Na+-dependent K+ uptake Ktr system from the cyanobacterium Synechocystis sp. PCC 6803 and its role in the early phases of cell adaptation to hyperosmotic shock.

Transmembrane ion transport processes play a key role in the adaptation of cells to hyperosmotic conditions. Previous work has shown that the disruption of a ktrB/ntpJ-like putative Na(+)/K(+) transporter gene in the cyanobacterium Synechocystis sp. PCC 6803 confers increased Na(+) sensitivity, and inhibits HCO(3)(-) uptake. Here, we report on the mechanistic basis of this effect. Heterologous expression experiments in Escherichia coli show that three Synechocystis genes are required for K(+) transport activity. They encode an NAD(+)-binding peripheral membrane protein (ktrA; sll0493), an integral membrane protein, belonging to a superfamily of K(+) transporters (ktrB; formerly ntpJ; slr1509), and a novel type of ktr gene product, not previously found in Ktr systems (ktrE; slr1508). In E. coli, Synechocystis KtrABE-mediated K(+) uptake occurred with a moderately high affinity (K(m) of about 60 microm), and depended on both Na(+) and a high membrane potential, but not on ATP. KtrABE neither mediated Na(+) uptake nor Na(+) efflux. In Synechocystis sp. PCC 6803, KtrB-mediated K(+) uptake required Na(+) and was inhibited by protonophore. A Delta ktrB strain was sensitive to long term hyperosmotic stress elicited by either NaCl or sorbitol. Hyperosmotic shock led initially to loss of net K(+) from the cells. The Delta ktrB cells shocked with sorbitol failed to reaccumulate K(+) up to its original level. These data indicate that in strain PCC 6803 K(+) uptake via KtrABE plays a crucial role in the early phase of cell turgor regulation after hyperosmotic shock.

Hyperosmolality caused by high salinity or drought constitutes a major challenge to the growth of prokaryotes, fungi, and plants (1)(2)(3)(4)(5). Many bacteria, including Escherichia coli, react to such a change to hyperosmotic conditions with a biphasic response. During the first phase, E. coli cells accumulate additional K ϩ from the medium through their K ϩ uptake systems Trk and Kdp and synthesize glutamate concomitantly. Thereby, they increase the ion content of their cytoplasm and counteract plasmolysis brought about by water efflux because of the reversal of the sign of cell turgor pressure caused by external hyperosmolarity. During its second phase of the cellular adaptation, E. coli replaces this internal potassium glutamate by synthesizing high amounts of the non-ionic compatible solute trehalose (6 -8). Bacillus subtilis follows a slightly different strategy in that after hyperosmotic shock, growing cells of this organism synthesize the compatible zwitterionic solute proline instead of trehalose (9,10), and Vibrio alginolyticus KtrAB makes major contributions to net K ϩ uptake (11). Other bacteria follow similar approaches (3,5,(12)(13)(14).
Cyanobacteria can adapt to high NaCl concentrations. Under these conditions a variety of these species synthesize high concentrations of glucosylglycerol and sucrose, which are accumulated as compatible solutes in their cytoplasm (15). During the early phase of this adaptation Synechocystis sp. PCC 6714 accumulates Na ϩ . Within a few minutes, Na ϩ is replaced by K ϩ . Subsequently, it takes many hours to days to replace the high K ϩ concentration in the cytoplasm by glucosylglycerol as well as minor amounts of sucrose (16). The K ϩ uptake system(s) involved in the early phases of this osmoadaptation have, however, not been identified in this early work. Since the elucidation of its genome nucleotide sequence, cyanobacterial research has focused on Synechocystis sp. PCC 6803 (//www. kazusa.or.jp/cyano/). During its long term adaptation to high NaCl concentrations, this organism also synthesizes and accumulates glucosylgylcerol (17)(18)(19). Determining from the genome sequence, strain PCC 6803 has been predicted to contain at least three types of K ϩ uptake systems, Kdp system, which is probably of minor importance (20), a Ktr system (21), which appears to play a role in salt stress by high NaCl concentrations (20) and K ϩ channels. However, the Ktr system has not been characterized any further. Recently, it has been reported that in Synechocystis sp. PCC 6803 the ntpJ gene (slr1509) is essential for both the adaptation to high NaCl concentrations and the bicarbonate transport via the SbtA system (22). SbtAmediated bicarbonate transport was also dependent on the presence of external Na ϩ (22). Since ntpJ has been described originally as the Na ϩ transport gene J, located downstream of a cluster of ntp genes encoding the Na ϩ -translocating V-type ATPase from Enterococcus hirae (23), Shibata et al. (22) interpreted their findings to mean that NtpJ functions as a Na ϩ efflux system required for the removal of Na ϩ from the cells after a hyperosmotic shock with NaCl and/or Na ϩ uptake due to Na ϩ /HCO 3 Ϫ symport via SbtA (22). However, ntpJ from * This work was supported by a grant-in-aid for Center of Excellence (COE) Research (to N. U.), the 21st Century COE Program (to N. U.), Grants-in-aid for Scientific Research (15380070, 15902677, and 16013219; to N. U.) from the Ministry of Education, Science, Sports and Culture of Japan, and grants from the Japan Society for the Promotion of Science Research of the Future Program (to N. U.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
** To whom correspondence should be addressed: Bioscience and Biotechnology Center, Nagoya University, Nagoya 464-8601, Japan. E. hirae has later been identified as a K ϩ -translocating subunit of the bacterial Ktr-type K ϩ uptake system (11, 21, 24 -27). Ktr from Vibrio alginolyticus is Na ϩ -dependent, and the same may apply to the Ktr system from E. hirae (26). KtrB belongs to a superfamily of K ϩ transporter proteins, with members in Archaea (KdpA, TrkH), bacteria (KdpA, TrkH, KtrB), fungi (TRK), and plants (HKT) (26 -38). In particular, the comparison between bacterial KtrB and HKT systems from some plants is important, because Ktr and HKT systems are the only members of the superfamily that mediate Na ϩ -dependent K ϩ transport.
Recently, Berry et al. (20) indicates that ntpJ/ktrB from Synechocystis sp. PCC 6803 has a function in K ϩ uptake. In this report, we confirm and extend their observation with a detailed study on the Ktr system from Synechocystis sp. PCC 6803. We show that NtpJ/KtrB is part of a more complex Ktr system, comprising of three kinds of subunits, KtrA, KtrB, and KtrE. In E. coli K ϩ transport via the Synechocystis KtrABE system is of moderately high affinity, is dependent on the presence of Na ϩ and on high membrane potential, but not on ATP. In this bacterium it neither transports Na ϩ inwards nor outwards. The measurement of the K ϩ content of the Synechocystis sp. PCC 6803 and the cell containing the mutation at ktrB gene revealed that the Ktr system was essential for the early phase of the adaptation of Synechocystis sp. PCC 6803 to high osmolality.
Reintegration of the Synechocystis ktrB Gene into the Strain PCC 6803 Genome-The strain Synechocystis ⌬ktrB, in which the slr1509 (ktrB) gene has been disrupted by insertion of a kanamycin resistance gene (22,42), was used to test for rescue of KtrB activity by integration of the gene into its genome. The ktrB gene and spectinomycin resistance gene were tandemly placed under the control of the iron transporter promoter (43) and inserted into the middle of the kanamycin resistance gene of strain ⌬ktrB by homologous recombination using the method of Ohkawa et al. (42). Cells resistant to spectinomycin (20 g/ml) were isolated, and the integration of the gene was confirmed by PCR with genomic DNA using the following primers: for ktrB, 5Ј-CCATTCTAC-GAACCTAAGT-3Ј (primer a) and 5Ј-CCTTGATATCGTTGTTATT-3Ј (primer b); for the kanamycin resistance gene, 5Ј-ACAGCGGCCGCTT-GAACTTTTGCTTTGCC-3Ј (primer c) and 5Ј-ACAGGTACCCACGGTT-GATGAGAGCTT-3Ј (primer d). This strain was named ϩktrB.
Measurement of K ϩ Uptake in E. coli and Synechocystis sp. PCC 6803-K ϩ influx was essentially measured as described by Tholema et al. (27). E. coli LB2003 cells were cultured in a synthetic medium in the presence of 30 mM KCl, 0.25 mM IPTG, 25 g/ml of ampicillin, and/or 15 g/ml of chloramphenicol at 30°C. E. coli LF12 were cultured in 1% tryptone, 0.5% yeast extract, 0.67% KCl, 1% glucose, 0.25 mM IPTG, and the required antibiotics at 30°C. Synechocystis sp. PCC 6803 was cultured in BG11 medium. The cells were collected by centrifugation, resuspended in 120 mM Tris-HCl (pH 8.0) and 1 mM EDTA, shaken for 10 min at 37°C, collected by centrifugation, and washed twice with 200 mM HEPES-NaOH (pH 7.5) (94 mM Na ϩ and 29 M K ϩ , as determined by flame photometry) or 200 mM HEPES-triethanolamine (pH 7.5) (1.9 M Na ϩ and 0.36 M K ϩ ), then resuspended in the same buffer. After shaking for 20 min at room temperature, the concentrations of cells were adjusted to an OD 578 of 3 with the same buffer. 10 mM glucose was added to the suspension 10 min prior to the start of the K ϩ uptake measurement. The net uptake of K ϩ was measured by the silicone filtration technique (21,27,45), the K ϩ contents of the cell pellets being determined by flame photometry (45).
Measurement of Na ϩ Uptake in E. coli-E. coli TO114 was grown in 1% tryptone, 0.5% yeast extract, 0.67% KCl, and the required antibiotics. Na ϩ uptake was measured as described above for K ϩ uptake. Na ϩ concentrations in the cells were determined by flame photometry.
Measurement of Na Efflux in E. coli-Na ϩ efflux was measured as described by Sakuma et al. (46). E. coli TO114 cells were grown in 1% tryptone, 0.5% yeast-extract, 0.67% KCl, and the required antibiotics to an OD 610 of 0.5, collected by centrifugation, resuspended in the same culture medium containing 150 mM NaCl, then kept on ice for 2 h. Na ϩ concentrations in the cells were measured as described above for Na ϩ content.

RESULTS
Three Synechocystis ktr Gene Products Are Required for K ϩ Transport Activity in E. coli-Since previous reports have suggested that Synechocystis KtrB plays a role in K ϩ uptake (20,21,30), rather than in Na ϩ transport (22, 47) (www. kazusa.or.jp/cyano/), we investigated the role of Synechocystis ktr genes in K ϩ transport and analyzed the Ktr gene family. We cloned the Synechocystis Ktr genes into the E. coli mutant strain LB2003. This strain carries mutations in genes encoding the three major E. coli K ϩ uptake systems, Kdp, Trk, and Kup. As a consequence, it possesses negligible K ϩ uptake activity at K ϩ concentrations in the low millimolar range and, therefore, does not grow at K ϩ concentrations Յ 10 mM (13). In agreement with previous results in that only together the cloned ktrA and ktrB genes from V. alginolyticus enable E. coli LB2003 to transport K ϩ (21), Synechocystis ktrB alone did not restore K ϩ transport in the strain (Fig. 1). The Synechocystis genome contains a ktrA orthologue (sll0493) at a site quite distant from that of the ktrB (slr1509) gene. However, coexpression of ktrA and ktrB in E. coli LB2003 also did not lead to growth at low K ϩ concentrations (Fig. 1). The ktrB gene (slr1509) forms a cistron with gene slr1508 (47). Coexpression of slr1508 with ktrB and ktrA in E. coli LB2003 resulted in growth at low K ϩ concentrations ( Fig. 1). To exclude the possibility of a polar effect on the expression of ktrB, the initiation codon of slr1508 was removed from the plasmid containing the slr1508-slr1509 cistron, giving plasmid pPAB404-KtrB⌬E. Together with the ktrA-containing plasmid, pSTV28-KtrA, it failed to restore growth of the LB2003 strain at low K ϩ concentrations (Fig. 1). These results showed that three Synechocystis ktr genes, ktrA, ktrB, and slr1508 are essential for K ϩ transport activity in E. coli. Gene product KtrA is a peripheral membrane protein thought to regulate K ϩ transport via KtrB by NAD ϩ /NADH binding (21,48). KtrB is the K ϩ -translocating subunit discussed above, and slr1508 is a novel type of ktr gene, which we have named ktrE rather than ktrC or ktrD, because in B. subtilis orthologues of its ktrA and ktrB genes have been named ktrC and ktrD (11). The subunit of ktrE has not been found in any other Ktr/Trk/HKT/Kdp transporter. Amino acid sequence alignment studies showed sequence similarity between KtrE and glycosyl transferases (Fig. 1D). KtrE is 25 and 23% identical to WbbP of Shigella dysenteriae (49) and WbgM of E. coli (50), respectively.
Na ϩ Activates KtrABE-mediated K ϩ Uptake in E. coli-To further characterize the Ktr system, we examined the ion transport activity of plasmid-containing, K ϩ -depleted E. coli LB2003 cells suspended in 200 mM HEPES-NaOH buffer (94 mM Na ϩ and 29 M K ϩ , as measured by flame photometry) in the presence of glucose. Addition of 1 mM KCl led to rapid and extensive net K ϩ uptake only when all three Synechocystis ktrABE were coexpressed in the cells ( Fig. 2A), confirming the results of the growth assays ( Fig. 1, B and C).
The K ϩ transport activity of both wheat HKT1 and V. alginolyticus KtrAB is enhanced by Na ϩ (27,29). To test whether this was also the case for Synechocystis KtrABE, the LB2003 cells were suspended in low Na ϩ buffer (200 mM HEPEStriethanolamine buffer; 1.9 M Na ϩ and 0.36 M K ϩ , as measured by flame photometry). Addition of 1 mM KCl to the cells with the empty plasmid led to a small amount of K ϩ uptake, which was not enhanced by the subsequent addition of 1 mM NaCl (Fig. 2B). Cells expressing ktrABE took up a similar amount of K ϩ after the addition of the KCl. However, the subsequent addition of NaCl to these cells led to continuous and extensive K ϩ uptake, indicating that in E. coli K ϩ uptake via Synechocystis KtrABC depends on the presence of Na ϩ . In addition, the reversal sequence of the addition of 1 mM Na ϩ and 1 mM K ϩ showed K ϩ uptake transport in the cells by addition of 1 mM KCl (Fig. 2C). The stimulatory effect was Na ϩ -specific, since addition of 1 mM LiCl, RbCl, CsCl, or CaCl 2 failed to stimulate K ϩ uptake (Fig. 2D). Because in Fig. 5, we show that KtrABE in Synechocystis sp. PCC 6803 is involved in adaptation to higher osmolalities of both NaCl and sorbitol, we tested the effect of osmolality on the transport process. When either 200 mM sorbitol or 100 mM NaCl (of equal osmolality) was added to the low Na ϩ buffer (200 mM HEPES-triethanolamine buffer) before the addition of 1 mM KCl, K ϩ was taken up by cells preincubated with NaCl, but not by those preincubated with sorbitol (Fig. 2, E and F), showing that the stimulation of KtrABE transport in E. coli was because of the ionic effect of Na ϩ and not to its osmotic effect.
Further Characterization of KtrABE in E. coli-Na ϩ -dependent K ϩ uptake had a pH optimum at pH 7.5 and did not occur at pH 5.5 (Fig. 2G), arguing against the active participation of FIG. 2. Characterization of K ؉ uptake by the Synechocysistis KtrABE system in E. coli. A, K ϩ uptake requires all three Ktr gene products. Energized K ϩ -depleted cells of strain LB2003 expressing different combinations of Ktr proteins were tested for their ability to take up K ϩ . At t ϭ 0 min 1 mM KCl was added to the medium containing 94 mM Na ϩ (by flame photometry). Symbols: KtrABE (filled circles), KtrAB (triangles), KtrBE (squares), or empty vector (empty circles). B, KtrABE system requires Na ϩ for activity. Cells of E. coli LB2003 containing ktrABE (filled circles) or empty vectors (empty circles) were suspended in 200 mM HEPES-triethanolamine buffer (pH 7.5) (containing 1.9 M Na ϩ ). KCl and NaCl were added at t ϭ 0 min and t ϭ 7 min, respectively. C, KtrABE-mediated K ϩ uptake in the presence of 1 mM NaCl. Cells of E. coli LB2003, containing ktrABE (filled circles) or empty vectors (empty circles) were suspended in the same buffer. NaCl and KCl were added at t ϭ 0 min and t ϭ 10 min, respectively. D, only Na ϩ stimulates K ϩ uptake via KtrABE. NaCl, LiCl, RbCl, CsCl, or CaCl 2 was added at 1 mM at t ϭ 0 min into 200 mM HEPES-triethanolamine buffer supplemented with 1 mM KCl. The initial rate of K ϩ uptake during the first minute after KCl addition is shown as a percentage of that by KtrABE-expressing E. coli in the presence of NaCl. E, K ϩ uptake in the presence of 100 mM NaCl added 10 min prior to addition of 1 mM KCl at t ϭ 0 min; other conditions and symbols as in B. F, absence of K ϩ uptake in the presence of 200 mM sorbitol added to 200 mM HEPES triethanolamine buffer (pH 7.5) (1.9 M Na ϩ ) 10 min prior to addition of 1 mM KCl at t ϭ 0 min; other conditions and symbols as in B. G, pH dependence of KtrABEmediated K ϩ transport. K ϩ influx was measured at pH 5.5-8.5 in the presence of NaCl. 1 mM KCl was added at t ϭ 0 min into 200 mM MES-NaOH (pH 5.5 and 6.5), 200 mM HEPES-NaOH (pH 7.5) or 200 mM Tricine-NaOH (pH 8.5). The initial velocity of K ϩ uptake during the first minute after KCl addition is shown as a percentage of that measured in pH 7.5. H, kinetics of K ϩ uptake by KtrABE. Data are plotted according to the method of Lineweaver Burk. The assay was performed in the buffer containing 94 mM Na ϩ . I, kinetics of K ϩ uptake with respect to its stimulation by Na ϩ . The assay was performed in 200 mM HEPES triethanolamine buffer (pH 7.5) (containing 1.9 M Na ϩ ) containing 1 mM KCl. The concentrations of added NaCl was varied between 0.2 and 50 mM.
protons (H ϩ ) in this process. In the presence of Na ϩ , the K m of KtrABE for K ϩ was ϳ60 M and its V max 110 nmol⅐min Ϫ1 mg Ϫ1 of cell protein (Fig. 2H). This K m -value for K ϩ is similar to those of 50 M K ϩ for V. aliginolyticus KtrAB in E. coli (21). These data indicate that in E. coli Synechocystis KtrABE functions as a rapid K ϩ uptake system with a moderately high affinity for K ϩ . The affinity of KtrABE for Na ϩ was also measured in an experiment where Na ϩ (0.2-50 mM) was added in the presence of 1 mM K ϩ . The K m value for Na ϩ in the stimulation of K ϩ uptake was 2.1 mM (Fig. 2I).
Mode of Energy Coupling to K ϩ Transport in E. coli-Next, we examined the mode of energy coupling to KtrABE-mediated K ϩ uptake in E. coli. Omission of the energy source, glucose, from the standard assay medium led to severe inhibition of net K ϩ uptake (Fig. 3A), suggesting that ATP and/or the transmembrane proton motive force (pmf) or one of its components, the membrane potential (⌬) or the transmembrane pH difference (⌬pH), is involved in KtrABE transport activity. The protonophore, carbonylcyanide m-chlorophenylhydrazone (CCCP), which abolishes the pmf, inhibited K ϩ uptake (Fig. 3B), suggesting that KtrABE activity depends on the pmf. However, in F 1 F o ATPase-containing E. coli, such as strain LB2003, CCCP is known to decrease cellular ATP levels because of enhanced ATP hydrolysis via this enzyme (e.g. Ref. 51). To clarify this point, we used E. coli strain H500 (⌬hemA), which is deficient in heme synthesis and so lacks a functional respiratory electron transport chain (Futatsugi doctoral thesis, 2001). 1 When grown on glucose in the absence of ␦-aminolevulinate, this strain did not show the cytochrome Soret (␥) absorption bands in the 410 -440 nm region, but these appeared when ␦-aminolevulinate was added. In the absence of ␦-aminolevulinate, neither ⌬ nor ⌬pH could be detected in the mutant and it grew at a slightly slower rate than the wild type on glucose. These results suggest that in these cells, the pmf is absent, but glycolysis provides enough ATP for cell growth. In E. coli, Kdp is the only K ϩ uptake system active in the absence of a pmf (52,53). We therefore constructed the mutant strain LF12, which is both ⌬hemA and ⌬kdpDE. As shown in Fig. 3C, LF12 cells containing empty vector grown in the presence of ␦-aminolevulinate showed some K ϩ uptake activity, presumably due to the other two K ϩ uptake systems, Kup and Trk, still present in this strain, but uptake was enhanced at least 5-fold when the KtrABE were also present. No detectable K ϩ uptake was observed in LF12 containing KtrABE grown in the absence of ␦-aminolevulinate. These results show that KtrABE requires the pmf for activity and does not use ATP as the sole source of energy. At pH 7.5, ⌬ is the major component of the pmf, because at this pH value, the difference in pH between the cytosol and the external space is very small. Thus, ⌬ appears to couple energy to K ϩ transport via the KtrABE system. KtrABE Does Not Have Na ϩ Transport Activity-Since it has been suggested that NtpJ/KtrB functions as a Na ϩ -efflux system (22), we tested whether we could detect Na ϩ transport activity of Synechocystis KtrABE in E. coli. For this purpose we used the E. coli TO114 as a host, which possesses negligibly low Na ϩ extrusion activity (40). As shown in Fig. 4A, E. coli TO114 cells transformed with a plasmid encoding the E. coli Na ϩ /H ϩ antiporter, NhaA, extruded Na ϩ , whereas, under the same conditions, cells expressing KtrABE did not. This is confirmed by the growth test of the cells (Fig. 4B). When 100 mM NaCl was incorporated into the solid medium, cells expressing KtrABE did not grow, whereas cells expressing NhaA grew well (Fig. 4B). This failure to rescue the deficiency in Na ϩ extrusion activity in E. coli supported the notion that KtrABE does not alleviate Na ϩ toxicity in E. coli. We then tested whether KtrABE mediates Na ϩ uptake. KtrABE failed to stimulate Na ϩ uptake by E. coli TO114 (Fig. 4C). It should be noted that this experiment was done in the same HEPES-triethanolamine buffer in which KtrABE exhibited Na ϩ -dependent K ϩ uptake (Fig. 2B), showing that this buffer does not inactivate KtrABE. Taken together, the data presented in Fig. 4 show that, in E. coli, Synechocystis KtrABE neither mediated Na ϩ efflux nor Na ϩ uptake.
Hypersensitivity to High Osmolality of Synechocystis sp. PCC 6803 Conferred by ktrB Mutation-The data from Figs. 1-4 indicate that the Ktr system is a Na ϩ dependent and rapid K ϩ uptake system and is unlikely to contribute to removal of Na ϩ toxicity under conditions of hyperosmotic stress. To understand the precise physiological role of the Ktr system in Synechocystis sp. PCC 6803, we performed cell growth and K ϩ transport experiments with the wild type, the ⌬ktrB strain, generated by insertion of a kanamycin resistance cassette into ktrB (22), and a newly constructed ϩktrB strain, which caries a copy of ktrB under the control of an iron-inducible promoter, in the middle of the kanamycin resistance gene of the ⌬ktrB strain (Fig. 5, A  and B). In particular, because the study using the ⌬ktrB strain (20, 22) did not check that only the ktrB gene was affected, we constructed the ϩktrB strain. On plates (Fig. 5C) and in liquid medium (Fig. 5D) growth of the ⌬ktrB strain was completely inhibited by the addition of 300 mM sorbitol or 300 mM NaCl to the BG11 medium (20.44 mM Na ϩ and 14.77 mM K ϩ , as determined by flame photometry). Both wild-type and the ϩktrB strain continued to grow under these conditions (Fig. 5), showing that it is only the lack of the Ktr system that gave rise to the osmosensitive phenotype. Since the ⌬ktrB strain was also sensitive to high concentrations of non-ionic compound, sorbitol, we conclude that KtrB plays a role in the response to hyperosmotic stresses in general.
KtrB-mediated K ϩ Uptake by Synechocystis sp. PCC 6803 Has Similar Properties to KtrABE in E. coli-Before exploring the role of the Ktr system in osmoadaptation of Synechocystis sp. PCC 6803 further, we examined whether this system behaved similarly in its native organism as in the E. coli hosts. For this purpose we measured net K ϩ uptake by K ϩ -depleted energized Synechocystis cells. At low external Na ϩ concentrations (1.9 M), addition of 5 mM KCl hardly led to K ϩ uptake; however, the subsequent addition of 5 mM NaCl caused a rapid and substantial, KtrB-mediated net K ϩ uptake (Fig. 6A). This effect was due to the addition of 5 mM NaCl but not because of the small increase in osmolality (from 312 to 334 mOsm), since it was not seen using an equivalent osmolality of sorbitol (10 mM) in the absence of Na ϩ (Fig. 6B). Moreover, addition of 200 mM sorbitol, which increased the osmolality from 312 to 549 mOsm, did not lead to net K ϩ uptake in the absence of Na ϩ (Fig. 6C). This indicates that as in E. coli (Fig. 2B), KtrABE requires Na ϩ for K ϩ uptake activity in Synechocystis sp. PCC 6803.
Second, we examined the energy dependence of KtrB-mediated K ϩ transport in Synechocystis sp. PCC 6803 (Fig. 7). As shown in E. coli (Fig. 3B), this process was prevented by the presence of the protonophore CCCP (Fig. 7B), suggesting a role of the pmf in the coupling of energy to the K ϩ transport via the Ktr system. Taken together these data show that the properties of the Ktr system in Synechocystis sp. PCC 6803 are similar to those in E. coli, in which a more extensive characterization of the system was possible. FIG. 4. KtrABE neither mediates Na ؉ efflux nor Na ؉ influx in E. coli. A, Na ϩ efflux. E. coli TO114, which has mutations in the genes encoding three Na ϩ /H ϩ antiporters, NhaA, NhaB, and ChaA, was used as the host strain to measure efflux of Na ϩ mediated by KtrABE. The bacteria containing plasmids encoding KtrABE (filled circles), vector plasmids (empty circles) or a plasmid encoding E. coli Na ϩ /H ϩ antiporter, NhaA, (triangles) were loaded with Na ϩ as described under "Experimental Procedures." At t ϭ 0, cells were diluted into a Na ϩ -free buffer, and the Na ϩ content of the cells was measured as a function of time. B, E. coli TO114 containing KtrABE failed to grow at high Na ϩ concentrations (100 mM NaCl). Cells containing NhaA grew well under the conditions (lower). The control experiment without addition of NaCl (upper). C, KtrABE fails to stimulate Na ϩ uptake by the E. coli. E. coli TO114 was used as the host strain to measure influx of Na ϩ mediated by KtrABE (filled circles). Transformants containing empty vector are indicated as empty circles. The assay was performed with energized cells depleted from Na ϩ and K ϩ in the 200 mM HEPES triethanolamine buffer, pH 7.5. 1 mM NaCl was added at t ϭ 0.
The Role of the Ktr System in Osmoadaptation of Synechocystis sp. PCC 6803-We then examined the role of a functional Ktr system on the K ϩ content of cells after hyperosmotic shock.
The effect of the addition of 100 mM NaCl or 200 mM sorbitol to HEPES-NaOH buffer (pH 7.5, 94 mM NaCl), on KtrB-mediated transport in Synechocystis sp. PCC 6803 was tested (Fig. 8). The ⌬ktrB strain did not accumulate much K ϩ under these conditions, and some of their K ϩ was lost rapidly after a hyperosmotic shock with either 100 mM NaCl (Fig. 8A) or 200 mM sorbitol (Fig. 8B). Wild-type and ϩktrB cells behaved similarly in that after an initial net loss of K ϩ after the hyperosmotic shock, they either accumulated an extensive amount of additional K ϩ by hyperosmotic shock with NaCl (Fig. 8A) or returned their K ϩ to a level slightly higher than that before the osmotic shock (Fig. 8B).
To further evaluate the role of the Ktr system in the osmoadaptation of Synechocystis sp. PCC 6803, the cells were hyperosmotically treated for 10 min with addition of 200 mM sorbitol before K ϩ addition (Fig. 9). Both the wild type and the ⌬ktrB strain lost a considerable amount of K ϩ after the shock. After the addition of K ϩ the wild type reaccumulated K ϩ to a level clearly exceeding that before the osmotic upshock (Fig.   FIG. 5. Hyperosmotic stress inhibits growth indicated by a, b, c, and d and arrows, respectively. B, PCR products from the DNA of WT, ⌬ktrB strain, and ϩktrB strain. The primer combinations (a ϩ b or c ϩ d) for PCR are indicated. C, growth of the WT, ⌬ktrB strain, and ϩktrB strain on solid medium containing the indicated concentrations of NaCl or sorbitol. D, growth of the three strains in liquid culture; the OD 730 was measured after 48 and 96 h of culture.
FIG. 6. KtrB-mediated K ؉ uptake by Synechocystis sp. PCC 6803 depends on Na ؉ . Energized cells depleted of Na ϩ and K ϩ were incubated in the 200 mM HEPES triethanolamine buffer (pH 7.5, containing 1.9 M Na ϩ and 0.36 M K ϩ ). At t ϭ 0, 5 mM KCl was added, followed at t ϭ 10 min by 5 mM NaCl (A), 10 mM sorbitol (B), or 200 mM sorbitol (C). The osmolality of the medium under the different conditions is shown in parenthesis. Symbols: circles, wild type; triangles, ⌬ktrB strain; squares, ϩktrB strain. 9A). By contrast the ⌬ktrB cells did not reaccumulate any K ϩ under these conditions (Fig. 9B). Fig. 9C shows initial K ϩ uptake rates after the addition of 5 mM KCl to wild type, ⌬ktrB cells, and ϩktrB cells. After a hypertonic shock, the Ktr system enabled the cells to take up K ϩ much faster. From the results of Figs. 8 and 9, we conclude that the role of the Ktr system in osmoadaptation of Synechocystis sp. PCC 6803 lies in its protection of the cells against K ϩ loss immediately after the hyperosmotic shock and in the subsequent readjustment of the cell turgor pressure by additional accumulation of K ϩ under these conditions. The Ktr system can fulfill this role as long as enough Na ϩ for the activation of the Ktr system and enough K ϩ for uptake is present in the medium.

DISCUSSION
Previous work on the function of slr1509 (ktrB/ntpJ) in Synechocystis sp. PCC 6803, has given conflicting results. On the one hand, inactivation of this gene results in both a high NaCl-sensitive phenotype of the organism and inhibition of its Na ϩ /HCO 3 Ϫ cotransport activity via SbtA. These observations have been explained by assuming that slr1509 extrudes Na ϩ from the cells (22). On the other hand, amino acid alignment studies have shown that slr1509 encodes a protein belonging to the KtrB family of K ϩ translocator proteins superfamily (20,21), and preliminary studies have shown that KtrB plays a role in K ϩ uptake by Synechocystis sp. PCC 6803 (20). One of the aims of our study was to find out the function and physiological role of KtrB. Our experiments expressing the Synechocystis ktr genes in E. coli demonstrated that KtrABE is a K ϩ uptake system dependent on Na ϩ but is neither involved in Na ϩ uptake nor Na ϩ extrusion (Figs. 2, B and C and 4). The KtrBmediated K ϩ transport with respect to Na ϩ dependence and protonophore-sensitivity was also analyzed in Synechocystis sp. PCC 6803 (Figs. 2B, 3, 6A, and 7). Hence we consider it unlikely that in Synechocystis sp. PCC 6803 the KtrB-mediated system extrudes Na ϩ as predicted by Shibata et al. (22), and we propose to modify their hypothesis as follows: In Synechocystis sp. PCC 6714, the earliest phase of adaptation to a hyperosmotic shock with NaCl consists of extensive Na ϩ uptake into the cells (16). Cells lacking an active Ktr system would not be able to carry out the subsequent phase of adaptation consisting of the exchange of the high cytoplasmic Na ϩ concentration against K ϩ (16). If the same applies to strain PCC 6803, the ⌬ktrB strain would continue to contain high Na ϩ concentrations in its cytoplasm. The Na ϩ /H ϩ antiporters, for which the Synechocystis sp. PCC 6803 genome contains several genes (47,54) would be unable to remove the Na ϩ from the cytoplasm, because Na ϩ /H ϩ antiport activity leads to overacidification of the cytoplasm. Only when this Na ϩ /H ϩ -antiport process is coupled to a combination of electrogenic H ϩ extrusion by the H ϩ pumps of the cells (thereby counteracting overacidification of the cytoplasm, but also leading to the formation of a large ⌬ and thereby to inhibition of further H ϩ extrusion) and electrogenic K ϩ uptake (via the KtrB-dependent system, which will decrease ⌬, thereby allowing the cells to extrude more protons), will the cells be able to remove their Na ϩ effectively from the cytoplasm (55,56) (Fig. 10). Failure to extrude Na ϩ in the ⌬ktrB cells gives the cells low Na ϩ motive force across the cytoplasmic membrane. Since the force is believed to be the driving force for HCO 3 Ϫ uptake via SbtA, low Na ϩ motive force in the ⌬ktrB cells explains the inhibition of HCO 3 Ϫ transport in the ⌬ktrB cells observed by Shibata et al. (22). If Na ϩ binds to the Ktr system, and this stimulates the K ϩ uptake, which is proposed in plant Ktr/HKT-type transporters (57), there is an alternative explanation on the requirement of Na ϩ and the Ktr system for SbtA-mediated HCO 3 Ϫ uptake. It is possible that SbtA-mediated HCO 3 Ϫ uptake is coupled to K ϩ uptake by KtrABE and that the dependence of HCO 3 Ϫ uptake on the presence of Na ϩ is a result of the Na ϩ requirement for the function of KtrABE (Figs. 6, 8, and 9). CO 2 uptake is driven by the pmf generated from photosynthetic PSI electron trans-port in Synechococcus sp. PCC 7942 (58). Assuming that SbtA and KtrABE would form the coupling system for HCO 3 Ϫ and K ϩ , the pmf may serve as an energy source for Na ϩ -dependent K ϩ uptake by KtrABE in Synechocystis sp. PCC 6803.
The results in this study indicate that the primary role of the Ktr system in osmoadaptation of Synechocystis sp. PCC 6803 to high osmolality occurs during the early phase of this process. We showed for the first time that this system is not only essential for the adaptation of the cells to osmotic stress caused by ionic compounds like NaCl, but also to that by the non-ionic compound sorbitol (Fig. 5). Both the results reported by Shibata et al. and this study show in growth experiments on plates that the Ktr system is also required for long term osmoadaptation, since the ⌬ktrB strain do not grow at high concentrations of either NaCl or sorbitol (Fig. 5). The loss of K ϩ by the addition of 100 mM NaCl (Fig. 8A) or 200 mM sorbitol (Figs. 8B and 9) was the unexpected because we thought that only water loss occurred by osmotic upshock without K ϩ loss. The loss of K ϩ induced by hyperosmotic stress may have an adverse effect on cell growth, and may lead to the cessation of cell growth as shown in Fig. 5, C and D. The basis of the K ϩ loss in the ⌬ktrB cells may be similar to that observed in ⌬ktrAB cells of B. subtilis, which lost all of K ϩ about 0.4 h after a hyperosmotic shock with NaCl (11). In this case, although 4 mM K ϩ was present in the medium, the mutant failed to accumulate K ϩ in the cells. Taken together, we predict that one of the roles of the Ktr system is to refill K ϩ , which is lost because of hyperosmotic shock. Interestingly, other K ϩ uptake systems in the ⌬ktrB cells, like Kdp or putative K ϩ channels, could not take over the function of the Ktr system with respect to K ϩ uptake (Figs. 5, C and D and 9). During the late phase of the osmoadaptation, the cells synthesize mainly the osmolyte glucosylglycerol, which replaces the previously accumulated Na ϩ and K ϩ in the cytoplasm and accumulates slowly to high concentrations in many cyanobacteria, including Synechocystis sp. PCC 6803 and Synechocystis sp. PCC 6714 (15)(16)(17)(18)(19). The other indication of the involvement of the KtrB system in long term osmoadapatation ( Fig. 5 and data by Shibata et al.,Ref. 22) is that the reaccumulation of K ϩ to the original or higher concentration level by the Ktr system in the cells after K ϩ loss by hyperosmotic shock is needed for the shift in metabolism of the cells to glucosylglycerol synthesis during the later phases of osmoadaptation. Such a process occurs in E. coli, for which a mutant strain, which is defective in additional K ϩ uptake after hyperosmotic shock has been shown to be impaired in the synthesis of its compatible solute trehalose as well (6). Very recently, this intriguing effect has been explained by the demonstration of the release of transcriptional suppression of osmoregulated genes by high potassium glutamate concentrations in the cytoplasm of E. coli in the early phase of osmoadaptation (59). It remains to be elucidated in Synechocystis sp. PCC 6803.  (47,54) The Na ϩ /H ϩ -antiport process is coupled to a combination of electrogenic H ϩ extrusion by the H ϩ pumps of the cells. The resultant overacidification forms the pmf, which serves an energy source for K ϩ uptake by KtrABE.
The second goal of this work was to characterize the KtrBmediated K ϩ uptake system from Synechocystis sp. PCC 6803. We wanted to know whether in this organism KtrB acts alone, or whether it requires additional gene products for its activity. For this purpose, we cloned putative ktr genes from the Synechocystis sp. PCC 6803 into an E. coli K ϩ uptake negative strain, and examined its growth at low K ϩ concentrations. Unlike the Ktr system from V. alginolyticus, which requires both ktrA and ktrB for K ϩ uptake activity in E. coli (21), Synechocystis Ktr system needs besides these two ktr genes, a third gene, ktrE for the activity (Fig. 1). We confirmed that this result was not an artifact caused by the polar effect of ktrE on ktrB expression in the ⌬ktrE strain by use of the construct with deletion of only the start codon from ktrE (Fig. 1). The ktrE gene encodes a product with sequence similarity to glycosyl transferases (Fig. 1D). This is the first report that KtrE is involved in a function in K ϩ transport, but it is not known how it exerts this function. Dependence of ⌬ component of the pmf for activity for the Ktr system suggests that the Ktr system functions coordinately with the energy generation system like photosynthesis and/or respiratory chain (Figs. 3 and 7). In this study, we found several requirements for exertion of the Ktr system, which are cellular subunits of KtrA and KtrE, ⌬ component, and Na ϩ . This indicates that the Ktr system is highly regulated in Synechocystis sp. PCC 6803.
Because the instant activation of KtrB-mediated K ϩ uptake occurs by the addition of Na ϩ in Synechocystis sp. PCC 6803 (Fig. 6), and the Ktr system is the main K ϩ uptake system in Synechocystis sp. PCC 6803 (Figs. 5, 8, and 9), we predict that a major reason that Synechocystis sp. PCC 6803, and possibly also other cyanobacteria require Na ϩ for their growth, lies in the Na ϩ dependence of their Ktr system. The system from Synechocystis sp. PCC 6803 is the third Ktr system for which Na ϩ dependence has been established (26,27). Because at present no Na ϩ -independent Ktr system is known, Na ϩ dependence may be a general feature of Ktr systems. Our data suggest that the Ktr system does not transport Na ϩ (Fig. 4). However HKT systems from plants, which are close relatives of KtrB, all transport Na ϩ (29, 33-38, 60 -63). Moreover, under certain conditions KtrB proteins from both E. hirae and V. alginolyticus transport Na ϩ as well (Ref. 26). 3 Hence we could not still rule out the possibility that the Ktr system mediates coupled K ϩ /Na ϩ cotransport. Because of the lack of a proper mutant (which should be inactivation of both the activities of Na ϩ /H ϩ antiport and of K ϩ uptake), we could not test this property on the Synechocystis KtrABE system in E. coli.
The genome of Synechocystis sp. PCC 6803 contains 43 genes that encode putative histidine kinases. A histidine kinase has been identified as a sensor of osmotic stress (64), using deletions in each of these genes (65). We have screened the corresponding mutant strains for growth under hyperosmotic conditions because of either 300 mM NaCl or 300 mM sorbitol and found that all of the mutants grew under high osmolyte conditions at which the ⌬ktrB strain did not survive. We have not identified any cross connection between the pathways of KtrBdependent osmoadaptation and a two-component system (data not shown).