D1 Degradation by Chloroplast Proteases

Photosynthetic organisms have evolved the so-called PSII repair cycle, in which a reaction center protein D1 is degraded rapidly in a specific manner. Two proteases that respectively perform processive or endopeptidic degradation, FtsH and Deg, participate in this cycle. To examine the cooperative D1 degradation by these proteases, we engaged mutants lacking FtsH2 ( var2 ) and Deg5/Deg8 ( deg5 deg8 ) in detecting D1 cleaved fragments. We detected several D1 fragments only under var2 background, using N-terminal or C-terminal specific antibodies of D1. The appearance of these D1 fragments was inhibited by serine protease inhibitor and by deg5 deg8 mutations. Given the localization of Deg5/Deg8 on the luminal side of thylakoid membranes, we inferred that Deg5/Deg8 cleaves D1 at its CD luminal loop, and that the cleaved products of D1 are the substrate for FtsH. These D1 fragments detected in var2 were associated with PSII monomer, dimer, and partial disassembly complex, but not with PSII supercomplexes. It is particularly interesting that another processive protease, Clp, was upregulated and appeared to be recruited from stroma to the thylakoid membrane in var2 , suggesting compensation for FtsH deficiency. Together, our data demonstrate in vivo cooperative degradation of D1, in which Deg cleavage assists FtsH processive degradation under photoinhibitory conditions. Healthcare) for D1 and D1 degradation fragments, and the ECL Western Blotting Detection Kit (GE Healthcare) for other proteins, and recorded with a LAS100-mini system (Fuji Photo Film Co. Ltd.). To improve immunoblot sensitivity, the cleavage products of the D1 protein were redetected after cutting off the membrane area above 25-kD. Experiments were repeated more than three times. Representative results are shown. The anti-D1 (C-term) antibody was designed to recognize the C-terminus (amino acids 303–315 and 329–344) of Arabidopsis D1 polypeptide. Synthetic peptide cocktails were used to raise antibodies in rabbit (Operon Biotechnology Co. Ltd.).

6 unfolding and translocation of membrane proteins (Ogura and Wilkinson, 2001;Ito and Akiyama, 2005;Wagner et al., 2012). It is clear that FtsH plays an important role in D1 degradation in both cyanobacteria and chloroplasts (Bailey et al., 2002;Silva et al., 2003;Komenda et al., 2006;Kato et al., 2009). Of 12 FtsH homologues in Arabidopsis, nine were found to be located in chloroplasts (Sakamoto et al., 2003). A major FtsH complex in chloroplasts is localized to thylakoid membranes as a hetero-hexamer complex, with their catalytic region facing the stromal side of the membrane (Lindahl et al., 1996;Sakamoto et al., 2003;Yu et al., 2004). Mounting evidence has demonstrated that FtsH1, FtsH2, FtsH5, and FtsH8 are four major isomers of the FtsH complex that is localized in the thylakoid membrane through one N-terminal transmembrane domain (Yu et al., 2004;Rodrigues et al., 2011). FtsH2 is the most abundant isomer, followed by FtsH5, FtsH8, and FtsH1 (Sinvany-Villalobo et al., 2004). Mutants respectively lacking FtsH2 and FtsH5, yellow variegated2 (var2) and var1, show a leaf-variegated phenotype, which is more enhanced in var2 (Chen et al., 2000;Takechi et al., 2000;Sakamoto et al., 2002). In chloroplast, FtsH hetero-complexes are formed by at least two-type isomers (A and B, represented by FtsH1/5 and FtsH2/8, respectively) that are functionally distinguishable with each other (Sakamoto et al., 2003;Yu et al., 2004Yu et al., , 2005Zaltsman et al., 2005a). The loss of the two isomers from either type engenders seedling lethality with incomplete chloroplast development (Zaltsman et al., 2005a). Thus, although var2 and var1 show clear phenotype, the mutants still have a certain level of the FtsH complex (Sakamoto et al., 2003;Zaltsman et al., 2005b). One notable feature in var1 and var2 mutants, in addition to their variegated phenotype, is their high vulnerability to photoinhibition under strong illumination (Sakamoto et al., 2002;Sakamoto et al., 2004). Furthermore, in vivo assessment of D1 degradation activity in these mutants clearly demonstrates that FtsH participates in the PSII repair not only under photoinhibitory but also non-photoinhibitory conditions (Kato et al., 2009).
Deg protease in bacteria is the periplasmic ATP-independent serine-type endoprotease. Most Deg family members contain more than one PDZ-domain, which is necessary for the formation of functional oligomeric complexes (Clausen et al., 2002). Several Deg proteases have been shown to affect D1 degradation in chloroplasts (Haussuhl et al., 2001;Kapri-Pardes et al., 2007;Sun et al., 2007;Sun et al., 2010a), although their function in the PSII repair seems to be less important in cyanobacteria (Barker et al., 2006). Of 16 Degs identified in Arabidopsis, five (Deg1, Deg2, Deg5, Deg7, and Deg8) have been reported as peripherally attached to the thylakoid membrane of chloroplasts: Deg1, Deg5, and Deg8 are localized on the lumenal side; Deg2 and Deg7 7 are localized on the stromal side (Huesgen et al., 2009;Schuhmann and Adamska, 2012). Initially, involvement of Deg2 in the cleavage between helices D and E of the D1 (DE loop) was proposed by in vitro studies conducted in Arabidopsis (Haussuhl et al., 2001). However, the rate of D1 degradation in deg2 mutants is comparable to that in wild type under light stress conditions (Huesgen et al., 2006). A recent report described that Deg7 participates in the cleavage of PSII core proteins including the damaged D1 and that it contributes the efficient PSII repair under photoinhibitory conditions (Sun et al., 2010a). Of the lumenal Degs, Deg5 and Deg8 are involved in the cleavage within the CD loop of the damaged D1. High-light-sensitive phenotypes in deg5 and deg8 mutants were shown to be enhanced in deg5 deg8 double mutants, suggesting the synergistic function of Deg5 and Deg8 in the PSII repair (Sun et al., 2007). The other lumenal Deg protease, Deg1, seems to participate in the cleavage of D1 protein at the CD loop and the downstream of transmembrane helix E (Kapri-Pardes et al., 2007). In addition, Deg1 appears to be fundamentally important for chloroplast development because Deg1 homozygous knockout lines were unobtainable (Kapri-Pardes et al., 2007;Sun et al., 2010b). Deg1 knockdown mutants show impaired plant growth compared with the wild type, even under non-photoinhibitory growth conditions. These knockdown lines caused a concomitant reduction of FtsH and Deg2.
Based on numerous studies described previously and proteolytic properties of FtsH (processive) and Deg (endopeptidic), a model in which Deg proteases have a supplementary role that increases the recognition site for FtsH in the D1 degradation has been proposed (Itzhaki et al., 1998;Kato and Sakamoto, 2009). For example, an earlier biochemical experiment shows that a purified recombinant FtsH can degrade a high-light-induced 23-kD D1 fragment in an ATP-dependent manner (Lindahl et al., 2000). This observation suggests that a partial D1 fragment, possibly generated by Deg, can be degraded by FtsH. However, the in vivo evidence to support cooperative D1 degradation mediated by FtsH and Deg is lacking. It should be examined using mutant analysis.
To address this question in this study, we assessed D1 degradation in var2 and deg5 deg8 mutants. Results showed that the several cleavage products of the D1 under photoinhibitory conditions accumulated in var2, which implies that FtsH is indeed required for in vivo degradation of D1 cleavage products. A line of evidence was provided that the accumulation of several D1 cleavage products in var2 depends on Deg5 and Deg8. These results supported our model showing that FtsH plays a fundamental role in D1 degradation and that Degs supplement it under photoinhibitory conditions.

In vivo D1 Degradation Assay in deg5 deg8 under Non-photoinhibitory Light Conditions
In this study, we specifically examined lumenal Deg5 and Deg8 and investigated their cooperative role in D1 degradation with FtsH. Involvement of these Degs in the PSII repair has been demonstrated previously (Sun et al., 2007), but not under low-light conditions. To evaluate the rates of D1 degradation in high-light, growth-light, and low-light conditions (1,200,180, and 20 µmol photons m -2 s -1 respectively), detached leaves from the wild-type Columbia (Col) and deg5 deg8 double mutant were incubated with lincomycin. Immunoblot analysis of D1 protein levels was performed, and the rate of D1 degradation was estimated based on the ratio of D1 signal to CBB stained light-harvesting complexes of PSII (LHCII). The results as presented in Fig. 1 indicate that D1 degradation rates under growth-and low-light conditions were similar between wild type and deg5 deg8, although the D1 degradation in deg5 deg8 was significantly slower than that in wild type when plants were exposed to high-light ( Fig. 1). Together with results of the previous study (Sun et al., 2007), these results indicated that Deg5 and Deg8 do not contribute significantly to D1 degradation under non-photoinhibitory conditions.

Accumulation of D1 Cleavage Products in var2 under Photoinhibitory Light Conditions
To examine D1 degradation mediated by FtsH and Deg further, it is important to assess D1 partial degradation products. Given the role of each protease (Deg in endopeptidic cleavage and FtsH in processive digestion), we assumed that var2 contains more D1 fragments that are generated by Deg. Our previous attempt to detect fragmented D1 using immunoblotting was unsuccessful, probably because of i) insufficient high-light intensity to cause photoinhibition, and of ii) our limited sensitivity in immunoblots. When we used extreme high light (2,500 µmol photons m -2 s -1 ) and improved immunoblot sensitivity (see Materials and Methods), however, several D1 fragments were detectable (Fig. 2). Specific antibodies that recognize the N-terminus, DE loop, and C-terminus of D1 protein (abbreviated as N-term, DE-loop, and C-term, respectively) were employed to detect D1 fragments. To allow detection of the fragments, which were regarded as much less abundant than full-length D1, the part of the blots corresponding to full length D1 (ca. 32 kDa) was discarded before immunodetection. We therefore minimized the background resulting from the signal corresponding to the full-length D1 protein. After exposure to extreme high light, wild type and var2 leaves showed rapid decreases in PSII activity that was 9 measured by maximum quantum yield of PSII (Fv/Fm); approximately 50% and 30% of the maximum PSII efficiency were lost, respectively, after 1-h exposure in wild type and var2 (Supplemental Table S1). Under this experimental condition, immunoblot analysis using anti-D1 (N-term) antibodies showed that a band of 18-kD accumulated significantly in var2. Similarly, immunoblot analysis using anti-D1 (C-term) antibodies showed that two bands of 12-kD and 16-kD accumulated in var2. These products were only slightly detectable in wild type under the high light condition ( Fig. 2A). We also used anti-D1 (DE-loop) antibodies that recognize DE loop of D1 proteins, but failed to detect any specific cleavage products that had been crossreacted (Supplemental Figure S2).
The other important observation was that although D1 degradation is impaired in var2 not only under high light but in non-photoinhibitory conditions (Kato et al., 2009), the D1 cleavage products detected by N-term and C-term D1 antibodies did not accumulate under normal-light conditions ( Fig. 2A). Overall, these results demonstrated that FtsH participates in the degradation process of the high-light-induced D1 cleavage products.
To estimate whether the cleavages are dependent on serine protease activity, we compared the accumulation of the D1 cleavage products in the presence of a serine protease inhibitor. Following pretreatment of var2 leaves with 4-(2-aminoethyl)-benzenesulfonyl fluoride (AEBSF) protease inhibitor, leaves were incubated for 1 h under extreme high light. Immunoblot results obtained using anti-D1 (N-term) and anti-D1 (C-term) antibodies showed that levels of the cleavage products decreased markedly in var2 in the presence of AEBSF (Fig. 2B). The result demonstrated that the D1 cleavage products that accumulated in var2 leaves under a high-light condition were likely caused by serine protease.

Characterization of var2 deg5 deg8 Triple Mutant
To assess the involvement of FtsH and Deg proteases in D1 degradation in a cooperative fashion in vivo, we generated a var2 deg5 deg8 triple mutant. As depicted in Fig. 3, var2 deg5 deg8 showed a variegated phenotype similar to that of var2: we observed no distinguishable leaf variegation phenotype and growth rate between var2 single and the triple mutants. In addition, we considered that deg5 and deg8 did not affect FtsH accumulation, because the triple mutant had levels of FtsH isomers comparable to var2 (Supplemental Figure S3).
These results suggest that Deg5 and Deg8 have little additional effect on thylakoid formation or maintenance, which has been proposed to be closely related to FtsH.
We next measured Fv/Fm in var2, deg5 deg8, var2 deg5 deg8, and wild type to assess whether additive PSII photoinhibition occurred in triple mutants. Detached leaves of var2, deg5 deg8, var2 deg5 deg8, and wild-type were exposed to high light for up to 4 h. The result that Fv/Fm values in var2 and deg5 deg8 mutants were lower than those in wild type was consistent with the previous reports (Sakamoto et al., 2002;Sun et al., 2007). Furthermore, Fv/Fm values in var2 deg5 deg8 were decreased significantly more than those in var2 and deg5 deg8 during high-light irradiation (Fig. 3B). These results demonstrate the synergistic effect of the reduced FtsH and Deg protease activities on photosensitivity to high light, suggesting increased photoinhibition in var2 deg5 deg8.

Cleavage Products of the D1 in var2 deg5 deg8
To assess whether the D1 cleavage products in var2 resulted from Deg proteases, we performed our D1 degradation assay using the triple mutant along with other control plants. The D1 cleavage products did not accumulate in wild type and mutants under normal-light conditions (Supplemental Figure S4). However, consistent with the result presented in Fig. 2, the 18-kD N-terminal D1 fragment and two C-terminal D1 fragments that corresponded to 16-kD and 12-kD were observed in var2 (Fig. 4). In contrast, these bands were at the undetectable level in deg5 deg8, implying that FtsH alone can degrade photodamaged D1 without Deg5 and Deg8. In the triple mutant, interestingly, the 18-kD N-terminal D1 fragment and only one of the two C-terminal fragments (16-kD) decreased in var2 deg5 deg8, although the 12-kD C-terminal fragment remained unchanged and was comparable between the triple mutant and var2 (Fig. 4). Given the fact that these D1 fragments are i) only detectable under var2 background and ii) likely generated by serine protease activity, the results suggest that the 18-kD N-terminal fragment and the 16-kD C-terminal fragment were generated by lumenal Deg5/Deg8. The 12-kD C-terminal fragment is likely to be generated by other Deg proteases. Taken together, we considered that the 18-kD N-terminal fragment and the 16-kD C-terminal fragment represented fragments cleaved at the lumenal CD loop of D1 protein, whereas the 12-kD C-terminal fragment represented a C-terminal fragment cleaved at the stromal DE loop of D1 protein.

D1 Cleavage Products in the PSII Complex Assessed by BN/SDS-PAGE
In chloroplasts, functional PSII core is a dimer that forms a large complex with LHCII antenna.
Degradation of photodamaged D1 mediated by FtsH and Deg is a critical step in the PSII repair cycle, which involves i) migration of the damaged PSII core from grana stacks to the stroma-exposed thylakoids, ii) partial disassembly of the PSII core, iii) D1 proteolysis as described previously, iv) synthesis and processing of D1 nascent chain, and v) migration of repaired PSII back to grana stacks (Baena-Gonzalez and Aro, 2002;Aro et al., 2005). Two-dimensional blue native (BN)/SDS-PAGE is a powerful approach to monitor the different status of PSII assembly/disassembly in the PSII repair cycle. Therefore, we performed BN/SDS-PAGE and subsequent immunoblots to detect D1 cleavage fragments. Purified thylakoid membranes were solubilized using n-dodecyl-ß-maltoside, and protein complexes were separated on the first dimension by 4% to 16% gradient native gel with subsequent the separation of protein subunits on second dimension with SDS-PAGE (Fig. 5).
Although the assembly of PSII complexes in wild type and var2 showed no great difference under growth-light conditions, PSII supercomplexes were only slightly detectable in var2 under high-light conditions. In addition, the level of PSII core complex lacking CP43 (termed RC47), which can be regarded as the PSII repair cycle intermediate at the dissociation step, increased in var2 under extreme high-light conditions. These results were consistent with those of our previous study (Kato et al., 2009), demonstrating compromised PSII repair with a reduced amount of FtsH under photoinhibitory light conditions. Immunoblot analysis using anti-D1 antibodies also supported these results: we detected the N-terminal D1 fragment and the two C-terminal D1 cleavage products, which co-migrated at the positions corresponding to PSII dimer, monomer, and RC47, except that the signal of 16-kD cleavage product was undetectable at positions corresponding to PSII dimer (Fig. 5). Importantly, even when the blots were overexposed, the specific signals of the anti-D1 (N-term) and anti-D1 (C-term) antibodies with the extreme high-light-treated thylakoid membrane did not appear at the positions corresponding to PSII supercomplexes in var2 and wild type.

Dynamics of Chloroplast Proteases in var2
Previous studies showed that D1 turnover occurred even in the depletion of major FtsH complex, which is comprised by FtsH2 and FtsH5 (Kato et al., 2009). This fact suggests that an alternative mechanism to degrade photodamaged D1 seems to act when the proteolytic activity proceeded by FtsH is limited. Assuming that other chloroplast proteases might compensate for FtsH deficiency, we examined the levels and thylakoid-membrane localization of other proteases in var2. Proteins isolated from purified chloroplast and from stromal and membrane fractions were subjected to immunoblot analysis using antibodies against various chloroplast protease www.plantphysiol.org on September 5, 2017 -Published by Downloaded from Copyright © 2012 American Society of Plant Biologists. All rights reserved.
( Fig. 6). Anti-D1 and Rubisco large subunit antibodies were used, respectively, as a control of thylakoid membrane protein and stroma protein. Results showed that levels of Clp protease were up-regulated in var2 because both ClpP6 (representing Clp protease complex) and ClpC1/C2 (representing ATP-dependent unfoldase complex) increased. It is particularly interesting that these subunits appeared to be recruited into thylakoid membranes when FtsH2 was deficient. SppA, which is closely associated with the stromal side of thylakoid membranes and enriched in stroma thylakoid, is known to be up-regulated by high light (Lensch et al., 2001).
The immunoblot analyses showed that the steady state level of SppA increased in var2 compared to wild type.
By contrast, levels of Deg2 and Deg8 in the thylakoid membranes of var2 and wild type were not significantly different. These results support the possibility that processive degradation of photodamaged D1, whether or not cleaved by Deg, proceeds with ATP-dependent proteases at the stromal surface of thylakoid membranes. FtsH and Clp are structurally and functionally related to one another (Kato and Sakamoto, 2010). Although FtsH is localized in thylakoid membranes and plays a major role in D1 degradation in the PSII repair cycle, Clp might participate in the PSII repair when FtsH is limited. We also examined the levels of other proteases in deg5 deg8.
In contrast to the results in var2, however, no significant changes between wild type and deg5 deg8 in the level and the localization of other chloroplast proteases were detected (Supplemental Figure S5).

DISCUSSION
Many studies of the degradation pattern of D1 under photoinhibitory conditions have been conducted because such studies can be expected to provide us with important clues to elucidate photoinhibition mechanisms.
Both in vivo and in vitro results have demonstrated two degradation patterns, proposing that primary D1 cleavage takes place in the stromal DE and the lumenal CD loops (Greenberg et al., 1987;De Las Rivas et al., 1992;Salter et al., 1992;Kettunen et al., 1996). However, over the last decade, considerable effort has revealed proteases responsible for D1 degradation under photoinhibitory conditions (Lindahl et al., 2000;Haussuhl et al., 2001;Bailey et al., 2002;Kapri-Pardes et al., 2007;Sun et al., 2007;Kato et al., 2009;Sun et al., 2010a 13 be the substrates for FtsH. First, our improved D1 degradation assay enabled us to detect several D1 cleavage products in var2 under excess light exposure. Second, inhibitor analysis showed that these fragments resulted from serine protease activities (Fig. 2). Finally, our attempt to detect D1 cleavage products in the var2 deg5 deg8 triple mutant demonstrated that the 18-kD N-terminal and 16-kD C-terminal cleavage products were produced by lumenal Deg proteases under photoinhibitory conditions (Fig. 4). Together, these results provide in vivo evidence that stromal FtsH and lumenal Deg cooperatively degrade photodamaged D1 in the PSII repair cycle.
To integrate our observations in this study into the current understanding in the PSII repair cycle in chloroplasts, we present a model that explains cooperative D1 digestion between FtsH and Deg proteases (Fig. 7).
In this model, two pathways for D1 degradation in chloroplasts can be assumed. One is constant processive D1 degradation mediated predominantly by the FtsH complexes, irrespective of light intensity. The other is an 'escape pathway', in which D1 degradation is conducted by multiple proteases under photoinhibitory condition.
Particularly, D1 subfragments generated by Deg accelerate D1 degradation proceeded by FtsH and/or other proteases, thereby facilitating efficient PSII repair. In this regard, it is noteworthy that we measured the rate of D1 degradation in the mutant lacking both FtsH2 and lumenal Deg5/Deg8 proteases, based on our previous non-variegated leaf-disc assay (Supplemental Figure S6)  14 suggested that Deg2 cleaves D1 at the DE loop (Supplemental Figure S1) (Haussuhl et al., 2001). We compared these cleavage events (particularly at the CD and DE loops) with the D1 subfragments detected in var2 background, based on their molecular size. First, we inferred that two subfragments, N-terminal 18-kD and C-terminal 16-kD fragments, correspond to the products from the cleavage at the CD loop. Second, the C-terminal 12-kD fragment corresponds to the product from the cleavage at the DE loop (schematically portrayed in Fig. 7). No N-terminal product from the cleavage at the DE loop was detected in our assay, probably because the cleavage at the CD loop is conducted very efficiently by lumenal Degs, or because the titer of anti-D1 (N-term) antibodies was too low to detect subfragments. Detection of the C-terminal 16-kD implies that some subfragments exist that escape from the cleavage at the DE loop, which might be operated by stromal Degs.
Given the fact that these subfragments accumulate only in var2 background and that they are generated by serine protease activity, we conclude that FtsH and Deg cooperatively degrade D1 in the PSII repair cycle: we therefore provide in vivo evidence that D1 fragments generated by Deg proteases can be degraded processively by FtsH. In addition, our results strongly imply that the cleavage at the CD loop is manipulated by Deg5 and Deg8. It is particularly interesting that BN/SDS-PAGE analysis showed that the D1 fragments generated by Deg proteases were only detectable in PSII dimer, PSII monomer, and RC47 complex, but that they were never detected in PSII supercomplexes (Fig. 5). The level of PSII supercomplexes was significantly decreased in var2 under extreme high-light conditions. Therefore, we cannot rule out the possibility that fragmentation of D1 by Deg occurred before PSII dimerization and partial disassembly. Nevertheless, considering that processive degradation of D1 by FtsH is active at the disassembly step, it is plausible that the cooperative D1 degradation mediated by Deg and FtsH is more significant in stroma thylakoids. In this regard, further study of the migration and the disassembly of PSII complexes is necessary to elucidate the PSII repair cycle.

Loss of chloroplast protease often causes defect of chloroplast function and chloroplast development, like
that of var2, which shows the variegated phenotype with undifferentiated plastid (Kato et al., 2007). To alleviate defects of a loss of a chloroplast protease, it is reasonable that other proteases substitute for the protease function.
Indeed, reduced accumulation of Clp protease triggers up-regulation of plastid chaperones and causes dramatic accumulation of SppA, another protease associated with thylakoid membranes (Rudella et al., 2006). This study revealed that the level of Clp protease is upregulated in var2, as evidenced by immunoblots of ClpP6 and ClpC (Fig. 6). It is particularly interesting that a significant portion of the Clp protease complexes appeared to be both Clp and FtsH are structurally similar, harboring ATPase and protease domains, and perform processive degradation (Kato and Sakamoto, 2010). Clp has been suggested as degrading both soluble and membrane-bound substrates ( Sjögren et al., 2006;Kim et al., 2009;Stanne et al., 2009;Zybailov et al., 2009).
We also found upregulation of SppA in var2, which was similar to the case in clp mutants. SppA plays a role in quality control of periplasmic and membrane-bound proteins (Lensch et al., 2001). In chloroplasts, SppA was shown to respond to high-light stress and to contribute to long-term high-light acclimation (Wetzel et al., 2009), implying that SppA upregulation in var2 reflects a general defect in proper thylakoid formation. Collectively considering that evidence, we infer that the complementary functions between FtsH and Clp protease complexes exist for protein quality control in thylakoid membranes. Supporting this inference is the fact that FtsH is upregulated in clpr2 knockdown transgenic lines (Rudella et al., 2006). In this scenario, the lack of significant change in Deg accumulation in var2 is explainable by the functional difference between Deg, Clp, and FtsH.
Over the past decade, research into D1 degradation has revealed the contribution of several proteases in this process, but a question remains as to how the proteases recognize photodamaged D1  Further studies examining substrate recognition mechanisms must be undertaken to support our understanding of PSII repair.

Plant Materials and Growth Conditions
Arabidopsis (Arabidopsis thaliana) Col was used as the wild type. The mutant lines used for this study, var2-1 (contains a nonsense point mutation) and deg5 deg8 (T-DNA insertion lines), were described previously (Takechi et al., 2000;Sun et al., 2007). To isolate the triple mutant var2 deg5 deg8, var2-1 were initially crossed with deg5 and deg8, respectively. The following crossing was performed between isolated homozygous double mutants (var2 deg5 and var2 deg8). The genotype of var2-1 was determined using dCAPS-assisted PCR using primers dCAPS(2-1) (5'-GGACCATGGTCTTTGATGGATTCTTCGTCA-3') and KT304 (5'-TCACGATTGTTCTTTATGTGGCTTAG-3') and digestion with Tsp45I. The genotype of deg5 and deg8 was confirmed by PCR using the following gene-specific and T-DNA-specific primers: LP To isolate thylakoid proteins, three leaf discs were collected using a 5-mm diameter biopsy punch (Kai Medical, Japan). Leaf discs were frozen immediately in liquid nitrogen and pulverized using a microtube homogenizer. Samples were suspended in ice-cold extraction buffer (330 mM sorbitol, 50 mM HEPES, pH 7.5, 5 mM MgCl 2 and 10 mM NaCl 2 ) and centrifuged at 2,500g for 5 min. Then, pellets were resuspended in SDS-PAGE sample buffer ( immunoblots were quantified using the ImageJ program (http://rsbweb.nih.gov/ij/) and normalized to the amount of CBB-stained LHCII.

D1 Fragments Detection
Leaves from approximately 6-week-old plants Col and mutants were used for detection of D1 fragments.
For high-light treatment, detached leaves were incubated for 1 h under extreme high-light irradiation (2,500 µmol photons m -2 s -1 ). To isolate membrane proteins, leaves that had been frozen rapidly in liquid nitrogen were pulverized in a mortar, and were homogenized in a solution containing 330 mM sorbitol, 50 mM HEPES, pH 7.5, 5 mM MgCl 2 and 2 mM Na 2 EDTA. After centrifugation at 3,000g for 5 min, the pellet was resuspended in the same buffer. Chlorophyll was extracted in 80% (v/v) acetone and the absorbance of chlorophyll extracts was measured using a spectrophotometer (Ultrospec 2100 pro; Amersham Biosciences). Chlorophyll contents were calculated applying the following equation: total chlorophyll (mg/L) = 7.12A 660 + 16.8A 642.5 . Membrane suspensions containing 100 µg of chlorophyll were centrifuged at 3,000g for 5 min, and the pellet were resuspended in SDS-PAGE sample buffer to a final concentration of 0.5 mg mL -1 . Before loading, samples were centrifuged at 15,000g for 5 min, and equally loaded supernatants (based on chlorophyll) were subjected to additional analysis. In inhibitor experiment, serine protease activity was blocked using PefaBloc SC (4-(2-Aminoethyl) benzenesulfonyl fluoride hydrochloride; AEBSF) protease inhibitor (Roche). Detached leaves were preincubated with their petioles submersed in 4 mM solutions of AEBSF in the dark for 3 h. To ensure the effect of inhibitor, the leaves were placed in a glass vial containing buffer with newly added 4 mM AEBSF. Then AEBSF was infiltrated for 1 min into the leaves using a syringe needle with a rubber cap. After infiltration of AEBSF, leaves were incubated for 1 h under extreme high-light irradiation (2,500 µmol photons m -2 s -1 ).

Protein Extraction for Chloroplast Proteases
To isolate chloroplast proteins, 2 g of wild-type and var2-1 leaves were harvested from approximately 8-week-old plants. Chloroplasts were isolated according to the protocol described previously (Miura et al., 2007).
Intact chloroplasts were fractionated into stromal and thylakoid fractions by centrifugation after osmotic shock in a hypotonic buffer (50 mM Tris-HCl, pH 7.5, and 5 mM MgCl 2 ). Chloroplast suspensions and stroma fractions were mixed with equal volumes of 2 x SDS-PAGE sample buffer. Then thylakoid membranes were resuspended in SDS-PAGE sample buffer. Before loading, samples were centrifuged at 15,000× g for 5 min. Then equally loaded supernatants (based on chlorophyll) were subjected to additional analysis.

Preparation of Thylakoid Membranes and BN/SDS-PAGE
For BN-PAGE, thylakoid membranes were isolated from growth-light-adapted leaves and from leaves illuminated for 1 h at 2,500 µmol m -2 s -1 by abbreviated thylakoid isolation method. After high-light treatment, leaves were ground in a blender with homogenization buffer (330 mM sorbitol, 50 mM HEPES, pH 7.5, 5 mM MgCl 2 and 2 mM Na 2 EDTA). Homogenates were then filtered though gauze and centrifuged at 2,500g for 5 min.
The pellet was resuspended in 1 mL homogenization buffer and was overlaid on two Percoll gradients (10 and 80%) and centrifuged at 2,500g for 5 min. The sediment at interface between the 10% and 80% gradient was recovered and diluted in five times volume homogenization buffer. After centrifugation at 2,500 × g for 5 min, the pellet was resuspended in 1 mL homogenization buffer and the total chlorophyll concentration was measured. To solubilize membrane proteins, thylakoid membrane suspensions were centrifuged at 2,500g for 3 min. Then pellets were resuspended using NativePAGE Sample Prep Kit (Invitrogen). The final concentration of chlorophyll was 0.5 mg mL -1 and the final concentration of n-dodecyl-ß-D-maltoside was 0.5% (w/v). After centrifugation at 15,000g for 10 min, the supernatant was mixed with CBB G-250; a final concentration of CBB G-250 was 0.125% (w/v). The samples were loaded onto a 4% to 16% gradient native gel. Electrophoresis was performed at 4°C overnight at 50 V. Second-dimension and further analysis were performed as described previously (Kato et al., 2009).
Whole serum was used.

Fluorescence Measurements
Mature leaves of 5-week-old plants were used for measurements of high-light sensitivity. The leaves were incubated for 0, 60, 120, and 240 min under high-light irradiation (1,200 µmol photons m -2 s -1 ). The changes in the maximum quantum yield of PSII (Fv/Fm) were measured. Before the measurements, leaf discs were maintained in the dark for 10 min to oxidize the plastoquinone pool fully.

Supplemental Material
The following materials are available in the online version of this article.

Supplemental Table S1
The ratio of variable to maximum quantum yield of PSII (Fv/Fm) measured from mature leaves using the FluorCam 700MF.
Supplemental Figure S1 Schematic summary of the degradation of D1 protein till this study.      were incubated in extreme high-light conditions (2,500 µmol photons m -2 s -1 ) for 1 h. A representative immunoblot (normalized by chlorophyll content) using anti-D1 (N-term) and anti-D1 (C-term) antibodies and the bands corresponding to CBB-stained LHCII are depicted. A selective detection of the cleavage products of D1 protein is shown in the middle panels.