PSB33 protein sustains photosystem II in plant chloroplasts under UV-A light

Plants can quickly and dynamically respond to spectral and intensity variations of the incident light. These responses include activation of developmental processes, morphological changes, and photosynthetic acclimation that ensure optimal energy conversion and minimal photoinhibition. Plant adaptation and acclimation to environmental changes have been extensively studied, but many details surrounding these processes remain elusive. The photosystem II (PSII)-associated protein PSB33 plays a fundamental role in sustaining PSII as well as in the regulation of the light antenna in fluctuating light. We investigated how PSB33 knock-out Arabidopsis plants perform under different light qualities. psb33 plants displayed a reduction of 88% of total fresh weight compared to wild type plants when cultivated at the boundary of UV-A and blue light. The sensitivity towards UV-A light was associated with a lower abundance of PSII proteins, which reduces psb33 plants’ capacity for photosynthesis. The UV-A phenotype was found to be linked to altered phytohormone status and changed thylakoid ultrastructure. Our results collectively show that PSB33 is involved in a UV-A light-mediated mechanism to maintain a functional PSII pool in the chloroplast.


Introduction
The light energy absorbed from the sun is not always beneficial for plants-in excess or under periods of environmental stress, the light causes inhibition of photosystem II (PSII) activity that may result in damage to the photosynthetic apparatus, a phenomenon referred to as photoinhibition (Kok, 1956;Li et al., 2018). To minimize photoinhibition, plants have evolved protective mechanisms that ensure functional photosynthesis under light stress conditions (Long et al., 1994;Scheller and Haldrup, 2005;Sarvikas et al., 2006;Yamamoto, 2016). These responses include limiting and dissipation of excess of absorbed light energy, and damage control by scavenging reactive oxygen species (ROS).
The thylakoid PSII auxiliary protein PHOTOSYSTEM II PROTEIN 33 (PSB33, AT1G71500) plays a specific role in the regulation of photoinhibition. PSB33 influences PSII-LHCII supercomplex organization in Arabidopsis (Fristedt et al., 2015), and plants devoid of PSB33 display reduced state transition, i.e. ability to distribute excitation energy between the two photosystems (Fristedt et al., 2017). Furthermore, psb33 mutants show stunted growth in moderate, high, and fluctuating light (Fristedt et al., 2015(Fristedt et al., , 2017. When we started cultivating psb33 mutant plants in our new growth facility, we were unable to reproduce our previous finding that the mutant grows more slowly than wild type under standard conditions (Fristedt et al., 2015). Instead of the expected 50% reduction in fresh weight compared with wild type, psb33 mutants grew to almost similar size (see Supplementary Fig. S1 at JXB online). We hypothesized that this variability in the size of psb33 is attributed to light spectral differences since the light intensity was identical between these two experiments.
Photoinhibition in plants is initiated by photoinactivation of PSII and is followed by a repair cycle of the damaged subunits. There is currently no consensus regarding how photoinactivation of PSII is initiated and sensed by the plant. The two leading hypotheses, which are not mutually exclusive, are (i) the acceptor-side PSII photoinhibition mechanism induced by singlet oxygen in excessive light, and (ii) the donorside PSII photoinhibition occurring via light absorbed by the Mn 4 O 5 Ca cluster of PSII and occurring even in very low light intensities (Aro et al., 1993;Zavafer et al., 2015). However, from studies on the action spectrum of PSII photoinactivation, it is evident that the extent of photoinhibition is wavelength dependent and that light in the UV region of the spectrum (100-400 nm) causes the strongest photoinactivation of PSII (Jones and Kok, 1966;Sarvikas et al., 2006;Takahashi et al., 2010), predominantly targeting the Mn 4 O 5 Ca cluster of the reaction center and the acceptor side of PSII (Hakala et al., 2005;Ohnishi et al., 2005). Thus, non-photochemistry wavelengths can substantially affect PSII photoinhibition.
The relative excitation pressure on the two photosystems, PSI and PSII, changes depending on light intensity and spectral properties, and the inability to appropriately distribute the light energy between the photosystems may lead to photoinhibition (Bellafiore et al., 2005). Plants can reallocate a fraction of the light-harvesting antenna complex II (LHCII) between PSI and PSII through a process known as state transition (Allen et al., 1981;Adamska et al., 1992). The phosphorylation of the LHCII antenna proteins turns on this reallocation. . High excitation pressure on PSII activates the serine-threonine kinase STN7 that phosphorylates LHCII proteins (Bellafiore et al., 2005;Fristedt and Vener, 2011;Trotta et al., 2016). This results in enhanced delivery of excitation energy from LHCII to PSI, thus increasing its light-harvesting capacity (state II). Over-excitation of PSI, on the other hand, triggers the dephosphorylation of LHCII by the PPH1/ TAP38 phosphatase (Pribil et al., 2010;Shapiguzov et al., 2010). Dephosphorylated LHCII associates with PSII to increase its capacity to absorb energy (state I). Under artificial illumination, state I is induced by low-intensity light (Rintamäki et al., 2000) as well as red (636-700 nm) and blue (400-500 nm) light (Bellafiore et al., 2005;Trotta et al., 2016). Far-red light (700-780 nm) preferentially excites PSI and triggers state II (Adamska et al., 1992;Bonardi et al., 2005).
Repair of photodamaged PSII proteins is efficient and under strict quality control, especially of the D1 subunit. This photoinhibition-repair cycle follows a sequential series of highly regulated steps: monomerization of damaged PSII dimers, migration of PSII monomers from appressed to unappressed thylakoid regions, degradation of the damaged D1 protein by FtsH protease, and finally, de novo production of D1 subunit insertion into PSII monomers (Ohad et al., 1984;Yoshioka-Nishimura and Yamamoto, 2014). Phosphorylation of the PSII core proteins by the STN8 kinase is important for the repair cycle, influencing the disassembly of the damaged PSII dimer and protecting D1 protein from excess degradation (Tikkanen et al., 2008;Fristedt et al., 2009;Kato and Sakamoto, 2014).
Apart from directly fueling photochemistry, light of different colors and intensities serves as cues during plant development and growth. A large number of photoreceptors allow plants to sense light over a broad spectrum of wavelengths, from UV-B (280-320 nm) to far-red. Phytochromes absorb red and far-red light in the 600-750 nm region and play central roles in plant germination, de-etiolation, stomata development, flowering, shade avoidance, and senescence (Wang and Wang, 2015). Blue and UV light are perceived and effectively transduced as a signal by the photoreceptor families cryptochromes (400-500 nm), phototropins (320-500 nm), zeitlupes (450-520 nm) and UV-B photoreceptors (280-320 nm) (Christie et al., 2015). These photoreceptors control processes that include photomorphogenesis, flowering, circadian period, phototropism, and stomata opening. Signaling pathways downstream of the photoreceptors often integrate with hormone networks and influence their levels (Chaiwanon et al., 2016). Although well studied, there is limited evidence linking photoreceptors directly to regulation of photosynthesis. Photoprotective mechanisms identified in green algae and cyanobacteria have been tied to UV and/or blue light perception by photoreceptors (Kirilovsky and Kerfeld, 2013;Allorent et al., 2016;Petroutsos et al., 2016). Further, blue light perception in diatoms increases their photoprotective potential (Schellenberger Costa et al., 2013). However, these are aquatic organisms exposed to an environment where blue light dominates. In land plants, phototropins activated by blue light stimulate the relocation of chloroplast (Goh, 2009). This process acts to distribute PSII damage among the different cell layers of the leaf in high light (Cazzaniga et al., 2013). More indirectly, blue and UV-A (320-400 nm) light play a role in the regulation of the PSII complex as they stimulate the expression of genes encoding lightharvesting chlorophyll a/b-binding proteins (Lhcbs) (Folta and Kaufman, 1999) and a special class of PSII core proteins (D2 and CP43; PSBD-PSBC) (Christopher and Mullet, 1994).
Here, we aimed to investigate if changes in light quality affect growth and photosynthetic responses in psb33 mutant plants. We found that the psb33 plants show a conditional dwarf phenotype when cultivated under blue light, specifically in UV-A or near UV-A light. This UV-A light-dependent phenotype was found to be associated with low levels of PSII proteins, abnormal thylakoid ultrastructure, and changed phytohormone levels. We suggest that PSB33 is involved in a UV light-mediated mechanism to maintain an efficient PSII repair by regulating the de novo synthesis of PSII.

Plant material and growth conditions
Arabidopsis seeds were sown on soil and vernalized at 4 °C for 2 d before transfer to climate-controlled growth chambers (CLF PlantMaster, Plant Climatics, Wertingen, Germany). The plants were grown under a 12 h photoperiod at 20 °C/18 °C light/dark in 60% relative humidity. Seedlings were transplanted into individual pots after 10 d and moved to chambers with LED lamps after an additional 4 d. All mutants and wild type used in the study were in the Columbia-0 genetic background, previously described for psb33-3 (Fristedt et al., 2015), stn7 (Bellafiore et al., 2005), stn8 and stn7stn8 (Bonardi et al., 2005), tap38 (Pribil et al., 2010), amiLhcb1 and amiLhcb2 (Pietrzykowska et al., 2014), psbw (García-Cerdán et al., 2011), psal (Lunde et al., 2000), pgr5 (Munekage et al., 2002), and npq4 (Johnson and Ruban, 2010). The LED lights were provided by Heliospectra L4A lamps (Heliospectra, Gothenburg, Sweden) with diode peak wavelengths at 400, 420, 450, 530, 630, 660, and 735 nm. For the experiment with the 385 nm LEDs, Heliospectra RX30 lamps were used. Full light spectra were recorded with a JAZ spectrometer (Ocean Optics, Dunedin, FL, USA). Lamps were set up to ensure an even light distribution of the different colors. For example, when blue and red light were combined, 50% of the total photon flux was provided by blue LEDs and 50% by red LEDs.

Pigment content, hormone analysis, and chlorophyll a fluorescence
Chlorophyll and carotenoids were extracted from leaf discs in 95% (v/v) ethanol for 10 min at 60 °C, and concentrations were determined spectrophotometrically (Lichtenthaler and Wellburn, 1983). Plant hormone levels were determined from leaf discs (12 mm diameter) as described (Floková et al., 2014).
Chlorophyll a fluorescence parameters were recorded with a Pocket PEA Chlorophyll Fluorimeter (Hansatech Instruments Ltd, King's Lynn, UK). The maximum quantum efficiency of Photosystem II (F v /F m , calculated as (F m− F 0 )/F m ) and performance index (PI ABS ) (Strasser et al., 2004) were determined in attached leaves dark-acclimated for 20 min in room temperature with 1 s measuring light with the intensity of 3500 µmol m −2 s −1 .

Transmission electron microscopy
Leaves from plants grown in fluorescent white, red, or blue light were collected 2 h after the onset of light following a 14 h night period. Tissue preparation and imaging were as described (Herdean et al., 2016).

PSII electron transfer properties
Electron transfer properties in PSII were measured by flash-induced variable fluorescence, thermoluminescence, and low-temperature electron paramagnetic resonance (EPR) spectroscopy in thylakoid membranes isolated from wild type and psb33 mutant plants. Fluorescence decay kinetics were measured using an FL3000 dual modulation kinetic fluorometer, and the thermoluminescence signal was measured using a TL200/PMT thermoluminescence system (both from Photon System Instruments, Brno, Czech Republic) as described in Volgusheva et al. (2016).
Low-temperature EPR was measured in an X-band ELEXSYS 500 spectrometer using a SuperX EPR049 microwave bridge and a Bruker 4122SHQE super high-Q cavity (Bruker BioSpin). The system was fitted with a 900-cryostat and an ITC-503 temperature controller from Oxford Instruments Ltd. The S 2 state multiline signal was induced by illumination at 200 K for 6 min, and full oxidation of cytochrome b 559 was induced by illumination at 77 K as described in Chen et al. (2011). Processing and analysis of EPR spectra were performed with the Bruker Xepr program.

Statistical analyses
Data were processed with Microsoft Excel 2010 with the add-in Daniel's XL Toolbox (https://www.xltoolbox.net/) or with IBM SPSS Statistics version 25 (IBM Corp., Armonk, NY, USA). The normality of data was tested using the Shapiro-Wilk method. Non-normally distributed data were log-transformed before parametric tests were performed. Results were considered significant if two-tailed P<0.05.

psb33 mutant plants show a conditional light quality phenotype
We set out to investigate if the lack of a slow-growth phenotype of psb33 in our new plant growth facility might be attributed to specific qualitative properties of the fluorescent lights in these chambers (see Supplementary Fig. S1). Therefore, plants were grown under fluorescent white light at a total photon flux density of 120 µmol photons m −2 s −1 (PFD) for 2 weeks and then transferred to blue (400, 420, 440 nm), red (630, 660 nm), or a combination of blue and red LED lights for an additional 3 weeks (Fig. 1). Wild type and control plants were maintained in the fluorescent white light throughout the growth period. The total light intensity was kept constant at 120 µmol m −2 s −1 in all the different spectral conditions (Fig. 1, left panel). To test if the growth phenotype of psb33 was associated with its inability to induce state transition, the double mutant stn7stn8 was included as a negative control in the experiment (Tikkanen et al., 2010). After 2 weeks of growth under fluorescent white light, before transfer to LED lights, there was no apparent difference in size between the wild type, psb33, and stn7snt8 ( Supplementary Fig. S2).
The stn7stn8 plants displayed an approximately 20% reduction in fresh weight compared with wild type after 5 weeks in all light conditions except in red light where they grew to a similar size (Fig. 1). psb33 plants, on the other hand, showed similar fresh weight to wild type in fluorescent white and red light, while a dramatic and statistically significant (oneway ANOVA, P<0.001) decrease was observed in blue light. The psb33 plants grown under a combination of red and blue light displayed an intermediate reduction in fresh weight compared with those cultivated in white or blue light. The growth phenotype of psb33 in the blue light did not correlate with changes in chlorophyll or carotenoid content (Table 1).
These indicate that the slow-growth phenotype of the psb33 mutant is related to sensitivity towards blue light. Further, the phenotype appeared not to be a consequence of impaired state transition, since plants lacking the kinases STN7 and STN8 (stn7stn8 plants) did not respond similarly to psb33 to the tested light regimes.

Chlorophyll fluorescence characteristics of psb33 in blue and red light
Next, we determined the photosynthetic performance of plants grown under white, blue, and red or a combination of blue and red light ( Fig. 2A). The maximum quantum efficiency of photosystem II (F v /F m ) was similar between the control plants (wild type and the stn7stn8) and showed only small variations in the different light conditions. Following our previous results (Fristedt et al., 2015), F v /F m in psb33 was slightly, yet statistically significantly (one-way ANOVA, P<0.001) lower than in wild type in white light. However, in blue, red, and combined blue/red light, the F v /F m values were reduced by approximately 20% in psb33, indicating decreased PSII efficiency in these conditions ( Fig. 2A). In repeated experiments, F v /F m was consistently lower in psb33 plants grown in blue light compared with plants grown in red light. However, this difference was not statistically significant in all experiments (data not shown). The performance index parameter (PI ABS ) is an overall parameter of plant vitality and a sensitive index to evaluate stress (Kalaji et al., 2014). PI ABS takes into account the concentration of active reaction centers and the force of the light and dark reactions (Strasser et al., 2000). Although there was some variability between replicates, stn7stn8 plants largely displayed similar PI ABS to the wild type in all light conditions (see Supplementary Fig. S3). However, psb33 plants grown in blue, red, and the combination of blue and red light displayed a decrease in PI ABS in comparison to psb33 plants grown in fluorescent white light.
To further explore the response of psb33 to blue and red light. Plants grown in fluorescent white light were transferred to blue and red light, and F v /F m was determined at intervals for up to 3 d (Fig. 2B, C). Recovery of the plants was assessed by monitoring the chlorophyll fluorescence after the transfer of plants back to white light. Both blue and red light induced a fast decrease of the F v /F m ratio, which reached similar values as in the long-term acclimated plants after 3 d. Following 3 d of recovery in white light, F v /F m completely reverted to steadystate values again, showing that these effects were reversible (Fig. 2B, C).  Chlorophyll fluorescence phenotype of psb33 is not found in other related photosynthetic mutants To gain insight into the mechanism behind the strong response of the psb33 plants to blue and red light, we compared the performance of psb33 under these light treatments to that of well-characterized photosynthesis mutants. Specifically, mutants lacking components of the photosynthetic machinery associated with excitation energy transfer were selected and compared with psb33 (Fig. 2D, E). The kinases STN7 and STN8, and phosphatase TAP38 regulate the phosphorylation status of PSII core and LHCII proteins (Bellafiore et al., 2005;Bonardi et al., 2005;Pribil et al., 2010;Shapiguzov et al., 2010). The stn7, stn8, te stn7stn8, and tap38 mutants have reduced state transition capacity and differ in sensitivity to high or fluctuating light (Fristedt et al., 2009;Tikkanen et al., 2010). The lhcb1 and the lhcb2 mutants, corresponding to plants deficient in Lhcb1 and Lhcb2, respectively, show altered antenna and thylakoid structure, impaired state transition, and reduced NPQ (Pietrzykowska et al., 2014). The npq4 mutant is devoid of the protein PsbS and lacks the qE component of NPQ; it is sensitive to photoinhibition (Li et al., 2002) and fluctuating light (Külheim et al., 2002). PsbS has also been proposed to play a role in energy distribution between the two photosystems (Jajoo et al., 2014). PsbW plays an important role in PSII function, and the corresponding psbw mutant shows destabilized PSII supercomplex organization, decreased phosphorylation of PSII core proteins, and faster changes of the redox state of the plastoquinone pool (García-Cerdán et al., 2011). The psal mutant lacks a PSI component, shows reduced PSI electron transport rate, and has normal phosphorylation of LHCII but an inability to increase the PSI antenna size in state II light (Lunde et al., 2000). Finally, PGR5 encodes a protein previously thought to be involved in cyclic electron transport around PSI (Munekage et al., 2002), but more recently shown to be involved in regulating proton flux across the thylakoid membrane Kanazawa et al., 2017). The pgr5 mutant is unable to induce NPQ, is sensitive to high light, which results in PSI damage (Munekage et al., 2002), shows stunted growth in normal conditions, and is unable to cope with fluctuating light resulting in wilting (Tikkanen et al., 2010). Comparing the F v /F m parameter of all these mutant lines and wild types in response to blue and red light, it is clear that the psb33 mutant was the only line showing a substantial reduction in photosynthetic performance (Fig. 2D, E). There was no statistical difference in F v /F m among the psb33 plants kept for 3 or 7 d in blue or red light. These results indicated that the observed phenotype of psb33 is not connected to thylakoid ΔpH-dependent regulation mechanisms since pgr5 and npq4 were unaffected under the tested light regimes. The results further strengthened the notion that the phenotype of psb33 cannot be explained by the lack of photosystem energy distribution-dependent mechanisms since none of the included mutants with impaired ability to phosphorylate PSII core or with altered antenna proteins displayed stress symptoms under red or blue light.

psb33 plants display altered phytohormone levels in blue and red light
Plant growth restrictions are commonly associated with defective hormone control. Therefore, we tested if the response of psb33 plants to blue and red light was accompanied by changes in leaf phytohormone levels (Fig. 3). Our analysis specifically targeted hormones involved in growth and stress signaling: jasmonates (including JA, isoleucine-conjugated JA (JA-Ile), and 12-oxo-phytodienoic acid (cis-OPDA)), salicylic acid (SA), indole acetic acid (IAA), and IAA derivatives. SA levels were 3.6 and 5.5 times lower (P<0.01, one-way ANOVA, Bonferroni-Holm post hoc test) in the mutant compared with wild type in blue and red light, respectively. However, levels were relatively low in both the wild type and psb33 in all conditions (Fig. 3F). The only statistically significant hormonal change exclusively apparent in psb33 grown in blue light was a strong accumulation of aspartic acid-conjugated IAA (IAA-Asp) (Fig. 3C).

Blue light induces changes to psb33 thylakoid ultrastructure
To further explore the psb33 growth phenotype and to resolve whether it relates to morphological changes at the microscopic level, the thylakoid ultrastructure of psb33 plants grown in white, blue, and red light was examined by transmission electron microscopy (Fig. 4). No difference in thylakoid structure was observed between wild type and psb33 plants grown in fluorescent white or red light. Thylakoids of psb33 grown in blue light were found to be swollen, showing large luminal spaces at non-appressed regions of the thylakoid membrane (Fig. 4B). Such swollen thylakoids could be observed in a few chloroplasts also from wild type plants taken from blue light. However, virtually all thylakoids from psb33 plants from blue light displayed such abnormal structures.

UV-A light causes the psb33 growth phenotype
Next, we set out to more specifically delineate the light wavelengths causing the deleterious growth phenotype of psb33 in blue light. Three blue light conditions provided by three different LEDs were tested at the PFD of 120 µmol m −2 s −1 : 400, 420/440, and 400/420/440 nm (Fig. 5). psb33 plants cultivated under only 400 nm light only reached 12% of wild type fresh weight at the end of the growth period (Fig. 5A, B), while psb33 plants in 420/440 nm and 400/420/450 nm light reached 66% and 40%, respectively, of the weight of wild type.
Due to the fact that the 400 nm LED lamp emits light ranging from 391 to 421 nm, with a peak at 404-405 nm, we suspected that light in the UV-A region was the main cause of the dwarf phenotype of psb33 grown in blue light (Figs 1, 5; Supplementary Fig. S4). Growing plants in 385 nm UV-A light corroborated this finding: psb33 plants displayed strong growth restriction compared with the wild type in this light condition ( Supplementary Fig. S5). Chlorophyll a fluorescence measurements showed a drastic reduction in quantum efficiency (F v /F m ) of PSII and PI ABS in psb33 plants grown in lights with UV-A ( Fig. 5C; Supplementary Fig. S5). The F v /F m ratio of psb33 plants grown in 420/450 nm light was significantly higher (one-way ANOVA, P<0.01) than in plants grown in 400 nm or 400/420/440 nm light (Fig. 5C). Thus, plants cultivated in light conditions lacking UV-A wavelengths performed better in terms of photosynthetic performance than plants grown in light containing UV-A.

Photosystem II core protein levels are reduced in psb33
The low F v /F m measured in the psb33 mutant indicated decreased PSII efficiency and possibly damage to the PSII complex. We therefore examined potential photodamage to PSII in psb33 grown in various blue light conditions. To investigate if PSII and PSI protein levels were influenced by the quality of light, whole-leaf proteins were extracted from plants grown in either fluorescent white light or blue light with or without the 400 nm UV-A component (see Supplementary  Fig. S4; 400/420/440 or 420/440 nm, respectively), and red light. Further, the fresh weight of plants was determined, and whole leaf proteins extracted from wild type and psb33 plants grown under blue and red light in combination with far-red light (400/420/440/730 or 630/660/730 nm, respectively; Supplementary Fig. S6). Far-red light was included in the experimental set-up since it predominantly excites PSI. The farred light treatment could thus reveal if increased excitation pressure of PSI would balance the relative energy distribution between the two photosystems and reduce photodamage. Extracted whole leaf proteins were separated on SDS-PAGE, immunoblotted, and labeled with antibodies specific for PSII (D1 and D2), PSI (PsaA and PsaB), LHCII (Lhcb1 and Lhcb2) and its phosphorylated forms (Lhcb1-P and Lhcb2-P) and Rubisco large subunit (RbcL) (Supplementary Fig. S7A). PsaA and PsaB levels varied in the different light conditions but were largely similar between wild type and psb33. So was the abundance of RbcL. In agreement with the impaired state transition phenotype of psb33, phosphorylation of Lhcb1 and Lhcb2 was almost completely lost in the psb33 mutant in blue light. Notably, in red light and light supplemented with far-red, Lhcbs were phosphorylated to a certain level in psb33 plants. Furthermore, very little phosphorylation of Lhcb proteins was observed in psb33 in fluorescent white light, corroborating our previous findings (Fristedt et al., 2017). Our compiled data thus indicated that the redistribution of energy between the two photosystems through rearrangement of the LHCII lightharvesting antenna is impaired in psb33. Yet, this deficiency cannot alone explain the blue light growth phenotype. The PSII core proteins D1 and D2 were both reduced in psb33 in blue light conditions. The corresponding reduction in protein abundance occurred in psb33 exposed to 400 nm LED light, whereas plants exposed to 420/440 nm light displayed higher levels of D1 than plants from only 400 nm or 400/420/440 nm light (Fig. 6A, B).

Photosystem II electron transport properties are modified in psb33
The reduced levels of the D1 and D2 proteins in psb33 cultivated under UV-A light may indicate an altered redox state of the acceptor side of PSII. To investigate this, we performed flash-induced variable fluorescence decay and thermoluminescence measurements in isolated PSII membranes from wild type and psb33 plants. Analysis of the flash-induced fluorescence decay kinetics demonstrated faster electron transfer from Q A − to Q B in psb33 mutant (Fig. 7A, left panel) (Vass et al., 1999;Mamedov et al., 2000;Roose et al., 2010). Also, thermoluminescence measurements, which are a useful complement to the flash-induced fluorescence decay measurements (Vass and Govindjee, 1996;Ducruet and Vass, 2009), also showed the presence of only a slightly up-shifted B-band (which represents the Q B − →S 2 state recombination) at 38 °C. This is different compared with a mixture of the Q-band (which represents the Q A − →S 2 state recombination) and B-band, present in thylakoids from wild type at 14 and 36 °C (Fig. 7B, left panel). Thus, both measurements confirm the presence of a more oxidized acceptor side of PSII and a higher redox potential of Q B in psb33 mutant.
The fluorescence decay kinetics and thermoluminescence bands measured in the presence of 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) (and in the case of thermoluminescence indicating only the Q A − →S 2 state recombination) were virtually the same in the wild type and psb33 mutant ( Fig. 7A and B, respectively, right panel). This indicated that the redox properties of Q A and the water oxidizing complex on the donor side of PSII were not affected in the psb33 mutant. This was also corroborated by EPR spectroscopy measurements of the S 2 state multiline signal, which were found to be similar in wild type and psb33 mutant (Fig. 7D).
Interestingly, cytochrome b 559 , an integral part and an alternative electron donor/acceptor of the PSII reaction center (Shinopoulos and Brudvig, 2012) was found to be predominantly in the low potential form in the psb33 mutant (Fig. 7E) again, indicating a more oxidized acceptor side in PSII. However, the overall amount of the PSII reaction centers, as judged from the tyrosine D radical on a chlorophyll basis, was only slightly diminished in the psb33 mutant (Fig. 7C).

psb33 reactive oxygen species levels are unaffected by UV-A light
The redox potential of Q B is important for the regulation of electron transfer for both forward and backward electron transfer, which functions as a protective mechanism for PSII (Kato et al., 2016). The modulated potential of Q B Fig. 4. Thylakoid ultrastructure of psb33 and wild type plants grown in different light conditions. Representative transmission electron micrographs are shown for leaf chloroplasts from 5-week-old wild type (WT) and psb33 mutant plants fixed 2 h after the onset of illumination. The thylakoids of the wild type and psb33 mutant are similar when grown in fluorescent white light or under LED emitting red light at the wavelength of 630/660 nm. When grown under LED emitting blue light (400, 420, 450 nm), psb33 displays swollen thylakoid, rarely observed in the wild type. Scale bars, in combination with elevated excitation pressure on PSII via UV-A light, could induce ROS-mediated damage to D1 and D2 in psb33 mutant plants. To test this, we investigated if psb33 mutant plants differed in ROS levels compared with wild type plants in UV-A light. Plants grown in 400 or 420/450 nm light were stained for hydrogen peroxide using 3,3′-diaminobenzidine and superoxide (O 2 · − ) radicals using nitroblue tetrazolium (NBT). Surprisingly, no visual differences in levels of either H 2 O 2 by DAB or O 2 · − by NBT were observed between wild type and psb33 (Fig. 8).

Discussion
The mechanisms by which plants maintain and optimize photosynthesis in fluctuating light environments are rapidly being unraveled (Tikkanen et al., 2010(Tikkanen et al., , 2012Cruz et al., 2016;Vialet-Chabrand et al., 2017). But as natural light fluctuates in intensity, it also varies considerably in quality, and the influence of light color in photosynthetic regulation has so far largely been overlooked (Allorent and Petroutsos, 2017). In this study, we demonstrate that plants lacking PSB33 are particularly sensitive to light in the UV-A spectral region. UV-A light was found to induce a dwarf phenotype in psb33 plants, which was associated with diminished PSII protein levels.

PSB33 has a specific spectral influence on photosystem II and linear electron flow
Since the growth phenotype of the psb33 mutant appeared to be very variable and strictly dependent on light conditions in our different growth facilities, we decided to place particular emphasis on the light quality in inducing the psb33 phenotype in this work. By cultivating the mutant in different light settings, it was found that the reduced growth of psb33 was specifically induced by blue light (Fig. 1; Supplementary Fig. S3), particularly the spectral region that comprises the UV-A (Fig. 5). We therefore suggest that the previously reported growth phenotype, reaching approximately 50% the size of the wild type under normal growth conditions (Fristedt et al., 2015), was due to the presence of short-wavelength blue light in the light source of the growth chambers.
The slow-growth of psb33 in blue light was found to be connected to low levels of PSII core proteins D1 and D2 (Fig. 6), which inevitably leads to reduced photosynthesis rate. The effect was more pronounced in UV-A (400 nm) blue light than in 420/450 or 400/420/450 nm blue light at the same total photon flux (Fig. 6B). The red light did not cause any difference in D1 or D2 protein abundance between wild type and mutant. It is interesting to note that both blue and red lightgrown plants displayed a similar reduction in maximal PSII quantum yield (F v /F m ) (Fig. 2) and PI ABS (see Supplementary  Fig. S3). However, psb33 plants grown in exclusively blue light at 400 nm showed further reduced F v /F m values (Fig. 5C).
Supplementing UV-A containing blue light with far-red light rescued the phenotype to some extent in terms of both growth and PSII protein levels ( Fig. 6A; Supplementary Fig.  S6). As far-red light predominately excites PSI, a possible explanation for this rescue is that PSB33 has a role in the linear electron transport chain to balance the redox components between the two photosystems. In the absence of PSB33, an unbalance of redox components in the electron transport chain would initiate the production of ROS, resulting in elevated PSII damage. However, we did not observe an increase of ROS in leaves from plants grown in UV-A light that would support an overexcitation of PSII and consequently increased photodamage. Another, perhaps more likely explanation is that the rescue effect of far-red is a consequence of reduced UV-A blue light as the total photon flux was kept constant in the experiments (i.e. 50% of the total light was provided by far-red and 50% by blue light).
When we tested mutants with reported impaired ability for PSII superstructure (psbw), state transitions (lhcb1, lchb2, psal, and stn7/8), cyclic electron transport and proton gradient (pgr5), or NPQ (npq4), none exhibited a slowgrowth phenotype comparable to that of psb33 in blue light, indicating that PSB33 has a unique role in blue light perception unrelated to these known pathways to maintain photosynthesis.
Saliccylic acid and aspartic acid-conjugated indole-3-acetic acid hormone levels differ in psb33 plants exposed to different lights SA levels are well known to influence plant growth negatively and to be important for light acclimation (Rivas-San Vicente and Plasencia, 2011). Surprisingly, the SA level decreased in response to both blue and red light treatments in psb33, whereas it increased in the wild type (Fig. 3F). SA and glutathione interplay to regulate antioxidant response through ROS scavenging (Herrera-Vásquez et al., 2015). The low levels of SA in psb33 could accordingly explain the accumulation of different ROS in the mutant line, as previously reported (Fristedt et al., 2015). However, we could not observe any differences between wild type and psb33 in or the absence of UV-A light (400 and 420/450 nm, respectively) in terms of H 2 O 2 and superoxide radicals (Fig. 8), suggesting that changed SA and ROS levels did not contribute to the deleterious effect of UV-A light. The only hormonal change exclusively observed in psb33 grown in blue light was a strong accumulation of aspartic acid-conjugated IAA (IAA-Asp) (Fig. 3C). This is believed to be an inactive form of IAA that is destined for catabolism (Woodward and Bartel, 2005;Ludwig-Müller, 2011).

psb33 plants grown in blue light show an abnormal thylakoid ultrastructure
PSB33 is primarily located in non-appressed thylakoid regions (Fristedt et al., 2015;Fristedt et al., 2017;Kato et al., 2017). Quality control of PSII is dependent on reversible thylakoid structure dynamics (Yamamoto, 2016). Light induces a reversible swelling and unstacking of the thylakoids, which is necessary for proper PSII repair and protects PSII from damage (Khatoon et al., 2009). Furthermore, the increased surface area of the membranes influences its fluidity, which is important for lateral movement of proteins and other compounds within the membrane regulating NPQ, state transition, and PSII repair (Yamamoto, 2016). Integrated regulation of the many processes involved in the dynamics of the thylakoid membrane is not yet fully understood . It has been shown that the PSII-LHCII supercomplex phosphorylation level affects the distance between the grana stacks as well as their size, and it been suggested that complexes in opposite orientation repel each other, causing unstacking of the grana (Fristedt et al., 2009). Thus, the reduced levels of PSII (D1 and D2) observed in psb33 mutant plants exposed to blue light (Fig. 6) could explain changes to the thylakoid structure of psb33, seen by transmission electron microscopy (Fig. 4). The distinctly swollen stroma lamella might negatively influence the rate of movement of proteins and compounds, restrict PSII quality control, and enhance damage to the complex (Kirchhoff, 2014). The altered PSII superstructure and PSII level in psb33 might have further enhanced the growth phenotype.

UV-A light reduces photosystem II proteins in psb33
Two previous observations have been linked to blue and UV-A light that can explain psb33 dwarf phenotype. First, the Mn 4 CaO 5 can directly absorb UV light and initiate the photoinactivation process (Ohnishi et al., 2005). It is possible, but not likely, that PSB33 protects the Mn cluster, due to restricted access by the (C-E) EPR spectroscopy measurements of the tyrosine D radical (C), the S 2 state of the water oxidizing complex (D), and the oxidized form of cytochrome b 559 in the g z region (E) before and after illumination at 77 K. The low and high potential of cytochrome b 559 are marked as HP and LP, respectively. EPR conditions: microwave frequency: 9.35 GHz, microwave power 1.3 μW, and temperature 15 K for (C), 10 mW and 7 K for (D), and 5 mW and 15 K for (E). (This figure is available in color at JXB online.) OEC proteins. However, our data on intra-PSII electron flow do not support this hypothesis (Fig. 7). Second, blue and UV-A light-mediated signals have been linked to increased D2 and CP43 gene expression in mature chloroplasts (Christopher and Mullet, 1994) and this is consistent with our results: we observed an up-regulation of D1 and D2 PSII proteins in wild type plants exposed to blue/UV-A light (Fig. 6A). In contrast, D1 and D2 were down-regulated in psb33 mutant plants in blue/UV-A.
PSB33's location is largely in non-appressed thylakoid regions, with its head domain exposed to the stroma (Fristedt et al., 2015), and it is part of a larger protein complex (Fristedt et al., 2017). PSB33 is likely part of a signaling mechanism involved in the rejuvenation of PSII. A hypothetical scenario would be that this mechanism is either triggered by UV-A light or develops as a consequence of plant exposure to blue and UV light, possibly as a result of re-adjusted redox regulation. Such a signaling cascade may eventually induce a conformational change on PSB33, which triggers downstream signaling for transcription of various PSII proteins.

Conclusions
The accumulated data show that PSB33 has an important role in sustaining a functional PSII pool in blue light, especially in UV-A light. Although the severe growth phenotype of psb33 plants in blue light is connected to misregulated phosphorylation and migration of Lhcb proteins, this appears not to be a causal relationship. The exact molecular mechanism of how PSB33 protein senses UV-light and initiates the photosynthetic response remains to be investigated.

Supplementary data
Supplementary data are available at JXB online.