Alterations in Global Patterns of Gene Expression in Synechocystis sp. PCC 6803 in Response to Inorganic Carbon Limitation and the Inactivation of ndhR, a LysR Family Regulator*

The cyanobacterium Synechocystis sp. PCC 6803 possesses multiple inorganic carbon (Ci) uptake systems that are regulated by Ci availability. The control mechanisms of these systems and their integration with other cell functions remain to be clarified. An analysis of the changes in global gene expression in response to Ci downshift and the inactivation of ndhR (sll1594), a LysR family regulator of Ci uptake is presented in this report. Mild Ci limitation (3% CO2 (v/v) in air to air alone) induced a dramatic up-regulation of genes encoding both inducible CO2 and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} uptake systems. An induction of ndhD5/ndhD6 and other genes in a probable transcriptional unit was observed, suggesting a function in inducible Ci uptake. The expression of slr1513 and sll1735, physically clustered with sbtA and ndhF3/ndhD3/cupA, respectively, were also coordinated with upstream genes encoding the essential components for \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} and CO2 uptake. Ci limitation induced the regulatory genes slr1214, sll1292, slr1594, sigD, sigG, and sigH, among which slr1214, a two-component response regulator, showed the earliest induction, implying a role for the early response to Ci limitation. Opposite regulation of genes encoding the assimilation of carbon and nitrogen demonstrated a striking coordination of expression to balance C- and N-fluxes. The analyses revealed that ndhR inactivation up-regulated the expression of sbtA/sbtB, ndhF3/ndhD3/cupA/sll1735, and slr2006-13 including ndhD5 and ndhD6, indicating a vital role of this regulatory gene in both CO2 and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document}acquisition of the cyanobacterium. We therefore suggest that ndhR be renamed ccmR to better represent its broader regulatory characteristics.

Inorganic carbon (Ci) 1 is an essential and often limiting inorganic substrate for oxygenic photosynthesis. For aquatic photosynthetic species including cyanobacteria, Ci is acquired either as HCO 3 Ϫ or as dissolved CO 2 . Cyanobacteria have evolved a complex Ci concentrating mechanism (CCM) that functions to increase the intracellular concentration of CO 2 to overcome the low affinity of the carbon-fixing enzyme, Rubisco (1)(2)(3). The CCM is regulated to optimize growth under varying Ci availabilities. Expression of the CCM is maximal under conditions of Ci limitation, whereas constitutive levels of CCM activity are observed even under Ci replete conditions. Four Ci acquisition systems have been described for Synechocystis sp. PCC 6803 and Synechococcus sp. PCC 7942 (hereafter Synechocystis 6803 and Synechococcus 7942, respectively) (1). An ABCtype HCO 3 Ϫ transporter encoded by cmpABCD operon and a potential Na ϩ /HCO 3 Ϫ symporter encoded by sbtA have been identified and thus represent specific HCO 3 Ϫ transporters driven by ATP and the trans-cytoplasmic membrane Na ϩ gradient, respectively (4,5). CO 2 uptake (CUP) relies on a carbonic anhydrase-like entity that catalyzes the active CO 2 hydration and is energized by type I NAD(P)H dehydrogenase (NDH-1) complexes presumed to be situated on the thylakoid membrane (1,6). The CUP-mediated hydration of CO 2 inside cells maintains an inwardly directed CO 2 concentration gradient across cytoplasmic membranes such that external CO 2 continuously diffuses into cells, perhaps mainly through water channels (7). Thus, in addition to its well established roles in respiratory and cyclic photosynthetic electron transport, the cyanobacterial NDH-1 complex is also essential for CUP activity. Elegant genetic analyses in the cyanobacteria Synechococcus 7942 and Synechocystis 6803 reveal that cyanobacterial CUP activity requires specialized forms of the NDH-1 complex that appear to incorporate evolutionary modifications in the subunit composition of the canonical type I NDH respiratory complex (1, 8 -10). These alternative forms of the NDH-1 complex contain the products of variant ndhD and ndhF genes, which are found as members of paralogous multigene families. This contrasts with other core ndh genes, which are typically found as single copies in the genomes of Synechococcus 7942 and Synechocystis 6803. The CUP systems possess a third type of gene (alternatively designated as chp or cup) encoding a novel protein hypothesized to be responsible for the actual CO 2 hydration activity (8 -10). These genes are organized into two systems, one constitutive, low affinity system (ndhF4/ndhD4/cupB) and one inducible high affinity system (ndhF3/ndhD3/cupA) (11). Although the products of the ndhD1, ndhD2, ndhF1, and ndhF2 genes appear to function in cyclic and respiratory electron transport, the roles of other ndhD and ndhF paralogs, such as ndhD5 and ndhD6, remain to be elucidated (1).
Although considerable progress has been obtained in defining the structural genes required for the CCM, some progress toward understanding the regulatory circuits governing their expression has also been made. A LysR-type transcription reg-ulator, NdhR, negatively controls the expression of its own gene and NDH-1 genes (ndhF3/ndhD3) in Synechocystis 6803 (12). Interestingly, NdhR also regulates the expression of a gene encoding a Na ϩ /H ϩ antiporter (i.e. nhaS1 (slr1727)), suggesting that it may exert control over other Na ϩ or H ϩ translocation systems. The CO 2 -and Na ϩ -dependent HCO 3 Ϫ uptake activities in cyanobacteria demand Na ϩ at micromolar and millimolar levels, respectively (3). For Synechocystis 6803 cells grown at low Ci, Na ϩ is indispensable because of its low capacity for Na ϩ -independent HCO 3 Ϫ transport, in contrast to other cyanobacteria, which are found to perform efficient Na ϩ -independent HCO 3 Ϫ transport under the same conditions (13). A Na ϩ gradient drives the Na ϩ -dependent HCO 3 Ϫ transporter (SbtA). Sodium also acts as a counter ion for Na ϩ /H ϩ antiporters operating to maintain pH homeostasis during the light-dependent episodes of proton release and uptake as a result of the cytoplasmic hydration of CO 2 and the dehydration of HCO 3 Ϫ by carbonic anhydrase activity in the carboxysome, where CO 2 is carboxylated by Rubisco.
The coordinated expression of the genes encoding the key CCM components under relatively severe Ci starvation has been more recently investigated (14). To further contribute to this effort, the present paper describes the genome-wide expression patterns in Synechocystis 6803 cells subjected to relatively mild Ci limitation as analyzed using DNA microarrays. The investigation not only reveals the expression of both known and previously unknown genes likely involved in the CCM, it also provides greater insight into the genome-wide expression modulation of genes involving the processes, such as N-assimilation and photosynthesis, that allow the cyanobacterium to maximize metabolic fitness during Ci limitation. Additionally, an ndhR mutant was constructed and the analysis of the mutant has led to the better definition of critical components of a CCM regulon. The results indicate that NdhR regulates a gene cluster that harbors the ndhD5 and ndhD6 genes, as well as the Na ϩ /HCO 3 Ϫ symport genes, sbtA, and its downstream putative ORF, slr1513 (designated sbtB, more recently), thus extending the group of known NdhR-regulated Ci genes beyond the initially determined ndhF3/ndhD3/cupA operon (12).

EXPERIMENTAL PROCEDURES
Growth Conditions-The wild-type and mutant cells of Synechocystis 6803 were grown photoautotrophically at 32°C (water bath) in BG-11 medium buffered with 20 mM TES-KOH at pH 8.0, and bubbled with 3% CO 2 in air (v/v). Experimental cultures were grown in 600-ml batches in 1-liter Roux bottles (Corning). For Ci-limited culture, Na 2 CO 3 was omitted from BG-11 medium. The modified BG-11 was also buffered with 20 mM TES-KOH at pH 7.0, to inhibit the formation of HCO 3 Ϫ . The pre-cultures of the wild-type and mutant cells grown at logarithmic phase under CO 2 -enriched conditions (3% in air, v/v) were used to inoculate the experimental cultures at starting cell densities with an OD 750 nm of 0.2. A relatively mild Ci stress was applied by switching the aeration from 3% CO 2 in air (v/v) to air alone (350 ppm CO 2 , Ci-limited). Continuous illumination was provided by Cool White (General Electric) fluorescent lamps with an incident flux of ϳ50 mol m Ϫ2 s Ϫ1 as measured with a sensor (LI-COR, Lincoln, NE). Continuous measurement of relative cell density was obtained using modulated beam (850 nm) industrial turbidity monitors (Banner Engineering, Minneapolis, MN) connected to a computer to record the turbidity data. The relative optical density measured by this instrument closely parallels the more typically used OD 750 nm measured in a standard spectrophotometer because there are no major pigment absorptions in whole cells at either of these wavelengths and scattering properties of cells are similar. However, spectrophotometric OD 750 nm determinations were used when absolute optical densities were used to facilitate standardization of the experiments. Solid medium contained BG-11 buffered with 10 mM TES-KOH, pH 8.0, and was supplemented with 1.5% agar and 5 mM sodium thiosulfate.
Construction of the ndhR Deletion Mutant-Gene deletion was accomplished by transformation of the wild type with mutagenic DNA fragments prepared using a fusion PCR approach (15,16). The genomic sequence of Synechocystis 6803 was obtained through Cyanobase (www.kazusa.or.jp/cyano/cyano.html) to design the primers used to amplify the ndhR (sll1594) gene-flanking segments (Table I). Care was taken to ensure that only the target gene ORF was knocked out. The plasmid pRL563, supplied by the Wolk group (17), was used as template to amplify the selectable cassette encoding the genes for resistance to streptomycin/spectinomycin (Sm/Sp) (17). Purified genomic DNA served as the PCR template to verify the complete segregation of the ndhR deletion strain (18).
Isolation of Total RNA-After incubation under designated conditions, a 100-ml aliquot of culture was immediately combined with an equal volume of ice-cold mixture of phenol and ethanol (1:10, w/v) in an ice bath to prevent degradation of RNA (19). The resultant cells were collected by centrifugation at 1000 ϫ g for 10 min at 4°C. Total RNA was isolated with RNeasy Midi Kit (Qiagen, Valencia, CA) following the protocol from the manufacturer. The resulting RNA was further treated with the DNA-free kit (Ambion, Austin, TX) to remove trace amounts of contaminating genomic DNA.
RT-PCR Analysis-RT-PCR primers were designed to amplify 350 -400 bp of internal coding region of each gene (Table I). Reverse transcription reactions were performed with the reverse primer of each gene using Superscript II (Invitrogen). The resulting cDNA was used as the template for RT-PCR (20). Amplified products were electrophoretically examined on 1% agarose gels. The transcript abundance of rnpB encoding RNase P in each sample served as a control.
Preparation of Fluorescently Labeled Probes for DNA Microarray Analysis-Total RNA extraction for DNA microarray hybridization was as described above. Fluorescently labeled cDNA was produced via a two-step procedure involving cDNA synthesis in a reverse transcriptase reaction incorporating aminoallyl-modified deoxynucleotide, followed by the second step involving chemical coupling of fluorescent dye (either Cy3 or Cy5) to the introduced amino moieties of the synthesized cDNA (21). Briefly, 16 g of total RNA served as the template of cDNA synthesis using 10 g of random 8-base oligonucleotides (Sigma-Genosys, The Woodlands, TX) and Superscript II reverse transcriptase (Invitrogen). The resultant cDNA was purified with Millipore Microcon 30 centrifugal filter devices (Billerica, MA) and resuspended in 0.1 M Na 2 CO 3 at pH 9.0. The cDNA labeling reaction with Cy3 or Cy5 dyes (Amersham Biosciences) was carried out at room temperature in the dark for 1 h and then quenched by the addition of 4 M hydroxylamine, followed by incubation for an additional 5 min. The Cy3 or Cy5 dyecoupled cDNA samples were combined and purified using a Qiagen PCR product purification kit according to the specifications from the manufacturer.
Hybridization of DNA Microarray and Data Analysis-Labeled probes were adjusted to 14.75 l, and the remainder of the hybridization components containing 2.5 l of 10 g l Ϫ1 salmon sperm DNA, 8.75 l of 20ϫ SSC, 0.25 l of 10% SDS, and 8.75 l of formamide were added. The mixture was then heated for 2 min at 99°C and maintained at 42°C until hybridization. 3168 genes or putative ORFs of Synechocystis 6803 were printed in quadruplicate on Telechem (Sunnyvale, CA) slides (21). Printed slides were baked at 80°C for 1 h and then subjected to a UV-cross-linking at a dose of 150 mJ cm Ϫ2 . The slide was washed at room temperature in 0.1% SDS, washed in deionized water to remove unbound material, and then boiled in deionized water for 3 min to denature the printed DNA. Prehybridization was performed by incubating the slide in 25% formamide, 5ϫ SSC, 0.1% SDS, and 1% bovine serum albumin for at least 45 min at 42°C. The slide was then rinsed with distilled water and dried by low speed centrifugation.
The slide was preheated at 42°C. The pre-warmed probe sample was pipetted and spread uniformly onto a 24 ϫ 60-mm glass coverslip (Fisher Scientific, Hampton, NH), and the pre-warmed slide was inverted and placed with the arrayed surface contacting the probe sample on the coverslip surface. The slide was incubated in a static incubator at 42°C for 12-16 h and washed by placing in a 250-ml solution of 2ϫ SSC and 0.1% SDS at 42°C for 5 min with gentle agitation provided by rotation of a magnetic stir bar. The slide was transferred quickly to a 250-ml solution of 0.1ϫ SSC and 0.1% SDS, incubated for 10 min at room temperature with gentle agitation, and washed five additional times in 0.1ϫ SSC for 1 min at room temperature. The slide was then rinsed briefly with deionized water, dried by low speed centrifugation, and stored in the dark until it was scanned.
Hybridization signals from the microarray were quantified using GenePix Pro 4.1 (Axon Instruments, Union City, CA). The quality control procedures were conducted in the image analysis software, and then data were saved to Acuity 3.1 (Axon Instruments). Each microarray hybridization was normalized with the Lowess print-tip option in the microarray informatics software. Four to six microarray hybridizations of two independent biological replicates were pooled to create data sets prior to the calculation of average log 2 ratio and standard deviation.

Cultures Subjected to Ci Downshift for DNA Microarray Analysis
This study investigated both the response of wild-type cells to Ci limitation and the impact of ndhR inactivation to better define its function in Ci uptake. The ndhR deletion strain, ⌬ndhR, was constructed by replacement of the ndhR coding sequence with a spectinomycin/streptomycin resistance gene cassette using established procedures, and complete segregation of the mutant allele was verified by PCR analysis (Fig. 1). To apply low Ci stress, the wild-type and mutant cultures were subjected to a downshift in CO 2 supply by switching the aeration with 3% CO 2 in air (CO 2 -enriched) to the aeration with air alone (CO 2 -limited). To enhance the impact of the CO 2 downshift, the pools of CO 3 2Ϫ and HCO 3 Ϫ dissolved in the culture media were minimized by growing cells in Na 2 CO 3 -free BG-11 medium buffered with 20 mM TES-KOH at pH 7.0.
The wild-type and ⌬ndhR cells showed similar rates of growth with 3% CO 2 enrichment under the given growth conditions (Fig. 2). After the switch from CO 2 -enriched to Cilimited aeration, ϳ3 h elapsed before Ci limitation became manifested as a sharp decline of growth rate as monitored by continuous recording of culture turbidity (Fig. 2). The CO 2 downshift was routinely performed when the cells reached a density of 0.95 at OD 750 nm . However, it was observed that when the downshift was performed at lower cell densities (data not shown), a longer time elapsed before the onset of growth rate decrease, suggesting that levels of Ci were drawn down in a cell density-dependent manner. Although this was not quantified in the present study, recent mass spectroscopic measurements demonstrate a cell-and light-dependent draw-down of [Ci] in the media of Synechocystis 6803 cultures subjected to a more severe Ci limitation protocol involving a switch to CO 2free aeration rather than a switch to ambient CO 2 (14). Interestingly, after CO 2 downshift there was not a gradual slow-down of growth prior to the dramatic drop of apparent growth rate before the inflection period is reached. This may indicate the existence of a threshold [Ci] required for unrestrained growth (Fig. 2). Close inspection of the inflection period during the transition to decreased growth rates revealed that the turbidity of the culture remained constant for the first 20 -30 min, and then declined during subsequent 3-5 h before resuming growth. Growth resumed approximately at 10 h after the CO 2 downshift, albeit at a lower rate than the rate prior to growth inflection. The growth rate of ⌬ndhR strain was observed to be slightly lower than that of wild-type cells during the post-inflection resumption of growth. Based on these growth characteristics, sampling time points for analysis of transcript abundances using RT-PCR and DNA microarray methods were established to reveal the gene expression events occurring at the different stages after the CO 2 downshift (Fig. 2).

Transcript Abundances of Key Ci Uptake Genes
Revealed by RT-PCR The transcript abundances of cupA (sll1734), ndhhF3 (sll1732), cmpA (slr0040), and sbtA (slr1512) at 0, 30, 60, 190,  (Table I). M, 1 kb plus DNA ladder (Invitrogen, Carlsbad, CA); 1, wild-type (2185 bp); 2, ⌬ndhR (2808 bp).  and 360 min after Ci downshift described above were evaluated using semiquantitative RT-PCR. This analysis served as a preliminary screen for establishing a chronological sampling protocol for subsequent microarray analysis. As shown in Fig.  3, a slight induction of cupA, ndhF3, and sbtA expression occurred at 60 min after CO 2 downshift. A strongly induced accumulation of these transcripts was observed at 190 min during the inflection period of the growth rate decline caused by CO 2 downshift. The high abundance of these transcripts remained through the 360 min of sampling. No increase in the level of cmpA transcripts in wild-type cells was observed at 60 min, but a strong accumulation of cmpA transcripts was observed at 190 and 360 min after CO 2 downshift.
The loss of ndhR resulted in distinctly different expression patterns of these genes as compared with wild type (Fig. 3). In the mutant cells, the accumulated transcripts of ndhF3 and cupA from ndhF3 operon were observed prior to CO 2 downshift, as expected from previous results demonstrating repressor activity of NdhR (12). Unexpectedly, the same pattern was observed for sbtA expression in ndhR mutant, hinting that sbtA is also a component of the NdhR regulon (Fig. 3). Interestingly, the accumulation of cupA, ndhF3, and sbtA transcripts in the mutant was further enhanced at 360 min after CO 2 downshift, implying that the expression of the key Ci uptake genes is subjected to other regulation mechanisms in the absence of the NdhR at the post-inflection stage. NdhR repressor activity upon cmpA expression is unlikely because no accumulation of cmpA transcript was observed in ndhR mutant prior to CO 2 downshift. In fact, the accumulation of cmpA transcripts in the mutant did not occur even at 190 min after CO 2 downshift (Fig.  3). The delayed expression of cmpA in the mutant compared with wild-type may result from the constitutive (de-repressed) sbtA expression that accounts for most activity of the HCO 3 Ϫ uptake in Synechocystis 6803 (4).

Transcriptome Dynamics of Synechocystis 6803 under Ci Limitation Revealed by DNA Microarray Analysis
Based on RT-PCR analysis, wild-type cells appeared to be still unchallenged by Ci limitation at 30 min after CO 2 downshift. Therefore, the time point was not further examined by microarray analysis. However, the investigation was extended to the recovery stage by including a sampling point at 720 min after CO 2 downshift (see Fig. 2). As shown in Fig. 4A, there was only a slight change in global expression pattern even at 180 min after CO 2 downshift. This point corresponds to the time immediately prior to the growth inflection period. In contrast, a dramatic alteration of global expression happened after another 20 min (200-min sampling point, Fig. 4B), indicating that a wide-reaching change of gene regulation in response to Ci downshift occurs over a short period of time. This also suggests that [Ci] in the medium dropped below a threshold level, thereby triggering the extensive changes of global gene expression at 190 -200 min at the given cell density and growth conditions.
Using the control sample as a reference, there were 456 ORFs that consistently exhibited expression changes of 2-fold or greater (log 2 ratio Ն 1 or Յ Ϫ1) in at least one of the five sampling time points (60,180,200,360, and 720 min) after CO 2 downshift (Fig. 5). Based on the chromosomal proximity and expression pattern, 50 gene clusters were counted among the 456 ORFs (Fig. 5, C1-50). The coordinated up and down-regulation of genes physically clustered on the chromosome may be correlated with operon structure, but this is impossible to evaluate at this level of experimental analysis (22). To show the features of coordinated expression of physically clustered genes, the transcriptional dynamics of the 465 ORFs are presented according to their linear order along the Synechocystis 6803 genome (Fig. 5).
Inorganic Carbon Acquisition-The Ci limitation protocol applied here induced a dramatic change in expression of known Ci uptake genes. The expression of ndhF3, ndhD3, and cupA of the previously studied ndhF operon (9, 10) encoding inducible CO 2 uptake system were all up-regulated at 180 min after CO 2 downshift, but maximal induction was observed at 200 min. The abundance of these transcripts continued to remain high during all subsequent time points. Interestingly, the two ORFs immediately downstream of cupA, sll1735 and sll1736, were also up-regulated (Fig. 5B, C11). These two putative genes may encode additional components of the high affinity CUP system that are not essential and were thus not identified during the initial genetic screens (9, 10). Transcript abundance for cupB (slr1302) of the constitutive CO 2 uptake system, in contrast, exhibited a continuous decline in abundance following the CO 2 downshift (Ϫ1.95-fold at 720 min). There was no significant change of transcript level of other two components of the constitutive system, ndhF4 (sll0026) and ndhD4 (sll0027), physically clustered on the chromosome (see supplemental data, available in the on-line version of this article).
The expression of sbtA, cmpA, cmpB (slr0041), cmpC (slr0043), and cmpD (slr0044) genes were all strongly up-reg- ulated in response to the CO 2 downshift. However, the genes for the Na ϩ -dependent (SbtA) and ATP-consuming (Cmp) HCO 3 Ϫ transporters showed the distinct temporal patterns of induction. Like the genes of ndhF3 operon, sbtA already showed an increased accumulation of transcripts at the 180min sampling point (i.e. prior to the growth inflection). However, no increase in transcript abundance was detected for any of the genes of the cmp operon at this point, suggesting that the two systems have distinct regulatory mechanisms. It is worth noting that the RT-PCR results indicate low constitutive levels of cmp transcript were present even prior to the CO 2 downshift, whereas the microarray experiment measured the change of transcript abundance relative to the initial reference point. The co-expression of sbtA with its downstream ORF, slr1513 (designated sbtB, more recently at the Kazusa data base), after CO 2 downshift was observed (Fig. 5C, C22), suggesting the functional relatedness of the two genes. The cmpABCD genes were also coordinately up-regulated with slr0042, a potential porin protein. This ORF thus may be a component of cmp operon in Synechocystis 6803 (Fig. 5D, C46), but it may not be essential for the activity of the ABC-type HCO 3 Ϫ transporter because it is absent from the cmp operon in Synechococcus 7942 (23).
The exact functions of ndhD5 (slr2007) and ndhD6 (slr2009) remain unknown, but the expression of these two ndhD genes was coordinately up-regulated in response to the CO 2 downshift, with kinetics similar to sbtA and the high affinity CUP genes suggesting the involvement of ndhD5 and ndhD6 genes in low Ci acclimation of Synechocystis 6803. Furthermore, these genes appear to be the elements of a large transcriptional unit of which the first putative ORF is slr2006 (Fig. 5A, C7). Based on the analysis of the ndhR knockout mutant, we found that this putative transcriptional unit is also negatively controlled by NdhR, as discussed in greater depth below.
Photosynthesis and Respiration-psaD (slr0737), psaF (sll0819), psaI (smr0004), psaJ (sml0008), psaK (ssr0390), and psaL (slr1655) genes encoding the polypeptides of photosystem I complex were down-regulated under low Ci conditions. Likewise, psbF (smr0006), psbJ (smr0008), psbL (smr0007), psbM (sml0003), psbO (sll0427), and psbX (sml0002) of photosystem II were also down-regulated in response to Ci downshift. This down-regulation is consistent with a decreased demand for the products of the light reactions as a result of the shortage of inorganic substrate for the carbon-fixing reactions. In contrast, we observed the up-regulation of psbA1 (slr1181), psbA2 (slr1311), and psbA3 (sll1867) encoding the D1 protein polypeptides. This indicates that the decreased availability of Ci leads to increased photodamage, presumably as a result of overreduction of the electron transport chain and diminished electron acceptor availability for the disposal of electrons generated by photosystem II. The apparent transcript abundances of psbA2 and psbA3 were much higher than that of psbA1, but the analysis was unable to distinguish the differential contribution of the three psbA genes to the increased psbA transcript populations because of the high sequence similarity among the three paralogous genes. The two psbD genes, sll0849 and slr0927, were also up-regulated by low Ci stress (1.58-and 1.81-fold at 720 min, respectively). The increased abundances of psbA and psbD transcripts are likely required to sustain the increased rates of synthesis of the D1 and D2 polypeptides, which are prone to rapid turnover in response to photodamage. In this regard, it is interesting that the two ftsH genes, slr1604 and slr0228, which show high homology to ftsH genes in other organisms capable of oxygenic photosynthesis, were also upregulated by Ci downshift (2.39-and 2.49-fold at 200 min). The FtsH proteases associated with the thylakoid membranes are involved in the degradation of photodamaged D1 protein in higher plants (24). One of these, slr0228, was recently shown to be critical for the proteolytic removal of the damaged D1 protein in Synechocystis 6803 (25). Additionally, the carboxylterminal protease gene, ctpA (slr008), the product of which is necessary for the maturation of nascent D1 protein, was also observed to be up-regulated with kinetics similar to the upregulation of psbA transcript abundance (26).
Light-harvesting genes encoding phycobilisome components were down-regulated in parallel with the declines of transcript abundances of photosystem I and II genes. The allophycocyanin genes apcC (ssr3383), apcE (slr0335), and apcF (slr1459) and the phycocyanin genes cpcA (sll1578), cpcB (sll1577), cpcC1 (sll1580), cpcC2 (sll1579), cpcD (ssl3093), and cpcG2 (sll1471) showed a down-regulation of transcripts in response to low Ci stress. These genes encode the structural subunits of phycobilisome light-harvesting antenna. cpcD, cpcC, cpcC2, cpcA, and cpcB form a potential operon (Fig. 5A, C6). The rpaC (sll1926), encoding a protein required for excitation energy state transitions in Synechocystis 6803, showed a 2.05-fold down-regulation at 200 min after CO 2 downshift. On the other hand, changes in transcript abundances were not observed for the major chromophore-bearing, allophycocyanin core proteins apcA (slr2067) and apcB (slr1986), or for the bilin lyases cpcE (slr1878)and cpcF (sll1051). In accordance with the down-regulation of phycobilisome structural genes described above, the two adjacent nblA genes (ssl0452 and ssl0453), which are involved in the degradation of phycobilisomes, were up-regulated by Ci limitation. This has been characterized as a general stress response mechanism and is important for adaptation under other nutrient deprivations, such as iron, nitrogen, and sulfur starvations (27). The products of nblA genes function to specifically activate phycobilisome degradation, thereby reducing photodamage to the photosystem II complex, as well as providing carbon skeletons under Ci limitation. The transcript level of nblB (slr1687) was also increased by Ci limitation in the same fashion. NblB possess lyase activity and catalyzes the removal of chromophores from holophycobiliprotein subunits, thereby facilitating phycobilisome degradation (27). Consistent with the indications for induction of light-stress responses, the genes encoding small 70-amino acid HLIPs, two hliA (ssl2542 and ssr1789), hliB (ssr2595), and hliC (ssl1633), were all up-regulated at 200 min after CO 2 downshift. It has been demonstrated that the expression of these genes is crucial for the survival of Synechocystis under high intensity light or nutrient limitation (28,29); however, the function of these genes remains controversial. The hliB gene is physically clustered and co-regulated with an unknown ORF, slr1544, that also encodes a predicted membrane protein (Fig. 5B, C13). Although this gene exhibits no obvious similarities to the HLIPs, the co-regulation may indicate that it participates in the same function.
In contrast to the down-regulation of genes encoding the polypeptides of the photosystems, light-harvesting phycobilisomes, and the ATP synthase, the abundance of transcripts for Rubisco genes, responsible for CO 2 fixation, increased in response to low Ci stress. The transcripts of rbcL (slr0009), rbcX (slr0011), and rbcS (slr0012) were constantly accumulated after the growth inflection induced by Ci limitation. RbcX acts as chaperonin to facilitate the proper folding of Rubisco (30). These three genes are the members of a cluster of low Ci activated genes in which the first ORF, slr0006, was most strongly induced (15.15-fold at 720 min) (Fig. 5, C and D, C37). The other two members of the up-regulated cluster, slr007 and slr0008 (ctpA), respectively, encode a probable sugar-phosphate nucleotidyltransferase and a carboxyl-terminal processing protease that is essential for the processing of D1 precursor mentioned above. These results contrast with previous analysis, in which it was found that 1-h aeration of CO 2 -free air, which causes the strong up-regulation of key Ci uptake genes, does not up-regulate the transcript levels of rbcL, rbcX, and rbcS (14). The discrepancy may reflect differences in the Ci limitation procedures used. Low (such as air level CO 2 ) but not extremely limited Ci supply seems necessary for the induction of rbc gene expression. The gene slr1347, encoding a carbonic anhydrase localized in the carboxysome, was slightly downregulated at 720 min after Ci downshift (Ϫ1.42 fold), whereas the expression of slr0051, encoding a periplasmic carbonic anhydrase, was slightly up-regulated at the same stage (1.43fold). Among the genes of the ccmK-N operon ccmM and ccmN transcript level declined at 720 min after Ci downshift (Ϫ1.52and Ϫ1.62-fold, respectively). Another ccmK homolog, slr1839, also showed lower transcript level at the same stage (Ϫ1.82-fold). Other ccm genes did not show significant change in RNA abundance.
The NDH-1 complex has multiple subunits, most of which are encoded by single genes in the Synechocystis 6803 genome. However, ndhD and ndhF exist as multiple copies thought to participate in the formation of distinct NDH-1 complexes with different functions as discussed above. The conventional NDH-I complex oxidizes NADPH and NADH, and reduces plastoquinone, thereby enabling cyclic electron transport around photosystem I and respiration. The genes encoding the common subunits of NDH-1 complexes, ndhA (sll0519), ndhB (sll0223), ndhC (slr1279), ndhE (sll0522), ndhG (sll0521), ndhH (slr0261), ndhI (sll0520), ndhJ (slr1281), and ndhK (slr1280), were up-regulated by Ci downshift. The ndhA, ndhI, ndhG, and ndhE genes are physically clustered and form an apparent transcriptional unit (Fig. 5D, C48). The ndhC, ndhK, and ndhJ genes also to form gene cluster on the chromosome (Fig. 5C,  C25). The ndhD2 gene (slr1291), which is associated with the formation of the conventional respiratory NDH-1 complex, was also up-regulated in response to low Ci stress (3.86-fold at 200 min). Collectively, these results are consistent with recent findings that NDH-1 activity is broadly increased under the conditions of Ci limitation (31).
Flavoproteins-The genome of Synechocystis 6803 possesses at least four putative ORFs encoding flavoproteins, i.e. sll0217, sll0219, sll0550, and sll1521. Recent evidence indicates that Sll0550 and Sll1521 mediate the dissipative transfer of electrons from NAD(P)H directly to molecular oxygen (32). A marked induction of sll0550 transcription was observed under Ci limitation, whereas sll1521 only showed slight induction (2.78-and 1.21-fold at 200 min, respectively). An unknown ORF, sll0449, located immediately at the upstream of sll0550, appeared to be co-transcribed with sll0550 (Fig. 5D, C42). In the context of the present experiment, we suggest that these proteins may function to protect the photosynthetic mechanism by regenerating the oxidized form of NADP ϩ , thereby preventing the over-reduction of the electron transport chain and the associated photodamage to photosystem II.
On the other hand, the transcription of sll0217 and sll0219 was even more strongly enhanced by low Ci stress (52.27-and 144.71-fold at 200 min, respectively). An unknown ORF, sll0218, physically clustered with sll0217 and sll0219, was strongly co-transcribed with sll0217 and sll0219 (263.93-fold at 200 min) (Figs. 5A (C3) and 6), indicating that the three genes may form a transcriptional unit. The interruptions of sll0217 and sll0218 apparently have no effect on O 2 photoreduction in previous experiments (32). However, these experiments may not have been performed under conditions allowing expression of these genes. We propose that these inducible genes are supplementary to the nearly constitutively expressed sll0550 and sll1521 genes and the dramatic accumulation of the sll0217 and sll0219 transcripts induced by both high light and Ci starvation afford protection from photodamage under stress conditions (33,34).
Nitrogen Acquisition and Assimilation-The coordination of carbon and nitrogen assimilation is crucial for the optimization of metabolism. Urea, ammonium, nitrate, and nitrite serve as the sources of nitrogen in Synechocystis 6803. Following CO 2 downshift, most of the genes encoding the transporters of these nitrogen compounds were dramatically down-regulated. The nrtA (sll1450), nrtB (sll1451), nrtC (sll1452), nrtD (sll1453), and urtD (sll0764) genes form a cluster (Fig. 5B, C15) that encodes the subunits of the major nitrate transporter. The urtA (slr0447), urtB (slr1200), and urtC (slr1201) genes, among which urtB and urtC are physically clustered, encode the urea transporter. In Synechocystis 6803 amt1 (sll0108) and amt2 (sll1017) encode high and low affinity ammonium/methylammonium permeases, respectively (35). All of the transporter genes for the uptake of urea, ammonium, and nitrate were down-regulated under Ci limitation. The ORF sll0374, a homologue of urtE of Anabaena urtABCDE encoding a urea transporter, is essential for urea transport activity in Synechocystis 6803 based on mutational analysis (36). It was slightly but constantly up-regulated in response to low Ci stress. Why the urtE homolog and urtABCD genes should exhibit a different pattern of transcriptional regulation under Ci limitation remains unclear.
All inorganic nitrogen sources are converted to ammonium in the cell prior to incorporation into carbon skeletons. Nitrate is reduced by the sequential action of nitrate reductase (NaR) and nitrite reductase (NiR). The expression of the genes encoding these two enzymes, nirA (slr0898) and narB (sll1454) in Synechocystis 6803, were highly repressed under Ci limitation. The narB and nrtABCD genes belong to the same transcriptional unit (Fig. 5B, C15), whereas nirA is physically clustered with the gene encoding cyanase (cynS, slr0899) on the chromosome (Fig. 5D, C40) and also exhibited a down-regulation under low Ci limitation. Cyanase catalyzes the decomposition of cyanate (NCO Ϫ ) into CO 2 and NH 3. The drop of cyanase activity was previously shown to be positively correlated with cynS transcript abundance under conditions of nitrogen repletion (37,38).
The accumulated ammonium in cyanobacterial cells is generally combined with glutamate by glutamine synthetase (GS) to form glutamine (Gln), and then glutamate synthase (GOGAT) converts Gln and 2-oxoglutarate to form two molecules of glutamate. Two forms of GS exist in Synechocystis 6803 and are encoded by glnA (slr1756) and glnN (slr0288). The glnN gene is highly transcribed only under nitrogen deficiency, whereas glnA expression is strongly expressed under both nitrogen depletion and repletion. Therefore, it has been proposed that the transcriptional regulation of glnA and glnN utilize distinct regulatory mechanisms (39). However, glnA and glnN expression exhibit similar responses to Ci downshift, with both decreasing after the growth inflection as a result of Ci limitation. Consistent with the transcriptional down-regulation of glnA and glnN, the gltS (sll1499) and gltB (sll1502) genes, encoding ferredoxin-dependent GOGAT and NADH-dependent GOGAT large subunit, respectively, were also down-regulated after CO 2 downshift (Ϫ2.72-and Ϫ2.40-fold at 200 min, respectively).
In accordance with the declines of acquisition and assimilation of carbon and nitrogen as described above, Synechocystis 6803 cells down-regulate the expression of genes for the translation machinery, which is a major consumer of the nitrogen and carbon assimilate. A stringent-type response appears to be gradually exerted by Ci limitation because the transcript abundance of most genes encoding the ribosome proteins exhibited a progressive decline during the course of the experiment with the maximal down-regulation occurring at 720 min after Ci downshift (40). Interestingly, many genes encoding translation machinery are physically clustered on the chromosome of Synechocystis 6803 (Fig. 5, C9, C10, C24, C26, and C43); the largest cluster harbors 26 genes (Fig. 5, A and B, C9). The decreased activity in general of gene expression may occur as indicated by the drop of transcript abundance for genes encoding RNA polymerase subunits, rpoA (sll1818) and rpoB (sll1787), under Ci limitation.
Regulatory Genes-Among various genes encoding regulatory products, rre15 (slr1214) showed an earliest accumulation of transcript after CO 2 downshift. Rre15 is a member of the PatA family of regulators that possess CheY-like response motifs, but do not have discernible DNA binding domains (41)(42)(43)(44). The induction of rre15 expression was triggered before the growth inflection provoked by CO 2 downshift (2.29-fold at 180 min) and steadily increased thereafter during the later stages of Ci limitation and showed a maximal increase of transcript abundance change among various regulator genes (21.75-fold at 720 min). The results suggest that rre15 may have a role in the low Ci acclimation of Synechocystis 6803. The other regulator gene in the patA subfamily, slr1594, was also up-regulated by Ci limitation. The preliminary experiments on slr1594 knockout mutant indicate that the regulator is involved in CO 2 uptake and associated pH homeostasis (data not shown).
Two-component response regulators, rre32 (slr0312), rre33 (sll0797), rre37 (sll1330), and rre38 (slr1584), were down-regulated by low Ci stress. The rre33 and hik30 (sll0798), putative response regulator and histidine kinase, respectively, are physically clustered and are proposed to form a two-component signal transduction system (Fig. 5D, C45), but an argument exists as to whether the two genes are involved in Ni 2ϩ sensing or redox control of the photosynthetic apparatus (45,46). We suggest that rre33 and hik30 may play multiple regulatory roles based on the fact that the expression of the two regulators is also strongly down-regulated by low Ci stress. The sole histidine kinase that was markedly up-regulated is hik1 (slr1393), a phytochrome-like protein (1.79-fold at 200 min). The putative two-component sensors of histidine kinases that showed a significant drop of transcript abundance include  , of ndhR, ndhD5, and sll0218 in the wild-type and ⌬ndhR strains. Cultures were grown in Na 2 CO 3 -free BG-11 media buffered with 20 mM TES-KOH, pH 7.0, and bubbled with 3% CO 2 in air. Cells were harvested and assayed at 0, 30,60,190, and 360 min after switching to air alone. The transcript level of RNase P in each sample serves as a control. hik10 (slr0533), hik20 (sll1590), hik23 (slr1324), and hik29 (slr0311).
Three sigma factors, sigD (sll2012), sigG (slr1545) and sigH (sll0856), exhibit increased transcript abundance starting at the growth inflection period elicited by Ci downshift. However, the respective temporal patterns of induction for each of these sigma factor genes were distinct. The transcription of sigD showed an earliest induction (3.59-fold at 200 min) among these sigma factors. The initial rise of transcript abundance for sigG and sigH was detected at 360 and 720 min, respectively. sigE (sll1689) is involved in the survival of Synechocystis 6803 under nitrogen stress (47), but no marked change of sigE expression was found under Ci starvation applied in the study. From the photosynthesis perspective, SigD is perhaps the most interesting because it has been implicated in the regulation of the psbA gene during light stress, which is consistent with up-regulation of the psbA genes despite the overall trend for decreased expression of photosynthesis genes (48).
In the genome of Synechocystis 6803, there are three homologs of cbbR, a proteobacterial transcription regulator of carbon metabolism: sll0998, sll0030, and sll1594 (12,49). The gene sll0998 seems to be essential for cell viability because the complete segregation of sll0998 knockout has not been achieved (12,49). A slight induction of sll0998 expression under low Ci stress was observed in the study (1.37-fold at 720 min). Sll0030 (CmpR) positively controls the expression of the cmp operon (49); Sll1594 (NdhR) negatively regulates transcription of ndhF3/ndhD3/cupA (12), ndhD5/ndhD6, and sbtA/ sbtB as described below. However, both cmpR and ndhR were up-regulated by Ci limitation (Figs. 5 and 6). It has been suggested that NdhR may not simply act as a negative regulator considering the finding that ndhR transcript rises under low Ci stress (14). The initial induction of cmpR and ndhR were observed at 200 and 180 min, respectively, in accordance with the early accumulation of ndhF3/ndhD3/cupA, ndhD5, and sbtA transcripts in the same time frame as compared with the expression of cmpABCD. The cmpR gene is located immediately upstream and divergently transcribed from the cmp operon (Fig. 5D, C46).
A conserved signal transduction protein in Synechocystis 6803, PII, encoded by glnB (ssl0707), is important for the integration of inorganic carbon and nitrate uptake (50). Although the transcript level of glnB did not show a marked change under CO 2 downshift, many of the nitrogen acquisition and assimilation genes were down-regulated as described above, and are probably under its control as reported (50,51). NtcA acts as a positive transcriptional activator of genes subjected to nitrogen control and also regulates the expression of genes implicated in carbon assimilation (52,53). NtcA protein levels in Synechocystis 6803 increase concomitantly with the accumulation of ntcA transcript, which is reported to be under redox control (54). Like glnB encoding PII protein required for NtcA-regulated gene expression (55), no marked change of the ntcA (sll1423) transcript level was observed under Ci limitation. NtcA activates the expression of glnA and glnB genes, depending on nitrogen availability (52). Meanwhile, NtcA represses transcription of gifA (ssl1911) and gifB (sll1515), which encode GS inhibitors in Synechocystis 6803 (56). In the present study, gifA and gifB showed maximal 66.49-and 5.82-fold inductions in response to Ci limitation, respectively. Taken together, we suggest that the regulatory activity of NtcA is exerted even though the ntcA transcript level remains relatively constant during the course of the Ci limitation experiment. We did observe a slight repression of ntcB (slr0395) transcription (Ϫ1.31-fold). NtcB acts as the transcriptional enhancer of two transcription units, nrtABCD-narB and nirA (57).
We found a number of unknown ORFs, which show dramatic changes in transcript abundance under Ci limitation (Fig. 5). The functions of these ORFs remain to be elucidated. Some of these ORFs were physically clustered on the chromosome of Synechocystis 6803, thereby forming potential transcription units. For instance, sll1307, sll1306, sll1305, sll1304, sll1785, sll1784, sll1783, slr1852, slr1853, slr1854, slr1855, slr1856, slr1857, slr1859, slr1860, and slr1861 are physically clustered to form two apparent transcriptional units, arranged in divergent transcriptional directions (Fig. 5B, C20). The two divergent units may therefore share a common promoter region as both were coordinately down-regulated in response to Ci limitation. Clues to the function of this group of genes come from previous analysis of slr1860 (IcfG) and slr1853, which encode carbon metabolism regulatory protein phosphatase (58) and several ORFs encoding probable carbohydrate metabolizing enzymes, hinting that the two transcriptional units are involved in carbon metabolism. Furthermore, the presence of a putative sigma factor gene and two anti-sigma factor antagonists, with similarities to the well studied Bacillus rsb genes, suggests that the cluster may be involved in integrating carbon metabolism with stress responses (59). Interestingly, cold stress also represses the expression of these genes (60).

Changed Patterns of Gene Expression as a Result of ndhR Inactivation
Based on the preliminary RT-PCR analysis, the expression of the key Ci uptake genes, cupA, ndhF3, and sbtA, was already up-regulated in ⌬ndhR cells grown in Na 2 CO 3 -free BG-11 strongly buffered with 20 mM TES-KOH at pH 7.0 and with the aeration of 3% CO 2 in air. This observation is consistent with the well characterized negative control of ndhF operon exerted by NdhR repressor activity (12). However, it was not anticipated that sbtA expression was also up-regulated by ndhR inactivation. The RT-PCR results imply that the loss of ndhR gene up-regulates not only CUP genes, but also genes responsible for the inducible HCO 3 Ϫ uptake under the given conditions. This unexpected finding led us to directly compare the global expression patterns of ⌬ndhR and wild-type cells grown with sufficient CO 2 supply to identify aberrantly regulated genes associated with the loss of ndhR.
Microarray analysis revealed that the expression of 49 genes exhibited 2-fold or greater changes of transcript abundances between wild-type and ⌬ndhR mutant cells grown in Na 2 CO 3free BG-11 buffered with 20 mM TES-KOH at pH 7.0 and with the aeration of 3% in air (Table II). One of the most important TABLE II Genes strongly affected by ndhR inactivation in Synechocystis 6803 ⌬ndhR and wild-type cells were bubbled with air supplemented with 3% CO 2 (v/v), and grown in Na 2 CO 3 -free BG-11 media buffered with either 20 mM TES-KOH, pH 7.0, or 20 mM TES-KOH, pH 8.0. Data shown are the log 2 ratios of transcript abundances of ⌬ndhR versus wild-type strains. Log 2 ratio and standard deviation calculations are based on pooled replicates of four to six microarray hybridizations of two independent cultures of each strain. Each microarray contained 4 replicate spots for each gene; therefore, each value represents a total 16-24 spots. All genes shown demonstrate 2-fold or greater changes (log 2 ratio Յ Ϫ1 or Ն 1) in transcript abundance for at least one growth condition and have been sorted by chromosomal gene order. Three potential gene operons are indicated with bold type. findings was that the loss of ndhR activated not only the expression of ndhF3/ndhD3/cupA operon, but also the transcription of other physically clustered genes including slr2006, slr2007, slr2008, slr2009, slr2010, ssr3409, ssr3410, slr2011, slr2012, and slr2013, of which slr2007 and slr2009 encode ndhD5 and ndhD6, respectively. DNA microarray analysis also revealed that the sbtA gene and its downstream ORF, sbtB, were also up-regulated as the result of ndhR inactivation under the given growth conditions (Table II).
Neutral pH significantly depresses the growth potential of Synechocystis 6803, which exhibits optimal growth at pH 8 -9. Furthermore, Na 2 CO 3 depletion may also lead to unfavorable regime because Na 2 CO 3 serves as a potential carbon source and substance involved in pH homeostasis via the conversion between CO 3 2Ϫ and HCO 3 Ϫ . To better distinguish effects caused by ndhR inactivation and those associated with growth conditions, the comparison of global expression patterns of ⌬ndhR mutant and wild-type cells were further explored by analysis of cells grown in BG-11 containing the normally added Na 2 CO 3 , buffered at pH 8.0, and with the aeration of 3% CO 2 in air (Table II). The results again indicate that the specific upregulation of the three transcription units, ndhF3/ndhD3/ cupA, slr2006-slr2013, and sbtA/sbtB, stem from the loss of ndhR rather than environmental cues, because expression patterns observed under the favorable growth conditions were quite similar to growth at pH 7 (Table II). The strong up-or down-regulation of several other genes, such as slr0829, slr1592, slr1099, sll0529, sll1605, sll0622, slr0787, and sll0916 in the mutant, was also consistent under the two different growth conditions.
Whether an altered expression a of certain gene directly derives from ndhR inactivation will require further investigation. Nevertheless, sequence analysis of the promoter regions of the genes reveals potential binding sites of LysR regulators. The T(N 11 )A sequence is often present in promoter regions of genes controlled by LysR-type regulators in proteobacteria. Based on the reported consensus sequence of NdhR binding motif, three presumptive NdhR binding sites were identified in the promoter region of the transcriptional unit, slr2006-slr2013, and four presumptive sites in the promoter region of sbtA/sbtB (Fig. 7). These results support the existence of a mechanism involving the direct binding of NdhR in the promoter regions to exert control over the expression of these genes.
Like wild-type cells, ⌬ndhR culture still exhibited a similar growth rate decrease resulting from the CO 2 downshift, despite the fact that the inducible Ci uptake systems involving cupA and sbtA were already activated in the mutant. Clearly, the increased Ci acquisition through the inducible Ci uptake systems do not completely satisfy the demand for [Ci] enabling optimal growth under the experimental conditions used here.
The loss of ndhR appeared not to activate the expression of the cmp genes as revealed by RT-PCR and DNA microarray analyses. Interestingly, the expression of the cmpA was even delayed in the mutant after CO 2 downshift (Fig. 3). A similar expression pattern was true for sll0218, which apparently forms a transcriptional unit with two flavoprotein genes (Fig.  6). The results suggest that ndhR inactivation has a secondary impact on the expression modulation of some other genes such as cmp and sll0218 under Ci limitation. The HCO 3 Ϫ transporters encoded by cmpABCD and sbtA in Synechocystis 6803 are both inducible under Ci limitation, but inactivation of cmp genes in the cyanobacterium has little effect on the HCO 3 Ϫ transport activity. Thus, it has been proposed that the Na ϩ -dependent HCO 3 Ϫ transporter encoded by sbtA plays a central role in HCO 3 Ϫ uptake in Synechocystis 6803 (4). A Na ϩ gradient is necessary to drive SbtA-mediated HCO 3 Ϫ uptake. However, so far no primary Na ϩ pump creating a Na ϩ gradient across membranes has been identified in the Synechocystis 6803. The slr1509 (ntpJ) gene is the sole homolog of gene encoding a putative Na ϩ -ATPase in Synechocystis 6803 genome, but recent evidence shows that slr1509 encodes Ktr-like system, a K ϩ /Na ϩ symporter rather than a primary Na ϩ pump (61). Therefore, the negative effect of slr1509 interruption on HCO 3 Ϫ uptake could indirectly derive from K ϩ /Na ϩ imbalance as a result of slr1509 inactivation (4). The function of ndhd5 and ndhD6 are unclear yet (14), but low Ci stress induces the expression of the two genes and other elements in the slr2006-slr2013 transcriptional unit in wild-type cells, implying that these genes somehow are involved in inducible Ci uptake. Several bacterial species possess ion transporting complexes encoded by a multi-gene operon of which two are the homologues of genes encoding the subunits of proton-translocating NADH dehydrogenase (62,63). It has been proposed that the complex may function as a primary ion extrusion or exchange system energized directly by electron transport through other components of the complex (64,65). slr2011 among slr2006-slr2013 cluster shows similarity to a Na ϩ /H ϩ antiporter. It would seem worthwhile to investigate the possibility that the gene cluster slr2006-slr2013 encodes a similar complex in which the activity of NADH dehydrogenase energizes the formation of a Na ϩgradient to drive HCO 3 Ϫ uptake. The fact that NdhR controls the expression of genes encoding both Na ϩ -translocation proteins (nhaS1 and sbtA) and NDH complexes may further support this hypothesis. Sequences are centered on the proteobacterial LysR-type T(N 11 )A motif (indicated by asterisks). The nucleotide positions relative to the start of slr2006 and sbtA translation are indicated in parentheses. // indicates that the motif is located on the minus strand of the promoter region. Highly conserved nucleotides are printed in bold uppercase, whereas variable ones are shown in lowercase. The previously reported consensus sequence is shown below (12).