Structure of the RuBisCO chaperone RbcX from the thermophilic cyanobacterium Thermosynechococcus elongatus

The crystal structure of the RuBisCO assembly chaperone RbcX from a thermophilic cyanobacterium has been determined at 1.7 Å resolution. The dimeric structure is capable of a hinge movement (probably connected with binding of the RuBisCO large subunit) pivoted on a kink in two long antiparallel α-helices.

Abstract

The crystal structure of TeRbcX, a RuBisCO assembly chaperone from the cyanobacterium Thermosynechococcus elongatus, a thermophilic organism, has been determined at 1.7 Å resolution. TeRbcX has an unusual cysteine residue at position 103 that is not found in RbcX proteins from mesophilic organisms. Unlike wild-type TeRbcX, a mutant protein with Cys103 replaced by Ala (TeRbcX-C103A) could be readily crystallized. The structure revealed that the overall fold of the TeRbcX homodimer is similar to those of previously crystallized RbcX proteins. Normal-mode analysis suggested that TeRbcX might adopt an open or closed conformation through a hinge movement pivoted on a kink in two long α4 helices. This type of conformational transition is presumably connected to RbcL (the large RuBisCO subunit) binding during the chaperone function of the RuBisCO assembly.

1. Introduction

Ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO), an enzyme that accounts for 30–50% of the total soluble protein in the chloroplast, is the most abundant protein on Earth (Dhingra et al., 2004 ▶). It fixes inorganic carbon dioxide, which enters the biosphere, and is therefore regarded as one of the most important proteins (Ellis, 1979 ▶).

RuBisCO form I, which is found in plants, green algae and cyanobacteria, is composed of eight large subunits (RbcL) and eight small subunits (RbcS), which are encoded by two genes, rbcL and rbcS, respectively. Chaperone factors enabling polypeptide-chain folding, as well as specific assembly chaperones (Ellis, 2006 ▶), are required for the RuBisCO assembly process (Andersson & Taylor, 2003 ▶; Tabita, 1999 ▶). Interestingly, in some cyanobacteria an extra gene, rbcX, is located between the rbcL and rbcS genes. Owing to its localization inside the RuBisCO operon, it has been suggested that the RbcX protein may act as a chaperone during RuBisCO assembly. Indeed, co-expression of rbcX with the rbcL and rbcS genes enhanced the formation of active RuBisCO from Anabaena sp. CA (Li & Tabita, 1997 ▶), Synechococcus sp. PCC 7002 (Onizuka et al., 2004 ▶) and Thermosynechococcus elongatus (Tarnawski, Gubernator et al., 2008 ▶). Thus, it seems that the presence of the rbcX gene within or outside the RuBisCO operon is closely related to its significance in RuBisCO assembly. In the case of Synechococcus sp. PCC 7942, in which the rbcX gene is located about 100 kb away from the rbcLS operon, insertional inactivation of the rbcX gene in the cyanobacterium showed no perturbation of the growth rate or RuBisCO content and activity (Emlyn-Jones et al., 2006 ▶), in apparent contrast to the situation in Synechococcus sp. PCC 7002 (Onizuka et al., 2004 ▶).

The RbcX proteins show a high level of sequential and structural conservation, invariably forming dimeric assemblies, which suggests that their mechanism of action must also be similar. Recently, it has been shown that folding of the RbcL subunit by the GroEL/GroES chaperonin is tightly coupled to an assembly process mediated by the RbcX chaperone. Further structural analysis of an RbcL₈–RbcX₈ assembly intermediate by cryo-electron microscopy provided a detailed insight into the mechanism of RuBisCO holoenzyme formation (Liu et al., 2010 ▶). In particular, Saschenbrecker et al. (2007 ▶) showed that RbcX acts as a specific RuBisCO assembly chaperone to promote the formation of the (RbcL₂)₄ core complex. The authors presented evidence that a specific C-terminal fragment of RbcL (E₄₅₉IKFEFD₄₆₅) binds in a central groove of the RbcX molecule. This fragment adopts an extended conformation, with the polar side chains facing outwards and two large hydrophobic side chains of phenylalanine extending into cavities formed by the side chains of Thr13, Tyr17, Tyr20, Ile50 and Gln51.

The crystal structures of RbcX from three mesophilic cyanobacteria, namely Synechococcus sp. PCC 7002 (Saschenbrecker et al., 2007 ▶), Anabaena sp. CA (Saschenbrecker et al., 2007 ▶) and Synechocystis sp. PCC 6803 (Tanaka et al., 2007 ▶), have been solved. Regardless of the species, each RbcX protein is composed almost exclusively of α-helices, which form an unusual four-helix bundle. In addition, all known RbcX proteins exist as homodimers.

The main goal of the present project is to elucidate the structure of the RbcX protein from the thermophilic cyanobacterium T. elongatus (TeRbcX). Amino-acid sequence analysis of the RbcX proteins highlighted the presence in the TeRbcX sequence of an unusual single Cys residue at position 103 that is only found in two other thermophilic RbcX proteins: those from Phormidium laminosum and T. vulcanus (Tarnawski, Krzywda et al., 2008 ▶). This cysteine residue maps onto a surface Arg of the known RbcX crystal structures (PDB entries 2peq, 2peo and 2py8; Saschenbrecker et al., 2007 ▶; Tanaka et al., 2007 ▶) but at a position that precludes its biological role. However, recombinant TeRbcX could not be crystallized owing to uncontrolled aggregation via intermolecular disulfide-bond formation (Tarnawski, Krzywda et al., 2008 ▶). Recently, we have constructed and crystallized two TeRbcX variants, which form the basis for the present structural analysis (Tarnawski, Krzywda et al., 2008 ▶). Whereas Cys103 pre­vented the crystallization of wild-type TeRbcX even in the presence of reducing agents, the C103A (TeRbcX-C103A) and C103R mutants did not form higher-order aggregates and could be readily crystallized. Importantly, it was shown that TeRbcX-C103A could still form homodimers analogously to the previously crystallized native RbcX proteins from other species.

Here, we report the crystal structure of TeRbcX-C103A, which is the first example of an RbcX structure from a thermophilic organism. Since the presence of the native Cys103 hampered crystallization, we describe the structure of TeRbcX containing the site-specific mutation C103A. TeRbcX-C103A could be obtained in two polymorphic modifications and crystals of the orthorhombic form diffracted X-rays to high resolution (1.7 Å). This is the highest resolution structure of any RbcX protein available to date.

2. Experimental procedures

2.1. Mutagenesis, expression and purification

The coding sequence of the rbcX gene from T. elongatus was cloned into pET-16b vector (Novagen). The Cys103 residue was mutated to Ala (C103A) by means of the QuikChange (Stratagene) method. Recombinant TeRbcX protein was expressed at 310 K in Escherichia coli BL21 (DE3) cells and purified by ammonium sulfate fractionation, thermal fractionation, ion-exchange and size-exclusion chromatography as reported previously (Tarnawski, Krzywda et al., 2008 ▶).

2.2. Protein crystallization

Crystallization of TeRbcX mutant proteins using the hanging-drop vapour-diffusion method was performed as described previously (Tarnawski, Krzywda et al., 2008 ▶). High-quality single crystals of TeRbcX-C103A were obtained using a reservoir solution consisting of 0.1 M sodium acetate pH 4.5, 1.0 M 1,6-hexanediol and 0.01 M CoCl₂.

2.3. X-ray diffraction data collection

A single crystal was harvested in a cryoloop; it was briefly soaked in a cryoprotectant solution consisting of the reservoir solution with 1,6-­hexanediol added to a final concentration of 1.8 M prior to flash-vitrification in a stream of cold nitrogen gas. X-ray diffraction data were collected using synchrotron radiation on the BL14.1 beamline at the BESSY synchrotron (Berlin). The diffraction data were indexed, integrated and scaled using HKL-2000 (Otwinowski & Minor, 1997 ▶). The final statistics are summarized in Table 1 ▶.

Table 1. Data-collection and refinement statistics.

Values in parentheses are for the highest resolution shell.

Data collection
Crystal dimensions (mm)	1.3 × 0.5 × 0.5
Space group	P212121
Unit-cell parameters (Å)	a = 45.78, b = 67.73, c = 94.54
Mosaicity (°)	0.6
Protein molecules per asymmetric unit	2
Matthews coefficient VM (Å3 Da−1)	2.51
Solvent content (%)	51.1
Radiation source	BESSY BL14.1
Wavelength (Å)	0.9184
Temperature (K)	100
Detector	MX-225 CCD
Resolution range (Å)	50–1.71 (1.76–1.71)
No. of observed reflections	161694
No. of unique reflections	31245
Multiplicity	5.2 (2.9)
Completeness (%)	96.2 (71.8)
Rmerge† (%)	4.7 (43.0)
〈I/σ(I)〉	29.9 (2.2)
Refinement
No. of reflections in working set	29886
No. of reflections in test set	1280
Resolution range (Å)	41.20–1.71
No. of non-H atoms	2075
Protein	1843
Solvent	232
R‡ (%)	18.2
Rfree§ (%)	21.1
R.m.s. deviations from ideal
Bonds (Å)	0.018
Angles (°)	1.59
Mean B factor (Å2)	45.8
Protein	44.9
Solvent	52.8
Ramachandran statistics (%)
Most favoured regions	92.5
Additional allowed regions	7.5

^† R _merge = , where I_i(hkl) is the ith measurement of the intensity of reflection hkl and 〈I(hkl)〉 is the mean intensity of reflection hkl.

^‡ R = for all reflections, where F _obs and F _calc are the observed and calculated structure factors, respectively.

^§ R _free is calculated analogously to R for test reflections which were randomly selected and excluded from the refinement.

2.4. Structure solution and refinement

The structure was solved by molecular replacement by application of the program REMO (Burla et al., 2007 ▶; Caliandro et al., 2006 ▶) using previously reported diffraction data (data set A-1) in the 50–2 Å resolution range (Tarnawski, Krzywda et al., 2008 ▶) and the RbcX model from Synechococcus sp. PCC 7002 (PDB entry 2peq; Saschenbrecker et al., 2007 ▶). Subsequently, the solution was con­firmed in MOLREP (Vagin & Teplyakov, 2010 ▶) using a new data set with a resolution range of 50–1.7 Å. Model rebuilding and refinement were carried out using Coot (Emsley & Cowtan, 2004 ▶) and REFMAC5 (Murshudov et al., 2011 ▶). TLS parameters calculated by the TLS Motion Determination server (Painter & Merritt, 2006a ▶,b ▶) and bulk-solvent correction were applied towards the end of the refinement. The final structure of TeRbcX-C103A is characterized by an R factor of 18.2% and an R _free of 21.1% (Table 1 ▶).

Atomic coordinates and structure factors have been deposited in the PDB with accession code 3q20. Structural figures were generated with PyMOL (DeLano, 2002 ▶).

2.5. Normal-mode analysis

Putative conformational changes of TeRbcX were detected by elNémo, the web interface to The Elastic Network Model, a fast and simple tool for the computation of low-frequency normal modes of a protein (Suhre & Sanejouand, 2004 ▶). The following program settings were applied: NMODES = 5, DQMIN = −100, DQMAX = 0, DQSTEP = 20.

3. Results and discussion

3.1. Overall structure

The overall architecture of the RbcX protein from T. elongatus is similar to other known cyanobacterial RbcX structures. TeRbcX is nearly entirely α-helical (88.2%). The fold consists of four α-helices (α1–α4) per monomer which form a characteristic helix bundle (Fig. 1 ▶ a). The N-terminal helix α1 consists of residues Val3–Thr33 and is followed by helix α2 (Pro35–Glu48) in an antiparallel orientation. The α1–α2 turn is formed by only one residue (Asp34). The junction between helices α3 (Gly53–Glu63) and α4 (Pro65–Gln111) is arranged in a similar way. The relatively short helices α2 and α3 are connected by a loop-like linker, the S₄₉IQD₅₂ loop, with the Ile50-Gln51-Asp52 segment in a reverse γ-turn conformation. The longest C-­terminal helix α4 is bent in the middle and the residues Ala83-Asp84-Tyr85 form a 3₁₀-helical turn. The C-terminal helix α4 extends away from the compact core of the bundle, so that its C-terminal half (beyond the kink) has no packing interactions with the core.

Figure 1.

Three-dimensional structure of the RbcX protein from T. elongatus. (a) The four-helix-bundle fold of a protomer (chain A) in rainbow colours from blue (N-terminus) to red (C-terminus). (b, c) The TeRbcX dimer, with chain A shown in green and chain B shown in cyan: (b) top view, (c) side view. (d) Ball-and-stick representation of the region boxed in (c). The mutated residue C103A of chain A is shown as 2F _o − F _c electron density contoured at 1.0σ.

The final model consists of two polypeptide chains in the asymmetric unit that form a boomerang-like homodimer (Figs. 1 ▶ b and 1 ▶ c). The crystal structure is consistent with the expectation of a dimer as the biological unit. Electron-density maps show clear tracing of residues 1–114 of polypeptide chain A and residues 2–107 of chain B. Additionally, two partially occupied molecules of 1,6-hexanediol and two partially occupied acetate ions are present in the asymmetric unit.

The quality of the electron density is poorer in the Thr33–Asp52 fragment of molecule A, where it is continuous but lacks the main- and side-chain details that are clearly visible for the rest of the structure. In particular, electron density is completely missing for the side chains of residues Leu39, Gln43, Ser46, Gln47 and Ser49. This fragment covers residues of helix α2 (residues 33–48) and the loop between helices α2 and α3 (residues 49–52), which are implicated in important contacts with the C-terminal peptide of RbcL (Saschenbrecker et al., 2007 ▶). Interestingly, a very similar picture is observed for the corresponding fragments of the 2peo structure, which are the most mobile regions of the protein (Saschenbrecker et al., 2007 ▶). The corresponding fragment of the present molecule B is well ordered and the electron density is very clear. Stabilization is achieved through several hydrogen bonds to a crystallographic copy of molecule B. Four of the five side chains that are dis­ordered in molecule A contribute to the stabilization of molecule B by forming six protein–protein hydrogen bonds and three hydrogen bonds to water molecules.

As in the other RbcX structures, residues 115–126 are not visible in the electron-density maps. It has been shown that the removal of these poorly conserved C-terminal residues does not affect RbcX activity in RuBisCO assembly (Saschenbrecker et al., 2007 ▶).

The two four-helix bundles of the dimeric structure occupy opposite ends of the boomerang-shaped molecule. The α4 helices of the two protomers in a dimer are arranged in an antiparallel fashion. The α4 helix is so long that its C-terminal half reaches the opposite bundle, creating in reality a five-helix bundle with one noncovalent intermolecular component.

Helices α1 and α2 of both protomers outline a diagonal groove on the surface of the molecule which is implicated in binding of the C-­terminal peptide of RbcL during RuBisCO assembly. The loops between helices α2 and α3 form a well defined clamp-like structure constraining access to the central groove, which is filled with several water molecules. The cleft leading to the groove is less than 10 Å wide, but the internal diameter of the groove (about 15 Å) is sufficient for binding of a polypeptide chain in an extended conformation.

Within the dimer, the protomers are related by a pseudo-twofold axis (179.3°) that is approximately parallel to the crystallographic b axis. The two polypeptide chains forming the dimer are structurally very similar. The r.m.s.d. value for their backbone C^α atoms is 0.45 Å, with the largest deviations being located in the region of helix α2. The atoms of the S₄₉IQD₅₂ loops are characterized by relatively high displacement parameters and must therefore be considered as flexible elements of the structure (Fig. 2 ▶ a). It is of note that the side chains of the Ile50 residues in the two polypeptide chains adopt different conformations, thus breaking the twofold symmetry of the dimer. Also, rigid-body TLS analysis (Painter & Merritt, 2006a ▶,b ▶) identified helices α2 and α3 as mobile elements of the structure (Fig. 2 ▶ b).

Figure 2.

Flexibility of the TeRbcX structure. (a) B factors of backbone C^α atoms of the TeRbcX dimer shown using a colour gradient (blue, low B values; red, high B values) and a chain diameter proportional to B. The arrows indicate the S₄₉IQD₅₂ loop between helices α2 and α3. (b) TLS analysis of TeRbcX chain A using 19 partition groups shown in different colours. Chain B (single segment) is shown in grey. (b) was generated with the TLSMD server (Painter & Merritt, 2006a ▶,b ▶).

A cluster of six glutamine residues (95, 96, 100, 101, 104 and 105), the side chains of which adopt alternative conformations, is located within the long C-terminal helix α4. The role of this glutamine tract is not clear. The side chains of four of these glutamine residues (95, 96, 100 and 104) are clearly solvent-exposed, with average relative residue accessible surface areas of about 50%.

3.2. Structural similarities

All cyanobacterial RbcX proteins crystallized to date are structurally similar and share a common fold (Table 2 ▶). Most of the differences in the conformation of the polypeptide chain are observed in the region of helices α2 and α3, including the loop-like linker (Fig. 3 ▶). In addition, electron-density maps showed that, apart from the highly disordered C-terminus, the S₄₉IQD₅₂ loop region is the least well ordered fragment of the polypeptide chain (see above). It is of note that in the model of RbcX from Anabaena sp. CA (PDB entry 2peo; Saschenbrecker et al., 2007 ▶), the residues Lys48-Val49 (corresponding to the K₄₈VQD₅₁ fragment of the present protein) are missing altogether owing to an absence of detectable electron density.

Table 2. Sequence identity (%, upper triangle) and backbone r.m.s.d. values (Å, lower triangle) for cyanobacterial RbcX proteins from T. elongatus (PDB entry 3q20; this work), Synechococcus sp. PCC 7002 (2peq), Anabaena sp. CA (2peo) and Synechocystis sp. PCC 8603 (2py8).

Values were obtained with the SSM program (Krissinel & Henrick, 2004 ▶). The resolution of each crystal structure determination is specified in parentheses.

Structure	3q20, chain AB	2peq, chain AB	2peo, chain AB	2py8, chain BC
3q20, chain AB (1.71 Å)	—	49.3	63.5	44.7
2peq, chain AB (1.90 Å)	1.39	—	51.6	51.3
2peo, chain AB (2.50 Å)	1.69	1.64	—	46.8
2py8, chain BC (2.45 Å)	1.40	1.42	1.55	—

Figure 3.

Structural alignment of RbcX dimers from T. elongatus (this work; PDB entry 3q20, chains AB, blue), Synechococcus sp. PCC 7002 (2peq, chains AB, green), Anabaena sp. CA (2peo, chains AB, yellow) and Synechocystis sp. PCC 6803 (2py8, chains BC, red). The superpositions were calculated in SSM (Krissinel & Henrick, 2004 ▶) using C^α coordinates and the structure of TeRbcX as the target.

Based on multiple alignment of 21 cyanobacterial RbcX sequences, the ConSeq (Berezin et al., 2004 ▶) and ConSurf (Glaser et al., 2003 ▶; Landau et al., 2005 ▶) servers identified a number of con­served residues in the TeRbcX structure, mostly in the central region (Met1, Thr10, Tyr17, Thr19, Gln21, Val74, Arg75, Glu76, Pro87 and Asn98) of the molecule, but also in the peripheral ‘corner’ regions (Glu32, Thr33, Pro35, Arg70, Arg102, Glu107 and Arg108) located at opposite ends of the dimer. The conserved residues forming the central core region are responsible for the stabilization of the dimeric RbcX fold and those lining the bottom of the central groove are involved in binding of the C-terminal RbcL peptide, as shown by Saschenbrecker et al. (2007 ▶). In addition, some residues in the peripheral regions have been shown to be functionally critical (Saschenbrecker et al., 2007 ▶) and their importance in surface interactions with the N-terminal domain of the RbcL subunit has been established (Liu et al., 2010 ▶).

3.3. Water structure

The electron-density maps allowed the identification of 208 water molecules, including 109 with full occupancy. The present TeRbcX structure is distinguished from the other RbcX structures by a high number of water molecules per dimer. Comparison of the solvent contents and Matthews coefficients of the present structure (51.1% and 2.51 ų Da⁻¹, respectively) and 2peq (30.8% and 1.78 ų Da⁻¹, respectively), solved at resolutions of 1.71 and 1.90 Å, respectively, showed a higher water content in the TeRbcX crystal structure. It is of note that in the TeRbcX structure water molecules form well ordered clusters or water chains around the side chains of certain residues, e.g. Arg102(A) and Gln29(B), at the protein surface (Supplementary Fig. 1 ▶ 1).

3.4. Cys103 of RbcX from T. elongatus

In the sequence of wild-type TeRbcX a single cysteine residue is present at position 103. Since its presence hampered our initial crystallization trials (Tarnawski, Krzywda et al., 2008 ▶), we sub­stituted this residue with alanine, creating the TeRbcX-C103A mutant, which readily crystallized. The side chain of the mutated C103A residue, which is located on the surface of the C-terminal part of helix α4, is clearly visible in electron density (Fig. 1 ▶ d).

The presence of Cys103 in the TeRbcX protein is rather puzzling. It might appear that it would be preferable to avoid a thermolabile and helix-disfavouring Cys residue in a protein such as TeRbcX, but it is remarkable that cysteine is also present in this position in two other thermophilic RbcX proteins, namely those from P. laminosum and T. vulcanus. The occurrence of cysteine at position 103 is directly connected with the observed TeRbcX aggregation via intermolecular S—S bond formation. Its presence must be justified by its function, but since the role of Cys103 in thermophilic RbcX proteins is currently unknown this interesting aspect needs to be investigated further.

3.5. Structure stabilization

The structure of RbcX is mainly stabilized by hydrophobic interactions. The dimer interface is predominantly uncharged and hydrophobic. The central region of the molecule is formed by symmetrically disposed residues from helices α1 and α4 of both protomers (Fig. 4 ▶ a). The highly hydrophobic core of the four-helix bundle is formed by residues from all four helices (Fig. 4 ▶ b). A tight packing of conserved hydrophobic residues (Leu14, Tyr17, Leu18, Val82, Leu86, Pro87, Met89 and Leu90) at the central junction between the two protomers is necessary for integrity of the dimer (Saschenbrecker et al., 2007 ▶).

Figure 4.

Stereoviews of the hydrophobic interactions in the TeRbcX structure. (a) The central core region at the dimer interface formed by helices α1 and α4 of both protomers (chains coloured as in Fig. 1 ▶ b): 2F _o − F _c electron density contoured at 1.0σ. (b) The core of the four-helix bundle of chain A in a 2F _o − F _c electron-density map contoured at 2.0σ.

The dimer stability appears to be enhanced by polar interactions around Arg75, which forms hydrogen bonds to Thr19, Glu54 and Asn98 from the complementary protomer (Fig. 5 ▶ a). A similar network has been described as contributing to dimer stability of the 2peq structure (Saschenbrecker et al., 2007 ▶). The system of inter­actions formed by Arg75 of TeRbcX is much stronger, as it involves a salt bridge with Glu54, which replaces the contact with Ser97 reported for 2peq (Table 3 ▶). In the 2peq structure the Asp54 side chain is too short for effective interaction with Arg75. In TeRbcX, the Ser97 position of 2peq is occupied by Ala, which makes no important contacts with the neighbouring residues. Other polar contacts are formed by residues from the opposing helices α4 with the participation of an extended network of water molecules (Fig. 5 ▶ b). It should be noted that in the TeRbcX structure an additional salt bridge is formed involving the Asp66–Arg37 ion pair. It staples the helices α2 and α4 within the helix bundle, thus increasing its integrity.

Figure 5.

Hydrogen bonds stabilizing the dimeric structure of TeRbcX (a) around Arg75 and (b) between the antiparallel helices α4. Water molecules are shown as red spheres. 2F _o − F _c electron-density maps are shown with the contour level set to 1.5σ in (a) and 1.0σ in (b). In (b), the kink in the antiparallel helix α4 is clearly visible. The protomers of the TeRbcX dimer are coloured green (molecule A) and cyan (molecule B). Hydrogen bonds are shown as dashed lines.

Table 3. Hydrogen bonding at the Arg75 side chain in the 3q20 and 2peq structures as calculated by WHAT IF (Hooft et al., 1996 ▶).

Distances are given in Å. Unrealistically short interatomic distances are indicated in italics.

	Chain A						Chain B
3q20	Arg75	N∊	Asn98	Oδ1	B	2.86	Arg75	N∊	Asn98	Oδ1	A	2.83
		Nη1	Glu54	O∊2	A	2.56		Nη1	Glu54	O∊2	B	2.72
		Nη1	Thr19	Oγ1	A	2.86		Nη1	Thr19	Oγ1	B	2.83
		Nη2	Glu54	O∊1	A	3.01		Nη2	Glu54	O∊1	B	3.14
		Nη2	Asn98	Oδ1	B	2.86		Nη2	Asn98	Oδ1	A	2.86
2peq	Arg75	N∊	Thr19	Oγ1	A	3.20	Arg75	N∊	Asn98	Oδ1	A	2.88
		Nη1	Ser97	Oγ	B	3.40		Nη1	Wat20	O	—	2.68
		Nη1	Asn98	Oδ1	B	3.24		Nη1	Thr19	Oγ1	B	2.92
		Nη2	Wat11	O	—	2.33		Nη2	Ser97	Oγ	A	2.90
		Nη2	Thr19	Oγ1	A	2.92		Nη2	Asn98	Oδ1	A	2.83

The organization of the structural elements in the RbcX dimer resembles the situation arising upon three-dimensional domain swapping (Bennett et al., 1995 ▶; Gronenborn, 2009 ▶; Jaskólski, 2001 ▶). However, in its bona fide meaning the term ‘three-dimensional domain swapping’ refers to cases in which the monomeric fold is recreated from elements contributed by two or more protein chains. Since a monomeric RbcX protein, in which helix α4 would fold upon itself (i.e. with the C-terminal half packed against the N-terminal half), is not known to exist, the term ‘three-dimensional domain swapping’ can only be used in a figurative sense in this case.

3.6. RbcX hinge regions

Upon binding to another molecule, protein motions that have functional importance may occur, including in particular motions of flexible hinge elements. Using the HingeProt server employing elastic network models (Emekli et al., 2008 ▶), we have identified a likely hinge region in all known RbcX structures. The hinge residue is Tyr85 of TeRbcX; Phe is present at the corresponding position in the other proteins. Similar results were obtained using the StoneHinge approach (Keating et al., 2009 ▶), which identified hinge regions around residues Leu86-Pro87 (chain A) and Asp84-Tyr85 (chain B) together with two hinge domains.

The dynamics and flexibility of TeRbcX was further examined by normal-mode analysis carried out using the ElNémo server (Suhre & Sanejouand, 2004 ▶). We found that one dominant mode described most of the conformational changes of the hinge motion. This hinge movement follows the first nontrivial lowest frequency normal mode (mode 7) with the perturbation set to 100 arbitrary units from the crystallographic state. Mode 7 has a collective character, as expected for functional conformational changes. The movement at the hinge element results in opening and closing of the binding groove of TeRbcX. Upon this ‘butterfly-type’ motion, which is pivoted on the kinks in helices α4, major rearrangements occur at the sites involved in RbcL binding. According to the obtained results, it appears that the RbcX protein can adopt two conformations. The present TeRbcX crystal structure is in fact very similar to the experimental ligand-bound form (PDB entry 2pem). In such a closed conformation the Ile50 side chains of the S₄₉IQD₅₂ loops of molecules A and B are about 10 Å apart, whereas in the open conformation resulting from the normal-mode analysis this distance is dramatically increased (Fig. 6 ▶). There is a good correlation between computed (derived form normal-mode analysis) and observed (crystallographic) B factors. The correlation factor of 0.749 for 221 C^α atoms (computed for 100 lowest frequency normal modes) indicates that the normal modes are a realistic description of the protein flexibility. An animation showing the structural changes leading from the closed to the open conformation is available as Supplementary Fig. 2 ▶ 1. It can be speculated that the closed–open conformational changes occur upon binding and release by TeRbcX of the C-terminal fragment of RbcL.

Figure 6.

RbcX from T. elongatus (PDB entry 3q20; red) and the open (yellow) conformation as modelled by ElNémo (Suhre & Sanejouand, 2004 ▶). This enlarged view shows the S₄₉IQD₅₂ loops and the Ile50 side chain (ball-and-stick representation).

4. Conclusions

In general, the present structure of the first RbcX protein from a thermophilic organism resembles those of previously described RbcX proteins from mesophilic species, implying a dimer as the minimal structural unit in the functional role of the protein as a chaperone in the assembly of the RbcL₈ module of the RuBisCO complex. The C ₂ dimeric structure of TeRbcX is supported by extended hydrophobic interactions between the two four-helix-bundle cores and by the exchange of the C-terminal parts of the long α4 helices between the two subunits.

The TeRbcX protein contains an unusual and unique Cys103 residue in its sequence. However, the location of this cysteine residue on the protein surface, in the distal part of helix α4, excludes its involvement in physiological dimerization or in interactions with the cognate C-terminal peptide of RbcL. The TeRbcX-C103A mutant remains functionally competent in the assembly of RbcL₈ in E. coli co-expression (data not shown).

Comparison with other RbcX models suggests that the TeRbcX dimer can adopt an open conformation with a very wide entrance to the substrate-binding groove that is formed at the dimer interface. It can be predicted that in order to facilitate substrate docking the RbcX molecule assumes a fully open conformation and then undergoes a conformational change through a hinge movement pivoted in the middle of the long α4 helices. It might thus be concluded that the RbcX molecule is capable of closing–opening motions upon ligand binding.

Supplementary Material

PDB reference: RbcX, 3q20

Animated version of Supplementary Fig. 2(a).. DOI: 10.1107/S1744309111018860/wd5153sup2.gif

Animated version of Supplementary Fig. 2(b).. DOI: 10.1107/S1744309111018860/wd5153sup3.gif

Animated version of Supplementary Fig. 2(c).. DOI: 10.1107/S1744309111018860/wd5153sup4.gif