Spontaneous Assembly of Photosynthetic Supramolecular Complexes as Mediated by the Intrinsically Unstructured Protein CP12*

CP12 is a protein of 8.7 kDa that contributes to Calvin cycle regulation by acting as a scaffold element in the formation of a supramolecular complex with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and phosphoribulokinase (PRK) in photosynthetic organisms. NMR studies of recombinant CP12 (isoform 2) of Arabidopsis thaliana show that CP12-2 is poorly structured. CP12-2 is monomeric in solution and contains four cysteines, which can form two intramolecular disulfides with midpoint redox potentials of –326 and –352 mV, respectively, at pH 7.9. Site-specific mutants indicate that the C-terminal disulfide is involved in the interaction between CP12-2 and GAPDH (isoform A4), whereas the N-terminal disulfide is involved in the interaction between this binary complex and PRK. In the presence of NAD, oxidized CP12-2 interacts with A4-GAPDH (KD = 0.18 μm) to form a binary complex of 170 kDa with (A4-GAPDH)-(CP12-2)2 stoichiometry, as determined by isothermal titration calorimetry and multiangle light scattering analysis. PRK is a dimer and by interacting with this binary complex (KD = 0.17 μm) leads to a 498-kDa ternary complex constituted by two binary complexes and two PRK dimers, i.e. ((A4-GAPDH)-(CP12-2)2-(PRK))2. Thermodynamic parameters indicate that assembly of both binary and ternary complexes is exoergonic although penalized by a decrease in entropy that suggests an induced folding of CP12-2 upon binding to partner proteins. The redox dependence of events leading to supramolecular complexes is consistent with a role of CP12 in coordinating the reversible inactivation of chloroplast enzymes A4-GAPDH and PRK during darkness in photosynthetic tissues.

The photosynthetic reduction cycle for carbon organication (Calvin cycle) is a finely regulated metabolism that plants keep tuned with light reactions of photosynthesis under variable environmental conditions. Thioredoxins and metabolic intermediates play essential signaling roles within this complex reg-ulatory system (1). Two nonconsecutive enzymes of the Calvin cycle, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) 2 and phosphoribulokinase (PRK), are regulated in this way, both individually and through the reversible formation of supramolecular complexes (2)(3)(4)(5)(6)(7)(8).
Chloroplast GAPDH is mainly heteromeric in land plants, with homologous A and B subunits occurring in stoichiometric ratio (9,10). The B-subunits confer regulatory properties, and the enzyme oscillates between a fully active A 2 B 2 tetramer at one extreme and a partially inhibited A 8 B 8 hexadecamer at the other (11). Partially polymerized intermediates like A 4 B 4 have also been reported (12)(13)(14). Thioredoxins and metabolites directly regulate AB-GAPDH activity and strongly affect the equilibrium between active tetramers and aggregated forms (8,10).
A second isoform of chloroplast GAPDH in land plants is a stable homotetramer of A-subunits (A 4 -GAPDH) (9,15), similar to Calvin cycle GAPDH of lower photosynthetic organisms (5,6,16). A 4 -GAPDH only accounts for a minor portion of total chloroplast GAPDH activity in land plants (14,15). Due to the absence of B-subunits, A 4 -GAPDH is not directly regulated by thioredoxins and metabolites (8,17,18), although the reversible glutathionylation of the active site cysteine 149 provides a mechanism of A 4 -GAPDH regulation that may be relevant under stress (19). Alternatively, reversible down-regulation of A 4 -GAPDH activity can be achieved through formation of a supramolecular complex with PRK and the regulatory peptide CP12 (2)(3)(4)(5)(6)(7)(8)20). In land plants, PRK itself undergoes a light/ dark modulation mediated by thioredoxins (21), but once incorporated into the complex, both A 4 -GAPDH and PRK activities are inhibited in a coordinated manner (7). Similar to A 8 B 8 -GAPDH, the stability of the supramolecular complex involving A 4 -GAPDH, CP12, and PRK is controlled by thioredoxins and several cofactors, including NAD(H) and NADP(H), ATP, and 1,3-bisphosphoglycerate (3)(4)(5)(6)(7). It is generally agreed that aggregated forms of GAPDH (A 8 B 8 ) and GAPDH-CP12-PRK complexes are prevalent in chloroplasts in the dark, whereas illumination favors the accumu-* This work was supported by Italian Ministry of University Grants FIRB 2003 and PRIN 2005. The 800-MHz spectrometer was funded by the European Union (Fonds Européen de Dé veloppement Ré gional), French Research Ministry, Region Nord-Pas de Calais, University of Sciences and Technology of Lille, and CNRS. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed E-mail: trost@alma.unibo.it. lation of fully active GAPDH tetramers (A 2 B 2 , A 4 ) and PRK dimers (4 -6, 8, 12, 13, 22, 23).
The acronym CP12 refers to small proteins of nearly 80 amino acids, widespread in oxygenic photosynthetic organisms (24,25), apparently lacking an ordered structure in solution (20). Interestingly, the C-terminal half of CP12 is closely related to the C-terminal extension (CTE) of GAPDH B-subunits (24). Both the CTE and the C-terminal part of CP12 contain a couple of conserved cysteines, which can form an intramolecular disulfide and are potential targets of thioredoxin regulation (3, 8, 10, 17, 26 -28). In addition, most CP12 proteins contain a second pair of conserved cysteines in the N-terminal half of the molecule also able to form a disulfide bond (2,25). However, the redox properties of CP12 disulfide bonds are unknown, and it is difficult to predict whether CP12 could exist in different redox states in vivo. Studies on the green algae Chlamydomonas reinhardtii (20) and the flowering plant Arabidopsis thaliana (7) help to trace the sequence of events during the formation of CP12-mediated supramolecular complexes. First, A 4 -GAPDH (in complex with NAD) interacts with oxidized CP12, and then oxidized PRK can participate in the assembly of the ternary complex, resulting in strongly inhibited enzyme activities.
In this paper, we shall investigate some biochemical-molecular features of A. thaliana CP12-2 and properties of the supramolecular complexes it forms with A 4 -GAPDH and PRK. The CP12-2 isoform, produced by one of three CP12 genes known for this species (8), was chosen as a model, since the expression pattern of the CP12-2 gene in different Arabidopsis organs and conditions strictly followed that of GAPDH, PRK, and other Calvin cycle genes (8,29). The resulting model supports the view that these supramolecular complexes represent an instrument for photosynthetic organisms to finely modulate Calvin cycle turnover in response, for example, to changes in light intensity as commonly occur in natural environments and to safely and reversibly store photosynthetic enzymes in an inactive conformation during the night.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-Heterologous expression and purification of recombinant A 4 -GAPDH (At3g26650), PRK (At1g32060), CP12-2 (At3g62410), and CP12-2 site-specific mutants of A. thaliana were performed as described (7). NMR analyses were performed on uniformly 15 N-labeled Histagged CP12-2 samples obtained by transformed Escherichia coli BL21(DE3) cells grown in M9 minimal medium containing 1 g/liter 15 NH 4 Cl (Euriso-top, France) as the sole nitrogen source. An overnight culture of 25 ml in M9 medium was transferred to 500 ml of fresh M9 medium, both supplied with kanamycin (50 g/ml) and grown at 37°C under shaking. When optical density at 600 nm reached 0.6 -1.0 units, expression was induced by the addition of 0.4 mM isopropyl-␤-D-thiogalactopyranoside. Since M9 minimal medium prevented rapid cell growth, the induction phase was prolonged for 15 h before cells were collected by centrifugation (10,000 rpm; 15 min). CP12-2 was purified from the resulting pellet as previously described (7). Purified proteins were quantified by absorbance at 280 nm (7), desalted in appropriate buffers, and stored at Ϫ20°C. Concentrations of purified proteins are all referred to native con-formations (CP12-2 monomers, A 4 -GAPDH tetramers, and PRK dimers).
Analysis of Thiol Groups and Redox Titration of CP12-2-Protein thiol analyses and redox titrations were performed with pure CP12-2 in 100 mM Tricine-NaOH, pH 7.9. Redox titration experiments were performed with 70 M CP12-2 incubated for 3 h at 25°C with variable ratios of reduced and oxidized DTT (20 mM total concentration) in a final volume of 500 l. After incubation, samples were desalted with PD10 columns (GE Healthcare) equilibrated with 100 mM Tricine-NaOH, pH 7.9. In order to avoid any possible DTT contamination, only the first 2 ml of eluted samples were collected. Control experiments were performed under the same conditions but in the absence of CP12-2. Absorbance at 280 and 412 nm was recorded immediately before and after the addition of 0.5 mM 5,5Ј-dithiobis(2nitrobenzoic acid) (DTNB). The number of solvent-accessible thiol groups under different redox conditions was calculated from the ratio between the absorbance at 412 nm (molar extinction coefficient of 14,150 M Ϫ1 for 2-nitro-5-thiobenzoate (thiolate) dianion) (32) and the absorbance at 280 nm (molar extinction coefficient of 8,370 M Ϫ1 for CP12-2) (7).
Redox titration results were fit by nonlinear regression (CoHort Software) to the Nernst equation for two redox components (26). Midpoint redox potentials are reported as average values Ϯ S.D. of triplicate experiments.
Isothermal Titration Calorimetry (ITC)-Calorimetric measurements were carried out using a VP-ITC MicroCalorimeter (MicroCal Inc., Northampton, MA). Each experiment was performed at a constant temperature of 30°C and consisted of 25 injections of 10-l aliquots, repeated every 200 s. All samples were degassed by stirring under vacuum before use. Heat of dilution, measured by control experiments in which samples were injected into a buffer-filled cell, was subtracted. Signals recorded in each experiment were integrated using OriginPro 7.5 software supplied with the instrument. The thermodynamic binding parameters (dissociation constant (K D ), variations of enthalpy (⌬H), Gibbs free energy (⌬G), entropy (⌬S), and the number of binding sites (n)) were obtained by nonlinear regression of the integrated heat plots, according to the "one set of sites" model of the software.
Calorimetric titrations of binary complex formation were carried out with 15 M oxidized CP12-2 in 25 mM potassium phosphate, 0.2 mM NAD, pH 7.5, in both sample and reference cells, whereas the syringe was filled with 52 M A 4 -GAPDH in the same buffer. Although CP12-2 behaved as a ligand, it was not filled in the syringe because of its very high heat of dilution.
The presence of ligand (CP12-2) in the cell and macromolecule (A 4 -GAPDH) in the syringe was taken into account in the elaboration of primary data.
Calorimetric titrations of ternary complex formation were performed with 5 M preformed binary complex in 25 mM potassium phosphate, 0.2 mM NAD, pH 7.5, in sample and reference cells, whereas the syringe was filled with 70 M PRK dissolved in the same buffer. Thermodynamic parameters of binary and ternary complex formation are reported as average values Ϯ S.D. of triplicate experiments.
Dynamic Light Scattering (MALS-QELS)-Purified single proteins and preformed binary and ternary complexes were analyzed by size exclusion chromatography connected to a multiangle light scattering (MALS) equipped with QELS module (quasielastic light scattering) for R H measurements. 100-l samples were loaded on a Superdex 200HR column (GE Healthcare) equilibrated in 25 mM potassium phosphate, pH 7.5, 1 mM EDTA, 150 mM KCl, with 0.2 mM NAD (for A 4 -GAPDH, binary, and ternary complexes) or without NAD (for PRK and CP12-2). A constant flow rate of 0.6 ml min Ϫ1 was applied. Elution profiles were detected by an Optilab rEX interferometric refractometer and a Dawn EOS multiangle laser light-scattering system at 690 nm (Wyatt Technology Corp.). Data acquisition and processing were carried out using ASTRA 5.1.9.1 software (Wyatt Technology). Determination of molecular masses and hydrodynamic radii are reported as mean values Ϯ S.D. of duplicate experiments.

CP12-2 of A. thaliana Is Intrinsically Unstructured and
Monomeric-NMR analysis of CP12-2 in the oxidized state revealed that most of the amide proton resonances are localized between 8.5 and 8.0 ppm, the so-called random coil region, strongly suggesting that CP12-2 is mainly unstructured (Fig. 1). Only a few residues located at the C terminus exhibit amide chemical shifts outside the random coil range, indicating structuration (33). Consistent with the importance of disulfide bridges in this respect, the reduction of oxidized CP12-2 by DTT led to typical random coil signals (not shown), as previously described for Chlamydomonas CP12 (20).
The calculated molecular mass of recombinant CP12-2 (after proteolytic cleavage of the His tag) was 8.7 kDa. Although oxidized CP12-2 behaves as a protein of 29 kDa in size exclusion chromatography (7), MALS-QELS analysis of the protein eluted from the size exclusion column yielded a molecular mass of 9 Ϯ 1 kDa, conclusively demonstrating that under native conditions, CP12-2 is a monomer ( Table 1).
is Avogadro's number, f/f 0 is the ratio of frictional coefficients (set to 1.2 for spherical proteins), and V k is the partial volume set to 0.73 for a spherical protein (45). b The experimental R H of CP12-2 was close to the lower detection limit of the QELS module. c Since chromatographic runs were performed in the presence of 0.2 mM NAD, molecular masses were calculated on the assumption that each A 4 -GAPDH tetramer bound four NAD molecules. d Calculated molecular mass for a binary complex with stoichiometry (A 4 -GAPDH)-(CP12-2) 2 . e Calculated molecular mass for a ternary complex with stoichiometry ((A 4 -GAPDH)-(CP12-2) 2 -(PRK)) 2 . JANUARY 25, 2008 • VOLUME 283 • NUMBER 4

JOURNAL OF BIOLOGICAL CHEMISTRY 1833
CP12-2 Redox Properties-To gain insight into the redox properties of CP12-2 cysteines, redox titrations were performed in the presence of DTNB as a probe to reveal free protein thiols under varying redox conditions. Fully reduced/oxidized samples were obtained following equilibration with 20 mM reduced or oxidized DTT, which was then removed by desalting. Although reduced CP12-2, whose amino acid sequence includes four Cys residues, was found by DTNB titration to contain four reactive thiols (4.5 Ϯ 0.5), oxidized CP12-2 had none (Ϫ0.3 Ϯ 0.1), indicating that both CP12-2 disulfides could be redox-titrated by DTT plus DTNB. Data from redox titrations of purified CP12-2 were therefore fitted to a Nernst equation for two different thiol/disulfide equilibria, equally contributing to the total redox response. At pH 7.9, the midpoint redox potentials (E m,7.9 ) of CP12-2 disulfides were estimated as Ϫ326 Ϯ 2 and Ϫ352 Ϯ 6 mV, respectively (Fig. 2).
The affinity between A 4 -GAPDH and CP12-2 was analyzed by isothermal titration calorimetry (ITC) (Fig. 3). Such interaction was characterized by a K D of 0.18 Ϯ 0.02 M. In agreement with MALS analysis, two binding sites for CP12-2 were detected per A 4 -GAPDH tetramer (n ϭ 1.9 Ϯ 0.2), with no evidence for different affinities between binding sites. Binding of two CP12-2 molecules to each A 4 -GAPDH tetramer was exothermic (⌬H ϭ Ϫ15 kcal mol Ϫ1 ), and although leading to a simultaneous decrease in entropy (T⌬S ϭ Ϫ5 kcal mol Ϫ1 ), the process was exoergonic overall (⌬G ϭ Ϫ9.4 kcal mol Ϫ1 ; Table 2).
The interaction between oxidized PRK and preconstituted (A 4 -GAPDH)-(CP12-2) 2 binary complexes was also investigated by multiangle light scattering. Binding of PRK (85 Ϯ 7 kDa; MALS) to the binary complex gave rise to a ternary complex of 498 Ϯ 6 kDa (MALS), likely to include two dimers of PRK plus two (A 4 -GAPDH)-(CP12-2) 2 binary complexes (calculated molecular mass ϭ 488 kDa; Table 1). The stoichiometric ratio of two A 4 -GAPDH subunits per PRK subunit in the 498-kDa ternary complex was well compatible with the relative intensities of Coomassie-stained bands in denaturing gel electrophoresis (not shown).

4).
As in experiments of binary complex formation, data were fitted to a simple model describing the interaction on n "ligands" (PRK) to one "macromolecule" (A 4 -GAPDH)-(CP12-2) 2 with n identical binding sites. Using this model, a K D value of 0.17 Ϯ 0.09 M and n ϭ 1.3 Ϯ 0.1 were obtained. Thermodynamic parameters indicated that ternary complex assembly was largely exothermic (⌬H ϭ Ϫ20 kcal mol Ϫ1 ) and exoergonic (⌬G ϭ Ϫ9.3 kcal mol Ϫ1 ), despite a significant decrease in entropy (T⌬S ϭ Ϫ11 kcal mol Ϫ1 ; Table 2). The entropic penalty associated with both binary and ternary complex formation (Table 2) suggests that CP12-2 adopted a folded structure upon binding to its structured partners, a typical behavior of intrinsically unstructured proteins (34). Indeed, the assembly of complexes involving disordered proteins is often associated with an appreciable entropic cost that is paid in terms of lower binding affinity (35). This situation seems also to apply to CP12-2 and its partners (A 4 -GAPDH and PRK in Arabidopsis).
CP12-2 Site-specific Mutants-CP12-2 was previously shown to bind A 4 -GAPDH under oxidizing conditions (3,7,20). The roles of the two active CP12-2 disulfides were therefore investigated by use of two site-specific mutants, having either the first (Cys 22 ) or the last (Cys 73 ) cysteine of the sequence mutated into serine. Under reducing conditions, both mutants (C22S and C73S) behaved as wild type CP12-2 in size exclusion chromatography, but they tended to form dimers under oxidizing conditions (not shown), possibly due to intermolecular disulfides between residual free cysteines (Cys 31 in C22S, Cys 64 in C73S). However, most of CP12-2 mutants remained in the monomeric state even when oxidized.
Although mutant C22S interacted with A 4 -GAPDH similarly to wild type CP12-2 (Fig. 5, A and B), mutant C73S did not form any complex with A 4 -GAPDH (Fig. 5C), suggesting an essential role of the C-terminal disulfide in CP12/GAPDH interaction. In the presence of PRK, neither mutant was able to promote the formation of a ternary complex reminiscent of the 498-kDa complex of wild type CP12-2 (7) (not shown). Clearly, the N-terminal disulfide plays a role in stabilizing the link between PRK and the binary complex within the final ternary complex.

DISCUSSION
CP12 is a widespread regulatory protein of oxygenic photosynthetic organisms that contributes to the regulation of carbon metabolism by producing supramolecular complexes with two enzymes, GAPDH and PRK, accounting for most of the energetic needs of the Calvin cycle (2,3). Why land plants, which possess distinct systems to regulate both GAPDH (mainly AB isoform) and PRK under variable light conditions (1,8), need to maintain this additional and more ancient regulatory system based on CP12 is a matter of debate. It has been argued that CP12 provides plasticity and coordination to the regulatory network modulating the Calvin cycle (7); the minor isoform A 4 -GAPDH is unregulated in the absence of CP12, and PRK too becomes more sensitive to modulators (e.g. 1,3bisphosphoglycerate and pyridine nucleotides) when embodied in the complex. Moreover, CP12 could link PRK also to the more abundant AB isoform of GAPDH (3,4), thereby further contributing to the coordinated control of the Calvin cycle. However, since antisense tobacco plants with reduced levels of CP12 show a severely altered leaf morphology (36), we should assume that the function of CP12 might well extend beyond the Calvin cycle. Interestingly, a recent analysis of protein-protein interaction maps from different eukaryotes revealed that poorly structured proteins (like CP12) commonly behave as hub proteins (i.e. they interact with many partners and often have regulatory functions) (37). The interaction of CP12 with partners other than just GAPDH and PRK, although speculative, may be a possible explanation for the aberrant leaf morphology of CP12 antisense tobacco plants (36). Although the fully oxidized form of chloroplast CP12-2 of A. thaliana contains two intrachain disulfide bridges (Fig. 1), solution NMR studies show that it is a poorly structured protein even under oxidizing conditions, attaining almost complete unstructuration when the cysteines are reduced. This lack of structure was supported by several bioinformatic predictors based on different algorithms (38) (not shown). Like CP12 of C. reinhardtii (21), also A. thaliana CP12-2 can thus be considered an intrinsically unstructured protein (35,39), and this property related to its regulatory function (20,34,37,40). In contrast with other reports (3,41) but consistent with its disordered nature, the CP12-2 failed to interact with itself, as shown by dynamic light scattering analysis (MALS-QELS). Previous and significantly higher estimates of CP12 molecular mass based on size exclusion chromatography (29 -70 kDa) (3,6,7) were apparently biased by the intrinsic disorder of CP12 (39).
Oxidized CP12-2 and A 4 -GAPDH from Arabidopsis interact together with a dissociation constant in the submicromolar range (K D ϭ 0.18 M measured by ITC) (i.e. a 450-fold lower affinity with respect to Chlamydomonas (K D ϭ 0.4 nM, surface plasmon resonance)) (20). The exceptionally high stability of the complex in Chlamydomonas may be related to the inhibitory effect of CP12 on GAPDH activity in this species (28,42), whereas little inhibition, if any, was observed in the plant protein unless PRK was also recruited in a ternary complex (7). A 4 -GAPDH is the only chloroplast GAPDH isoform in C. reinhardtii (25), and CP12 may uniquely provide a way to regulate GAPDH activity in these green algae, whereas Arabidopsis and land plants in general also contain an AB-GAPDH isoform, which is itself autonomously regulated (8,10). Thus, the relevant role of CP12 in land plants may be to create a direct connection between A 4 -GAPDH and PRK, leading to coordinate down-regulation of these key enzymes, both quite active in the free state (7).
The stoichiometry of the binary complex (A 4 -GAPDH)-(CP12-2) 2 was assessed by isothermal titration calorimetry (CP12-2/A 4 -GAPDH ϭ 1.9 Ϯ 0.2) ( Table 2) and confirmed by light scattering determination of the molecular mass (MALS, 170 Ϯ 14 kDa; calculated, 166 kDa; Table 1). That one A 4 -GAPDH tetramer is able to bind two CP12 monomers is also consistent with the homology between CP12 and the CTE of subunits B of GAPDH, whose primary sequence is in the main highly related to A subunits. The crystal structure of spinach A 2 B 2 -GAPDH in the oxidized state shows that each CTE, partially structured by a disulfide bridge, is located within a wide cleft delimited by each pair of A/B-subunits (10). GAPDH tetramers (A 2 B 2 or A 4 ) (10, 15) contain two symmetrical clefts of this type, and it is likely that CP12-2 may bind to A 4 -GAPDH in the same way that CTE is allocated in A 2 B 2 -GAPDH. This conclusion is also supported by site-specific mutant C73S of CP12-2 (Fig. 5), and analogous site-specific mutants of CP12 from other species (3,6,28), which also fail to form the C-terminal disulfide and are consequently unable to build up a stable complex with GAPDH. A chloroplast complex of proteins, including GAPDH, CP12, and PRK, with a molecular mass of 460 -640 kDa has been known for years (2-7) and has been explained by a model with (GAPDH-CP12-PRK) 2 stoichiometry, based on the purported capability of CP12 to dimerize (3,5). However, the observed A 4 -GAPDH binding to two CP12-2 monomers suggests a different interpretation, also compatible with the molecular mass of the Arabidopsis ternary complex (498 Ϯ 6 kDa; MALS), (i.e. two (A 4 -GAPDH)-(CP12-2) 2 binary complexes interacting with two PRK dimers (for a calculated molecular mass of 488 kDa ( Table 1). The hydrodynamic radius of this ternary complex (7.0 Ϯ 0.1 nm) was slightly higher than expected (6.2 nm, Table 1), indicating deviation from a theoretical spherical shape. Based on the apparent symmetry of both (A 4 -GAPDH)-(CP12-2) 2 binary complexes and PRK homodimers, we propose that the architecture of the ternary complex includes two binding sites for PRK on each binary complex and two binding sites for binary complexes on each PRK dimer. As a result, a toric supramolecular complex would finally be assembled as depicted in the model of Fig. 6. This model requires a cautious interpretation of microcalorimetric data, because ternary complex formation actually involves two steps: the first interaction between PRK and the binary complex and the following dimerization of the transient complex (Fig. 6). Since thermodynamic parameters of ternary complex formation (Table 2) are related to the whole process, the meaning of the stoichiometric parameter n is uncertain in this case.
Arabidopsis CP12-2 primary sequence contains four cysteines, and two internal disulfides are formed under oxidizing conditions (3,43). The C-terminal disulfide (Cys 64 -Cys 73 ) is homologous to the regulatory disulfide of A 2 B 2 -GAPDH (3,10,24). Redox titration analysis of CP12-2 revealed that the most negative midpoint redox potential (E m,7.9 ϭ Ϫ352 mV) was identical to that of spinach A 2 B 2 -GAPDH (Ϫ353 mV) (26) and was therefore attributable to the C-terminal disulfide. Surprisingly, the E m,7.9 of the second disulfide of CP12-2 (Ϫ326 mV, N-terminal) was identical to that of spinach PRK (Ϫ330 mV) (7), although no sequence similarities exist between the two proteins. The N-terminal disulfide was not required for binary complex formation with GAPDH, but the ternary complex of 498 kDa did not form in its absence (see also Refs. 3 and 28). Since PRK does not bind to CP12-2 alone (7,20), it is likely that the CP12-2 domain, including the N-terminal disulfide, forms a binding site for PRK together with A 4 -GAPDH. However, the cyanobacterium Synechococcus PC7942 presents a GAPDH-CP12-PRK ternary complex of about 500 kDa, although CP12 of this species has no N-terminal cysteines (6).
The different redox properties of the two disulfides of Arabidopsis CP12-2 and the sequential formation of the supramolecular complex (7,20) fit within a plausible physiological scenario. During transition from saturating light to suboptimal light availability, the incipient oxidation of chloroplast thioredoxins (E m,7.9 ϭ Ϫ351 mV for thioredoxin f) (44) would cause an initial oxidation of C-terminal cysteines of CP12-2 (E m,7.9 ϭ Ϫ352 mV) as well as partial inactivation of AB-GAPDH (E m,7.9 ϭ Ϫ353 mV) (26). Under these conditions, partially oxidized CP12-2 can bind A 4 -GAPDH containing bound NAD(H) at coenzyme sites (7,8). With further decreasing light, ongoing oxidation of thioredoxins would lead to formation of the N-terminal disulfide of CP12-2 (E m,7.9 ϭ Ϫ326 mV) and inactivation of PRK via disulfide formation (E m,7.9 ϭ Ϫ330 mV). A supramolecular complex of two (A 4 -GAPDH)-(CP12-2) 2 binary complexes bound to two PRK dimers could then form (Fig. 6), resulting in strong but reversible inhibition of both Calvin cycle enzymes in the dark.