Rubisco mutagenesis provides new insight into limitations on photosynthesis and growth in Synechocystis PCC6803

Abstract

Orthophosphate (Pi) stimulates the activation of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) while paradoxically inhibiting its catalysis. Of three Pi-binding sites, the roles of the 5P- and latch sites have been documented, whereas that of the 1P-site remained unclear. Conserved residues at the 1P-site of Rubisco from the cyanobacterium Synechocystis PCC6803 were substituted and the kinetic properties of the enzyme derivatives and effects on cell photosynthesis and growth were examined. While Pi-stimulated Rubisco activation diminished for enzyme mutants T65A/S and G404A, inhibition of catalysis by Pi remained unchanged. Together with previous studies, the results suggest that all three Pi-binding sites are involved in stimulation of Rubisco activation, whereas only the 5P-site is involved in inhibition of catalysis. While all the mutations reduced the catalytic turnover of Rubisco (K_cat) between 6- and 20-fold, the photosynthesis and growth rates under saturating irradiance and inorganic carbon (Ci) concentrations were only reduced 40–50% (in the T65A/S mutants) or not at all (G404A mutant). Analysis of the mutant cells revealed a 3-fold increase in Rubisco content that partially compensated for the reduced K_cat so that the carboxylation rate per chlorophyll was one-third of that in the wild type. Correlation between the kinetic properties of Rubisco and the photosynthetic rate (P_max) under saturating irradiance and Ci concentrations indicate that a >60% reduction in K_cat can be tolerated before P_max in Synechocystsis PCC6803 is affected. These results indicate that the limitation of Rubisco activity on the rate of photosynthesis in Synechocystis is low. Determination of Calvin cycle metabolites revealed that unlike in higher plants, cyanobacterial photosynthesis is constrained by phosphoglycerate reduction probably due to limitation of ATP or NADPH.

Introduction

Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco, EC 4.1.1.39), the main enzyme in nature that assimilates inorganic carbon (Ci) into organic compounds, catalyses the primary reactions of photosynthesis and photorespiration by carboxylation or oxygenation of ribulose-1,5-bisphosphate (RuBP), respectively. Both reactions initiate upon carbamylation of Lys201 at the enzyme large subunit and subsequent stabilization of the carbamate by a magnesium ion, which turns the enzyme catalytically active. Binding of RuBP stimulates closure of the catalytic site and facilitates the conversion of the pentose phosphate into its enediol form. This intermediate reacts either with CO₂ (carboxylation) to form two molecules of 3-phosphoglycerate (PGA), or with O₂ (oxygenation) to form one molecule of PGA and another molecule of 2-phosphoglycolate that enters the photorespiratory pathway. The products of both reactions are liberated as the catalytic site opens (Taylor and Andersson, 1996; Cleland et al., 1998; Duff et al., 2000).

Despite its central role in photosynthetic metabolism, Rubisco catalysis is considered slow and inefficient, as the catalytic turnover and affinity for CO₂ are low (120–720 carboxylations per catalytic site min⁻¹ and 10–300 μM, respectively). Moreover, the carboxylation reaction is competitively inhibited by O₂, and side reactions of the enzyme generate products inhibitory to activation and activity (reviewed by Kellogg and Juliano, 1997). On the basis of biochemical constraints and considering free energy differences in Calvin cycle reactions (Bassham and Krause, 1969; Farquhar et al., 1980; Dietz and Heber, 1984), as well as variations in the rate of photosynthesis, resulting from differential expression of Rubisco using antisense technology (Stitt et al., 1991; Furbank et al., 1996), Rubisco activity has been considered the main limiting factor of photosynthesis under saturating irradiance and limiting CO₂ concentrations.

Rubisco activity is tightly regulated by activation–deactivation and accessibility of substrates, as well as stimulatory and inhibitory effectors (Woodrow and Berry, 1988; Kellogg and Juliano, 1997). Orthophosphate (Pi) is a key regulatory effector of the photosynthetic machinery (Heber et al., 1986; Woodrow and Berry, 1988) and it affects Rubisco in an antagonistic manner. On the one hand, it stimulates Rubisco activation and, on the other hand, it inhibits enzyme activity by competition with RuBP (Heldt et al., 1978; Tabita and Colleti, 1979; McCurry et al., 1981; Parry et al., 1985; Anwaruzzaman et al., 1995; Marcus and Gurevitz, 2000). The crystal structure of Rubisco from Nicotiana tabaccum in complex with Pi reveals three Pi-binding sites: a positively charged pocket at the enzyme surface named the ‘latch’ site that also interacts with the C-terminus of the enzyme large subunit during catalysis to close the catalytic site, and two sites that bind the 1P and 5P of RuBP (Duff et al., 2000). Substitution of residues at the 5P- and latch sites of Rubisco from the cyanobacterium Synechocystis PCC6803 revealed that they play a role in stimulation of Rubisco activation, whereas only the 5P-site is involved in inhibition of the catalytic activity by Pi. These experiments, together with the biphasic kinetics of the Pi-stimulated activation of the enzyme (Anwaruzzaman et al., 1995), led to the suggestion that stimulation of Rubisco activation occurs via a multisite mechanism and that the inhibitory site only partially overlaps with the stimulatory site at the 5P-binding site (Marcus et al., 2005). However, the possibility that the 1P-binding site is involved in regulation of Rubisco activity by Pi has not been examined. In the closed state of the enzyme, phosphate at the 1P-binding site forms hydrogen bonds with the backbone amides of Gly381, Gly403, and Gly404, and with the side chain of Thr65 (Fig. 1). As the catalytic site opens, phosphate interaction with Thr65 is replaced by an interaction with Trp66 (Duff et al., 2000).

Fig. 1.

Diagram of RuBP interactions at the 1P- and 5P-binding sites. In the closed state of Rubisco, phosphate forms hydrogen bonds with Gly381, Gly403, Gly404, and Thr65 at the 1P-binding site (dashed lines), whereas in the open state of the enzyme the interaction with Trp66 replaces the interaction with Thr65 (not shown). Phosphate also forms hydrogen bonds with Arg295 and His327 at the 5P-binding site in the closed state of the enzyme (dashed lines). In the open state of the enzyme, the interaction of phosphate with His327 is replaced by interaction with His298 (not shown). The diagram is based on the crystal structure of non-activated spinach Rubisco in complex with RuBP (PDB code 1RCX), and was produced using RasMol software. The carbon, nitrogen, oxygen, and phosphate atoms are coloured in grey, blue, red, and orange, respectively.

Using a mutant of the cyanobacterium Synechocystis PCC6803, Syn6803Δrbc (Amichay et al., 1993), which enables site-directed mutagenesis of Rubisco and analysis of the effects in its natural photoautotrophic environment (Marcus et al., 2003, 2005), the effect of the substitutions at the 1P-binding site on catalysis and regulation of enzyme activation by Pi were examined. Surprisingly, although some of these substitutions substantially decreased the catalytic turnover of Rubisco, the photosynthetic and growth rates of the mutant cells were only slightly affected. Therefore, the content and kinetic properties of Rubisco in these mutant cells were examined and a search for an additional rate-limiting factor of carbon fixation in Synechocystis in comparison with plants was conducted.

Materials and methods

Growth conditions and photosynthesis assays

Synechocystis PCC6803 was grown on BG-11 medium as was previously described (Marcus et al., 2003). Nuphar lutea plants, collected in the basin of the Yarkon River in the coastal plain of Israel, were grown in an open pond. Net photosynthetic rates were determined using a Clark-type O₂ electrode (Rank Brothers, UK). Relative rates of gross photosynthesis of N. lutea leaves were measured under saturating irradiance (2000 μmol photons m⁻² s⁻¹) and ambient CO₂ concentration using a modulated fluorometer (Diving-PAM, Walz, Germany) as was previously described (Snir et al., 2006). Rubisco was partially purified from cell extracts by two steps of ammonium sulphate fractionation followed by dialysis (cut-off 12000 Da). Enzyme integrity was evaluated electrophoretically on non-denaturing and denaturing polyacrylamide gels followed by immunochemical analysis (Marcus et al., 2003, 2005). Neither dissociation nor proteolytic products of Rubisco were observed in these extracts (data not shown). The molar concentration of Rubisco catalytic sites was determined by incubating the enzyme with 15 μM [¹⁴C]CPBP (carboxypentitol bisphosphate; 66.6×10⁷ Bq mol⁻¹) for 10 min at 30 °C. [¹⁴C]CPBP was synthesized as previously described (Marcus and Gurevitz, 2000). [¹⁴C]CPBP-bound Rubisco was separated from the free ligand by gel filtration using Sephadex G-100. The ratio of CPBP binding to the wild-type and mutant Rubisco was compared with the densitometric ratio of bands of the wild-type and mutant enzymes separated on a native polyacrylamide gel. The similarity of the two ratios, irrespective of the mutant analysed, indicated that determination of the catalytic site number by [¹⁴C]CPBP binding was reliable. The activation and carboxylase assays of Rubisco were performed as described (Marcus and Gurevitz, 2000; Marcus et al., 2005).

Electron microscopy

Synechocystis cells were fixed in 2.5% glutaraldehyde and then in OsO₄, and dehydrated by increasing concentrations of ethanol. The fixed samples were embedded in Glycid ether and stained with uranyl acetate and lead citrate.

Mutagenesis and genetic analysis

Plasmid pSynR4.0, a pBluescript derivative containing the entire rbc operon and flanking sequences from Synechocystis PCC6803 (Amichay et al., 1993), was used for mutagenesis via PCR using three oligonucleotide primers (Supplementary Table S1 available at JXB online). A primer designed with a mutation was reacted with either an upstream or a downstream primer encompassing the restriction sites used for subcloning. Plasmid pSynR4.0 (Amichay et al., 1993) was the DNA template and Taq DNA polymerase was used for amplification. The resulting DNA fragment was used in a second PCR step with a complementary primer encompassing a restriction site. A 580 bp XbaI fragment was used for mutagenesis of Thr65 and a 516 bp NsiI–NruI fragment for mutagenesis of Gly404 (Supplementay Table S1). The final PCR product was cleaved by the appropriate enzymes and ligated into the pSynR4.0 vector that was used to transform the Syn6803Δrbc recipient (Amichay et al., 1993). Transformed cells were inoculated onto solid BG-11 medium at ambient CO₂ concentrations. Air-grown colonies appeared within 14–21 d. PCR analysis and DNA sequencing verified the full segregation of the introduced mutations in the few copies of the cyanobacterial genome.

Determination of the concentrations of Calvin cycle metabolites

Synechocystis cells (30–50 μg chlorophyll ml⁻¹) were incubated in a transparent O₂ electrode chamber under the indicated illumination and Ci concentration at 30 °C until the rate of photosynthesis stabilized. The cells were quickly sucked into a syringe containing 0.35 M perchloric acid that denatured the cell proteins and immediately stopped all metabolic activities. The acid was neutralized with K₂CO₃ and the potassium perchlorate precipitate was discarded after centrifugation. The concentrations of RuBP, PGA, GA3P (glyceraldehyde-3-phosphate), and DHAP (dihydroxyacetone-phosphate) were determined in the supernatant using published enzymatic assays (Bergmeyer, 1974).

Bioinformatic analysis

The evolutionary conservation scores of glycine residues in 400 homologues of Rubisco from Synechococcus (PDB code 1RBL) that were collected from the Swiss Protein Database using the PSI-Blast algorithm (Altschul et al., 1997) were calculated by a Bayesian method using ConSurf software (Landau et al., 2005). The structural model was drawn using RasMol software.

Results

Effects of substitutions at the 1P-binding site on regulation of Rubisco activation by Pi and its catalytic properties

To examine the role of the 1P-binding site in regulation of Rubisco activation and catalysis conserved residues involved in RuBP and Pi binding were substituted (Fig. 1). Whereas no cyanobacterial mutants were recovered for W66A, G381A, and G403A, presumably because of the low activity or instability of the modified enzymes, G404A and T65A/S mutant cells grew photoautotrophically at a rate similar to that of the wild type (data not shown). The main effects of these substitutions were a drastic (85–95%) reduction in the K_cat of Rubisco and a decrease in the K_m(RuBP) of mutants G404A and T65S, whereas the apparent K_m(CO₂) at ambient O₂ concentration for G404A and T65A/S was similar to that of the wild type (Table 1). To analyse the role of the 1P-binding site in Rubisco regulation by Pi, the kinetics of RuBP carboxylation catalysed by the mutant enzymes at various RuBP and Pi concentrations under saturating CO₂ concentration were compared with those of the unmodified enzyme (Table 1). Dixon analysis (Segel, 1975) of the data for RuBP revealed that Pi competitively inhibited the carboxylation reaction of the wild-type and mutant enzymes, with no significant alteration in K_i(Pi) (Table 1). However, while Pi-stimulated activation of the wild-type and T65S enzyme derivatives was similar under suboptimal concentrations of CO₂ and in the presence of MgCl₂ (0.07 mM and 5 mM, respectively), this stimulatory effect was abolished for mutants T65A and G404A (Fig. 2). The difference in Pi effect between T65A and T65S might be attributed to the ability of serine to maintain the hydrogen bond with Pi via the hydroxyl side chain (Morell et al., 1994). Notably, although the stimulation of enzyme activation by Pi was abolished for T65A and G404A, the activation by CO₂ and Mg²⁺ was not impaired for either mutant enzyme (data not shown).

Table 1.

Kinetic properties of Rubisco mutants. The enzyme was activated for 30 min by 280 μM CO₂ and 10 mM MgCl₂ at 30 °C. Aliquots of 10 μl (1.25–1.5 μM activated Rubisco) were added to 290 μl of the reaction mixture containing either various concentrations of ¹⁴CO₂ and 0.5 mM RuBP or various concentrations of RuBP and Pi and 660 μM ¹⁴CO₂ (4 Bq nmol⁻¹), 5 mM dithiothreitol, 10 mM MgCl₂, and 7.5 Wilbur-Anderson units of carbonic anhydrase in 50 mM HEPES, pH 8.0. The reaction was terminated after 2 min by the addition of 200 μl of 6 N acetic acid, and the acid-stable material was counted after 48 h. The catalytic site concentration was determined by [¹⁴C]CPBP binding (see the Materials and methods). Each assay was performed in triplicate. Data were analysed according to Hanes-Woolf or Dixon (Segel, 1975), and the best-fitting lines were calculated using the least-squares method. Correlation coefficients for the regression lines were >0.95.

Strain	Kcat (min−1)	Km(RuBP) (μM)	Km(CO2)a (μM)	Ki(Pi) (mM)
Wild type	857±163	146±24	268±48	8.3±1.1
T65A	40±5.6	193±29	283±57	9.02±1.2
T65S	132±17	47.3±7.6	258±51	NMb
G404A	71.7±13	46.3±8.1	296±56	8.35±0.9
Wild typec	545±95	140±11	181±43	5.8±0.4
H327Qc	92±15	490±12	175±57	0.67±0.07

^aApparent K_m(CO₂), determined at 21% O₂.

^bNM, not measured.

^cResults taken from Marcus et al. (2005).

Fig. 2.

Effect of Pi concentration on Rubisco activation. Wild type and Rubisco mutants G404A, T65A, and T65S were activated for 30 min at 30 °C with 1 mM dithiothreitol, 70 μM CO₂, 5 mM MgCl₂, and various concentrations of potassium phosphate added after a 5 min interval. Aliquots of activated enzyme were added to a reaction mixture in the presence of 0.5 mM RuBP, 70 μM ¹⁴CO₂ (11.3 Bq nmol⁻¹) (for further details see Table 1). Activation fold was defined as the ratio of Rubisco activity activated in the presence or absence of Pi, respectively. Rubisco activity in the absence of Pi was 40, 5.5, 2.56, and 22.5 min⁻¹ for the wild type and G404A, T65A, and T65S Rubisco mutants, respectively.

Effects of substitutions at the 1P-site on photosynthesis:

Despite the marked reductions in K_cat for the Rubisco mutants (6- to 20-fold), the photosynthesis rates of the cells were reduced only 40–50% (T65A/S mutants) or barely at all (G404A mutant) under saturating irradiance and varying Ci concentrations (Fig. 3). In an attempt to explain the difference between these mutant cells, the concentration of Rubisco per chlorophyll, their carboxylation capacity, and the concentrations of four Calvin cycle metabolites (RuBP, PGA, GA3P, and DHAP) were determined.

Fig. 3.

O₂ exchange rates of Synechocystis wild type and G404A, T65A, and T65S Rubisco mutants. Measurements were taken at various irradiances and 2 mM Ci (A) and at various Ci concentrations under 400 μmol photons m⁻² s⁻¹ (B) in 25 mM HEPES buffer, pH 7.0, using a Clark-type O₂ electrode (Rank Brothers, UK). Each assay was performed in triplicate. Standard deviations were <10%.

Rubisco content, determined by binding of radiolabelled [¹⁴C]CPBP, increased 2.6-, 3.5-, and 4.6-fold, respectively, in G404A, T65S, and T65A mutant cells grown under 20 μmol photons m⁻² s⁻¹ at ambient CO₂ concentration (Fig. 4). Electron micrographs of these mutants revealed an increase in the content of carboxysomes (Fig. 5), which encapsulate Rubisco in cyanobacteria (Codd and Marsden, 1984; Marcus et al., 1992). As a result, the carboxylation capacity per chlorophyll at saturating substrate concentrations reached 25, 49, and 30% of that of the wild-type in mutants G404A, T65S, and T65A, respectively (Fig. 4).

Fig. 4.

Rubisco activity, content, and catalytic turnover in Synechocystis wild type and mutants. The cells, grown on BG-11 medium at ambient CO₂ under continuous illumination (20 μmol m⁻² s⁻¹), were spun down and resuspended in 50 mM HEPES buffer, pH 8.0, containing 285 μM CO₂, 10 mM MgCl₂, 1 mM dithiothreitol, and a protease inhibitor cocktail (P2714, Sigma). The cells were broken under 24000 lb inch⁻² using a French press. The crude extract was centrifuged and the supernatant containing the enzyme was activated by incubation for 30 min at 30 °C. The content of Rubisco catalytic sites was determined by incubation of 200 μl of activated enzyme with 15 μM [¹⁴C]CPBP (185 Bq nmol⁻¹). The CPBP-bound enzyme was separated from the free ligand using a 10 cm Sephadex-G100 column. Rubisco activity was determined as described in Table 1 in the presence of 1 mM RuBP and 640 μM ¹⁴CO₂ (1.3 Bq nmol⁻¹) in the reaction mixture. Each assay was performed in triplicate. Standard deviations were <10%.

Fig. 5.

Transmission electron micrographs of Synechocystis cells. Wild type (A) and Rubisco mutants T65A, T65S, and G404A (B–D, respectively). Carboxysomes, designated by white arrows, are distinguished by their grainy composition and polyhedral shape.

Since Synechocystis cells are routinely grown in the laboratory under low irradiance (20 μmol photons m⁻² s⁻¹), it may be assumed that under these conditions the photosynthesis rate is limited to a lesser extent, as was shown for tobacco plants (Stitt et al., 1991). Therefore, the effects of varying irradiance during growth on Rubisco content and cell photosynthesis were examined. It was found that although Rubisco content doubled in the wild-type cells and did not change in the mutant cells grown under 100 μmol photons m⁻² s⁻¹, the photosynthesis rate was not affected by the elevated irradiance (data not shown). These results led to examination of the correlation between the carboxylation capacity and the kinetic properties, K_m(RuBP) and K_cat, of Rubisco variants (Marcus et al., 2003, 2005; Figs 6, 7) and the photosynthesis rate in the cells under saturating irradiance and Ci concentration (P_max).

Fig. 6.

Relations between K_m(RuBP) and K_cat of Rubisco (A) and between the rate of photosynthesis under saturating irradiance and Ci concentrations (P_max) and the K_m(RuBP) (B) and K_cat (C) of Synechocystis Rubisco derivatives in the wild type and C172A, C172A-C192A, R134A, K305E, C399A, G404A, H298A, H327Q, T65A, and T65S Rubisco mutants. The P_max was determined using an O₂ electrode at 400 μmol m⁻² s⁻¹ and 2 mM NaHCO₃, and the kinetic properties of Rubisco were determined as described in Table 1.

Fig. 7.

Relations between Rubisco activity per chlorophyll and the rate of photosynthesis in Nuphar lutea leaves and in the cyanobacterium Synechocystis PCC6803. Rubisco activity and the light-saturated relative rate of gross photosynthesis were determined during the day in Nuphar aerial leaves. The relative rate of gross photosynthesis was determined at ambient CO₂ concentration under 2000 μmol photons m⁻² s⁻¹ using a PAM fluorometer. The light- and Ci-saturated rate of photosynthesis of Synechocystis cells (wild type and C172A, C172A-C192A, R134A, K305E, H298A, H327Q, T65A, and T65S Rubisco mutants) was determined using an O₂ electrode at 400 μmol m⁻² s⁻¹ and 2 mM NaHCO₃. Rubisco activity was determined at saturating RuBP and CO₂ concentrations as described in Fig. 4. The maximal values of each parameter for Nuphar and for wild-type Synechocystis were defined as 100%.

Analysis of correlations between Rubisco kinetic properties and the rate of photosynthesis

In light of the inverse relations between K_cat and K_m(CO₂) of Rubisco from various species (Savir et al., 2010), the relationships between K_cat and K_m(RuBP) of Rubisco were examined in 11 Synechocystis mutants of the present work (Table 1) as well as of previous studies (Marcus et al., 2003, 2005). An optimum curve was obtained (Fig. 6A) in which the maximal K_cat values pertain to K_m(RuBP) values of 140–170 μM, which were also determined in the wild-type enzyme. At a higher or lower affinity for RuBP the catalytic turnover was lower. The optimum curve implies that there is no simple relations between the correlation of P_max and Rubisco K_cat and the correlation of P_max and Rubisco K_m(RuBP) if the relations between K_cat and K_m(RuBP) were reciprocal.

Correlation between K_m(RuBP) and the P_max in these 11 Rubisco mutants revealed that fluctuations in K_m(RuBP) in the range of 50–250 μM had little effect on the P_max. An exception to this was the K_m(RuBP) value for the H327Q Rubisco mutant (490 μM), which could be one of the reasons for its low P_max (Fig. 6B). As additional factors (e.g. Rubisco content and K_cat; Figs 4, 5, 6C; Table 1) could be affected by the substitutions and prominently affect the rate of photosynthesis, the variance in this correlation analysis was high. Since the RuBP concentration limits the P_max (Farquhar et al., 1980), the P_max should increase as K_m(RuBP) decreases. However, the decrease in K_m(RuBP) of Rubisco mutants G404A, T65S, and C172A (Fig. 6B) was not reflected by an increase in P_max probably due to the marked decrease in K_cat, which reduced the carboxylation rate (Fig. 6C; Table 1; Marcus et al., 2003). Consequently, the net effect of these alterations in K_cat and K_m(RuBP) on P_max was modest. This relation between K_cat and K_m(RuBP) would prevent further elevation in Rubisco affinity for RuBP because it may severely decrease the K_cat. Correlation between Rubisco K_cat and P_max of the cells (Fig. 6C) revealed that a decrease in K_cat from 700 min⁻¹ to 200 min⁻¹ essentially did not change the rate of photosynthesis, although a further decrease in K_cat resulted in a prominent reduction in P_max. As Rubisco K_cat decreased while its content increased in the mutants modified at the 1P-binding site (Fig. 4), the correlations between the carboxylation capacity (carboxylation per chlorophyll under saturating RuBP and CO₂ concentrations) and P_max were analysed in extracts of nine Rubisco variants (Marcus et al., 2003, 2005; Figs 3, 4), and compared with similar correlations in higher plants. Whereas the light-saturated rate of photosynthesis at ambient CO₂ concentration in the C₃ plant N. lutea (Fig. 7) or in the C₄ plant Flaveria bidentis (Furbank et al., 1996) decreased along with a reduced carboxylation capacity, up to an ∼66% reduction in the carboxylation capacity of Synechocystis hardly affected the rate of photosynthesis (Fig. 7). Under these conditions, the control coefficient [a quantitative measure of the limitation imposed by a single factor over the flux through the pathway, whose absolute value varies between 0 for a non-limiting to 1 for a factor that totally limits the flux (Kacser and Burns, 1973)] of Rubisco on photosynthesis was zero. However, any further decrease in carboxylation capacity resulted in a steep decline in the rate of Synechocystis photosynthesis (Fig. 7), raising the control coefficient of the enzyme on photosynthesis to 0.6. By interpolation, it seems that the minimal carboxylation capacity required to support Synechocystis photoautotrophic growth is ∼20% that of the wild type (Figs 6, 7).

Analysis of Calvin cycle metabolites

As the limitation imposed by Rubisco on the rate of photosynthesis is low (Figs 6, 7), the possibility was considered that another reaction in the Calvin cycle limits the rate of photosynthesis in Synechocystis. The concentration ratio of products to substrates for a limiting reaction generally differs from that calculated at equilibrium. Hence, large deviations from the equilibrium concentrations of the products and substrates of a given reaction could suggest a limiting reaction (Bassham and Krause, 1969). Therefore, the concentrations of four Calvin cycle metabolites (RuBP, PGA, GA3P, and DHAP, the products and substrates of the reactions catalysed by Rubisco, phosphoglycerate-kinase, glyceradehyde-3-phosphate dehydrogenase, and triose-phosphate isomerase, respectively) were determined, their product to substrate ratios were calculated, and how they would be affected by varying irradiance and CO₂ concentrations was examined.

Under limiting irradiance, the RuBP concentration was high, whereas the concentrations of PGA, GA3P, and DHAP were low (Fig. 8). As the irradiance increased, the RuBP concentration decreased and the concentrations of PGA, GA3P, and DHAP increased. Consequently, the product to substrate concentration ratios of PGA to RuBP and that of GA3P to PGA increased with higher irradiance (Fig. 8). The concentration of these metabolites was also determined for mutant G404A and for a formerly studied mutant, H327Q (Marcus et al., 2005). While Rubisco K_cat and content were similar in the two mutants, K_m(RuBP) was different (46 μM and 490 μM, respectively; Table 1 and Fig. 4; Marcus et al., 2005). As substitution H327Q increased the competitive inhibition of carboxylation by Pi, one would expect an elevated K_m(RuBP) in vivo, depending on the intracellular Pi concentration (Marcus et al., 2005). Indeed, the RuBP concentration in mutant H327Q was higher than that in the wild type at either irradiance, whereas in mutant G404A it was lower under limiting irradiance and equal to that of the wild type under saturating irradiance. In both mutants, the PGA concentration was lower than that of the wild type. Consequently, the highest PGA to RuBP concentration ratio was determined for the wild type, whereas the lowest was determined for mutant H327Q. The GA3P concentration in both mutants was lower than that of the wild type, and the DHAP concentration was higher in mutant G404A compared with the wild type and with mutant H327Q. The GA3P to PGA concentration ratio was similar in the wild type and in both mutants (Fig. 8).

Fig. 8.

Concentrations and concentration ratios of the Calvin cycle metabolites RuBP, PGA, GA3P, and DHAP in Synechocystis wild type and Rubisco mutants H327Q and G404A. Cells were incubated at 30 °C under the indicated irradiance in the presence of 3.75 mM NaHCO₃ until the rate of photosynthesis was stable. Aliquots (30–50 μg chlorophyll ml⁻¹) were rapidly sucked into 0.35 M perchloric acid. After neutralization of the acid with K₂CO₃, the concentration of metabolites was determined. RuBP was converted to PGA, and PGA, GA3P, and DHAP were reduced by NADH-dependent reactions to glycerol-3-phosphate.

The differences in concentrations of RuBP, PGA, GA3P, and DHAP in mutants G404A and H327Q apparently resulted from the large difference in Rubisco affinity for RuBP. The higher affinity for RuBP probably enabled mutant G404A to utilize low RuBP concentrations that exist in these cells at saturating irradiance and Ci concentrations and maintain their wild-type-like rate of photosynthesis (Fig. 3).

Discussion

Role of the 1P-binding site in Rubisco regulation by Pi

Mutational analysis highlighted the role of the 1P-binding site in stimulation of Rubisco activation. As shown in the crystal structure of the Rubisco–Pi complex (Fig. 1), phosphate interacts with Thr65, Trp66, Gly381, Gly403, and Gly404 (Duff et al., 2000). While substitution of Trp66, Gly381, and Gly403 with alanine appeared lethal to cell viability, analysis of mutants T65A/S and G404A demonstrates how the substitutions differentially effect the K_m(RuBP) and K_cat of the enzyme as well as its activation by Pi, but not the competitive Pi inhibition of its activity (Table 1, Fig. 2). These findings along with previous analyses of the 5P- and the latch-binding sites of Pi (Marcus et al., 2005) imply that all three Pi-binding sites play a role in stimulation of Rubisco activation, whereas only the 5P-site is involved in Pi inhibition of Rubisco activity. Whereas at the 5P- and latch-binding sites, phosphate forms hydrogen and electrostatic interactions with the side chains of positively charged residues (histidine, arginine, and lysine), at the 1P-binding site it interacts only via hydrogen bonds with the backbone amides of Gly381, Gly403, and Gly404, and with the side chain hydroxyl of Thr65 (Fig. 1; Taylor and Andersson, 1996; Duff et al., 2000). Although the interaction of phosphate with the backbone amides of glycine residues (Fig. 1) seems non-specific, Gly381, Gly403, and Gly404 are highly conserved, and substitutions of Gly381 and Gly403 with alanine were lethal. Evidently, even a change as minor as the change introduced to Gly404 (e.g. addition of methyl upon substitution with alanine) dramatically alters the kinetic properties of the enzyme (Table 1). Gly381 belongs to loop 7 between the β7 strand and the α7 helix, and Gly403 is located in loop 8 between the β8 strand and the αP helix. However, Gly404 resides at the edge of the αP helix (the numbering of helices and strands follows Knight et al., 1990) and is slightly distant from RuBP, which may explain the lack of lethality of the corresponding mutant cells. Interestingly, this structure resembles a ‘P-loop’—a common motif that has been found in the binding sites of nucleotide phosphates. This element consists of a glycine-rich sequence that connects a β-sheet with an α-helix (Rossman et al., 1974; Kinoshita et al., 1999). Comparison of 400 distinct rbcL sequences (see the Materials and methods) reveals that two-thirds of the glycine residues in the large subunit of Rubisco appear at loops and most of them are conserved (data not shown). Thus, the substitution of glycine may affect the loop structure or dynamics during catalysis. Alternatively, the side chain of alanine might sterically hinder RuBP binding.

The limiting factors for photosynthesis in Synechocystis

Since the discovery of the Calvin cycle in the middle of the 20^th century, extensive physiological, metabolic, and genetic studies have shown that Rubisco activity is the main limiting factor for photosynthesis under saturating irradiance and limiting Ci concentrations in C₃ (Farquhar et al., 1980; Dietz and Heber, 1984; Stitt et al., 1991; Snir et al., 2006) and C₄ (Furbank et al., 1996) plants, as well as in CO₂-concentrating green algae (Bassham and Krause, 1969). In contrast, in the G404A Synechocystis PCC6803 mutant, the 90% decrease in Rubisco K_cat hardly affected its photosynthesis and growth rates (Figs 3–6; Table 1). Similar to plants that respond to reduced Rubisco activity by elevating its activation status (Stitt et al 1991), it was found that Synechocystis PCC6803 cells increased their Rubisco content by as much as 4.6-fold (Fig. 4). Despite this increase, the carboxylation capacity was not returned to the wild-type level. In light of the minor effect of substitution G404A on cell growth and especially on P_max (Fig. 3), Rubisco activity does not seem to be the limiting factor for cyanobacterial photosynthesis under saturating irradiance and limiting Ci concentrations. In contrast to the G404A mutant, the rate of photosynthesis in the T65A/S mutants was modestly impeded under saturating Ci and irradiance (Fig. 3), suggesting that limitations such as RuBP regeneration, rather than Rubisco activity, were limiting photosynthesis.

To examine the extent of Rubisco limitation on the rate of photosynthesis, P_max was correlated with the carboxylation capacity of the cells, and with K_cat as well as with K_m(RuBP) of Rubisco, and the concentration ratio of PGA to RuBP, the product and substrate of the carboxylation reaction, was determined. On the one hand, the correlation between P_max and carboxylation capacity revealed that unlike in C₃ and C₄ higher plants (Fig 7; Stitt et al., 1991; Furbank et al., 1996), the rate of photosynthesis in Synechocystis PCC6803 is insensitive to large variations in Rubisco K_cat and K_m(RuBP) and in the carboxylation capacity of the cells (Figs 6, 7). Thus, the limitation imposed by Rubisco on the rate of photosynthesis is low. These results are in accordance with the findings of Daniell et al. (1989), who showed that overexpression of Rubisco did not influence the growth rate of the cyanobacterium Synechococcus, but are in conflict with those of Iwaki et al. (2006), who claimed that expression of foreign type 1 Rubisco stimulated photosynthesis in Synechococcus. On the other hand, the PGA to RuBP concentration ratio (Fig. 8) was affected by the irradiance and Ci concentration, and by mutations in the enzyme that reduced the carboxylation capacity. These effects demonstrated that the carboxylase reaction was not in equilibrium as shown for higher plants (Dietz and Heber, 1984) and algae (Bassham and Krause, 1969). Likewise, large alterations in the concentrations of fructose bisphosphatase and phosphoribulose kinase, which catalyse highly regulated non-equilibrium reactions in the Calvin cycle, had a negligible effect on the rate of photosynthesis of higher plants (for a review, see Raines, 2003).

The conflicting findings regarding the limitation imposed by Rubisco on the rate of photosynthesis can be rationalized should other reactions impose a greater limitation on the rate of photosynthesis. For that reason, the limitation imposed by PGA reduction and triose-phosphate isomerization on the rate of photosynthesis was examined by determining their substrate to product concentration ratios (PGA, GA3P, and DHAP). PGA formed by the carboxylation reaction of Rubisco is reduced to GA3P using NADPH and ATP produced by the light reactions in two subsequent steps catalysed by phosphoglycerate kinase (PGK) and glyceraldehyde phosphate dehydrogenase (GAPDH; Equation 1). Triose-phosphate isomerase converts GA3P to DHAP. The concentrations of triose-phosphates (GA3P and DHAP) increased in Synechocystis with elevation in irradiance, but they differed from the concentrations at equilibrium (Fig. 8), whereas in higher plants these metabolites are in equilibrium (Bassham and Krause, 1969). This could result from accumulation of phosphoglycolate or Calvin cycle metabolites (e.g. RuBP or PGA) that inhibit the activity of triose-phosphate isomerase (Leegood, 1990).

The equilibrium equation of PGA reduction implies that the GA3P to PGA concentration ratio equals the product of the mass–action ratio for ATP production and for NADPH production multiplied by the equilibrium constant (K_eq) as described in Equation 2 (Arnon et al., 1958; Heber et al., 1986).

1

2

The GA3P to PGA concentration ratio increased with irradiance in Synechocystis cells up to 500 μmol photons m⁻² s¹ (Fig. 8), whereas in higher plants it is constant or decreases (Heber et al., 1986). Assuming that the cytoplasmic pH is constant in the illuminated cells, photosynthesis in cyanobacteria, unlike in higher plants, is limited by ATP, NADPH, or by both under limiting irradiance. In contrast to higher plants, cyanobacteria concentrate CO₂ at the carboxylation site by uptake of CO₂ and HCO₃^– or CO₃^2– driven by ATP, ion gradients, and photosynthetic electron transport (Kaplan et al., 1987; Marcus et al., 1992; Badger et al., 2006). The excess energy required for Ci transport may decrease the pools of ATP and NADPH, and hence PGA reduction is limited. Conversely, competition between PGA reduction and Ci transport on limited resources of ATP and NADPH may limit the Ci uptake and, as a result, also the rate of photosynthesis. Indeed, it has been shown that high CO₂-grown Anabaena varibilis cells are limited by Ci uptake. However, elevation of the Ci uptake capacity in air-grown cells lifted this limitation (Kaplan et al., 1980). As the GA3P to PGA concentration ratio is stabilized at high irradiance (Fig. 8), it remains unclear whether under saturating irradiance ATP or NADPH limit PGA reduction or Ci uptake and consequently the rate of photosynthesis.

Conclusions

Mutagenic analysis of the phosphate-binding sites in Synechocystis Rubisco revealed that: (i) all three phosphate-binding sites play a role in Pi-induced stimulation of Rubisco activation, whereas only the 5P-binding site is involved in Pi inhibition of enzyme activity; (ii) under saturating irradiance and limiting Ci concentrations, Rubisco is not the main rate-limiting factor for cyanobacterial photosynthesis, as up to 90% reduction in its K_cat or two-thirds of the carboxylation capacity of the cells hardly affected the carbon assimilation and growth rates; and (iii) PGA reduction limits the light-limited rate of photosynthesis due to ATP or NADPH limitation as revealed by the analysis of Calvin cycle metabolites.

Supplementary data

Supplementary data are available at JXB online.