Introduction

The Photosystem II (PSII) complex, which functions as the light-driven water: plastoquinone oxidoreductase of oxygenic photosynthesis, is found in the thylakoid membrane system of cyanobacteria and chloroplasts. Crystal structures of isolated dimeric oxygen-evolving PSII complexes have been determined for the cyanobacteria Thermosynechococcus elongatus and Thermosynechococcus vulcanus (Kamiya and Shen 2003; Ferreira et al. 2004; Loll et al. 2005; Guskov et al. 2009; Umena et al. 2011). In the case of T. elongatus, each monomer is composed of 17 intrinsic and 3 extrinsic subunits and a variety of cofactors (Guskov et al. 2009). Recent studies have begun to reveal how PSII is assembled from its component parts. Current evidence suggests a modular assembly of PSII (Nixon et al. 2010; Boehm et al. 2011). First, a PSII reaction centre (PSII RC) complex is formed from PsbI-precursor D1 (Dobakova et al. 2007) and cytochrome b559-D2 (Komenda et al. 2004) sub-complexes. Then a CP47 module (Boehm et al. 2011) is attached to create the RC47 complex, after which a CP43 module (Boehm et al. 2011) is added to form the non-oxygen-evolving monomeric PSII complex. At this stage, the oxygen-evolving Mn4CaO5 cluster is assembled; the PsbO, PsbU and PsbV lumenal extrinsic proteins are attached and PSII can dimerise (reviewed by Komenda et al. 2012a). Removal of the C-terminal extension of the precursor D1 subunit (Komenda et al. 2007), which is required for assembly of the Mn4CaO5 cluster (Nixon et al. 1992), is initiated in vivo after formation of the PSII RC (Komenda et al. 2004).

The PSII complex is susceptible to light-induced damage, which can lead to chronic photoinhibition (Adir et al. 2003). Damaged subunits, predominantly the D1 subunit, are replaced during the so-called ‘PSII repair cycle’. Current models suggest that the damaged PSII complex partially disassembles to form the monomeric RC47 complex (Komenda et al. 2004); the damaged D1 protein is degraded by a hetero-oligomeric FtsH protease complex (Boehm et al. 2012a) and replaced by a newly synthesised copy of D1. Oxygen-evolving monomeric and dimeric complexes are then reformed as in PSII assembly (Komenda et al. 2012a).

A number of accessory factors have been identified in various types of PSII complex that are absent in the crystallised dimeric PSII complex (Nixon et al. 2010; Chi et al. 2012). Their functions are still unclear, but current evidence suggests roles in the assembly and repair of PSII or the optimisation of water oxidation. One of these, Psb28-1, encoded by sll1398 in Synechocystis sp. PCC 6803 (hereafter Synechocystis 6803), is known to bind to the cytoplasmic surface of PSII close to CP47 and PsbH (Dobakova et al. 2009) and in the vicinity of the HliB and HliC subunits that can attach to PSII in vivo (Promnares et al. 2006; Yao et al. 2007). Psb28-1 is mainly found in RC47 complexes in WT (Dobakova et al. 2009) and has been recently detected in isolated PSII core complexes containing the His-tagged Psb27 subunit (Liu et al. 2011). Analysis of a psb28-1 null mutant of Synechocystis 6803 has led to the hypothesis that Psb28-1 might be involved in the synthesis of chlorophylls and/or the apopolypeptides of the chlorophyll-binding proteins CP47 and PsaA/PsaB (Dobakova et al. 2009) although this phenotype appears to be dependent on the particular WT strain used (Sakata et al. 2013). In addition, Psb28-1 is involved in maintaining PSII activity during heat stress possibly at the level of repair (Sakata et al. 2013). Synechocystis 6803, like certain other cyanobacteria, also encodes a second Psb28 homologue, designated Psb28-2, encoded by slr1739, which has also been detected in RC47 complexes (Boehm et al. 2012b).

The structure of a C-terminally His-tagged derivative of Psb28-1 encoded by Synechocystis 6803 has been determined by nuclear magnetic resonance spectroscopy (NMR) (Yang et al. 2011). Here, we present the crystal structure of the single Psb28 homologue found in the thermophilic cyanobacterium T. elongatus which has also been reported to associate with PSII (Nowaczyk et al. 2012). As anticipated, the overall fold of Psb28 is conserved between the two species. However, our results suggest that Psb28-1 might actually exist as a dimer in vivo and that dimerization is impaired by addition of protein tags at the C-terminus.

Materials and methods

Cloning, expression and purification of Psb28

A PCR fragment containing the Psb28 gene of T. elongatus (tlr0493) was amplified from genomic DNA using Phusion polymerase (NEB, UK) and primers Psb28-BamHI-F (5′-TATATAGGATCCGGTGCAATGGCAGAAATT-3′) and Psb28-EcoRI-R (5′-TATATA GAATTCTTATCAAGAGTTCTCAGACTTCTGAAAGCCA-3′), purified, digested by BamHI/EcoRI and ligated (Quick Ligation Kit, New England Biolabs, UK) into a modified version of pRSET-A that directs the expression of a target protein with a thrombin-cleavable N-terminal His6-tag (Michoux et al. 2010). Soluble His6-tagged Psb28 was expressed in the KRX strain of Escherichia coli, purified using immobilised Ni2+-affinity chromatography and the His-tag removed by thrombin cleavage as described by Michoux et al. (2012). Size-exclusion chromatography was performed on a Superdex75 10/300 GL column (GE Healthcare) equilibrated with 50 mM Tris pH 8.0, 150 mM NaCl or 50 mM MES pH 6.5, 5 mM MgCl2, 5 mM CaCl2. The column was calibrated for each buffer with molecular weight markers for gel filtration chromatography from Sigma (BSA, carbonic anhydrase, cytochrome c and aprotinin).

Crystallization and structure solution

The protein lacking the hexahistidine tag was concentrated to 32 mg/ml and crystallisation screens were set up at 19 °C using a Mosquito® robot (TTP LabTech, UK). Rod-shaped Psb28 crystals appeared in sitting drops with the protein solution mixed with an equal volume of 20 % (v/v) PEG200, 20 % (w/v) PEG4000, 10 % (v/v) 2-propanol buffer and 100 mM HEPES sodium salt pH 7.5. A crystal was mounted in a loop and plunged into liquid nitrogen with no further cryoprotection. Data were collected on beamline I04-1 at Diamond Light Source on a Pilatus 6 M detector.

The structure was solved by molecular replacement in PHASER (McCoy et al. 2007). The NMR structure of Psb28-1 from Synechocystis (PDB code 2KVO) was truncated to match the T. elongatus sequence by CHAINSAW and the first model used as a search model. A partial solution with three copies showed a clear Psb28 dimer. A search with this dimer yielded a clear solution with two dimers in the asymmetric unit. The structure was completed and edited in Coot (Emsley et al. 2010), with cycles of refinement in REFMAC5 (Murshudov et al. 2011) and finally with Phenix (Adams et al. 2010). Non-crystallographic symmetry restraints were applied. The atomic model and structure factors have been deposited in the PDB under accession number 3ZPN.

Growth of Synechocystis strains and protein analysis

The glucose-tolerant wild type strain (WT-G) of Synechocystis 6803 (Williams 1988) and the previously constructed ΔCP43 (Vermaas et al. 1988), ΔCP43/CP47-His (Boehm et al. 2012b), ΔPSI (Shen et al. 1993), ΔPSI/ΔHliABCD (Vavilin et al. 2007) strains were used in this study. The Hlip- and Psb28-lacking derivatives of the ΔCP43 strain were obtained by the transformation of the ΔCP43 strain using genomic DNA isolated from previously described insertion mutants lacking HliA, HliB (Xu et al., 2002) and Psb28 (Dobakova et al. 2009). Similarly, the PSI-less strain lacking PsbH was obtained by transformation of the ΔPSI strain using genomic DNA isolated from the psbH deletion strain (Mayes et al. 1993). The C-terminally flagged strain was constructed by transformation of the WT-G strain by a synthetic construct bearing the nucleotide sequence of psb28-1 gene fused to a sequence encoding a 3 × flag sequence at the C-terminus followed by a zeocine-resistance cassette together with 300-bp upstream and 200-bp downstream region of the original psb28-1 gene. Strains were grown in liquid BG-11 mineral medium and maintained on solid BG-11 plates containing 1.5 % (w/v) agar, both containing 5 mM N-tris (hydroxymethyl)methyl-2-aminoethanesulfonic acid–KOH, pH 8.2, at a light intensity of 40 (for WT-G strains) or 5 (for PSI-less strains) μE m2 s−1 of white fluorescent light at 29 °C. The medium was supplemented with 5 mM glucose.

Two-dimensional protein analyses were performed by a combination of blue native and SDS-PAGE as described in Komenda et al. (2012b). 2D gels were stained by Sypro Orange, blotted onto PVDF membranes and probed with specific antibodies against D1, Psb28-1 and Psb28-2 of Synechocystis 6803 (Komenda et al. 2012b; Boehm et al. 2012b).

Results and discussion

Sequence alignment

In contrast to Synechocystis 6803 which contains two Psb28 homologues, there is only one Psb28 homologue encoded in the T. elongatus genome (Fig. S1). Sequence comparisons revealed 59 % identity (78 % similarity) and 33 % identity (57 % similarity) between Psb28 from T. elongatus and Psb28-1 and Psb28-2 from Synechocystis 6803, respectively (Fig. 1). The full length gene tlr0493 encoding Psb28 was expressed in E. coli as an N-terminal His-tagged derivative (Fig. S2a). After removal of the His-tag by thrombin cleavage (Fig. S2b), to leave two additional Gly-Ser residues at the N-terminus, and concentrating to 32 mg/ml, the protein was crystallised in two crystal forms only one of which was suitable for structure determination. Size-exclusion chromatography indicated that the E. coli-expressed Psb28 used for crystallisation could exist as monomers or dimers depending on the buffer (Fig. S2c and d).

Fig. 1
figure 1

Sequence alignment of Psb28 from T. elongatus (tlr0493) and Psb28-1 (sll1398) and Psb28-2 (slr1739) from Synechocystis 6803. Residues involved in salt bridges between different chains in putative dimer are highlighted in green and cyan. The picture was created by ESPript server (Gouet et al. 1999) and β-hairpins (⊃) were detected by ProMotif (Hutchinson and Thornton 1996)

Full size image

Structure description

The best X-ray diffraction data sets were collected for crystals obtained in 20 % (v/v) PEG200, 20 % (w/v) PEG4000, 10 % (v/v) 2-propanol buffer and 100 mM HEPES sodium salt pH 7.5 (Fig. S3a) and the structure was solved by molecular replacement using the NMR structure of Psb28-1 from Synechocystis 6803 (PDB: 2KVO); Table 1 shows the data collection and refinement statistics. The structure reveals that β-strands and α-helices account for 44 and 20 % of Psb28, respectively (Fig. 2). Two antiparallel β-sheets (A and B) are composed of β-strands, β1-4-5 and β2-3-6-7, with β-hairpins connecting β2 and β3, β4 and β5 and β6 and β7. One inverse γ-turn is present (Arg66Asp67Val68) between β5 and β6 (Fig. 1). Two α-helices are present, one short consisting of five residues (Lys42-Glu46) and one longer of 17 residues (Pro89-Ser105). The former is located between β3 and β4 and the latter is close to the C-terminus. The first 2 and last 6 residues are missing from electron density so that only residues 3–110 were modelled.

Table 1 Data collection and refinement statistics for the Psb28 crystal structure
Full size table
Fig. 2
figure 2

Structural features of Psb28 from T. elongatus. a 3D fold of protein b its topological cartoon representation and c protein–protein interactions in dimeric Psb28 from T. elongatus. Salt bridges and hydrogen bonds are shown as black and green dashed lines, respectively. For full list of residues involved in H-bonds refer to Table S1

Full size image

Comparison with the solution structure of Synechocystis Psb28-1

Structural comparison between the Psb28 crystal structure presented here, and Psb28-1 revealed a high overall similarity with a pairwise Cα RMSD of 1.6 Å as calculated by SSM (Krissinel and Henrick 2004). As shown in Fig. S4, the most significant difference between both the proteins lies in the loop between β-strands 3 and 4 (see below). In addition, Psb28 from T. elongatus lacks a 310 helix composed of residues 39–41 (2KVO original numbering) and the second 310 helix of Psb28-1 is replaced by a five-residue long α-helix (Lys42-Glu46). Only one significant hit, Psb28-1, with a Z score of 13.8, was detected by DALI analysis (Holm and Rosenstrom 2010).

In the crystal structure, four molecules are present in the asymmetric unit (Fig. S3b). As suggested by PISA (Krissinel and Henrick 2007), and in contrast to the solution structure of His-tagged Psb28-1 from Synechocystis 6803, Psb28 from T. elongatus forms two dimers composed of chains A–B and C–D. The interface is created between long C-terminal α-helices and β-strands 2 with numerous hydrogen bonds between chains (Table S1). Two salt bridges (Asp20-Arg22 and Arg27-Asp39) further stabilize the dimer (Fig. 2c). Sequence alignments suggest that a salt bridge may also be present between Glu17 and Arg19 of Psb28-1 from Synechocystis 6803. Our structural analysis also suggests that a second salt bridge might form between Lys24 and Glu37 of Psb28-1 despite the fact that the loop with Lys24 adopts different conformations to the corresponding region in the protein from the thermophilic cyanobacterium. In fact, the flexibility of this region may result from the lack of the stabilizing effect of an intradimeric salt bridge. Based on a structural alignment of the C-terminally tagged Psb28-1 monomer and the Psb28 dimer described here, we speculate that the presence of the His-tag at the C-terminus of Psb28 might impede dimerization (Fig. S5).

Analysis of conserved residues and implications for the structure of Psb28-2

Using the crystal structure of Psb28 from T. elongatus described here and sequences of cyanobacterial, algal and plant Psb28-1 and cyanobacterial Psb28-2, we mapped conserved residues on the protein structure using a Consurf analysis (Fig. 3a). This analysis revealed that the conserved hydrophobic residues shown in Fig. 3b create a hydrophobic core inside Psb28 and that most conserved residues are located in two clusters of charged residues. One of them, a patch of acidic residues, forms a β-turn between β4 and β5, whereas the other, composed of basic residues, is found in the middle part of the long C-terminal α-helix. Interestingly, these two clusters interact with each other by means of intramolecular hydrogen bonds and salt bridges (Fig. 3c). By bridging two distant parts of the protein. these interactions seem to enable the formation of the hydrophobic core and probably stabilize the protein structure. Given the relatively high sequence similarity between Psb28-1 and Psb28-2 and that sequences of both the protein sub-families were taken into consideration, this analysis supports the idea that Psb28-2 adopts a very similar 3D fold to Psb28-1.

Fig. 3
figure 3

Analysis of conserved residues distribution between cyanobacterial, algal and plant Psb28-1 and cyanobacterial Psb28-2 as calculated by Consurf (Ashkenazy et al. 2010). a Side chains of the most conserved residues are shown as sticks. b Side chains (shown as sticks) of conserved residues lining the hydrophobic core of the protein. c Interactions between the two absolutely conserved regions found in both the Psb28-1 and Psb28-2 with lengths given in Å. The conservation colouring scale spans from teal for variable to salmon for the most conserved residues. Compare with sequence alignment from Fig. 1

Full size image

Analysis of Psb28-1 and Psb28-2 in Synechocystis 6803

To explore the possibility that Psb28-1 and Psb28-2 of Synechocystis 6803 might exist in different oligomeric states in vivo, we separated detergent-solubilised membrane protein extracts by BN-PAGE in the first dimension and by SDS-PAGE in the second dimension, followed by immunoblotting using antibodies specific for Psb28-1 and Psb28-2 (Fig. 4). Using this technique, three distinct populations of Psb28-1 were detected. As reported previously some of the Psb28-1 co-migrated with the RC47 complex that accumulates in ΔCP43 mutants (Dobakova et al. 2009; Boehm et al. 2012b), but most (band 2 in Fig. 4) migrated with an apparent molecular mass distinctly larger than that of ‘unassembled’ Psb28-2 present in all strains but most abundant in the ΔPsb28-1 strain (Fig. 4). Some Psb28-1 did co-migrate with Psb28-2 (band 1 in Fig. 4), but its abundance was variable between samples and sometimes not detected (Boehm et al. 2012b). Control immunoblots confirmed that band 2 of Psb28-1 did not co-migrate with the CP47 subunit (data not shown), and its migration was unaffected in strains lacking either PsbH or four members of the Hli protein family (HliA-D), two of which are known to attach to RC47 under high-light conditions (Fig. S6) (Yao et al. 2007). Together these data suggest that band 2 is probably an oligomer. According to the crystal structure (Fig. S5), dimerisation of Psb28-1 is potentially impaired by addition of a tag at the C-terminus of Psb28-1. Consistent with this hypothesis, a C-terminal 3xFlag-tag derivative of Psb28-1 expressed in Synechocystis 6803 now migrated mainly as a monomer (band 1) in the same region of the BN gel as Psb28-2 (Fig. 5).

Fig. 4
figure 4

The occurrence of the monomeric and dimeric forms of Psb28-1 protein in various CP43-less strains. Membranes isolated from the strains cultivated under 10 μmol photons m−2 s−1 in the presence of 5 mM glucose were analysed by 2D CN/SDS-PAGE. The gel was stained by Sypro Orange and then blotted to PVDF membrane which was probed by antibodies specific to D1, D2 and Psb28-1. The dimeric (2) and monomeric (1) forms of Psb28 are designated by vertical arrows. For comparison, signals of Psb28-2 protein corresponding to the monomer obtained using a specific antibody are also shown and designated by horizontal arrows. The loaded samples contained 5 μg of Chl

Full size image
Fig. 5
figure 5

The occurrence of the monomeric form of Psb28-1 in the strain expressing the C-terminally Flagged Psb28-1 protein. Membranes isolated from both the strains were analysed by 2D BN/SDS-PAGE. The gel was stained by Sypro Orange and then blotted to PVDF membrane which was probed by antibodies specific to D1 and Psb28-1. The dimeric (2) and monomeric (1) forms of Psb28-1 are designated by vertical arrows. The loaded samples contained 5 μg of Chl

Full size image

Conclusion

In summary, we have determined the crystal structure of Psb28 from T. elongatus and have provided evidence that Psb28-1, but not Psb28-2, of Synechocystis 6803 forms a higher order structure, most likely a dimer, in vivo. The physiological importance of Psb28-1 dimerisation is currently unknown. Our data provide a structural model for how Psb28-1 might oligomerise in vivo which can be tested in the future by mutagenesis. The crystal structure determined here might prove to be useful in determining the site of attachment of Psb28 to PSII determined through cross-linking experiments.