Immunoelectron Microscopy for Locating Calvin Cycle Enzymes in the Thylakoids of Synechocystis 6803

Abstract

Unicellular cyanobacteria Synechocystis 6803 were fixed using high-pressure freezing (HPF) and freeze substitution without any chemical cross-linkers. Immunoelectron microscopy of these cells showed that five sequential enzymes of the Calvin cycle (phosphoriboisomerase, phosphoribulokinase, ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO), 3-phosphoglyceratekinase and glyceraldehyde-3-phosphate dehydrogenase) and the catalytic portion of the chloroplast H⁺-ATP synthase (CF1) are located adjacent to the thylakoid membranes. Cell-free extracts of Synechocystis were processed by ultracentrifugation to isolate thylakoid fractions sedimenting at 40 000, 90 000, and 150 000 g. Among these, the 150 000-g fraction showed the highest linked activity of the above five sequential Calvin cycle enzymes and also the highest coordinated activity of light and dark reactions as assessed by ribose-5-phosphate (R-5-P) +ADP dependent CO₂ fixation. Immunogold labeling of this membrane fraction confirmed the presence of the above five enzymes as well as the catalytic portion of the CF₁ ATP synthase. Notably, the protein A-gold labeling of the thylakoids was observed without use of chemical cross-linkers and in spite of the normal washing steps used during standard immunolabeling. The results showed that soluble Calvin cycle enzymes might be organized along the thylakoid membranes.

INTRODUCTION

Though the conventional perception of metabolism in biology relies on the concept that the cells are bags, full of freely diffusing molecules and precise chemical reactions related to life occur by a chance encounter of these molecules; it is known that the intracellular environment is crowded with proteins and has limited solvation capacity. Such milieu in vivo would necessitate that the sequential enzymes of a metabolic pathway should be organized to facilitate the delivery of the substrates at proper sites in the cells, ensuring an efficiency of the metabolic processes (Srere, 1987). Involvement of membranes was also envisaged in such organization. However, using the prevailing technologies, it has been difficult to prove such an organization and metabolic significance of such organization was hotly debated (Ovadi, 1991). As rightly pointed out by Ellis (2001), ‘although macromolecular crowding was obvious, it was not appreciated in the biochemical research’. The recent high throughput technologies in genomics and proteomics have resulted in a large amount of information about several genes, proteins, and enzymes in a variety of living systems. Concurrently, the limitations of reductionism in understanding complex biological processes are being realized. This has rejuvenated interest in understanding the molecular architecture of living organisms. The concept of biochemical networks, where a protein is an element in the network of interactions, is now getting recognition (Monk, 2003; Han et al., 2004). Several tools are being developed to map the interactomes among different metabolic pathways in silico.

Photosynthesis, the process of conversion of solar energy into chemical energy, requires coordinated activities of many protein molecules. The pigment protein complexes participating in electron transport of light reactions of photosynthesis are known to be organized as superstructures in thylakoid membranes. The enzymes of the photosynthetic carbon reduction cycle are water-soluble and considered to be randomly distributed in the stroma of the chloroplast or the cytosol of other photosynthetic bacteria. Nevertheless, the activities of the light and the dark reactions are interdependent and are finely regulated at various levels. The protein concentration in the stroma is around 400 mg ml⁻¹ (w/w) (Paech, 1986; Ellis, 1979). The accuracy in photosynthetic carbon assimilation will be unfeasible, unless there is an order in the arrangement among the components of light and dark reactions.

Calvin cycle enzymes are known to have a tendency to form multi-enzyme complexes (Sainis and Harris, 1986, Sainis et al., 1989; Persson and Johansson, 1989; Anderson et al., 1995). A ∼530-kDa functional five-enzyme complex containing ribose-5-phosphate isomerase (RPI), ribulose-5-phosphate kinase (PRK), ribulose-1,5-bisphosphate carboxylase/oxygenase, phosphoglycerate kinase (PGK), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and a 65-kDa polypeptide was isolated by Gontero et al. (1988) from spinach chloroplast. A similar stable CO₂ fixing complex consisting of RPI, PRK, ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO), GAPDH, sedoheptulose-1,7-bisphosphatase (SBPase), and ferredoxin NADP reductase (FNR) was also reported by Süss et al. (1993). Recently, nearest neighbor analysis experiments were used to study the co-localization of several Calvin cycle enzymes in the stroma of pea chloroplasts using immunoelectron microscopy. Using this technique, in vivo associations were predicted among several soluble enzymes such as carbonic anhydrase, PRK, PGK, and RuBisCO (Anderson and Carol, 2004); GAPDH, triose-P-isomerase, aldolase and SBPase (Anderson et al., 2005); transketolase, xylulose-5-P-3-epimerase and RPI (Anderson et al., 2006); NADP, FNR, and GAPDH (Negi et al., 2008). The existence of enzymes in supercomplexes was shown to have advantages such as higher catalytic efficiency, less susceptibility to auto oxidation and proteolysis in comparison to free enzymes (Gontero et al., 1988, 1993). However, all these studies dealt with the associations among Calvin cycle enzymes in the soluble fraction of chloroplasts. Membrane association of Calvin cycle enzymes was rarely investigated. There were some reports on membrane-bound particles active in the dark reactions of photophosphorylation (Howell and Moudrianakis, 1967) and on dual localization of some Calvin cycle enzymes (Raghavendra et al., 1981; Mori et al., 1984; Hermonso et al., 1992; Wolosiuk et al., 1993). The possibility of the involvement of thylakoid membranes in organization of Calvin cycle enzymes was suggested by Süss et al. (1993, 1995), Sainis et al. (2003), and Dani and Sainis (2005). Babadzhanova et al. (2002) detected free and membrane-bound multi-enzyme complexes of Calvin cycle in cotton leaves. Recent studies have shown that a small-molecular-weight protein CP12 is involved in a multi-enzyme complex of PRK, GAPDH, and aldolase (Erales et al., 2008). Thioredoxin and electron transport chain have been shown to play important role in the dissociation of this complex (Howard et al., 2008).

The present investigations focus mainly on the immunolocalization of Calvin cycle enzymes in the cells of unicellular cyanobacteria Synechocystis 6803 as well as in the isolated thylakoid fractions of this organism.

RESULTS

Immunolabeling of Cells of Synechocystis 6803

Synechocystis cells in log phase were subjected to HPF and freeze-substitution without using any chemical cross-linkers as described in Methods. Figure 1A and 1B show the ultrastructure of cryo fixed Synechocystis cells. It is worth noting that the ultrastructure could be preserved even without the use of any chemical cross-linkers during freeze-substitution. The cells showed cell wall, typical thylakoids in the periphery, carboxysomes and cytosol (Figure 1).

Figure 1.

Electron Micrographs of Synechocystis Cells after High-Pressure Freezing.

The log phase culture of Synechocystis 6803 was fixed through HPF and embedded in LR white.

(A) Entire cells.

(B) Portion of cell enlarged.

Cx, carboxysome; T, thylakoid membrane. Space bars = 400 nm.

Sections of cells of Synechocystis after HPF, freeze-substitution, and resin embedding were used for immunolabeling with antibodies against the catalytic portion of the chloroplast H⁺-ATP synthase (CF₁) and five Calvin cycle enzymes viz. RPI, PRK, RuBisCO, PGK, and GAPDH. Among these, PRK and PGK use ATP whereas GAPDH requires NADPH produced during assimilation of light energy by thylakoids. The conjugates were detected under electron microscope using protein A conjugated with 5 or 10 nm gold particles. Specificity of antibodies was checked on Western blots (Supplemental Figure 1). Figure 2A shows immunolabeling with antibodies against CF₁. CF₁ is known to be present in the thylakoids. Obviously, immunogold label for CF₁ was observed predominantly in the vicinity of thylakoid membranes. For a given protein, the intracellular distribution over 20 different cells was quantitated by counting the gold particles associated with different structures in the cells. In case of CF₁, around 92% of gold labeling was observed in the periphery of the membranes, about 7% were in the cytosol, and 1% appeared in the carboxysomes (Table 1). The differential labeling not only showed the intracellular distribution of CF₁, but also indicated the specificity of labeling with the antibodies. When immunolabeling was carried out with antibodies against five sequential Calvin cycle enzymes, a substantial amount of label was seen along the thylakoid membranes (Figure 2B–2F). For RuBisCO, about 47% gold particles were observed in the thylakoid fractions and 53% in the cytosol, of which 36% were in the carboxysomes. For other Calvin cycle enzymes like PGK and GAPDH, about 66%; and for PRK as well as RPI, over 72% label was seen around the thylakoids (Table 1). The percentage of membrane-associated fraction would increase to 79% for RPI and PRK, 51% for RuBisCO, 72% for PGK and GAPDH, if it were normalized to membrane labeling of CF₁. The percentage of the gold label for other Calvin cycle enzymes in the carboxysomes was 5.9 ± 0.8% for RPI, 5.0 ± 0.8% for PRK, 3.2 ± 1.2% for PGK, and 4.0 ± 1.2% for GAPDH.

Figure 2.

Immunogold Labeling of HPF Fixed Cells.

Immunolabeling was carried out with polyclonal antibodies raised in rabbit.

(A) anti-CF₁.

(B) anti-RPI.

(C) anti-PRK.

(D) anti-RuBisCO.

(E) anti-PGK.

(F) anti-GAPDH.

10 nm protein A-gold is shown for immunolabeling of CF₁, RPI, RuBisCO, and PGK, 5 nm protein A-gold for PRK and GAPDH. Space bars = 200 nm.

Table 1.

Frequency of Immunolabeling and In-Situ Localization of Enzymes in Synechocystis.

	Percent of protein A-gold particles
Protein localized	Thylakoid membranes percent ± SE	Cytosol percent ± SE	Carboxysomes percent ± SE
CF1	91.8 ± 1.6 (100%)	7.1 ± 1.6	1.1 ± 0.7
RPI	72.8 ± 2.1 (79.3%)	21.3 ± 1.9	5.9 ± 0.8
PRK	72.2 ± 1.5 (78.6%)	22.9 ± 1.7	5.0 ± 0.8
RuBisCO	46.8 ± 2.8 (50.9%)	16.9 ± 2.3	36.3 ± 3.3
PGK	66.1 ± 2.9 (72%)	30.7 ± 2.9	3.2 ± 1.2
GAPDH	66.1 ± 1.6 (72%)	29.9 ± 1.9	4.0 ± 1.2

70-nm thick sections were taken on formvar-coated copper grids and immunolabeled with respective antibodies. Gold particles on thin sections of 20 different cells were counted to calculate the distribution of CF₁ and Calvin cycle enzymes. The values in parenthesis indicate gold labeling percentage normalized to labeling for CF₁. All the values represent mean ± SE (n = 20).

We also carried out immunolocalization after chemical fixation and dehydration of cells using progressive lowering of temperature (data not given). Notably, significantly less immunogold labeling could be detected in such samples, indicating a superiority of the HPF method in retaining the antigenicity of the epitopes as well as the specificity of immunogenic reaction. No significant labeling was observed if pre-immune serum was used for immunolabeling (data not given).

Biochemical Characterization of Isolated Thylakoids of Synechocystis 6803

These associations were investigated further using isolated thylakoid membranes of Synechocystis. We had earlier used ultracentrifugation to isolate the different thylakoid fractions from cell-free extracts of Anacystis nidulans BD-1, a local isolate of Synechococcus spp. (Dani and Sainis, 2005). The cell-free extracts of Synechocystis 6803 were also subjected to similar sequential ultracentrifugation at 40 000, 90 000, and 150 000 g for 1 h each and the pellets designated as 40, 90, and 150-k were used for biochemical analysis as described in Methods. All the thylakoid fractions were analyzed for the presence of chlorophyll a, phycobilisomes, and Calvin cycle enzymes. Table 2 shows that the 40-k fraction had the highest chlorophyll (66%) followed by the 90-k fraction (23%). The 150-k fraction contained around 11% chlorophyll. About 54% of phycocyanins were present in the 40-k fraction and nearly 18 and 28% of total were present in the 90 and 150-k fractions, respectively (Table 2).

Table 2.

Linked Activities of Calvin Cycle Enzymes and Pigment Distribution among the Three Thylakoid Fractions.

Fraction	R-5-P+ADP+Pi-dependent CO2 fixation activity		R-5-P-dependent GAPDH activity		Phycocyanin		Chlorophyll
	Activity	Percent of total	Activity	Percent of total	mg/mg Chl	Percent of total	Percent of total
40-k	0.3 ± 0.1	16.0 ± 1.0	0.2 ± 0.12	21.5 ± 16.2	0.8 ± 0.2	54.3 ± 3.7	66.1 ± 6.0
90-k	2.5 ± 1.2	27.3 ± 3.7	0.99 ± 0.5	22.9 ± 13.5	1.3 ± 0.1	17.8 ± 3.1	23.3 ± 3.2
150-k	13.3 ± 5.6	57 ± 2.6	5.4 ± 1.2	55.5 ± 22.7	5.9 ± 0.7	27.5 ± 2.4	10.5 ± 3.0

R-5-P+ADP+P_i-dependent CO₂ fixation activity: sequential activities involved in this assay are Electron transport chain, CF₁, RPI, PRK, and RuBisCO. The linked activity is expressed as μmoles of CO₂ fixed, mg chlorophyll⁻¹, h⁻¹. The controls without ADP and DTT had specific activities in the range of 0.04 to 0.06 μmoles of CO₂ fixed, mg chlorophyll⁻¹, h⁻¹ for all the fractions.

R-5-P-dependent GAPDH activity: the sequential activities involved in this assay are RPI, PRK, RuBisCO, PGK, and GAPDH. The linked activity is expressed as μmoles of NADPH⁺ oxidized, mg chlorophyll⁻¹, h⁻¹. The NADPH reduction in absence of R-5-P was used as blank.

All the values represent mean ± SE (n = 3).

We measured R-5-P+ADP+P_i-dependent CO₂ fixation activity in the three thylakoid fractions (Table 2). This assay involves the electron transport in the presence of methyl viologen as an electron acceptor and light-dependent synthesis of ATP by CF₁. R-5-P is converted to Ru-5-P by RPI. The ATP produced in the light is used for conversion of Ru-5-P to RuBP by PRK. This RuBP and ¹⁴CO₂ are used by RuBisCO to form 3-PGA as explained in Supplemental Figure 2A. All the three thylakoid fractions showed R-5-P+ADP+P_i-dependent CO₂ fixation activity (Table 2). The 150-k fraction showed the highest percentage of total activity (57%) and also the highest activity based on chlorophyll as compared to the 40 and 90-k fractions. There was negligible CO₂ fixation activity in the reaction mixtures without ADP, indicating that there is no residual ATP. Additionally, in the absence of DTT, the rate of CO₂ fixation activity was very low, indicating that PRK was involved in the reaction. Thus, R-5-P+ADP+P_i-dependent CO₂ fixation activity confirmed the presence of fully functional components of the photosynthetic electron transport and the presence of at least three enzymes viz. RPI, PRK, and RuBisCO in the membrane fraction.

The presence of the five sequential enzymes viz. RPI, PRK, RuBisCO, PGK, and GAPDH was examined by using linked assays viz. R-5-P+ATP-dependent GAPDH activity, which involved sequential activities of the five Calvin cycle enzymes. The sequential enzymes functioning successively in this reaction are RPI, PRK, RuBisCO, PGK, and GAPDH. Each enzyme uses its substrate to form a product, which is used by the subsequent enzyme in the pathway as shown in Supplemental Figure 2B. Table 2 shows that 40, 90, and 150-k fraction had 22, 23, and 55% of the total R-5-P+ATP-dependent GAPDH activity, respectively, found in the thylakoid fraction. In this case also, the 150-k fraction had the highest percentage of total activity and also had higher specific activity on chlorophyll basis. The amount of RuBisCO and GAPDH in the three fractions and the soluble fraction was estimated using dot blots. It was observed that about 20–30% of these Calvin cycle enzymes could be detected with thylakoids (data not given).

Thus, both of these linked activities were highest on chlorophyll basis and as a percentage of activity in the 150-k fraction as compared to the 40 and 90-k fractions (Table 2). The 150-k fraction also had the highest amount of phycocyanin per mg chlorophyll (Table 2).

Immunolabeling of Isolated Thylakoids of Synechocystis 6803

The existence of Calvin cycle enzymes in the thylakoid membrane fraction was further substantiated by immunoelectron microscopy. Immunogold labeling of the 150-k thylakoid fraction was done with antibodies against the catalytic portion of CF₁ ATP synthase, RPI, PRK, RuBisCO, PGK, and GAPDH detected by protein A conjugated to 5-nm gold particles (Figure 3A–3F). The thylakoids were absorbed on a formvar-coated grid and immunolabeled with primary and secondary antibodies and stained negatively with uranyl acetate as given in Methods. Considerable gold labeling was detected around the membranes for all the proteins except for the controls, where 5 nm protein A-gold was used without any specific primary antibody treatment. This indicated that labeling observed with antibodies for Calvin cycle enzymes was specific. Substantial labeling with anti-CF₁ antibody confirmed that the 150-k fraction had the thylakoid components, which were earlier shown by the functional assay of photophosphorylation-dependent CO₂ fixation. When the numbers of gold particles were counted per μm² over 10 random fields, the 150-k thylakoid fraction showed significant labeling with antibodies of Calvin cycle enzymes indicating the presence of these soluble enzymes in this membrane fraction. In the case of GAPDH, the label appeared to be located on the side of thylakoid membranes in comparison to immunolabeling of thylakoids with other Calvin cycle enzymes. Since Calvin cycle enzymes are expected to be present as extrinsic proteins on membranes, we also performed immunolocalization in 150-k membranes that were washed with high salt (1 M KCl in buffer). In the case of RPI, over 80% of gold particles were washed, whereas, for GAPDH and PGK, 60% gold particles were washed. In the case of RuBisCO and PRK, only 15% of gold particles were washed. The results showed that some of the Calvin cycle enzymes could not be dissociated from membranes, even after high salt wash.

Figure 3.

Immunogold Labeling of the 150-k Thylakoid Membrane Fraction.

Immunolabeling was carried out with polyclonal antibodies raised in rabbit.

(A) anti-CF₁.

(B) anti-RPI.

(C) anti-PRK.

(D) anti-RuBisCO.

(E) anti-GAPDH.

(F) anti-PGK.

Labeling done with 5 nm protein A-gold without primary antibody served as control. Arrows indicate the gold particles. Bar = 100 nm. Number of gold particles μm⁻² were counted at 10 random locations. The average number of gold particles were 331.3 ± 38.8 for CF₁; 325.6 ± 39.6 for RPI; 359.3 ± 18.2 for PRK; 354.9 ± 53.0 for RuBisCO; 366.2 ± 14.3 for PGK; and 282.9 ± 44.2 for GAPDH. The frequency of immunolabeling when no primary antibody was used was 95.7 ± 21.7 for protein A-gold particles of 5 nm in this experiment.

The proteins in the three thylakoid fractions and cytosol were resolved using SDS–PAGE (Supplemental Figure 3A). The three thylakoid fractions showed heterogeneity in polypeptides detected on SDS–PAGE. There were proteins that were specific to each fraction. The analysis of the supernatant fraction showed that there were several proteins (marked by arrows) that were specifically detected only in this fraction. The Western blots for the Calvin cycle enzymes on the three thylakoid fractions also revealed that Calvin cycle enzymes can be detected mainly in 90 and 150-k fractions and soluble fraction (Supplemental Figure 3B). We also subjected the 150-k fraction and the soluble cytosolic fraction to preliminary ESI–LC–MS analysis. There were several peptides showing presence of Calvin cycle and other soluble proteins in the 150-k fraction (data not shown). Interestingly, the data in Supplemental Table 1 show that the peptides like acyl carrier protein, rehydrin, hypothetical protein sll1785, slr1852, sll1762, slr1513, peptidyl-prolyl cis-trans isomerase, cyanoglobin, negative aliphatic amidase regulator, ribosome releasing factor, heat shock protein, GrpE, aspartate aminotransferase, inorganic pyrophosphatase, etc. could be detected only in cytosolic fraction. These preliminary analyses of proteins showed that the associations of Calvin cycle enzymes detected by functional analysis and immunoelectron microscopy in 150-k fraction were not merely due to contamination of soluble proteins because many soluble proteins that were present only in cytosolic fraction were not detected in the 150-k fraction.

DISCUSSION

Calvin cycle enzymes are considered to be randomly distributed in stroma of chloroplasts or in cytosol of unicellular photosynthetic prokaryotes and function independently of each other. However, this assumption was put to question by several observations, wherein interaction among sequential enzymes was detected in vitro (Sainis and Harris, 1986, Sainis et al., 1989; Sainis and Melzer, 2005; Persson and Johansson, 1989; Anderson et al., 1995). Recently, the near-neighbor analysis using immunoelectron microscopy had also suggested non-random distribution and closeness of Calvin cycle enzymes to each other in vivo (Anderson and Carol, 2004, 2005, 2006; Negi et al., 2008). Occasionally, interaction of Calvin cycle enzymes with thylakoid membranes was envisaged and its significance was assessed indirectly (Sainis et al., 2003; Dani and Sainis, 2005).

In the present investigation, we addressed the question of association of Calvin cycle enzymes viz. RPI, PRK, RuBisCO, PGK, and GAPDH with thylakoids using a multipronged approach that included immunoelectron microscopy of HPF fixed cells and isolated native thylakoids. We expected that the Calvin cycle enzymes requiring ATP and NADPH should be positioned in the neighborhood of thylakoids. Synechocystis 6803 was used for our studies because the photosystems in this organism are similar to that of higher plants and their small unicellular structure is an important advantage for cryo-fixation in comparison to the multicellular higher plants. We employed cryofixation to avoid a possible dislocation of enzymes, as might be the case for chemical fixation.

The cyanobacteria were fixed by different methods that included cryo-fixation using metal mirror fixation, HPF, and also chemical fixation followed by dehydration and embedding by progressive lowering of temperature (PLT). The cells after metal mirror fixation showed several artifacts such as ice crystal formation, disorganized ultrastructure, as well as cell rupture. The cells treated with PLT protocols for dehydration and embedding resulted in good preservation of the ultrastructure but showed less immunolabeling, probably caused by cross-linking of epitopes by aldehydes (data not shown). Therefore, we applied HPF using yeast cells as filler. This resulted in superior preservation of ultrastructure and an increased preservation of the immunogenecity of the antigenic sites. Since the carboxysomes were found to be intact, it was also less likely that the there was any high-pressure-induced cell disruption phenomenon in HPF fixed cells (Figure 1). The ultrastructure observed in our experiments was similar to that observed by Liberton et al. (2006) and Van de Meene et al. (2006) who had chemically fixed Synechocystis cells before cryo-fixation. HPF is acknowledged to be the best method for keeping the living architecture closest to in vivo because the molecular movements are captured at their respective positions over fractions of milliseconds and all the cellular water is converted to paracrystalline ice, thus reducing the possibility of freezing-related distortion of the system. Indeed, the timescale of freezing has to be extremely narrow and subsequent procedures have to be done at liquid nitrogen temperatures to reduce the ice crystal formation. Therefore, this technique is extremely critical. Ice crystals, if formed, can be detected under electron microscope due to the contrast they provide to the sample. Our results showed no ice crystal formation, thus corroborating successful implementation of HPF.

Immunolocalization studies on Calvin cycle enzymes using these HPF fixed cells showed that over 70% gold particles for RPI, PRK, PGK, and GAPDH and 50% for RuBisCO could be located to thylakoid membranes. The percent association in the case of RuBisCO was less compared to other Calvin cycle enzymes (Figure 2 and Table 1). This may be due to the fact that about 36% of gold particles are found in carboxysomes. In the case of other enzymes, the label in carboxysomes was negligible. Michael et al. (1993) observed that in Synechococcus UTEX 625, if the cells were grown in cultures not enriched with CO₂, the RuBisCO could be detected in carboxysomes and PRK in the periphery along the thylakoids. In the present study using Synechocystis cells, RuBisCO was found to be distributed in thylakoids as well as in carboxysomes and PRK along the thylakoid membranes. The possibility that, due to high pressure, the enzymes may have been released from rupturing of carboxysomes is ruled out because this would lead to a random distribution of the released enzymes. As against this, we have observed specific distribution of Calvin cycle enzymes along thylakoids.

The results indicated that although Calvin cycle enzymes are extracted in aqueous buffers after disruption of cells, as shown classically by various researchers, they would be located in the neighborhood of thylakoid membranes in vivo. These in-situ immunolocalization studies were strongly complemented by the immunoelectron microscopy and functional analysis of isolated thylakoids.

The thylakoid membranes from Synechocystis 6803 were separated into three fractions, sedimenting at 40 000, 90 000, and 150 000 g, as in the case of Anacystis nidulans (Dani and Sainis, 2005). The functional assays were done by monitoring R-5-P+ADP+P_i-dependent CO₂ fixation (Table 2). The purpose was to evaluate the link between light and dark reactions and to understand the functional significance of the supramolecular organization of the photosynthetic apparatus in this context. The photophosphorylation-dependent CO₂ fixation reaction will involve the whole chain electron transport in light and subsequent formation of ATP by CF₁. Linked activities of RPI and PRK will generate RuBP using R-5-P and the ATP produced during light reaction (Supplemental Figure 2A). The controls included reaction mixtures, where ADP or DTT were not added. In the absence of ADP, the activity was negligible, indicating that no ATP was present in the preparation and the reaction was dependent on production of ATP using ADP and P_i. The dependence of the reaction on DTT indicated that phosphoribulokinase activation was required for this assay. Out of the three thylakoid fractions, the 150-k fraction showed highest photophosphorylation-dependent carboxylase activity (Table 2). We could also detect linked activities of the five sequential enzymes of Calvin cycle RPI, PRK, RuBisCO, PGK, and GAPDH, measured as R-5-P+ATP-dependent GAPDH activity in the three thylakoid fractions. Out of these, the 150-k fraction showed highest activity (Table 2). The results suggested that, although Calvin cycle enzyme activities could be detected in all the thylakoids, the 150-k fraction had an efficient functional super complex displaying higher linked activities of light and dark reactions.

Immunogold labeling of the isolated 150-k fraction showed a significant number of gold particles with antibodies of Calvin cycle enzymes as well as with anti-CF₁ antibody (Figure 3). Since immunolabeling could be seen using isolated membranes, this indicated that the enzymes were exposed and not enclosed in the vesicles. High salt washing of the membranes resulted in a decrease in the immunolabeling of this fraction. However, complete removal of RuBisCO and PRK was not possible, indicating that Calvin cycle enzymes were tightly bound to thylakoids. The possibility of formation of vesicles of membranes enclosing soluble enzymes was also ruled out, since some of the enzymes could be removed completely after washing the 150-k membrane fraction with 1 M KCl. The 40 and 90-k fractions also showed immunolabeling with antibodies of Calvin cycle enzymes (data not given), but the linked activities in these fractions were lower (Table 2), suggesting that the 150-k fraction probably had superior organization of these soluble enzymes as compared to the 40 and 90-k fractions. SDS–PAGE and proteomic analysis of the three thylakoids and soluble fraction showed that there were proteins, which were common to the 150-k fraction and the cytosolic fraction. If there is a chance sticking of proteins of soluble phase with the membranes, then they should stick to all the membrane fractions. There should be an equal probability for every protein to stick to the membrane. Therefore, the proteins present in the soluble phase should be detected in all the three membranes on SDS–PAGE. However, this was not observed (Supplemental Figure 3A and Supplemental Table 1). Preliminary LC–ESI–Q–TOF analysis of soluble fraction also showed presence of peptides unique to this fraction (Supplemental Table 1). These observations rule out the possibility that soluble proteins are sticking non-specifically to the membranes. Furthermore, if there was a contamination of the soluble proteins with the membranes during sample preparation, then these contaminants should appear in membranes in much lower amounts in comparison to the soluble phase when both of these samples are run on the gel on equal protein basis. However, bands could be observed in the gels that are equally intense in membrane fraction and in the soluble phase, which clearly depicts that this is not a mere contamination. Additionally, we have isolated these three thylakoid fractions several times for variety of analyses from cell-free extracts of Synechocystis 6803. These findings may have a subtle structural significance. Thus, the present study showed that the sequential enzymes of the Calvin cycle are organized adjacent to the thylakoid membranes. Such organization should provide functional facilitation to light and dark reactions of photosynthesis. The thylakoid membranes would provide a surface for the organization of sequential enzymes of the Calvin cycle and also offer a hydrophobic environment with very limited solvation capacity, so that the intermediates of metabolic pathway do not have a chance to diffuse randomly. The enzymes may be present on the surface of thylakoid membranes as a thin layer with the intermolecular distance such that the product of one enzyme would readily reach the next enzyme of the pathway, resulting in substrate channeling. Recent results by Howard et al. (2008) have shown the thioredoxin-mediated reversible dissociation of multiprotein complex involving Calvin cycle enzymes viz. PRK–CP12–GAPDH in response to changes in light intensity. These findings also suggest a role for thylakoids in regulating the action of Calvin cycle enzymes. The existence of the multi-enzyme complex of PRK–CP12–GAPDH in the neighborhood of thylakoid membrane will facilitate its regulation by thioredoxin.

It is worth noting that, in the case of mitochondria, interaction of Kreb's cycle enzymes with the inner surface of mitochondrial membranes was predicted based on cross-linking, histochemical, and biochemical studies (Srere, 1993). We could detect Calvin cycle enzymes in the isolated thylakoid membrane fractions without the use of chemical cross-linkers. The immunolabeling of isolated thylakoid membranes with antibodies of soluble Calvin cycle enzymes was observed even after extensive washes with buffers containing salt and detergent, which are normally used during the immunolabeling labeling procedure. Sherman et al. (1994) have performed immunolocalization of cytochrome oxidase in isolated membranes and Chen et al. (2004) have carried out similar studies on phosphate carrier (PIC) and adenine nucleotide carrier of isolated ATP synthase complex from rat liver mitochondrial membranes. However, in both these cases, a brief cross-linking treatment with glutaraldehyde was provided in spite of the fact that these proteins were components of the respective membranes. We have not used any cross-linker in our experiments on immunolabeling of isolated thylakoids. The existence of phycobilisomes and Calvin cycle enzymes in 150-k fraction even after harsh breakage and washing treatment indicated relative stability of interactions among these components, suggesting the structural significance to these interactions. Previously, the thylakoid membranes obtained through severe mechanical procedures like ultrasonication or passage through French press were normally found to be inactive in non-cyclic photophosphorylation (Krogmann, 1973) and needed various stabilizers to maintain their activity. In contrast to this, the 150-k thylakoid fraction showed photophosphorylation-dependent CO₂ fixation in light. Interestingly, among the three thylakoid fractions, the 150-k fraction showed the highest coordinated activities of the components of light and dark reaction. This may be due to optimization in organization of components of photosynthesis in this fraction.

The proteomic studies by Srivastava et al., 2005 on the purified thylakoid membranes prepared by a combination of sucrose density centrifugation and aqueous two-phase partitioning followed by washing with 0.1 M sodium bicarbonate from cell-free extracts of Synechocystis 6803 using MALDI–TOF MS analysis had also shown that several Calvin cycle enzymes such as PRK, RuBisCO, PGK, and GAPDH could be detected in the proteins obtained from thylakoid membranes. A comparative proteomic analysis of the 40, 90, and 150-k fraction of thylakoids by nano ESI–LC–MS confirmed the presence of these Calvin cycle enzymes (data not shown).

We have observed that only 20–30% of Calvin cycle enzymes are associated with membranes. In this connection, it is interesting to note that in the case of antisense transgenic plants, where concentration of several enzymes was reduced by over 80–90%, the overall rate of CO₂ fixation was not affected significantly (Stitt and Sonnewald, 1995). These observations have raised several new questions on the regulation of metabolism in vivo. Supramolecular organization and channeling must be the important aspects of in-vivo chemistry, and metabolic controls may be distributed over many enzymes in vivo. The structural and organizational aspects of enzymes, proteins, membranes, micro molecules, and water would be significant in deciding the kinetic parameters of metabolism in vivo in the living organisms.

Our observation on the possible involvement of thylakoid membranes in the organization of Calvin cycle enzymes also suggests a convenient approach for channeling of intermediates in a multistep pathway like the Calvin cycle. Sequential enzymes may be juxtaposed on the surface of membranes without forming the tight multi-enzyme complexes. The organization of sequential enzymes near each other on the surface of membranes will facilitate the transfer of intermediates among these enzymes because of the adjacent location of their active sites. The hydrophobic environment in the neighborhood of membranes will prevent unwanted diffusion of intermediates in the cytosol or stroma and thus indirectly helping in its diffusion to the active site of the next enzyme. Such organization would result in facilitated diffusion of metabolites among sequential enzymes, without having to tightly interlock the active sites in the multi-enzyme organization. These organizations would be dynamic and delicate and hence not preserved after cell disruption and extraction of soluble components in aqueous buffers and therefore not detectable by conventional biochemical procedures. However, in water-limited and protein-crowded environment in vivo, they would play an important role. Minor reorientation of the enzymes in this association would result in altering the channeling and, in turn, the efficiency of the pathway. This may be the reason why we do not see linked activities of Calvin cycle enzymes in 40 and 90-k fractions, though we could detect the enzymes immunologically in these fractions.

The present study showed for the first time the ultrastructure of Synechocystis cells, which were not fixed chemically. In these investigations, we have standardized an HPF protocol for cryo fixation of unicellular cyanobacteria. In-situ location of Calvin cycle enzymes was also explored in the cryo-fixed cells for the first time. Additionally, the immunolabeling of the isolated thylakoids with antibodies of Calvin cycle enzymes was also employed for the first time to show their association with membranes. Such cryofixed cells and isolated membranes can be used further to map the exact 3-D organization of molecular components of light and dark reactions taking visual approaches in proteomics such as mapping nearest neighbor distances (Anderson et al., 2003) and/or by cryo-3-D-electron tomography (Nickell et al., 2006) to discern the molecular architecture of the photosynthetic module.

METHODS

Chemicals

All the chemical reagents were from Sigma–Aldrich Chemie (Munich, Germany). The radioactive bicarbonate was from Board of Radiation and Isotope Technology, India.

Growth of Culture

Culture of Synechocystis 6803, obtained from Dr W. Vermaas, Arizona State University, was grown in BG-11 medium as described by Rippka et al. (1979) under continuous white light of intensity 21 W m⁻² and constant temperature of 30°C.

High-Pressure Freezing, Freeze-Substitution, Immunogold Labeling, and Transmission Electron Microscopy of Cells of Synechocystis 6803

Cells in the log phase were centrifuged in sealed yellow Eppendorf tips and the pellet was transferred into nitrocellulose tubes by capillary action. The filled tubes were cut into 2-mm pieces and placed into 0.2-mm platelets. Yeast paste was used as a filler and the flat side of a 0.30-mm platelet was used as a lid. For HPF, a Leica EM HPF unit (Leica, Benzheim, Germany) was used. Freeze-substitution with pure acetone was carried out in an automated freeze substitution unit (AFS, Leica, Benzheim, Germany) as follows: 72 h at –80°C, 12 h warming up to –70°C, 24 h at –70°C, 12 h warming up to –50°C, 24 h at –50°C, 12 h warming up to –35°C, 24 h at –35°C and 12 h warming up to 20°C. Resin infiltration was done stepwise with 10, 40, 60, and 80% LR White in acetone for 4 h each. After infiltration with 100% resin overnight, the samples were polymerized for 48 h at 60°C. Sections with a thickness of 70 nm were cut with a diamond knife and used for electron microscopic examination. Immunogold labeling was performed with polyclonal antibodies against CF₁, RPI, PRK, PGK, GAPDH, and RuBisCO as described previously by Teige et al. (1998). The anti-rabbit antibodies were purified on a Protein-A sepharose column. The specificity of antibodies was checked using Western blots. The data are shown in Supplemental Figure 1. For controls in place of primary antibody, preimmune serum was used. Sections were contrasted with uranyl acetate prior to examination in a Tecnai G² Sphera transmission electron microscope at 120 kV.

Quantitative Analysis and Spatial Localization of Immunogold Particles

The density of the gold labels was determined by counting the gold particles on the micrographs at a magnification of 22 000× and calculating the number of particles per cell. For these experiments, 5 nm protein A-gold was used. We have defined labeling of membrane-associated chloroplast enzymes as the number of immunogold particles residing over the thylakoid membranes or adjacent up to a distance of 2.5 nm to thylakoid membranes. Gold particles that were not located over or adjacent to thylakoid membranes have been considered to label enzymes not attached to thylakoid membranes. The percent of particles associated with thylakoid membranes was computed for each enzyme on thin sections of 20 different cells.

Preparation of Membrane Fractions

Cell cultures grown for 10 d, with 0.2–0.3 O.D. at 730 nm were used for isolation of thylakoid membranes as described by Murata et al. (1981) with the following modifications. Cells from 2-l culture were harvested, washed, and re-suspended in 20 ml of membrane isolation buffer (MIB-10 mM Tris-HCl, pH 7.8; 10 mM MgCl₂; 50 mM NaHCO₃; 1 mM EDTA; 12 mM β mercaptoethanol and 10% sucrose containing 3 mg PMSF). The cell-free extract was prepared by sonication at 4°C for 50 min in pulse mode. The extract was centrifuged at 8000 rpm at 4°C for 10 min to remove unbroken cell debris. Supernatant was subjected to ultracentrifugation at 40 000, 90 000 and 150 000 g for 1 h each sequentially at 4°C as described previously (Dani and Sainis, 2005). The pellets after each round were re-suspended in 0.2–1.0 ml of MIB and referred to as 40, 90, and 150-k membrane fraction. The supernatant obtained after centrifugation at 150 000 g was considered as cytosolic fraction.

Assays of Enzyme Activities

The thylakoid fractions containing 10 μg chlorophyll were assayed for the phosphorylation-dependent carboxylase activity and R-5-P+ATP-dependent GAPDH activity.

The reaction mixture for the photophosphorylation-dependent CO₂ fixation consisted of membranes containing 2–10 μg Chlorophyll, 50 mM Tricine–NaOH pH 7.5, 600 mM sucrose, 10 mM NaCl, 5 mM MgCl₂, 5 mM K₂HPO₄, 3 mM ADP, 20 mM NaH¹⁴CO₃ (specific activity 0.5 mCi), and 10 mM DTT. Assay was performed under white light of intensity 21 W m⁻², at room temperature. Reaction was started by addition of 2 mM R-5-P and 2 mM methyl viologen and was terminated after 10 min by transferring 100 μl of reaction mixture to 200 μl of 6N acetic acid in scintillation vials. Acid stable product was counted in liquid scintillation counter. The reaction mixtures without ADP, R-5-P, or DTT constituted controls for this assay.

The assay mixture for R-5-P+ATP-dependent GAPDH activity consisted of membranes in MIB (10 μg of chlorophyll), 40 mM NaHCO₃, 0.14 mM NADPH, 2 mM ATP, 10 mM DTT, 20 mM MgCl₂, and 2 mM R-5-P. Decrease in O.D. at 340 nm was monitored. The reaction mixtures without R-5-P constituted controls for this assay.

Pigment Estimation

Chlorophyll estimation was done according to Tandeau de Marsac and Houmard (1988) and phycocyanin was estimated as described by Glazer (1988).

Immunogold Labeling and Electron Microscopy of Isolated Thylakoid Membranes

The 150-k fraction was diluted to bring the chlorophyll content to 5–10 μg ml⁻¹. Copper grids coated with 0.4% formvar and poly-L-lysine were inverted on a 10-μl drop of the membrane fraction for 10 min and blot dried. The immunogold labeling was carried out as described for thin sections. For controls, the grids treated with membrane fractions, but not treated with any primary antibody, were used. The grids were negatively stained with 4% aqueous solution of uranyl acetate for 30 s and viewed under transmission electron microscope as described previously.

SUPPLEMENTARY DATA

Supplementary Data are available at Molecular Plant Online.

FUNDING

The International Bureau of Federal Ministry of Education and Research, Germany (BMBF), and the Department of Atomic Energy, India, funded this project under Indo-German Collaborative project IND05/009.