Photon Flux Density-dependent Gene Expression in Synechocystis sp. PCC 6803 Is Regulated by a Small, Redox-responsive, LuxR-type Regulator*

The expression of many cyanobacterial genes is regulated by the redox state of the photosynthetic electron transport chain. However, factors involved in this regulation have not been identified. In this study, we demonstrate that a small LuxR-type regulator in Synechocystis sp. PCC 6803, PedR (Ssl0564), senses the activity of photosynthetic electron transport to achieve the photon flux density-dependent transcriptional regulation. PedR is constitutively expressed in Synechocystis cells and exists as a dimer bridged by intermolecular disulfide bond(s). It activates the expression of chlL, chlN, chlB, and slr1957 and represses that of ndhD2, rpe, and the pedR (ssl0564)-sll0296 operon under conditions where the activity of photosynthetic electron transport is low. When the supply of reducing equivalents from photosynthetic electron transport chain increases upon the elevation of photon flux density, PedR is inactivated through its conformational change within 5 min. This mechanism enables transient induction or repression of the target genes in response to sudden changes in light environment. The fact that orthologs of PedR are conserved among all the cyanobacterial genomes sequenced so far indicates that this type of transcriptional regulation is essential for cyanobacteria to acclimate to changing environments.

The expression of many cyanobacterial genes is regulated by the redox state of the photosynthetic electron transport chain. However, factors involved in this regulation have not been identified. In this study, we demonstrate that a small LuxR-type regulator in Synechocystis sp. PCC 6803, PedR (Ssl0564), senses the activity of photosynthetic electron transport to achieve the photon flux density-dependent transcriptional regulation. PedR is constitutively expressed in Synechocystis cells and exists as a dimer bridged by intermolecular disulfide bond(s). It activates the expression of chlL, chlN, chlB, and slr1957 and represses that of ndhD2, rpe, and the pedR (ssl0564)-sll0296 operon under conditions where the activity of photosynthetic electron transport is low. When the supply of reducing equivalents from photosynthetic electron transport chain increases upon the elevation of photon flux density, PedR is inactivated through its conformational change within 5 min. This mechanism enables transient induction or repression of the target genes in response to sudden changes in light environment. The fact that orthologs of PedR are conserved among all the cyanobacterial genomes sequenced so far indicates that this type of transcriptional regulation is essential for cyanobacteria to acclimate to changing environments.
Following changes in environmental conditions such as photon flux density, temperature, or nutrient availability, photosynthetic organisms must balance energy supply through photosynthetic electron transport and its consumption by energy-demanding metabolic processes (1). For example, a shift from low to high light conditions triggers a decrease in the amount of the light-harvesting complexes and photosystem reaction centers to down-regulate the energy supply, whereas up-regulation of energy consumption is achieved by activation of metabolic pathways such as CO 2 fixation (2)(3)(4). This coordination is important not only to maximize the efficiency of photosynthesis but also to minimize the damage due to the over-reduction of the photosynthetic electron transport chain that may result in the formation of harmful reactive oxygen species (5).
Recent DNA microarray studies on high light acclimation of the cyanobacterium Synechocystis sp. PCC 6803 revealed the intimate relationship between these acclimation responses and transcriptional regulation (6 -8). Transcript levels of more than 100 open reading frames (ORFs), 3 including those involved in light absorption, photochemical reaction, protection from photoinhibition, and protein synthesis, were strongly affected by the changes in photon flux density. Interestingly, many of these ORFs were shown to be responsive to other environmental stresses such as low temperature (9), high salt (10), low CO 2 (11), hydrogen peroxide (12), and iron deficiency (13). These observations imply that transcription of high light-responsive genes is not regulated by sensing photon flux density per se but by monitoring the intracellular redox levels. It has been proposed that the redox state of the plastoquinone pool is important for transcriptional regulation of photosynthesis-related genes (14,15). However, DNA microarray analysis using inhibitors of the photosynthetic electron transport revealed that the redox state of components located downstream of the plastoquinone pool would be more critical for transcriptional regulation than that of the plastoquinone in Synechocystis sp. PCC 6803 (16). Similarly, it has recently been shown by DNA microarray technique that the redox state of the components on the acceptor side of photosystem I is important for light-dependent modulation of nuclear gene expression in higher plants (17). These results seem quite reasonable considering that the redox state of the downstream region of the photosynthetic electron transport chain can be directly affected by the energy balance of cells. However, actual mechanisms of redox sensing and subsequent processes of signal transduction are totally unknown.
In this study, we report that PedR (Ssl0564) from Synechocystis sp. PCC 6803, a small LuxR-type transcriptional regulator, works as a sensor for the availability of reducing equivalents supplied from photosynthetic electron transport chain. Thus, PedR establishes an important link between perception of environmental changes and transcriptional regulation to start acclimation processes.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-A glucose-tolerant wildtype strain of Synechocystis sp. PCC 6803 was grown at 32°C in BG-11 medium with 20 mM Hepes-NaOH, pH 7.0, under continuous illumination at 20 mol photons m Ϫ2 s Ϫ1 with bubbling of air. The ssl0564-disrupted mutant was grown under the same conditions, except that 20 g/ml kanamycin sulfate was added to the medium. Cell density was estimated as A 730 with a spectrophotometer (model UV-160A; Shimadzu). High light shift experiments were performed by transferring cells at the exponential growth phase (A 730 ϭ 0.1-0.2) from low light (20 mol photons m Ϫ2 s Ϫ1 ) to high light (300 mol photons m Ϫ2 s Ϫ1 ) conditions.
Construction of the ssl0564-disrupted Mutant-The geneflanking segment of ssl0564 (1000 bp) was amplified by PCR using primers 0564delF (5Ј-ACCGAGGATTGTGTTGG-3Ј) and 0564delR (5Ј-ATGATCACGGGGAATGCG-3Ј) and cloned into pT7Blue T-vector (Novagen). A kanamycin resistance cartridge, which had been excised from the plasmid pRL161 by digestion with HincII, was inserted into the coding region of ssl0564 at the ApaI site. Wild-type cells were transformed with the construct, and transformants were selected by 20 g/ml kanamycin sulfate. The complete segregation of the mutant genome was verified by PCR.
DNA Microarray Analysis-Total RNA used for DNA microarray analysis was isolated using the RNeasy Midi kit (Qiagen) as previously described (6). After the removal of trace amounts of contaminating genomic DNA by treatment with DNase I (Takara), RNA was labeled with Cy3-dUTP or Cy5-dUTP (Amersham Biosciences) using the RNA fluorescence labeling core kit Ver. 2.0 (M-MLV version; Takara). Hybridization of labeled cDNAs with CyanoCHIP (Ver. 1.6; Takara) was performed according to the manufacturer's instruction. Image acquisition with a ScanArray 4000 (GSI Lumonics) and data analysis by QuantArray Ver. 2.0 software (GSI Lumonics) were performed as previously described (16).
Overexpression and Purification of Ssl0564-The ssl0564 coding region was PCR amplified by using primers 0564F (5Ј-AACATATGCTGGCCGGGAACTAT-3Ј) and 0564R (5Ј-AAGGATCCTAAGACCCACTGCCTTG-3Ј) cloned into pT7Blue T-Vector (Novagen), digested with NdeI and BamHI (sites underlined), and subcloned into the same restriction sites in pET28a (Novagen) to create pET0564 for expression of a fusion protein with an N-terminal His tag. The nucleotide sequence was confirmed by DNA sequencing using the BigDye terminator method (Applied Biosystems). Escherichia coli BL21(DE3) harboring pET0564 was grown in 800 ml of culture containing 20 g/ml kanamycin sulfate for 9 h at 37°C without addition of isopropyl ␤-D-thiogalactoside. Cells were harvested by centrifugation, resuspended in 15 ml of 20 mM phosphate buffer, pH 7.4, containing 0.5 M NaCl and 60 mM imidazole, and disrupted by sonication for 30 s for 10 times at 4°C. After the removal of unbroken cells and insoluble materials by centrifugation, the soluble protein fraction was filtered through a 0.2-m filter (DISMIC-25cs; ADVANTEC). His-Ssl0564 was purified by nickel affinity column chromatography using a HiTrap chelating HP column (Amersham Biosciences). The soluble protein fraction was applied to the column equilibrated with 20 mM phosphate buffer, pH 7.4, containing 0.5 M NaCl and 60 mM imidazole, washed with the same buffer, and eluted with 20 mM phosphate buffer, pH 7.4, containing 0.5 M NaCl and 200 mM imidazole. Purified His-Ssl0564 was desalted by a HiTrap desalting column (Amersham Biosciences). Protein composition was examined by electrophoresis in a non-reducing 15% SDS-polyacrylamide gel followed by staining with Coomassie Brilliant Blue R-250.
Gel Mobility Shift Assay-For preparation of the probes and the specific competitor DNA fragments for gel mobility shift assays, the following DNA fragments corresponding to the promoter region of each gene were obtained by PCR amplification: ssl0564 (from nucleotides 2287333 to 2287068 according to the numbering in CyanoBase), ndhD2 (from nucleotides 285939 to 286163), rpe (from nucleotides 1713448 to 1713257), chlL (from nucleotides 3415526 to 3415988), slr1957 (from nucleotides 1755153 to 1755591), chlB (from nucleotides 2394872 to 2395102), and psaD (from nucleotides 126444 to 126638). The 3Ј-end of the DNA fragment for the probe was labeled with DIG-ddUTP by the terminal transferase according to the manufacturer's instructions (DIG gel shift kit; Roche Applied Science). Assays were performed using the DIG gel shift kit according to the manufacturer's instructions. Purified His-Ssl0564 protein was incubated with 30 fmol DIG-labeled DNA fragment in a 20-l reaction mixture containing 1 g of poly d[I⅐C], 0.1 g of poly-L-lysine, 20 mM Hepes-KOH, pH 7.6, 1 mM EDTA, 10 mM (NH 4 ) 2 SO 4 , 0.2% (w/v) Tween 20, and 30 mM KCl. After incubation overnight at room temperature, 5 l of gel loading buffer consisting of 60% (v/v) of 1ϫ Tris borate-EDTA and 40% (v/v) glycerol was added to the reaction mixture. Samples were then applied onto a 6% polyacrylamide gel and subjected to electrophoresis at 95 V for 2.5 h at 4°C. DNA and protein were transferred to a nylon membrane (Hybond Nϩ; Amersham Biosciences) by capillary transfer method and fixed by baking at 80°C for 2 h. Detection of DIG-labeled probe was performed according to the standard protocol for the DIG luminescent detection kit (Roche Applied Science).

Small LuxR-type Regulators Are Highly Conserved among
Cyanobacterial Species-In this study, we initiated the characterization of a small LuxR-type regulator (SLTR), Ssl0564, in a cyanobacterium Synechocystis sp. PCC 6803. The LuxR subfamily is one of the major groups of bacterial transcriptional regulatory proteins that bind DNA via a helix-turn-helix motif. It includes a large number of proteins consisting of two domains, a C-terminal DNA-binding domain with a helix-turnhelix motif and an N-terminal receiver domain that can be activated by either phosphorylation or binding of an effector molecule (19,20). In addition, there exists a small group of SLTRs consisting solely of a DNA-binding domain. SLTRs, ϳ100 amino acid residues in length, are found in a wide range of bacteria such as proteobacteria, firmicutes, and cyanobacteria (smart.embl-heidelberg.de/). They have not been characterized yet except for GerE, which functions in Bacillus subtilis at the late stages of sporulation (21)(22)(23). Fig. 1 shows the phylogenetic tree of SLTRs constructed with the neighbor-joining algorithm. It is noticeable that most of the cyanobacterial SLTRs form a discrete clade (shown in gray). All the cyanobacterial species for which the genomic sequencing has been completed possess one SLTR in this clade. The notable feature of SLTRs in the cyanobacterial clade is recognized in the C-terminal extension containing three conserved cysteine residues (Fig. 2). This region is only conserved in the cyanobacte-  rial subfamily in SLTRs, suggesting that it may be involved in regulatory mechanisms characteristic of photosynthetic organisms. Several cyanobacterial species such as Gloeobacter violaceus, Synechococcus sp. WH8102, and Prochlorococcus marinus MIT9313 have multiple SLTRs. In such cases, only one regulator is in the cyanobacterial clade, whereas the other SLTRs are rather divergent and do not have the three cysteine residues unique to SLTRs in the cyanobacterial clade.
Ssl0564 Acts as Both a Positive and a Negative Regulator-Reverse transcription PCR analysis suggested that ssl0564 is weakly expressed in cells of Synechocystis sp. PCC 6803 incubated under dark, low light (20 mol photons m Ϫ2 s Ϫ1 ), and high light (300 mol photons m Ϫ2 s Ϫ1 ) conditions (not shown). To investigate the physiological role of Ssl0564, first we made a gene-disrupted mutant of ssl0564 (the ⌬ssl0564 mutant) by the insertion of a kanamycin resistance cartridge. The growth characteristic of the ⌬ssl0564 mutant was similar to that of the wildtype strain under all growth conditions examined, including low light, high light, high temperature (42°C), and presence of ethanol (3%), glucose (5 mM), or NaCl (0.4 M) (not shown).
To identify target genes of Ssl0564, we compared the gene expression profiles in the wild-type and the ⌬ssl0564 mutant cells under low light conditions by DNA microarray analysis (Fig. 3). Most genes were similarly expressed in both strains, but the transcript levels of several genes were reproducibly affected by the disruption of ssl0564 (Table 1). It was suggested that Ssl0564 represses the transcription of ndhD2, rpe, and the ssl0564-sll0296 operon and activates that of chlL, chlN, chlB, and slr1957 under low light conditions. Ssl0564 Becomes a Dimeric Form through the Formation of Intermolecular Disulfide Bond(s)-To characterize the gene product of ssl0564, His-tagged Ssl0564 expressed in E. coli was purified by nickel affinity column chromatography and used for subsequent analyses. When the purified protein was subjected to non-reducing SDS-PAGE, we observed two bands with molecular mass of 12 and 24 kDa (Fig. 4A, lanes 2 and 9). To examine the possibility that dimerization occurs via the formation of intermolecular disulfide bond(s), we treated Ssl0564 with a thiolalkylating reagent, NEM, before incubation with hydrogen peroxide (Fig. 4B). The alkylated Ssl0564 could no longer form dimer (lane 4), suggesting that the formation of intermolecular disulfide bond(s) is essential for dimerization.
The Dimeric Form of Ssl0564 Can Bind to the Promoter Region of Its Target Genes-To test whether genes listed in Table 1 are indeed targets of Ssl0564, we performed a DNA gel mobility shift assay of the promoter segments of these genes with purified His-Ssl0564 protein in the presence of potassium ferricyanide (Fig. 5A). Although clear shifted bands were not observed, the intensity of the band corresponding to the free probe (lane 1) decreased with the increase of Ssl0564 protein in the reaction mixture (lanes 2-4). It is notable that the addition of non-labeled competitor DNA together with Ssl0564 protein could suppress the decrease of the free probe band (lane 5). When the promoter segment of psaD, which does not function as a target gene, was used as a probe, the band pattern was not affected by the addition of Ssl0564 protein. These data supported the notion that specific complexes were formed between Ssl0564 and promoter segments of putative target genes.
Next, we tested the binding ability of Ssl0564 protein in various redox environments. In Fig. 5B, the binding activity of  . Although Ssl0564 could bind to ndhD2 probe in the absence of redox agents, the oxidizing environment promoted the formation of the complex of Ssl0564 with ndhD2 probe. On the other hand, addition of DTT prevented the formation of such a complex. When ndhD2 probe was incubated with ferricyanide or DTT in the absence of Ssl0564 protein, no changes in band pattern was observed (lanes 5-7). These results indicate that dimerization is a prerequisite for stable binding of Ssl0564 to its target sequences. Transcriptional Regulation by Ssl0564 Is Achieved depending on the Activity of Photosynthetic Electron Transport-We then followed the transcript levels of rpe, ndhD2, and slr1957 in the wild-type and the ⌬ssl0564 mutant under different light conditions by RNA gel blot analysis (Fig. 6). Under low light conditions, the level of the transcripts originating from rpe and ndhD2 was low in the wild-type cells (lane 1), whereas such a down-regulation was fully (rpe) or partially (ndhD2) abolished in the mutant (lane 2). In the case of slr1957, its transcript was highly accumulated in the wild-type (lane 1) and was scarcely observed in the mutant (lane 2). These observations agree with the results of DNA microarray analysis showing that Ssl0564 acts as a negative regulator for rpe and ndhD2 and as a positive regulator for slr1957 under low light conditions. Within 15 min after the shift to high light conditions, increase of rpe and ndhD2 transcripts and decrease of slr1957 transcript were observed in the wild-type cells (lane 3), showing that light response of these genes is attained by the inactivation of Ssl0564 under high light conditions. As expected, in the ⌬ssl0564 mutant, high light response of these genes was much less prominent than that in the wildtype cells (lane 4).
Then, which is the critical factor for Ssl0564-dependent transcriptional regulation, changes in photon flux density per se or those in the redox state of the photosynthetic electron transport chain? To answer this question, 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) or methyl viologen was added to cultures upon the shift to high light conditions. By the addition of DCMU or methyl viologen, the supply of reducing equivalents from the photosynthetic electron transport chain to stromal components is perturbed even under high light conditions because DCMU and methyl viologen inhibit electron transfer  For lane 5, 300-fold excess amounts of the non-labeled promoter segment were added as a competitor. B, the DIG-labeled promoter segment of ndhD2 was incubated overnight with purified His-Ssl0564 under various redox conditions. Lane 1, ndhD2 probe without additives. Lanes 2-4, ndhD2 probe and 1 g of His-Ssl0564 incubated without redox agents, with 100 M ferricyanide, or with 1 mM DTT, respectively. Lanes 5-7, ndhD2 probe incubated without redox agents, with 100 M ferricyanide, or with 1 mM DTT, respectively. from photosystem II to the plastoquinone pool and that from photosystem I to ferredoxin, respectively (24,25). In the presence of these reagents, induction of rpe and ndhD2 and repression of slr1957 were hardly observed in the wild-type cells upon the shift to high light conditions (Fig. 6, lanes 5 and 7). The only exception was marked down-regulation of slr1957 treated with methyl viologen, which may be attributed to unidentified regulatory factors other than Ssl0564. In the case of the ⌬ssl0564 mutant, the transcript levels of the target genes were not affected by the addition of DCMU or methyl viologen (lanes 6 and 8). The above observations clearly indicate that Ssl0564 works as a transcriptional regulator under conditions where the activity of the photosynthetic electron transport is low and supply of reducing equivalents is limited, e.g. low light conditions. Ssl0564 Protein Undergoes Conformational Change depending on the Activity of Photosynthetic Electron Transport-To elucidate the mechanism by which the activity of Ssl0564 is modulated, we examined the physiological state of Ssl0564 in vivo by immunoblot analysis. The wild-type cultures were incubated under the same conditions used for RNA gel blot analysis shown in Fig. 6. After cultures were treated with 10% trichloroacetic acid to fix cellular proteins, thiols were alkylated by NEM or AMS to inhibit their nonspecific interaction, which is an essential step for successful detection of Ssl0564. The proteins were separated by non-reducing SDS-PAGE, and Ssl0564 was detected using its specific antibody. As shown in Fig. 7, we observed that the constant amount of Ssl0564 exists in Synechocystis cells exclusively as a dimeric form both under low and high light conditions. It thus appears that neither oligomeric state nor amount of Ssl0564 is critical for the light-dependent regulation of its activity in vivo.
Unexpectedly, we observed that the dimeric form of Ssl0564 was separated into two bands (Fig. 7). In low light-grown cell extracts, the fast migrating band was mainly detected (lane 2). Exposure of cells to high light for 15 min resulted in significant increase of the slowly migrating band (lane 3). On the other hand, in extracts of cells incubated with DCMU or methyl viologen during the high light treatment, only the fast migrating band was detected (lanes 4 and 5), indicating that elevation of photosynthetic electron transport activity upon the shift to high light caused the increase in the relative intensity of the slowly migrating band. Furthermore, these data, together with the results of RNA gel blot analysis (Fig. 6), suggest that the slow or fast migrating band corresponds to the inactive or active form of Ssl0564, respectively. It must be noted that alkylation of samples with either NEM or AMS gave the same pattern of band shift (not shown). If Ssl0564 has any free thiol groups, the addition of AMS moiety (500 Da)/thiol would cause a larger band shift than the addition of NEM moiety (125 Da). The band shift of Ssl0564 independent of alkylating agents indicates that three cysteine residues in Ssl0564 are always in oxidized form and the conformational change of Ssl0564 is not due to redox active cysteines.
The Upshift of Photon Flux Density Causes Transient Inactivation of Ssl0564-We examined the time course change of the conformation of Ssl0564 and expression level of its target genes, rpe, ndhD2, and slr1957, upon the shift to high light conditions (Fig. 8). Within 5 min after the shift to high light, most of the Ssl0564 protein turned into the inactive form with slower electrophoretic mobility, which was followed by the accumulation  of ndhD2 and rpe transcripts and the disappearance of slr1957 transcript. After the temporal increase of the inactive form, the active form of Ssl0564 with faster electrophoretic mobility increased again.
Next, cell cultures grown at photon flux density of 20 mol photons m Ϫ2 s Ϫ1 were exposed to different levels of illumination for 5 min. As shown in Fig. 9A, upshift of photon flux density over 100 mol photons m Ϫ2 s Ϫ1 resulted in the significant increase of the slowly migrating form, i.e. the inactive form of Ssl0564. As expected, this change in conformation of Ssl0564 was accompanied by the induction of ndhD2 and rpe and the repression of slr1957 (Fig. 9B). Several repeats of similar experiments indicated that the threshold for the inactivation of Ssl0564 exists around 60 -80 mol photons m Ϫ2 s Ϫ1 .
As mentioned in the Introduction, the redox state of the plastoquinone pool is recognized as a critical factor for transcriptional regulation of photosynthesis-related genes. To check the possibility that the activity of Ssl0564 is modulated by the redox state of the plastoquinone pool, we compared the effect of addition of two photosynthetic inhibitors, DCMU and 2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone (DBMIB), which inhibits electron transport from the plastoquinone pool to cytochrome b 6 /f complex. These two inhibitors have opposite effects on the net redox state of the plastoquinone pool, more oxidized in the presence of DCMU and more reduced in the presence of DBMIB (24). Upon the shift to high light conditions, both inhibitors completely suppressed the appearance of the inactive form of Ssl0564 together with the light response of its target genes (Fig. 10), indicating that activity of Ssl0564 is regulated irrespective of the redox state of the plastoquinone pool.

DISCUSSION
Transcriptional Regulation by PedR Coupled with the Activity of Photosynthetic Electron Transport-In this study, we have shown that one of the cyanobacterial orthologs of SLTR, Ssl0564, works as a sensor of photosynthetic activity and achieves the photon flux density-dependent transcriptional regulation of a set of genes in Synechocystis sp. PCC 6803. Thus, we propose that Ssl0564 is designated PedR (Photosynthetic Electron transport-Dependent Regulator).
Upon the transfer of the wild-type cultures from low to high light conditions, we observed the inactivation (Fig. 6) and the conformational change (Fig. 7) of PedR. Addition of DCMU, DBMIB, or methyl viologen extinguished the effect of the upshift of photon flux density (Figs. 6, 7, and 10), indicating that the activity of the photosynthetic electron transport rather than photon flux density per se is critical for the regulation of PedR activity. Considering that similar results were obtained by the addition of these reagents, which perturb different points of the photosynthetic electron transport chain, the activity of PedR is likely to be modulated by the availability of reducing equivalents provided at the acceptor side of photosystem I (Fig. 11). Under low light conditions where activity of the photosynthetic electron transport is low, a large fraction of PedR is in the active compact form and regulates expression of its target genes. Upon the sudden elevation of photon flux density, supply of reducing equivalents from the photosynthetic electron transport chain abruptly increases, leading to the conformational change and inactivation of PedR. As the acclimation responses take place to mitigate the over-reduction of the photosynthetic FIGURE 8. Time course change of the conformation of Ssl0564 and the expression level of its target genes rpe, ndhD2, and slr1957 upon the shift to high light conditions. A, trichloroacetic acid-fixed proteins from cells incubated under high light were alkylated, and 7.5 g of total protein/lane was separated by non-reducing 15% SDS-polyacrylamide gel electrophoresis with 8 M urea. Ssl0564 was detected by its specific antibody. B, total RNA was isolated from cells incubated under high light conditions. 2 g of total RNA/ lane was loaded for the detection of rpe and ndhD2 transcripts, and 6 g was loaded for slr1957 transcript. . Photon flux density dependence of the conformational change of Ssl0564 and the expression level of its target genes rpe, ndhD2, and slr1957. A, trichloroacetic acid-fixed proteins from cells exposed to various levels of illumination for 5 min were alkylated, and 4 g of total protein/lane was separated by non-reducing 15% SDS-polyacrylamide gel electrophoresis with 8 M urea. Ssl0564 was detected by its specific antibody. B, total RNA was isolated from cells exposed to various levels of illumination for 15 min. 2 g of total RNA/lane was loaded for the detection of rpe and ndhD2 transcripts, and 6 g was loaded for slr1957 transcript. electron transport chain, PedR returns to the active form again. This mechanism seems to enable transient induction or repression of the target genes in response to sudden changes in photon flux density.
The Physiological Significance of Transcriptional Regulation by PedR-It is notable that genes involved in various cellular functions are coordinately regulated by PedR (Table 1). When the activity of photosynthetic electron transport is low, PedR positively regulates the expression of chlL, chlN, and chlB encoding subunits of light-independent protochlorophyllide reductase and an uncharacterized ORF, slr1957. Upon the upshift of photon flux density, transcription of these genes is down-regulated due to the inactivation of PedR. DNA microarray analysis showed that genes involved in biosynthesis of photosynthetic pigments including chlL, chlN, and chlB were remarkably down-regulated under high light conditions (6). This is quite reasonable considering that absorption of excess light energy should be avoided under high light conditions. PedR seems to take a part in this acclimation response. As for slr1957, its physiological role is totally unknown. Judging from its expression pattern, it may have some roles under dark or low light conditions. We observed that expression of slr1957 is sometimes not fully activated under the conditions where PedR is in the active state (Fig. 6, HLϩMV; Fig. 9, 20 mol photons m Ϫ2 s Ϫ1 ). There must be a negative regulation of slr1957 by a certain factor in addition to the positive regulation by PedR.
Transcription of ndhD2, rpe, and pedR-sll0296 operon is negatively regulated by PedR under low light conditions and derepressed upon the sudden increase of photosynthetic electron transport activity. ndhD2 is one of six ndhD genes in Synechocystis sp. PCC 6803 and encodes NdhD2 subunit involved in electron transport from NADPH to the plastoquinone pool (26). This gene was remarkably up-regulated under high light conditions (6). Although the precise role of NdhD2 subunit has not been identified, it may contribute to cycling of excess reducing equivalents under high light conditions. Tu et al. (8) reported that a sensor histidine kinase, Hik33 (DspA) is involved in the repression of ndhD2 under low light conditions. This agrees well with our observation that the expression of ndhD2 was not fully derepressed in the ⌬ssl0564 mutant exposed to low photon flux density (Fig. 6, lane 2). Hik33 and PedR may belong to independent pathways of signal transduction for light response of the ndhD2 gene. Another gene negatively regulated by PedR, rpe, encodes pentose-5-phosphate-3-epimerase, which catalyzes the interconversion of ribulose-5-phosphate and xylulose-5-phosphate in the Calvin cycle and in the oxidative pentose phosphate pathway. Although this enzyme in cyanobacteria has not been characterized, it is plausible that expression level of rpe has some regulatory role on cellular metabolism. It was shown that PedR binds to its own promoter to repress the transcription of pedR itself and sll0296 locating just downstream of pedR. The transcript level of pedR is maintained at low level irrespective of the growth condition, which seems to be attained by negative autoregulation.
The Structural Characteristics of PedR-Among SLTRs conserved in many bacterial species, only GerE in B. subtilis has been extensively characterized and its crystal structure was solved at 2.05 Å resolution (23). Comparison of the structure of cyanobacterial SLTRs with that of GerE may provide additional insights into the function of cyanobacterial SLTRs. GerE con-  (HLϩDBMIB) were alkylated, and 7 g of total protein/ lane was separated by non-reducing 15% SDS-polyacrylamide gel electrophoresis with 8 M urea. Ssl0564 was detected by its specific antibody. B, total RNA was isolated from cells incubated for 15 min under the same conditions used for panel A. 2 g of total RNA/lane was loaded for the detection of rpe and ndhD2 transcripts, and 6 g was loaded for slr1957 transcript.
sists of four ␣-helices: the central pair, ␣-2 and ␣-3, forms a helix-turn-helix motif whereas ␣-4 in the C-terminal region is involved in dimerization. These ␣-helices, especially the central pair involved in DNA binding, are well conserved in cyanobacterial SLTRs (Fig. 2), suggesting that the DNA binding mechanism of PedR is similar to that of GerE. In the promoter regions of target genes of PedR, there exist multiple inverted repeats consisting of the 12-mer sequence 5Ј-RWWTRGGYNNYY-3Ј typical of GerE binding motifs. It has been shown that the positioning and multiplicity of GerE binding sites are variable (23) and GerE can act both as a positive and as a negative regulator depending on its binding position (27). By a similar mechanism, PedR could work for positive and negative regulation. When we examined the binding ability of His-PedR to the promoter regions of its putative target genes, a distinct shifted band was not observed in most cases (Fig. 5). This is probably due to the heterogeneous population of the His-PedR⅐DNA complex. Variable numbers of His-PedR bound to multiple binding sites located on a promoter fragment, which may result in a smearing of the shifted band.
SLTRs in the cyanobacterial clade ( Fig. 1) have a unique C-terminal region not conserved in any other bacterial SLTRs (Fig. 2). The region extends from ␣-4 helix and contains three conserved cysteine residues. Characterization of His-PedR protein suggested that some of these cysteines are involved in dimerization through the formation of intermolecular disulfide bond(s) (Fig. 4). By the addition of reducing agents, PedR becomes a monomeric form in vitro (Fig. 4) and also in vivo (not shown). Under physiological conditions, however, PedR constitutively exists as a dimer and the monomeric form is scarcely observed (Fig. 7). It seems that the supply of reducing equivalents from photosynthetic electron transport chain is not strong enough to reduce the intermolecular disulfide bond(s) but is sufficient for the observed conformational change, which is presumably critical for the regulation of the PedR activity. The nature of the conformational change has not been identified. Three cysteine residues were insensitive to AMS modification, indicating that they are constitutively in the oxidized form and not responsible for the redox-dependent conformational change of PedR. Temporal modification of a certain amino acid residue could occur upon the supply of reducing equivalents, leading to the conformational change and inactivation of PedR. To identify the amino acid residues critical for the regulation of PedR activity, we are now trying to introduce random amino acid substitution into PedR protein and examine the consequence of the mutagenesis.