Identification of an Atypical Membrane Protein Involved in the Formation of Protein Disulfide Bonds in Oxygenic Photosynthetic Organisms*

The evolution of oxygenic photosynthesis in cyanobacteria nearly three billion years ago provided abundant reducing power and facilitated the elaboration of numerous oxygen-dependent reactions in our biosphere. Cyanobacteria contain an internal thylakoid membrane system, the site of photosynthesis, and a typical Gram-negative envelope membrane system. Like other organisms, the extracytoplasmic space in cyanobacteria houses numerous cysteine-containing proteins. However, the existence of a biochemical system for disulfide bond formation in cyanobacteria remains to be determined. Extracytoplasmic disulfide bond formation in non-photosynthetic organisms is catalyzed by coordinated interaction between two proteins, a disulfide carrier and a disulfide generator. Here we describe a novel gene, SyndsbAB, required for disulfide bond formation in the extracytoplasmic space of cyanobacteria. The SynDsbAB orthologs are present in most cyanobacteria and chloroplasts of higher plants with fully sequenced genomes. The SynDsbAB protein contains two distinct catalytic domains that display significant similarity to proteins involved in disulfide bond formation in Escherichia coli and eukaryotes. Importantly, SyndsbAB complements E. coli strains defective in disulfide bond formation. In addition, the activity of E. coli alkaline phosphatase localized to the periplasm of Synechocystis 6803 is dependent on the function of SynDsbAB. Deletion of SyndsbAB in Synechocystis 6803 causes significant growth impairment under photoautotrophic conditions and results in hyper-sensitivity to dithiothreitol, a reductant, whereas diamide, an oxidant had no effect on the growth of the mutant strains. We conclude that SynDsbAB is a critical protein for disulfide bond formation in oxygenic photosynthetic organisms and required for their optimal photoautotrophic growth.

The formation of disulfide bonds, required for the functions of cysteine containing extracellular proteins and exoplasmic domains of membrane proteins, is a catalyzed process in all organisms (1,2). The core pathways for disulfide bond forma-tion in the periplasm of prokaryotes and lumen of the endoplasmic reticulum (ER) 2 in eukaryotes involve thiol-disulfide oxidoreductases characterized by a CXXC active site embedded in a domain with structural similarity to thioredoxins. In prokaryotes, formation of disulfide bonds in substrate proteins is catalyzed by a minimum of four Dsb proteins, DsbA, DsbB, DsbC, and DsbD (1). DsbA, a 21-kDa soluble periplasmic protein with a canonical CXXC motif in its active site, acts as a disulfide bond carrier and rapidly oxidizes cysteine residues in protein substrates (3). In eukaryotes, this process is catalyzed by protein-disulfide isomerase (PDI) (4). These disulfide bond carriers receive oxidizing equivalents from membrane-localized protein thiol oxidoreductases, DsbB in prokaryotes (5), and Ero1 and Erv2 in eukaryotes (6,7). In Escherichia coli, two additional proteins, a soluble thioredoxin-like protein DsbC and a cytoplasmic membrane protein DsbD, are involved in the isomerization of incorrectly paired cysteines (8,9). The active site cysteines of DsbC are maintained in a functional reduced state by DsbD that shuttles the reducing equivalents from NADPH and thioredoxin reductase (1).
Despite the basic requirement of thiol-disulfide oxidoreductases, not all prokaryotes appear to have complete sets of proteins involved in disulfide bond formation as delineated in E. coli. DsbA and DsbB homologues are found in most Gramnegative bacteria. In contrast, DsbC and DsbD appear to be restricted to the ␤and ␥-subdivisions of eubacteria (1). Grampositive bacteria contain few extracytoplasmic proteins with disulfide bonds. Nonetheless, genetic studies have shown that disulfide bond formation is a catalyzed process in these organisms and genes essential for disulfide bond formation have been identified (10). Cyanobacteria are oxygenic photosynthetic bacteria with a typical Gram-negative envelope membrane system as well as elaborate internal thylakoid membranes that include a lumen, the site for photosynthetic water oxidation. Proteomic analyses of the extracytoplasmic space of the cyanobacterium Synechocystis 6803 have identified several cysteine containing proteins (11,12). However, the existence of a system for disulfide bond formation in either the periplasm or the thylakoid lumen of cyanobacteria remains unknown. In the present work, we report the identification and characterization of SynDsbAB, a novel fusion protein in Synechocystis 6803, which is required for protein disulfide bond formation.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-Synechocystis 6803 cultures were grown at 30°C in BG11 medium (13) buffered with 10 mM TES-KOH (pH 8.2) with constant shaking. Illumination was at 30 mol of photons m Ϫ2 s Ϫ1 unless otherwise noted. Mutant strains were supplemented with either 25 g/ml of erythromycin (Em) or 40 g/ml of kanamycin (Km). Cell growth was monitored by measuring OD at 730 nm on a Quant plate reader (Bio-Tek Instruments, Winooski, VT). When indicated, glucose (5 mM), dithiothreitol (DTT) (0.2 mM), or diamide (10 M) was added to the cultures. Growth of cyanobacteria under manganese depletion was carried out essentially as described (14).
DNA Manipulations and Plasmid Constructions-Details of the bacterial strains and plasmids used in this study are listed in Table 1. Plasmids pCR2.1 (Invitrogen) and pUC119 (15) were used as basic cloning vectors. Standard cloning methods were used as described (16). Gene knock-out strains were made using a deletion inactivation approach. The primers used in this study are listed in Table 2. The SyndsbAB gene was PCR amplified from Synechocystis 6803 genomic DNA using primers Dsb1p and Dsb2p. The resulting PCR product (2.0 kb) was cloned into pCR2.1 resulting in pSL1889. A transcription terminator-less Em r antibiotic cassette (0.85-kb fragment) was PCR amplified from plasmid pNE131 (gift from Dr Howard Berg) using prim-ers Em1p and Em2p. SalI restriction sites were engineered at the 5Ј-end of primers to facilitate cloning. The resulting 0.85-kb PCR products were cloned into pCR2.1, digested with SalI, and cloned into SalI-digested pUC4K that replaced the Km r cassette with the Em r cassette resulting in pSL1875. This Em r cassette excludes the transcriptional termination signals present in the flanking region of pNE131.
To construct plasmid pSL1878 (missing the coding region of the SyndsbAB gene), the 5Ј (upstream) and 3Ј (downstream) untranslated regions of the SyndsbAB gene were PCR amplified separately. The upstream region (ϳ0.6 kb) was amplified using primers Dsb1p and Dsb3p that introduced EcoRI and BamHI sites at the two ends, respectively. The downstream region (ϳ0.5 kb) was amplified using primers Dsb2p and Dsb5p that introduced HindIII and BamHI sites at the two ends, respectively. The resultant PCR products corresponding to the upstream and downstream regions of the SyndsbAB gene were each cloned into pCR2.1, resulting into pSL1827 and pSL1828, respectively. These plasmids were then digested with EcoRI/ BamHI (pSL1827) and BamHI/HindIII (pSL1828) and were ligated into EcoRI/HindIII-digested pUC119 resulting in pSL1878 missing the entire coding region of the SyndsbAB gene. The Em r cassette was isolated from pSL1875 and cloned into the unique BamHI site of pSL1878 resulting in plasmid pSL1881. The orientation of the Em r cassette was determined by PCR using primers Dsb1p and either Em1p or Em2p.
To construct plasmid pSL1880 (⌬SyndsbA), a region encoding the first 165 amino acid residues of SynDsbAB was PCR amplified from Synechocystis 6803 genomic DNA using primers Dsb1p and Dsb7p (introducing a BamH1 restriction site), and cloned into pCR2.1 resulting in pSL1829. The EcoRI/BamHI fragment from pSL1829 and the BamHI/HindIII fragment from pSL1828 were ligated into EcoRI/HindIII-digested pUC119 resulting in pSL1880. The Em r cassette was cloned into the unique BamHI site resulting in plasmid pSL1883. The orientation of the Em r cassette was again determined by PCR.
To construct plasmid pSL1879 (⌬SyndsbB), a downstream region encoding the last 173 amino acids of SynDsbAB was PCR amplified from Synechocystis 6803 genomic DNA using primers Dsb2p and Dsb8p and cloned into pCR2.1 resulting in pSL1830. An EcoRI/BamHI fragment from pSL1827 and a BamHI/ HindIII fragment from pSL1830 were ligated into EcoRI/ HindIII-digested pUC119 resulting in pSL1879. The Em r cassette was cloned at the unique BamHI site of pSL1879 resulting in plasmid pSL1882. The orientation of the Em r cassette was determined by PCR.
To construct a vector to express the mntCAB-phoA fusion reporter at the cryptic psbA1 genomic location in Synechocystis 6803, we amplified the psbA1 gene using four primers. Primers Psb1p and Psb2p amplified the N-terminal half of the gene and introduced two restriction sites, EcoRI and PstI, respectively. A second pair of primers (Psb3p and Psb4p) amplified the C-terminal half and introduced PstI, SalI, and HindIII restriction sites. The PCR products (ϳ0.5 kb) were cloned into pCR2.1 resulting in plasmids SL1800 and SL1801. These plasmids were then digested with either EcoRI/PstI (pSL1800) or PstI/ HindIII (pSL1801) and the resulting psbA1 fragments were cloned into EcoRI/HindIII-digested pUC119 resulting in pSL1805. A Km r antibiotic cassette (1.3 kb) from pUC4K was ligated into the PstI site of pSL1805 resulting in pSL1809. XhoI-digested DNA fragments containing mntCAB and phoA from pSL1124, pSL1125, and pSL1127 (17) were cloned at the SalI site of pSL1809 resulting in pSL1886, pSL1887, and pSL1888, respectively.
Generation of Synechocystis 6803 Mutant Strains-To generate strains missing either the entire coding region or specific domains encoded by the SyndsbAB gene, plasmids pSL1881 (missing the entire ORF), pSL1882 (missing SyndsbB), or pSL1883 (missing SyndsbA) were used to transform the Synechocystis 6803. Transformants were selected on Em (25 g/ml) under 30 mol of photons m Ϫ2 s Ϫ1 light. To test for E. coli alkaline phosphatase (AP) activity in wild type and ⌬SyndsbAB strains of Synechocystis 6803, the mntCAB-phoA fusion reporter construct from pSL1888 was introduced at the psbA1 locus of Synechocystis 6803. Transformants were selected on Km (40 g/ml). Segregation was verified by PCR using primers Psb1p and Psb4p.
Measurement of Alkaline Phosphatase Activity-AP activity in E. coli was measured using p-nitrophenyl phosphate (Sigma) as described (18). To assay E. coli AP activity expressed in Synechocystis 6803, 2-ml cultures (1 ϫ 10 8 cells/ml) were harvested by centrifugation at 6,000 ϫ g for 5 min. The cell pellet was resuspended in 0.2 M Tris-HCl (pH 8.5), 2 mM MgCl 2 and p-nitrophenyl phosphate (Sigma) were added to 4 mM. AP activity was determined as described (19). The values shown represent the average of at least three separate experiments.
Computational Analysis-Hydropathy analysis of SynDsb-AB was performed using the TopPred program (20). Subcellular localization of the protein was predicted using the TargetP program (21). Secondary structure prediction was done using the PSIPRED program (22). All conserved amino acid residues discussed in this work are based on their positions in SynDsbAB of Synechocystis 6803, unless otherwise stated.

RESULTS
Identification of the SyndsbAB Gene-We utilized E. coli AP translationally fused to Synechocystis 6803 MntB protein to examine if a system for disulfide bond formation is present in Synechocystis 6803. E. coli AP, encoded by the phoA gene, is a periplasmic protein that requires the presence of a disulfide bond for its activity (3,5). The mntB gene is co-transcribed with the mntA and mntC genes (14). The mntCAB operon is tightly controlled by the availability of manganese in the BG11 growth medium and transcripts from this operon are transcribed only under manganese limiting conditions (14). The three mntCAB-phoA fusion reporter constructs (pSL1886, pSL1887, and pSL1888) used in this study results in localization of AP into either the cytoplasm or periplasm of E. coli (17). These mnt-CAB-phoA constructs were individually introduced into Synechocystis 6803 by transformation to replace the cryptic psbA1 gene by the reporter gene. The resulting Synechocystis 6803 strains hereafter named SynAS1, SynAS2, and SynAS3 contain E. coli AP localized either in the cytoplasm (SynAS1), anchored in the cytoplasmic membrane facing the cytoplasm (SynAS2) or the periplasm (SynAS3). When SynAS1, SynAS2, and SynAS3 were grown in BG11 containing manganese, only low levels of AP activity were detected (Fig. 1). Importantly, the endogenous AP, which is expressed only under phosphate limiting conditions (19), did not show any activity under our experimental conditions. When these strains were grown in manganese-depleted BG11, the activity of AP was unaltered in wild-type Synechocystis 6803 and in the SynAS1. In contrast, we found a significant increase in the activity of AP in the SynAS3 when grown in manganese-depleted BG11. Some AP activity could also be observed in the SynAS2. This is related to a limited degradation of hybrid proteins during translocation across the membrane that allows the retention of AP in the periplasm (17).
To identify cellular components involved in disulfide bond formation in Synechocystis 6803, we used sequence similarity searches using amino acid sequences of E. coli Dsb proteins. Such searches failed to identify proteins with similarity to DsbA and DsbB. However, limited sequence similarity to DsbC and DsbD with three proteins viz, Sll0621, Sll0686, and Slr0313 were found in Synechocystis 6803 genome. It has been reported that components involved in disulfide bond formation in both prokaryotes and eukaryotes are analogous to each other and show structural similarities despite a lack of primary sequence similarity (23). Therefore, we searched for proteins containing the CXXC motif in the PEDANT data base to identify possible component(s) in Synechocystis 6803. Among 270 such proteins identified, we examined all 132 hypothetical proteins, and one such protein, hereafter named SynDsbAB, showed structural similarities to proteins involved in disulfide bond formation in prokaryotes and eukaryotes (Fig. 2). SynDsbAB contains 325 amino acids with a predicted mass of 35 kDa and pI of 5.11. Orthologs of SynDsbAB are present in most cyanobacteria with fully sequenced genomes (supplemental Table S1) including Gloeobacter violaceus PCC 7421, which lacks a differentiated thylakoid membrane system and represents an early branching lineage in cyanobacteria (24). Notably, similar orthologs with the putative chloroplast target signal sequence are found in Chlamydomonas (25) and in higher plants including Arabidopsis (Fig. 2).
Domain Structure of SynDsbAB-SynDsbAB consists of two distinct domains: an N-terminal membrane domain (SynDsbB) and a C-terminal soluble domain (SynDsbA) (Fig. 2). SynDsbB contains five predicted transmembrane segments (TMS) and has two pairs of cysteines, including a canonical CXXC conserved in all cyanobacterial and plant sequences. The strategically located pair of cysteines (Cys 44 and Cys 55 ) in the periplasmic loop connecting the 1st and 2nd TMS as well as residues Cys 138 and Cys 141 between the 3rd and 4th TMS in the SynDsbB domain are located similarly to those found in other oxidoreductases (23). A similar domain structure has also been predicted for VKORC1 (26), a subunit of the VKOR holoenzyme that catalyzes the conversion of vitamin K epoxide to vitamin K and subsequently to vitamin KH 2 . Vitamin KH 2 is required for post-translational modification of glutamate to ␥-carboxyl glutamate in blood clotting proteins by ␥-glutamyl carboxylase (27).
The SynDsbA domain has a typical TRX-fold motif (␤ 1 ␣ 1 ␤ 2 -␣ 2 -␤ 3 ␤ 4 ␣ 3 ). It contains two pairs of cysteines that are absolutely conserved in cyanobacteria and Arabidopsis. These include a canonical CXXC found between ␤ 1 and ␣ 1 (similar to PDI and DsbA) and residues Cys 268 and Cys 280 localized between ␤ 2 and ␣ 2 . Interestingly, the second pair of cysteines present in the SynDsbA domain is missing in PDI and DsbA. These cysteines are replaced by an additional ␣-helix (␣*) in DsbA. The presence of the ␣* helix in DsbA has been suggested to prevent unwanted interaction between DsbA and other cysteine(s) containing proteins (1). In addition, amino acids known to be important in the function of DsbA, PDI, Ero1p, and Erv2 are conserved in SynDsb-AB. For example, a cis-proline residue, highly conserved among the thioredoxin superfamily members, is localized close to the active-site cysteines of DsbA. Replacement of this proline by other amino acids results in the formation of conjugated complexes with either DsbA substrates or with DsbB (28,29). Sequence alignment of SynDsbAB from cyanobacteria and Arabidopsis show the presence of a conserved  JUNE 6, 2008 • VOLUME 283 • NUMBER 23

JOURNAL OF BIOLOGICAL CHEMISTRY 15765
proline between ␣ 2 and ␤ 3 similar to that found in DsbA and PDI (Fig. 2). We have also identified two conserved aromatic amino acids, Trp 244 and Trp 291 (supplemental Fig. S1), previously suggested to aid in the binding and orientation of the cofactor FAD in Ero1 and Erv2 (23). The FAD cofactor plays an important role in the transfer of reducing power to electron acceptors. We also found a conserved Ser 57 in SynDsbB (Fig. 2), known to be conserved in all VKORC1 subunits (26). SynDsbAB Complements E. coli Dsb Null Strains-To test the role of SynDsbAB in disulfide bond formation, we performed functional complementation assays of E. coli strains defective in disulfide bond formation. The ⌬dsbA and ⌬dsbB strains of E. coli exhibit a number of phenotypes caused by a generalized inability to form disulfide bonds. On minimal media, both strains are non-motile due to their inability to introduce a single disulfide bond critical for proper folding of the flagellar motor protein FlgI (30). Fig. 3A shows the results of a complementation assay using pSL1889, a plasmid that expresses the SyndsbAB gene. Wild type cells were motile after 24 h incubation on minimal plates containing 0.3% agar, whereas both ⌬dsbA and ⌬dsbB cells were non-motile. Transformation of ⌬dsbA and ⌬dsbB strains with pSL1889 resulted in the restoration of motility in these strains. In addition, expression of the SyndsbAB gene restored the AP activity in both ⌬dsb strains to levels found in wild type cells (data not shown). However, expression of the SyndsbAB gene did not complement the functions of DsbC and DsbD (supplemental Table S2). In addition, expression of the genes encoding Slr0313, Sll0621, and Sll0686 proteins with sequence similarity to DsbC and DsbD were unable to complement the function of any Dsb protein (supplemental Table S2).
Next, we tested the ability of individual domains of SynDsbAB to complement the ⌬dsbA and ⌬dsbB strains. To express SyndsbB, we replaced the region encoding amino acids 166 to 325 of SynDsbAB with the Em r cassette. The expressed SynDsbB domain has four TMS segments as well as both cysteine pairs known to be important for the activity of thiol oxidoreductases. To express SyndsbA, we replaced the region encoding the initial 153 amino acids of SynDsbAB with the Em r cassette, which allows transcription downstream of the insertion. We included the fifth TMS segment in the SynDsbA domain so that translocation of SynDsbA into the extracytoplasmic space would occur. The ⌬dsbA and ⌬dsbB strains were both transformed with either SL1882 (⌬SyndsbB) or SL1883 (⌬SyndsbA) and examined for the restoration of motility (Fig.  3B). SynDsbA was able to completely restore motility (Fig. 3B) and AP activity (data not shown) in the ⌬dsbA strain. The ability of this domain was specific to the function of DsbA as it was unable to restore the motility of the ⌬dsbB strain. However, SynDsbB did not restore the motility of either ⌬dsbA or ⌬dsbB strains. To assess the possibility that the absence of the fifth TMS is linked to the inability of SynDsbB to restore motility in ⌬dsbB cells, we constructed two additional plasmids containing SyndsbB that expressed either the first 195 amino acids (including all five transmembrane helices) or the first 244 amino acids (without the CXXC motif of SynDsbA). Neither of these constructs could restore motility of the ⌬dsbB strain (data not shown).
Photoautotrophic Growth of SyndsbAB Mutant Strains Is Significantly Impaired-The function of SynDsbAB in Synechocystis 6803 was studied by creating targeted deletions in the SyndsbAB gene. The entire gene or regions encoding individual domains were deleted and replaced with the Em r cassette (Fig. 4A). Details of the cloning process for creating targeted deletions in the SyndsbAB gene are provided in supplemental Fig. S2. ⌬SyndsbAB was created by PCR amplification of the upstream and downstream untranslated regions of the SyndsbAB gene. ⌬SyndsbA was created by combining the upstream region of the SyndsbAB gene to the region encoding the SynDsbB domain. ⌬SyndsbB was created by combining the  Complete segregation of Synechocystis 6803 strains was confirmed by PCR and restriction digestion with BamHI. The SyndsbAB gene does not contain BamHI sites, whereas the mutated genes have two BamHI sites introduced by insertion of the Em r cassette as described under "Experimental Procedures." As seen in Fig. 4B, digestion of PCR products obtained from mutant strains with BamHI resulted in three bands corresponding to the Em r cassette (0.85 kb) and the two flanking regions. The size of upstream and downstream flanking regions in ⌬SyndsbAB was ϳ0.6 and ϳ0.5 kb, respectively (see also supplemental Fig. S2). In the ⌬SyndsbA and ⌬SyndsbB mutant strains, the ϳ1.0-kb fragment corresponds to a flanking region and one of the SynDsb domains, whereas the smaller fragments correspond to the opposite flanking region as obtained in ⌬SyndsbAB (supplemental Fig. S2). These null strains will be referred to as ⌬SyndsbAB (missing both domains), ⌬SyndsbA (missing SynDsbA domain), and ⌬SyndsbB (missing SynDsbA domain). All mutant strains showed significant growth impairment when compared with the wild type strain under photoautotrophic conditions at normal light intensity (30 mol of photons m Ϫ2 s Ϫ1 ) (Fig. 5A). Changing the light intensity to either 100 or 5 mol of photons m Ϫ2 s Ϫ1 did not improve growth of these mutant strains. In fact, the doubling time of the mutants increased under high light and decreased under low light when compared with growth under normal light intensities (data not shown). This suggests that growth impairment observed in these strains is not dependent upon light intensity. Glucose can often supplement growth defects in Synechocystis 6803 mutant strains. When glucose (5 mM) was included in the growth medium, all mutant strains died rapidly (Fig. 5B). Growth under high CO 2 (3%) as well as inclusion of NaHCO 3 in the BG11 medium did not restore growth of these mutant strains (data not shown).

Growth of Mutant Strains under Different Redox Conditions-We
next tested the effects of DTT, a membrane-permeable reductant and diamide, a membrane-permeable oxidant, on the growth characteristics of SyndsbAB mutant strains. DTT (0.2 mM) was added directly to cells in BG11 to test the sensitivity of mutant strains to reductant. The mutant strains were hypersensitive to DTT and within 24 h, all cells were dead (Fig. 5C). In contrast, addition of DTT did not adversely affect the growth of wild type Synechocystis 6803. The sensitivity of SyndsbAB mutant strains to reductant is similar to those found in E. coli and eukaryotes (5,6). In contrast, we found that addition of diamide (10 M) did not have any appreciable impact on the growth of mutant strains (Fig. 5D). This is in marked contrast to mutant strains of E. coli and yeast, where addition of oxidants had an appreciable improvement in the growth pattern of mutant strains defective in disulfide bond formation (5,6).
Alkaline Phosphatase Activity in SyndsbAB Mutant Strains-The role of SynDsbAB on the activity of E. coli AP expressed in cyanobacteria was determined by transformation of the SyndsbAB mutant strains by plasmid pSL1888. Plasmid pSL1888 contains the E. coli phoA gene translationally fused to the mntCAB operon (17). This construct allows for the controlled expression of E. coli AP and its localization to the periplasm of Synechocystis 6803. We expected the fusion proteins to be responsive to external manganese concentrations, because the Synechocystis 6803 mntCAB operon is expressed only under manganese limiting conditions (14). As shown in Fig. 6, cells grown in manganese replete BG11 show low levels of AP activity. In manganese-depleted BG11, AP activity in pSL1888 transformed wild type Synechocystis 6803 increased significantly. In contrast, the activity of the E. coli AP in the three mutant strains of SyndsbAB increased but to significantly lower levels compared with wild type. The low level of AP activity in all three mutant strains is most likely due to spontaneous formation of disulfide bonds in the AP enzyme in the oxygenrich redox environment of cyanobacterial cells.

DISCUSSION
Disulfide bond formation in both prokaryotes and eukaryotes share a common mechanistic process involving two proteins; a membrane-associated oxidoreductase that generates disulfide bonds and a soluble oxidoreductase that carries oxidizing equivalents from the membrane-associated oxidoreductase to substrates. Here, we report a novel protein, SynDsbAB, which exhibits the properties of both enzymes, and is essential for the formation of disulfide bonds in the extracytoplasmic spaces of the cyanobacterium Synechocystis 6803. To our knowledge, this is the first time that a disulfide bond formation system analogous to those present in E. coli and yeast has been identified and characterized in an oxygenic photosynthetic organism. Loss of SynDsbAB renders Synechocystis 6803 cells hypersensitive to perturbations in redox potential induced by the reductant DTT and glucose. Importantly, the enzymatic activity of E. coli AP expressed in Synechocystis 6803, which is absolutely dependent upon the correct formation of disulfide bond, is significantly reduced in Synechocystis 6803 strains lacking Syn-DsbAB. Furthermore, expression of the SyndsbAB gene in ⌬dsb strains of E. coli restores the ability to form disulfide bonds in the periplasm. These findings together suggest that SynDsbAB is central to the formation of disulfide bonds in the cyanobacterial periplasm.
A clue to the potential function of SynDsbAB in cyanobacteria came from an analysis of its secondary structure. The SynDsbB domain shares significant structural similarity to proteins that generate disulfide bonds in the periplasm of E. coli viz. DsbB (5), and in the lumen of ER in eukaryotes, viz. Ero1 and Erv2 (6, 7) as well as VKORC1 (Fig. 2). Indeed, our results show that SynDsbAB can complement the function of DsbB. In contrast, despite a strong similarity with VKORC1 (26), we were unable to detect any VKORC1 activity in Synechocystis 6803 (data not shown). Similarly, the SynDsbA domain shares significant structural similarity to the disulfide carrier proteins, DsbA in E. coli and domains of PDI in eukaryotes ( Fig. 2 and 4). Our results show that SynDsbAB can functionally complement the function of DsbA. When individual domains (i.e. either SynDsbB or SynDsbA) were expressed in E. coli, we found that only SynDsbA was able to complement the function of DsbA. Furthermore, SynDsbAB could not complement the function of DsbC and DsbD as determined by a copper toxicity test (supplemental Table S2). Synechocystis 6803 contains three genes (sll0621, sll0686, and slr0313) that encode proteins with sequence similarity to DsbC and DsbD, each with multiple cysteines. Expression of these genes also did not complement the function of Dsb proteins in ⌬dsb strains of E. coli (supplemental Table S2). These results together suggest that SynDsbAB acts catalytically and that the restoration of activity is not due to the manipulation of periplasmic redox status that might have resulted from the overexpression of the respective proteins.
Complementation of ⌬dsbA by expression of SyndsbA alone suggests that SynDsbA can act independently of SynDsbB in the oxidation of cysteines in substrate proteins within E. coli by accepting oxidizing equivalents from DsbB. The ability of SynDsbA to accept oxidizing equivalents from DsbB is not surprising because DsbB is known to have broad substrate specificities and can oxidize substrates such as PDI, thioredoxin, and a mutant version of DsbC (31)(32)(33). In contrast, expression of SyndsbB alone in E. coli failed to complement ⌬dsbB, suggesting that E. coli DsbA is unable to accept oxidizing equivalents from SynDsbB. Structural analysis of the SynDsbB domain suggests a possible reason for its inability to functionally interact with DsbA. Compared with E. coli DsbB, both cysteine pairs in SynDsbB are positioned differently on the transmembrane helices ( Fig. 2) (23). The canonical CXXC in SynDsbB is present between the 3rd and 4th TMS compared with the 1st and 2nd TMS in DsbB. Similarly, the second pair of cysteines is present in large periplasmic loops connecting the 1st and 2nd TMS in SynDsbB compared with the 3rd and 4th TMS in DsbB. These structural differences may lead to inefficient interactions between SynDsbB and DsbA. It is known that Ero1 and Erv2, both DsbB-like proteins, exhibit structural similarity (23) and both generate disulfide bonds for PDI in the lumen of the ER (6, 7). However, it appears that only interaction of EroI with PDI is efficient as a null mutation in Ero1 is fatal (6), and only overexpression of Erv2 allow for the viability of Ero1 mutant cells (7).
A remarkable phenotypic parallel is observed between the properties of Synechocystis 6803 mutant strains of the SyndsbAB gene and null mutant strains of dsbA, dsbB, and ERO1 genes. Loss of SynDsbAB function render cells to be extremely sensitive to the presence of DTT and glucose, a phenotype also observed in dsbA, dsbB, and ERO1 null strains (5,6). Similarly, loss of SynDsbAB resulted in significant reduction of E. coli AP activity targeted to the periplasm of Synechocystis 6803, a finding similar to that found in ⌬dsbA and ⌬dsbB strains (3,5). Importantly, deletion of either SyndsbA or SyndsbB had similar effects on the activity of AP, suggesting that these two domains act in concert to form disulfide bonds in substrate proteins in Synechocystis 6803. An additional significant finding relates the impact of SynDsbAB to the growth of Synechocystis 6803. Growth of SyndsbAB null strains is significantly reduced under photoautotrophic conditions. Loss of either DsbA or DsbB had no physiologically visible impact on growth of E. coli under normal conditions whereas loss of Ero1 function renders yeast cells non-viable (3,5,6). Our efforts to restore the growth of mutant strains by altering culture conditions and/or nutrient concentrations have been unsuccessful, suggesting that loss of SynDsbAB function leads to pleiotropic effects. Thus it can be argued that the disulfide bond formation pathway in cyanobacteria is essential, but that spontaneous formation of disulfide bonds that occur in the presence of oxygen in absolutely required proteins allow for limited growth of mutant strains. This is noteworthy because cyanobacteria are oxygenic organisms and intracellular oxygen concentrations are much higher compared with non-photosynthetic organisms.
As mentioned previously, SynDsbB shows significant structural similarity to VKORC1 (26). VKORC1 is a subunit of the VKOR holoenzyme, whose function in mammalian systems is to recycle vitamin K epoxide to vitamin K and provide vitamin KH 2 to ␥-glutamyl carboxylase for the carboxylation of glutamate present in blood clotting proteins (27). Synechocystis 6803 does not appear to have ␥-glutamyl carboxylase and therefore a similar functional role of VKORC1 in cyanobacteria is unlikely. However, reduction of vitamin K to vitamin KH 2 by VKOR is kinetically similar to various oxidoreductases involved in generation of disulfide bonds. In fact, DsbB, an oxidoreductase that generates disulfide bonds in the periplasm of E. coli, has been shown to use vitamin K as an electron acceptor under anaerobic conditions in vivo (34). Regardless, whether SynDsbB is a DsbBlike or VKORC1-like protein, it is sufficient to provide SynDsbA with oxidizing equivalents required for disulfide bond formation in cysteine containing nascent proteins in vivo. Recent results by Wajih et al. (35) indicate that PDI provides electrons obtained from the oxidation of newly transported cysteine containing proteins in the ER to reduce the thioredoxin-like CXXC center in VKORC1. This further supports our results that SynDsbB acts like oxidoreductases involved in disulfide bond formation such as DsbB and Ero1. Finally, SynDsbAB orthologs have remained unaltered during billions of years of evolution from ancestral cyanobacteria to modern land plants. This suggests that the fusion of these two activities is advantageous to photosynthetic organisms where intracellular oxygen concentration is high compared with that in other organisms.