioxidants. Moreover, although ctr1-3 exhibited an increase in α-tocopherol content, it showed no increase in ascorbic acid content (Fig 3).

Restoring VDE activity reverses the small growth phenotype imposed by a moderate increase in ethylene signaling
Exposure to high light induces ROS synthesis and can result in growth retardation [125]. Therefore, the higher rate of O 2 .production and increased photoinhibition experienced by in eto1-1 and ctr1-3 may negatively affect their growth [49,107]. ctr1-3 is considerably smaller than eto1-1, correlating with its higher level of ethylene signaling, and their small stature has been attributed to a smaller cell size [14,80,81]. If the greater photosensitivity of eto1-1 and ctr1-3 contributes to their small growth phenotype when grown under high light, restoring VDE activity should not only reduce ROS generation and photoinhibition but may also reverse the small growth phenotype of eto1-1 and ctr1-3 to some extent. To test this, we grew WT, eto1-1, and ctr1-3 plants with or without the 35S::NPQ1 transgene under high light and compared their growth characteristics. eto1-1 T:NPQ1 plants were substantially larger than eto1-1 (i.e., 1900 PFD) was determined by measuring F v /F m over time following the transfer of plants to darkness to facilitate recovery. The data was expressed relative to dark-adapted F v /F m value in order to make direct comparisons between lines. WT (filled diamonds); WT T:NPQ1 (open diamonds); eto1-1 and ctr1-3 (open squares); eto1-1 T:NPQ1 and ctr1-3 T:NPQ1 (filled squares). The data reported are the average of six replicates.
doi:10.1371/journal.pone.0144209.g008 Table 6. Restoring VDE expression in eto1-1 and ctr1-3 does not alter α-tocopherol content. plants when grown under high light (Fig 9A) which resulted from an increase in leaf size ( Fig  9B) and cell size (S4A Fig). Increasing VDE expression in WT plants increased plant size to a much smaller extent than it did in eto1-1 plants (Fig 9A and 9B, S4A Fig). Quantitative measurements revealed that the fresh and dry weights of eto1-1 T:NPQ1 were 4.6-fold and 4.2-fold greater, respectively, than eto1-1 plants (Fig 9C). Consequently, eto1-1 T:NPQ1 plants were similar in stature and biomass to WT plants. In contrast, increasing VDE expression in WT plants increased their fresh and dry weights by just 30% and 42%, respectively (Fig 9C). Comparison of all leaves from eto1-1 T:NPQ1 and eto1-1 plants revealed that increasing VDE expression in eto1-1 did not alter cotyledon size but it did increase the size of all true leaves (Fig 9G). Increasing VDE expression in ctr1-3 also increased plant stature when grown at 250 PFD but to a smaller extent than observed in eto1-1 plants (  If the restoration of VDE activity reversed the small growth phenotype of eto1-1 because of a reduction in photodamage during growth under high light, restoring VDE activity in eto1-1 would be predicted to have a substantially smaller effect on growth under low light which limits ROS production and photodamage (Table 3). To test this hypothesis, we grew the same lines under low light (i.e., 50 PFD). eto1-1 T:NPQ1 plants were similar in size to eto1-1 plants ( Fig  9D) as was the size of leaves under these growth conditions (Fig 9E). The fresh and dry weights of eto1-1 T:NPQ1 were not significantly different from eto1-1 under these growth conditions (Fig 9F). The presence of the transgene also had little effect on the growth of WT plants ( Fig  9D and 9E) or the fresh or dry weights of WT plants (Fig 9F). The fresh weight of eto1-1 plants, however, was 66% of WT when grown under low light versus 25% of WT when grown under high light (compare Fig 9F to 9C) indicating that high light has a disproportionately negative effect on eto1-1 growth. Similarly under low light, ctr1-3 T:NPQ1 fresh weight (0.010 g) was not significantly different from that of ctr1-3 (0.010 g). Thus, the reversal of the small stature of eto1-1 and ctr1-3 following restoration of VDE activity was observed specifically under high light conditions. These results suggest that increases in ethylene responses cause an increase in ROS production during exposure to high light that contributes to a reduction in stature. The data also suggest that restoring VDE activity lowers ROS production and reduces their deleterious effect on plant stature.

Increasing VDE expression does not alter ethylene production or ethylene responses
The effect that increasing VDE expression had on reversing the small cell size and plant stature of eto1-1 raised the possibility that the increase in VDE expression may reduce ethylene production. To examine this, we measured the production of ethylene in eto1-1 and eto1-1 T: NPQ1 plants. Ethylene production was nearly 4-fold higher in eto1-1 plants than in WT but ethylene production in eto1-1 T:NPQ1 plants was not significantly different from that in eto1-1 (S6A Fig). Increasing VDE expression in WT plants also did not affect ethylene evolution (S6A Fig). We next investigated the effect of VDE on ethylene responsiveness by examining the triple response of seedlings to ethylene when grown on medium with or without ACC, the precursor to ethylene. The triple response of Arabidopsis is an ethylene-mediated response of darkgrown seedlings characterized by the radial expansion of the hypocotyl, inhibition of root and hypocotyl elongation, and the presence of an exaggerated apical hook [126]. In the absence of ACC, seedling growth is influenced by the endogenous production of ethylene only. The hypocotyls and roots of eto1-1 were substantially shorter than WT and exhibited a more pronounced apical hook than WT (Table 7 and   ethylene (S6A Fig). Growth of eto1-1 T:NPQ1 seedlings was not significantly different from eto1-1 seedlings and the pronounced apical hook remained (Table 7 and S6B Fig). Increasing VDE expression in WT plants also did not alter the triple response of WT seedlings (Table 7 and S6B Fig). However, because VDE is light-activated, growth of light-grown seedlings was also examined. Although light-grown seedlings differ in their response to ethylene from those grown in the dark, the light-grown seedling response is characterized by a reduced cotyledon size, a delay in true leaf emergence, and an inhibition of root growth [127]. To examine the effect that VDE has on ethylene responsiveness in light-grown seedlings where VDE would be active, we germinated the same lines on medium with or without 20 μM ACC and grown in low light (i.e., 100 PFD) to avoid photoinhibition. In the absence of ACC, eto1-1 T:NPQ1 seedlings exhibited a similar reduction in cotyledon size and root length as eto1-1 (S6C and S6D

Discussion
In this study, we show that eto1-1, in which ethylene production is increased moderately, and ctr1-3, in which ethylene responses are constitutive exhibit an impaired qE that is due, in part, to a reduction in VDE activity. This conclusion is supported by the observation that eto1-1 and ctr1-3 exhibit significantly lower rates of light-induced de-epoxidation that could be corrected by restoring VDE activity. The reduced de-epoxidation activity in eto1-1 was explained by a reduction in VDE activation while the larger reduction in de-epoxidation in ctr1-3 was explained by a 2-fold reduction in the levels of NPQ1 mRNA, VDE protein, VDE enzyme activity, as well as a reduction in the activation of VDE activity. NPQ1 promoter activity was repressed following an increase in ethylene responses, imposed either pharmacologically or genetically (Fig 4). The repression of NPQ1 promoter activity correlated with the level of ethylene responses in that its activity was repressed to greater extent in ctr1-3 than in eto1-1. This conclusion is consistent with the level of ethylene required to repress NPQ1 expression in WT plants in that NPQ1 promoter activity was repressed approximately 2-fold following a 10.6-fold increase in ethylene production by exogenous application of ACC compared to the more modest reduction in promoter activity observed in eto1-1 which exhibits less than a 4-fold increase in ethylene production under moderate light. Although the reduction in NPQ1 promoter activity in ctr1-3 partially accounted for the reduction in NPQ1 transcript and VDE protein levels, the small reduction in NPQ1 promoter activity in eto1-1 was not reflected in reduced NPQ1 transcript or VDE protein levels (Fig 3). Nevertheless, the substantially lower level of de-epoxidation activity present in eto1-1 indicated a reduction in VDE enzyme activity, which requires an acidified thylakoid lumen generated by light-induced proton pumping from the stroma for its activation.
Direct measurements revealed that eto1-1 and ctr1-3 are unable to establish a full transthylakoid membrane pH gradient during exposure to saturating light and that restoring VDE activity substantially reversed this defect ( Table 4). The lower transthylakoid membrane ΔpH in eto1-1 and ctr1-3 was not a result of a reduction in electron flow as their ETR was actually higher than WT, suggesting a defect in establishing or maintaining a transthylakoid membrane pH gradient. Under nonsaturating light, however, eto1-1 was able to establish a transthylakoid membrane ΔpH equivalent to WT while the transthylakoid membrane ΔpH in ctr1-3 was closer to the WT value under nonsaturating light than it was under saturating light. This suggests that the defect in establishing a full transthylakoid membrane pH gradient in eto1-1 occurs specifically during exposure to elevated light levels and is exacerbated in ctr1-3 under these same conditions.
These observations also indicate that despite the higher rate of electron flow through PSII in these mutants, an increase in ethylene responses results in a lower transthylakoid membrane ΔpH, a reduction in VDE activation, and a greater proportion of electrons being used for ROS generation, particularly under saturating light. This possibility was supported by the substantially higher rate of O 2 .generation in eto1-1 and ctr1-3 and the accompanying higher level of photoinhibition experienced by these mutants under saturating light but not under low light relative to WT (Table 3). npq1, which lacks VDE expression, as well as npq4, which lacks PsbS expression, also exhibit higher rates of O 2 .production when exposed to saturating light (Table 3), supporting the conclusion that the increase in O 2 .production in the ethylene mutants results, in part, from a reduction in xanthophyll cycle functioning. That the level of O 2 .production in eto1-1 and ctr1-3 is higher than the levels in npq1 or npq4, however, suggests that the reduced functioning of the xanthophyll cycle in eto1-1 and ctr1-3 only partially accounts for the increase in ROS production during exposure to saturating light. Nevertheless, the partial reversal of the aberrantly high level of O 2 .production in eto1-1 (Table 3) and the generation of a normal transthylakoid membrane pH gradient ( Table 4) following restoration of VDE activity specifically during exposure to high light supports the notion that the ethylene-mediated reduction in VDE activity does contribute to these defects. The generation of a transthylakoid membrane ΔpH is necessary for the activation of VDE enzyme activity and the generation of qE [51,52,105,128]. The light-mediated acidification of the thylakoid lumen causes protonation of specific amino acid residues in VDE that alters its conformation and causes the enzyme to dock with the thylakoid membrane where it comes in contact with V and de-epoxidates it to A and Z [51,52]. A low thylakoid lumen pH may also promote a conformational change in the membrane [129]. A more basic thylakoid lumen and more acidic stroma would have the potential to affect several chloroplast functions critical to NPQ, including lower activation of VDE, reduced de-epoxidation, and impaired qE.
As protonation of PsbS under conditions of excess light is necessary for its function [130], its activity, like VDE, requires an acidified thylakoid lumen. Defects in establishing or maintaining a protein gradient across the thylakoid membrane would be expected to reduce PsbS function, as well as that of VDE, thus leading to an impaired qE. Although the more acidic stroma of eto1-1 prevents full activation of VDE activity as measured by its decreased rate of de-epoxidation, it is unknown whether it is sufficient to affect PsbS activity. PsbS expression, however, does determine qE capacity [107] and a reduction in PsbS levels are accompanied by changes in the structural organization of PSII-LHCII arrays in the thylakoid membrane [131,132,133]. Therefore, the observed reduction in PsbS expression and perhaps activation may contribute to the lower qE observed in eto1-1 and ctr1-3.
Acidification of the thylakoid lumen during light exposure also results in protonation of certain light-harvesting antenna complexes (LHCs) as well as causing conformational changes in the LHCs that facilitate qE [134,135,136]. Zeaxanthin is important in promoting the dissociation of LHCII from PSII and its aggregation [136]. Zeaxanthin has also been proposed to bind LHCII proteins to either quench excited chlorophyll directly or to function as an allosteric modulator of the ΔpH-sensitive qE inherent in LHCII proteins [56,66,137]. The link between pH-induced LHC II aggregation and qE can be observed in vitro with isolated LHCs [134].
A more acidic stroma would also be expected to reduce the level of soluble CO 2 , which in turn would be expected to reduce the rate of CO 2 assimilation. The observation that the observed lower rate of CO 2 assimilation in eto1-1 can be corrected by increasing the internal CO 2 concentration (Fig 5A) supports the notion that a more acidic stroma limits CO 2 solubility. Thus, a more acidic stroma in eto1-1 (and ctr1-3) could affect multiple processes that contribute to the observed defects in VDE activation, xanthophyll de-epoxidation, qE, CO 2 assimilation, as well as the increase in ROS generation. The observation that restoring VDE activity to eto1-1 restores near WT levels of qE is again consistent with the conclusion that the ethylene-mediated reduction in VDE activation contributes to the reduction in qE.
Although the defects in the expression and activation of VDE in eto1-1 and ctr1-3 were consistent with the observed reduced initial induction of NPQ, the level of NPQ eventually overtook the steady-state WT level following prolonged exposure to light (S1 Fig). The initial induction of NPQ typically represents qE whereas qI largely contributes to the additional accumulation in NPQ following its initial rapid rise. The increase in NPQ in eto1-1 and ctr1-3 that eventually exceeds the WT steady-state level, therefore, can be understood as an increase in photoinhibitory processes during prolonged exposure to higher light levels. An increase in qI was consistent with a lower relative F v /F m (Fig 8B and 8C), a lower NPQ f and higher NPQ s (Figs 1G and 8A), and a slower rate of recovery following high light stress (Fig 8B and 8C). That an increase in ethylene responses was responsible for the accumulation in qI was supported by the observation that inhibiting ethylene perception in eto1-1 largely prevented qI accumulation (S1D Fig). Moreover, preventing repair of damaged PSII reaction centers through the inhibition of new protein synthesis revealed that repair plays a disproportionately larger role in eto1-1 and ctr1-3 ( Fig 8D).
A reduction in photoprotection caused by an impaired xanthophyll cycle would be expected to be exacerbated by a reduction in CO 2 solubility in the stroma as both would contribute to an increase in ROS and photoinhibition. Accordingly, the increase in qI in eto1-1 and ctr1-3 could be prevented by either increasing the concentration of CO 2 (Figs 5B and 7B) or by restoring VDE activity (Figs 7A and 8). The notion that an increase in ethylene responses reduces the level of soluble CO 2 in the stroma is also supported by photosynthetic measurements in which the reduced rate of CO 2 assimilation in eto1-1 could be corrected by increasing the level of soluble CO 2 (Fig 5A). A similar stomatal conductance in eto1-1 and WT excluded the possibility of reduced gas diffusion. The observation that eto1-1 required a higher C i in order to achieve a ½V max rate of CO 2 assimilation equivalent to WT was consistent with a more acidic stroma as suggested by the transthylakoid pH gradient measurements in this mutant.
These observations indicate that the level of soluble CO 2 present in the stroma of eto1-1 and ctr1-3 is lower than in WT. As CO 2 serves as the final electron sink in photosynthesis, a reduction in the level of soluble CO 2 may result in the over reduction of the photosystems and an accumulation of photoinhibition during prolonged exposure to light, thus contributing to the increase in NPQ in eto1-1 and ctr1-3 that eventually overtakes the steady-state WT level. Such a scenario can also occur during drought conditions where water stress-induced stomatal closure limits the diffusion of CO 2 into chloroplasts. This leads to decreased CO 2 assimilation, increased photorespiration, and elevated H 2 O 2 generation. Increased production of singlet oxygen and O 2 .-, the latter of which can be generated by the water-water-cycle associated with PSI or by oxygen reacting with reduced quinones on the acceptor-side of PSII [44,138,139,140], also occurs under conditions of limiting CO 2 [43,141]. The elevated level of O 2 .in eto1-1 and ctr1-3 specifically during exposure to high light correlates with the accumulation of photoinhibition in these mutants whereas the restoration of VDE activity reduced the level in photoinhibition (Fig 8) as it did the aberrantly high level of O 2 .-( Table 3).
The reduced cell size and plant stature of eto1-1 and ctr1-3 are likely a consequence of a combination of regulation by ethylene directly as well as indirectly through a reduction in qE, an increase in O 2 .generation and qI, and a reduction in net photosynthetic gain. The extent to which photoinhibitory processes and ROS are responsible for the reduction in cell size and plant stature should be observed only under those conditions resulting in their elevated production, e.g., following exposure to excess light. The observation that restoring VDE expression in eto1-1 reversed its small cell size and plant stature specifically under moderate to high light, but not low light when ROS levels are low (Fig 9), supports the conclusion that increased ethylene signaling results in photodamage from elevated ROS to which an impaired xanthophyll cycle likely contributes. Although eto1-1 plants are also smaller than WT when grown under non-photoinhibitory conditions, they are considerably closer in size to WT during growth under low light (66% of WT fresh weight) compared to growth under high light (25% of WT fresh weight) (Fig 9), demonstrating that the degree to which increased ethylene signaling reduces plant stature is exacerbated by growth under high light. These observations suggest that ethylene regulates cell size in part through impairment of the xanthophyll cycle under high light conditions in addition to its role in regulating cell size under lower light conditions. The higher level of ethylene signaling in ctr1-3 results in an even smaller cell size and reduced plant stature than in eto1-1. VDE and PsbS expression are repressed in ctr1-3, resulting in a reduced rate of de-epoxidation and qE. Restoring VDE expression increased cell size when the mutant was grown under moderate light but this was offset by a decrease in the number of cells per leaf such that no change in plant stature was observed. As ctr1-3 could not be grown to adulthood under high light, it wasn't possible to examine the effect of increasing VDE expression on ctr1-3 growth under the same conditions used for eto1-1. It is possible, however, that the direct regulation of cell size by the high level of ethylene signaling in ctr1-3 makes a greater contribution to its small stature than does its impaired xanthophyll cycle.
The reduction in VDE activity in eto1-1 and ctr1-3 is not solely responsible for the small growth phenotype of these mutants as npq1 (or npq4) is not substantially smaller than WT [124]. However, the fact that restoring VDE activity in eto1-1 can fully reverse its small stature and partially reverse the small stature of ctr1-3 suggests that a reduction in VDE activity in the context of increased ethylene responses does contribute to a reduction in growth specifically under conditions of high light. The effect of restoring VDE activity on growth of in eto1-1 and ctr1-3 was specific to growth under high light as demonstrated by its failure to reverse the small growth phenotype of these mutants under low light. This finding also supports the conclusion that ROS contributes to the small growth phenotype of eto1-1 and ctr1-3 under high light as ROS production increases substantially in both mutants to a greater extent than in WT but is reversed by restoring VDE activity.
The negative effect that ROS has on growth was also observed in Arabidopsis silenced for the zinc finger transcription factor ZAT10, which was characterized by elevated levels of H 2 O 2 and O 2 .and reduced photochemistry when exposed to high light, resulting in substantial growth retardation [142]. Similarly, loss of ETHYLENE RESPONSE FACTOR 6 expression resulted in elevated levels of H 2 O 2 , photosensitivity, and reduced growth [143,144]. The importance of VDE activity in photoprotection is most clearly demonstrated with the npq1 mutant which lacks VDE expression altogether. npq1 exhibits substantially greater photoinhibition following transfer to high light than WT as measured by a lower F v /F m , reduced quantum yield of electron transport (φPSII), lower rate of CO 2 assimilation (despite a similar rate of transpiration), and some reduction in growth [49]. These phenotypes were even more pronounced when combined with the npq4 mutation [49,107]. npq1 plants also exhibited photooxidative damage, including bleaching and areas of necrosis following sudden exposure to high light [145]. npq1 (and npq4) plants exhibited reduced fitness when grown under field conditions involving exposure to full sun or in rapidly fluctuating moderate light [146], demonstrating that changes to VDE (or PsbS) expression affect growth under variable light environments. When grown in high light, however, npq1 exhibited growth similar to WT suggesting acclimation that may have been supported by the higher levels of α-tocopherol in young leaves of this mutant that protected against ROS-mediated damage [73,124,147]. npq4 also exhibits an increase in the level of α-tocopherol as well as an increase in ascorbic acid content [124]. The significant increase in α-tocopherol content in npq1 (and to a lesser extent npq4) was confirmed in our study (Table 6). In contrast, eto1-1 exhibited no such increase in α-tocopherol or ascorbic acid so that the reduction in its VDE activity and PsbS expression occurred in the absence of any additional compensating antioxidants. Moreover, zeaxanthin and α-tocopherol exhibit synergistic protection against photodamage [15,148,149,150] so a decrease in one may be amplified in the absence of an increase in the other. ctr1-3 exhibited an increase in α-tocopherol content (but not in ascorbic acid content) similar to npq1, raising the possibility that the increase in α-tocopherol may have partially compensated for the greater reduction in VDE activity and PsbS expression relative to eto1-1 despite its much higher level of ethylene signaling.
Although the reduction in VDE activity in eto1-1 and ctr1-3 is responsible for their reduced de-epoxidation and likely contributes to their increased ROS levels, it is not solely responsible for their increased oxidative load as restoration of VDE activity in eto1-1 and ctr1-3 only partially reversed the increase in ROS. Although the level of ROS is elevated in npq1, it does not reach the levels observed in eto1-1 and ctr1-3 (Table 3). Additionally, restoration of VDE activity in ctr1-3 only partially reversed its small cell phenotype. Such observations indicate that the reduction in VDE activity in the context of increased ethylene signaling does increase photosensitivity, ROS generation, and growth impairment as demonstrated by the reversal of these effects following the restoration of VDE activity. These observations also show the effects of restoring VDE activity was specific to eto1-1 and ctr1-3 as increasing VDE expression in WT plants did not have the same effect. However, because restoring VDE expression failed to fully reverse ROS generation and photosensitivity in eto1-1 and ctr1-3, ethylene likely affects other factors that also contribute to the oxidative load. We can conclude, however, that ethylene does negatively regulate the expression and activity of VDE and this repression impairs the function of the xanthophyll cycle while contributing to the increased photosensitivity and reduced plant growth specifically under high light conditions. An important area for future work will be to determine whether ethylene regulates NPQ and VDE expression through similar mechanisms in other species to establish whether the observations made in Arabidopsis are common throughout plant species.