Intramolecular cross-linking of the extrinsic 33-kDa protein leads to loss of oxygen evolution but not its ability of binding to photosystem II and stabilization of the manganese cluster.

The extrinsic 33-kDa protein of photosystem II (PSII) was intramolecularly cross-linked by a zero-length cross-linker, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide. The resulting cross-linked 33-kDa protein rebound to urea/NaCl-washed PSII membranes, which stabilized the binding of manganese as effectively as the untreated 33-kDa protein. In contrast, the oxygen evolution was not restored by binding of the cross-linked protein, indicating that the binding and manganese-stabilizing capabilities of the 33-kDa protein are retained but its reactivating ability is lost by intramolecular cross-linking of the protein. From measurements of CD spectra at high temperatures, the secondary structure of the intramolecularly cross-linked 33-kDa protein was found to be stabilized against heat treatment at temperatures 20 degrees C higher than that of the untreated 33-kDa protein, suggesting that structural flexibility of the 33-kDa protein was much decreased by the intramolecular cross-linking. The rigid structure is possibly responsible for the loss of the reactivating ability of the 33-kDa protein, which implies that binding of the 33-kDa protein to PSII is accompanied by a conformational change essential for the reactivation of oxygen evolution. Peptide mapping, N-terminal sequencing, and mass spectroscopic analysis of protease-digested products of the intramolecularly cross-linked 33-kDa protein revealed that cross-linkings occurred between the amino group of Lys48 and the carboxyl group of Glu246, and between the carboxyl group of Glu10 and the amino group of Lys14. These cross-linked amino acid residues are thus closely associated with each other through electrostatic interactions.

electrostatically interact with negative charges on PSII intrinsic proteins (27).
Reconstitution experiments have indicated that the three extrinsic proteins bind to the PSII complex in the order of the 33-, 23-, and 17-kDa proteins (28 -30). Among these three extrinsic proteins, the 33-kDa protein is required for stoichiometric and functional binding of the 23-kDa protein, whereas both the 33-and 23-kDa proteins are required for functional binding of the 17-kDa protein. The extrinsic proteins cannot, however, directly bind to each other when they are not associated with the PSII complex (24). These results suggest that binding of the extrinsic proteins to the PSII complex alters the conformation of the extrinsic proteins themselves and/or that of the intrinsic part of the complex so as to create the binding sites for the other extrinsic proteins (24).
To examine whether conformational changes occur with the 33-kDa protein accompanying its binding to PSII and the possible importance of such conformational changes, we performed intramolecular cross-linking of the 33-kDa protein with a water-soluble carbodiimide, EDC, by which the conformational changes of the protein are expected to be suppressed. The intramolecularly cross-linked 33-kDa protein was found to retain the rebinding and manganese-stabilizing capabilities but not the reactivating ability, suggesting that a suitable flexibility of the 33-kDa protein is needed for its full functioning in oxygen evolution.

MATERIALS AND METHODS
Preparation and Cross-linking-Oxygen-evolving PSII membranes were prepared from spinach chloroplasts with Triton X-100 as in Ref. 31, with slight modifications as described in Ref. 21. The PSII membranes were suspended in medium A containing 40 mM Mes (pH 6.5), 0.4 M sucrose, 10 mM NaCl, and 5 mM MgCl 2 and stored in liquid nitrogen until use. The extrinsic 33-kDa protein was extracted from the PSII membranes by 1 M CaCl 2 treatment (16) and purified according to Refs. 27 and 32. For cross-linking, the purified 33-kDa protein was passed through a Sephadex G-25 column equilibrated with distilled water. The concentration of the 33-kDa protein was determined using an extinction coefficient of 16 mM Ϫ1 cm Ϫ1 at 276 nm (33). Intramolecular and intermolecular cross-linking of the 33-kDa protein was carried out in a solution containing 4 M 33-kDa protein and 1 mM EDC at 25°C for 12 h. After the reaction was stopped by adding 100 mM sodium acetate, the reaction mixture was passed through a Sephadex G-25 column equilibrated with 20 mM phosphate buffer (pH 6.6). The pH of the protein mixture was then adjusted to pH 10 with concentrated NaOH solution and incubated at 25°C for 2 h to recover carboxyl groups from modification by EDC. After the pH was returned to 6.5 by adding 500 mM Mes (pH 6.5), the cross-linked 33-kDa protein was concentrated by ultrafiltration and then passed through a Sephacryl S-100HR column equilibrated with 100 mM Mes (pH 6.5) to separate the intramolecularly and intermolecularly cross-linked products.
Reconstitution, Oxygen Evolution, and Electrophoresis-For reconstitution, native PSII membranes were washed with 2.6 M urea, 0.2 M NaCl to remove the three extrinsic proteins of 33, 23, and 17 kDa (17). The resulting PSII membranes were incubated with either the native or the intramolecularly cross-linked 33-kDa protein at a protein to Chl ratio of 0.6 (w/w), in medium A at 0°C for 30 min in the dark at a Chl concentration of 0.5 mg/ml. The reconstituted PSII membranes were collected by centrifugation at 35,000 ϫ g for 10 min, then washed once with and resuspended in medium A. Oxygen evolution was measured at 25°C with a Clark-type oxygen electrode in medium A, to which 5 mM CaCl 2 and 0.4 mM phenyl-p-benzoquinone were supplemented. Chl concentration was determined by the method of Porra et al. (34). SDS-polyacrylamide gel electrophoresis was carried out according to Laemmli (35), with a slab gel of 11.5% acrylamide containing 6 M urea. Samples were solubilized with 5% lithium lauryl sulfate and 75 mM dithiothreitol. After electrophoresis, gels were stained with Coomassie Brilliant Blue and photographed.
Measurements of Manganese Contents and CD Spectra-For determination of manganese released from urea/NaCl-washed PSII membranes reconstituted with the native or intramolecularly cross-linked 33-kDa protein, membrane suspensions after incubation for 0 -48 h at 0°C in the dark were centrifuged at 35,000 ϫ g for 10 min and manganese remaining in the supernatant was assayed with a Hitachi polarized Zeeman atomic absorption spectrophotometer (Z-8000). Amounts of manganese per PSII were estimated by assuming the antenna size of PSII as 250 Chl. The circular dichroism (CD) measurements of the native or intramolecularly cross-linked 33-kDa protein were performed with a JASCO J-500A spectropolarimeter as described in Ref. 36.
Protease Digestion, HPLC Separation, Mass Spectroscopic, and Nterminal Sequencing Analysis-The untreated and intramolecularly cross-linked 33-kDa proteins were denatured in 8 M urea and 25 mM Mes (pH 6.5) at 37°C for 15 h, and then 3 M Tris-HCl (pH 8.5) was added to a final concentration of 2 M urea. The denatured 33-kDa protein was digested first with lysyl endopeptidase at a protein to lysyl endopeptidase ratio of 50 (w/w) for 15 h at 37°C, and then with another lysyl endopeptidase at the same concentration for 15 h at 37°C. The digested mixture was subjected to a reversed phase column (Bondasphere 5 C4 300A, Waters Inc.) in an HPLC set-up (LC-9A, Shimadzu Inc., Japan). The column was eluted with a gradient of 0 -75% acetonitrile in 0.1% trifluoroacetic acid at a flow rate of 1 ml/min, and the elution pattern was monitored at 210 nm. Each fraction was collected, dried, and resolubilized in 2 l of 67% acetic acid out of which, 1 l was mixed with a same volume of matrix (a mixture of 1:1 volume of glycerol and 3-nitrobenzyl alcohol), and analyzed with a fast atom bombardment mass spectrometer (JEOL JMS HX-110) at a voltage of 10 kV with xenon as the ionization gas. The resulting mass spectra were analyzed with a DA5000 data system and assigned to the known protein sequence. The peptides obtained from HPLC were also analyzed for their N-terminal sequence by Edman degradation of the peptides followed by sequence analysis with a protein sequencer (Applied Biosystem, model 477A and 476A).

Preparation of Intramolecularly Cross-linked 33-kDa Pro-
tein-To suppress the formation of intermolecularly crosslinked products, a diluted solution of the 33-kDa protein (4 M) was used for treatment of EDC. A small amount of intermolecularly cross-linked product was, however, formed as shown in Fig. 1. The polypeptide pattern of the EDC-treated 33-kDa protein showed two Coomassie Brilliant Blue-stained bands with apparent molecular masses of about 55 and 27 kDa (lane 1), which correspond to dimer and monomer of the 33-kDa protein resulting from intermolecular and intramolecular cross-linking, respectively. No trimer and polymer are formed under the conditions employed here. The intermolecularly (dimer) and intramolecularly (monomer) cross-linked products of the 33-kDa protein could be clearly separated by a Sephacryl S-100HR column (lanes 2 and 3); upon electrophoresis, the intramolecularly cross-linked 33-kDa protein migrated faster and appeared as a broader band as compared with the untreated protein.
Reconstitution of the Intramolecularly Cross-linked 33-kDa Protein to PSII-The three extrinsic proteins of 33, 23, and 17  1, 2, Fig. 2) (17). The native 33-kDa protein was able to rebind to urea/NaCl-washed PSII completely (lane 3). Because the intramolecularly cross-linked 33-kDa protein migrated faster than the native 33-kDa protein and thus comigrated with the large amount of LHC II bands, detection of the 33-kDa protein reconstituted was done as follows. PSII membranes reconstituted with the native or intramolecularly crosslinked 33-kDa protein were again treated with 2.6 M urea plus 0.2 M NaCl and centrifuged at 35,000 ϫ g for 10 min. The supernatants were concentrated 5-fold by ultrafiltration as the intramolecularly cross-linked 33-kDa protein showed a much weaker Coomassie Brilliant Blue staining intensity than the native 33-kDa protein. The result shown in lane 4R of Fig. 2 indicates that the intramolecularly cross-linked 33-kDa protein rebound to PSII to a significant amount. To estimate the amount of the intramolecularly cross-linked 33-kDa protein rebound, the 33-kDa protein corresponding to the amount when the protein was completely rebound was electrophoresed together (lanes 3C and 4C). The comparison of Coomassie Brilliant Blue staining intensities in lanes 4C and 4R revealed that the intramolecularly cross-linked 33-kDa protein was completely rebound to PSII. It should be noted that the band of the intramolecularly cross-linked 33-kDa protein rebound (lane 4R) showed a similar broad band as the original cross-linked protein (lane 4C). If this is due to heterogeneous cross-linking, the results suggested that the 33-kDa protein was able to rebind to PSII irrespective of the heterogeneous cross-linking. These results indicate that the binding sites of the 33-kDa protein remain intact after intramolecular cross-linking. Table I shows oxygen-evolving activity of urea/NaCl-washed PSII membranes reconstituted with the native or intramolecularly cross-linked 33-kDa protein. Removal of the three extrinsic proteins by urea/NaCl wash reduced the oxygen evolving activity to 4% of the original activity. The activity was restored to 66% by reconstitution with the native 33-kDa protein, whereas it was scarcely restored by reconstitution with the intramolecularly cross-linked 33-kDa protein. This indicates that the intramolecularly cross-linked 33-kDa protein lost its reactivating ability of oxygen evolution even though it was completely rebound to PSII.
When urea/NaCl-washed PSII membranes lacking the three extrinsic proteins were incubated at 0°C in the dark, two out of the four manganese per PSII released after 48 h (Fig. 3) (20). Reconstitution of the native 33-kDa protein effectively suppressed the release of manganese (Fig. 3). Reconstitution of the intramolecularly cross-linked 33-kDa protein also suppressed the manganese release equally effectively (Fig. 3). These results indicate that the cross-linked 33-kDa protein still retained its ability to stabilize binding of the manganese cluster.
Structural Flexibility of the 33-kDa Protein-To examine the changes on structural flexibility of the 33-kDa protein by intramolecular cross-linking, the far-ultraviolet circular dichroism spectra (CD spectra) of the native and the intramolecularly cross-linked 33-kDa protein were measured at high temperatures (Fig. 4). The secondary structure of the native 33-kDa protein was appreciably affected at 60°C and its random conformation appeared at 70°C (Fig. 4A). On the contrary, the CD spectral changes of the intramolecularly cross-linked 33-kDa protein required much higher temperature than those of the native 33-kDa protein, and the random conformation of the FIG. 3. Stabilization of manganese binding by reconstitution of the native or intramolecularly cross-linked 33-kDa protein with urea/NaCl-washed PSII membranes. The urea/NaCl-washed PSII before or after reconstitution with the native or intramolecularly cross-linked 33-kDa protein was incubated at 0°C in the dark, and its manganese content bound was determined at the designated time points.  intramolecularly cross-linked 33-kDa protein appeared at 90°C (Fig. 4B). These results indicate that structural flexibility of the 33-kDa protein was significantly decreased by intramolecular cross-linking. .  5 shows the peptide maps of the native (A) and the intramolecularly cross-linked 33-kDa protein (B) which had been digested with lysyl endopeptidase and separated by reversedphase HPLC. Peptide peaks 2, 3, 4, and 6 drastically decreased and new peptide peaks A and B appeared by intramolecular cross-linking. Therefore, peaks A and B are expected to be the peptides containing intramolecularly cross-linked sites. To identify the intramolecularly cross-linked sites, N-terminal amino acid sequences and molecular masses of these two peptides were determined (Table II). Two amino acid sequences were detected by Edman degradation of peptide peak A, which completely agreed with amino acid sequences from Tyr 45 -Lys 49 and Ile 237 -Gln 247 of the 33-kDa protein, in which only Lys 48 and Glu 246 were not detected. Peptide peak A had a measured mass of 1927.10 Da, which is consistent with the predicted mass of Tyr 45  Peptide peaks 2, 3, 4, and 6, which were significantly decreased by intramolecular cross-linking, were found to be Gly 160 -Lys 186 , Ile 237 -Gln 247 , Thr 15 -Lys 44 , and Leu 77 -Lys 101 , respectively, by mass spectrometeric analysis (data not shown). Peptide peaks 3 and 4 are involved in the intramolecularly cross-linked peptide peaks A and B. The cross-linked products containing peptide peaks 2 and 6 were not, however, found on the peptide map of lysyl endopeptidase digests of the intramolecularly cross-linked 33-kDa protein (Fig. 5B). It is likely that these cross-linked products were retained in and not eluted from the reversed-phased HPLC column. Thus, in addition to the intramolecular cross-linking between Lys 48 and Glu 246 and between Glu 10 and Lys 14 , intramolecular cross-linkings containing the peptide of Gly 160 -Lys 186 and Leu 77 -Lys 101 seem to be formed.

Reactivation Mechanism of Oxygen Evolution by Binding of the 33-kDa Protein-
The present results demonstrated that the rebinding and manganese-stabilizing capabilities of the 33-kDa protein were retained but its reactivating ability was lost by intramolecular cross-linking of the protein with EDC. This implies that different mechanisms exist for the stabilization of manganese binding and reactivation of oxygen evolution; the latter but not the former was impaired by the intramolecular cross-linking. Several possibilities may be considered as responsible for loss of the reactivating ability, e.g. an inhibition of homodimerization of the 33-kDa protein which might be essential for its functioning, a loss of the function of the 33-kDa protein in maintaining binding of Ca 2ϩ and/or Cl Ϫ (37), and a suppression of conformational changes of the 33-kDa protein possibly accompanying its binding to PSII. Our present results support the hypothesis that a conformational change occurred

FIG. 4. CD spectra of the native (A) and intramolecularly cross-linked (B) 33-kDa protein at various temperatures.
Characterization of Intramolecularly Cross-linked 33-kDa Protein accompanying binding of the 33-kDa protein, and this conformational change was suppressed by the intramolecular crosslinking, since the secondary structure of the protein was significantly stabilized against heat treatment by intramolecular cross-linking (Fig. 4). This suggests that the structural flexibility of the 33-kDa protein was remarkably reduced. In addition, while urea/NaCl-washed PSII membranes reconstituted with the native 33-kDa protein completely rebound the 23-kDa protein, the PSII reconstituted with the intramolecularly cross-linked 33-kDa protein scarcely rebound the 23-kDa protein (data not shown). Since the 23-kDa protein cannot directly associate with the 33-kDa protein in solution (24), these results suggest that binding of the 33-kDa protein to PSII alters the conformation of the protein itself which is essential for binding of the 23-kDa protein. The occurrence of a structural change of the protein is consistent with results of pH-dependent structural changes (38), effects of genetic or chemical modification of its disulfide-forming cysteines (39,40), or the effects of conformational constrains resulting from other amino acid substitutions (41).
From the present and previous results, however, we cannot determine whether the conformational change of the 33-kDa protein itself or a further structural rearrangement of intrinsic PSII proteins allosterically induced by binding of the 33-kDa protein is responsible for the reactivation of oxygen evolution. The possible structural changes of intrinsic PSII proteins upon binding of the 33-kDa protein have been previously reported; for example, we recently showed that removal of the 33-kDa protein makes the C-terminal region of CP43 accessible to trypsin, thus suggesting that removal of the protein at the lumenal side induces a conformational change of the CP43 protein at the stromal side (42). An effect on properties of the acceptor side of PSII upon either biochemical removal of the 33-kDa protein (43) or genetic deletion of the psbO gene encoding the 33-kDa protein (44,45) has also been reported based on thermoluminescence and fluorescence measurements. Based on these results, we propose the following mechanism for reactivation of oxygen evolution by binding of the 33-kDa protein: binding of the 33-kDa protein to PSII alters the conformation of the 33-kDa protein itself which allosterically results in structural changes of intrinsic PSII proteins ligating the manganese atoms, leading to formation of a functional conformation of the manganese cluster and then the reactivation of oxygen evolution.
Structure of the 33-kDa Protein-Although the primary structure of the 33-kDa protein has been determined in various species of plants (46), there is only very limited information concerning the tertiary structure of the protein. Two Cys residues (Cys 28 and Cys 51 ) of the protein form a disulfide bond important for maintaining the functional structure of the protein (40,45,47). The secondary structural analysis of the 33-kDa protein in solution by far-UV CD spectroscopy revealed that the protein contains a large proportion of ␤-sheet and a relatively small amount of ␣-helical structure (48). Recently, we reported that the positive charges of ⑀-amino groups on Lys 4 , Lys 20 , Lys 66 or Lys 76 , Lys 101 , Lys 105 , Lys 130 , Lys 159 , Lys 186 and one or two Lys in Lys 230 -Lys 236 in the 33-kDa protein electrostatically interact with negative charges on PSII intrinsic proteins (27). This implies that these Lys residues are located on the surface of the protein which interacts with PSII intrinsic proteins when it binds to PSII. The present study revealed that the positive charges of the amino group of Lys 48 and Lys 14 electrostatically associate with the negative charges of the carboxyl group of Glu 246 and Glu 10 , respectively, when the protein is free in solution. This is very interesting, since we have reported that the carboxyl group of Glu 246 or C terminus electrostatically interacts with the amino group of Lys 190 of the 33-kDa protein when the protein is associated with PSII (49). This indicates that the carboxyl group in the C-terminal region of the 33-kDa protein electrostatically interacts with different amino groups, depending on whether the protein is free in solution or binds to PSII, which again suggests that the 33-kDa protein alters its conformation upon binding. Fig. 6 summarizes the localization of amino acid residues interacting with each other in the spinach 33-kDa protein which have been determined in the present and previous studies (27,40,47). Amino acids associated with each other through electrostatic interaction or disulfide bond are connected with a line when the protein is free in solution and with a dotted line when the protein binds to PSII. Domains involving Lys residues which are located in the regions interacting with PS II intrinsic proteins are boxed. These results provide an important information for the tertiary structure of the 33-kDa protein.  6. Localization of amino acids closely associated with each other (connected with a line) and Lys residues located in the regions that interact with PSII intrinsic proteins (boxed) in the primary structure of the 33-kDa protein from spinach. Residues interacting with each other through either electrostatic interaction or disulfide bond are connected with a solid line when the protein is free in solution and with a dotted line when the protein is associated with PSII. Domains involving Lys residues interacting with PSII intrinsic proteins are boxed.