Essential role of the plasmid hik31 operon in regulating central metabolism in the dark in Synechocystis sp. PCC 6803

Authors


Summary

The plasmid hik31 operon (P3, slr6039-slr6041) is located on the pSYSX plasmid in Synechocystis sp. PCC 6803. A P3 mutant (ΔP3) had a growth defect in the dark and a pigment defect that was worsened by the addition of glucose. The glucose defect was from incomplete metabolism of the substrate, was pH dependent, and completely overcome by the addition of bicarbonate. Addition of organic carbon and nitrogen sources partly alleviated the defects of the mutant in the dark. Electron micrographs of the mutant revealed larger cells with division defects, glycogen limitation, lack of carboxysomes, deteriorated thylakoids and accumulation of polyhydroxybutyrate and cyanophycin. A microarray experiment over two days of growth in light-dark plus glucose revealed downregulation of several photosynthesis, amino acid biosynthesis, energy metabolism genes; and an upregulation of cell envelope and transport and binding genes in the mutant. ΔP3 had an imbalance in carbon and nitrogen levels and many sugar catabolic and cell division genes were negatively affected after the first dark period. The mutant suffered from oxidative and osmotic stress, macronutrient limitation, and an energy deficit. Therefore, the P3 operon is an important regulator of central metabolism and cell division in the dark.

Introduction

Cyanobacteria grow in the natural environment in alternating light–dark (LD) cycles. Such a growth pattern requires sensing the change in light intensity and adjusting to this change through regulation of genes in various pathways for utilization of appropriate energy sources for metabolism. Synechocystis sp. PCC 6803 is a model cyanobacterium for many studies that involve photosynthesis, responses to environmental changes and biofuel production. This organism can grow either in autotrophic conditions using light and CO2 from the air or on an organic source of carbon such as glucose (Singh and Sherman, 2005). Central metabolism in the light with glucose involves the photosynthetic electron transport chain and the Calvin cycle for carbon fixation, the storage of carbon in the form of glycogen granules, glycolysis and a TCA cycle with N assimilation through the glutamine synthase-glutamate synthase (GS-GOGAT) pathway. In the dark, glycogen can be metabolized through the upper half of glycolysis, the oxidative pentose phosphate (OPP) pathway, lower half of glycolysis, the oxidative and reductive branches of the TCA cycle and the C4 cycle (Yang et al., 2002a; Steuer et al., 2012). Thus, Synechocystis can grow on glucose under (i) continuous light and LD cycles through mixotrophic (MT) growth; (ii) light activated heterotrophic growth (LAHG) that mainly consists of dark periods with brief pulses of light; and (iii) heterotrophic (HT) growth in continuous dark (DD) as shown by Yang et al. (2002b).

The regulatory schemes that control glucose utilization in alternating LD cycles are very complex due to the large number of genes, metabolites and pathways involved. In this condition, there is catabolism and anabolism of storage compounds like glycogen, polyhydroxybutyrate, polyphosphate and cyanophycin. The means by which cells co-ordinate these metabolic processes and the regulatory genes responsible for controlling targets in the central metabolic pathways are not completely known. Over 25 mutants with phenotypic defects in glucose metabolism in Synechocystis have been described previously (Nagarajan, 2013). Many of these mutants show defective expression of glucose catabolic genes or to a lesser extent in photosynthesis (PS) and carbon fixation genes and some have defects when grown in glucose in continuous low light (LL) or high light (HL). Only a few mutants in regulatory genes;, e.g. hik8 (Singh and Sherman, 2005), rre37 (Tabei et al., 2007), sigE (Osanai et al., 2005), the double mutants sigBsigE and sigDsigE, the triple mutant sigBsigDsigE (Summerfield and Sherman, 2007); and the aquaporin mutant aqpZ (Akai et al., 2011), have been found to have defects when grown on glucose in alternating LD cycles or LAHG conditions. A hik31 double mutant has a glucose defect in LL and in the dark when grown in high CO2 (HC) due to lower expression of icfG, constitutive Glc-6-P dehydrogenase activity and no glucokinase activity (Kahlon et al., 2006). We have begun an intensive study of the hik31 operons (Nagarajan et al., 2012) and are focusing on the dark-associated glucose defect.

The genomes of cyanobacteria are quite diverse and the majority of the sequenced strains do not contain plasmids. Synechococcus elongatus PCC 7942 and S. elongatus PCC 6301, each contain two plasmids, and six other strains contain between 3–9 plasmids. The few studies undertaken on the functions of individual plasmid genes in cyanobacteria have revealed roles in antibiotic and metal resistance, production of toxins and gas vacuoles, sulphur metabolism, signal transduction and thermal tolerance (Chen et al., 2008; Kimura et al., 2002 and references therein; Steinmuller et al., 1991; Lopez-Maury et al., 2009). Some mutants in cyanobacterial plasmid genes have been lethal for the host cell suggesting that these genes encode some essential functions (Kimura et al., 2002; Nagarajan et al., 2012). Due to the availability of sophisticated microarray slides containing genes from all seven plasmids of Synechocystis, more studies are being undertaken that focus on the expression and importance of plasmid genes (Singh et al., 2008; 2009).

In a previous study, we showed that the plasmid operon mutant ΔP3, with a deletion of three genes in the plasmid pSYSX (slr6039-slr6041) grew similar to the WT in photoautotrophic (PA) LL, MT LL and PA HC conditions, slower than the WT in PA LD, MT LD, HT DD and MT HC conditions and poorly in PA HL conditions (Nagarajan et al., 2012). The cells for this mutant were also larger than the WT and showed varying degrees of division and shape defects after the 3rd day in all growth conditions. Metal stress experiments revealed a defective phenotype in cobalt chloride salts at a final concentration of 5 mM, suggesting that P3 may control the signal transduction of the neighbouring cation transporter genes. The P3 operon was also shown to be important for low O2 adaptation as only the mutants lacking P3 did not show increased growth in low O2 conditions (Summerfield et al., 2011). Our results suggested that P3 encodes elements that act as a positive regulator of PS, carbon breakdown, cell division and cation transporter genes. Although chromosomal homologues are found in many other cyanobacteria, plasmid homologues for the hik31 operon are only found in Anaebaena azollae 0708 on plasmid pAzo01, Anabaena sp. PCC 7120 on plasmid Beta, and in S. elongatus PCC 7942 on plasmid pANL.

We also showed previously that the two duplicated hik31 operons P3 (slr6039-6041) on the plasmid, and C3 (sll0788-sll0790) on the chromosome are not redundant, and are differentially regulated in the light and dark. In addition, mutants for both copies had similar phenotypes in PA LL and MT LL, but differed when exposed to high light, dark periods with and without glucose, or high CO2 with glucose. This suggests that P3 and C3 are functionally divergent, have common targets in the light and separate targets in the dark, and also share a mutually beneficial regulatory relationship. The ΔC3 mutant was similar to the WT, only slightly affected in glucose growth in LL and LD, but had defective growth in MT HL. P3 transcript expression was lower than C3 and like C3, was induced by glucose and high light in active growth conditions. Unlike C3, P3 expression was upregulated in the dark with or without glucose, indicating that it has an additional role in regulating metabolism in the dark.

The goal of this work was to identify the role of hik31 and the neighbouring response regulator in response to various environmental conditions utilizing the mutants described previously (Nagarajan et al., 2012). We chose to work primarily with the P3 mutant under MT LD conditions, since this generated the worst growth among all of the mutants. In this study, we explore in detail the role of P3 in regulation of LD and glucose metabolism through transcriptomics, functional phenotyping and metabolic complementation. Our results show that P3 is an important regulator in the dark that helps maintain and co-ordinate photosynthetic metabolism with carbon breakdown and nitrogen assimilation; and that P3 mediates macronutrient, redox and pH homeostasis as well as cell division. We also show a relationship between P3 and the previously studied regulators hik8, hik34, rre37, sll0822, sigma factors and their targets, thereby expanding on known signal transduction networks.

Results

Effect of glucose on growth and pigmentation in the light and the dark

We studied the effect of glucose on growth through viable count determination for the WT and the P3 mutant cells grown in PA LL, PA LD, MT LL and MT LD for 3 days and plated on media with and without glucose. WT and mutant PA LL cells grew better on plates without glucose than with glucose (1.7- and 2-fold respectively), suggesting that autotrophically grown cells were stressed on exposure to glucose and could not adapt as quickly to glucose media (Table 1). The WT MT LL and WT PA LD cells continued to grow well on media with glucose. WT PA LD cells grew more than 1000-fold better than ΔP3 PA LD cells on plates with glucose. In addition, the WT MT LD grown cells grew 2.5-fold better on plates without glucose than with glucose, more than 100-fold better than the mutant on media with glucose, and more than 1000-fold better on media without glucose. The results indicated that growth under LD, especially in the presence of 5 mM glucose, led to severe problems of survival on plates, especially when glucose was present. Pigment content also was assessed for Chl a, phycbillins and carotenoids. The results shown in Fig. S1 demonstrated that there were pigment alterations in the mutant and that glucose worsened the pigment defect in LL and even more so in LD.

Table 1. Viable counts for the WT and ΔP3 grown for 3 days in PA and MT conditions
Growth conditionStrainViable counts in cfu ml−1 (×107)a
Control, no glucoseGlucose (5 mM)
  1. aCells were grown from same starting cell density in labelled condition for 3 days and then serially diluted and spread plated onto control and glucose media. Plates were incubated in LL conditions for 8 days before counting colonies.
PA LLWT10.06.0
ΔP310.05.0
MT LLWT6.010.0
ΔP38.07.0
PA LDWT1.02.0
ΔP30.8<0.001
MT LDWT10.04.0
ΔP30.06<0.001

Growth with other sugars

Spotting results with various other sugars as carbon sources for the WT and ΔP3 are presented in Fig. S2. We compared the inhibitory effect of glucose on the mutant with the known effects of these other sugars. 3OMG is permeable, not phosphorylated by glucokinase and not metabolized; fructose is permeable and lethal; mannose is phosphorylated by glucokinase, but not further metabolized as it inhibits the production of G-6-P, NADPH and ATP; and mannitol and sucrose are non-permeable solutes (Hihara et al., 1998; Lee et al., 2005). Most of these sugars were tolerated by the cells (except for fructose and mannose that were lethal), so that they grew in the presence of these sugars similar to the control PA LD cells. The effect of glucose for the P3 mutant was in between that obtained for a permeable, not phosphorylated, not metabolized sugar effect (3OG), and a sugar that is permeable, phosphorylated, not further metabolized and inhibits energy production (mannose). Thus, we concluded that glucose is permeable, phosphorylated and partially metabolized by the mutant. These results showed that the defect caused by glucose was unique from the other sugars and was due to insufficient metabolism after entry.

Effect of addition of glycolysis and TCA components

Growth experiments were carried out for the WT and the mutant in LD with glucose and supplemented individually with six carbon sources from the TCA and glycolysis pathways (pyruvate, acetate, oxaloacetate, succinate, fumarate and malate) to see if they could complement the growth defect in glucose metabolism (Fig. S3). For the WT, none of the additions increased growth above the control. Growth for the plasmid operon mutant was improved by the addition of all these compounds compared with growth with glucose alone, with the best growth seen with acetate, succinate and fumarate. This suggested a defect in metabolism for the plasmid operon mutant in both the glycolysis and TCA pathways, since addition of pyruvate and TCA components partly complements the defect. Such improved growth on addition of TCA metabolites has been seen previously for the spkD and icfG regulatory mutants (Beuf et al., 1994; Laurent et al., 2008).

Microarray experiment

A microarray experiment was performed in MT LD to determine the different genes affected by the removal of P3 that led to the defects in growth, pigment content, and carbon breakdown in the dark. We also performed the same experiment with the C3 operon mutant to assess the role of C3, since growth of ΔC3 was hardly perturbed at all under similar growth conditions. The microarray time points were chosen to identify genes more directly affected by P3 and C3 (day 1, L1 and D1) and those eventually affected (day 2, L13 and D13).The growth of the mutants was little affected in day 1, and affected significantly for ΔP3, but not ΔC3 in day 2. ΔP3 cells showed a 1.3-fold reduced growth at L13 and a 1.7-fold reduced growth at D13 compared with the WT (data not shown). The transcriptional response of the two operon mutants in the same time points was striking and reflected the growth patterns. Both strains had a relatively similar number of differentially transcribed genes at L1, but even at D1, fewer genes were affected in ΔC3. The major differences between the two mutants were seen in the day 2, where ΔC3 had very few differentially expressed genes (data not shown, manuscript in preparation). On the contrary, ΔP3 showed extreme changes in the transcriptional patterns in the second day, with many hundreds of genes affected. We interpret this result to mean that ΔC3 and ΔP3 respond similarly to the addition of glucose in the light, but face different consequences once they initiate growth in the dark. By the second day, ΔC3 had recovered and grew as well as WT, whereas ΔP3 could not grow and a large host of metabolic pathways were affected. These results demonstrated the important role of the P3 operon in MT LD conditions that is not shown by C3. We will describe the metabolic and regulatory fate of gene transcription in ΔP3 in detail, as cells attempt to compensate for the loss of the P3 operon under MT LD conditions.

Table 2 shows the gene categories that changed in the different time points and the large number of genes significantly upregulated and downregulated in the mutant compared with the WT. The full microarray dataset and adjusted P-values for all genes are provided in Table S1A and arranged according to general pathways. Subsequent Tables S2 to S5 have genes arranged in sections A to F and more based on functional categories. Gene categories that contained more genes upregulated in both days of the experiment included cell envelope and transport and binding proteins. Gene categories that were mainly downregulated included amino acid biosynthesis; central intermediary metabolism; energy metabolism; PS and respiration; and purines, pyrimidines, nucleosides and nucleotides (Fig. S4). Certain gene categories in Table 2 had oppositely expressed genes in day 1 and day 2. Cellular processes and transcription genes were upregulated in day 1, but downregulated by day 2. Some genes in biosynthesis of cofactors, prosthetic groups and carriers; DNA replication, restriction, modification, recombination and repair; other categories; and regulatory functions were downregulated in day 1 and upregulated by day 2.

Table 2. Gene categories significantly changed (upregulated or downregulated) in the P3 mutant (FDR ≤ 0.01, fold change ≥ 1.5)
General pathwaysNumber of genesΔP3/WT
L1D1L13D13
image image image image image image image image
Amino acid biosynthesis972611316282026
Biosynthesis of cofactors, prosthetic groups and carriers12541031128253421
Cell envelope67418312111811
Cellular processes7763112626826
Central intermediary metabolism31101356212
DNA replication, restriction, modification, recombination and repair730015237257
Energy metabolism93344517181526
Fatty acid, phospholipid and sterol metabolism39222169810
Hypothetical124329626048237246289247
Other categories2531014121340464844
Photosynthesis and respiration142739416684688
Purines, pyrimidines, nucleosides and nucleotides430316711910
Regulatory functions1532108938204326
Transcription303031516815
Translation168227281014832078
Transport and binding proteins199141413745235729
Unknown6112431602992117114146
Total3444113226220182597776724822
Total genes and per cent of genome changed on slide 339, 9.8%402, 11.7%1373, 39.9%1546, 44.9%

Gene expression for both hik31 operons and validation of the microarray experiment

We tested the expression of the individual hik31 operon genes in our microarray time points for both cultures through RT-PCR and restriction digests to differentiate the expression of both C3 and P3 operon copies. The plasmid copy P3 op and hikP were progressively upregulated in the WT from L1 to D13 and missing in the mutant (Fig. 1A). We selected differentially expressed genes from the microarray in several categories and tested the correlation with RT-PCR. There was an excellent match between the two for almost all the genes tested. A few selected gene results are included in Fig. 1 and Table S1B.

Figure 1.

Expression of the hik31 operon genes and selected differentially transcribed genes from the microarray experiment for the WT and ΔP3.

A. Gels showing RT-PCR amplification of both hik31 operons and individual genes. The RT-PCR products for both copies of the operons were separated using restriction digests and separate primers were used for the hiks as previously described (Nagarajan et al., 2012). Thirty-five cycles were used for all the genes and rnpB was used as a control.

B. Genes from different categories that showed altered behaviour were selected for validation of the microarray experiment. The number of cycles used is indicated next to the gene name in the figure.

Photosynthesis, pigment biosynthesis, redox and stress responses

The genes referred to in this section are listed in Table S2. Many photosynthesis (PS) genes were downregulated in L1 and day 2 in the P3 mutant, and this included the subcategories for ATP synthase, cytochrome b6/f complex, photosystem I and II, and phycobilisomes (Table S2A). Operons co-ordinately downregulated in ΔP3 included psaAB, psbCD1, psbEFLJ, apcABC, cpcDC1C2AB and atpIHGFDACBE. The respiratory electron transport terminal oxidases slr1136-slr1138 (ctaC1D1E1) and slr1379-slr1380 (cydAB) were slightly upregulated in L1, possibly due to enhanced respiration based on glucose addition, whereas most of the genes in the thylakoid were downregulated (Schultze et al., 2009).

Pigment biosynthesis genes in the cobalamin, haem, phycobilin and porphyrins group were mostly downregulated and cellular protection genes of the high light-inducible polypeptide Hli proteins for PS were downregulated (Table S2B). General protection responses in day 1 included the upregulation of many heat shock proteins and chaperones, and photoprotection genes of the flv operon (sll0216-sll0219) in day 2 (Table S2C).

Many genes that were affected by oxidative stress after H2O2 addition (Li et al., 2004) were similarly affected in the P3 mutant (Table S2D). This included many downregulated PS and detoxification genes in day 1 and day 2 that affected responses to redox or oxidative stress; e.g., sll5104 (arsenate reductase), peroxidases and glutathiones. The V4R proteins, encoded within the gene cluster slr0144 to slr0151 were mainly downregulated in peroxide stress, in HL and in L1 and day 2 (Table S2D), but upregulated in D1 (see Summerfield and Sherman, 2008).

Specific genes upregulated in peroxide stress and in our study included heat shock proteins and sll0247 (isiA) (Singh et al., 2004). slr1516 (sodB), the peroxiredoxins (tpx) sll0755 and sll0221 (prxQ), thus indicating the presence of high levels of reactive oxygen species (ROS) in the P3 mutant. Correspondingly, genes encoding ROS scavenging molecules (e.g. carotenoids) were activated in day 2 (Table S2D). Numerous biosynthesis genes for riboflavin, pantothenate, quinolinate, folic acid, menaquinone, putative ubiquinone and cobalamine were upregulated in day 2 (Table S1A). Various hyp genes were also upregulated in either the first day or second day, and may be responding to changes in redox in the P3 mutant. The hoxEFUYH operon was downregulated in day 2 along with sll0359, an AbrB-like transcriptional regulator that has been known to activate this operon (Table S2E).

There is a close relationship between iron homeostasis and oxidative stress in cells (Shcolnick et al., 2009). Upregulation of the PerR regulator, slr1738 was seen in D1 and day 2 along with the adjacent and divergent sll1621 (ahpC) gene that encodes a peroxiredoxin (Table S2F). Iron uptake genes are upregulated in oxidative stress, and antioxidant genes are upregulated to repair Fe-S clusters that may be damaged due to ROS (Houot et al., 2007). Genes that encode ferric iron ABC transporters and related proteins, comprised of slr1295, slr0513, slr0327 and sll1878 (futA1, futA2, futB and futC respectively), and sll1404-sll1406 were upregulated mainly in day 1. Similarly, slr1319 (fecB), ssl2250 (bfd) and sll0221 (bcp) were also upregulated in D1 and day 2. Together, these genes may serve to take up more iron for repair or sequester ferric iron (Katoh et al., 2001). On the other hand, slr1894 (mrgA) was downregulated in L1 and day 2. This gene product may be involved with iron storage and mobilization and an mrgA mutant was highly sensitive to peroxide stress (Li et al., 2004; Shcolnick et al., 2009). The downregulated bacterioferritins slr1890 (bfrB) and sll1341 (bfrA), along with the downregulated mrgA and sll0567 (fur) regulator could worsen the oxidative stress due to limited iron storage, scavenging and detoxification.

Excess metals have also been known to cause protein denaturation, oxidative stress and damage PS proteins (Blasi et al., 2012). The C3 operon (termed copMRS by Giner-Lamia et al., 2012) was previously shown to regulate the slr6042-slr6044 operon for copper efflux in the presence of copper. However, the authors could not detect an effect on plastocyanin sll0199 (petE), the complementary cytochrome sll1796 (petJ), or on genes for copper import into the thylakoid, slr1950 (ctaA) and sll1920 (pacS). petE has more copper content, is present in both photosynthetic and respiratory electron transport chains and is also essential in glucose growth whereas petJ has more haem content and operates primarily in the PS electron transport chain (Giner-Lamia et al., 2012 and references therein). In our study, the copper efflux genes slr6042-slr6044 were upregulated in most time points together with pacS and ctaA in day 2 in the P3 mutant (Table S2G). P3 may repress slr6042-slr6044 and petE (upregulated in D1) and be responsible for activating petJ (downregulated in day 1 and day 2). It is possible that these differences are due to genotype differences in our lab strain from that used by Giner-Lamia et al., 2012. We found no significant changes in growth for our operon mutants compared with the WT when grown on CuCl2 at 5 μM (data not shown). Our results suggest a direct effect for P3 in regulating these genes, even though there was no change in copper concentration in this study. Nonetheless, similar conditions may result from the oxidizing environment found in the P3 mutant relative to those caused by copper addition.

In a similar microarray experiment done with the C3 mutant, we found an opposite effect on petE (downregulated in day 1) and petJ (upregulated in day 1 and day 2) and the slr6042-slr6044 operon (downregulated in day 1 and day 2, S. Nagarajan and L.A. Sherman, unpubl. data) compared with the P3 mutant microarray results. Thus, C3 and P3 and their oppositely regulated petE, petJ and copper transport genes may be affected by redox imbalance of the electron transport system and oxidative stress and not just stress caused by copper. This would explain the large number of phenotypes we see for mutants in this system in glucose, high light and in the dark in our lab strains. Furthermore, hik31 was found in both the thylakoid and plasma membranes (Giner-Lamia et al., 2012). The removal of P3 may therefore result in the early downregulation of many PS genes in this study. Another cluster sll1783-sll1785 found to play a role in copper transport previously was downregulated in ΔP3 (Tottey et al., 2008; Singh et al., 2010).

Other metal transporter genes were activated in the P3 mutant and may indicate metal limitations and a response to ROS (Table S2G). Nickel transporters, zinc tranporters, manganese transporters and regulators, and a magnesium transporter were also upregulated in day 2. The sll0797 (nrsR), regulator for nickel transport, may be responding to changes in the plastoquinone redox poise as seen before (Li and Sherman, 2002). A possible cobalt regulon sll0381-sll0384 containing a cobalamine biosynthesis protein cbiM was seen to be downregulated in D1 and day 2 and may serve to explain the defective phenotype for the P3 mutant when grown in CoCl2 salts (Nagarajan et al., 2012). Thus, metal homeostasis for cobalt, nickel, zinc, magnesium and manganese was affected in ΔP3 and could be due to multiple reasons- to compensate for the downregulation of PS genes, due to oxidative stress, or in order to export toxic metals that may accumulate in the cell based on other defective processes. In all, these results suggest a decrease in photosynthesis and pigment production in the P3 mutant and an increase in oxidative stress that may affect iron and metal homeostasis for the cell.

Regulatory, transcription factors, translation and cellular processes genes

Regulatory genes affected in the P3 mutant included many genes shown to be involved with oxidative stress, motility or in the preservation of membrane integrity (Table S3). slr1285 (hik34) was upregulated in day 1 and L13, slr1147 (hik2) in day 2 and slr1783 (rre1) was downregulated in day 2. Genes shown to be regulated by the hik34/hik2-rre1 system that behaved similarly in the P3 mutant are presented in Table S3A (Paithoonrangsarid et al., 2004; Shoumskaya et al., 2005). Hik34 is a negative regulator of heat shock proteins in normal conditions, but a positive regulator under stress conditions, like in ΔP3, as both hik34 and its target chaperones were upregulated in day 1. The hik31 operon genes were all downregulated in the hik34 mutant, but hik34 was upregulated in the P3 mutant, thus suggesting a reciprocal impact of these genes on each other (Suzuki et al., 2005). We conclude that hik34 may activate P3, whereas P3 may repress hik34.

Some regulatory genes known to be affected in different stress conditions were similarly affected in the P3 mutant (Table S3B). These genes were downregulated or upregulated in response to oxidative stress, redox stress or in C, iron limiting and oxidative stress conditions. The slr0947 (rpaB) regulator was downregulated in D1 and D13 and so were the PSI, apc and cpc genes as possible targets (Seino et al., 2009). A putative signalling protein slr6110 on the pSYSX plasmid was also downregulated in both days and may be activated by P3.

Many sigma factors were induced in our study and their known targets were also affected similarly (Table S3C, Osanai et al., 2008; Los et al., 2010). The sll0306 (sigB) gene may regulate other genes in the P3 mutant in day 1 (Table S3D) (Imamura et al., 2003; Singh et al., 2006; Osanai et al., 2008). sigB and sll1689 (sigE) are important for gene regulation in mixotrophic conditions and mutants in both these genes did not grow in 8L/16D (Summerfield and Sherman, 2007). Both genes were downregulated by D13 and ΔP3 could not grow in similar conditions. It is possible that P3 negatively controls the expression of sigB, sigH, sigI, sigG and sigF in glucose as these were all induced in day 1. The genes hik31 and rreC were induced in the dark in the sigB mutant, decreased in the sigD mutant in the light and in the sigBE mutant in both light and dark (Summerfield and Sherman, 2007). Downregulated sigB and sigE could be contributing to the growth defect of the P3 mutant in MT LD from day 2 onwards. Numerous genes in most categories of translation also were downregulated in the P3 mutant and could lead to reduced protein synthesis and increased protein misfolding (Table S3E).

Type 4 pili are important for twitching motility, cell adherence and phototaxis. In ΔP3, some phototaxis genes and regulators were repressed (Table S3F). Downregulated sll1694 (pilA1) and sll1695 (pilA2) may affect chlorophyll biogenesis, assembly and delivery to new photosystems, competence and motility (He and Vermaas, 1999; Yoshihara et al., 2001). Other downregulated genes in day 2 could affect motility, orientation and directional movement towards the light as well as membrane fluidity (Bhaya et al., 2001). Alternatively, other pilins were transiently upregulated in ΔP3 like the pilA4-pilA11 genes and may be a stress response. Many genes in the hexosamine pathway, murein sacculus and peptidoglycan genes were upregulated in ΔP3 and may lead to more peptidoglycan synthesis (Singh et al., 2008, Table S3G). Overall, these results suggest a gradual slowing down of key metabolic processes in the P3 mutant by day 2 with reduced translation, motility and an increase in peptidoglycan synthesis that may worsen cell division.

Ultrastructure of ΔP3 in MT LD and cell division genes

ΔP3 cells were larger in size and showed improper septation after the third day of growth in all conditions. Two doubling cells attached to each other resulted in a clover leaf-shaped tetrad appearance due to division of each daughter cell even before separation from its twin (Nagarajan et al., 2012). Figure 2 shows the P3 mutant cells in comparison to the WT in MT LL and MT LD conditions. Notably, the mutant appeared less abnormal in MT LL and had the same number of carboxysomes as the WT.

Figure 2.

Electron micrographs of the WT (A and C) and ΔP3 (B and D) grown in 5 mM glucose for 4 days in LL (A and B) and 12L/12D (C and D). CG, cyanophycin granules; PHB, polyhydroxybutyrate granules. Magnification was 11 500× to 21 000×.

In contrast, the mutant cells in MT LD demonstrated a dense cytoplasm with fewer and disorganized thylakoids. No lipid bodies and very few glycogen granules were seen. Importantly, no carboxysomes were found in over 200 mutant cells. ΔP3 cells each had many storage organelles including 2–6 polyphosphate bodies, 1–4 cyanophycin granules and a few large PHB granules. In contrast, the WT cells had a normal size and shape, thylakoid arrangement and intracellular organelles like lipid bodies, glycogen granules, 1–3 carboxysomes, 2–3 polyphosphate bodies and no cyanophycin granules.

There have been relatively few studies on cell division in cyanobacteria. Table S3G shows the cell division genes downregulated in day 2 in the P3 mutant. For example, sll1833 (ftsI/pbp4) was downregulated in ΔP3 in D13 and may affect the inward synthesis or incorporation of peptidoglycan in septation allowing the separation of daughter cells. Another gene slr0804 (pbp8) known to be important for complete septation and separation of daughter cells, was downregulated in D1 in ΔP3. A pbp4 mutant had giant cells and both pbp4 and pbp8 mutants had septation defects and clover leaf-shaped cells (Marbouty et al., 2009).

The sll0202 (gidA) and sll0288 (minC) genes were also downregulated in ΔP3 and minC was shown to have an important role in proper cell shape and size (Mazouni et al., 2004). gidA encodes a glucose inhibited division protein and may also contribute to the cell division defect of ΔP3, as seen for a gidA mutant in the rod-shaped strain Salmonella enterica Serovar typhimurim (Shippy et al., 2012). Finally, three ftsH genes associated with the thylakoid membrane were downregulated in day 2, and may have a role in cell division as seen in other bacteria. These downregulated genes could also contribute towards reduced photoprotection of the D1 protein of PSII (Muramatsu and Hihara, 2012). Overall, these downregulated cell division genes may lead to the division defects seen in the mutant.

Nitrogen metabolism and transporters

Table S4 lists selected genes from nitrogen metabolism that were affected in the P3 mutant. Synechocystis can use nitrogen in the preferred form of ammonium; or as nitrate, nitrite and urea; or as amino acids glutamine, glutamate and arginine. After uptake, urea, nitrate and nitrite are converted to ammonium that is then assimilated into carbon skeletons through the GS-GOGAT pathway or using glutamate dehydrogenase and incorporated into organic nitrogen compounds. These processes need sufficient ATP, Mg2+ or Mn2+, NADH, NADPH and ferredoxin for proper functioning (Flores and Herrero, 2005; Muro-Pastor et al., 2005).

The main regulators affecting N metabolism and amino acid homeostasis include NtcA, NtcB, PII, PphA and PamA. In this study, sll1423 (ntcA), slr0395 (ntcB) and sll0985 (pamA) were unchanged, and ssl0707 (glnB) encoding PII and sll1771 (pphA) were downregulated in day 2 (Table S4A). Although the known targets for these genes were not affected in agreement with their mode of function, these regulators could be active in their stable protein forms in different states even if gene expression can vary, as described previously (Forchhammer, 2004).

Nitrogen metabolism is fine-tuned by the careful co-ordination among the different regulators that sense the levels of the different metabolites in a sophisticated manner. In low/limited N conditions (high C to N ratio), the NH4+ or NO3 content inside the cells is lower, the 2OG content is high, PII is phosphorylated and activates NtcA. Other N transport and assimilation genes are then activated by NtcA, and gifA and gifB are repressed. Importantly, the amt1 and amt2 genes, urtC, ureF, nrtCDnarB and glnN all were upregulated in L1 and in some cases in D1 in the P3 mutant, thus indicating N limitation in the initial day after glucose addition (Tables S4B and S4C).

Alternatively, in high/excess N conditions, (low C to N ratio), the 2OG content is low, PII is dephosphorylated by PphA and inhibits NtcA so that other N transporters and conversion genes are not activated by NtcA, and gifA and gifB are de-repressed and inactivate Glutamine Synthase (GS) (Garcia-Dominguez et al., 1999). PamA binds PII, argB is activated and increases arginine synthesis, and cyanophycin is produced. Low C levels are likely to be present in the P3 mutant after D 1 and maybe caused by inhibited CO2 fixation and HT conditions with glucose as seen before (Kurian et al., 2006). The icd gene for 2OG production was downregulated in the dark, possibly due to reduced carbon fixation. A higher N content could be present from D1 onwards due to the upregulation of amt1, amt2, and amt3, urtD and ureE, ureF genes that may bring in and convert N in various forms, upregulated argB for arginine synthesis and gifA and gifB (Table S4B, C and E). The expression of the gif, glnA, glnN, glsF, gltB, gdhA and slr0899 (cynS) genes also suggest ammonium accumulation from D1 in ΔP3 (Herrero et al., 2001). The gif genes were upregulated at D1 and the other genes were downregulated at day 2 time points. Other gene responses that indicated reduced N demand and counter indicated N starvation in day 2 included downregulated nblA1, nblA2, nblB2, sll0783 and sll0784 (merR) (Li and Sherman, 2002) (Table S4D).

The nrtABCDnarB operon genes were downregulated by day 2 (Table S4B). The molybdopterin biosynthesis genes slr0900-slr0903 and ssr1527 associated with narB were upregulated in day 1 in the P3 mutant and downregulated in day 2 like narB (Herrero et al., 2001; Flores and Herrero, 2005). slr0898 (nirA) was downregulated throughout all 4 time points and P3 may be needed to activate this gene for efficient conversion of nitrite to ammonium. Reduced nirA expression could also be a result of decreased nitrate transport or ferredoxin availability (Reyes et al., 1993). Downregulated GOGAT genes in the mutant may reflect the lack of 2OG, ferredoxin and NADH and the presence of NH4+ for conversion of glutamine to glutamate. Many ferredoxin genes were downregulated in the mutant, especially in day 2 (Table S1A). The gene for gdhA was downregulated and may lead to insufficient glutamate production in day 2 (Chavez et al., 1999).

Cyanophycin is a polymer made up of multi-l-arginyl-poly-l-aspartate and can be stored in granules. Cyanophycin synthetase slr2002 (cphA) was upregulated at L1 in the ΔP3 mutant and this could lead to gradual accumulation of cyanophycin. Growth on media containing arginine with or without nitrate leads to round-shaped cyanophycin granules in the WT, largely reduced thylakoids, reduced pigment content and photosynthetic activity (Stephan et al., 2000). This is similar to what is seen for ΔP3 in Fig. 2 in both MT LL (similar to growth on arginine plus nitrate media with few cyanophycin granules and slightly reduced thylakoid content) and MT LD (similar to growth on arginine as a sole N source with many cyanophycin granules and much lower thylakoid content) (Stephan et al., 2000). Arginine is used for protein synthesis in the WT in MT LD, but in the ΔP3 mutant, it may accumulate, and together with aspartate, become stored as cyanophycin. Thus, the behaviour of the N metabolism genes suggests an initial N limitation that is followed by an accumulation of ammonia and cyanophycin and limitation of glutamate.

Amino acid, sulphur, phosphate metabolism and transporters

There are three main pathways for arginine metabolism in Synechocystis containing many different routes and interconnecting side branches with shared genes (Quintero et al., 2001; Schriek et al., 2007; KEGG database). Two major routes include the arginine deiminase and oxidase/dehydrogenase pathways and a minor pathway includes arginine decarboxylase and all were mainly affected in D1 and day 2 (Table S4E). Overall, these gene expression results indicated increased arginine production though upregulated slr1898 (argB), and reduced or incomplete arginine catabolism that could result in decreased conversion to succinate, fumarate and glutamate with possible accumulation of ornithine, citrulline, proline and ammonia in day 2 in the P3 mutant. Similarly, aspartate breakdown pathway branches were either downregulated or incomplete (Table S4F). Aspartate synthesis may be increased [upregulated slr1476 (pyrB) and slr1705] and there could also be reduced production of threonine and lysine and accumulation of pantetheine and oxaloacetate by day 2. Taken together, these gene expression changes indicated that aspartate and arginine accumulated in the P3 mutant to produce cyanophycin.

Other processes affected included synthesis and transport of other amino acids (downregulated genes, Table S4G), as well as sulphate and phosphate metabolism (upregulated genes, Tables S4H and S4I) (Wang et al., 2004; Juntarajumnong et al., 2007). The hik31 operon genes were upregulated after S deprivation and may mediate S acclimation in day 2 in the P3 mutant (Zhang et al., 2008). Additionally, sphR could be negatively controlled by P3 as it was upregulated in day 1 in response to phosphate limitation for activating the pho regulon genes.

Organic carbon metabolism, energy and reducing sources

Table S5 lists the genes affected in central carbon metabolism in ΔP3, including glucose anabolic and catabolic pathways. Most genes in these pathways were downregulated in day 2. Studies to understand the metabolic flux and proteome through these pathways were explored previously for the WT in MT LL, HT DD and LAHG conditions, but a detailed study of MT LD conditions has not been performed as yet (Yang et al., 2002a,b; Tabei et al., 2009). We compared the information in these studies with our microarray experiment to understand the extent of the deviations in ΔP3.

The glucose transporter sll0771 (glcP) was downregulated in L1 and D13 (Table S5C). Glucokinase activity is maximal in heterotrophic conditions (Knowles and Plaxton, 2003). Glucokinase slr0329 (xylR) was upregulated in day 2 and may help phosphorylate glucose in the P3 mutant in LD conditions (Lee et al., 2005). In ΔP3, downregulated PS and the icd gene may lead to lower NADPH. The zwf and gnd genes in the OPP were upregulated for NADPH production (Table S5B). Also, the opcA gene, essential for dark NADPH production, behaved similar to zwf in day 2 in ΔP3. The slr0400 (an NAD kinase) gene was upregulated in day 2 and may be another attempt to increase NAD(P)H for the P3 mutant (Gao and Xu, 2012). The OPP is important to generate ribose-5P as a precursor to nucleotides and nucleic acids, amino acid histidine synthesis and cofactors like folate and riboflavin (Knoop et al., 2010). Accordingly, we found upregulated slr0194 (rpiA) and ssl2153 (rpiB), and several histidine, riboflavin and folate genes in day 2 (Table S1A).

The P3 mutant in MT LD already behaved as if it was experiencing HT DD conditions in day 2, since it had reduced expression for genes in PS (Table S2), CO2 fixation (Table S5J), gluconeogenesis (gap2, fba2, fbp1, Tables S5J and C), and glycogen formation and breakdown (pgm, glgC, malQ and glgP2; Tables S5A and C) that could cause problems in the dark as early as D1 (Kurian et al., 2006). Inorganic carbon fixation to 3PGA could be lower in the mutant leading to a reduced lower half of the glycolysis pathway with downregulated gpmB, eno and the pdh genes in Tables S5C and D (Osanai et al., 2006). Downregulation of glgP2 could affect responses to Ci limitation and downregulation of thioredoxins slr0623 and slr1139 may cause adaptation problems for the mutant in LD and redox transitions (Table S5A, Fu and Xu, 2006). Downregulated gap2 as the major GAPDH is significant to the P3 mutant phenotype in MT LD and also suggests insufficient levels of NADH and NADPH (Table S5J, Koksharova et al., 1998). Furthermore, the icdA gene was downregulated in D1 and D13 and maybe due to reduced carbon fixation. Other TCA cycle genes sucC and fumC were upregulated and citH was downregulated (Table S5E) and may compensate for the reduced production of succinate and fumarate and accumulation of oxaloacetate as suggested by the analysis of arginine and aspartate metabolism genes in Table S4.

The NADH dehydrogenase genes ndhD1, ndhD2, ndhB and ndhF1 genes (Table S5F) were downregulated by D13, and this could affect respiration, cyclic electron flow, ATP production, as well as CO2 uptake in ΔP3 (Ohkawa et al., 2000; Zhang et al., 2004; Ogawa and Mi, 2007; Ma and Mi, 2008). The ndbC gene was upregulated and may help to regulate the redox state of the PQ pool (Howitt et al., 1999). The ATPase operon may be downregulated in the P3 mutant due to incomplete electron transport leading to deficient H+ translocation and cytosolic alkalinization (Lee et al., 2007; Table S2A). Polyhydroxyalkanoate synthesis genes were upregulated in L1 and L13 but downregulated by D13 (Table S5H); and this induction may be responsible for production of PHB granules seen in Fig. 2.

The sugar catabolic genes induced in our study were also induced in nitrogen limitation and belong to the NtcA and SigE regulons (Osanai et al., 2006). Genes induced in glycogen metabolism and the OPP in Table S5 could be due to a combination of N depletion in the mutant in L1 when the OPP is used for reducing power, and an attempt to use the glycogen reserves from PA LL growth for energy. The sll0750 (hik8) gene was upregulated in day 2 and hik8 activated genes pfkB1, gap1, zwf, gnd were upregulated in L13 by > 1.4-fold (Singh and Sherman, 2005). These and additional genes pyk1, tal, glgX, glgP2, cph1 and glnN were downregulated in D13 when sigE was downregulated, suggesting that sigE takes precedence over hik8 in D13 for overlapping regulated genes (Osanai et al., 2005). Genes affected similarly to sll1330 (rre37) that may be regulated by this gene include fba2, fbp1 and gpmB in day 1 (Azuma et al., 2011). Genes pfkB1 and pyk1 thought to be regulated by rre37 were upregulated in L13 when rre37 was downregulated and this could be due to upregulated hik8 and an unidentified regulator for pyk1, normally induced in the dark and in N depletion (Tabei et al., 2007). Thus, there was a hierarchy of the known regulated genes in ΔP3 in day 2 with sigE taking precedence over hik8, and hik8 in turn, over rre37 (Table S5I). Taken together, these results indicate that there is downregulation of organic carbon genes for glucose catabolism by day 2 and an upregulation of glucosylglycerol genes (Table S5G) and pha genes (Table S5H), leading to the production of sucrose and PHB respectively.

Inorganic carbon metabolism and photorespiration

The carbon concentrating mechanism (CCM) comprised of the carboxysome structural proteins, enzymes and inorganic carbon (Ci) transporters, serves to enhance the cell's internal concentration of Ci so that Rubisco is saturated. Otherwise, the oxygenase activity of Rubisco results in accumulation of the toxic metabolite 2PG that inhibits enzymes in the Calvin cycle.

The carboxysome genes, sll1028-sll1032 (ccmK2K1LMN), comprise an operon that was downregulated in ΔP3, in a similar fashion to many PS genes (Table S5J). The negligible carboxysomes and reduced shell structure in the P3 mutant may lead to lower content of enzymes for rubisco (downregulated slr0009-slr0012 rbcLSX) and carbonic anhydrase, and their presence in the cytoplasm rather than being confined to the carboxysome. This may cause the induction of periplasmic carbonic anhydrase slr0051 (ecaB) and slr0436 (ccmO) needed for carboxysyome assembly. ΔP3 also has a strikingly similar phenotype to the ccmM mutant (no carboxysomes) and similar gene expression patterns. Impaired carbon fixation in the ccmM mutant led to excess N over C with time in LC, less 2OG and other amino acids, and less N assimilation through genes in N transport and GS-GOGAT (Hackenberg et al., 2012), similar to our analysis for ΔP3.

In the P3 mutant, high affinity Ci transporters were upregulated, whereas low affinity systems bicA and cupB and the regulator sll1594 (ndhR) were downregulated (Table S5K). In addition, sbtA, a sodium dependent bicarbonate transporter, and sodium transporter genes were upregulated in the dark (Price et al., 2008). The slr1860 (icfG) and sll0776 (spkD) regulatory genes co-ordinate the assimilation of Ci and are needed for growth when external Ci levels are low (Table S5I). icfG was downregulated in day 2 and spkD was upregulated by 1.4-fold in day 1 for adaptation to low Ci levels as seen before (Beuf et al., 1994; Laurent et al., 2008).

The ndhR gene (downregulated; see Table S5K) is a negative regulator that impacts many Ci transporters, slr2006-slr2013 (mrp cluster- putative cation/H+ antiporters), and the sll0217- sll0219 flv operon that together are thought to be activated by low Ci levels. Several genes affected in Ci limitation that play a role in co-ordinating carbon metabolism, outer membrane and cell wall permeability, and stress responses, were consistent with the P3 mutant microarray gene changes and previous results (Wang et al., 2004; Eisenhut et al., 2007), suggesting a Ci limitation for the P3 mutant.

Both cyAbrB-like transcriptional regulators, sll0359 and sll0822 (Kaniya et al., 2013), were downregulated at least twofold in day 2 (Table S5I). Gene sll0822 induces many N assimilation genes, and nrtA, amt1, glnB, urtA and sigE were downregulated in day 2 (Ishii and Hihara, 2008, Table S4B). This gene can also repress Ci transporters sbtA, ndhF3 and cmpA in HC, but these genes were all upregulated in mutant P3 in day 2 even in LC (Lieman-Hurwitz et al., 2009). Recent work has shown that cyAbrB2 (sll0822) is essential for transcription of genes involved with carbon and nitrogen metabolism after a shift to phtomixotrophic conditions (Kaniya et al., 2013). Notably, electron micrographs indicated some of the same ultrastructural changes in the cyAbrB2 mutant as in the P3 mutant (Kaniya et al., 2013), but the P3 mutant changes were more severe. Taken together, the results of our work and those referenced above lead us to suggest that RreP is an important regulator that influences the expression of key C and N metabolic regulators sll0822 and PII.

More photorespiratory flux was found in LC conditions in an earlier study, and we find evidence of this in ΔP3 (Huege et al., 2011, Table S5L). Pathway analysis for photorespiratory cycles in ΔP3 suggested accumulation of oxalate, glycine, 2PG and glycolate; with reduced breakdown of glycine to ammonium and CO2; and reduced serine, and glycerate content. This may reduce 3PGA production from glycerate in the absence of sufficient carbon fixation. Together, these results strongly support a Ci deficiency in ΔP3 and indicate incomplete carbon fixation, despite the upregulation of the high affinity CCM transporters, and the existence of active photorespiratory cycles to detoxify 2PG (Eisenhut et al., 2008).

The effect of pH on the glucose defect of ΔP3

Some mutants have shown growth improvement at certain pH levels in the presence of glucose in Synechocystis (e.g. an aqpZ mutant, Akai et al., 2011). To determine if this is the case for ΔP3, we performed an experiment with buffered media with pH at 7.5, 8, 9 and 10 for WT and ΔP3 in MT LD (Fig. 3). The WT grew best at pH 7.5 and progressively worse with increasing pH. ΔP3 on the other hand, grew better at pH 8 and had increased pigment content. The mutant grew almost 2–3 folds better at pH 8 than at pH 7.5, slightly better than the WT at pH 8, and grew worse at pH 9 and 10. This supports an inorganic carbon (Ci) defect in the ΔP3 mutant in pH 7.5 that is functionally complemented by growth at pH 8. It could be that the availability and transport of HCO3 ions by the CCM genes is ideal for the P3 mutant at pH 8 and enhances growth. An active photosynthetic metabolism, C/N status and redox state of the plastoquinone pool and increasing pH from 7 to 8 were found to activate glutamine synthase (Reyes et al., 1995; Muro-Pastor et al., 2001). The increased inorganic carbon availability at pH 8 would potentially lead to more carbon fixation, 2OG production and an active GS-GOGAT pathway to convert the excess ammonium to glutamate in ΔP3. Thus, the glucose defect of ΔP3 is pH-dependent, as pH 8 helped partially reverse the growth defect and problems with pH homeostasis at pH 7.5.

Figure 3.

Growth of the WT and ΔP3 at the end of 3 days in MT 12L/12D using liquid media of different pH (7.5, 8, 9 and 10 as indicated in the legend). Error bars for cell counts are for triplicate measurements.

Growth with added N sources and metabolites

Upon examining the N metabolism genes in Table S4, C metabolism genes in Table S5, the predicted nitrogen limitation in day 1, and Ci limitation in Fig. 3, we reasoned that growth of ΔP3 may improve on addition of N and C sources. We tested the growth of the mutant and WT in MT LD conditions for 3 days with addition of NH4Cl, NaHCO3, 2OG and sodium glutamate (Fig. 4). 2OG is known to be important for many functions, including ammonia assimilation, nitrogen and carbon metabolic regulation, and many of the added nitrogen compounds lead to production of precursors for pigment synthesis. Figure 4 shows improved growth for the WT in NaHCO3, followed by 2OG, and not much change in growth with NH4Cl and glutamate. The growth for the mutant improved significantly with all of the added compounds with bicarbonate leading to the best growth (4-fold), followed by 2OG (3.5-fold), NH4Cl (3-fold) and glutamate (2-fold). In fact, the mutant cells with bicarbonate grew almost as well as the WT in glucose. Pigment production improved for the mutant with all the added compounds, with ammonium addition resulting in double the pigment content (data not shown). Thus, these results support our analysis of the microarray results for C and N limitation in the P3 mutant and interestingly, reveal that the demand for inorganic C is more than that for N. These results also suggest that the addition of N compounds along with glucose may reduce the initial N limitation, leading to the recovery of growth, and may not lead to the detrimental N accumulation in day 2. The hik31 operon genes were upregulated more than twofold in N limited growth (Aguirre von Wobeser et al., 2011), suggesting a role in regulation of N metabolism.

Figure 4.

Growth of the WT and ΔP3 at the end of 3 days in MT 12L/12D with added nitrogen and carbon sources (NH4Cl, 2OG, NaHCO3 and sodium glutamate as indicated in the legend). Error bars for cell counts are for triplicate measurements.

Discussion

Glucose, the preferred organic carbon compound for Synechocystis, has many effects on cell growth and physiology. Cells need both light and dark incubation periods to use glucose fully as some important catabolic genes and regulatory genes are active only in the dark and others only in the light (Yang et al., 2002b; Tabei et al., 2009). In this study, we investigated the role of the P3 operon during growth in LD cycles in the presence of glucose and examined in detail the effects of glucose on metabolism for the cell. Our results showed that there was an incomplete metabolism of glucose, especially in the dark, that affected the growth and survival properties for ΔP3 (Table 1, Figs S1 and S2). Genes affected in the microarray experiment in LD cycles with glucose in the mutant indicated reduced photosynthesis and increased oxidative stress that affected metal and iron homeostasis; (Table S2); reduced protein synthesis, motility and defective cell division (Table S3); a transition between N limitation in day 1 and excess in day 2 that affected TCA metabolites, sulphur and phosphorus levels (Table S4); and decreased organic carbon metabolism and inorganic carbon limitation (Table S5). The ultrastucture of the mutant revealed reduced thylakoids; lack of carboxysomes; and the storage of polyphosphate, PHB and cyanophycin in ΔP3 (Fig. 2). The results in Figs 3 and 4 indicated that ΔP3 suffered from Ci and N deficiency that was overcome by addition of supplemental C and N compounds.

Physiological changes in day 1 versus day 2

The microarray data in Table 2 and Fig. S4 revealed two main phases in gene responses in day 1 and day 2 in our microarray study. Sixteen groups of around 10 genes or more displayed various transcriptional temporal patterns at different time points in the P3 mutant (Table S6). There were many more genes downregulated in day 2 for the P3 mutant than in day 1. In day 1, PS genes and carbon fixation genes were downregulated (groups 1, 3 and 4), and this was followed by a downregulation of many N assimilation, transport and regulatory genes (groups 1, 5, 8 and 9), and eventually by decreased energy metabolism, amino acid biosynthesis, transcription, translation, cellular processes, and central intermediary metabolism (groups 2, 4, 6, 7 and 10) in day 2. Consequently, there was increased transcription for genes encoding biosynthesis of cofactors, regulatory and transport genes (groups 12–16) in day 1 and day 2. Additionally, genes in the PHB and cyanophycin biosynthesis categories were upregulated (groups 11, 4 and 14). It appears that the mutant is trying to cope by upregulating the flavoprotein and transport and binding proteins for many high affinity transporters, the biosynthesis of cofactors genes, along with regulatory genes to relieve the stress caused by glucose. Upregulated chaperones were an indication of general stress responses. Nevertheless, the upregulated genes do not overcome the defective growth of the mutant. These gene expression patterns explain the reasons behind the growth defect, the defect in the dark, the pigment defect, and the cell division defect for the P3 mutant.

Many genes in PS and respiration were downregulated as early as L1 in the first day and continued to be downregulated in day 2, indicating that this is likely to be an inherent defect of the mutant even in autotrophic growth conditions. The P3 mutant may be defective in both linear electron flow required for bicarbonate transport and cyclic electron flow required for CO2 uptake, and may have lower ATP levels. Some genes in the cell division, chemotaxis, glycogen metabolism, glycolysis, and OPP pathways were mainly downregulated in the second day, thus making this a delayed response of the mutant and a conditional defect that takes place in late log phase of growth. Such a delayed phenotypic defect in glucose conditions has been seen previously with other mutants, like the double hik31 mutant and the pamA mutant (Haimovich-Dayan et al., 2011). The differential transcription results from the second day reflected the changing energy and nutrient status of the mutant as the cells exhausted the pre-existing pools from growth in PA LL before being subjected to MT LD conditions. It has been reported that cells grown in log conditions can better adapt to HT DD, even without light, as their glucose catabolic machineries are still active from PA LL growth (Tabei et al., 2009).

Growth defects and a role in heterotrophic metabolism in the P3 mutant may be generally explained by the large downregulation of amino acid biosynthesis, photosynthesis, energy metabolism, translation and other downregulated gene categories. Many high affinity Ci transporters along with sodium ion transporters were upregulated in the P3 mutant and could be due to downregulated ndhR regulator, sll0822 or the PII protein. Cyanobacterial Ci transporters known to be inactive in darkness were active in the dark in the P3 mutant, and may be depleting the cell of energy reserves quickly. Together with downregulation of the rbc and ccm operons for the carboxysome, these results suggest a severe Ci limitation for the P3 mutant in MT LD conditions and that carbon fixation is of paramount importance to the cell even in the presence of glucose. The Ci limitation in ΔP3 may increase the production of toxic metabolites of the photorespiratory pathways and 2 of the 3 photorespiratory cycles showed signs of being upregulated. The glycolate produced could also be the signal to increase the expression of high affinity CCM transporters by upregulated cmpR (Hackenberg et al., 2012).

Genes affected by the removal of P3

Genes that could be activated by P3 and that are downregulated in the mutant in day 1 and/or beyond in day 2 are highlighted in Fig. 5 with green asterisks and also summarized in Table S7. Genes that were upregulated in ΔP3 are indicated with red asterisks in Fig. 5. The red and green arrows indicate enhanced or reduced processes. In ΔP3, glucose could be phosphorylated by enhanced expression of xylR, converted to 3PGA and channeled into sucrose (upregulated sps, ggt, ggp, glp) as an osmotic stress response, into PHB rather than glycogen, and into amino acid pools ornithine/arginine/citrulline (Huege et al., 2011). Key genes affected in day 1 that could be activated by P3 include rbcLSX, ccmK2K1L, malQ, glgC, gap2, fbp1, fba2, gpmB, pdhA and icd. Genes similarly affected in L1 and/or D1 and in day 2 as well are likely to be due to the removal of P3. Conversely, genes affected oppositely in day 2 are likely to be due to other secondary regulatory effects other than P3. Therefore, we have only selected those genes that were consistently downregulated in day 1 and/or day 2 for Table S7. P3 has a definite role in activating the petJ gene and repressing the slr6042-slr6044 operon for copper efflux and C3 has the opposite impact on these genes. Regulatory genes affected in the P3 mutant and downregulated could be downstream of the signal transduction cascade. These results strongly show the impact of the P3 operon as a potential activator of key metabolic genes in most important central metabolism pathways.

Figure 5.

Pathway map of central metabolism showing genes and processes upregulated (red asterisks, red arrows respectively) and downregulated (green asterisks, green arrows respectively) in ΔP3 compared with the WT in day 1 (L1 and/or D1). The majority of the genes shown were downregulated in the mutant by D13 and this included genes listed in Table S5: glgX, glgP, malQ, glgC, pgm, zwf, pgl, gnd, rpe, talB, tktA, pgi, pfkB1, fbp1, fba1, fba2, gap1, pgk, gpmB, eno, pyk1, ppsA, pdhB, pdhD, icdA, citH, phaA, phaB phaC and phaE.

RreP regulates central metabolism in the dark

In this study, we analysed the growth phenotype, functional complementation of the P3 mutant, and differentially expressed genes to determine the role of the P3 operon in regulating central metabolism in the dark. MT conditions use many metabolic pathways like photosynthesis, nitrogen assimilation, carbon metabolism using both CO2 through the Calvin cycle and glucose through OPP, glycolysis and the TCA cycle so that the constant switching between these pathways caused problems due to downregulated genes in the P3 mutant (Fig. 5, Table S6). The hik31 operon genes were induced in growth with HL (Nagarajan et al., 2012), high salt and sorbitol and in peroxide stress supporting a role for this operon in mediating the redox, electron transport, osmotic and oxidative stress responses (Kanesaki et al., 2002).

Hence, rreP may be involved in the signal transduction of multiple stresses like dark, HL, N limitation, glucose and redox. P3/RreP is an important multifunctional system involved in integrating photosynthetic, N and C metabolism; pH and metal homeostasis; redox and osmotic balance; and cell division in the dark. Therefore, we conclude that an operon on a plasmid in the cyanobacterium Synechocystis regulates a number of key metabolic pathways when cells are grown on an organic carbon source in typical diurnal conditions.

Experimental procedures

Bacterial strains and culture conditions

The Synechocystis WT glucose-tolerant strain and mutant ΔP3 (constructed as described previously in Nagarajan et al., 2012) were grown in triplicate at 30°C in 12 h light 12 h dark (12LD) cycles for 3 days in BG-11 medium buffered with 25 mM HEPES-NaOH (pH 7.5) in 100 ml flasks and shaken at a constant speed of 125 rpm. The starting cell density for these experiments was around 1 × 107 cells ml−1. A light intensity of about 30–40 μE m−2 s−1 was used for all studies. 5 mM glucose was used for both cultures and 25 μg ml−1 kanamycin was added for the ΔP3 mutant. Both absorbance at 730 nm and doubling times from cell counts using a Petroff-Hausser counting chamber were used to measure growth. Pigment content was estimated from whole-cell absorption values and normalized at 800 nm. We compared the peaks for the pigments for chlorophyll a at 436 nm and 686 nm, carotenoids at 480 nm and 530 nm and phycocyanin at 630 nm. Growth on plates for both cultures was carried out by spotting 5 μl of fourfold serial dilutions of cultures on appropriate duplicate unbuffered BG-11 plates with 5 mM glucose and antibiotics that were incubated in LL or 12LD conditions for 8 days. For the viable count growth experiment, cells were grown for 3 days in PA LL, PA LD, MT LL or MT LD conditions in liquid media and then serially diluted and spread- plated on to BG-11 plates, with and without glucose, and incubated in LL for 10 days. The following sugars (5 mM) were added to monitor growth of cells spotted on BG-11 plates to compare and assess the effect of glucose in LL and 12LD – 3 ortho methyl d-glucose (3OMG), fructose, sucrose, mannose and mannitol. Spotting with TCA intermediates was performed in LL and 12LD and incubated for 8 days using BG-11 plates supplemented with 5 mM glucose and 5 mM of each of the following glycolysis or TCA components (sodium salts) in separate plates- pyruvate, acetate, oxaloacetate, succinate, fumarate and malate. Cells were also grown in liquid media for 3 days in LL and 12LD to test the growth with 5 mM glucose and one of the above glycolysis or TCA compounds. Additional liquid growth experiments were carried out with BG-11 supplemented with 5 mM glucose along with 5 mM of one of the following in separate flasks- NH4Cl, 2-OG sodium salt, NaHCO3 and sodium glutamate for 3 days in 12LD. pH sensitivity experiments were carried out in liquid BG-11 media with 5 mM glucose and buffered with HEPES-NaOH pH 7.5 and 8 and CAPS-NaOH pH 9 and 10 and grown for 3 days in LL and 12LD.

Growth conditions for the microarray experiment

Starter WT and ΔP3 cells from plates were grown in liquid BG-11 for 4 days in PA LL conditions. They were re-suspended to about 1 × 107cells ml−1 in 100 ml of BG-11 with 25 mM HEPES-NaOH pH 7.5, then grown for 1.25 days in photoautotrophic conditions till the cell count reached around the mid-log level. 5 mM glucose was then added and the cultures were grown for another 2 days in 12LD cycles. Samples for RNA extraction were taken after adding glucose at 1 h in day 1 in the light (L1) and in the dark (D1), and 1h in day 2 in the light (L13), and in the dark (D13). Two biological replicates and four technical replicates were used per time point per culture including a dye swap. Cells from the four time points were spun down and the pellet was frozen in STET buffer at −80°C. RNA was extracted using Tri-reagent (Ambion), Dnase-treated, and column purified (Zymo research). Labelling of the purified RNA and hybridization was carried out at MOgene, LLC in St Louis, MO. as previously described (Toepel et al., 2008). The arrays used were Agilent 2x11K 60 oligomer arrays. RNA from both biological replicates mixed in equal concentrations for each culture were used. Dye biases were taken into account by using the dye swap replicates for each time point in this dual colour microarray. The slides were scanned and feature extracted using an Agilent scanner. Data analysis was initially done by the Rosetta Luminator software package that included P-values for fold changes of gene expression.

Statistical analysis of microarray data

The differential gene expression analysis was done using open-source software, R (R Development Core Team, 2012) and Bioconductor (Gentleman et al., 2004). For the differential gene expression analysis, microarray data was normalized using the Variance Stabilizing Normalization (VSN) method (Huber et al., 2002) using the Bioconductor package. VSN uses a robust statistical model that controls for any systematic background, experimental, or technical errors that affect microarray data and minimizes the chances of detecting false positives in the differential gene expression analysis later. We used an analysis of variance (anova) for detecting differentially expressed genes between mutant and wild type separately across four time points in the experiment; the baseline treatment effect was the wild type. Specifically, at each time point, we used the moderated two sample t-test implemented in Bioconductor package limma (Smyth, 2004), which is a robust extension of the two sample t-test targeted for microarray data for testing the null hypothesis that there is no difference in a gene's expression between the mutant and wild type. The differentially expressed genes at each time point are determined from the P-values of these hypotheses tests after controlling the false discovery rate (FDR) at1% (Benjamini and Hochberg, 1995). The stringent FDR cut-off bounds the number of genes that are falsely declared as differentially expressed by the statistical analysis. To make the results more reliable and practically significant, we filtered only those differentially expressed genes that had a fold change ≥ 1.5 or ≤ 0.7. The microarray data contained 8635 probes for 3494 genes. We selected a gene only if all its probes satisfied the aforementioned filtration criterion. 2020 genes were differentially expressed in at least 1 time point. For our data analysis of the microarray results, we also included genes changed by 1.3-fold or more and 0.8-fold or less for genes that were in an operon, or in the same pathway for related functions. The transcriptome data from this study was submitted to the ArrayExpress database at the European Bioinformatics Institute (Accession No. E-MTAB-1539).

RT-PCR primers and conditions

RNA was Dnase-treated with DNase I Amplification grade (Invitrogen) for 15 min. The treatment was checked with PCR using standard PCR conditions and the rnpB control primer set. Dnase-treated RNA was used to make cDNA using random primers and Superscript II using the manufacturer's instructions. RT-PCR was carried out at 94°C for 1 min and 20–40 cycles of 94°C for 15 s, 54°C for 30 s and 68–72°C for 60–180 s depending on size of the amplified region and level of gene expression. Primers for testing the expression of the C3 and P3 operon genes and the restriction digests to differentiate between both copies were described previously (Nagarajan et al., 2012). Selected genes from different categories were chosen for microarray data validation with RT-PCR and the primers for these genes and their amplicon sizes are listed below in 5′ to 3′ direction – atpG (310 bp): F – TGATGCCACCCTGCCCCTGAT and R – CGCCGCCGTTTCCTTTTGCC; ccmK1 (176 bp): F – GCGGCCGAGTCACCGTGATT and R – CACCGCCTCGGTGTAGCGAA; psaL (208 bp): F – TGGGCATGGCCCACGGTTACT and R – TGGCTCCAACCATCGGCTGT; psbO (363bp): F – AAGCGCCAAAAGGCTGAGTACGT and R – ATCCGCCGCTGAAGGCAGAG; apcA (418 bp): F – ATGCTGATGCAGAAGCCCGCT and R – GCTTCGGCGGCATCATCGGA; pilT2 (386 bp): F – TCTAGTGACGGGGCCCACCG and R – CCGTCCGCATGGCTGGTTGT; fumC (395 bp): F – GGGGAGCGCAAACCCAACGT and R – CACCGCCGCAATGTGCATCG; cmpA (353 bp): F – AGCCAACTGGGCCTCCGCTA and R – CAGGATCAACGCCTCCGGCC; czcB (127 bp): F – GGTCAGCCAACGGGCCAGTC and R – TGGCGGTGCTCCAGGCAATG. The above genes were tested as indicated for the requisite number of cycles in Fig. 1B.

Electron microscopy

Cells were grown in MTLL and MT 12LD conditions in liquid media for 4 days before being processed for electron microscopy as preciously described (Nagarajan et al., 2012).

Acknowledgements

We would like to acknowledge MOgene, LLC (St Louis, MO) for performing the microarray hybridization and initial analysis, as well as Debra M. Sherman and the Purdue Life Sciences Microscopy Facility for the electron micrographs. S.S. acknowledges financial support for this work from the Statistical Consulting Service, Department of Statistics at Purdue University. This work was supported by a grant from the DOE Genomics: GTL program and in part by a Purdue PRF fellowship to S.N.

Ancillary