Cyclic Nucleotides, the Photosynthetic Apparatus and Response to a UV-B Stress in the Cyanobacterium Synechocystis sp. PCC 6803*

Cyclic nucleotides cAMP and cGMP are ubiquitous signaling molecules that mediate many adaptative responses in eukaryotic cells. Cyanobacteria present the peculiarity among the prokaryotes of having the two types of cyclic nucleotide. Cellular homeostasis requires both cyclases (adenylyl/guanylyl, for their synthesis) and phosphodiesterases (for their degradation). Fully segregated null mutants have been obtained for the two genes, sll1624 and slr2100, which encode putative cNMP phosphodiesterases. We present physiological evidence that the Synechocystis PCC 6803 open reading frame slr2100 could be a cGMP phosphodiesterase. In addition, we show that Slr2100, but not Sll1624, is required for the adaptation of the cells to a UV-B stress. UV-B radiation has deleterious effects for photosynthetic organisms, in particular on the photosystem II, through damaging the protein structure of the reaction center. Using biophysical and biochemical approaches, it was found that Slr2100 is involved in the signal transduction events which permit the repair of the UV-B-damaged photosystem II. This was confirmed by quantitative reverse transcriptase-PCR analyses. Altogether, the data point to an important role for cGMP in signal transduction and photoacclimation processes during a UV-B stress.

decreases in biomass production, secondary sugars, and chlorophyll content as well as to inactivation of ATPase (9).
In cyanobacteria more than 99% of the UV-B is absorbed by chlorophyll-binding proteins and phycobilisomes, the light-harvesting complexes (10). One of the main targets of the UV-B radiation is Photosystem II (PSII), 3 whose electron transport is inhibited and D1 and D2 reaction center proteins are degraded (11)(12)(13)(14)(15)(16)(17)(18). The ensuing degradation of the D1 and D2 proteins, which compose the heart of the PSII reaction center, leads to an inhibition of the PSII activity (19). Among the strategies developed to cope with UV-B, cyanobacteria increase their de novo synthesis of the D1 and D2 subunits to repair PSII (20,21).
To adapt to new conditions, cells must first perceive the environmental signals and then transduce them to the response apparatus so as to modify their metabolism accordingly. Second messengers such as cyclic nucleotides (cNMPs) play key roles in the transduction steps (22). Cyanobacteria are the only prokaryotes that, like eukaryotes, possess both cAMP and cGMP, their precise role being, however, poorly documented at present (23,24). Information has been obtained from different strains, and currently only an incomplete picture of the role of cNMPs is available. For Spirulina cells, which excrete cAMP under standard growth conditions, cAMP stimulates respiration and gliding motility and triggers mat formation (24). In Anabaena flos-aquae, the extracellular cAMP concentration was shown to be 10-fold higher in stationary than in exponential phase (25). In Anabaena cylindrica, lightoff and light-on signals modulate the intracellular cAMP concentration (26). cAMP-mediated photosignaling through phytochrome-like proteins has now been demonstrated in Anabaena PCC 7120 (27). In contrast to what happens in A. cylindrica, the cellular cAMP concentration increases upon a shift from dark to light, especially blue light, in Synechocystis PCC 6803 (24,28). We also know that in that strain, both cAMP and its receptor protein (Sycrp1) are required for motility. Mutants in either the adenylyl cyclase or Sycrp1 are indeed sessile, motility for cya1 mutants being recovered by the addition of exogenous cAMP (29,30). Finally, in that same strain the cGMP concentration increases when cells are grown photoheterotrophically and starved for nitrogen (31).
Cyclic nucleotide homeostasis requires both cyclases to achieve their synthesis from ATP or GTP and phosphodiesterases for their degradation to AMP or GMP. By complementation of Escherichia coli cya mutants, a few cyanobacterial genes coding for adenylyl cyclases have been cloned (32). The recent availability of complete genome sequences for 13 cyanobacterial strains has revealed that many putative adenylyl cyclases may exist within a single species, with at least 6 present in Anabaena/Nostoc PCC 7120 (33) and up to 13 in Trichodesmium erythraeum. 4 Both an adenylyl and a putative guanylyl cyclase have been described for Synechocystis PCC 6803 (29,34). Although phosphodiesterase enzymatic activities have been measured for cAMP degradation in a few species (35), no orthologue of bacterial-type phosphodiesterases could, however, be recognized by in silico approaches in Synechocystis PCC 6803 (36).
Two Synechocystis PCC 6803 open reading frames (slr2100 and sll1624) carry a so-called HD (or phosphohydrolase) domain. Belonging to the HD protein family, they were proposed to be putative cNMP phosphodiesterases (36). The questions we addressed were: do these molecules regulate the intracellular level of cAMP and/or cGMP, and what is their in vivo function? To get an answer, the two genes were inactivated, and the phenotype of the corresponding mutants was studied. Under steady-state standard conditions, both grow at a rate similar to that of the wild type strain. However, the slr2100 mutant contains slightly lesser amounts of the photosynthetic apparatus components and higher amounts of cGMP and is more sensitive to a UV-B stress than the wild type. We found that it is impaired in its ability to repair the damaged PSIIs under UV-B radiation. Altogether, the data show that cGMP could play an important role in the adaptation, regulation, and functioning of the photosynthetic apparatus in Synechocystis PCC 6803.

MATERIALS AND METHODS
Strains and Growth Conditions-Synechocystis sp. PCC 6803 was obtained from the Pasteur Culture Collection. Liquid cultures were routinely grown photoautotrophically at 30°C under white light (30 mol of photon⅐m Ϫ2 ⅐s Ϫ1 , cool-white Philips fluorescent tubes) in BG11 medium (37) supplemented with 10 mM NaHCO 3 and buffered with 10 mM Hepes-NaOH, pH 8.0, under a 1% (v/v) CO 2 -enriched atmosphere in an illuminated rotatory shaker (120 rpm). For growth on plates, 1% of separately autoclaved agar was added. Whenever appropriate, antibiotics (spectinomycin (30 g⅐ml Ϫ1 ), kanamycin (75 g⅐ml Ϫ1 ), or chloramphenicol (7 g⅐ml Ϫ1 )) were added to the plates.
Cyanobacterial growth was followed by recording optical densities at 750 nm. Cyanobacterial cell mass was estimated by measuring the Chla concentration of the cultures. Chla was determined in 90% acetone extracts (38) or using the Chlorophyll 1.03 program (bilbo.bio.purdue.edu/www-cyanosite/protocols/chl103.html). Whole cell spectra were recorded from 400 to 750 nm with an Aminco DW-2 spectrophotometer. The E. coli strain routinely used as host, DH5␣, was grown at 37°C in Luria broth medium with appropriate antibiotics whenever necessary (ampicillin and/or chloramphenicol at 50 or 30 g⅐ml Ϫ1 , respectively).
UV-B Treatment-UV-B irradiation was performed at 30°C in open, rectangular glass containers in which 11-mm-height layers of cells were maintained in suspension by magnetic agitation. UV-B light was provided by a Vilbert-Lourmat VL-215M lamp in combination with a 0.1-mm cellulose acetate filter (Clarfoil, Courtaluds Chemicals, UK), yielding an intensity of 6 mol of photons⅐m Ϫ2 ⅐s Ϫ1 at the surface of the samples. Incident wavelengths range from 290 to 340 nm, with a maximum at 312 nm. Cells were used at a concentration of 6.5 g Chla⅐ml Ϫ1 .
Oxygen Evolution Measurements-Steady-state rates of oxygen evolution were measured using a Hansatech DW2 O 2 electrode at a light intensity of 1000 mol of photons⅐m Ϫ2 ⅐s Ϫ1 of photosynthetically active radiation (400 -700 nm) in the presence of 0.5 mM 2,5-dimethyl-p-benzoquinone as electron acceptor. Light was provided by a Halogen-Bel-laphot (12 V, 100 watts, Osram, Germany) lamp. Typically, 2 ml of cells at 6.5 g of Chla⅐ml Ϫ1 were used in each measurement. Results are expressed in relative value, 100% as the steady-state oxygen evolution at time 0.
Fluorescence Relaxation Decay-Flash-induced increase and subsequent decay of chlorophyll fluorescence yield was measured with the P.S.I. double-modulated fluorometer FL-100 (P.S.I., Brno, Czech Republic) and was performed in the 150 s to 100 s range as described (19).
Plasmid Constructions-Standard cloning procedures were used according to Sambrook et al. (39). A 2680-bp DNA fragment containing the entire slr2100 (rre20) open reading frame and two flanking regions (704 bp upstream and 870 bp downstream, i.e. from nucleotides 1567613 to 1570293 in Cyanobase; www.kazusa.or.jp/cyano) was amplified by PCR from genomic DNA using two specific primers that were designed from the sequences available in Cyanobase (TABLE ONE). For oriented cloning into pBluescriptSK ϩ , BamHI and PstI restriction sites were added at the 5Ј-and 3Ј-ends of the sequence, respectively. The PCR product was obtained using 0.1 g of purified Synechocystis genomic DNA after 35 cycles of amplification with the Expand High Fidelity system (Roche Applied Science), with cycles set up as 94°C (30 s), 63°C (30 s), and 68°C (3 min). The PCR fragment was digested by BamHI and PstI and cloned into a pBluescript SK ϩ vector (Stratagene) digested by the same enzymes, producing pSlr2100. A similar protocol was used to construct pSll1624, the BamHI-PstI insert (2427 bp) containing the entire sll1624 (rre18) open reading frame and two flanking regions (782 bp upstream and 640 bp downstream, i.e. from nucleotides 1320195 to 1322621 in Cyanobase).
The p⌬Slr2100 derivative was constructed by substituting the entire slr2100-coding sequence with the ⍀ cassette (aadA gene), which confers resistance to both spectinomycin and streptomycin. The slr2100coding region was removed by performing a reverse PCR on pSlr2100 using primers specific for the slr2100-flanking regions Del2100-Fw and Del2100-Rev and the above-described protocol. Both primers were designed to generate a XmaI site at the extremity of the PCR product site used to insert the ⍀ cassette (2-kilobase-long XmaI fragment from pHP45) after dephosphorylation with shrimp alkaline phosphatase.
The p⌬Sll1624 derivative was constructed by substituting the entire sll1624-coding sequence with the cassette from pUC4K, which confers resistance to kanamycin. The two flanking sequences were prepared by PCR and ligated after restriction by SmaI to pBluescript SK ϩ . Then the kanamycin cassette was inserted into the SmaI site.
Construction of the Synechocystis Mutants-Wild type Synechocystis PCC 6803 cells were transformed separately with plasmids p⌬Slr2100 and p⌬Sll1624, which do not replicate in the cyanobacterium, according to Golden et al. (40). Cells were incubated for 48 h on nitrocellulose filters (Nuclepore REC-85) without any selection, and the transformants were selected after transfer to antibiotic-containing BG11 plates.

cGMP and Adaptation to a UV-B Stress
Cells were repeatedly subcultured until full segregation of the mutation was obtained. Total segregation was ascertained by PCRs performed using the flanking region oligonucleotides. cNMP Determination-Cells (100 ml) between 6 and 6.5 g of Chla⅐ml Ϫ1 were harvested by filtration under vacuum on glass fiber prefilters (Millipore: APF B04700). Filters were immediately transferred to a tube containing 10 ml of 1 mM NaH 2 PO 4 buffer at pH 6.5 preheated at 100°C. After 5 min at 100°C, a known volume of the filtrate was filtered through a 0.45-m polyvinylidene difluoride membrane (Millipore) before transfer to a Vivaspin M r 50,000 column (Vivascience) and centrifugation for 30 min at 5000 ϫ g. An aliquot of the eluate was freeze-dried, and the lyophilisate was resuspended in 600 l of 30 mM, pH 6.5, of NaH 2 PO 4 buffer. After centrifugation for 5 min at 12,000 ϫ g, 100 l were injected onto a high performance liquid chromatography XTerra rp18 (3.5-m 4.6 ϫ 150 column, Waters) equipped with a 20-mm-long precolumn. The column was developed at a flow rate of 0.8 ml⅐min Ϫ1 with solvent A (30 nM sodium phosphate buffer at pH 6.5) and solvent B (acetonitrile/H 2 O (99:1, v/v). Quantification was achieved by calibrating the system with precisely known quantities of the different cyclic nucleotides.
Thylakoid Preparation and Protein Analyses-Thylakoid membranes were prepared by breakage of the cells with glass beads (150 -200 m in diameter) at 4°C followed by differential centrifugations according to Komenda et al. (41). Protein composition was assessed by electrophoresis in a denaturing 12-20% linear gradient polyacrylamide gel containing 6 M urea. After solubilization (41), thylakoid extracts adjusted at 0.7 g of Chla per lane were loaded, and the gel was run overnight at 18°C. Proteins were transferred onto nitrocellulose membranes (0.45 m, Schleicher and Schuell) by wet blotting. The membrane was incubated with an antibody raised against the C terminus of pea D1 protein (a kind gift of P. Nixon) and then with secondary antibody-alkaline phosphatase conjugate. The antigen-antibody complexes were visualized by colorimetric reaction using a BCPIP-NBT system. Membranes were scanned, and the bands were quantified using the ImageJ program (a public domain image processing and analysis program provided by the National Institutes of Health).
Macroarray and Quantitative PCR-Total RNAs were isolated by the hot-phenol method adapted from Mohamed and Jansson (42) and treated with DNase I (Invitrogen, 1 unit/g of RNA) according to the manufacturer's instructions before use. The absence of DNA products from these RNA preparations in standard PCR assays was checked to ascertain the removal of any DNA contaminants.
Probes for the macroarray analyses were prepared as follows. RNA was reverse-transcribed with a set of hexanucleotides (CGATCG, GGCGAT, CAAAAT, CAATGG, GGCAAT, AAATCC, CTTTTT, ACCAAT, GGCCAC, AAAACC, CCAGCA) kindly designed by Anne Morgat (INRIA Rhône-Alpes, France) so as to prime all of the 400 genes loaded on the nylon membrane while minimizing transcription from the rRNAs. The labeling reaction was performed in a total volume of 20 l containing 1 g of RNA, 5 pmol of each primer, 0.5 mM dCTP, dGTP, and dTTP, 5 M dATP, 40 Ci of ␣-[ 32 P]dATP, 200 units of SuperScript II reverse transcriptase, and 20 units of RNasOUT in 1ϫ manufacturer's reaction buffer (Invitrogen). RNA and primers in RNase-free H 2 O were first incubated at 55°C for 15 min and cooled down. The reaction mix containing [␣-32 P]dATP was then added, and incubation was done for 1 h at 37°C before the addition of 1 mM dATP (final concentration) followed by a further 15 min of incubation at 37°C. RNase (2 g) was added, and after 30 min at 37°C probes were purified using the QIAquick nucleotide removal kit (Qiagen). Hybridizations were performed overnight at 65°C. Membranes were then washed as follows: 2ϫ SSC (1ϫ SSC ϭ 0.15 M NaCl and 0.015 M sodium citrate) for 5 min at room temperature, 1% SDS in 2ϫ SSC for 30 min at 60°C, and 1% SDS in 0.1ϫ SSC for 30 min at room temperature. Quantifications of the RNA levels were performed with a Fuji Fla3000 reader.
For quantitative PCRs, primers were designed using the LightCycler software (Roche Applied Science) so as to get products with size about 150-bp long with a T m of 60°C. The cDNA synthesis was achieved in a final volume of 20 l containing 1 g of RNA, 2 pmol of the primers for each of the genes to be further amplified, 0.5 mM concentrations of each dNTP, 200 units of SuperScript II reverse transcriptase, and 20 units of RNasOUT in 1ϫ manufacturer's reaction buffer (Invitrogen). Real-time PCRs were performed according to the manufacturer's instructions using the QuantiTect SYBR Green kit (Qiagen) and a Roche LightCycler system on 0.4 l of the cDNA solution prepared above. Amplifications were done by incubating the reaction mixtures at 95°C for 30 s before 45 cycles of 30 s at 95°C (melting), 30 s at 60°C (annealing), and 30 s at 72°C (extension). At the end of the runs, a melting curve was generated and analyzed to ascertain that the recorded data correspond to only one PCR product of expected T m . The relative abundance of each transcript was determined by comparison of the threshold cycle values (C T ) recorded by the apparatus for the different genes to that for the rnpB product. The ratio of expression (R) of the gene of interest (goi) to rnpB was calculated using the simple equation: R goi/rnpB ϭ 2 (CT rnpB Ϫ CT goi ) .

Isolation and Growth Properties of the Mutants
Targeted substitution of the sll1624 and slr2100 genes by the antibiotic resistance cassettes was achieved by transforming wild type Synechocystis PCC 6803 cells with the non-replicative plasmid p⌬Sll1624 and p⌬Slr2100, respectively. Full segregation was easily obtained for the two mutations and ascertained by Southern blot experiments using slr2100, sll1624, and antibiotic resistance genes as hybridizing probes (data not shown). Under standard conditions, no significant differences in terms of growth rate could be observed between the null mutants (gene deletions) and wild type strain, with doubling times ϳ14 h at 30°C and 30 mol of photon⅐m Ϫ2 ⅐s Ϫ1 .

Characterization of the Mutants
Pigment Composition-Because of an apparent color difference between cultures, whole cell absorption spectra were recorded (Fig. 1A). The phycocyanin over chlorophyll ratio (PC/Chl) was consistently 10% lower for the slr2100 mutant (1.01 Ϯ 0.004) compared with the sll1624 and wild type strains (1.12 Ϯ 0.013 and 1.09 Ϯ 0.007, respectively). No significant difference was, however, observed in the carotenoid absorption bands under standard growth conditions, and the three strains exhibit similar rates of oxygen evolution.
Effect of the Light Regime on Oxygen Evolution-Transfer of wild type Synechocystis PCC 6803 cells from a moderate photon flux density (30 mol of photon⅐m Ϫ2 ⅐s Ϫ1 ) to Ͼ300 mol of photon⅐m Ϫ2 ⅐s Ϫ1 produces photoinhibition. Oxygen evolution is then reduced as a consequence of partial PSII inactivation. The two mutants behave similarly to the wild type upon a shift to 1500 mol of photon⅐m Ϫ2 ⅐s Ϫ1 (only slr2100 is shown, Fig. 1B). In contrast, differences were observed in the response to a UV-B stress. After a 2-h exposure to UV-B (at an intensity in the range found in natural environments), oxygen evolution decreased by ϳ29% for the wild type and ϳ40% for slr2100 ( Fig. 2A). The sll1624 mutant, however, does not differ from the wild type (Fig. 2B).
To determine the origin of this increased sensitivity of the slr2100 mutant to UV-B radiation, the same experiment was repeated in the presence of lincomycin, a translation inhibitor. For the wild type the cGMP and Adaptation to a UV-B Stress OCTOBER 7, 2005 • VOLUME 280 • NUMBER 40 relative decrease in O 2 evolution was much larger in the presence of lincomycin (50 versus 29%) than for the slr2100 mutant (55 versus 40%). The impairment of O 2 evolution for the wild type in the presence of lincomycin resembles that observed for slr2100 without the inhibitor. The simplest explanation is that the mutation has an effect on the cascade of events required for the repair of the damaged PSII centers, which is known to require de novo protein synthesis (43). Upon transfer back to medium intensity visible light (50 mol⅐m Ϫ2 ⅐s Ϫ1 ) after the 2-h UV exposure, both wild type and mutant strains recover the same level of O 2 evolution (90 -100% of the initial value within 1 h). PSII inactivation is, thus, fully reversible for the two strains ( Fig. 2A).
Effect of UV-B on Fluorescence Relaxation-To ascertain the previous results, the relaxation of the fluorescence induced by a saturating flash delivered to dark-adapted cells in the presence of 3-(3,4-dichlorophenyl)-1,1-dimethylurea was monitored for both the wild type and slr2100 strains. Recordings were made with cells sampled before the treatment, 2 h after the UV-B exposure, and after 1 h of recovery under white light conditions (Fig. 3A). As previously described (19), sigmoidal type curves are obtained when the fluorescence intensity was plotted against the log of the time. This decay was assigned to the back recombination of Q A Ϫ with the S 2 state of the water oxidation complex. Such experiments, thus, give information on the functional integrity of the redox components on the donor side of PSII.
The half-life for relaxation is 0.6 s for the two strains when grown under standard conditions (Fig. 3A). After 2 h of a UV-B treatment, the decay was faster for the wild type (half-life, 0.4 s), indicating that Q A Ϫ cannot recombine with the S 2 state in part of the centers and exchanges with redox components closer to Q A Ϫ like the redox-active tyrosine (Tyr-Z) (19,44). The difference was even more pronounced for the slr2100 mutant (half-life, 0.3 s). This result indicates that the manganese cluster of the PSII centers was more damaged by the UV-B treatment in the mutant than in the wild type. After the recovery phase, the fluorescence relaxation profiles are similar for both strains (Fig. 3C), confirming that the mutation likely affects the PSII repair process that takes place during the UV-B exposure. D1 Contents of the Thylakoid Membrane-The similarity in the O 2 evolution profile found between the slr2100 mutant and the lincomy-

cGMP and Adaptation to a UV-B Stress
cin-treated wild type strain ( Fig. 2A) pointed to a defect in the PSII repair process. The content in mature D1 protein of the thylakoid membrane was, thus, determined by immuno-reaction with a specific antiserum. As shown in Fig. 4, there is a significant difference between the two strains, showing faster loss of the D1 protein in the mutant, which confirms an impairment of D1 turnover in slr2100.
Cyclic Nucleotide Contents and Effect of a Specific Phosphodiesterase Inhibitor-The intracellular levels of cyclic nucleotides were determined for the two strains after various periods of UV-B exposure. No significant difference was detected in the cAMP concentration in the cells kept under standard conditions (T 0 ) as well as during the first 10 min of the UV-B treatment (Fig. 5A). In contrast, under standard conditions the steady-state level of cGMP is significantly higher in slr2100 than in the wild type strain, and it almost does not vary after the UV-B exposure (Fig. 5B). For the wild type, the cGMP concentration drops rapidly by about 40% within the first 10 min of the UV-B treatment. The UV-B stress, thus, specifically affects the intracellular cGMP concentration, and slr2100 is impaired in its ability to regulate it.
To confirm that the observed mutant phenotype could be linked to its high and not regulated level of cGMP, which may result from the lack of a phosphodiesterase activity, inhibitors were tested. Dipyridamole is known to inhibit cGMP phosphodiesterases (45). O 2 evolution was monitored in wild type cells after a UV-B exposure in the presence of dipyridamole. Under these conditions cells were more sensitive than without the inhibitor (Fig. 2C). The extent of inhibition of O 2 evolution observed under the UV-B stress in the presence of dipyridamole (47%) is similar to that recorded for the slr2100 mutant without inhibitor (42%). In both instances cells fully recover after white light exposure, as does the mutant.

Characterization of the slr2100 Gene
Because the slr2100 gene product can modify the response to a UV-B stress and the intracellular level of cGMP, we looked for putative regulatory sequences in front of slr2100. In Cyanobase (www.hazusa.or.ip) (46,47), the Synechocystis PCC 6803 slr2100 sequence was annotated as an open reading frame composed of 368 amino acids, with a calculated molecular mass of 40,673 Da. No obvious ribosome binding site lies in front of the proposed ATG. The latter is nevertheless likely to be the start codon because: (i) the CheY-like phospho-acceptor domain (see below) begins immediately downstream of this ATG, and (ii) there is no other potential initiation codon preceded by a recognizable Shine-Dalgarno sequence in front of this domain. A lack of consensus ribosome   OCTOBER 7, 2005 • VOLUME 280 • NUMBER 40 binding site has already been observed in cyanobacteria, even in front of the psbA genes, which encode the very high turnover D1 protein of photosystem II and phycobiliprotein genes (48).

cGMP and Adaptation to a UV-B Stress
Primer extension analysis was performed using a synthetic 32 P endlabeled primer (nucleotides 6 -26 from the slr2100 ATG) that allowed to map a stable 5Ј end for the slr2100 transcripts (Fig. 6A). This potential transcriptional start site is located 22 bp upstream from the slr2100 ATG and is preceded by the sequence 5Ј-TTGCCC-N 19 -TATACA-N 7 -T-3Ј (Fig. 6B). Given its sequence and the spacing between the putative Ϫ10 and Ϫ35 boxes, this promoter could be either a poor SigA-dependent or a class II-dependent promoter. No consensus DNA sequence known as the target for transcriptional regulators (SYCRP1 or NtcA; see for example Refs. 49 and 50) could be detected in the slr2099-2100 intergenic region. Only an imperfect match (GTAATTTTTCTCAC) to the consensus E. coli CRP binding site (TGTGAT-N 4 -ATCACA) can be found 280 bp downstream from the slr2100 transcriptional start site.

Transcriptome Analysis
Preliminary experiments were performed using macroarrays that contain PCR products for about 400 genes (ϳ20% of the transcription units). Most of the genes chosen encode products for photosynthetic related processes and stress responses or known regulatory proteins. Representatives of the main cellular functions have also been included (see Supplemental Table 1s). Transcription profiles of slr2100 and wild type strains grown under standard conditions and after a 30-min UV-B exposure have been compared. Transcript levels were determined by measuring counts from the nylon membrane after hybridization with 32 P-labeled retrotranscription products directly obtained from total RNAs without any amplification step. As could be expected, mRNA levels vary more than 4 orders of magnitude, photosynthesis-related (psa, psb, cpc, glgA, and ftrC) and ribosomal protein (rps2) genes being the most expressed. Comparing slr2100 and wild type strains adapted under standard growth conditions, changes by a factor of 2 or more  were observed for 30 genes. In particular, the mRNA level for slr0093 (dnaJ), slr2076 (groEL1), sll1932 (dnaK3), and sll0430 (htpG), which code for molecular chaperones, is 2-3 times lower in slr2100. This means that the mutant is likely affected in its abilities to properly fold or unfold polypeptides. After a 30-min UV-B exposure, no very significant differences were noticed for the wild type compared with the global analysis reported during the course of this work by Huang et al. (9). The observed changes likely result from the use in the present study of a lower and more "natural" UV-B irradiance. Changes by a factor of 2 or more between slr2100 and the wild type were observed for 17 genes (data not shown).
These results were used to select genes for a more precise analysis. We more specifically looked at genes linked to cNMPs, related to PSII and its turnover, as well as at a few genes that have been proposed to participate in light-sensing and adaptation (TABLE TWO). Transcript levels were measured using quantitative reverse transcription-PCR for 32 genes, with transcript levels normalized to that of rnpB, the gene coding for the RNA component of the RNase P (Fig. 7). This gene is classically used as an internal standard, its transcription assumed to be the less affected by the growth conditions.
As expected, because the gene was deleted, no slr2100 transcript was found in the mutant. For 23 genes we did not detect significant differences between the wild type and slr2100 strains grown under standard conditions (Fig. 7A). Thus, when the cells are adapted to their environment, the higher cGMP steady-state concentration present in the slr2100 cells does not affect the transcript level for these genes. A decrease by a factor of about 2 was observed for: (i) two photosynthesisrelated genes (psbA3 cpcB), (ii) two genes involved in PSII turnover, phb1 (slr1106) and ftsH (slr0228), and (iii) a gene encoding a two-component system histidine kinase (hik33) proposed as a global regulator that helps coordinate cellular metabolism with growth limitations imposed by environmental conditions (51).
After the UV-B exposure, transcript levels are affected for 22 of the 32 genes in the slr2100 mutant (Fig. 7C). An increase similar to or higher than that of the wild type was monitored for cya2, psbA2, psbA3, ftsH, nblA1, sigD, hliD, hik14, and gifB. For psbA3 and ftsH, which were less expressed in slr2100 than in the wild type under standard conditions, the extent of increase is more pronounced, with levels similar in the two strains after 15 min of UV-B. Enhanced levels were also observed for 10 genes that were not affected in the wild type: phb2 (second prohibitin), clpC (ATP-dependent Clp protease ATPase subunit), sigA, sigB, plpA (hik3, phytochrome-like protein), hik33, katG (catalase peroxidase), and interestingly, three genes related to cAMP, cya1 (adenylate cyclase), sycrp1 (cAMP receptor protein), and sll1624 (the second putative cNMP phosphodiesterase). Only two of the genes for which decreased levels were monitored in the wild type after UV-B showed the same pattern; they are cpcF and cph1. The amount of stable mRNAs that were about two times lower for cpcB and higher for slr2098 in slr2100 than in the wild type before the UV-B treatment, were not modified by it. In contrast, an opposite regulation was observed in slr2100 compared with wild type cells for five genes: phb1, cph2, rre28, and sodB showed higher levels, whereas that of slr2098 was lower.
The transcription pattern of slr2100 cells after the UV-B treatment clearly differs from that of the wild type for 20 of the 32 genes (Fig. 7D). Because both cAMP synthesis (achieved by Cya1) and that of its receptor protein (SYCRP1) increase, the expression of the genes transcribed under the control of the cAMP-SYCRP1 complex likely is specifically modified in the slr2100 mutant. The later also tends to increase protective mechanisms against damaging reactive oxygen species by producing more superoxide dismutase (SodB) and catalase peroxidase (KatG). Other important differences concern genes related to: (i) PSII repair

DISCUSSION
Because of the presence of a HD domain and of their similarity with the cAMP-dependent phosphodiesterase RegA from Dictyostelium discoideum, the two open reading frames, Sll1624 and Slr2100, have been proposed as the putative phosphodiesterases needed to ensure cyclic nucleotide homeostasis in Synechocystis PCC 6803 cells (36). This hypothesis is supported for Slr2100 by the data obtained in this study: (i) the steady-state cGMP level under standard growth conditions is two times higher in the slr2100 null mutant than in the wild type cells, (ii) after an exposure to UV-B, the decrease in intracellular cGMP that occurs in wild type does not happen in slr2100 cells, and (iii) the increased level of slr2100 mRNAs monitored in wild type cells after the UV-B treatment correlates with the decreased concentration of cGMP observed. We have observed that the mutant cells contain more cAMP after 30 min of UV-B than the wild type (data not shown), which can be correlated with the 2-fold increase in the adenylate cyclase cya1 mRNAs detected in the mutant compared with the wild type (Fig. 7). Cross-talks between cAMP and cGMP regulatory networks could, thus, exist.
When adapted to standard growth conditions, both the slr2100 and the wild type strains exhibit similar generation time, although the mutant contains less phycocyanin, meaning that light-harvesting is not rate-limiting. Accordingly, we detected a 2-fold lower steady-state level of cpcB transcripts in the mutant (Fig. 7). The expression of 3 of the five group 2 factor genes, sigA (house-keeping), sigB, and sigD, was monitored. An accumulation of sigB transcripts has already been reported to occur under stress conditions like heat shock and high salt concentration (52), and SigD has been shown to contribute to the light-induced transcription of psbA2/A3, the two genes that produce the PSII reaction center D1 polypeptide (53). A UV-B treatment leads to increased expression of sigD in wild type cells and of the three genes in slr2100. To achieve a similar growth rate, the mutant must compensate the lack of Slr2100 protein by establishing new equilibria and likely has different enzyme and regulatory protein complements. Accordingly, not all of the genes are similarly affected by the UV-B exposure.
One such difference between the wild type and mutant strain concerns the level of expression of slr2098 (hik21) and slr2099 (hik40), two genes that encode two-component hybrid histidine kinases located immediately upstream from slr2100. It was proposed that these three genes together with slr2104 (the fourth open reading frame downstream of slr2100) could form a phosphorelay cascade (36). Slr2098 is a 1261-amino acid-long polypeptide with three MHYT domains at its N-terminal end followed by PAS-PAC, histidine kinase, two response regulators, and a histidine phospho-transfer domain. The newly identified MHYT conserved domain consists of six transmembrane segments, three of which contain conserved methionine, histidine, and tyrosine residues that are projected to lie near the outer face of the cytoplasmic membrane (54). A model of the membrane topology of the MHYT domain indicates that its conserved residues could coordinate one or two copper ions, suggesting a role in sensing oxygen, CO, or NO. Slr2099 is a 366-amino acid-long response regulator preceding the histidine kinase. Slr2104 is another more complex hybrid kinase (950 amino acids long) with the succession of GAF, PAC, histidine kinase, response regulator, and histidine phospho-transfer domains. Interestingly, as the acronym refers to (cGMP phosphodiesterase-adenylyl cyclases-bacterial transcription factors FhlA), GAF domains could bind cGMP. Based on two-hybrid experiment data, direct interactions and possible phospho-transfers have been proposed between Slr2098 (Hik21) and Slr2104 (Hik22) as well as from Slr2099 (Hik40) to Slr2100 (Rre20) (www.genome.jp/dbget-bin/show_pathway?syn02020). The lack of Slr2100 in the mutant clearly has consequences on the expression of slr2098 and slr2099 (Fig. 7).
Excess light energy as well as UV-B radiation provokes cell damages, in particular at the level of the photosynthetic apparatus. For PSII repair, photodamaged D1 polypeptides must be degraded and replaced by newly synthesized polypeptides. Compared with the wild type, neither slr2100 nor sll1624 exhibits a different behavior toward photoinhibition, but the former mutant clearly shows an increased sensitivity to UV-B radiation. The experiments of O 2 evolution and relaxation of variable fluorescence demonstrate that PSII in slr2100 is damaged to an extent similar to that found for wild type cells in which translation is inhibited during the UV-B exposure. The decreased amount of D1 protein found in the thylakoids of the mutant strain exposed to UV-B compared with that of the wild type (Fig. 4) demonstrates that the degradation part of the repair cycle is not affected by the lack of the slr2100 gene. This idea is also consistent with the response of the slr0228 gene, which codes for FtsH, a key protease for PSII repair after high light (55) and UV-B light (56) damages. The level of slr0228 and psbA3 is 2-fold lower in slr2100 than in the wild type under standard conditions. The mutant cells are, thus, partly deprived of essential components for the PSII repair when exposed to UV-B radiation. However, the transcription of ftsH increases more in the mutant than in the wild type so that the ftsH mRNA level in slr2100 reaches that of the wild type after 15 min of UV-B treatment, ensuring efficient D1 degradation. The replacement of UVdamaged and degraded D1 by new copies requires an enhanced transcription of psbA genes, especially of psbA3 (21), followed by translation and incorporation of new D1 copies into the PSII reaction center complex. Although the psbA mRNA levels are lower in the mutant than in the wild type under standard conditions, the higher extent of UV-B induction compensates for this effect in the slr2100 mutant under UV-B exposure. The similar and higher mRNA amounts in UV-B-stressed wild type and mutant cells show that it is not the abundance of psbA transcripts that limits D1 synthesis. Therefore, the unregulated cGMP concentration observed in the mutant under the conditions of UV-B exposure should affect either the translation of psbA mRNA or the incorporation of newly synthesized D1 into the PSII reaction center. Altogether, our data fit with the observation that the repair of damaged PSII differs depending on whether the inhibition was induced by UV-B or high light (17), since the difference in the repair efficiency between the slr2100 and wild type strains only exists when cells are exposed to UV-B and not during photoinhibition by visible light.
This work points to a role for the cyclic nucleotides, more specifically cGMP, in the regulation and adaptation of the Synechocystis PCC 6803 photosynthetic apparatus to a UV-B stress. Even though functions such as UV-photo-protective mechanisms are common to many species, the regulatory networks by which different cyanobacteria respond to such stress seem to be achieved by different sets of proteins, or at least the domains that interact may belong to proteins that do not have the same multimodular arrangements. These differences highlight the plasticity of the ancestral cyanobacterial genome and the success of the cyanobacteria in colonizing quite different ecosystems throughout the ϳ3 billion years of evolution.