Crystal structure of a type III Rubisco in complex with its product 3-phosphoglycerate

Abstract

The crystal structure of the complex of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco; EC 4.1.1.39) from Archaeoglobus fulgidus (afRubisco) with its products 3PGAs has been determined to a resolution of 1.7Å and is of the closed form. Type III Rubiscos such as afRubisco have eighteen out of the nineteen essential amino acid residues of canonical Rubisco; the nineteenth is Tyr rather than Phe. Superposition with the structure of a complex of the similar tkRubisco with the six-carbon intermediate analogue 2CABP shows the same conformation of the nineteen residues except for Glu46 and Thr51. Glu46 adopts a unique conformation different from that in other Rubiscos and makes two H-bonds with the ligand 3PGA. Similar to other closed state Rubiscos, the backbone of Thr51 is rotated and the side chain makes an H-bond with the ligand 3PGA. Two product 3PGA molecules are bound at the active site, overlapping well with the 2CABP of tkRubisco/2CABP. The positions of the P1 and P2 phosphate groups differ by 0.4Å and 0.53Å, respectively, between 2CABP and the two 3PGAs. This afRubisco/3PGA complex mimics an intermediate stage of the carboxylation reaction which occurs after the production of the two 3PGA products but before the reopening of the active site. The stability of this complex suggests that the Rubisco active site will not reopen before both 3PGA products are formed.

1. Introduction

Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) catalyzes the addition of carbon dioxide (CO₂) to ribulose-1,5-bisphosphate (RuBp) in the first step of the photosynthetic Calvin-Benson-Bassham (CBB) cycle. A molecule of CO₂ is added to the five-carbon substrate RuBp, and the resulting six-carbon intermediate is cleaved to produce two molecules of 3-phosphoglycerate (3PGA). More than 90% of the inorganic carbon that is converted into biomass is fixed by Rubisco¹. Despite its importance in the global carbon cycle, Rubisco is not a good enzyme because of its two main drawbacks: low specificity and low catalytic rate. Rubisco is also involved in photorespiration. In this pathway, its competing oxygenase activity catalyzes the addition of an O₂ molecule to RuBp, leading to the subsequent release of CO₂ and waste of energy (ATP and NADH). The typical catalytic rate of a plant Rubisco is less than 10 carboxylation reactions per second^1,2. Rubisco makes up 30-50% of the soluble protein in plant leaf and is the most abundant protein on earth^{3, 4}.

Based on amino acid sequence alignment, Rubiscos are classified into four groups⁵. Type I Rubiscos are widely distributed in plants, algae, cyanobacteria and some autotrophic bacteria. They are composed of eight large subunits (L) and eight small subunits (S) with tetragonal symmetry (L₈S₈)⁶. Type II Rubiscos are found in some autotrophic bacteria and are usually composed of only two large subunits (L₂). Both Type I and Type II Rubiscos catalyze the carboxylation reaction in the CBB cycle; the L₂ dimers are their catalytic units. Type III Rubiscos have been found only in archaea and are composed of only large subunits^7–10. Usually they are dimers (L₂), but in some cases they can assemble into higher oligomers, such as the Rubisco from Thermococcus kodakarensis (tkRubisco), that forms a pentagonal ring of dimers (L₂)₅. The L₂ dimer of a Type III Rubisco has carboxylase/oxygenase activities; however, the CBB cycle has not been found in archaea, and the biological function of Type III Rubiscos is unclear at present¹¹. It has been found that tkRubisco is involved in nucleoside metabolism¹². Type IV Rubiscos do not have the carboxylase/oxygenase activities and are thus called Rubisco-like proteins (RLPs). They have only the large subunits and the dimer (L₂) is their catalytic unit^13,14.

The importance of Rubisco in photosynthesis and hence in improving crop yields has been recognized and the enzyme itself has been studied extensively by biochemistry, molecular biology and crystallography for decades. However, only limited progress has been made toward a complete understanding of the details of the catalysis mechanism at the molecular level. This is mainly due to the complexity of the multiple-step reaction catalyzed by Rubisco, involving deprotonation and protonation, tautomerization, addition of gas molecules and water, and change of H-bond networks and conformation. Today it is thought that the carboxylation of RuBp involves the following sequence of steps (Fig. 1): the substrate RuBp first binds to the active site of activated Rubisco; a proton is removed from C3 and the substrate undergoes tautomerization from the ketose form to the enediol form (enolization); a CO₂ comes in and displaces the water molecule coordinated to the Mg²⁺; the CO₂ molecule, polarized by the Mg²⁺ and surrounding amino acid residues, makes an electrophilic attack on C2 of the enediolate to form a six-carbon intermediate (carboxylation); a water molecule is added to this intermediate at the C3 position (hydration); hydration results in the C2-C3 bond cleavage to produce a 3PGA and a carbanion; a proton is stereo-specifically added to the carbanion to produce another3PGA; the 3PGA products are released from the active site^5,15,16. Significant protein conformational change accompanies carboxylation. It is thought that, after the addition of CO₂, loop6 folds over the active site, followed by folding of the C-terminal and N-terminal regions to seal the active site (closed form). After the reaction is complete, the C-terminal and N-terminal regions and loop6 move away from the active site so that the products can be released (open form)⁵. Note that most of these steps are speculative because of the lack of good model complexes for all steps. Therefore, finding complexes with stable analogues of the intermediate steps is essential for our understanding of the reaction catalyzed by Rubisco. Such mimics have not yet been found for most intermediates.

Figure 1.

Steps in the conversion of RuBp to two 3PGA by Rubisco.

For this report we determined the crystal structure of a Type III Rubisco from Archaeoglobus fulgidus (afRubisco) in complex with its two 3PGA products. The afRubisco/3PGA complex mimics an intermediate stage of the carboxylation reaction that is after the production of the two 3PGA products but before the reopening of the active site.

2. Experimental Procedures

2.1. Protein expression and purification

The cDNA encoding the afRubisco (Af_1638) was PCR-amplified from the genomic DNA of Archaeoglobus fulgidus and inserted into the expression vector pSUMO. The recombinant plasmid pSUMO-Rubisco was transferred into E. coli BL21(DE3) for protein expression, which was induced at 18ºC for 20 hours with 0.3mM IPTG. The cells were harvested by centrifugation and then suspended in binding buffer (500mM NaCl, 50mM Tris-HCl, pH8.5, 10mM imidazole). Cell lysis was carried out by sonication. After centrifugation, the supernatant was applied to a nickel affinity column. After protein binding, the column was washed thoroughly with 100 volumes of binding buffer followed by 10 volumes of washing buffer (500mM NaCl, 50mM Tris-HCl, pH8.5, 40mM imidazole). The protein was then eluted from the column with 5 volumes of elution buffer (200mM NaCl, 300mM imidazole-HCl, pH7.5). ULP1ase was added to the eluate and incubated at 4°C overnight to cleave off the SUMO tag. The buffer was then changed to the carbamylation buffer (100mM Tris-HCl, pH8.3, 10mM MgCl₂, 20mM NaHCO₃) using a Centrifugal Filter tube (Millipore) and incubated at 4°C overnight. After being concentrated, the protein was purified using a Superdex 200 column (GE) mounted on FPLC with an elution buffer containing 150mM NaCl and 5mM Tris-HCl (pH7.5). The peak containing afRubisco was pooled and concentrated to around 20mg/ml, and stored at −80°C.

2.2. Rubisco activity assay

The carboxylase activity assay was performed according to the method described by Maeda et al. with slight modification⁷. 5μg of activated afRubisco was added to the enzymatic reaction buffer (final volume 100μl) containing 100mM Tris-HCl pH 8.2, 20mM MgCl₂, 10mM NaHCO₃, 0.5mM DTT and 2mM RuBp and incubated at room temperature for 30 minutes or at 65°C for 5 minutes. After stopping the reaction, 100μl of coupling reaction mixture was added to quantify the 3PGA formed. The coupling reaction mixture contained 100mM Tris-HCl pH 8.2, 20mM MgCl₂, 10mM NaHCO₃, 0.5mM DTT, 2mM NADH, 2mM ATP and 10 mM reduced glutathione, 50 U/ml 3-phosphoglycerate kinase, 50 U/ml glyceraldehyde-3-phosphate dehydrogenase, 400U/ml triose phosphate isomerase and 50 U/ml glycerol-3-phosphate dehydrogenase. The difference in absorbance at 340nm before and after the coupling reaction was measured. The activity was calculated from the UV absorption change and the ε value of NADH (6.22 mM⁻¹cm⁻¹).

2.3. Crystallization and data collection

Crystal screening was carried out using commercial crystallization kits by the sitting drop vapor diffusion method at 20°C. All the initial conditions were optimized using the hanging drop vapor diffusion method. Crystals that diffracted to the highest resolution were grown from the condition containing 26% PEG4000 (w/v), 200mM NaAc, 100mM Tris-HCl, pH8.0. Crystals reached a typical size of 10x30x30 μm³ one week later.

Before data collection, the crystals were soaked in cryoprotectant solution (well solution : glycerol = 4:1) containing 10mM RuBp for 1-30 minutes. X-ray diffraction data were collected at the Cornell High Energy Synchrotron Source (CHESS) on beamline F1 (Pilatus 6M detector) at 100 K. The diffraction data were processed using the HKL package¹⁷; statistics of data collection and processing from selected crystals are summarized in Table 1.

Table 1:

Statistics of diffraction data collection and structure refinement

Data collection
Space group	P21
Cell dimensions
a/b/c (Å)	80.33/116.21/83.29
β (°)	96.32
Resolution (Å)	50-1.7 (1.73-1.70)*
Unique reflections	166579 (8243)
Redundancy	3.4 (3.3)
Completeness (%)	100.0 (99.9)
Average I/σI	16.50 (2.09)
Rmerge	0.062 (0.377)
Rmeas	0.074 (0.452)
Rpim	0.040 (0.247)
CC1/2	0.993 (0.836)
CC*	0.998 (0.954)
Refinement
Rwork/Rfree	0.1438/0.1774
RMSD
Bond length (Å)	0.007
Angle (°)	0.929
Ramachandran statics
Favored regions (%)	96.95
Allowed regions (%)	2.87
Outliers (%)	0.17
PDB ID	8DHT

^*Values in parentheses are for the highest-resolution shell.

2.4. Crystal structure determination

The crystal structure of afRubisco was solved by molecular replacement using the crystal structure of tkRubisco (PDB code: 3KDO) as a search model. Refinement was performed using Phenix_refine¹⁸. Model building was performed using Coot¹⁹. Statistics of structure refinement are summarized in Table 1.

3. Results

3.1. AfRubisco is a bona fide Rubisco

The archaea Archaeoglobus fulgidus contains two Rubisco-like genes, a.fulgidus 1 and a.fulgidus 2 (AF_1638). It has been suggested that both belong to the Rubisco-like protein (RLP) subgroup (Form IV)⁵, but sequence comparisons indicate that a.fulgidus 2 may instead be a Form III Rubisco²⁰. The deduced amino acid sequence of the protein product of a.fulgidus 2 (Af-Rubisco) has 72% identity with the Form III Rubisco from T. kodakarensis (tkRubisco) and has 37% identity with the Form I Rubisco from spinach (spRubisco) (Fig. S1). Hanson and Tabita proposed 19 amino acid residues of photosynthetic Rubisco as essential for catalysis²⁰. AfRubisco has 18 (E46, T51, N108, K160, K162, G181, D183, K186, D188, E189, H278, R279, H311, K319, S364, G366, G388 and G389) of the 19 residues, differing only at Y184 (corresponding to F198 in spinach Rubisco). AfRubisco and tkRubisco are identical at all 19 of these positions, and tkRubisco has been shown to have carboxylase activity⁸. This suggests that afRubisco may also be a functional Rubisco. Therefore, we determined the carboxylase activity of afRubisco. Activated afRubisco has very low carboxylase activity at room temperature but has much higher carboxylase activity at 65°C (0.096μmol of CO₂ fixed/min/mg of protein). This may be explained by the fact that the optimal growth temperature for Archaeoglobus fulgidus is around 83° C. For the similar tkRubisco, its carboxylase activity at its optimal temperature (100° C) is 160 times higher than that at 25°C⁸.

3.2. Overall structure of afRubisco

The crystal structure of afRubisco has been determined at 1.7Å using molecular replacement method. The final model contains all 441 residues except the first one. The N-terminal region, the C-terminal region and “loop6” (the loop connecting β-strand 6 and α-helix 6 in the C-terminal α/β barrel) are all well-defined with strong electron density. AfRubisco has the typical Rubisco large subunit fold that comprises a small N-terminal domain (residues 2-132) and a large C-terminal domain (residues 133-441) (Fig. 2a). The N-terminal domain has a conserved ferredoxin-like fold, and the C-terminal domain has a TIM (triosephosphate isomerase) barrel fold.

Figure 2. Crystal structure of afRubisco.

(a) AfRubisco contains a large C-terminal domain and a small N-terminal domain. AfRubisco exists as a dimer in solution, and the active site is located at the interface between the monomers (black ellipse). The flexible fragments (N-terminal, C-terminal and loop6) that will move when the active site closes and reopens are highlighted in yellow. (b) Superposition of afRubisco (green) on tkRubisco (magenta). The rmsd between alpha carbons is 0.53Å.

AfRubisco exists as a dimer in solution and in the crystal. The two active sites are located at the interface between the large domain of one subunit and the small domain of the other subunit (Fig. 2a). The structure of afRubisco is of the closed form of Rubisco, with the C-terminal domain and loop6 from one subunit and the N-terminal domain from the other subunit folding onto the active site. Superimposition with the structure of tkRubisco/2CABP shows only small differences between the two structures, with an rmsd of 0.53Å between main chains (Fig. 2b & Fig. S2). The major differences are found in the N-terminal region: the first seven residues of tkRubisco are disordered, while only the first residue of afRubisco is missing; and tkRubisco has an inserted residue in the first loop (around residue 20).

The most notable feature of the new structure is the two 3PGA molecules bound in the active site (Fig. 3a).

Figure 3. The active site of afRubisco.

(a) The carbamylated lysine is shown as Kcx186. The Mg²⁺ is hexagonally coordinated by carbamylated Kcx186, D188, E189 and the two 3PGA molecules. The coordination between the Mg²⁺ and the ligands are shown as red dashed lines with length in Angstrom. (b) Superposition of tkRubisco/2CABP (magenta) and afRubisco/3PGA (cyan) shows that the side chains of both E46 and T51 of afRubisco form H-bonds with the ligand, while the side chains of the corresponding residues in tkRubisco do not. The H-bonds are shown as red dashed lines with length in Angstrom. (c) Superposition of afRubisco/3PGA (green) with tkRubisco/2CABP (3DKN; yellow), spRubisco/3PGA (1AA1; cyan), crRubisco/2CABP (1RK8; magentas), spRubisco/RuBp (1RXO; blue) and rrRubisco/RuBp (9RUB; red) shows different conformations of residues corresponding to E46 and T51.

3.3. Active site of afRubisco

As shown in Figure 3a, residue Lys186 has been carbamylated and stabilized by a Mg²⁺ ion. Its side chain carboxyl group has H-bond interactions with both 3PGA molecules. The Mg²⁺ is hexagonally coordinated by carbamylated K186 (named as Kcx186), D188, E189 and the two 3PGA molecules. Superposition with the structure of tkRubisco/2CABP complex (PDB ID: 3KDN) shows that the 19 essential residues, and other residues of the active site, are almost all at the same position, with the same conformation, except for E46 and T51 (Fig. 3b). The side chain of E46 forms two H-bonds with the carboxyl group of p1 3PGA in afRubisco, while it is distant (>3.77Å) from the carboxyl group of the inhibitor 2CABP in tkRubisco. The side chain of T51 forms a H-bond with the phosphate group of p1 3PGA in afRubisco, while it is distant (>4.15Å) from the phosphate group of the inhibitor 2CABP in tkRubisco. Two residues (D188 and E189) in the Rubisco motif, in addition to coordinating Mg²⁺, have H-bond interactions with 3PGA (Fig. 3a), similar to the interactions of the corresponding residues in tkRubisco with 2CABP. Most of the other residues in the Rubisco motif surround the ligands in the active site, with the exception of G181, D183, and Y184. The first two are distant from the active site and have no interactions with ligands. Tyr 184 (Tyr 187 in tkRubisco) corresponds to Phe in the canonical Rubisco motif. This residue does not interact with 3PGA or 2CABP; it resides in a local hydrophobic region and has only hydrophobic interactions with nearby hydrophobic residues, so the difference between Tyr and Phe is immaterial.

3.4. Binding of 3PGA at the active site

The most notable feature of this structure is that there are two 3PGA molecules binding at the pocket-shaped active site with strong and clear density (Fig. 4a). The 3PGA binding at the bottom of the pocket is assigned as p1 3PGA (at position 1), and the 3PGA close to the entrance of the pocket is assigned as p2 3PGA (position 2). These two 3PGA molecules have extensive interactions with protein, water and the Mg²⁺ ion (Table S1). Superposition with the structure of tkRubisco/2CABP shows that these two 3PGA molecules align well with the intermediate analogue 2CABP (Fig. 4b & Fig. S3). For comparison purposes, we neglect the C2-C3 bond of 2CABP and assign the C1-C2 part of 2CABP as part 1 (p1 2CABP) and the C3-C5 part of 2CABP as part 2 (p2 2CABP). p1 3PGA matches well with p1 2CABP, except for the C2 atom which has a D configuration in 3PGA, but an L configuration in p1 2CABP. Both p1 3PGA and p1 2CABP provide two coordinating oxygens to the Mg²⁺ through the carboxyl group and the hydroxyl group, respectively. The interactions between p1 3PGA and protein are the same as those of p1 2CABP except for residue E46 (E49 in tkRubisco) which forms a H-bond with the carboxyl group of 3PGA in afRubisco but not with the carboxyl group of 2CABP in tkRubisco. P2 3PGA and p2 2CABP match well and provide one coordinating oxygen to the Mg²⁺. Both the C2 atom in p2 3PGA and the corresponding atom in p2 2CABP are in the D configuration. The C1 end of p2 3PGA is a carboxyl group while the corresponding part of p2 2CABP is a hydroxyl group; thus, p2 3PGA has three more H-bond interactions, two with the protein (S364 and Kcx186) and one with p1 3PGA (Table S1).

Figure 4. Binding of 3PGA at the active site.

(a) A 2Fo-Fc omit electron density map (contoured at 2.0σ) clearly shows that two 3PGA molecules are bound to the active site. (b) Superposition of afRubisco/3PGA (green) and tkRubisco/2CABP (magenta) shows that the two 3PGA molecules overlap well with the 2CABP molecule. (C) Superposition of afRubisco/3PGA (green) and spRubisco/3PGA (magenta) shows that the two p1 3PGA molecules overlap with each other very well, while p2 3PGA and p3 3PGA have almost no overlap.

The crystal structure of spinach Rubisco (spRubisco) has been solved in complex with two 3PGA molecules by soaking at a high concentration of 3PGA (100 mM) (PDB ID: 1AA1)²¹. Superposition of afRubisco with the structure of this spRubisco shows that the p1 3PGA molecules align well, but the p2 3PGA molecules have almost no overlap (Fig. 4c). This spRubisco structure is of the open form. Its p2 3PGA is far away from the Mg²⁺ and very close to the exit of the pocket. Therefore, we re-assign this 3PGA as p3 3PGA, so that it will not be confused with the p2 3PGA in afRubisco. The interactions between the p1 3PGA and the protein are almost the same in afRubisco and spRubisco with a few differences: in afRubisco loop6 folds onto the active site so that the side chain of K319 has two H-bonds with p1 3PGA, while in spRubisco loop6 is placed so that the residue corresponding to K319 is not near the active site; in afRubisco the side chain of E46 has two H-bonds with p1 3PGA, while in spRubisco the corresponding residue (E60) is distant (5.73Å) from p1 3PGA; in afRubisco p1 3PGA has three H-bonds with p2 3PGA, while in spRubisco there is no interaction between p1 3PGA and p3 3PGA. Therefore, the interaction between p1 3PGA and the protein is stronger in afRubisco than in spRubisco. In contrast to p2 3PGA in afRubisco, p3 3PGA in spRubisco is flipped almost 180° about the phosphate group (Fig. 4c), so that the carboxyl group of p3 3PGA is pointing to the exit of the pocket, rather than towards the Mg²⁺. Because p2 3PGA and p3 3PGA bind to different positions at the active site, their interactions with the protein are almost all different, except for the two H-bonds with R279 (R295 in spRubisco). The interaction between p2 3PGA and afRubisco is much stronger than the interaction between p3 3PGA and spRubisco, because of the greater number of protein-ligand H-bond interactions.

Discussion

Due to its importance, Rubisco has been investigated extensively by genetic, microbiology, molecular biology, biochemistry, evolution and structural methods. The carboxylation catalyzed by Rubisco involves multiple discrete steps and corresponding intermediates⁵. Crystal structures of complexes of Rubisco with intermediate analogues will provide valuable information about the mechanism of the catalytic reaction^8,15,21–23. However intermediate analogues for most reaction steps that are stable and suitable for crystallographic study have not yet been found. Today more than one hundred crystal structures of Rubisco and Rubisco-ligand (substrate, product, inhibitors, positive effectors) complexes have been determined, and most of them are in complex with the inhibitor 2CABP that mimics the six-carbon intermediate 3-keto-2′-carboxyarabitol 1,5-bisphosphate⁸. Due to its stability and tight binding to Rubisco, 2CABP is ideal for crystallographic study. In contrast, the natural substrate RuBp is not stable at the active site. In order to obtain a complex with RuBp, the Mg²⁺ of Rubisco must be displaced by other metal ions such as Ca²⁺ (15) or Zn²⁺ (unpublished data). For those ligands that bind less tightly at the active site, such as the product 3PGA, it is difficult to obtain crystals of the complex. To date, there is only one crystal structure of activated Rubisco in complex with 3PGA in the PDB, in which the complex was obtained by soaking the apo-crystals in a high concentration of 3PGA²¹. In this spRubisco/3PGA structure, two 3PGA molecules are bound at the active site: one binds tightly and coordinates directly to the Mg²⁺, while the other binds loosely and is far from the Mg²⁺. In the afRubisco/3PGA crystal structure reported here, activated afRubisco was soaked in the substrate RuBp. Both 3PGA molecules bind tightly at the active site and coordinate the Mg²⁺.

Comparison of the structure of afRubisco/3PGA with structures of spRubisco/3PGA and other Rubisco-ligand complexes shows that the major differences in interaction between protein and ligand are located at residues E46, T51 and S364 (Fig 3c & Fig S3c). Residue E46 and the corresponding residues in other Rubiscos are located at the carboxyl end of an α-helix with their main chains adopting nearly same conformation. However, the conformation of the side chain of E46 is different from those of corresponding residues of other Rubiscos, flipping nearly 180° compared to the side chain of the corresponding residue of tkRubisco/2CABP. Hence the side chain of E46 is much closer to the ligand and forms two H-bonds with the carboxyl group of p1 3PGA. In contrast, the side chains of the corresponding residues of other Rubiscos are too far away from the ligands to form any H-bond with the bound ligands. Residue T51 is located in a loop with slight flexibility, and there is significant difference among the conformations of T51 and the corresponding residues of other Rubiscos (Fig 3c). The side chain of T51 forms one H-bond with the p1 phosphate group of the ligand 3PGA. The side chain of residue T65 in Chlamydomonas reinhardtii Rubisco (crRubisco, pdb ID: 1GK8) adopts a conformation similar to T51 of afRubisco and forms one H-bond with the ligand 2CABP. The side chains of the corresponding residues of other Rubiscos flip away from the ligands and are too far away from the ligands to form any H-bond with the ligands. These three additional H-bonds make the p1 3PGA bind more tightly to the active site of afRubisco. The difference at S364 is due not to the protein - S364 and the corresponding residues of other Rubiscos adopt nearly the same conformation – but to the bound ligands. Only the side chain of afRubisco S364 forms a H-bond with the bound ligand p2 3PGA (Fig S3c); because 2CABP doesn’t have an electron rich donor atom at this site and the p3 3PGA of spRubisco/3PGA is too far away, neither 2CABP nor the p3 3PGA can form any H-bond with the residues corresponding to S364.

Crystal structures of Rubisco and Rubisco-ligand complexes adopt two different conformation states, designated the open and closed forms, (Table 2). Transition of the structure from the open state to the closed state involves serial movements: rotation of the small domain towards the large domain; folding of loop6 over the active site; packing of the C-terminal fragment onto loop 6; and ordering of the N-terminal fragment. The timing and mechanism of the transition between the two states are not yet known in detail. The complex of activated Rubisco with 2CABP is of the closed form, while the complex of activated Rubisco with the substrate RuBp is of the open form. Therefore, the binding of the substrate alone can’t induce the closure of the active site; the addition of CO₂ to form the six-carbon intermediate is required to trigger the closing. There is no structural evidence about when the active site reopens, although it is speculated that this occurs after the cleavage of the C2-C3 bond. In this report, the crystal structure of afRubisco/3PGA has been determined to be of the closed form (Fig. 2a). The two 3PGA molecules mimic the status right after the cleavage of C2-C3 bond. Superposition of the structure of afRubisco/3PGA with the structure of tkRubisco/2CABP shows that the two 3PGA molecules overlap well with 2CABP. The distances from the Mg²⁺ to C2 and C3 are 2.79Å and 3.12Å, respectively, in afRubisco/3PGA, while they are 2.82Å and 3.00Å in tkRubisco/2CABP. The two phosphates of the two 3PGA molecules bind at the same sites as the P1 and P2 phosphate groups of the 2CABP molecule in tkRubisco/2CABP, but their positions are slightly shifted: the distances from the Mg²⁺ to P1 and P2 are 6.44Å and 7.57Å, respectively, in afRubisco/3PGA, while they are 6.39Å and 7.41Å in tkRubisco/2CABP. The C2-C3 distance is 1.35Å in tkRubisco/2CABP (carbon-carbon single bond) and 3.22Å in afRubisco/3PGA, indicating that the carbon-carbon single bond is broken in afRubisco/3PGA. This result suggests that the active site doesn’t reopen simultaneously with the cleavage of the C2-C3 bond, but rather after both products are formed. The driving force behind the reopening of the active site could be provided by the additional entropic freedom of motion gained by the cleavage of the substrate²¹.

Table 2:

Conformation of Rubisco and Rubisco-ligand complex

Rubisco or complex	Source	PDB ID	Conformation	P1-P2 distance (Å)
Apo activated Rubisco	Different sources		open	N/A
Activated Rubisco-2CABP	T. kodakarensis	3DKN	closed	8.81
Activated Rubisco-NADPH (or 6PG)	Rice	3AXK	open	N/A
Activated Rubisco-3PGA	Spinach	1AA1	open	9.70
Activated Rubisco-3PGA	A. fulvidus	(this paper)	closed	9.29
Activated Rubisco-CA2+-RuBp	Spinach	1RXO	open	9.4
Non-activated Rubisco-RuBp	Spinach	1RCX	closed	9.1
Non-activated Rubisco-XuBp	Synechococcus	1RSC	closed	8.84

Duff et al. found that the P2 phosphate group binding site consists of two subsites, and that the P2 phosphate group of all closed state structures occupies the lower subsite, while the P2 phosphate group of all open state structures is in the upper subsite²⁴. The P1 phosphate group has three sub-binding-sites, and in all closed state structures the P1 phosphate group occupies the proximal subsite (the position closest to the P2 site), while in open structures it is in the distal subsite. Examination of the H-bonds between 3PGA and protein in afRubisco/3PGA shows that the P2 phosphate group is in the lower subsite, making two H-bonds with R279 and a third one with H311; while the P1 phosphate group is in the proximal subsite, making H-bonds to the backbone amides of G366, G388 and G389 and a H-bond to the side chain of T51. The structure of afRubisco is of closed state, which is consistent with Duff et al.’s finding. Duff et al. suggested the distance between the P1 and P2 phosphate groups signals the transition between open and closed status of the active site. If the P1-P2 distance is more than 9.4Å, the protein is of open conformation; while the protein is of closed conformation if the P1-P2 distance is less than 9.1Å (24) (Table 2). The P1-P2 distance in afRubisco/3PGA is 9.29Å, and the protein is of closed form, supporting the correlation between open vs. closed form and P1-P2 distance but raising Duff et al.’s 9.1 Å upper limit for the closed form to 9.3 Å.

All known products of the carboxylation catalyzed by Rubisco are D-3PGA, which means that the stereochemistry around C2 is reversed during the catalysis reaction. After carbon-carbon bond cleavage in the six-carbon intermediate, a C2 carbanion is formed. A stereospecific protonation of this carbanion is required to produce only D-3PGA, i.e. the attacking proton can come from only one direction. It is supposed that in spinach Rubisco it is K175 donating a proton to the carbanion²¹. In afRubisco, the corresponding residue is K160 and it is well-positioned to donate that proton. This residue is conserved among almost all Rubiscos (Fig. S3a).

Conclusion

The structure of afRubisco/3PGA, determined to 1.7 Å, provides strong support for its identification as a genuine Type III Rubisco. Given the similarity of its active site to that of other Rubiscos, the observation of two 3PGA molecules in the closed form of afRubisco implies that the common mechanism of Rubisco catalysis involves cleavage of the C2-C3 bond of RuBp prior to opening of the site and release of the products.

Abbreviations

Rubisco

ribulose 1,5-bisphosphate carboxylase/oxygenase

3PGA

3-phospho-D-glycerate

2CABP

2-carboxy-D-arabinitol 1,5-bisphosphate

RuBp

D-ribulose 1,5-bisphophate

XuBp

D-xylulose 1,5-bisphosphate