Plasma Membrane Aquaporin AqpZ Protein Is Essential for Glucose Metabolism during Photomixotrophic Growth of Synechocystis sp. PCC 6803*

The genome of Synechocystis PCC 6803 contains a single gene encoding an aquaporin, aqpZ. The AqpZ protein functioned as a water-permeable channel in the plasma membrane. However, the physiological importance of AqpZ in Synechocystis remains unclear. We found that growth in glucose-containing medium inhibited proper division of ΔaqpZ cells and led to cell death. Deletion of a gene encoding a glucose transporter in the ΔaqpZ background alleviated the glucose-mediated growth inhibition of the ΔaqpZ cells. The ΔaqpZ cells swelled more than the wild type after the addition of glucose, suggesting an increase in cytosolic osmolarity. This was accompanied by a down-regulation of the pentose phosphate pathway and concurrent glycogen accumulation. Metabolite profiling by GC/TOF-MS of wild-type and ΔaqpZ cells revealed a relative decrease of intermediates of the tricarboxylic acid cycle and certain amino acids in the mutant. The changed levels of metabolites may have been the cause for the observed decrease in growth rate of the ΔaqpZ cells along with decreased PSII activity at pH values ranging from 7.5 to 8.5. A mutant in sll1961, encoding a putative transcription factor, and a Δhik31 mutant, lacking a putative glucose-sensing kinase, both exhibited higher glucose sensitivity than the ΔaqpZ cells. Examination of protein expression indicated that sll1961 functioned as a positive regulator of aqpZ gene expression but not as the only regulator. Overall, the ΔaqpZ cells showed defects in macronutrient metabolism, pH homeostasis, and cell division under photomixotrophic conditions, consistent with an essential role of AqpZ in glucose metabolism.

The genome of Synechocystis PCC 6803 contains a single gene encoding an aquaporin, aqpZ. The AqpZ protein functioned as a water-permeable channel in the plasma membrane. However, the physiological importance of AqpZ in Synechocystis remains unclear. We found that growth in glucose-containing medium inhibited proper division of ⌬aqpZ cells and led to cell death. Deletion of a gene encoding a glucose transporter in the ⌬aqpZ background alleviated the glucose-mediated growth inhibition of the ⌬aqpZ cells. The ⌬aqpZ cells swelled more than the wild type after the addition of glucose, suggesting an increase in cytosolic osmolarity. This was accompanied by a down-regulation of the pentose phosphate pathway and concurrent glycogen accumulation. Metabolite profiling by GC/TOF-MS of wild-type and ⌬aqpZ cells revealed a relative decrease of intermediates of the tricarboxylic acid cycle and certain amino acids in the mutant. The changed levels of metabolites may have been the cause for the observed decrease in growth rate of the ⌬aqpZ cells along with decreased PSII activity at pH values ranging from 7.5 to 8.5. A mutant in sll1961, encoding a putative transcription factor, and a ⌬hik31 mutant, lacking a putative glucose-sensing kinase, both exhibited higher glucose sensitivity than the ⌬aqpZ cells. Examination of protein expression indicated that sll1961 functioned as a positive regulator of aqpZ gene expression but not as the only regulator. Overall, the ⌬aqpZ cells showed defects in macronutrient metabolism, pH homeostasis, and cell division under photomixotrophic conditions, consistent with an essential role of AqpZ in glucose metabolism.
Since identification of the first aquaporin from red blood cells (1), genes encoding aquaporins have been found in both prokaryotic and eukaryotic cells. In animals, abundant major intrinsic protein isoforms are involved in a number of diseases and are known to have a role in the regulation of water homeostasis (2). In plants, aquaporins regulate water permeability and transport in response to external changes in water supply (3). The first aquaporin to be isolated from plant cells was the tonoplast intrinsic protein, ␥-TIP (4). This finding established that aquaporins reside not only in the plasma membrane but also in endomembranes, presumably to coordinate water transport inside the cell. In Synechocystis sp. PCC 6803 (henceforth referred to as Synechocystis) a single copy gene encoding an aquaporin homolog, aqpZ, is present in the genome. The functional characteristics of AqpZ and its subcellular localization in Synechocystis have not been determined, although microarray experiments have identified a list of genes induced by hyperosmotic stress in both the wild type (WT) and a ⌬aqpZ strain (5). Moreover, loss of aquaporins in microorganisms in general does not result in growth defects under a range of environmental conditions (6). Hence, the question as to the physiological role of aquaporins in microbial cells remains open.
In microorganisms, the best studied aquaporin is the AqpZ protein from Escherichia coli. The aqpZ null mutant forms smaller colonies and has reduced viability in medium with low osmolarity compared with the parental wild-type cells (7). However, another study failed to detect any growth defects of an aqpZ disruption mutant under any condition tested (8). Although wild-type E. coli cells have higher water permeability compared with an aqpZ null mutant, it has not been demonstrated that aquaporins are important for proper osmotic adjustment (9). Although the physiological relevance of E. coli AqpZ remains unclear, other functions of aquaporins that are related to specific ecological lifestyles or developmental stages have received increased attention (6,10). Some aquaporin isoforms mediate permeation of glycerol, H 2 O 2 , CO 2 , silicon, or boron in addition to water (11,12). The range of specificities of aquaporins implies that they are involved in processes as diverse as nutrient acquisition, control of development, and growth and defense responses against environmental stress.
Cyanobacteria are prokaryotic microorganisms that perform oxygenic photosynthesis and are adapted to a regular cycle of light and dark periods, in which they are different from nonphotosynthetic microorganisms. In most species of cyanobacteria, glycogen accumulated during the day serves as the predominant metabolic fuel at night. Glucose derived from glycogen or supplied exogenously is catabolized via the oxidative pentose phosphate pathway, glycolysis, and the tricarboxylic acid (TCA) cycle, leading to the production of ATP and carbon skeletons. A glucose-tolerant strain of the cyanobacterium Synechocystis has been isolated previously (13). These cells grow photoautotrophically under light conditions but are also capable of photomixotrophic growth or light-activated heterotrophic growth in glucose-supplemented media (14).
In the present study, we determined the membrane localization and investigated the physiological role of aquaporin AqpZ in Synechocystis. The addition of glucose to ⌬aqpZ cells triggered structural aberrations and morphological abnormalities. Moreover, ⌬aqpZ cells growing on medium containing glucose accumulated more glycogen, and their glucose catabolysis was down-regulated. These data suggest that AqpZ plays a crucial role in the regulation of glucose metabolism under photomixotrophic conditions. To our knowledge, this is the first evidence of a physiological role of AqpZ in addition to its role in the osmotic stress response.

EXPERIMENTAL PROCEDURES
Plasmid Construction-The aqpZ coding region of Synechocystis slr2057 was amplified from genomic DNA by PCR using gene-specific primers (sense, 5Ј-CAGTAGATCTATGA-AAAAGTACATTGCTG-3Ј; antisense, 5Ј-CAGTGCTAGCT-CACTCTGCTTCGGGTTCG-3Ј). The resulting PCR product was cloned into the BglII and NheI sites of pX␤G-ev1 (1). To create Myc-tagged AqpZ, another set of primers (sense, 5Ј-CATGGAATTCCATGAAAAAGTACATTGCTG-3Ј; antisense, 5Ј-CAGTGCTAGCTCACTCTGCTTCGGGTTCG-3Ј) was used to amplify the coding region of aqpZ from genomic DNA by PCR, and the resulting PCR product was cloned into the EcoRI and NheI sites of pX␤G-ev1, placing it in frame with the N-terminal Myc tag contained in the vector. The correct frame was verified by sequencing. Myc-Y69 (AQP-3) from C. elegans and the human aquaporin hAQP1 were used as controls (1).
Expression in Xenopus Oocytes and Measurement of Water Permeability-Capped cRNAs were synthesized in vitro from XbaI-linearized pX␤G-ev1 plasmids using the mMESSAGE mMACHINE T3 kit (Ambion, Austin, TX). Defolliculated Xenopus laevis oocytes were injected with 5 or 10 ng of cRNA or diethyl pyrocarbonate-treated water (1,15). Injected oocytes were incubated for 2-3 days at 18°C in 200 mosM modified Barth's solution (10 mM Tris-HCl (pH 7.6), 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 0.3 mM Ca(NO 3 ) 2 , 0.4 mM CaCl 2 , 0.8 mM MgSO 4 ). An oocyte swelling assay was used to determine the osmotic water permeability, P f (1). Oocytes were transferred to modified Barth's solution diluted to 70 mosM with distilled water, and the time course of volume increase was monitored at room temperature by video microscopy with an on-line computer (16,17). The relative volume (V/V 0 ) was calculated essentially as described by Preston et al. (1). The relative volume (V/V 0 ) was calculated from the area at the initial time (A 0 ) and after a time interval (At): V/V 0 ϭ (At/A 0 ) 3 ⁄ 2 . The coefficient of osmotic water permeability (P f ) was determined from the initial slope of the time course (d(V/V 0 )/dt), initial oocyte volume (V 0 ϭ 9 ϫ 10 Ϫ4 cm 3 ), initial oocyte surface area (S ϭ 0.045 cm 2 ), and the molar volume of water (V w ϭ 18 cm 3 /mol): Oocyte Immunofluorescence and Confocal Microscopy-Oocytes were incubated in fixing solution (80 mM Pipes, pH 6.8, 5 mM EGTA, 1 mM MgCl 2 , 3.7% formaldehyde, 0.2% Triton X-100) at room temperature for 4 h, transferred to methanol at Ϫ20°C for 24 h, equilibrated in PBS (3.2 mM Na 2 HPO 4 , 0.5 mM KH 2 PO 4 , 1.3 mM KCl, 135 mM NaCl, pH 7.4) at room temperature for 2 h, incubated in PBS with 100 mM NaBH 4 at room temperature for 24 h, and bisected with a razor blade (18). Fixed oocytes were blocked in PBS containing 2% BSA for 1 h at room temperature and then incubated at 4°C with the anti-AqpZ antibody (see below) for 24 h followed by incubation with Alexa Fluor 488 goat anti-rabbit IgG in PBS containing 2% BSA for 24 h. Samples were mounted in Fluoromount-G (Southern Biotechnology Associates) and visualized with a PerkinElmer UltraView LCI confocal laser-scanning microscope.
Oocyte Membrane Extraction and Immunoblotting-For each sample, 10 oocytes were homogenized together by pipetting up and down in hypotonic lysis buffer (7.5 mM sodium phosphate, 1 mM EDTA, pH 7.5) containing a protease inhibitor mixture (Sigma-Aldrich) (19). The oocyte yolk was removed by discarding the pellet after a centrifugation at 735 ϫ g and 4°C for 10 min. The supernatant was then centrifuged at 200,000 ϫ g, 4°C, for 1 h. The pellet containing the oocyte membrane fraction was solubilized in buffer (50 mM Tris-HCl (pH 8.0), 50 mM NaCl, 50 mM EDTA-2Na, 10% (w/v) glycerol, and 2% SDS). Total protein content was determined by the bicinchoninic acid (BCA) assay method (Pierce). Equal amounts of protein were separated by SDS-PAGE on a 12% gel. Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane, probed with either anti-AqpZ antibody (see below) or anti-Myc antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) followed by horseradish peroxidase-conjugated donkey anti-rabbit IgG (Amersham Biosciences). The enhanced chemiluminescence detection system (Amersham Biosciences) was used to visualize the immunoreactive proteins by exposure to x-ray films.
Isolation of Synechocystis Membranes-Thylakoid and plasma membrane fractions were prepared from Synechocystis cells as described previously (20,21). An anti-AqpZ antibody was raised against two synthetic peptides with the sequences NH 2 -GSNPLATNGFGDHS-COOH and NH 2 -VLEDLGRPEP-EAE-COOH (Operon, Tokyo, Japan). Polyclonal antibodies raised against the plasma membrane nitrate transporter NrtA (22) or against the thylakoid membrane proteins NdhD 3 and NdhF 3 (23,24) were used to identify the Synechocystis plasma membrane, or thylakoid membrane fractions, respectively. Proteins were separated by SDS-PAGE on a 12.5 or 15% gel and then transferred to PVDF membranes. Membranes were incu-bated for 1 h with the primary antibody (1:2000 in blocking buffer), followed by incubation for 30 min with the secondary antibody (horseradish peroxidase-conjugated goat anti-rabbit IgG (Amersham Biosciences; 1:1000)) and subsequently developed by ECL (Amersham Biosciences).
Immunolabeling and Electron Microscopy-Synechocystis cells grown to an OD 730 of about 1.0 in BG11 medium were fixed with 50 mM phosphatase buffer containing 5% glutaraldehyde and 2% osmium tetroxide (pH 7.2). The samples were dehydrated through a graded acetone series and embedded in Spurr's resin and polymerized. Ultrathin sections were first labeled with AqpZ antiserum (1:20) in Tris-buffered saline and then with 12-nm colloidal gold particles coupled to goat antirabbit IgG. IgG was purified from the serum using the Melon TM Gel IgG spin purification kit (Pierce). The sections were stained with uranyl acetate followed by lead citrate solution and examined with a transmission electron microscope (H-7500, Hitachi).
Bacterial Strains and Culture Conditions-The GT (glucose tolerance) strain of Synechocystis sp. PCC 6803 and its derivatives were grown at 30°C in BG11 medium (25) buffered with 20 mM TES 2 -KOH (pH 8.0) under aerobic conditions. Solid medium consisted of BG11 buffered at pH 8.0, with 1.5% agar, and 0.3% sodium thiosulfate. To test the effects of pH on growth of Synechocystis cells, a range of BG11 media were prepared by replacing TES buffer with MES buffer for pH 6.5, TES buffer for pH 7.0 -8.0, Bicine buffer for pH 8.5, and CHES buffer for pH 9.0 -10.0. Continuous illumination was provided by white fluorescent lamps (50 mol of photons m Ϫ2 s Ϫ1 ). For growth in the dark, culture vessels were wrapped with aluminum foil. For light-activated heterotrophic growth, BG11 plates supplemented with 5 mM glucose were incubated in the dark with a daily 15-min pulse of white light. Unless otherwise stated, cells were grown in volumes of 30 ml in flasks. In all experiments, cells grown in fresh BG11 media as precultures were transferred to BG11 (photoautotrophic growth condition) or BG11 containing 5 mM glucose (photomixotrophic growth condition) at an OD 730 ϭ 0.05.
Construction and Isolation of Mutants-In this study, several mutant strains (⌬glcP (sll0771), ⌬sll1961, and ⌬aqpZ/⌬glcP) were generated by homologous recombination using plasmids described elsewhere (21). The plasmids for disruption of the glcP or sll1961 gene contained an insertion of a kanamycin (K m r ) resistance cassette in the middle of the respective coding sequence. For reintegration of the Synechocystis aqpZ (slr2057) gene into the genome of the ⌬aqpZ cells, the full-length coding sequence of aqpZ was amplified by PCR using an NdeI sitecontaining forward primer 5Ј-CAGTCATATGATGAAAAA-GTACATTGCTG-3Ј and SalI site-containing reverse primer 5Ј-CAGTGTCGACTCACTCTGCTTCGGGTTCG-3Ј. The NdeI-SalI DNA fragment encoding AqpZ was inserted into the corresponding sites in p68TS4OxCm and integrated by homologous recombination into targeting site 4 of the chromosomal DNA of Synechocystis ⌬aqpZ cells (21).
Immunoblot Analysis-Cultures were grown to a cell density of OD 730 ϭ 1.0 and then transferred to fresh BG11 medium either with or without 5 mM glucose. The cells were lysed mechanically by vortexing with glass beads, and the proteins were separated by SDS-PAGE (12.5 or 15% gels) and then transferred to PVDF membranes. The membranes were probed with anti-AqpZ antibody and anti-OpcA antibody (26,27). The membranes were incubated for 1 h with the primary antibody (1:2000 in blocking buffer) and then incubated for 30 min with the secondary antibody (horseradish peroxidase-conjugated goat anti-rabbit IgG (1:1000)) and subsequently subjected to chemiluminescence detection.
Assay of Glucose Uptake-Cells were collected at mid-exponential phase by centrifugation at 5,000 ϫ g for 5 min and then resuspended in BG11 supplemented with 5 mM glucose at an OD 730 of 0.05. Glucose uptake was determined by measuring the concentration of glucose in the medium enzymatically using the D-glucose kit (Roche Applied Science, R-Biopharm), following the manufacturer's instructions. Aliquots (200 l) of the cultures were harvested every 12 h and centrifuged at 1,500 rpm for 2 min, followed by the determination of glucose concentration in the supernatant.
Optical and Transmission Electron Microscopy-Optical microscopy was performed with a microscope (Axioskop FL, Carl Zeiss, Gottingen) equipped with a high definition image capture camera (HC-1000, Fujix, Tokyo) as described previously (21,28). The cell size was analyzed using ImageJ software (National Institutes of Health, Bethesda, MD) and beads of a defined diameter (3.005 Ϯ 0.027 m) as size standards. Cells grown under either photoautotrophic or photomixotrophic conditions were harvested. Cells were fixed in 4% paraformaldehyde and 2% glutaraldehyde in 100 mM cacodylate buffer (pH 7.2) and then postfixed in 1% osmium tetroxide in 50 mM cacodylate buffer. Samples were dehydrated through a graded methanol series, then embedded in Epon812 resin (28). The ultrathin sections were stained with uranyl acetate followed by lead citrate solution and examined with a 1010 transmission electron microscope (JEOL) at 80 kV (28).
Analysis of Cell Size Distribution-Cell size distribution was measured using a particle size distribution analyzer (CDA-1000X, Sysmex). Samples (100 l) were taken from the cultures at the indicated intervals and diluted 100 -500 times with modified physiological saline (0.45% sodium chloride, 144 mosM) prior to performing the measurements.
Determination of Glycogen Content-Analysis of the intracellular glycogen content was performed as described elsewhere (29). The glucose produced by acid hydrolysis was quantified using a colorimetric assay (30).
Metabolite Profiling-Cultures were harvested by centrifugation at 15,000 ϫ g at 4°C for 2 min at the times indicated and washed twice with 1 ml of ice-cold distilled water to remove dead cells, and then cell pellets were frozen in liquid nitrogen. Each sample was extracted, derivatized, and analyzed by gas chromatography-time-of-flight mass spectrometry (GC/ TOF-MS) as described (33)(34)(35)(36). We used lyophilized samples for analysis. Each sample was prepared using 2 mg, dry weight, of cells for GC/TOF-MS analysis. Finally, 22 g of fresh weight of the derivatized extract was injected into the GC/TOF-MS. Nonprocessed data were exported to the netCDF format using ChormaTOF software (version 3.22, Leco) for further data analysis. The raw data were processed using a custom script as described by Jonsson et al. (33,34) to perform base-line correction, alignment, and peak deconvolution. Metabolites were identified by comparing their mass spectrum and retention time index with those generated for reference compounds analyzed on our instrumentation as well as those in the MS and retention time index libraries in the Golm Metabolome Database (37,38).
Analysis of Metabolite Data-Analytical bias in metabolite measurements was controlled using cross-contribution-compensating multiple standard normalization (39) with 16 internal standard peaks. Metabolite peaks with more than 30% missing values were removed. Analysis of variance for examining correlation with glycogen levels and cell diameter was fitted for each metabolite. The formula for examining correlation with duration of culture, glycogen content, and cell diameter is shown in Fig. 8A. Model parameters were calculated using LIMMA (40) for the statistical environment R project (see the R Project site on the World Wide Web). p values were false discovery rate-adjusted to correct for multiple testing. The model used for studying the effect of genotype, g, glycogen level, d, and cell diameter, p, on the estimated level, m, of metabolite i in sample j was as follows, where g:d and g:p are the interactions between genotype and glycogen level and between genotype and cell diameter, respectively, ␤ is the intercept, and ⑀ is the error. We furthermore considered the following, to examine interaction between the same metabolite levels and the duration of culture a. Oxygen Evolution Measurements-Cells grown under photoautotrophic or photomixotrophic conditions for 5 days were harvested by centrifugation at 5,000 ϫ g at room temperature and resuspended in 20 mM TES buffer, pH 8.0, at an OD 730 ϭ 1.0. Cell suspensions were illuminated with white light at 1000 mol of photon m Ϫ2 s Ϫ1 , and oxygen evolution was measured using a Clark-type electrode at 25°C. For the determination of whole chain photosynthetic electron transport activities, cell suspensions were supplemented with 2 mM sodium bicarbonate. PS-II-mediated electron transport activities were measured in the presence of 1 mM 1,4-benzoquinone and 0.8 mM K 3 Fe(CN) 6 in the cell suspension (41).

AqpZ-mediated Water Transport Was Insensitive to Inhibition by Mercury in Xenopus
Oocytes-In order to evaluate its water transport activity, Synechocystis AqpZ was expressed in Xenopus oocytes. Correct expression of wild-type AqpZ or a Myc-tagged version in this system was confirmed by Western blot using either an antibody generated against Synechocystis AqpZ (see "Experimental Procedures") or an anti-Myc epitope antibody. Both proteins, wild-type AqpZ and Myc-tagged AqpZ, were detected in protein extracts of oocytes expressing the respective protein (Fig. 1A, left). Myc-tagged C. elegans aquaporin (Myc-Y69) was used as a positive control (Fig. 1A,  right). The specificity of the anti-AqpZ antibody was also confirmed in this experiment. Confocal microscopy of immunolabeled oocytes showed that AqpZ was localized to the oocyte plasma membrane (Fig. 1B). Oocyte swelling assays were performed to measure AqpZ activity. Oocytes expressing either wild-type AqpZ or the Myc-tagged version exhibited a significant increase in osmotic water permeability (P f ) when compared with water-injected cells (Fig. 1, C and D). This indicated that Synechocystis AqpZ conferred water permeability.
AqpZ Is Localized to the Plasma Membrane-Fractions of plasma and thylakoid membranes were prepared by sucrose density gradient fractionation followed by aqueous polymer two-phase partitioning (20,21,24). As shown in Fig. 2A, a single protein band of the corresponding molecular mass (25.5 kDa) of AqpZ was detected by Western blot in the plasma membrane fraction, which was identified by the presence of the plasma membrane marker protein NrtA (22). No AqpZ protein was detected in the thylakoid membrane fraction, which was identified by the presence of NdhD 3 and NdhF 3 (23,24). In addition, the membrane localization of the AqpZ protein was determined by immunogold labeling followed by electron microscopy. A cross-section of Synechocystis wild-type cells grown under non-stress conditions showed gold particles decorating the plasma membrane when incubated with the AqpZ antiserum (Fig. 2B). Only a small amount of the label was found in other locations. Control experiments with the ⌬aqpZ strain did not show any significant labeling (supplemental Fig. S1). These results indicate that AqpZ was primarily localized in the plasma membrane of Synechocystis.
Synechocystis Cells Lacking the Aquaporin AqpZ Were More Sensitive to Glucose in the Medium-When a glucose-tolerant wild-type strain of Synechocystis PCC 6803 and the aquaporindeficient mutant, ⌬aqpZ, derived from it were grown on BG11 plates in the absence of glucose, both grew equally well (Fig. 3A,  left). However, the addition of 5 mM glucose to the BG11 medium strongly inhibited the growth of the ⌬aqpZ strain (Fig.  3A, right). To confirm that lack of AqpZ was the reason for this sensitivity of ⌬aqpZ cells to glucose, we reintroduced the aqpZ gene into a neutral site of the ⌬aqpZ genome. The complementation construct consisted of aqpZ under control of the trc promoter, the resulting cells were designated ϩaqpZ. By using antibodies against AqpZ, we confirmed the presence of AqpZ protein in extracts from ϩaqpZ cells and the absence of AqpZ protein in extracts from ⌬aqpZ cells (Fig. 3B). The complemented cells were able to grow on glucose medium as well as the wild type, whereas the ⌬aqpZ cells did not grow (Fig. 3C). As a control and to exclude the possibility that the growth inhi-bition was caused by the change in osmolarity due to the addition of 5 mM glucose, we added sorbitol or L-glucose at the same concentration instead of glucose to the cells (Fig. 3C and supplemental Fig. S2). Only glucose caused growth inhibition of the ⌬aqpZ cells. Synechocystis wild-type cells are able to grow heterotrophically in the dark if provided with a short pulse of white light in the presence of glucose (14). We examined the growth of ⌬aqpZ cells under these heterotrophic conditions. An aliquot of a mid-logarithmic phase culture of either wild-type or ⌬aqpZ cells grown on medium without glucose in the light was spotted onto solid BG11 plates with or without 5 mM glucose (Fig.  3D) and grown in the dark. In contrast to the wild-type cells, ⌬aqpZ cells were incapable of growing on medium containing 5 mM glucose in the dark (Fig. 3D). Similar to this, after 48 h of growth in liquid medium with glucose in the light, the growth rate of ⌬aqpZ cells was much slower than that of wild-type cells (Fig.  3E). These results indicate that loss of aqpZ caused hypersensitivity to glucose and that ⌬aqpZ cells were unable to effectively use glucose as a carbon source for growth.
We observed individual Synechocystis cells grown in liquid medium with glucose using phase-contrast imaging. The number of dark cells was higher than the number of bright cells in the ⌬aqpZ culture compared with the wild type (Fig. 4, A and

FIGURE 2. Membrane localization of AqpZ in Synechocystis.
A, plasma membrane localization of AqpZ. Plasma membranes (PM) and thylakoid membranes (TM) were isolated by sucrose density fractionation followed by aqueous polymer two-phase partitioning. Shown are Coomassie Brilliant Blue staining (left) and immunoblotting (right) of the plasma membrane and the thylakoid membrane fractions from wild-type Synechocystis. NrtA (marker for the plasma membrane fraction) and NdhD 3 and NdhF 3 (markers for the thylakoid membrane fraction) were detected on Western blots using the corresponding antibodies. B, crosssection of Synechocystis wild-type cell immunolabeled using an anti-AqpZ antibody. AqpZ protein, indicated by the presence of gold particles (arrowheads), was localized in the plasma membrane. Bar, 500 nm. B). Those cells appearing dark using phase-contrast microscopy did not show autofluorescence under UV light (Fig. 4, C  and D), indicating that they lacked chlorophyll. Based on these results, dark cells were considered to be dead cells. After growth for 48 or 72 h, the number of dark cells in cultures of the ⌬aqpZ strain was much higher than in wild-type cultures (Fig. 4E). These results showed that AqpZ is essential for survival of Synechocystis under photomixotrophic conditions. Fig. 4, we noticed a difference in size and morphology between the ⌬aqpZ and the wild-type cells. We therefore compared wild-type and ⌬aqpZ cells during incubation with or without 5 mM glucose under the microscope (Fig. 5, A-D). There was no significant difference in the size of wild-type and ⌬aqpZ cells grown in liquid medium without glucose (Fig. 5, A and B). In contrast, after 48 h of growth on medium containing 5 mM glucose, both wild-type and ⌬aqpZ cells became larger (Fig. 5, C and D). However, the diameter of the ⌬aqpZ cells increased more than that of the wild-type cells (Fig. 5, C and D). The distribution of cell sizes of the wild type, ⌬aqpZ, and cells lacking the glucose uptake transporter GlcP (⌬glcP) was evaluated over the course of 72 h using a particle counter (Fig. 5, E-H). The size distribution of wildtype, ⌬aqpZ, and ⌬glcP cells was similar when cells were grown without glucose (Fig. 5, E and F). After 24 h of incubation in the presence of 5 mM glucose, ⌬aqpZ cells were larger than the wildtype cells. The size of the ⌬glcP cells remained largely unchanged over the course of the experiment (Fig. 5, G and H). In the ⌬aqpZ cell culture, a significant number of small cells (less than 1.5 mm in diameter) were observed after 48 h; these were most likely dead cells.

Deletion of aqpZ Resulted in an Increase in Cell Volume in the Presence of Glucose-While performing the microscopic observations shown in
Cell Division Was Affected in ⌬aqpZ Cells Grown in the Presence of Glucose-To assess the effects of glucose on cell morphology in ⌬aqpZ cells, we examined the cells by transmission  electron microscopy. When both wild-type and ⌬aqpZ cells were cultured in medium without glucose, their appearance was similar (Fig. 6, A and E). After incubation in liquid medium with glucose for 24 h, glycogen granules (white spots) were visible in sections of both wild-type and ⌬aqpZ cells to a similar extent (Fig. 6, B and F). However, at 48 and 72 h, a number of ⌬aqpZ cells containing glycogen granules showed dramatic changes compared with the wild type (Fig. 6, G-I). The structure of the thylakoid membranes in these ⌬aqpZ cells was disordered, and oval-shaped cells containing bodies of high electron density were observed (Fig. 6, I and J). Fig. 6, G-J, shows the unusual appearance of these cells that also were unable to complete cell division. These structural aberrations and morphological abnormalities of ⌬aqpZ cells were correlated with the presence of glucose in the medium.
Glucose Uptake Is Required for Glucose-induced Growth Inhibition of ⌬aqpZ Cells-To examine whether ⌬aqpZ cells were sensitive to intracellular or extracellular glucose, we studied the effect of deletion of the glucose uptake transporter, GlcP (42)(43)(44). Consistent with previously reported results, the ⌬glcP mutant showed strongly reduced glucose uptake from the medium (Fig. 7A). When GlcP was deleted in the ⌬aqpZ background (⌬glcP/⌬aqpZ mutant), these cells were no longer sensitive to glucose (Fig. 7B). Next, glucose analogs were tested to assess the cause of glucose toxicity in ⌬aqpZ cells. L-Glucose (an enantiomer of glucose, which is very poorly transported into cells), 3-o-methyl-glucose (a glucose analog, which can be taken up into the cell but not phosphorylated by hexokinase), 2-deoxyglucose (which is phosphorylated by hexokinase but inhibits the production of glucose 6-phosphate, NADPH, and ATP), or fructose was added to solid media and growth was tested. Unlike glucose, none of the analogs resulted in a significant difference in growth between wild-type and ⌬aqpZ cells, either on solid (Fig. 7C) or in liquid media (data not shown) (44). These data indicated that there is no difference in glucose uptake or hexokinase activity between the wild-type and ⌬aqpZ cells.
The Activity of the Pentose Phosphate Pathway was Decreased in ⌬aqpZ Cells-The expression of genes encoding enzymes in the glycolytic and the pentose phosphate pathways has been studied under various nutrient conditions (29, 44 -50). Following its uptake into the cell, glucose is converted to glucose 6-phosphate by glucokinase, and glucose 6-phosphate is then further metabolized by glucose 6-phosphate dehydrogenase, an enzyme belonging to the pentose phosphate pathway. In Synechocystis, the OpcA protein is required for glucose-6-phosphate dehydrogenase activity (26,27). Changes in the amount of OpcA protein in response to the addition of glucose to the medium were determined by immunological analysis (Fig. 8, A  and B). Expression of OpcA protein was induced by glucose after 48 h (Fig. 8, A and B). However, the level of induction of OpcA protein in ⌬aqpZ was approximately half as high as that in the wild type (Fig. 8, A and B). This indicates that the aqpZ mutation inhibited the glucose-mediated activation of the pentose phosphate pathway.
⌬aqpZ Cells Contained More Glycogen-The decrease in induction of OpcA protein in ⌬aqpZ cells in response to glucose (Fig. 8) and the existence of glycogen granules in the cells (Fig. 6) suggested that glucose catabolism was decreased in these cells as well. We therefore hypothesized that the cells may convert excess glucose to glycogen in order to prevent an increase in intracellular osmolarity. The glycogen content of wild-type and ⌬aqpZ cells during log phase was the same before the addition of glucose (0 h in Fig. 9). After the addition of glucose to the medium, the amount of glycogen reached its peak at 48 h in the wild type and then decreased again. The ⌬aqpZ cells accumulated significantly more glycogen over the course of the experiment than the wild type, and no decrease in glycogen content was seen at 72 h (Fig. 9), at which point the mutant contained 5.6 times more glycogen than the wild type at 0 h. Metabolite Profiles of Synechocystis Cells Grown under Photomixotrophic Conditions-To elucidate the impact of loss of aqpZ on the metabolite content, we determined the levels of 72 metabolites in extracts of wild-type and ⌬aqpZ cells (Fig. 10 and supplemental Figs. S2-S5). We applied two-way analysis of variance to find possible interactions between incubation time, intracellular glycogen content, or cell size and metabolite levels ( Fig. 10A and supplemental Fig. S2). The statistical analysis revealed significant differences in metabolite levels between wild-type and ⌬aqpZ cells (Fig. 10B and supplemental Fig. S2). The culture time-based profiles, which were verified at 48 h under photomixotrophic conditions, showed reduced amounts of isocitrate, citrate, glutamate, sucrose, succinate, ornithine, and spermidine in ⌬aqpZ cells and a higher abundance of glucose and glucose 6-phosphate, (Fig. 10B). Interestingly, the lower level of spermidine, known as a compound that enhances cell proliferation, was correlated with the decrease in cell division of the ⌬aqpZ cells (51). When we compared the correlation between intracellular glycogen levels or cell sizes and changes in metabolite levels, some of the interaction effects were overlapping (Fig. 10A). The overall changes in the levels of metabolites showed that glucose metabolism was inhibited in the ⌬aqpZ cells ( Fig. 10 and supplemental Figs. S3-S6).
Synechocystis AqpZ Does Not Transport Glycerol-One possible explanation for the finding that products of the early steps of glycolysis accumulated in the ⌬aqpZ mutant would be that  AqpZ has a function in the export of glycerol, one of the products of glucose metabolism. Glycerol transport activity has been shown for some other aquaporins (52,53). We examined whether AqpZ is able to transport glycerol by using heterologous expression in E. coli (supplemental Fig. S7). The E. coli MA05 (glpF Ϫ ) or MA06 (aqpZ Ϫ , glpF Ϫ ) strain complemented with the E. coli glycerol transporter gene glpF was able to grow on medium containing 2 mM glycerol as the sole carbon source. In contrast, Synechocystis AqpZ was not able to rescue the growth defects of strain MA05 or MA06. The same was true for E. coli AqpZ and the empty vector (supplemental Fig. S7). These data showed that Synechocystis AqpZ did not function as a glycerol transporter in the same way as E. coli GlpF.
pH Dependence of ⌬aqpZ Cells-Because organic compounds contribute to the cellular pH buffering system (54,55), the reduced amount of certain metabolites found in ⌬aqpZ cells (Fig. 10) may result in a disturbance of pH homeostasis in the cells. It has also been reported that glucose affected the pH sensitivity of Synechocystis (41,56). Therefore, we tested the growth of ⌬aqpZ cells in liquid medium with pH values ranging from 6.5 to 10. In medium without glucose, ⌬aqpZ cells grew identical to the wild type at all pH values tested (Fig. 11, A and  B). Oxygen evolution as a measure of PSII activity was the same for wild-type and ⌬aqpZ cells in medium without glucose at all pH values tested (Fig. 11E). In contrast, under photomixotrophic conditions, when glucose was included in the medium, neither the wild-type nor the ⌬aqpZ cells grew at pH 6.5 (Fig. 11, A, B, and D). The most significant difference was that the growth rate of the ⌬aqpZ cells was strongly decreased after 72 h at pH values from 7.5 to 8.5 (Fig. 11, B and D). Oxygen evolution of the ⌬aqpZ cells was also strongly decreased at these pH values (Fig. 11, E and F). These results indicated that hypersensitivity of the ⌬aqpZ cells to glucose was pH-dependent.
Comparison of the Glucose Sensitivity of ⌬sll1961 or ⌬hik31 Mutants with ⌬aqpZ Cells-DNA microarray and Northern blotting analysis conducted previously revealed that deletion of sll1961, encoding a putative transcription factor, resulted in a decrease in the transcription level of aqpZ (57). The sll1961 mutant was also very sensitive to glucose in the medium (58). The aqpZ (slr2507) gene is adjacent to sll1961 but in the opposite orientation on the chromosome (Fig. 12A). In addition,   inactivation of the hik31 gene, encoding a sensory histidine kinase, also suppressed cell growth in the presence of glucose (44). We compared the phenotypes of these two mutants with that of the ⌬aqpZ cells (Fig. 12B). Both mutants, ⌬sll1961 and ⌬hik31, exhibited more severe growth inhibition in the presence of glucose than the ⌬aqpZ cells. Immunoblot analysis showed that in the presence of glucose in the medium, the amount of AqpZ protein was slightly decreased in the ⌬sll1961 cells but unchanged in the ⌬hik31 cells (Fig. 12C). High light conditions increased the amount of AqpZ protein in the wild type, which was consistent with the results of the DNA microarray (57). No increase in AqpZ protein expression level in high light conditions was seen in the ⌬sll1961 mutant (Fig.  12D). These data suggest that sll1961 may partially regulate the expression of AqpZ protein.

DISCUSSION
Loss of function of the plasma membrane-localized aquaporin AqpZ made Synechocystis cells unable to grow in medium containing glucose. ⌬aqpZ cells showed aberrant cellular morphology and were affected in cell division (Figs. 2 and 6). This glucose sensitivity was dependent on glucose uptake, and in the absence of a functioning glucose transporter, glucose in the medium had no negative effect on ⌬aqpZ cells (Fig. 7). This is compelling evidence because not much information on the phenotypes of aquaporin-deficient strains is available (6). Synechocystis uses the degradation of glucose into smaller molecules via the pentose phosphate pathway and glycolysis to increase intracellular osmolarity (44,50). Consistent with this, both wild-type and ⌬aqpZ cells grown on medium with glucose swelled more compared with cells grown in the absence of glucose (Fig. 5). The ⌬aqpZ cells were larger in size than the wild type, which may be the result of a decrease in water permeability due to the loss of AqpZ function. Based on OpcA protein expression (Fig. 8) and metabolite profiling of the ⌬aqpZ cells ( Fig. 10 and supplemental Figs. S3-S6), glucose in the medium led to a decrease in glucose degradation and to enhanced glycogen synthesis, most likely to limit swelling of the cells (Fig. 9).  This is in agreement with previous reports that the glycogen content was elevated or that glucose catabolism was impaired in Synechocystis when cells were exposed to excess glucose (29,41,48,59). The observed decrease in the content of spermidine, a compound known to be involved in the enhancement of cell division, is consistent with decreased cell growth after glucose addition to the medium (51) (Fig. 10).
Water transport across the membrane is coordinated with the transport of various solutes, which are in turn closely associated with the formation of proton motive force consisting of pH gradient and membrane potential across the membrane (12,60). The potassium transport system, consisting of Ktr-type transporters and potassium channels, and the sodium extrusion system of Na ϩ /H ϩ antiporters in the plasma membrane of Synechocystis function to control the turgor pressure and the electrochemical membrane potential in response to extracellular or intracellular environmental changes (21,(61)(62)(63)(64)(65)(66)(67)(68)(69). The immediate adjustment to changes of membrane potential or cellular osmolarity requires fast water flux across the plasma membrane in response to the conversion of glucose into various types of molecules through the carbon metabolism pathway. In addition, intermediate metabolites like citrate and malate derived from glucose through glycolysis and the TCA cycle are known to be important for regulation of pH homeostasis, ⌬pH, and the proton motive force (70). Synechocystis mutants impaired in tocopherol biosynthesis, which is an antioxidant small metabolite, are also affected in their pH dependence (41). The ⌬aqpZ mutant was sensitive to pH values between pH 7.5 and pH 8.5 in glucose-containing medium and showed inactivation of PSII (Fig. 11). This may be the result of changed levels of metabolites detected in the ⌬aqpZ cells.
We analyzed potential effects on carbon metabolism by GC/TOF-MS profiling of ⌬aqpZ and wild-type cells and compared their metabolite contents. Metabolite profiling data were used to analyze possible interactions between glycogen content, cell size, and metabolite levels (supplemental Figs. S4 -S6). The content of glucose 6-phosphate and glucose increased in the ⌬aqpZ cells after 48 h of growth, whereas the levels of products of the tricarboxylic acid cycle and some amino acids remained unchanged or were lower compared with the wild type. This is in agreement with the finding that the influx of glucose into the pentose phosphate pathway was reduced (Fig.  8). Consistent with this, statistical analyses showed that the ⌬aqpZ cells overall contained lower levels of metabolites compared with the wild type ( Fig. 10 and supplemental Figs. S3-S6). The reduction in the amount of metabolites derived from the degradation of glucose prevented an increase of the intracellular osmolarity and promoted glycogen synthesis in the cells.
Fujimori et al. (57) isolated the sll1961 mutant based on its hypersensitivity to high light conditions. In a subsequent study, the same group found that glucose in the medium reduced the growth of the sll1961 mutant (58). DNA microarray analysis revealed that loss of function of sll1961 resulted in decreased expression of aqpZ at the transcriptional level, indicating that sll1961 is a positive transcription regulator for aqpZ gene expression. This functional relationship between sll1961 and aqpZ is further supported by their adjacent position in the Synechocystis genome. We compared the sensitivity of the ⌬sll1961 and the ⌬aqpZ mutant to glucose and the level of AqpZ protein in cells grown without and with glucose and found some evidence for a regulatory role of sll1961 on aqpZ expression (Fig.  12). However, because loss of function of sll1961 did not result in a lack of AqpZ protein, we concluded that sll1961 only had a limited role in aqpZ expression.
Like other aquaporins, AqpZ can mediate both inward and outward water flow (Fig. 1) (71). Because the genome of Synechocystis PCC 6803 contains a single gene encoding an aquaporin, aqpZ, and our results showed that AqpZ resides in the plasma membrane ( Fig. 2 and supplemental Fig. S1), this leads to the following questions. Is AqpZ the only water transport system? What is the water permeation route across the thylakoid membrane? The detailed structural analysis of potassium channels revealed that they allow the passage of water molecules through the conducting pore (72). Considering these findings, we predict that other membrane proteins in addition to AqpZ may conduct the diffusion of water across membranes.
It is also possible that AqpZ may be required for the transport of products synthesized during the process of glucose metabolism. Many aquaporins have been documented to act as facilitators of water, glycerol, peroxide, CO 2 , and other small, uncharged, ubiquitous molecules (11,12). We therefore tested the possibility that glycerol may be able to pass through AqpZ by heterologous expression of aqpZ in E. coli glpF mutants but did not observe glycerol transport activity (supplemental Fig.  S7). It cannot be excluded that passage of small metabolites through AqpZ may be required in order to discharge cytotoxic compounds from cells growing on glucose-containing media. This would explain why cells lacking AqpZ were hypersensitive to glucose and were larger than the wild type in the presence of 5 mM glucose (Fig. 5).
In summary, our efforts in characterizing ⌬aqpZ cells under photomixotrophic conditions have yielded new insights into the role of aquaporin in glucose metabolism in Synechocystis. Loss of function of aqpZ resulted in a down-regulation of glucose hydrolysis after glucose addition into the medium and an increase of glycogen synthesis. Therefore, in addition to its role as a water channel, AqpZ plays a role in regulating cell volume, macronutrient homeostasis, and cell division under photomixotrophic conditions.