MS-based cross-linking analysis reveals the location of the PsbQ protein in cyanobacterial photosystem II

Significance

Cyanobacteria, photosynthetic prokaryotes, were primarily responsible to create an oxygen-rich atmosphere on our planet. Photosystem II (PSII), a large membrane-bound pigment protein complex in cyanobacteria, uses light energy to oxidize water to dioxygen. X-ray crystal structures of cyanobacterial PSII have recently been determined. PsbQ, a protein that optimizes PSII-mediated oxygen evolution activity, is ubiquitously present in cyanobacteria. However, PsbQ is absent in the determined structures of PSII. By using protein cross-linking, mutagenesis, and advanced MS techniques, we showed that PsbQ in the model cyanobacterium Synechocystis 6803 binds to CP47 and PsbO, two known protein components of PSII. These results helped determine the location of PsbQ on the lumenal side, near the water oxidation site, of cyanobacterial PSII.

Abstract

PsbQ is a luminal extrinsic protein component that regulates the water splitting activity of photosystem II (PSII) in plants, algae, and cyanobacteria. However, PsbQ is not observed in the currently available crystal structures of PSII from thermophilic cyanobacteria. The structural location of PsbQ within the PSII complex has therefore remained unknown. Here, we report chemical cross-linking followed by immunodetection and liquid chromatography/tandem MS analysis of a dimeric PSII complex isolated from the model cyanobacterium, Synechocystis sp. PCC 6803, to determine the binding site of PsbQ within PSII. Our results demonstrate that PsbQ is closely associated with the PsbO and CP47 proteins, as revealed by cross-links detected between ¹²⁰K of PsbQ and ¹⁸⁰K and ⁵⁹K of PsbO, and between ¹⁰²K of PsbQ and ⁴⁴⁰D of CP47. We further show that genetic deletion of the psbO gene results in the complete absence of PsbQ in PSII complexes as well as the loss of the dimeric form of PSII. Overall, our data provide a molecular-level description of the enigmatic binding site of PsbQ in PSII in a cyanobacterium. These results also help us understand the sequential incorporation of the PsbQ protein during the PSII assembly process, as well as its stabilizing effect on the oxygen evolution activity of PSII.

Photosystem II (PSII) functions as a light-driven, water-plastoquinone oxidoreductase in oxygenic photosynthesis. PSII is a membrane protein complex containing more than 20 protein subunits. Early biochemical investigations established that at least seven major intrinsic proteins are required for oxygen evolution: CP47, CP43, D1, D2, the α- and β- subunits of cytochrome b₅₅₉, and PsbI (1). Additionally, a number of low molecular mass intrinsic polypeptides are associated with these seven major polypeptides.

Despite the relatively conserved overall functioning of PSII over billions of years of evolution, the composition of the luminal extrinsic PSII proteins varies significantly across different phyla (2–5). In higher plants and green algae, four extrinsic proteins, PsbO, PsbP, PsbQ, and PsbR, are required to support maximal rates of oxygen evolution under physiological conditions. In contrast, PsbO, PsbU, and PsbV (cytochrome c₅₅₀) play analogous roles in cyanobacteria and red algae (5). More recently, it was discovered that homologs of PsbQ and PsbP exist in cyanobacteria (2, 3, 6). These differences are especially noteworthy given that the oxygen-evolving complex itself has remained practically unaltered throughout the evolution from cyanobacteria to green algae and higher plants (7).

The presence of PsbQ in cyanobacterial PSII was discovered by Kashino et al. (3) when analyzing the complete protein complement of isolated PSII complexes from the HT3 strain of Synechocystis sp. PCC 6803, which contains a C-terminal His₆-tag on the CP47 protein (8). The physiological role of PsbQ was subsequently investigated in several laboratories through phenotypic analysis of single or double mutants lacking PsbQ as well as other extrinsic PSII proteins (2, 9, 10). Direct evidence for the binding of PsbQ to PSII was obtained by using isolated PSII complexes from a QHis strain containing a C-terminal His₈-tag on PsbQ (11). It was shown that PsbQ defines cyanobacterial PSII complexes with high activity and stability. Although the X-ray crystal structure of isolated cyanobacterial PsbQ has been solved to near-atomic resolution, PsbQ has not been identified in any of the crystal structures of cyanobacterial PSII from the thermophilic cyanobacteria Thermosynechococcus elongatus BP-1 (12) and Thermosynechococcus vulcanus (7). Because of the absence of PsbQ in these structures, the binding site of this protein within the PSII complex has remained unknown.

In the present study, we have used chemical cross-linking followed by immunodetection and MS to investigate the structural location of PsbQ in PSII from Synechocystis 6803. Based on our analysis, we propose a model in which PsbQ binds to PSII through a close association with PsbO and CP47, thereby stabilizing the PSII dimer and permitting high rates of oxygen evolution in this cyanobacterium.

Results

By using a Protein BLAST (BLASTp) search, we examined the prevalence of PsbQ across the cyanobacterial phylum. Homologs to Synechocystis 6803 PsbQ were detected in 97 other cyanobacterial species (Fig. 1 and Fig. S1). PsbQ is present in an evolutionarily diverse variety of diazotrophic and nondiazotrophic strains, but was not found in Gloeobacter species, an ancient lineage of cyanobacteria that do not have a thylakoid membrane. We therefore propose that the psbQ gene evolved in cyanobacteria concurrently with, or soon after, the development of the thylakoid system. Notably, PsbQ homologs are present in the thermophiles T. elongatus and T. vulcanus, although the PSII crystal structures obtained from these species lack PsbQ (as detailed earlier) (7, 12).

Fig. 1.

Phylogenetic tree showing evolutionary relationship between PsbQ proteins from selected cyanobacterial species; Arabidopsis thaliana serves as an outgroup. Green, nitrogen-fixing; red, thermophilic. A. marina MBIC11017, Acaryochloris marina MBIC11017; M. aeruginosa NIES-843, Microcystis aeruginosa NIES-843. (Scale bar: substitutions per nucleotide.)

We have previously described the isolation and characterization of highly active and stable PSII from the QHis strain of Synechocystis 6803 (11). In the present study, we took advantage of its availability to pursue the structural location of PsbQ within functional PSII. Oxygen-evolving PSII is usually found as a dimer and monomer, with the isolated dimer being more active than the monomer (13). Blue native (BN) gel and SDS/PAGE protein profile analysis of HT3PSII and QHisPSII (Fig. 2) showed that the HT3PSII preparation could be resolved into two major green bands (Fig. 2A), corresponding to PSII dimer and PSII monomer, respectively. Interestingly, only the dimeric form was observed for the QHisPSII preparation. This observation may partially explain the higher oxygen evolving activity of QHisPSII compared with HT3PSII (11), presumably because the His₆-tag on CP47 in the HT3 strain enables the isolation of a mixture of a more active PSII dimer and a less active PSII monomer.

Fig. 2.

(A) BN gel analysis of HT3PSII and QHisPSII. Arrows indicate the dimeric (marked as “D”) and monomeric (marked as “M”) forms of PSII. (B) Polypeptide profiles of isolated HT3PSII and QHisPSII. The positions of major PSII protein components are indicated on the left. The C-terminally His-tagged PsbQ and the Psb28 protein (asterisk) are indicated. Each lane was loaded with sample containing 5 µg of chlorophyll a (Chla).

Chemical Cross-Linking.

Chemical cross-linkers can covalently link amino acid pairs found in close proximity to each other in a protein or a protein complex (14–18). In this study, we used 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide (EDC), a zero-length cross-linker that can link carboxylate groups [from aspartate (D) and glutamate (E) side chains, and protein C termini] to primary amine groups [from lysine (K) or protein N termini]. We also used the thiol-cleavable 12-Å cross-linker 3,3′-dithiobis(sulfosuccinimidylpropionate) (DTSSP) and the noncleavable bis(sulfosuccinimidyl)suberate (BS³), both of which can link two amino acid residues containing primary amine groups. Similar cross-linking approaches have been used to reveal a close association between PsbO and CP47 in PSII (19, 20), a result that was subsequently validated by crystallographic studies (21). To identify the binding site of PsbQ in the PSII dimer, we chose to cross-link QHisPSII instead of HT3PSII, because the former contains a more biochemically homogeneous population than HT3PSII (as detailed earlier).

We first probed our cross-linked samples with anti-PsbQ antibodies and found that treatment of QHisPSII with EDC and DTSSP generated similar, but not identical, cross-linked products, with apparent molecular masses of 23 kDa (product a), 27 kDa (b), 36 kDa (c), 52 kDa (d), 57 kDa (e), 75 kDa (g), 80 kDa (f), and 96 kDa (h; Fig. 3 A and B). In the absence of EDC and DTSSP, C-terminally His₈-tagged PsbQ migrated at 17 kDa (Fig. 3 A and B). PsbQ is N-terminally lipid-modified, and thus anchored to the thylakoid membrane or to the transmembrane domain of PSII (2, 9). However, the observed PsbQ-containing cross-linked products (Fig. 3 A and B) indicated that PsbQ is in close contact with other proteins, and possibly binds to PSII via interprotein interactions. We further examined the DTSSP cross-linked products by using anti-CP47 (Fig. 3C), anti-PsbH (Fig. 3D), anti-D1 (Fig. 3E), and anti-PsbO (Fig. 3F) antibodies. Products d and i may be identical and represent a cross-link between PsbQ and CP47, or products d and m may be identical and represent a cross-link between PsbQ and PsbO or PsbU and PsbO. Similarly, products h and n may also represent a cross-linked complex containing PsbQ and PsbO, and, alternatively, any of the PsbQ-containing cross-linked products observed in Fig. 3B may be linked to any of the numerous other PSII proteins. Based on these data, we suggest structural interactions between PsbQ and CP47 and/or PsbQ and PsbO.

Fig. 3.

Chemical cross-linking of PsbQ with other PSII subunits. (A) Detection of cross-linked species of PsbQ. Lane 1, QHisPSII; lane 2, QHisPSII plus EDC. Immunoblots were probed with anti-PsbQ antibodies. (B–F) DTSSP cross-linking in QHisPSII. Immunoblots were probed with anti-PsbQ, anti-CP47, anti-PsbH, anti-D1, and anti-PsbO antibodies, respectively. Lane 1, QHisPSII; lane 2, QHisPSII plus DTSSP. Each lane was loaded with sample containing 0.2 µg of Chla. Non-cross-linked proteins are labeled with asterisks.

To determine cross-link partner(s) of PsbQ more definitively, we used HPLC coupled with MS (LC-MS/MS). After more than a decade of method development, a platform for identification of cross-linked species using MS and bioinformatics techniques is currently relatively mature (17, 18, 23–25). MS-based techniques can be used to identify cross-linked species with high resolution and mass accuracy. Careful interpretation of the mass and the product-ion spectra by using database searches can reveal the exact amino acid pairs that are cross-linked (26). To minimize sample losses, we subjected our cross-linked products to direct in-solution trypsin digestion instead of in-gel digestion, followed by LC-MS/MS (Materials and Methods). We detected two cross-links between PsbQ and PsbO, as seen in the product-ion (MS/MS) spectra for a representative PsbO-PsbQ cross-linked species (Fig. 4A) identified by a database search (Fig. S2). Examining the product-ion spectrum, we could verify that the identified link between residues ¹²⁰K of PsbQ and ¹⁸⁰K of PsbO is a confident search result (Fig. 4A). [Note the peptide-sequence numbering used in this study hereafter is based on the PSII structure (7) (Protein Data Bank [PDB] ID code 3ARC) and the PsbQ structure (27) (PDB ID code 3LS1) for literature consistency, unless otherwise stated.] Residue ¹²⁰K of PsbQ was also linked to residue ⁵⁹K of PsbO (Fig. S2). Considering the cross-linker arm span of DTSSP (12 Å), we were able to localize ¹²⁰K of PsbQ approximately between ⁵⁹K and ¹⁸⁰K of PsbO.

Fig. 4.

Product-ion (MS/MS) spectra obtained for cross-linked peptides. (A) Product-ion (MS/MS) spectra of the intermolecular cross-link between peptide GLYTGYDNAVALPSAAD¹⁸⁰KFR (PsbO) and LDAAA¹²⁰KDR (PsbQ). Cross-links were formed specifically between ¹⁸⁰K of PsbO (in peptide 163–182) and ¹²⁰K of PsbQ (in peptide 115–122) upon treatment with DTSSP. (B) Product-ion (MS/MS) spectra of the intermolecular cross-link between peptides AQLGEGFDFDTETFNS⁴⁴⁰DGVFR (CP47) and A¹⁰²KTLAK (PsbQ); ⁴⁴⁰D of CP47 was cross-linked to ¹⁰²K of PsbQ upon treatment with EDC.

In addition to the interprotein cross-links between PsbQ and PsbO, we detected two interprotein cross-links between CP47 and PsbO, as well as many intraprotein cross-links in CP47 and PsbO (Table S1). Considering the difference in protein sequence between PsbO in the crystal structure (PDB ID code 3ARC) from T. vulcanus and PsbO from Synechocystis 6803, we generated a homology model of Synechocystis 6803 PsbO by using I-TASSER (28) and compared the model to the crystal structure version by using the PyMOL and APBS programs (29, 30) (Fig. S3). Our cross-linking results are consistent with the structural relationship between CP47 and PsbO observed in the PSII crystal structures from the two thermophilic cyanobacteria (7, 21). We consider these results as positive controls. By using immunological methods, Bricker et al. determined earlier that PsbO forms cross links to CP47 in PSII (19). With the current MS-based cross-link analysis and the availability of PSII crystal structures, we have confirmed this early finding and added finer molecular detail.

Analysis of the EDC-treated sample revealed a cross-link between ¹⁰²K of PsbQ and ⁴⁴⁰D of loop E of CP47 (Fig. 4B and Fig. S4). Given that EDC is a carboxylate-to-amine zero-length cross-linker, this cross-link implies that these two residues interact by complementary charges and are within van der Waals contact of each other. This result particularly helps to define a close association between PsbQ and CP47.

PsbO Deletion Mutant.

To gain further insight into the in vivo structural and functional relationships between PsbO and PsbQ, we generated a psbO deletion mutation in the genetic background of HT3 (henceforth HT3∆O). By using the HT3 (8), HT3∆Q (9), and newly generated HT3∆O strains, we then isolated three types of PSII complexes, with the normal complement of extrinsic proteins (HT3), with PsbQ deleted (HT3∆Q), and with PsbO deleted (HT3∆O), respectively. BN gel analysis (Fig. 5A) of the HT3PSII, HT3∆QPSII, and HT3∆OPSII preparations showed that HT3PSII was resolved to two major green bands (Fig. 5A, lane 1), corresponding to PSII dimer and monomer (31). The PSII dimer/monomer ratio in HT3∆QPSII was not notably different from that of HT3PSII (Fig. 5A, lane 2), except that RC47 (the CP43-less PSII precursor) was more pronounced than in HT3PSII, indicative of the possible role of PsbQ in protection of functional PSII or during PSII biogenesis. SDS/PAGE polypeptide profile (Fig. 5B) and immunodetection (Fig. 5C) analysis indicate that the PsbO protein level did not significantly change in the absence of PsbQ. Deletion of psbO (HT3∆O), however, resulted in complete elimination of dimeric PSII (Fig. 5A) and the absence of PsbQ (Fig. 5 B and C). It appears that binding of PsbO to PSII is independent of PsbQ, but PsbQ is unable to bind to PSII in the absence of PsbO, a protein that presumably contributes to the intermonomer interactions with CP47 and stabilization of the PSII dimer (32, 33).

Fig. 5.

(A) BN gel analysis of HT3PSII, HT3∆QPSII, and HT3∆OPSII. Dimeric (marked as “D”) and monomeric (marked as “M”) PSII are indicated. Note the increase in RC47 (CP43-less monomeric PSII) in HT3∆OPSII. (B) Polypeptide profiles of isolated HT3PSII, HT3∆QPSII, and HT3∆OPSII. The positions of the major PSII protein components are indicated on the left. PsbO position is marked with an asterisk. Each lane was loaded with sample containing 5 µg of Chla. (C) Immunodetection of PSII polypeptides in the isolated complexes. Samples containing 0.2 µg of Chla were fractionated by SDS/PAGE, transferred to a PVDF membrane, and probed by using anti-CP47, anti-PsbO, and anti-PsbQ antibodies.

Discussion

The mechanism of water oxidation in PSII, initially developed in ancient cyanobacteria, has remained virtually unaltered during the evolution of algae and plants. Various extrinsic proteins, such as the well-defined PsbO, PsbU, and PsbV proteins in cyanobacteria, are believed to fine-tune the specific needs of PSII activity, helping modulate its requirements for inorganic cofactors (manganese, calcium, and chloride) and optimize water oxidation activity under different environmental conditions. In cyanobacterial PSII, PsbQ is an enigmatic protein. Homologs of PsbQ are present in all thylakoid-containing cyanobacterial strains (Fig. 1 and Fig. S1), and it is a stoichiometric component of highly active PSII preparations from Synechocystis 6803 (11). Thus, the absence of PsbQ in the PSII crystal structures from T. vulcanus and T. elongatus might have resulted from the loss of this protein during the purification of PSII from these thermophilic cyanobacteria.

Our present study demonstrates that PsbQ is closely associated with PSII via interactions with PsbO and CP47. Based on our cross-linking data (Fig. 4 A and B and later discussion), we propose a molecular model of the location of PsbQ within a PSII (PDB ID code 3ARC) dimer (Fig. 6A). In this model, PsbQ is positioned in a valley formed by the long luminal loop E of CP47 and PsbO from each monomer. In this valley, the positively charged surface formed by helices 2 and 3 (H2 and H3) of PsbQ (27) interacts with the negatively charged surface of the PsbO protein (Fig. S3C). By adopting this orientation, two conserved helices (H2a and H2b) from the two PsbQ proteins in the dimer are arranged in an antiparallel configuration, with their N-termini pointing toward the interface of loop C of CP47 and PsbO, which is close to the peripheral interface of the two PSII monomers. This structural location of PsbQ does not cause any apparent structural conflicts between PsbQ and the other extrinsic proteins, i.e., PsbO, PsbU, and PsbV. An implication of the spatial relationship between PsbQ, PsbO, and loop E of CP47, however, is that only after PsbO is recruited to the luminal side of PSII is PsbQ able to bind. This hypothesis is supported by our observation that deletion of PsbO leads to complete absence of PsbQ in PSII (Fig. 5 B and C). In contrast, loss of PsbV reduces the PsbQ level to nearly 40% of that in the WT cells (35). Our model is consistent with the experimental observation that PsbO is the first luminal extrinsic protein recruited to PSII (31, 35). It is also supported by the earlier observation that deletion of the psbO gene abolishes the dimerization process of PSII in vivo (34), as observed in our PSII preparations as well (Fig. 5A).

Fig. 6.

Schematic model for the binding of PsbQ to the interface of dimeric PSII (PDB ID codes 3LS1 and 3ARC, respectively). PsbQ is shown in lime green; CP47 is shown in wheat color; PsbO is shown in yellow and blue; PsbU is shown in pink; PsbV is shown in orange; PsbM is shown in magenta; and other PSII components are shown in gray. (A) Luminal side view (tilted) showing PsbQ binding to the cove formed by CP47 and PsbO. (B) A magnified view of the interactions between PsbQ and PsbO shown in A; ¹²⁰K of PsbQ as well as ⁵⁹K and ¹⁸⁰K of PsbO are represented as spheres; ⁹⁶K from the second PsbQ protein in the PSII dimer is also shown. (C) Same as B with clockwise rotation (60°), showing interactions between PsbQ and CP47, as deduced from Fig. 4B; ¹⁰²K of PsbQ and ⁴⁴⁰D of CP47 are represented as spheres. (D) Side view of the interactions between the two PsbQ proteins and the PSII dimer. All images were prepared by using PyMOL software (30).

The present study addresses whether luminal domains of CP47, CP43, and three known extrinsic proteins are able to accommodate additional extrinsic proteins. The data using the zero-length cross-linker EDC (Figs. 3A and 4B) suggest that regions of complementary charge exist between PsbQ and other PSII proteins. It was previously suggested that PsbQ is associated directly with a small (4 kDa) intrinsic PSII core subunit in a green-algal PSII complex (36). The identity of this protein, however, remains unknown. It seems likely that the small cross-linked products observed in Fig. 3A (products a, b) and Fig. 3B (products a, b) could result from cross-links between PsbQ and small subunits located in the interface between two PSII monomers (e.g., PsbM, PsbL, and PsbT). This assumes, however, a closer contact of PsbQ to the transmembrane domain of PSII, or the floor of the valley formed by CP47 and PsbO. Such a close contact seems reasonable (Fig. 4B) for a compact association of PsbO and its binding partners including CP47, CP43, D1, D2, and PsbU. To address such a hypothesis, however, more elegant protein footprinting experiments e.g., hydrogen–deuterium exchange or fast photochemical oxidation of proteins (37), need to be designed.

PsbO interacts with a number of other PSII subunits (4). Our MS-based cross-linking analysis also detected two CP47–PsbO cross-links that are consistent with the available crystal structures (7, 12) (Table S1). Moreover, PsbO also participates in an intermonomer interaction with CP47, stabilizing the PSII dimer. In silico analysis indicates that ⁵⁹K of PsbO could form a charge-pair interaction with ³⁰⁷E of CP47 (4, 38). Taken together with the structural proximity of PsbQ to PsbO and the close association between PsbQ and CP47 (Fig. 4 A and B), we speculate that PsbQ increases the stability of the PSII dimer by interacting with PsbO and CP47, thus decreasing the solvent exposure of those interaction interfaces. This structural model is consistent with the results of previous studies that deletion of the psbQ gene in Synechocystis 6803 results in photosynthetic defects under Ca²⁺ and Cl⁻ limiting conditions (2, 10). The phenotype of this deletion mutant, however, was less severe relative to that of other cyanobacterial extrinsic protein mutants, indicative of the auxiliary role of PsbQ in PSII photochemistry under nutrient-replete conditions.

Genetic and physiological data indicate that PsbQ stabilizes the binding of PsbV in PSII (9, 39), and in silico protein docking analysis seems to support this idea as well (40). Our LC-MS/MS analysis, however, failed to detect any confident PsbQ–PsbV cross-links. Interestingly, PsbQ–PsbQ cross-linked species (DMLGLASSLLP⁹⁶KDQDK, LDAAA¹²⁰KDRNGSQAK; cross-link between ⁹⁶K and ¹²⁰K; Fig. S5) were consistently detected in our cross-linked samples. It is unlikely that this cross-linked species originated from one copy of PsbQ (i.e., an intraprotein link), because the arm span of the cross-linker used in this study is 12 Å and the distance between ⁹⁶K and ¹²⁰K is 41 Å (Fig. S6). More importantly, the two Ks are located on opposite ends of helix 3 in the crystal structure (27) rather than in a loop area that might have more structural flexibility. Our immunodetection of cross-linked species seems also to support a dimeric PsbQ from both EDC and DTSSP cross-linker results (Fig. 3A, label c and Fig. 3B, label c). In our model, helix 3 is buried in the interface formed by loop E of CP47 and PsbO. Our data indicate that the two copies of PsbQ are located in structural proximity to each other and probably share the same symmetrical axis as the two PSII monomers. Bioinformatic analysis using the ConSurf server indicated that amino acid residues that are involved in the association between the extrinsic proteins (PsbO, PsbU, PsbV) and corresponding intrinsic PSII components are highly conserved (41–43). Our proposed binding surfaces of the two copies of PsbQ, and of PsbQ to its binding partners, are also conserved relative to the fully exposed areas. In line with previous studies (18, 32), our model (Fig. 6A) assumes there is only one copy of PsbQ per PSII monomer, but we cannot completely exclude a more complicated scenario of more than one copy per PSII monomer. In this context, it is noteworthy that the chlorophyll c containing diatom Chaetoceros gracilis has two different PsbQ homologs, PsbQ′ and Psb31, in each PSII monomer (44). Our observation of PsbQ–PsbQ interactions in QHisPSII also does not necessarily exclude the possibility that PsbQ could associate with monomeric PSII, in which, after the recruitment of PsbO, PsbO and PsbQ synergistically facilitate a rapid PSII dimerization process during the dynamic assembly of PSII.

PsbQ is present in a highly active and stable form of PSII in the mesophilic cyanobacterium Synechocystis 6803, but its location within the complex has remained unknown as a result of its absence in the PSII crystal structures obtained from thermophilic cyanobacteria. This discrepancy may reflect the different environmental conditions to which the organisms adapt. Our study presents a molecular-level model for the binding site of PsbQ to dimeric PSII. In addition, we detected a cross-link between two copies of PsbQ, which suggests strongly the presence of PsbQ in the PSII dimer interface. Our results help elucidate the stage of incorporation of PsbQ during the PSII assembly process and its stabilizing effect on the PSII dimer, and thus provide a basis for further investigation of its role in optimizing PSII function.

Materials and Methods

BLAST Analysis, Sequence Alignment, and Phylogenetic Analysis.

National Center for Biotechnology Information BLASTp software (45) was used to detect homologs of the Synechocystis 6803 PsbQ protein (Sll1638) in other sequenced cyanobacterial strains. A protein was considered a homolog if its E-value was <10⁻⁶. Search parameters were as follows: database, nonredundant protein sequences; organism, cyanobacteria; word size, 3; maximum matches in a query range, 0; matrix, BLOSUM62; gap costs, existence:11 extension:1; compositional adjustments, conditional compositional score matrix adjustment. BLAST results were aligned by using Clustal Omega (46, 47) (www.ebi.ac.uk/Tools/services/web/toolform.ebi?tool=clustalo) by using the default parameters. Phylogenetic trees were viewed by using TreeView version 1.6.6 (48).

Cyanobacterial Culture and PSII Purification.

Cyanobacteria strains were grown in BG11 medium. Generation of the QHis strain was previously reported (11). The HT3 strain was a generous gift from Terry M. Bricker (Louisiana State University, Baton Rouge, LA) (8). The HT3∆Q strain was previously reported (9). The HT3∆O strain was generated by transforming the HT3 strain by using the ∆O construct previously reported (49). Purification of tagged PSII complexes was performed as described previously (3) with minor modifications.

Protein Analysis.

Protein electrophoresis was performed as described previously (3, 50) unless otherwise indicated. After transfer of the fractionated proteins onto PVDF membranes (Millipore), PSII subunits were identified by using specific antisera. Bands were visualized by using enhanced chemiluminescence reagents (West Pico; Pierce) on an ImageQuant LAS-4000 imager (GE Healthcare), and image quantification was performed by using ImageQuant TL software. Levels of PSII monomers and dimers were determined by BN gel electrophoresis (51).

Chemical Cross-Linking.

PSII preparations were resuspended at 0.2 mg/mL chlorophyll a in 25% (wt/vol) glycerol, 10 mM MgCl₂, 5 mM CaCl₂, and 50 mM MES buffer (pH 6.5). Cross-linking of PSII samples with EDC was performed essentially as described previously (52). DTSSP and BS³ (Thermo Scientific) cross-linking was performed according to the manufacturer’s protocol (with minor modifications) followed by desalting by using Zeba spin columns (Thermo Scientific).

Proteolytic Digestion and Peptide Cleanup.

Modified samples were precipitated by using acetone, and the resuspended samples were directly subjected to trypsin digestion. One reason in-solution digestion of the cross-linked samples was preferred after chemical cross-linking and quenching was that cross-linked species usually represent only very small fractions of total proteins, especially for protein complexes like PSII that contain at least 40 subunits per PSII dimer (21). Thus, the goal was to avoid over–cross-linking the sample, which could lead to protein aggregation and denaturation. In-solution digestions also avoided the loss of cross-linked products during the postgel handling, and reduced the artifacts of gel adduction.

MS Analysis.

The peptide mixture from trypsin digestion was dissolved in water with 0.1% formic acid. The peptide samples were analyzed by using our LC-MS proteomics workflow (53). The peptide sample was loaded onto an Ultimate 3000 Nano LC system (Thermo Scientific Dionex) coupled with an LTQ Orbitrap mass spectrometer (Thermo Fisher Scientific). The peptides were trapped by a guard column (Acclaim PepMap100, 100 µm × 2 cm, C18, 5 µm, 100 Å; Thermo Scientific Dionex) through which solvent A (water with 0.1% formic acid; Sigma-Aldrich) was pumped at 6 µL/min. The peptide mixture was fractionated on a custom-packed Michrom Magic C 18 RP column, as previously reported (53). The peptides were eluted at a flow rate of 260 nL/min, ramping a gradient from 5% to 60% solvent B (80% acetonitrile, 20% water, and 0.1% formic acid) in 110 min. The eluted samples were directly introduced into the mass spectrometer via a PicoView nano electrospray source (New Objective). Ion source and other parameters of the mass spectrometer were optimized by tuning with peptide standards. The mass spectrometer was operated in data-dependent mode by using previously reported parameters (53).

Database Search.

LC-MS data in Thermo Xcalibur .raw files were converted into mzXML and .mgf format by MM File Conversion from the MassMatrix package. Product-ion mass spectra were searched against the UniProt database to generate the protein list for cross-linked peptide identification. The cross-linked peptides were identified by using the search algorithm MassMatrix (54). For MassMatrix, each protein sequence pair was established and searched against all LC/MS data. The MassMatrix search parameters were as follows: variable modification, oxidation of Met; maximum number of variable PTM per peptide, 1; peptide tolerance, 15 ppm; MS/MS tolerance, 0.8 Da; mass type, monoisotopic; C13 isotope ions, yes; enzyme, trypsin; missed cleava+ges, 3; fixed modification, none; peptide length, from 3 to 50; cross-link search mode, exploratory; cross-link sites cleavability, noncleavable by enzyme; and maximum number of cross-links per peptide, 2. The search results were viewed using by XMAP (version 0.5.1; MassMatrix).