ChlR Protein of Synechococcus sp. PCC 7002 Is a Transcription Activator That Uses an Oxygen-sensitive [4Fe-4S] Cluster to Control Genes involved in Pigment Biosynthesis*

Background: A four-gene operon, acsFII-ho2-hemN-desF, is most highly expressed under microoxic conditions. Results: A chlR mutant cannot express this operon, and ChlR forms homodimers ligating a [4Fe-4S] cluster. Conclusion: ChlR is a transcription activator that utilizes an oxygen-labile [4Fe-4S] cluster to sense oxygen. Significance: ChlR is a simple regulatory element that could facilitate expression of O2-sensitive proteins. Synechococcus sp. PCC 7002 and many other cyanobacteria have two genes that encode key enzymes involved in chlorophyll a, biliverdin, and heme biosynthesis: acsFI/acsFII, ho1/ho2, and hemF/hemN. Under atmospheric O2 levels, AcsFI synthesizes 3,8-divinyl protochlorophyllide from Mg-protoporphyrin IX monomethyl ester, Ho1 oxidatively cleaves heme to form biliverdin, and HemF oxidizes coproporphyrinogen III to protoporphyrinogen IX. Under microoxic conditions, another set of genes directs the synthesis of alternative enzymes AcsFII, Ho2, and HemN. In Synechococcus sp. PCC 7002, open reading frame SynPCC7002_A1993 encodes a MarR family transcriptional regulator, which is located immediately upstream from the operon comprising acsFII, ho2, hemN, and desF (the latter encodes a putative fatty acid desaturase). Deletion and complementation analyses showed that this gene, denoted chlR, is a transcriptional activator that is essential for transcription of the acsFII-ho2-hemN-desF operon under microoxic conditions. Global transcriptome analyses showed that ChlR controls the expression of only these four genes. Co-expression of chlR with a yfp reporter gene under the control of the acsFII promoter from Synechocystis sp. PCC 6803 in Escherichia coli demonstrated that no other cyanobacterium-specific components are required for proper functioning of this regulatory circuit. A combination of analytical methods and Mössbauer and EPR spectroscopies showed that reconstituted, recombinant ChlR forms homodimers that harbor one oxygen-sensitive [4Fe-4S] cluster. We conclude that ChlR is a transcriptional activator that uses a [4Fe-4S] cluster to sense O2 levels and thereby control the expression of the acsFII-ho2-hemN-desF operon.

Microorganisms acclimate continuously to changes in their physicochemical environments, including changes in nutrient availability, energy sources, salinity, pH, and temperature. Light is the energy source for phototrophic organisms; and thus, light wavelength and total irradiance are usually among the most important environmental factors for phototrophic organisms. Cyanobacteria perform oxygenic photosynthesis; and therefore, the supply of reducing equivalents generated by the photosynthetic apparatus increases with increasing irradiance, but the oxygen levels inside cells as well as in the immediate micro-environment can fluctuate significantly because of photosystem II activity and changes in respiratory oxygen uptake (1). Thus, cyanobacteria can rapidly acclimate to diurnal shifts in light availability, to rapidly changing irradiance levels throughout the day, and to the accompanying changes in oxygen levels that result. Adjustments can be made by several mechanisms, including changes in gene expression; protein maturation, assembly and stability; post-translational modifications of enzymes; and even substrate availability for O 2 -dependent enzymes.
For the model cyanobacterium Synechococcus sp. PCC 7002 (hereafter Synechococcus 7002), the impact of physicochemical parameters has been extensively studied using global systems biological approaches, including transcriptomics, proteomics, and metabolomics (2)(3)(4)(5)(6). Some changes in culture conditions, such as dark incubation, cause dramatic changes in the transcriptome (2), whereas other conditions, such as limitation for nutrients, cause more specific and limited short term responses (3,4). Large changes in transcript levels for the acsF II -ho2-hemN-desF operon, which encodes key enzymes required for chlorophyll a, phycocyanobilin, heme, and lipid biosynthesis, were observed in response to changes in oxygen (2). A similar gene cluster comprising the open reading frames (ORFs) sll1874 (acsF II /chlA II ), sll1875 (ho2), and sll1876 (hemN) occurs in Synechocystis sp. PCC 6803 (hereafter Synechocystis 6803), and transcript levels for these three genes also coordinately increase under microoxic conditions (7).
The acsF II , ho2, and hemN genes encode important enzymes that are involved in pigment biosynthesis: Mg-protoporphyrin monomethyl ester oxidative ring cyclase, heme oxygenase, and coproporphyrinogen III oxidase, respectively. Many cyanobacterial genomes encode two genes that produce alternative enzymes for these functions. For example, in Synechocystis 6803, there are two Mg-protoporphyrin monomethyl ester oxidative ring cyclases, AcsF I /ChlA I (sll1214) and AcsF II /ChlA II (sll1874), respectively. AcsF I /ChlA I catalyzes the formation of the isocyclic ring of chlorophyll a (converting Mg-protoporphyrin IX monomethyl ester into 3,8-divinyl protochlorophyllide) under atmospheric oxygen levels, whereas under microoxic growth conditions, the same reaction is mainly performed by AcsF II /ChlA II (7). Interestingly, both enzymes are monooxygenases and require molecular O 2 as a substrate. These enzymes are unrelated to Mg-protoporphyrin monomethyl ester oxidative ring cyclases of the BchE type, which are O 2 -independent enzymes of the radical S-adenosylmethionine superfamily that use water as the oxygen donor for formation of the 13 1 -oxo group of (bacterio) chlorophylls and are widely distributed in anoxygenic phototrophs (8 -11).
Cyanobacteria as well as rhodophyte, glaucophyte, and cryptomonad algae use phycobiliproteins as major antenna pigments for photosynthesis (12,13). Because phycobiliproteins have numerous linear tetrapyrrole (phycobilin) chromophores, these organisms are critically dependent on heme oxygenase activity for the production of the bilin chromophores of the light-harvesting phycobiliproteins. Synechocystis 6803 has two heme oxygenases, Ho1 (sll1184) and Ho2 (sll1875), which can oxidatively cleave the heme macrocycle in a reaction that requires both O 2 and reducing equivalents and which produces carbon monoxide and biliverdin, the precursor of all linear tetrapyrrole pigments (14). Like Ho1 of Synechocystis 6803 and the heme oxygenases of other organisms, Ho2 also requires molecular oxygen for its catalytic activity (15). Under microoxic conditions and especially at high irradiance levels, Ho2 is the major enzyme cleaving heme to form biliverdin. Ho1 catalyzes the same reaction under atmospheric O 2 levels (16,17).
The third gene of the cluster expressed under microoxic conditions, hemN (sll1876), encodes an oxygen-independent coproporphyrinogen III oxidase (HemN type), which belongs to the radical S-adenosylmethionine protein superfamily (18). It harbors an oxygen-sensitive [4Fe-4S] cluster and requires S-adenosylmethionine for catalysis (19,20). Besides hemN, the Synechocystis 6803 genome includes a gene for an oxygen-dependent coproporphyrinogen III oxidase (HemF, sll1185), which is a monooxygenase with a binuclear iron center (21,22). In Synechocystis 6803, hemN has been shown to be induced under microoxic conditions, but hemF is required for growth at atmospheric O 2 levels (23). The role of a second hemN-like gene (sll1917) in Synechocystis 6803 remains unclear (23).
A microarray study in Synechocystis 6803 showed that the psbA1 gene, encoding an alternative D1 subunit of photosystem II, and the acsF II -ho2-hemN operon were the only four genes for which transcript levels increased substantially under microoxic conditions (24). Recently, a MarR-type transcriptional activator (sll1512) that apparently controls the expression of these four genes was described (25).
While Aoki et al. (25) were studying the product of sll1512, we identified a similar transcriptional regulator, the product of ORF SYNPCC7002_A1993 in Synechococcus 7002. In the stud-ies presented here, we show that the product of this ORF, ChlR, is a transcription activator that controls the expression of a single operon encoding four genes in Synechococcus 7002. Additionally, we show that this transcription factor activates transcription from the acsF II promoter of Synechocystis 6803 in the absence of any other cyanobacterium-specific factors in Escherichia coli. Finally, we show that this transcription activator harbors a single oxygen-sensitive [4Fe-4S] cluster per homodimer that acts as the O 2 sensory prosthetic group.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-Synechococcus sp. strain PCC 7002 wild-type and mutant strains were maintained in liquid culture and on 1.5% agar plates in medium A supplemented with 1 mg of NaNO 3 ml Ϫ1 (designated as medium A ϩ ) as described previously (2,26). Cultures of the SynPCC7002_ A1993 (ChlR) overexpression strain (under control of an ammonia-repressible promoter) that were used for RNA extractions were grown in a HEPES-buffered A medium (25 mM HEPES, pH 8.0 replaced 8.3 mM Tris-HCl, pH 8.2) containing either 12 mM NaNO 3 or 10 mM NH 4 Cl (4). Liquid cultures were grown in tubes containing medium (25 ml) at 38°C with continuous irradiation with 250 mol of photons m Ϫ2 s Ϫ1 . The cultures were sparged with either 1% (v/v) CO 2 in air (standard conditions) or 1% (v/v) CO 2 in N 2 (microoxic conditions). The following antibiotic concentrations were added to the medium when appropriate: 50 g ml Ϫ1 for spectinomycin and/or 20 g ml Ϫ1 for gentamycin. Cultures for growth rate determination were cultivated without antibiotics, whereas cultures for RNA analyses were grown in the presence of the respective antibiotics. Cell growth was monitored by measuring the optical density at 730 nm (OD 730 nm ; 1.0 OD 730 nm ϭ 1.0 Ϯ 0.2 ϫ 10 8 cells ml Ϫ1 ) with a Genesys 10 spectrophotometer (Thermo-Spectronic, Rochester, NY). Cultures for RNA analyses were inoculated at an OD 730 nm between 0.05 and 0.1 from precultures that had been grown under the same conditions. When these cultures reached an OD 730 nm of 0.7, three independently grown, replicate cultures were pooled. Cells derived from 25-ml aliquots of the cultures were rapidly centrifuged (5 min, 5000 ϫ g, 4°C), and the cell pellets were rapidly frozen in liquid nitrogen and stored at Ϫ80°C until required.
Inactivation of chlR and Construction of Expression Systems-To inactivate the chlR (SynPCC7002_A1993) gene of Synechococcus 7002, ϳ1000-bp regions immediately upstream and downstream of chlR were amplified by PCR with primers 1 and 2 for the upstream sequence and primers 3 and 4 for the downstream sequence (see Table 1 for oligonucleotide sequences). Oligonucleotides 2 and 3 introduced EcoRV sites 2 bases downstream from the start codon and 6 bases upstream of the stop codon. The PCR products for the upstream and downstream flanking regions were digested with EcoRV, and the aadA gene conferring streptomycin and spectinomycin resistance was excised as a 1091-bp Eco53kI fragment from plasmid pSRA2 (plasmids used and constructed in this study are listed in Table 2). The flanking regions and the aadA cassette were purified after electrophoresis of DNA fragments on agarose gels. The fragments were mixed at a 3:1:3 ratio of the upstream flank to the antibiotic resistance cassette to the downstream flank and ligated with T4 DNA ligase. The ligation products were directly used to transform Synechococcus 7002 as described previously (27).
For inactivation of desF, the flanking regions were amplified by PCR using primers 5 and 6 (upstream) and primers 7 and 8 (downstream), respectively. The resulting PCR products were subsequently cloned as XbaI/EcoRV (upstream flanking region) and HindIII/XhoI fragments (downstream flanking region), respectively, into pBluescript II KS(ϩ); finally, the cat gene conferring chloramphenicol resistance was cloned as a 1213-bp EcoRV/HindIII fragment from pRL409 into this plasmid, resulting in pLM1. For deletion of the entire region comprising chlR, acsF II , ho2, hemN, and desF, the flanking regions were amplified using primers 3 and 9 (3Ј of chlR) and primers 7 and 8 (3Ј of desF). The resulting PCR fragments were cloned as XbaI/ EcoRV and HindIII/XhoI fragments, respectively, into pBluescript II KS(ϩ), and ermC conferring erythromycin resistance was cloned as a 1503-bp EcoRV/HindIII fragment from pRL409 into the same plasmid, resulting in pLM2. XhoI-linearized pLM1 and ScaI-linearized pLM2 were used to transform Synechococcus 7002.
To introduce the chlR gene into plasmid pAQ1 of Synechococcus 7002 under control of an inducible promoter, an expression vector having the P nrtABCD promoter of Synechocystis 6803 and aacC1 gene conferring gentamycin resistance was constructed. A 478-bp EcoRI/NdeI fragment comprising the P nrtABCD promoter region of Synechocystis sp. PCC 6803 and a His tag sequence from pAQ1Ex-ntR (28) was cloned into pAQ1cpcEx that had aacC1 as a drug marker, resulting in pLM3. The chlR gene from Synechococcus 7002 was amplified using primers 10 and 11, and the resulting PCR product was cleaved with NcoI and BamHI, resulting in a 199-bp NcoI fragment and a 187-bp NcoI/ BamHI fragment. The NcoI/BamHI fragment was first cloned into NcoI/BamHI-digested pLM3 followed by cloning the NcoI fragment into the NcoI-digested pLM3 derivative, finally resulting in pLM4. Both the orientation of the NcoI fragment and the whole PCR-amplified region were confirmed by sequencing. Through NcoI cloning, the His tag-coding sequence of pLM3 was removed, resulting in a non-tagged gene product. However, introduction of the NcoI site at the 5Ј-end of the chlR gene caused a threonine to  alanine exchange of the second amino acid in the resulting protein. The spectinomycin-resistant chlR deletion mutant strain was transformed with ScaI-linearized pLM4. A Strep-tagged variant of ChlR was constructed for expression and subsequent purification from E. coli. An N-terminal Strep tag-coding sequence and restriction sites were fused to the chlR coding sequence by PCR using primers 12 and 13, and the resulting PCR product was cleaved with NdeI and BamHI, resulting in a 422-bp NdeI/BamHI fragment. This fragment was cloned into NdeI/BamHI-digested pET-42b expression vector, yielding pLM5.
A reporter construct based on the yellow fluorescent protein (YFP) was designed for testing expression levels from the acsF II promoter sequence. To probe expression in Synechococcus 7002, the acsF II promoter region of Synechocystis 6803 was amplified by PCR using primers 14 and 15. The PCR product was digested with EcoRI and NcoI, and the resulting 581-bp EcoRI/NcoI fragment was cloned into EcoRI/NcoI-digested pLM3, replacing the P nrtABCD promoter and yielding pLM6. Finally, ScaI-linearized pLM6 was transformed into wild-type Synechococcus 7002. For expression experiments in E. coli, a 1515-bp EcoRI/XbaI fragment from pLM6, including the Synechocystis 6803 P acsF-II sequence and yfp, was cloned into an EcoRI/XbaI fragment of pCDFDuet TM -1, which only included the origin of replication and the aadA resistance marker, yielding pLM7. For expression experiments in E. coli, plasmids pLM5 and pLM7 were transformed either separately or together into E. coli BL21(DE3). All expression constructions for Synechococcus 7002 and E. coli were verified by DNA sequencing.
RNA Preparation, RNA Sequencing, and Data Analysis-RNA samples for subsequent cDNA library construction were prepared from frozen cell pellets derived from 25-ml aliquots of liquid culture (pooled from three independent cultures). The RNA preparation and quantitation were performed as described previously (2). Construction of cDNA libraries and SOLiD TM sequencing was performed in the Genomics Core Facility at The Pennsylvania State University (University Park, PA). The cDNA libraries were constructed using the SOLiD Whole Transcriptome Analysis kit (Applied Biosystems) and were barcoded by using the SOLiD Transcriptome Multiplexing kit (Applied Biosystems). The SOLiD ePCR kit and SOLiD Bead Enrichment kit (both Applied Biosystems) were used for processing the samples for sequencing, and the SOLiD 5500 protocol (Applied Biosystems) was used for sequencing. The sequence data have been submitted to the NCBI Sequence Read Archive (SRA) under accession numbers SRX275952 to SRX275958 and SRX268824.
Mapping of cDNA sequences was performed by using the Burrows-Wheeler algorithm (29). The Synechococcus 7002 genome and the sequences for the various drug marker cassettes used were used as the reference genome; four mismatches within the 50-bp reads were allowed (Ͼ90% sequence identity). Sequences mapping to ribosomal RNA-coding regions and reads that did not map uniquely were disregarded. The methods for counting the sequences covering each ORF, for calculating the relative transcript abundance (RTA) 3 for each ORF, and for comparing RTAs between different data sets were performed as described previously (2). The probability (p values) for equal RTA in the respective comparisons was calculated for each ORF either by using the z-test or the 2 test as appropriate (for more details, see Ref. 2). The data for all protein-coding ORFs derived from these analyses are listed in supplemental Tables S1 and S2.
Overexpression and Purification of N-terminally Strep-tagged ChlR-For production of N-terminally Strep-tagged ChlR (N Strep ChlR), E. coli cells were cultivated in 1 liter of Luria-Bertani medium (10 g liter Ϫ1 tryptone, 5 g liter Ϫ1 yeast extract, 10 g liter Ϫ1 NaCl) containing 30 g ml Ϫ1 kanamycin in 2.8-liter flat bottom flasks at 37°C with shaking at 120 rpm. When the cultures reached an OD 600 nm of 0.6, chlR expression was induced by adding 0.1 mM isopropyl 1-thio-␤-D-galactopyranoside, and the culture was further incubated for 4 h. Cells were harvested by centrifugation (10 min, 5000 ϫ g, 4°C), washed in Buffer A (100 mM Tris/HCl buffer at pH 8.0, 150 mM NaCl), frozen in liquid nitrogen, and stored at Ϫ80°C until required.
Purification of N Strep ChlR was performed under oxic conditions. Cells were resuspended in Buffer A at a ratio of 8 ml of buffer/1 g of cells (wet weight), and lysozyme was added to a final concentration of 1 mg ml Ϫ1 . The cell suspension was stirred at room temperature for 20 min, and cells were disrupted by sonication (Branson Sonifier 450, three intervals at 30 s at level 2.5 and 70%). The sample was kept on ice during sonication. The cell extract was cleared by ultracentrifugation (100,000 ϫ g, 60 min, 4°C). The resulting soluble extract was loaded on a column containing Strep-Tactin Superflow resin (Iba, Göttingen, Germany) (5-ml bed volume equilibrated with Buffer A). The column was washed with 10 column volumes of Buffer A and eluted with 5 ϫ 3 ml of elution buffer (Buffer A containing 5 mM D-desthiobiotin). Fractions were collected, and protein-containing fractions were pooled and concentrated using a centrifugal ultrafiltration device (Amicon Ultra centrifugal filter, Ultracel 10,000, EMD Millipore, Billerica, MA; 5000 ϫ g). The sample was washed twice using Buffer A and concentrated again to dilute the D-desthiobiotin. The protein concentration was determined by the Bradford method (30) using bovine serum albumin as a standard. The purity of the samples was estimated by visual inspection of SDS-polyacrylamide gels stained by Coomassie Brilliant Blue G-250 (31). Quantitative amino acid analysis of N Strep ChlR samples was performed at the Molecular Structure Facility at the University of California-Davis.
Fe-S Cluster Reconstitution and Chemical Determination of Iron-Reconstitution of Fe-S clusters was performed under anoxic conditions as described previously (32). In brief, dithiothreitol (10 mM final concentration), FeCl 3 (0.5 mM final concentration), and Na 2 S (0.8 mM final concentration) were added stepwise to a solution of N Strep ChlR (at a final concentration of 0.1 mM in Buffer A). After reconstitution, the sample was concentrated by ultrafiltration (Amicon Pressure Cell, EMD Millipore) with a 10,000-molecular weight cutoff membrane and subjected to ion-exchange chromatography. Molecular sieve chromatography of N Strep ChlR was carried out according to a procedure described previously (33) using an ÄKTA (GE Healthcare) liquid chromatography system in a Coy anaerobic chamber. The sample was purified on a HiPrep 16/60 Sephacryl S-200 HR column (GE Healthcare) equilibrated with Buffer A at a flow rate of 0.5 ml min Ϫ1 . This column was previously calibrated with molecular weight markers as described (34). Protein-containing fractions were pooled and concentrated using a centrifugal ultrafiltration device (Amicon Ultra centrifugal filter, Ultracel 10,000, EMD Millipore). Samples prepared for Mössbauer spectroscopy were reconstituted using 57 Fe by the same procedure. Chemical iron determination was performed as described previously (35,36).
UV-Visible Spectroscopy, Low Temperature EPR, and Mössbauer Spectroscopy-UV-visible spectra were recorded using a modified Cary 14 spectrophotometer (On-Line Instrument Systems, Bogart, GA). Diluted, reconstituted N Strep ChlR samples were aliquoted into air-tight quartz cuvettes under anoxic conditions. After spectra of the anoxic samples were recorded, the cuvettes were opened to expose the sample to oxygen. After mixing the sample with air and incubating the sample for 20 min at room temperature, spectra of the oxygen-exposed samples were recorded. N Strep ChlR (300 l, 57 Fe-labeled) that had been reconstituted anoxically for Mössbauer spectroscopy was aliquoted into Mössbauer cups under anoxic conditions, frozen, and stored in liquid nitrogen until measured. A sample of O 2 -exposed reconstituted N Strep ChlR was prepared by exposing the above sample to air, pipetting to mix the solution several times, and then incubating it at ambient atmosphere on ice for 15 min before freezing the sample in liquid nitrogen. Samples for EPR spectroscopy were prepared by diluting the samples 10-fold with Buffer A, placed in calibrated EPR tubes (Wilmad Lab Glass, Vineland, NJ), and frozen anoxically by slow immersion of the tube in liquid nitrogen. EPR samples of reduced N Strep ChlR were prepared by addition of 9 l of 100 mM sodium dithionite in 1 M HEPES buffer, pH 7.5 to a 300-l sample. After adding dithionite solution, the sample was mixed, incubated for a few minutes on ice, and then frozen as described above.
EPR spectra were acquired on a Bruker ESP300 CW X-Band spectrometer (operating at approximately 9.48 GHz) equipped with a rectangular cavity (TE102) and a continuous flow cryostat (Oxford 910) with a temperature controller (Oxford ITC 503). Spin quantitation was carried out relative to a Cu 2ϩ -EDTA standard.
Mössbauer spectra were recorded on spectrometers from WEB Research (Edina, MN). The spectrometer used to acquire the weak field spectra was equipped with a Janis SVT-400 variable temperature cryostat. The spectrometer used to acquire the strong field spectra was equipped with a Janis 8TMOSS-OM-12SVT variable temperature cryostat. The external magnetic field was applied parallel to the ␥ beam. All isomer shifts quoted are relative to the centroid of the spectrum of ␣-iron metal at room temperature. Simulation of the Mössbauer spectra was carried out by using the WMOSS spectral analysis software (WEB Research, Edina, MN).
ChlR-dependent YFP Expression in Synechococcus 7002 and in E. coli-Synechococcus 7002 cells harboring the reporter P acsF-II-6803 ::yfp were grown under the same conditions described above for standard growth conditions or microoxic conditions. Cells harboring a yfp reporter gene on plasmid pAQ3 that was under the control of the very strong P cpcBA promoter (37) was the positive control, and wild-type Synechococcus 7002 cells served as the negative control. Cultures were grown to an OD 730 nm between 1.5 and 2.0, and the cell suspension was then adjusted with A ϩ medium to an OD 730 nm of 1.0 for the YFP fluorescence measurements.
For YFP fluorescence measurements in E. coli, BL21(DE3) cells were used that harbored either the expression plasmid for the regulator (pLM5), the YFP reporter plasmid (pLM7), or both plasmids. Cultures were grown in LB medium supplemented with 30 g kanamycin ml Ϫ1 for pLM5 and 50 g spectinomycin ml Ϫ1 for pLM7. An E. coli culture harboring a pAQ1Ex::P cpcBA ::yfp derivative with aacC1 as drug marker served as the positive control; 50 g gentamycin ml Ϫ1 was added to the LB medium for these cells. Primary cultures (20 ml of medium in 100-ml Erlenmeyer flasks) were incubated at 37°C on a shaker at 240 rpm. When these cultures reached an OD 600 nm of 0.3, the cultures were split. One-half was transferred into a 10-ml Erlenmeyer flask, which had almost no headspace to produce microoxic conditions. The other half was left in the 100-ml flask to provide strong aeration and thus oxic conditions. For the co-expression experiments in E. coli, expression of N Strep ChlR was not additionally induced by addition of isopropyl 1-thio-␤-D-galactopyranoside, but expression of the regulator (N Strep ChlR) occurred at a basal level due to inherent leakiness of the lacZ promoter system. The cultures were further incubated at 37°C and 240 rpm shaking for 1.25 and 2.5 h (for oxic and microoxic conditions, respectively), and the cultures were then rapidly chilled on ice to stop further growth and slow down cellular processes (final OD 600 nm values were between 1.2 and 1.9). The cells were collected by centrifugation (5 min, 5000 ϫ g, 4°C), washed in 50 mM Tris-HCl buffer at pH 8.0, and resuspended in the same buffer. Cell suspensions were adjusted with buffer to an OD 600 nm of 0.5 for YFP fluorescence measurements.
Fluorescence spectra were taken with an SLM-Aminco 8100C spectrofluorometer, which was modernized for computerized data acquisition by On-Line Instrument Systems. The cell suspensions were excited at 488 nm, and the average of three scans from 500 to 600 nm was calculated.

RESULTS
Deletion of chlR Affects Growth at Lower O 2 Levels-To investigate the function of ORF SynPCC7002_A1993, which was annotated as a putative transcriptional regulator, a deletion mutant was constructed by completely replacing the ORF by a drug resistance cassette (aadA) (Fig. 1A). After transformation, the wild-type and mutant alleles segregated completely (Fig.  1B), which indicated that chlR is not an essential gene under the conditions used for selection of transformants and segregation of alleles. Growth of the chlR mutant strain was indistinguishable from that of the wild type when agar plates were incubated in air (Fig. 2A). However, when plates were incubated under an O 2 -free atmosphere (in a glove box with an atmosphere of CO 2 : H 2 :N 2 , 10:10:80, v/v/v), the chlR mutant failed to grow, and only when traces of oxygen became available from wild-type cells could very small colonies form (Fig. 2B). When cultivated in liquid cultures under microoxic conditions (i.e. vigorous sparging with 1% CO 2 and 99% N 2 (v/v)), the chlR only grew slightly slower than the wild type (data not shown). Mutants in which desF alone or chlR and the entire acsF II -ho2-hemN-desF operon had been deleted also failed to grow on plates in an anoxic environment (Fig. 2B). These data suggest that DesF, which is likely to replace the essential ⌬9-acyl-lipid desaturase DesC (38), may play an essential role in lipid synthesis under microoxic conditions.
Deletion of chlR Abolishes O 2 -dependent Regulation of the Low O 2 -induced Operon-In a previous study on the effects of light, darkness and oxygen on the transcriptome of Synechococcus 7002, we showed that transcript levels of an operon comprising acsF II , ho2, hemN, and desF clearly responded to changing O 2 levels. Compared with standard conditions, the relative transcript abundance for these four genes increased when cells were grown under microoxic conditions (2) (see Fig. 3A). Besides these four genes, a few other genes showed increased mRNA levels under microoxic conditions, notably isiA and isiB (Table 3). Global transcriptomes for the chlR mutant strain obtained for cultures grown under the same conditions (microoxic versus standard) were essentially identical (Fig. 3B). Transcript levels for acsF II , ho2, hemN, and desF in cells grown under these two conditions did not differ significantly (p values Ͼ0.1; Table 3). The transcriptomes of the chlR mutant and the wild type were identical when cultures were grown under standard conditions (Fig. 3C). However, when the two strains were grown under microoxic conditions, the transcript levels for acsF II , ho2, hemN, and desF were much lower in the chlR mutant compared with the wild type ( Fig. 3D and Table 3).
Controlled Expression of chlR Restores Expression of the acsF II -ho2-hemN-desF Operon-For the expression of ChlR in the chlR deletion mutant strain, the gene was reinserted by taking advantage of the previously described expression system (37) that inserts a gene of interest into pAQ1, which is the smallest plasmid of Synechococcus 7002 and which has the highest copy number of the six plasmids in this strain (37). For expression experiments, the controllable nrtABCD promoter of Synechocystis 6803 that is induced in the absence of ammonia (39) was selected to be able to control the expression levels of chlR. Fig. 3E compares the transcriptomes obtained for permissive conditions for chlR expression (i.e. cells were grown with nitrate as the nitrogen source) with the transcription data obtained for the chlR mutant when both cultures were grown with nitrate under microoxic conditions. Transcript levels for acsF II , ho2, hemN, and desF were 90 -250-fold higher in the cells expressing chlR compared with the "background" levels for these transcripts observed in the chlR deletion mutant ( Table 3). The much higher transcript levels for the acsF II -ho2-hemN-desF operon in the complemented strain clearly showed that the transcriptional regulation of these genes had been fully restored (Fig. 3E). Notably, transcript levels for isiA and isiB, which were increased in the wild-type strain under microoxic conditions, did not respond to complementation (Table 3).
A comparison of the transcriptome obtained for cells of the complemented chlR expression strain that were grown under microoxic conditions with that of cells grown at an atmospheric Replacements of chlR, desF, and the entire region encoding chlR to desF by aadA, cat, and ermC cassettes, respectively, by homologous recombination are also illustrated. B, electrophoretic analysis of PCR amplicons produced using DNA from the wild-type strain (WT) or the ⌬chlR::aadA deletion mutant of Synechococcus 7002 as template before and after digestion by NcoI. The amplicon produced from a wild-type DNA template was ϳ2400 bp, which could be digested into nearly equally sized 1200-bp fragments by NcoI. The ϳ3000-bp amplicon produced from the ⌬chlR::aadA mutant template had no NcoI sites. The primer pair used to analyze the mutation produced some nonspecific products (that could not be cleaved by NcoI) when the mutant DNA was used as template.  desF mutant (⌬desF), and a strain in which the entire region from chlR to desF was deleted (⌬chlR-desF). Cells were streaked on medium A ϩ plates without antibiotics and grown in air or in an anaerobic chamber in an atmosphere of 10% CO 2 , 10% H 2 , and 80% N 2 . All strains could grow on atmospheric O 2 levels, but only the wild type grew under anoxic conditions when the only oxygen available to the cells was that produced by photosynthesis. O 2 level with both cultures grown with nitrate as nitrogen source showed that transcript levels for acsF II , ho2, hemN, and desF were much higher under microoxic conditions, whereas isiA and isiB levels were essentially unchanged (Fig. 3F). This result confirms that regulation of only the acsF II -ho2-hemN-desF operon is highly O 2 -dependent. When the chlR expression strain was grown with ammonia as the nitrogen source (i.e. chlR expression is repressed), the transcriptomes of cultures grown under microoxic and standard (oxic) conditions were nearly identical (Fig. 4A). When the transcriptome of cells of the chlR expression strain that were grown with nitrate as nitrogen source was compared with the transcriptome of cells grown with ammonia (when both cultures were grown under standard (oxic) conditions), only a few genes had increased transcript levels: besides chlR, these genes encoded the nitrate reductase (narB), the nitrite reductase (nirA), a nitrate/nitrite transporter (nrtP), and SynPCC7002_A0125 (annotated as formate/nitrite transporter) (Fig. 4B). Very high transcript levels for acsF II , ho2, hemN, and desF, however, were observed when the transcriptomes of the chlR expression strain were compared for cells grown on nitrate versus ammonia when the conditions were microoxic (Fig. 4C). In the same comparison, transcript levels for isiA and isiB were slightly lower in nitrate-grown cells compared with ammonia-grown cells, indicating that these genes are not part of the ChlR regulon (Table 3). These experiments show that ChlR is a transcriptional activator that is required only for expression of the acsF II -ho2-hemN-desF operon.
The acsF II Promoter from Synechocystis 6803 Interacts with ChlR from Synechococcus 7002-To characterize this gene regulation mechanism further and to assess the compatibility of the oxygen-controlled systems from Synechococcus 7002 and Synechocystis 6803, we fused the acsF II (sll1874) promoter sequence from Synechocystis 6803 to the yfp gene, which encodes the yellow fluorescent protein (37,40), as a reporter gene. The expression construct was recombined into the endogenous plasmid pAQ1 of the wild-type Synechococcus 7002 (yielding pAQ1::P acsF-II-6803 ::yfp). The YFP content in cells of Synechococcus 7002 was monitored as YFP-derived fluorescence at 526 nm. In cultures harboring pAQ1::P acsF-II-6803 ::yfp, relatively strong YFP fluorescence was observed when the cultures were grown under microoxic conditions (Fig. 5). This result clearly shows that ChlR of Synechococcus 7002 recognized the Synechocystis 6803 promoter. The signal intensity reached about 25% of that observed for the positive control having yfp under the control of the strong cpcBA promoter and recombined into the plasmid pAQ3 of Synechococcus 7002 (37). Much weaker YFP fluorescence was also observed from cells that harbored the plasmid pAQ1::P acsF-II-6803 ::yfp and that were grown under standard growth conditions at atmospheric O 2 levels (Fig. 5), but the intensity was about 6-fold lower than that for cells grown under microoxic conditions, which is consistent with the much lower transcript levels observed for the acsF II -ho2-hemN-desF operon under atmospheric oxygen conditions.
Interaction of ChlR with P acsF-II-6803 Is Independent of Other Cyanobacterium-specific Elements-After demonstrating that ChlR could interact with and regulate the acsF II promoter from Synechocystis 6803, we investigated whether this interaction relies on other cyanobacterium-specific elements. Therefore, chlR (as an N-terminally Strep-tagged derivative) and yfp (under control of P acsF-II-6803 ) were expressed from separate plasmids (see "Experimental Procedures") in E. coli BL21(DE3). Maturation of the chromophore of YFP does not occur in the absence of O 2 (41). Conveniently, the conditions to allow maturation of the YFP chromophore are incompatible with transcriptional activation, so these two processes can be temporally separated. Chromophore maturation was incomplete in E. coli cultures that were grown under the nearly anoxic conditions used. To achieve full YFP fluorescence signals in E. coli cells, a prolonged incubation (about 2 h) of the cells in a buffer at room temperature and in the presence of O 2 was required (data not shown).
E. coli strains with only one of these two components (either chlR or P acsF-II-6803 ::yfp) did not show any YFP-derived fluorescence, and only the strains carrying both plasmids allowed YFP  expression in E. coli (data not shown). The YFP-derived fluorescence signal was intense in cultures grown under microoxic conditions. The maximal fluorescence intensity was about half of that obtained with the very strong cpcBA promoter, which also supports YFP expression in E. coli, for equal cell numbers (Fig. 6). Similar to the situation in Synechococcus 7002, when E. coli cells were grown under oxic conditions with strong aeration, the fluorescence signals were 4 -5-fold lower compared with cells grown under microoxic conditions. These results indicate that YFP expression controlled by ChlR relies on the O 2 level and interaction between the transcription activator and the promoter. No other components specific to cyanobacteria are apparently required.
ChlR Harbors an Oxygen-labile Fe-S Cofactor-N-terminally Strep-tagged ChlR was heterologously overproduced in E. coli BL21(DE3) and purified to homogeneity under oxic conditions by Strep-Tactin affinity chromatography (Fig. 7A). Preliminary data showed that heterologously produced N Strep ChlR contained substoichiometric amounts of iron, and UV-visible spectroscopy revealed a faint absorbance that suggested the presence of an Fe-S cofactor. The small Fe-S cluster yield suggested either that the Fe-S cluster biosynthesis machinery of the host cells could not provide sufficient Fe-S clusters to allow for complete maturation of ChlR or that the purification under oxic conditions led to its complete degradation. By enhancing the iron supply during cultivation and by conducting the purification of N Strep ChlR under anoxic conditions, the iron content and the characteristic UV-visible absorbance could be increased to some extent, but ChlR was still mostly devoid of Fe-S clusters. To determine the nature of the metal-containing cofactor, N Strep ChlR was next purified under oxic conditions from cells grown oxically, and the protein was then chemically reconstituted under anoxic conditions. After repurification of the protein from the reconstitution solution, the protein had a dark brownish color, and a broad absorption band between 370 and 440 nm was observed in the UV-visible spectrum immediately after reconstitution under an O 2 -free atmosphere (Fig. 7B). This absorption spectrum is typical of the S3 Fe charge trans-fer bands characteristic of Fe-S proteins. Quantitative amino acid analysis showed that the protein concentration values obtained by the Bradford assay overestimated the protein concentration and that a correction factor of 0.54 was required. An iron analysis for reconstituted N Strep ChlR revealed the presence of 4.06 Ϯ 0.07 iron atoms per N Strep ChlR homodimer (see below).
When solutions containing reconstituted N Strep ChlR were exposed to oxygen (by opening the cuvette, mixing the protein solution with air, and incubating at room temperature for 20 min), the absorbance in the 370 -440-nm region of the UVvisible spectrum decreased dramatically (Fig. 7B). Subsequent removal of oxygen and reduction of the same sample with 1 mM sodium dithionite did not restore the original signal but led to a further decrease of the absorbance. These data suggest that the Fe-S cluster in N Strep ChlR is irreversibly modified by oxygen.  cells that carried both plasmids pLM7 and pLM5 and that were grown under microoxic conditions or oxic conditions. E. coli cells (strain TOP10FЈ) with plasmid pAQ1Ex::P cpcBA ::yfp grown under oxic conditions served as the positive control. Cells were harvested by centrifugation, washed in 50 mM Tris-HCl buffer at pH 8.0, resuspended in the same buffer, and adjusted to an OD 600 nm of 0.5 for the measurements. Cell suspensions were incubated at room temperature and exposed to air; measurements were taken at several times until the YFP fluorescence signal developed fully (data not shown). The fluorescence spectra showing the maximum signal for all condition are as follows: P cpcBA-6803 ::yfp (pAQ1Ex::P cpcBA ::yfp) (positive control; dotted line), P acsF-II-6803 ::yfp (pLM7) and N Strep ChlR (pLM5) microoxic sample (solid line), P acsF-II-6803 ::yfp and N Strep ChlR oxic conditions (dashed line), and Tris-HCl buffer (negative control; dash/dotted line).

ChlR Is a Homodimer That Binds One [4Fe-4S] Cluster per
Homodimer-To determine the stoichiometry and type of Fe-S cofactors in ChlR, we used a combination of analytical and spectroscopic methods (42). Recombinant apo-N Strep ChlR was reconstituted under anoxic conditions, and the resulting protein was chromatographed on a Sephacryl S-200 column to remove excess Fe-S reconstitution reagents and to determine the molecular weight of the protein (Fig. 8). By comparison of the elution time with those for proteins of known molecular weight, the molecular weight of N Strep ChlR was determined to be ϳ33,000. Considering that the calculated mass of an N Strep -ChlR monomer is 15,732 Da, this result establishes that reconstituted N Strep ChlR is a homodimer in solution.
The 4.2-K/53-mT Mössbauer spectrum of N Strep ChlR reconstituted with 57 Fe under anoxic conditions (Fig. 9A, top, (43,44). In addition, the pronounced shoulder at ϳ0.5 mm/s (green arrow) suggests the presence of a small amount of [2Fe-2S] 2ϩ clusters (quadrupole doublet with ␦ ϭ 0.28 mm/s and ⌬E Q ϭ 0.54 mm/s, ϳ15% of signal intensity; green line). The remaining ϳ10% of the absorption is broad and featureless (see black arrows) and could emanate either from Fe-S clusters with half-integer spin ground state or from unspecifically bound iron. The EPR spectrum of N Strep ChlR reconstituted under anoxic conditions (Fig. 9B, red line) demonstrated largely the absence of Fe-S clusters with half-integer spin ground state. The weak signal with g av ϭ 2.006 could emanate from a small amount (ϳ0.01 eq) of [3Fe-4S] ϩ cluster, but this small amount is beyond the detection limit of Mössbauer spectroscopy. Therefore, the broad features in the Mössbauer spectrum are assigned to unspecifically bound iron. The 4.2-K/6-T Mössbauer spectrum (Fig. 9A, bottom) confirms the presence of The EPR spectrum of dithionite-reduced, reconstituted N Strep ChlR (Fig. 9B, blue line) exhibits a nearly axial spectrum  with principal g values of g x ϭ 2.04, g y ϭ 1.92, and g z ϭ 1.89 and relaxation properties characteristic of those of [4Fe-4S] ϩ clusters (45). Spin quantitation using a Cu 2ϩ standard yielded 0.77 spin per N Strep ChlR homodimer, suggesting that the [4Fe-4S] 2ϩ cluster is (nearly) quantitatively reduced.
Exposure of N Strep ChlR reconstituted anoxically to O 2 was monitored by Mössbauer and EPR spectroscopies. The Mössbauer spectra recorded before and after O 2 exposure (Fig. 10, top and middle, respectively) reveal pronounced changes. The progress of the reaction can be demonstrated from the difference spectrum (Fig. 10, bottom)

DISCUSSION
In this study, we showed that the product of ORF SYNPCC7002_A1993, annotated as a putative transcriptional regulator, is an oxygen-sensitive transcriptional activator for an operon of four genes, acsF II , ho2, hemN, and desF. Three of the resulting enzymes catalyze important reactions of photosynthetic pigment biosynthesis under microoxic conditions. HemN acts in heme biosynthesis and converts coproporphyrinogen III to protoporphyrinogen IX, AcsF II (ChlA II ) converts Mg-protoporphyrin IX monomethyl ester into 3,8-divinylprotochlorophyllide in chlorophyll a biosynthesis, and Ho2 oxidatively cleaves the tetrapyrrole ring of heme to produce biliverdin (7,15,16,23). Under atmospheric O 2 levels, these same reactions are catalyzed by HemF, AcsF I (ChlA I ), and Ho1 (Hox), respectively. Although the exact function of desF has not yet been demonstrated, based on sequence similarity, it is likely that this protein substitutes for the essential fatty acid desaturase DesC under microoxic conditions (46). The desF gene is often absent from similar operons in other cyanobacteria (e.g. Synechocystis 6803). The paralogous enzymes encoded in such operons, which are presumably optimized to function under microoxic conditions, occur in numerous cyanobacteria. Similar gene clusters like the acsF II -ho2-hemN-desF cluster of Synechococcus 7002 are present in many cyanobacterial strains from different habitats (data not shown); however, the gene composition of these clusters is variable. For example, Leptolyngbya sp. strain PCC 7376 has a similar five-gene operon that includes a psbA paralog.
A very similar MarR-type transcriptional regulator (sll1512) has recently been reported in Synechocystis 6803 (25). That study demonstrated that a constitutively activated variant of ChlR binds to the promoter sequences of acsF II (chlA II ) and psbAI. Transcript levels of these genes and of hemN and ho2, the other two genes of the acsF II (chlA II ) operon, are expressed at higher levels under microoxic conditions (24). In Synechococcus 7002, however, no changes of the transcript levels of photosystem-related genes and in particular of the three paralogs of psbA were observed in response to microoxic or anoxic conditions (2). Changes of the transcript level of a few other genes related to iron uptake or stress, isiA and isiB in particular, were observed in some microoxic samples (Table 3). This appears to be a secondary effect due to differences in iron availability under low O 2 levels because our experimental setup did not allow rigorous control of the O 2 level in liquid cultures. The results of our complementation experiments, however, showed clearly that isiA and isiB are not regulated by ChlR (Table 3).
Only the four genes of the acsF II -ho2-hemN-desF operon specifically responded to changing O 2 levels.
MarR-type transcription regulators typically form homodimers (47), and it was proposed that ChlR of Synechocystis 6803 similarly forms a homodimer that binds to the acsF II (chlA II ) promoter region when activated (25). In this study, we show experimentally that ChlR from Synechococcus 7002 forms homodimers in vitro and can activate transcription from the acsF II (chlA II ) promoter of Synechocystis 6803 in E. coli in vivo and that no other cyanobacterium-specific factors are required for O 2 sensing and transcription activation. By placing a yfp reporter gene under the control of the inducible acsF II (chlA II ) promoter and by transferring the system into both Synechococcus 7002 and E. coli, we demonstrated that this system can be used to express genes in a manner that is solely controllable by the oxygen level. FIGURE 11. Sequence comparison of ChlR homologs from diverse cyanobacteria. Cys residues are shaded in gray. The shaded aspartic acid residue in the Synechococcus 6803 sequence causes constitutive activation of ChlR when mutated to His (25). Hyphens denote insertions/deletions introduced to optimize the alignment. In the consensus line, stars indicate absolutely conserved residues, and colons and periods indicate conservative replacements. This adds to a growing list of expression systems for cyanobacteria, and this system should be especially useful for controlling the expression of hydrogenase genes in cells that are fermenting or in cells that at least have minimal intracellular O 2 levels.
MarR-type regulators comprise a broad family of transcription activators and repressors that regulate a variety of stress responses and other cellular processes, and specific binding of ligands has been shown for some examples (47). The purified, reconstituted MarR-type transcription activator ChlR of Synechococcus 7002 harbors an Fe-S cluster as an oxygen-sensitive prosthetic group. A combination of spectroscopic and analytical methods revealed the presence of 0.74 [4Fe-4S] 2ϩ clusters and 0.30 [2Fe-2S] 2ϩ clusters, i.e. a total of 1.04 clusters per ChlR homodimer. The presence of some [2Fe-2S] clusters and a minute amount of [3Fe-4S] ϩ clusters in ChlR that had been reconstituted and purified under anoxic conditions was presumably due to minor O 2 contamination during the purification process and manipulations required to prepare the protein for spectroscopic measurements. Our findings together with the previously described homodimeric structure of other MarR-type regulators as well as the ChlR homolog of Synechocystis 6803 (ChlR) (25,47) strongly suggest that one [4Fe-4S] cluster is bound per ChlR homodimer. Database searches show that ChlR-type regulators occur in numerous cyanobac-teria derived from a wide variety of habitats, but ChlR homologs do not seem to occur in all cyanobacteria. MarR family members also occur in cyanobacteria that do not have acsF II -ho2-hemN-desF operons, but it is presently unknown what these other transcription factors control.
If one considers all cyanobacterial MarR family members, only one Cys residue (the second cysteine; residue 14 in ChlR from Synechococcus 7002) is conserved (see Fig. 11). This led Aoki et al. (25) to suggest that this cysteine residue might form a sulfenic acid or S-thiolated derivative under oxic conditions (non-activating ChlR) that could be reduced to a free thiol under microoxic conditions (activating ChlR) (25). However, if one only considers those MarR family members that are most similar to ChlR (also restricting the analysis to organisms that have operons similar to acsF II -ho2-hemN-desF), a different pattern emerges (Fig. 11). Comparison of this more limited set of ChlR homologs shows that these proteins have four conserved Cys residues; two occur near the N terminus, and two occur near the C terminus. The C-terminal pair of Cys residues always occurs in the pattern Cys-X-Cys, whereas the N-terminal pair of Cys residues occurs in the motif Cys-X 4 -6 -Cys-Pro. The N-terminal Cys motif is reminiscent of a portion of the Cys motif for [4Fe-4S] ferredoxins (i.e. Cys-X 2 -Cys-X 2 -Cys-X 3 -Cys-Pro), and furthermore, these Cys residues occur in close prox- cluster (B) by the Fe-S cluster biogenesis machinery, ChlR can bind to its operator near the acsF II promoter (C) and activate transcription (D). Exposure to oxygen (E) converts the [4Fe-4S] cluster to a [2Fe-2S] cluster, and the transcription activator loses its ability to bind to the acsF II promoter (F). It is presently not clear whether an Fe-S cluster can be restored or reinserted into the inactivated protein or whether this protein turns over and must be resynthesized.