Zeaxanthin Protects Plant Photosynthesis by Modulating Chlorophyll Triplet Yield in Specific Light-harvesting Antenna Subunits*

Background: The plant carotenoid zeaxanthin is accumulated under excess light. Results: Zeaxanthin induces a red shift in the carotenoid triplet excited state spectrum and reveals a higher efficiency in controlling chlorophyll triplet formation. Conclusion: Binding of zeaxanthin to specific proteins modulates the yield of dangerous chlorophyll excited states and protects photosynthesis from over-excitation. Significance: Functional dissection of zeaxanthin-dependent photoprotective mechanisms is crucial for understanding how plants avoid photoinhibition. Plants are particularly prone to photo-oxidative damage caused by excess light. Photoprotection is essential for photosynthesis to proceed in oxygenic environments either by scavenging harmful reactive intermediates or preventing their accumulation to avoid photoinhibition. Carotenoids play a key role in protecting photosynthesis from the toxic effect of over-excitation; under excess light conditions, plants accumulate a specific carotenoid, zeaxanthin, that was shown to increase photoprotection. In this work we genetically dissected different components of zeaxanthin-dependent photoprotection. By using time-resolved differential spectroscopy in vivo, we identified a zeaxanthin-dependent optical signal characterized by a red shift in the carotenoid peak of the triplet-minus-singlet spectrum of leaves and pigment-binding proteins. By fractionating thylakoids into their component pigment binding complexes, the signal was found to originate from the monomeric Lhcb4–6 antenna components of Photosystem II and the Lhca1–4 subunits of Photosystem I. By analyzing mutants based on their sensitivity to excess light, the red-shifted triplet-minus-singlet signal was tightly correlated with photoprotection in the chloroplasts, suggesting the signal implies an increased efficiency of zeaxanthin in controlling chlorophyll triplet formation. Fluorescence-detected magnetic resonance analysis showed a decrease in the amplitude of signals assigned to chlorophyll triplets belonging to the monomeric antenna complexes of Photosystem II upon zeaxanthin binding; however, the amplitude of carotenoid triplet signal does not increase correspondingly. Results show that the high light-induced binding of zeaxanthin to specific proteins plays a major role in enhancing photoprotection by modulating the yield of potentially dangerous chlorophyll-excited states in vivo and preventing the production of singlet oxygen.

Plants are particularly prone to photo-oxidative damage caused by excess light. Photoprotection is essential for photosynthesis to proceed in oxygenic environments either by scavenging harmful reactive intermediates or preventing their accumulation to avoid photoinhibition. Carotenoids play a key role in protecting photosynthesis from the toxic effect of over-excitation; under excess light conditions, plants accumulate a specific carotenoid, zeaxanthin, that was shown to increase photoprotection. In this work we genetically dissected different components of zeaxanthin-dependent photoprotection. By using time-resolved differential spectroscopy in vivo, we identified a zeaxanthin-dependent optical signal characterized by a red shift in the carotenoid peak of the triplet-minus-singlet spectrum of leaves and pigment-binding proteins. By fractionating thylakoids into their component pigment binding complexes, the signal was found to originate from the monomeric Lhcb4 -6 antenna components of Photosystem II and the Lhca1-4 subunits of Photosystem I. By analyzing mutants based on their sensitivity to excess light, the red-shifted triplet-minussinglet signal was tightly correlated with photoprotection in the chloroplasts, suggesting the signal implies an increased efficiency of zeaxanthin in controlling chlorophyll triplet formation. Fluorescence-detected magnetic resonance analysis showed a decrease in the amplitude of signals assigned to chlorophyll triplets belonging to the monomeric antenna complexes of Photosystem II upon zeaxanthin binding; however, the amplitude of carotenoid triplet signal does not increase correspondingly. Results show that the high light-induced binding of zeaxanthin to specific proteins plays a major role in enhancing photoprotection by modulating the yield of potentially dangerous chlorophyll-excited states in vivo and preventing the production of singlet oxygen.
Plants are particularly prone to photo-oxidative damage for the same reasons that they are effective at photosynthesis, namely because the primary pigment chlorophyll (Chl) 2 is a very efficient sensitizer. The singlet excited states of Chl molecules ( 1 Chl*) are efficiently quenched by photochemical reaction centers. Nevertheless, environmental conditions easily unbalance the ratio between energy capture and utilization; e.g. at high photon flux densities, accumulation of excitons in the light-harvesting complexes (Lhc) of both photosystems (PS) increases the amount of 1 Chl*. This raises the probability of intersystem crossing to the Chl triplet state ( 3 Chl*), a species that reacts with molecular oxygen (O 2 ) to yield singlet oxygen ( 1 O 2 ) molecules (1). Because of the high reactivity and low diffusion radius of 1 O 2 , this reactive oxygen species (ROS) induces damage in its local environment by (2), destroying lipids and nucleic acids and proteins (3)(4)(5), thus leading to a loop of ever-increasing 1 O 2 production and further oxidation until photobleaching (6). This photooxidative damage leads to a dramatic depression of photosynthetic efficiency called photoinhibition (7)(8)(9).
A second pigment class essential for photosynthesis is represented by carotenoids (Car), whose photoprotective action in the photosynthetic apparatus is well established (10). Carotenoids are either carotenes, bound to PSI and PSII core complexes, or their oxygenated derivatives, xanthophylls, mainly bound to Lhc proteins. Xanthophylls are involved in a number of photoprotection mechanisms, being active in (i) preventing over-excitation of reaction centers by quenching 1 Chl* states (11), (ii) quenching 3 Chl* through carotenoid triplet ( 3 Car*) formation, thus avoiding ROS formation (12), and (iii) scavenging ROS (13). Among xanthophylls, zeaxanthin (Zea) is of particular interest because it is absent in dark or low light conditions and only accumulates in excess light (EL), where it is produced from the diepoxide xanthophyll violaxanthin (Vio) (14) through the action of the violaxanthin de-epoxidase enzyme (15,16). Zea is known to be involved in several types of photoprotection events of the PSII reaction center, which occur on varying timescales. A number of these mechanisms quench 1 Chl*, namely (i) feedback de-excitation quenching (qE) (17,18), which occurs on the timescale of seconds to minutes, (ii) a slowly inducible quenching or qZ (19), which is likely caused by Zea binding to Lhc upon exchange with Vio (20), and (iii) a long term, irreversible quenching that reflects a photoinhibitory state of PSII (21). An increase of thermal dissipation of 1 Chl* can effectively protect reaction centers from over-excitation, thus reducing the probability of intersystem crossing to 3 Chl* and 1 O 2 formation in the Lhcs.
Previous results (22) demonstrated that the npq1 mutant of Arabidopsis thaliana, which is defective in the light-dependent Vio to Zea interconversion, shows increased photoinhibition and lipid peroxidation with respect to wild type (WT) in EL, leading to a decreased fitness (23). Early suggestions pointed to the decreased capacity for quenching of singlet excited chlorophylls (qE) (11). Further work by comparing the npq1 to the npq4 mutant, the latter lacking qE but retaining the ability for Zea synthesis, showed that protection of thylakoid membrane lipids against photooxidation was provided by a mechanism different than 1 Chl* quenching (22). Moreover, besides its role in singlet energy dissipation, Zea has been proposed to act as an antioxidant by scavenging 1 O 2 and thus preventing lipid peroxidation (24) upon its release from the pH-dependent V1 binding site of the major LHCII complex (25,26) into the lipid phase. However, a fundamental understanding of the Zea-dependent photoprotection mechanism, its location and whether its significance in providing overall photoprotection in vivo is mainly dependent on either 1 Chl*/ 3 Chl* quenching or ROS scavenging remains and awaits a more detailed analysis of the contributions of the different mechanisms involved in photoresistance.
Once synthesized, Zea has a dual location; it can be either free into thylakoid membrane lipids or bound to the Lhc proteins (27,28). The effect of the former pool was investigated in Arabidopsis showing that Zea has a distinct capacity for scavenging ROS with respect to other xanthophyll species (29). Fur-thermore, it has been reported (30,31) that the photoprotective effect of xanthophylls is greatly enhanced by their binding to Lhc proteins. Therefore, besides ROS scavenging in the lipid phase (32) and enhancing of qE (11), a third photoprotective effect is provided by a mechanism specifically exerted by the Lhc-bound Zea pool.
Here, we show that Zea bound to specific Lhcs is directly involved in the modulation of 3 Chl*. By means of time-resolved spectroscopy in vivo, we found that the photoresistance of Arabidopsis to EL treatment tightly correlates with the detection of a red shift in the major carotenoid signal in triplet-minus-singlet (TmS) spectra upon Zea synthesis, as measured in intact leaves. Fractionation of thylakoid membranes from EL-treated plants allowed detection of the red-shifted TmS signal in the monomeric Lhcb subunits of PSII (Lhcb4 -6) and the dimeric Lhca subunits of PSI but not in the major trimeric LHCII complex. In each case the Zea-dependent red shift correlates with a reduced yield of 1 O 2 production from purified Lhc proteins or from photosynthetic supercomplexes containing these subunits. Analysis by fluorescence-detected magnetic resonance (FDMR) at 4 K showed decreased amplitude of 3 Chl* upon Zea binding, implying a Zea-specific effect in decreasing the yield of dangerous excited states. We conclude that, in addition to the previously described effects in quenching 1 Chl* and scavenging 1 O 2 , an additional photoprotection mechanism is elicited by Zea binding to specific Lhc protein subunits consisting of a reduction in the yield in harmful 3 Chl*.

Thylakoid Isolation and Sample Preparation
Stacked thylakoid membranes were isolated from either dark-adapted or excess light (EL)-treated leaves (33). Grana membranes have been isolated from dark-and EL-treated samples using ␣-dodecylmaltoside (␣-DM) solubilization of stacked thylakoids, as described in Ref. 34. Membranes corresponding to 500 g of chlorophylls were washed with 5 mM EDTA, solubilized with 0.6% ␣-DM, and then fractionated by ultracentrifugation in a 0.1-1 M sucrose gradient as previously described (25). Purified monomeric Lhcb proteins (band 2 in the sucrose gradient) were further fractionated by flat-bed isoelectric focusing at 4°C (35). Purified LHCI complexes were obtained as described in Ref. 36.

Pigment Analysis
Pigments were extracted either from whole leaves, thylakoid membranes, or isolated antenna complexes with 80% acetone then separated and quantified by HPLC (37).

Determination of the Sensitivity to Photooxidative Stress
Short term EL treatment was performed for 1 h at 1200 mol photons m Ϫ2 s Ϫ1 at RT (22°C) to obtain the maximum Zea accumulation on detached leaves floating on water. Light was provided by 150 watt halogen lamps (Focus3, Prisma, Italy). Samples for HPLC analysis were rapidly frozen in liquid nitrogen before pigment extraction. EL-treated leaves used for thylakoid isolation were vacuum-infiltrated with 50 M Norfluorazon (a zeaxanthin-epoxidase inhibitor) upon EL treatment to slow down Zea-Vio conversion (38) during thylakoid isolation procedure. Photo-destruction of pigment-protein complexes in vivo was induced by a strong light treatment; leaf discs floating on water were exposed to high light (2500 mol of photon m Ϫ2 s Ϫ1 ) at RT for 31 h. For the quantification of Lhcb subunits, frozen leaf discs were homogenized in liquid nitrogen, and protein was extracted in 62.5 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, 5% ␤-mercaptoethanol. For each sample, the same volume of leaf extract (corresponding to 0.5 g of chlorophylls of t 0 sample) was loaded on SDS-PAGE. Immunoblot assays with antibodies against different polypeptides were performed as described previously (39). To avoid any deviation between different immunoblots, samples were compared only when loaded in the same slab gel.
Time-resolved Spectroscopy-absorbance changes were monitored in purified proteins or intact Arabidopsis leaves with a home-built pump and probe laser spectrophotometer basically described in Ref. 40) or with a laser flash photolysis spectrometer for isolated pigments. Transient absorption spectra were recorded in the laboratory of H. A. Frank at room temperature using an Edinburgh Instruments Model LP920KS flash photolysis spectrometer consisting of a Xe 920 450-watt arc lamp as a probe light source. The excitation pulse was generated by a Continuum Nd:YAG-pumped optical parametric oscillator laser tuned to excite selectively the Chl a Q y band at 662 nm. The pump laser and probe light beams were configured perpendicular to each other. A Tektronix digital oscilloscope model TDS 3012B was used for signal averaging. All transient absorption profiles were the average of 20 scans. The solutions contained micromolar concentrations of Chl a (as a triplet donor), and the xanthophylls (as triplet acceptors) and were degassed using at least five freeze-pump-thaw cycles before the spectroscopic measurements.
Quantification of Singlet Oxygen Yield-Measurements of 1 O 2 production on leaves and purified pigment-protein complexes were performed using Singlet Oxygen Sensor Green (SOSG, Invitrogen). SOSG is a fluorescent probe highly selective for 1 O 2 that increase its 530-nm emission band in presence of this ROS; it was shown to be a useful and reliable probe for the detection of 1 O 2 in vivo and in purified Lhcs (30,(41)(42)(43). Leaves were vacuum-infiltrated with the dye solution (SOSG 200 M) and illuminated with red light ( Ͼ 600 nm, 1200 or 400 mol m Ϫ2 s Ϫ1 ) at RT. The increase of SOSG-specific fluorescence emission was followed to quantify ROS release into leaves ( exc 480 nm, emis 530 nm). Fluorescence emission spectra were measured at RT using a Jobin-Yvon Fluoromax-3 spectrofluorimeter equipped with a fiber optic to measure emission of fluorogenic probes on leaves. For the measurements of 1 O 2 yield on either Lhc, PSII supercomplex, and PSI-LHCI, pigment-protein complexes were harvested from sucrose gradient and diluted in a reaction buffer (10 mM Hepes, pH 7.5, 0.06% ␣-DM, 2 M SOSG) to the same absorption area in the wavelength range 600 -750 nm (about 2.2 g of Chls/ml). Isolated complexes were illuminated with red light ( Ͼ 600 nm), and fluorescence yield of SOSG were determined before and during EL treatment to quantify 1 O 2 -dependent fluorescence increase (30). Upon EL treatment, we did not observe photobleaching of either chlorophyll or carotenoid in leaves or purified pigmentprotein complexes.
FDMR-For all the experiments, the samples were dissolved in buffer to a concentration of 200 g Chl/ml. Degassed glycerol was added (60% v/v) just before freezing the samples by direct immersion in liquid helium into the pre-cooled cryostat. FDMR experiments were performed in the laboratory-built apparatus, previously described in detail (44 -46). In the specific experiments, the modulation frequency and the microwave power were chosen depending on the triplet state. The temperature of all the experiments performed was 1.8 K. All the FDMR spectra are presented as ⌬I/I, where ⌬I is the fluorescence change induced by the resonant microwave field, and I is the steady state fluorescence detected by the photodiode. Because in all the experiments I remains substantially unmodified upon light treatments, the change of the intensity of ⌬I/I signal after EL is a measure of the change in triplet yield.

Statistics
Significance analysis was performed using either Student's t test or analysis of covariance in GraphPad Prism (see the figure legends for details). Error bars represent the standard deviation.
Sequence data from this article can be found in the EMBL/ GenBank TM data libraries under the following accession numbers: lut2 (line N656231 with insertion into the lycopene-⑀cyclase gene, At5G57030); npq1 (line N624757 with insertion into the violaxanthin-deepoxidase gene, At1G08550); npq2 (line N559469 with insertion into the zeaxanthin-epoxidase gene, At5G67030); ch1 (line N524295 with insertion into the chlorophyll-a oxygenase gene, At1G44446).

Effect of Zea-Lhc Interaction in Limiting 1 O 2 Release from
Arabidopsis Leaves under EL-When photosynthetic organisms are exposed to EL, photo-oxidative stress occurs with the production of 1 O 2 . Among the xanthophylls, Zea has been proposed to be the most active into scavenging of 1 O 2 and preventing lipid peroxidation (29). To investigate the function of the Lhc-bound zeaxanthin pool, we analyzed the photoprotective efficacy of A. thaliana mutants that were deficient in their ability to synthesize Zea (npq1) either in a WT or ch1 genetic background. The ch1 mutation inactivates Chl a oxidase (47) and thus Chl b synthesis; see supplemental Table S1 for pigment composition of these genotypes. Because Chl b is an essential co-factor for the assembly of Lhc pigment-protein complexes (48), a de-stabilization of Lhc proteins is obtained; the ch1 mutation leads to a strong reduction of PS antenna size (49), whereas xanthophylls bound to Lhc in the WT are released into the membrane (29).
We evaluated 1 O 2 release on leaf discs using SOSG (41), a 1 O 2 highly selective fluorescence probe. After illumination of discs with EL (1200 mol photons m Ϫ2 s Ϫ1 , RT), ch1 leaves showed a far higher release of 1 O 2 with respect to WT at each time point (supplemental Fig. S1) despite a 4-fold lower PSII antenna size (29). This observation is consistent with recent reports showing that the enhanced lipid peroxidation in ch1 is due to 1 O 2 attack on lipids (50) and that higher 1 O 2 yield in ch1 is attributed to the lack of xanthophyll binding complexes (30). To evaluate the photoprotection capacity of zeaxanthin in WT versus ch1 background, intensity of EL was chosen for npq1 and ch1npq1 leaf discs to obtain the same 1 O 2 yield; thus, WT and npq1 discs were illuminated with 1200 mol of photons m Ϫ2 s Ϫ1 , and ch1 genotypes were illuminated with 400 mol of photons m Ϫ2 s Ϫ1 . Results are reported in Fig. 1. The WT leaves exhibited a lower 1 O 2 yield per chlorophyll relative to the npq1 in EL (Fig. 1A), thus confirming the enhanced photoprotection capacity of Zea with respect to the other xanthophylls (29). However, Zea-dependent photoprotection is lower in ch1 plants; clearly, the effect of the npq1 mutation in increasing 1 O 2 release was significant only in non-ch1 genotypes (Fig. 1). These findings with ch1 plants imply that, although Zea has a high photoprotective effect, its performance is strongly enhanced through binding to the Lhc proteins.
Measurements of Light-induced 1 O 2 Yield on Purified Pigment-Protein Complexes-The results reported above suggest a role for Zea-Lhc interactions in the photoprotection of thylakoid membranes by preventing 1 O 2 release into the lipid phase. Previous investigation on recombinant Lhcs (20) and in vivo (51) showed that Zea can bind to specific sites of Lhc subunits; in particular, xanthophyll exchange occurs in the inner L2 site of the minor antennae of Lhcb4 (CP29), Lhcb5 (CP26), Lhcb6 (CP24), and LHCI (51-53), whereas trimeric LHCII binds Zea to the external site V1 (25).
The differential level of 1 O 2 production with versus without Zea is the signature of the Zea-dependent photoprotection mechanism and can be used for tracking the pigment-protein complexes responsible for this effect. To this aim, we compared the capacity for photoprotection of Lhcbs binding Vio or Zea by analyzing the amount of 1 O 2 released by pigment-protein complexes isolated from either dark-adapted (Vio) or EL-treated (Zea) leaves. The 1 O 2 yield was measured upon illumination of the complexes in the presence of SOSG. The pigment composition of purified complexes was analyzed by HPLC (supplemental Table S2). Binding of Zea to monomeric Lhcbs (a preparation of monomeric PSII antenna proteins that contains Lhcb4, Lhcb5, and Lhcb6 along with a fraction of monomerized LHCII complex) and LHCI (PSI antennae) significantly reduces the amount of 1 O 2 produced in EL. Fig. 2, A and C, show that 1 O 2 yield by monomeric Lhcb-Zea is reduced by 65% with respect to the Lhcb-Vio sample, the reduction being even greater in purified Lhcb5; likewise, the reduction in 1 O 2 yield of LHCI-Zea versus LHCI-Vio is 40% at 700 mol photons m Ϫ2 s Ϫ1 (Fig. 2D). These results imply that the Zea-containing complexes have a stronger capacity for photoprotection with respect to the same complexes binding Vio. Such an effect was not observed in trimeric LHCII, i.e. there was no change in 1 O 2 yield based on its Vio or Zea content (Fig. 2B). When the production of 1 O 2 was measured on both PSII supercomplexes (C 2 S 2 , see Ref. 54) and PSI-LHCI, complexes isolated from dark-adapted leaves (Vio binding) exhibited a 2-fold higher 1 O 2 yield with respect to complexes isolated from EL-treated leaves (Zea binding) (Fig. 2, E and F, respectively). These results show that the lack of bound Zea to Lhc complexes negatively affects the photoprotective efficiency of the whole photosystem for both PSI and PSII. Singlet oxygen production from WT and xanthophyll mutants lacking Zea (npq1, panel A) and/or Lhc complexes (ch1, panel B) was measured on intact leaves upon illumination (600 Ͻ Ͻ 750 nm, RT, 1200 mol of photons m Ϫ2 s Ϫ1 for WT and npq1, 400 mol of photons m Ϫ2 s Ϫ1 for ch1 background). The highly selective fluorescent probe SOSG was used to quantify light-dependent 1 O 2 release, as its 530-nm emission band increases in proportion to the amount of this ROS released by the photosynthetic apparatus. Each experimental point corresponds to a different sample; these data are representative of two independent experiments. Experimental points were modeled with a linear regression. Statistical analysis (analysis of covariance) revealed that npq1 leaves showed significantly higher 1 O 2 production than wild type (p ϭ 0.0013, panel A); instead, Zea-dependent photoprotection is less evident in ch1 plants (differences between the slopes of panel B are not significant, p ϭ 0.55).
Among Lhcs, the components of the monomeric Lhcb fraction showed the highest reduction in 1 O 2 yield upon binding Zea. To distinguish between the contributions of individual Lhc gene products, monomeric antennae fractions isolated from either dark-adapted and EL-treated WT leaves were further fractionated by preparative isoelectrofocusing; this purification procedure was shown to be affective in removing loosely bound pigments, such as xanthophyll bound to the external V1 site of LHCII, whereas xanthophylls of inner binding sites were retained (25). The pigment composition of purified complexes were analyzed by HPLC (supplemental Table S2); fraction 1 contained only monomeric LHCII and had a low content of Zea, whereas the fractions containing minor antennae had higher levels of Vio-Zea exchange. The fractions were further analyzed for their 1 O 2 release upon EL treatment. Results are reported in Fig. 3; monomeric LHCII-Zea (fraction 1) did not show significant differences in 1 O 2 yield with respect to LHCII-Vio (Fig. 3A); instead, fractions enriched in monomeric anten-nae (fractions 2-4) showed clear differences, with complexes from dark-adapted leaves yielding from 25% (CP29) to 65% (CP24) more 1 O 2 with respect to the same complexes isolated from EL-treated leaves (Fig. 3, B-D).
Role of Zea-Lhc Interaction in the Photoprotection of PSII Antenna Subunits-The hypothesis that the lack of Zea might prevent activation of photoprotective mechanisms localized within monomeric Lhcs, thus enhancing ROS production specifically at the level of monomeric Lhcbs, was further investigated by analyzing WT and npq1 genotypes for their capacity to resist degradation of Lhcb complexes in vivo upon exposure to EL. The kinetics of Lhcb photodegradation under strong light was quantified on total leaf extracts by immunoblotting with specific antibodies directed against different polypeptides supplemental Fig. S2). We observed that upon EL treatment the abundance in individual Lhcb polypeptides underwent a decay with rates that are affected by the presence/absence of Zea. A pronounced effect was clearly observed in the decay of minor FIGURE 2. Light-dependent singlet oxygen production from isolated pigment-proteins complexes. SOSG was used to follow light-dependent 1 O 2 release of pigment-protein complexes isolated from either dark-adapted or EL-treated leaves. Fluorescence of SOSG has been measured upon illumination of a solution containing monomeric Lhcb (A), trimeric LHCII (B), CP26 (C), and LHCI (D) with either Vio bound (solid line, closed symbols) or Zea bound (dashed line, open symbols) at different light intensities, 10 min each intensity. Trimeric LHCII isolated from either dark-adapted or EL-treated plants did not show significant differences in 1 O 2 yield at all light intensities tested (B). Instead, Zea binding monomeric Lhcb, CP26, and LHCI showed significantly lower 1 O 2 release than the Vio-binding complexes. Zea bound to these specific proteins plays a role in enhancing photoprotection by preventing the release of 1 O 2 from either PSII (E) and PSI-LHCI (F) supercomplexes. *, p Ͻ 0.05 by Student's t test of Zea binding complexes relative to the corresponding Vio binding.
antennae that depended on the WT or npq1 genetic background, whereas the decay rate of the major LHCII polypeptides was only marginally affected if at all (supplemental Fig.  S2). The decay for Lhcb4 -6 degradation was clearly faster in npq1 leaves (no Zea) than WT (ϩZea) during EL stress, whereas no difference was observed in the case of LHCII (supplemental Fig. S2). We conclude that the Zea-dependent enhancement of photoprotection is located in monomeric Lhcs rather than in the major trimeric LHCII complex.
Time-resolved Laser Spectroscopy Measurements of the Car Triplet-excited States-Previous work has shown that 3 Chl* quenching and 1 O 2 scavenging by xanthophylls ligands are the major determinants for the level of 1 O 2 release by Lhc proteins upon EL treatment in vitro and in vivo (55,56). To identify the physical mechanism(s) underlying the Zea-dependent photoprotective effect in plants, we measured the light-induced formation of triplet states (TmS spectra) upon excitation of Chl at 650 nm. These spectra are denoted TmS because they show the intersystem crossing of a pigment from its singlet state to its triplet state. The disappearance of the singlet state (S 0 ) and the appearance of the triplet state (T 1 ) are observed as a bleaching of the singlet absorbance bands (S 2 4 S 0 ) and the appearance of triplet absorbance bands (T 2 4 T 1 ). Such measurements of light-induced absorbance changes were done on a reconstituted in vitro suspension containing a mixture of Chl a and purified xanthophylls and in vivo on intact leaves from Arabidopsis genotypes differing in their capacity to synthesize Zea in EL (see supplemental Fig. S3, which compares the in vivo spectra with the triplet state spectra of Vio and Zea in solution, and supplemental Table S1 for pigment composition of the different genotypes). As expected, the kinetics and spectral response of the in vitro and in vivo systems were different because of the different environments (solvent or protein, respectively) surrounding the pigments. As commonly reported, the absorption peaks of pigments in solution are shifted toward shorter wavelengths as compared with what they are in the proteins, and due to diffusion, energy transfer between pigments in solution is slower than in their native environment. The data obtained on isolated pigments show that Zea is the "red-most carotenoid" (Table 1). It suggests that the substitution of Vio into Zea must translate into a red shift of the in vivo triplet spectrum of carotenoids.
On intact leaves a strong absorbance increase is instantly (Ͻ100 ns) observed at 520 nm after a flash of 650 Ϯ 10-nm light (supplemental Fig. S3). The decay of this signal is biexponential, with half-times of ϳ2 s and ϳ200 ms. Measurements of the dependence of each component on the intensity of the excitation light and the corresponding spectra at approximately maximum amplitude are shown in supplemental Fig. S3, C and D, respectively. The fast (ϳ2 s) component dominantly results from 3 Car*, whereas the slow component (ϳ200 ms) is due to the electrochromic band-shift of carotenoids induced by transmembrane charge separation (57). Therefore, the TmS spectra of carotenoids can be deconvoluted from the electrochromic shift of carotenoids on the basis of their very different lifetimes. The amplitude of the triplet signal largely exceeds the amplitude of the bandshift signal, itself largely exceeding the P700 ϩ signal; indeed, in the spectral region 500 -560 nm (Fig. 4) DECEMBER 7, 2012 • VOLUME 287 • NUMBER 50 signal is very flat and of small amplitude (58). In the subsequent text, the fast microsecond component is referred as TmS (triplet-minus singlet) difference spectrum.

Effect of Zeaxanthin in the Modulation of Chl Triplet Yield
The TmS transition of WT, npq1, and lut2 leaves, darkadapted and after exposure to strong actinic light (to trigger the synthesis of Zea), in the 500 -560-nm range that includes the 3 Car* transition are shown in Fig. 4. In comparing dark-adapted versus EL-treated leaves, a loss of signal was observed upon illumination in WT (Fig. 4A) and lut2 (Fig. 4C) leaves between 500 and 520 nm (minimum at 505 nm), whereas a gain of signal was observed above 520 nm, with a maximum between 530 and 535 nm. Such an EL-induced red shift in 3 Car* transition level is expected from the de-epoxidation of Vio into Zea and from the red-most spectral contribution of the latter (supplemental Fig.  S3). Consistently, the npq1 mutation (Fig. 4B) abolished the red shift. It is worth noting that the amplitude of the 3 Car* red shift measured on intact leaves was proportional to the amount of Zea accumulated during EL treatment, which was not existent in npq1, greatest in lut2, and in-between the changes observed in npq1 and lut2 for WT leaves. These results show that Zea synthesis correlated with the formation of the red-shifted 3 Car* spectral form on intact leaves.
The red-shifted spectral form is likely to arise via 3 Chl* quenching rather than scavenging of 1 O 2 because, although scavenging of 1 O 2 by Zea occurs in the lipid phase (22,29), it is very unlikely that it would be detected on timescales as short as 100 ns. To test this hypothesis, ch1 mutants were investigated, as ch1 plants lack Lhcb and have eight times more lipid-free xanthophylls than WT plants (30), implying that the redshifted 3 Car* spectral form should be more evident in these mutants if it results from scavenging. Upon light-induced Zea synthesis, the ch1lut2 plants (lacking both Lhcbs and Lut) did not undergo changes in the carotenoid TmS spectrum (Fig. 4D), although they do synthesize twice as much Zea than lut2 plants (29). It seems plausible that ␤-carotene is the only carotenoid contributing to the TmS spectra of ch1lut2 leaves; indeed in this genotype all xanthophylls are released into the thylakoid membrane, and ␤-carotene is the only carotenoid bound to chlorophyll-protein complexes, thus in condition to perform direct quenching of 3 Chl*. This is consistent with TmS spectrum reported for ch1lut2 dark-adapted leaves, which showed a maximum even redder than that of EL-treated WT leaves; indeed, similar 3 Car* transition maxima have been reported for Zea and ␤-carotene (59). Additional measurements on mutants with different abilities for synthesizing Zea under EL treatment, namely npq1lut2 and npq2lut2, showed that the formation of the red-shifted 3 Car* signal was associated with the extent of Vio to Zea conversion (Fig. 4, E and F). Moreover, TmS red shift still holds in a mutant lacking qE (npq4lut2, see supplemental Fig. S4) implying it is independent from PsbS, the protein essential for qE. Therefore, we conclude that the Car red shift observed is only due to Zea bound to the Lhc proteins serving as Chl a/b-xanthophyll binding antennas of photosystems.
Identification of Photosynthetic Subunits Involved in the Red Shift of Carotenoid Triplet Signal-To find the binding site for the Zea responsible for the red shift observed in the 3 Car* spectrum of leaves, we performed TmS spectroscopy measurements on proteins isolated from WT leaves that were either darkadapted or treated with EL ( Fig. 5; see supplemental Table S2 for pigment composition of purified complexes). Changes were observed in the monomeric Lhcb subunits (Fig. 5A) and LHCI (Fig. 5D)-containing fractions, whereas trimeric LHCII did not show any changes in the 3 Car* spectrum when isolated from EL-treated plants (Fig. 5B). These results suggest that both PSII and PSI have a specific Lhc target for Zea binding. A further fractionation of monomeric Lhcbs into antenna components revealed that changes in the 3 Car* spectrum were measured in all minor antennae (CP29, CP26, and CP24). Given by way of example, Fig. 5C compares TmS spectra from samples of CP26 containing either Vio or Zea (CP26 dark-adapted and CP26 EL, respectively) at the same Chl concentration. The peak amplitude of the CP26 dark-adapted sample is reduced by 45% with respect to CP26 EL, whereas the amount of absorption at wavelengths longer than 520 nm is ϳ60% higher in the CP26 EL sample containing Zea. In the LHCI samples containing either Vio or Zea (LHCI dark-adapted and LHCI EL, respectively) the peak was observed at 515 and 525 nm, respectively (Fig. 5D). We conclude that the TmS spectral shift observed in vivo tightly correlates with that detected in the LHCI and Lhcb4 -6 subunits but not in the major trimeric LHCII complex. Moreover, both of these spectral features are associated with the major component of Zea-dependent photoprotection.
Investigation on the Chlorophyll-to-carotenoid Triplet Transfer-The mechanism(s) underlying the decreased production of 1 O 2 upon Zea binding could be either (i) an increased efficiency in 3 Chl* quenching by xanthophylls bound to Lhc proteins or (ii) a direct down-regulation of the chlorophyll triplet yield, as compared with other de-excitation pathways. To distinguish between these two hypotheses, we assessed both the total amount of 3 Car* and the kinetics of 3 Car* formation in isolated pigment-protein complexes by time-resolved TmS spectroscopy. For this experiment, we chose the monomeric CP26 protein purified from dark-adapted or EL-treated WT leaves; in the latter, both (i) the Zea-dependent reduction of 1 O 2 release (Fig. 2C) and (ii) the red shift in TmS spectra (Fig. 5C) were observed.
For an estimation of 3 Car* formed upon chlorophyll excitation, the TmS spectra of CP26 was fitted with 3 Car* spectral forms in a protein environment (Fig. 6), obtained by measuring recombinant LHC reconstituted with single xanthophyll species. 3 Although this deconvolution problem might have more 3 R. Bassi, unpublished results. than one solution, the fits given in Fig. 6 give a likely estimation of the total amount of 3 Car* formed upon excitation of Chls in the two samples. Both TmS spectra were described using single 3 Vio*, 3 Lutein* ( 3 Lut*), and 3 Zea* forms; the spectra of the CP26-Vio and CP26-Zzea complexes fit well by using similar spectral forms of 3 Vio* and 3 Lut*, whereas a 3 Zea* form was necessary to fit the CP26-Zea TmS spectrum (Fig. 6, A and B).
The contributions of each xanthophyll to the total amount of 3 Car* formed were determined by using both the amplitudes in TmS spectra and 3 Car* extinction coefficients (⑀ T ) measured on isolated xanthophylls (Table 1) and are shown in Fig. 6C. In the CP26-Zea sample, the contributions of both 3 Vio* and 3 Lut* (which are reduced with respect to CP26-Vio sample) are compensated by formation of 3 Zea* (Fig. 6C).
To assess the kinetics of 3 Car* formation on purified CP26, time-resolved absorbance changes were recorded at two distinct wavelengths, corresponding to the maximum absorbance of 3 Vio* ϩ 3 Lut* (510 nm) and 3 Zea* (530 nm). Similar halftimes for both 3 Car* rise and decay (supplemental Fig. S5) were determined at both wavelengths, implying no differences in the 3 Chl* quenching capacity of CP26-bound Zea versus Vio.
In light of these findings, we attempted to verify the alternative hypothesis of direct modulation of 3 Chl* yield upon bind-ing of Vio versus Zea independent from the quenching by xanthophylls. To this aim, we used FDMR. FDMR is a double resonance technique based on the principle that when a triplet steady state population is generated by illumination, application of a resonant electromagnetic field between a couple of spin sublevels of the triplet state induces a change in the steady state population of the triplet state itself due to anisotropy of the decay and population rates of the three spin sublevels. In FDMR experiments, the change induced in the triplet population is detected as a corresponding change in the emission of the system (60,61). The activity of the carotenoids in quenching triplet states was investigated at low temperature (1.8 K), when the funneling of the excitation toward the low lying excited states of the Chl molecules is fast.
The low temperature emission spectrum of thylakoids from Arabidopsis WT showed the characteristic Chl a fluorescence bands peaking at 685-695 and 735 nm due to the different antenna pools belonging to PSII and PSI, in agreement with literature data (data not shown). The FDMR signals, detected at 680 -690 (PSII) and 720 nm (PSI), which represent transitions of 3 Chl* states, are shown in Fig. 7A. The three FDMR transitions with the polarization pattern usually found for 3 Car*, detected at 690 and 730 nm, are presented in Fig. 7B (44 -46, 62). Two 3 Chl* components (visible in the 690 nm FDMR spectra), with ͉D͉ Ϫ ͉E͉/͉D͉ ϩ ͉E͉ transitions at about 725/945 and 742/965 MHz, have been previously assigned to the PSII outer antennae, whereas the 767/992-MHz component has been attributed to the core complex (45) (note that the signals have been reversed in sign compared with Ref. 48 for better representation). The 3 Car* states detected at 690 nm has been assigned to the PSII Lhcs (45). The negative signal at ϳ1000 MHz in the ͉D͉ Ϫ ͉E͉ spectrum of 3 Car* in thylakoids (Fig. 7B) is the contribution of the 3 Chl* state from the PSII core complex (992 MHz component). The amplitude of this signal is enhanced due to the high frequency modulation (325 Hz) used for the 3 Car* measurement because of its fast decay (48). Other slower 3 Chl* components are not detected at this frequency (63).
The FDMR spectra of the thylakoids isolated from darkadapted npq1 leaves were identical to those of the sample from WT in the entire detection range explored (Fig. 7C). When measurements were performed on thylakoids isolated from WT EL-treated leaves (see supplemental Table S3 for pigment composition of thylakoids used), a decrease (Ϫ25%) in the intensity of the FDMR signals assigned to two 3 Chl* states belonging to PSII (͉D͉ Ϫ ͉E͉ transitions at 742 and 767 MHz) was observed. Also, a new component (719/992 MHz) appeared that can be assigned to the P 680 recombination triplet state (64). Interestingly, in the thylakoids isolated from npq1 EL-treated leaves, the intensity of the 742 MHz component did not undergo any decrease. A decreased amplitude was observed for the 767-MHz component, with a corresponding increase in the 720 MHz component (Fig. 7C). Except for the lack of the 3 P680* state formation, similar results were obtained on thylakoids isolated from dark-adapted WT and npq1 leaves and de-epoxidated in the dark by incubating thylakoids at pH 5.2 (supplemental Fig. S6), implying that Zea itself, rather than the light treatment, is responsible for the spectral changes observed.
The FDMR signals of the 3 Car* states were sensitive to the treatments performed on the sample; the prolonged EL exposure induced a decrease of the intensity of the signals in WT thylakoids, whereas the effect was not evident in the npq1 thylakoids (Fig. 7D). It should be noticed that carotenoids triplet states were observed in the presence of Zea, however, not with an increased yield. Lut and Zea contributions to the FDMR signal cannot be distinguished, as they have the same resonance frequencies as detected by comparing spectra from npq2lut2 genotypes, containing Zea as the only carotenoid with the npq2 containing Lute and Zea (not shown). Although FDMR cannot ascertain if Zea triplet states are or not populated at all, it is clear that the carotenoid triplet yield does not increase and does not correlate with the decrease of the 3 Chl* population in the presence of Zea. Once identified, the Zea-dependent effect on modulation of 3 Chl* populations in thylakoids was tracked to the individual pigment-protein components by performing the same measurements on the different proteins isolated from WT leaves, either dark-adapted or treated with EL (Fig. 7E).
The ratio between the intensities of ͉D͉ Ϫ ͉E͉/͉D͉ ϩ ͉E͉ transitions of 3 Chl* are different in isolated Lhcs (Fig. 7E) with respect to thylakoids (Fig. 7A). Amplitudes of these transitions are sensitive to the environment of the triplet states; indeed 3 Chl* states show variable amplitudes in the two transitions due to changes in both populations and decay probabilities of the triplet sublevels. In thylakoids, the presence of interacting complexes creates a protein environment different than the deter-gent micelles of the isolated Lhcs. Moreover, FDMR spectra of thylakoids also contain the contribution from the PSII inner antenna subunits (767/992 MHz), absent in purified Lhcs.
Again, the Lhcbs affected by the light irradiation were the monomeric Lhcb4 -6 proteins rather than the major LHCII complex. In Fig. 7E, the FDMR signals of 3 Chl* states detected at two different wavelengths show a decrease in intensity of 40% in the monomeric antenna fraction when binding Zea versus Vio. We did not observe any increase of the 3 Car* states correlated to the decrease of the 3 Chl* states. No change on the triplet states FDMR signals was detected in the samples from npq1, and the FDMR spectra of isolated trimeric LHCII from WT and npq1 were both insensitive to EL exposure of leaves previous to isolation (Fig. 7F). Establishing if the subset of chromophores-forming triplets is the same in both thylakoids and Lhcs is not straightforward. However, the effect of EL on the 3 Chl* yield is the same in both the minor Lhcbs and in thylakoids, suggesting these triplets have the same origin.

DISCUSSION
In this work we have scrutinized the mechanisms that contribute to the photoprotective effect of Zea, the xanthophyll specially synthesized in response to EL conditions. Previous work indicated that Zea synthesis has multiple effects. Earlier reports have emphasized the enhancing effect on qE, the PsbSdependent thermal dissipation of 1 Chl* excited states (11,14), and the uncoupler-resistant component called irreversible quenching or, more recently, qZ (51). A second photoprotective effect was reported to consist of a scavenging effect of lipidfree Zea for ROS released from Chl binding complexes (22,29). A Zea-enhanced scavenging effect was also reported within Lhc proteins (30,65). Although these functions contribute to the Zea photoprotective effect, genetic dissection showed that thermal dissipation of excess energy dissipation has a relatively small effect as assessed using the npq4 mutant lacking qE (66) and likewise for the scavenging effect of lipid-free Zea with respect to the Lhc-bound fraction (Fig. 1). This implies that the Zea-dependent photoprotection effect is associated to the binding of Zea to Lhc proteins, where it has a strong effect in decreasing 1 O 2 evolution during illumination. Here, we used high sensitivity laser spectroscopy to investigate changes in the optical properties of leaves associated with EL treatment. Illumination of Arabidopsis wild type plants induced a spectral red shift of the T 2 4 T 1 transition of 3 Car*. Using a range of Arabidopsis mutants, this spectroscopic feature was observed only in genotypes able to accumulate Zea, either upon EL exposure or constitutively (Fig. 5), irrespective of their the ability to perform qE (supplemental Fig. S4) and was coupled to a modulation of 3 Chl* yield, suggesting this mechanism is a component of photoprotection.
A Mechanism of Photoprotection Based on 3 Chl* Downregulation-Because the TmS red shift involves the major 3 Car* transition, we first investigated if Zea had an enhanced 3 Chl* quenching capacity with respect to the pre-existing Viola, thus yielding a lower level of 3 Chl* formation. However, this was not confirmed by experimental evidences as (i) although not entirely resolved due to the 10 ns limitation in time resolution of our spectrophotometer, the kinetics of 3 Car* population was the same in Vio (510 nm) and Zea (530 nm) binding CP26 (supplemental Fig. 5), (ii) the relatively similar extinction coefficients of 3 Zea* and 3 Vio* ( Fig. 6 and Table 1) imply that the decrease in the 510-nm band ( 3 Vio*) appears to be compensated by a similar increase of the 530-nm ( 3 Zea*) component in the spectra of Fig. 5C without a major difference in quenching efficiency between Zea and Vio, and (iii) a slight decrease in the 3 Car* level was detected by both laser spectroscopy (Fig. 6) and FDMR (Fig. 7D) in Zea versus Vio binding complexes. Rather, FDMR measurements showed that Zea binding to Lhc proteins induces a down-regulation of the 3 Chl* yield of the complexes (Fig. 7E) and whole thylakoids (Fig. 7C), likely through a mechanism different from the previously described triplet energy transfer to xanthophyll ligands (56,67). The spectra are taken at 1.8 K; however, the samples adapted at RT were frozen by direct immersion on liquid helium into the cryostat, and therefore, the spectra are representative of the conformational distribution present at RT. It should be noted that the decrease in the concentration of 3 Chl* on both thylakoids and minor antennae cannot be ascribed to the residual quenching of 1 Chl* observed (68) upon EL treatment, as the procedure for thylakoid isolation is long enough to allow complete relaxation of NPQ induced by EL on leaves (19); the small quenching of 1 Chl* measured upon binding of Zea to minor antennae (about 10% than the corresponding sample from dark-adapted WT leaves, data not shown) cannot account for a 40% decrease in the concentration of 3 Chl* in these complexes (Fig. 7E). Moreover, FDMR signals are normalized to the amplitude of the steady state fluorescence. Thus, if the lower 3 Chl* yield was only due to the quenching of the corresponding singlet states, a decrease in the FDMR signal of the same level would be expected; instead, the decrease in the 3 Chl* yield in thylakoids upon EL treatment was ϳ25% (Fig. 7C) without a change in steady state fluorescence (not shown).
It is well known that CP26 and other Zea binding Lhc subunits undergo a conformational change upon exchange of Vio to Zea (51). We suggest that this conformational change, by affecting Chl-Chl and/or Chl-protein interactions, does modulate the triplet yield of the chlorophyll ligands and thus downregulates the probability of reaction with O 2 and, subsequently, 1 O 2 formation. Zea binding to Lhc proteins has at least two effects that may be correlated to each other or may not; that is, (a) quenching of 1 Chl* excited states and (b) decrease in 3 Chl* yield.
The relationship between the small change in fluorescence yield and the large change in triplet yield undergone by monomeric Lhc proteins upon binding of Zea is not clearly understood. An efficient thermal deactivation of 3 Chl* once they are formed appears unlikely because it requires intersystem crossing, making it a relatively slow process. A weak 1 Chl* quenching effect can be accounted for by an effect limited to a subset of chlorophylls in the pigment-protein complex. Such an effect could be produced by one of the following mechanisms; (i) the conformational change induced by Zea binding causes excitons in Chls bound to such a domain to be diverted from the sites where 3 Chl* are formed with higher yield (i.e. the red-most sites such as Chl A2, A4), possibly due to changes in the energy level of the former Chls (69), or (ii) alternatively, Chls generating triplet states with higher yield might become more readily quenched when still at their 1 Chl* state via thermal relaxation (70,71). In such case, the singlet quenching would be limited to a subset of Chls of limited amplitude when refereed to the whole complex (52,68,72), whereas the effect on triplet state would be stronger.
Quenching of singlet states during short term light adaptation (qE) has been shown to be correlated to energy transfer from the Chl Q y transition and the short-living 1 Car* state followed by thermal deactivation via the formation of a Chl Ϫ / Car ϩ radical pair (18,69). Such a mechanism could also favor 3 Chl* to 1 Car energy transfer. However, the radical pair formation has been shown to require both the synthesis of Zea and the presence of a trans-membrane pH gradient, whereas its yield coefficient is far lower in isolated Lhc (18); this evidence is not consistent with the 40% reduction in 3 Chl* yield detected in isolated pigment-protein complexes (Fig. 7).
Alternatively, the modulation of 3 Chl* yield might be independent from the formation of radical pairs. Whatever the physical mechanism involved, the physiological effect is strong and produces effective photoprotection. 3 Chl* Down-regulation and PSII Supercomplex Organization-The search for fractions of thylakoid membranes exhibiting the red-shifted TmS spectrum and reduced FDMR 3 Chl* signal yielded two major targets: monomeric Lhcb4 -6 antenna subunits of PSII and LHCI. This is in agreement with previous work on Vio versus Zea exchange in different Lhc proteins during the operation of xanthophyll cycle (20,53), as CP26 (Lhcb5), CP24 (Lhcb6), and Lhca4 were found to be the best "exchangers" among all Lhcs. Previous work with recombinant Lhc proteins (20) and in vivo (51) has shown that xanthophyll exchange is operated at the level of binding site L2 in both CP26 and LHCI. It would, therefore, appear that binding of Zea to Lhc proteins is sufficient for the modification of the triplet-state properties of the pigment-protein complexes, whereas the actual expression of this effect is controlled by the capacity for xanthophyll exchange in site L2 of the individual gene products. It should be noted that monomeric Lhcbs are located in between the PSII core complex and the outer antenna layer made by 2-4 copies of trimeric LHCII (75), each binding 42 Chls and 12 xanthophylls (26), implying most excitation energy collected by LHCII is funneled to the PSII reaction center through monomeric Lhcbs. Supplemental Fig. S2 clearly shows that, under EL conditions, Lhcb monomers get destroyed faster than LHCII unless they bind Zea that makes them more resistant to photodestruction.
A recent report demonstrated that, although xanthophyll ligands are very efficient in 3 Chl* quenching, a limited fraction of 3 Chl* cannot be quenched. In their original work, Mozzo et al. (56) measured a small amount of unquenched 3 Chl* at physiological temperatures in LHCII. This is consistent with evidence that some Chls in purified LHCII, e.g. Chl 611 (26) were located too far apart from xanthophylls for efficient triplet transfer and can explain the bleaching of the complex when challenged with EL (76) (supplemental Fig. S2). In LHCII, ϳ5% of the triplets reside on Chls, corresponding to a 95% efficiency for Chl to Car triplet transfer (56); this value was lower for Lhcb5 and for Lhcb6 (52), reaching ϳ20% of unquenched 3 Chl* in Vio binding complexes.
Results described here (Fig. 7) clearly show that even Chls in minor antennae, although active in singlet energy transfer (as assessed by fluorescence emission measurements), are not equally active in transferring triplets. It is well known that triplet transfer requires shorter distances between the chromophores than singlet transfer. The absence of a triplet transfer enhancement upon Zea synthesis on one hand and regulation of 3 Chl* on the other, however, suggests that the binding of Zea in site L2 has an effect on the structure of the complex; it might result in an altered organization of protein domains to modulate the triplet yield of some chlorophylls rather than in a conformational change that reduces the proximity between unprotected chlorophylls and xanthophylls.
Why is Zea only present during stress conditions when it is effective in photoprotection? At least one reason, if not the only reason, is that massive binding of Zea to Lhc (such as in npq2 mutant of Arabidopsis) is well known to strongly decrease the excited singlet state lifetime (77) of the plants. Therefore, plants that constitutively accumulate Zea can utilize less photons for charge separation, which decreases the growth rate in limiting light (51). We conclude that monomeric Lhcbs in WT PSII are present in two states, dark-adapted and EL, binding Vio and Zea, respectively, in their L2 sites. The dark-adapted state ensures a longer fluorescence lifetime and relatively high 3 Chl* yield, which is not a problem due to efficient photochemical quenching in low light conditions. The EL state has a lower 3 Chl* level and a shorter 1 Chl* lifetime (68) than the darkadapted state, which is well suited for photoprotection under these conditions. Besides Lhcbs, Zea also binds to Lhca polypeptides, which is so far unexplained under the widespread understanding that the major Zea function is to enhance qE due to the low level of excited states (short fluorescence lifetime) of PSI. Nevertheless, Lhca subunits host the red-most absorption forms that make a low energy trap for excitation energy before photochemical quenching by P700 (78). Because of the 3 Chl* down-regulating effect of Zea, its binding to Lhcas may be understood as a photoprotection mechanism for highly localized excitons in Lhca3 and Lhca4, the red-most subunits in the PSI-LHCI supercomplex (79).
We have shown that Zea accumulation in chloroplasts under EL produced a red shift of the major carotenoid transition in the TmS spectra. These changes were measured on intact leaves and correlate with the amount of Zea accumulated in specific Lhc subunits located in between the major LHCII antenna complex and the reaction center in PSII supercomplexes and in LHCI. The resistance/sensitivity of these proteins to EL conditions indicated that the red shift is correlated with the dominant component of photoprotection, in absence of which monomeric Lhcbs are preferentially destroyed. The protection effect appears to consist with a direct down-regulation of 3 Chl* without a corresponding triplet quenching enhancement by nearby xanthophylls. High resolution crystallography of Zea binding complexes will further assist with elucidating this new mechanism of photoprotection. After this manuscript was finished, the report by Carbonera et al. (73) was published, showing that 3 Chl* concentration is strongly reduced upon high light treatment. Although this result is consistent with the present report, it refers to a component of photoprotection distinct from that discuss here. In fact, in Ref. 73 the protective effect was abolished by the lhcsr1 lhcsr2 psbS mutation abolishing qE. On the contrary, the zeaxanthin-dependent photoprotection component here described is fully active in the npq4 mutant lacking PsbS (supplemental Fig. S4).