|Home | About | Journals | Submit | Contact Us | Français|
Chaperonins are large ATP-driven molecular machines that mediate cellular protein folding. Group II chaperonins use their “built-in lid” to close their central folding chamber. Here we report the structure of an archaeal group II chaperonin in its prehydrolysis ATP-bound state at subnanometer resolution using single particle cryo-electron microscopy (cryo-EM). Structural comparison of Mm-cpn in ATP-free, ATP-bound, and ATP-hydrolysis states reveals that ATP binding alone causes the chaperonin to close slightly with a ~45° counterclockwise rotation of the apical domain. The subsequent ATP hydrolysis drives each subunit to rock toward the folding chamber and to close the lid completely. These motions are attributable to the local interactions of specific active site residues with the nucleotide, the tight couplings between the apical and intermediate domains within the subunit, and the aligned interactions between two subunits across the rings. This mechanism of structural changes in response to ATP is entirely different from those found in group I chaperonins.
Maintaining proper protein folding and homeostasis is critical for the growth and survival of living cells (Dobson, 2004). Failure in these processes in human is implicated in many diseases such as type II diabetes, cancer, neurodegeneration, and heart diseases (Balch et al., 2008). Chaperons are a group of molecular machines that help maintaining correct folding and the clearance of misfolded proteins by binding to nonnative polypeptides and facilitating their folding in the cell (Hartl and Hayer-Hartl, 2002). Among these are the ATP-driven group II chaperonins from eukaryotes and archaea that bear unique structural and functional characteristics (Cong et al., 2010; Ditzel et al., 1998; Zhang et al., 2010). The eukaryotic group II chaperonin TRiC/CCT is required to fold about 10% essential proteins newly synthesized from the ribosomes including tumor suppressors, cell cycle regulators, cytoskeletal proteins (Yam et al., 2008), and β sheet proteins (Knee et al., 2011). More intriguingly, a subset of its substrates, such as actin, can only be folded by TRiC but not other chaperon systems (Chen et al., 1994; Spiess et al., 2004), which implies its unique structural features and underlying mechanisms in protein folding.
Group II chaperonins, ~1 megadalton in size, are composed of two back-to-back rings with eight to nine subunits in each ring. Each subunit consists of three domains: a substrate-binding apical domain, an ATP-binding equatorial domain and an intermediate domain that connects the apical and equatorial domains. Chaperonin-assisted substrate folding is closely associated with its ATP-driven conformational changes. Group II chaperonins possess a long flexible helical protrusion at the tip of each apical domain, which can be transformed into a β-iris and serves as a “built-in lid” (Cong et al., 2010; Ditzel et al., 1998; Zhang et al., 2010). This unique feature allows them to form a closed folding chamber without cochaperons such as the GroES for the group I chaperonins (Xu et al., 1997). Little is known about the detailed molecular mechanism how the lid in the group II chaperonins is driven to close by the ATP, because of a lack of high resolution structures at different states during the folding cycle.
We use the 16 subunit homo-oligomeric chaperonin from archaea Methanococcus maripaludis (Mm-cpn) to investigate the structural features of group II chaperonins. Like all other group II chaperonins, Mm-cpn folds proteins in an ATP-dependent manner. It shares similar allosteric regulation properties of eukaryotic chaperonins such as TRiC (Kusmierczyk and Martin, 2003; Reissmann et al., 2007) while its homo-oligomeric composition makes it a tractable system for structural study. Recent study by high-resolution single particle cryo-EM and X-ray crystallography reveals structural details of Mm-cpn and its variant in its ATP-free open state and ATP-hydrolysis closed state (Pereira et al., 2010; Zhang et al., 2010). ATP hydrolysis leads to an alteration of intersubunit contacts within and across the two rings and ultimately causing a rocking motion of the subunit to close the ring. However, the structural detail of Mm-cpn upon ATP binding before hydrolysis is yet unknown. Here we use single particle cryo-EM to solve the structure of the lidless Mm-cpn variant in the ATP-bound, prehydrolysis state at 8 Å resolution. This intermediate structure along the lid-closure process reveals the mechanism of how the local effect of ATP binding and ATP hydrolysis are signaled throughout this protein-folding machine to complete the lid closure.
In order to study the Mm-cpn in the ATP bound state, we chose a lidless Mm-cpn variant (D386A Δlid Mm-cpn). The mutation of Asp386 to an Alanine makes the chaperonin ATPase deficient. This variant of Mm-cpn can still bind but cannot hydrolyze ATP (Reissmann et al., 2007), thus producing a uniform ATP-bound state of the chaperonin. A similar mutation was introduced in Group I chaperonins to study the ATP-bound GroEL structure (Ranson et al., 2001, 2006). A previous NMR study showed the protruding helix in the apical domain of the group II chaperonin can partially unwind and become disordered in the open state (Heller et al., 2004), thus making it unresolvable in the open state. The protruding lid segment (Ile241-Lys267) in the apical domain has been shown to be dispensable for polypeptide substrate binding and ATP binding/hydrolysis, we thus deleted and replaced it by a short linker (ETASE) (Reissmann et al., 2007) for the present study. This Mm-cpn variant offers an advantage for the particles to be distributed in random orientations in ice-embedded samples, which would facilitate angular data sampling needed for 3-D reconstruction (Zhang et al., 2010).
Cryo-EM raw images and two-dimensional class-averages show D386A Δlid Mm-cpn is open in the ATP bound state as represented by rectangular-shaped side view and ring-shaped top view with clear subunit boundaries (Figure 1). This appears similar to those images of the D386A Mm-cpn without the lid deletion in the ATP-bound state (Douglas et al., 2011) (see Figure S1 available online).
Figure 2A shows the 8 Å resolution density map of D386A Δlid Mm-cpn in the prehydrolysis ATP-bound state (Figure S2). Upon ATP binding, the chaperonin is still in the open conformation as illustrated by the top and side views of Mm-cpn. The atomic model of the 16-subunit chaperonin complex (Figure 2B) is built using Rosetta (DiMaio et al., 2009). This modeling protocol would fit an initial model built from a homolog (PDB code: 1Q3Q) to the cryo-EM density map with the correct protein geometry. The fitting has been applied to not only one subunit but also the entire complex with a D8 symmetry constraint to eliminate any steric clashes between subunits. A slice of the equatorial domain, viewed down the 8-fold symmetry axis, clearly shows the nice match of the modeled α helices to our density map (Figure 2C). The resolvability of the map is not uniform as the α helices are best resolved in the equatorial domains and less resolved (with weaker density) in the apical domains due to the flexibility in these regions (Figure 2D). This is in agreement with our previously studied ATP-free open-state Mm-cpn (Zhang et al., 2010) and the larger B-factor in the GroEL apical domains in the crystal structure (Braig et al., 1995; Chaudhry et al., 2004). Strong density is observed in the ATP binding pocket (Figure 2E, marked by blue dotted circle, and viewed along the blue arrow in Figure 2D), suggesting the presence of ATP and its chelated metal ions in the map.
The ATP-bound structure of D386A Δlid Mm-cpn provides a good intermediate state along the chaperonin's ATP hydrolysis process to allow us to investigate in detail the effect of ATP binding to the group II chaperonin and the subsequent ATP hydrolysis step. In determining this effect, we compared this new map with the previously determined Δlid Mm-cpn maps in both ATP free open state (EMDB ID 5140) and ATP/AlFx induced ATP-hydrolysis closed state (EMDB ID 5138) (Zhang et al., 2010). In order to make fair comparisons, the models for these two maps are generated using the same modeling method with Rosetta as described above (Figure S3).
Modeling accuracy depends on the resolution of the density map as well as the modeling method. For a 4 Å resolution map like the wild-type Mm-cpn in the closed state, the modeling accuracy can be assessed by the backbone connectivity, visibility of the bulky side-chain densities and the peptide geometry (Zhang et al., 2010). We have confirmed the accuracy of our Rosetta-built Cα backbone model for the closed-state Δlid Mm-cpn subunit map, which has 1.7 Å root mean square deviation (rmsd) from that of the crystal structure (PDB ID: 3KFE, Pereira et. al., 2010). However, for an 8 Å resolution map like that of the Δlid Mm-cpn free from ATP, it is harder to evaluate the model accuracy. We opted to assess the model accuracy by comparing the models obtained by two different modeling tools: Rosetta and DireX (Schröder et. al., 2007). Since any cryo-EM map may have variable resolvability throughout the protein chain equivalent to the different B factors in a crystal structure, the model accuracy may vary throughout the polypeptide chain. For instance, the equatorial domain and the region close to the ATP binding site are much better resolved than the apical domain in our map of the ATP-free Δlid Mm-cpn as evidenced by the difference in the resolvability of the α helices in these regions. The overall Cα RMSD between the Rosetta model and DireX model (PDB ID: 3IYF) for the ATP-free Δlid Mm-cpn is 2.8 Å. However, for the individual domains, the RMSD between the two Cα models breaks down to 1.7 Å in the equatorial domain, 2.3 Å in the intermediate domain, and 4.3 Å in the apical domain. The high discrepancy at the apical domain, which is known to be highly flexible, may partly reflect the difference in the modeling tools in handling poorly resolved density. These quantifications are indeed useful to indicate the level of accuracy of the model for that map. Moreover, these comparisons indicate that both the DireX and Rosetta provide equivalent models for subnanometer resolution map within 2–4 Å difference on average.
Figure 3 shows the overlap of the lidless Mm-cpn models in the ATP-free state (cyan) and ATP-bound state (purple). Upon binding of the ATP molecule, the apical domains of the chaperonin complex moves slightly toward the 8-fold symmetry axis, causing a small shrinkage of the folding chamber entrance. In addition, there is about 45° (which is an average of the 16 subunits due to the D8 symmetry in our reconstruction) of counterclockwise rotation of the apical domain as viewed from the top (Figure 3A; Movie S1). However, the apical domain from each subunit remains separated from its neighbors as in the ATP-free state and the chaperonin still stays in the open conformation with the potential apical domain substrate-binding sites accessible. The overall conformation of the equatorial domain is almost unchanged upon ATP binding. However, the intermediate domain tilts downward around the hinge region (Figure 3B, solid triangle mark) while the apical domain moves along (Figure 3B; Movie S2). As a result, the distal C-terminal end of Helix L in the intermediate domain slightly tilts toward the stem-loop, brings the key hydrolytic residue aspartic acid of Helix L into proximity of the ATP binding pocket. Thus, the system becomes poised for hydrolysis.
Figure 4 shows the overlap of the lidless Mm-cpn models in the ATP-bound state (purple) and ATP-hydrolysis state (yellow). Upon ATP hydrolysis, the apical domains of the chaperonin complex move all the way toward the 8-fold symmetry axis and form interactions with adjacent apical domains within the same ring (Figure 4A; Movie S1). Little apical domain rotation is observed from the transition from ATP-bound state to the hydrolysis state. During this step, all three domains move concertedly in a forward rocking motion hinged in the equatorial domain (Figure 4B; Movie S2). It is the ATP hydrolysis that triggers this large rocking motion of the subunit to completely close the folding chamber.
From these results (Figures 3 and and4),4), we see that the chamber closure process in the group II chaperonin are achieved in two steps: first, ATP binding induces a local active site conformational change and triggers the intermediate domain tilting and apical domain rotation; second, ATP hydrolysis causes more active site changes leading to a large subunit rocking. In order to further investigate the changes occurred in the ATP-binding pocket, we compare the computationally extracted single subunits from the ATP-free, prehydrolysis ATP-bound, and ATP-hydrolysis state. Their Cα models are superimposed (Figure 5). In this way, we can examine the local conformational changes within the subunit, which is independent of the global subunit movement. In addition, the ATP molecule is docked into the active site in the ATP-bound and ATP-hydrolysis state models (Figure 6; see Experimental Procedures).
One apparent conformational change upon ATP binding is in the loop region connecting Helices F and G (herein referred to as “FG-loop”) on the exterior surface of the intermediate domain (indicated by a gray arrow in Figure 5A). By interacting with the nucleotide in the ATP-binding pocket (Figure 6A, Movie S3), the FG-loop pulls the intermediate domain relatively downwards around the hinge point connecting the equatorial domain (Figure 5A, solid triangle) and brings the lower surface of intermediate domain onto the nucleotide (Movie S3). Several residues in this loop region have been implicated to contribute to the binding of the nucleotide according to a previous crystallographic study on a homolog (Shomura et al., 2004). The residue 386 (Asp in wild-type Mm-cpn) on Helix L moves toward the γ-phosphate and is placed for ATP hydrolysis (Figures 5A and and6A;6A; Movie S3).
In the ATP-hydrolysis state, the elongated γ-phosphate bond allows the nucleotide to reach both residue 386 in Helix L and residue 60 (Asp) in the stem-loop to get hydrolyzed. We observed an upward movement of the stem-loop triggered by this interaction (Figures 5B and and6B;6B; Movie S3). This rearrangement of the stem-loop has a large influence on the N and C termini of the neighboring subunit that together forms a β sheet (Zhang et al., 2010). As a result, the N and C termini from that neighboring subunit react to pull the entire subunit and perform a forward rocking motion toward the folding chamber. This “communication” between the stem-loop and the N/C termini of adjacent subunits may deliver the signal for positive cooperativity through the ring (insets in Figures 3B and and4B;4B; Movie S3).
The unique behaviors of Mm-cpn reaction to ATP are intrinsically built into its quaternary structural design. The aligned arrangement of subunits across two rings (Zhang et al., 2010) and loose interaction between adjacent open-state subunits in the apical and intermediate domains spatially allow this large rocking freedom for the group II chaperonins. The strong molecular interactions between apical and intermediate domains within one Mm-cpn subunit “glue” these two domains together and make them move almost as a rigid body during lid closure as in TRiC/CCT (Booth et al., 2008). These unique quaternary structural characteristics lead to its significantly different mode for lid closure in group II chaperonins from the group I chaperonins such as GroEL although they share similar tertiary structures (Figure S4) and a highly conserved ATP-binding pocket (Chaudhry et al., 2003). The staggered arrangement of the two rings in the group I chaperonins limits the conformational flexibility in the equatorial domain and directs the movements toward the two upper domains. Even in the nucleotide-free state, the three domains of GroEL already have interactions with domains of neighboring subunits within the same ring. By switching on and off salt bridges between adjacent apical and intermediate domains of neighboring subunits, GroEL's apical and intermediate domains move in the opposite directions upon folding chamber closure (Ma et al., 2000; Ranson et al., 2001; Xu et al., 1997) (Movie S4). This is also another example in nature that even with similar “power engine” (ATP-binding pocket) and “parts” (domain architectures), with slightly different ways of assembling and arrangement of these “parts,” the molecular machine would operate in a dramatically different fashion.
Both D386A and D386A Δlid were purified in buffer (20 mM HEPES [pH 7.4], 50 mM NaCl, 5 mM MgCl, 0.1 mM EDTA, 1 mM dithiothreitol (DTT), 10% glycerol, and 0.1 mM phenylmethylsulphonyl fluoride [PMSF]). They were then diluted in ATPase buffer (20 mM Tris-HCl [pH 7.5], 100 mM KCl, 5 mM MgCl and 1 mM DTT) to lower the glycerol concentration to less than 2% for EM imaging.
ATP was added to final concentration of 1 mM. The sample was then incubated in a water bath at 37°C for 5 min before freezing onto the grid. Although it has been suggested that the deleted lid-segment can possibly regulate the negative cooperativity between the two rings under low ATP concentration (Reissmann et al., 2007), under the current ATP concentration, this negative cooperativity has been overcome. It also gives a more homogeneous Mm-cpn population for high-resolution structure determination.
Detergent octyl glucoside (OG) (0.05%) was added to the D386A Mm-cpn buffer before cryo-EM freezing so to increase the yield of side views on the grid. For D386A Δlid sample, OG was not needed, thus producing better contrast when imaged in the microscope.
Samples were embedded in vitreous ice as follows: a 2.5 μl aliquot of the samples as prepared above was applied onto a 400 mesh R1.2/1.3 Quantifoil grid (Quantifoil Micro Tools, Germany). The grid was previously washed and glow discharged. After applying the sample, the grid was blotted and rapidly frozen in liquid ethane using a Vitrobot (FEI, The Netherlands), and then stored in liquid nitrogen before imaging.
The D386A Mm-cpn variant was imaged on a JEM2010F and the D386A Δlid Mm-cpn variant was imaged on a JEM2200FSC electron cryo-microscopes, respectively, both operated at 200kV (JEOL, Japan). Both microscopes have a field emission gun. Images of D386A and D386A Δlid were recorded at a detector magnification of × 83,100 and × 112,000, respectively, on Gatan 4k × 4k CCD cameras (Gatan, Pleasanton, CA). For the JEM2200FSC data collection, an in-column omega energy filter was used with a slit width of 10 eV. The particle images for the 8 Å resolution map of the D386A Δlid ATP-bound state were obtained from 64 CCD frames with a defocus range of 1–2.5 μm.
The cryo-EM data processing just followed the protocol we published recently (Baker et al., 2010).
A total of 12,761 particles were used for map reconstruction with D8 symmetry applied. The final resolution of D386A Δlid ATP-bound state Mm-cpn density maps was 8 Å, according to the Fourier shell correlation (Saxton and Baumeister, 1982) with 0.5 cutoff criterion (Figure S2).
Map segmentation, visualizations and animations were done with UCSF Chimera (Pettersen et al., 2004).
Initial models of the Δlid variants of Mm-cpn subunit in different states were constructed using the SWISS-MODEL server (Arnold et al., 2006) using a homologous thermosome (PDB code: 1Q3Q) as a template. This homology model was rigid-body docked into the cryo-EM density map to generate a 16 mer complex with D8 symmetry.
Rosetta was then used to refine the complete symmetric model inside the density map by optimizing an energy function that includes both chemical and statistical terms; additionally, a term that measures real-space correlation of the model with experimental density data was used (DiMaio et al., 2009) to flexibly fit the model into the map. The final cross-correlation scores between the fitted models and the density maps are: 0.94 (ATP-free state), 0.92 (ATP-bound state), and 0.96 (ATP-hydrolysis state) at a threshold of 8 Å resolution. The capability of visualizing secondary structure elements around the ATP-binding pockets in all these three states (Figure 2E; insets of Figure S3) allow us to confidently establish the mechanism upon ATP binding and ATP hydrolysis.
The ATP was docked into the ATP-bound state D386A Δlid and ATP-hydrolysis state Δlid Mm-cpn using Rosetta to give a general idea of the nucleotide location in the active site and its interactions with the neighboring loops and secondary structures elements.
This research has been supported by NIH grants (PN2EY016525 and P41RR002250) and NSF grant (IIS-0705644). N.R.D. purified the chaperonin. J.Z. and B.M. collected the cryo-EM images and processed the data. J.Z. built the atomic models and analyzed the results. F.D. developed the Rosetta protocol for cryo-EM density based atomic model building and advised on building models for Mm-cpn. L.J. modeled the nucleotide in the chaperonin's ATP-binding pocket.
SUPPLEMENTAL INFORMATION Supplemental Information includes four figures and four movies and can be found with this article online at doi:10.1016/j.str.2011.03.005.
All authors contributed to the preparation of this manuscript. We declare no competing financial interests.
ACCESSION NUMBERS The cryo-EM density map of Mm-cpn D386A Δlid ATP-bound state has been deposited to the EBI-MSD EMD databank with accession code EMD-5258. The atomic models of Mm-cpn D386A Δlid ATP-bound state and Mm-cpn Δlid ATP/AlFx induced state have been deposited to Protein Data Bank with accession code 3J02 and 3J03.