Conformational cycle of the archaeosome, a TCP1-like chaperonin from Sulfolobus shibatae.

The major heat shock proteins in the archaeon Sulfolobus shibatae are similar to the cytosolic eukaryotic chaperonin and form an 18-subunit bitoroidal complex. Two sequence-related subunits constitute a functional complex, named the archaeosome. The archaeosome exists in two distinct conformational states that are part of chaperonin functional cycle. The closed archaeosome complex binds ATP and forms an open complex. Upon ATP hydrolysis, the open complex dissociates into subunits. Free subunits reassemble into a two-ring structure. The equilibrium between the complexes and free subunits is affected by ATP and temperature. Denatured proteins associate with both conformational states as well as with free subunits that form an intermediate complex. These unexpected observations suggest a new mechanism of archaeosome-mediated thermotolerance and protein folding.

We report here that the major heat shock protein in S. shibitae (called also TF55 (23) and rosettasome (1a)), unlike the bacterial GroEL, exists in two distinct, stable conformational states (closed and open) that appear to be part of the protein folding pathway. This protein will be referred to here as the "archaeosome." The closed archaeosome complex binds ATP and forms an open complex, which upon ATP hydrolysis specifically dissociates into subunits. Free subunits reassemble to an 18-subunit bitoroidal complex and complete the cycle. The equilibrium between complex and subunits is affected by ATP and temperature. Heat-denatured proteins associate with both conformational states. The subunits also bind denatured proteins and form an intermediate complex.
Dephosphorylated bovine ␣-S1-casein and pure ␣ and ␤ subunits of the archaeosome were labeled with 32 P using [␥-32 P]ATP and human casein kinase II. Excess of proteins over [␥-32 P]ATP were used to ensure a single labeling event per polypeptide chain. Human casein kinase II was inactivated by heat and 32 P-labeled proteins were purified from unreacted [␥-32 P]ATP on excellulose GF5 gel filtration columns from Pierce. The specific radioactivity of radiolabeled proteins was determined by liquid scintillation counting and UV spectroscopy. Protein concentration was determined by spectroscopy using calculated extinction coefficients. Molar extinction coefficients of chaperonin and its subunits were calculated from the protein sequence using the method of Gill and von Hippel (24) and were 9.08 ϫ 10 3 M Ϫ1 cm Ϫ1 at 278 nm for the ␣ subunit, 3.26 ϫ 10 4 M Ϫ1 cm Ϫ1 for the ␤ subunit at 280 nm, and 3.75 ϫ 10 5 M Ϫ1 cm Ϫ1 for the ␣/␤ archaeosome at 280 nm.
Purification of archaeosome and glutamate dehydrogenase from Sulfolobus shibatae was done using the following procedure. S. shibatae (DSM strain 5389) was grown at 80°C in liquid medium (containing yeast extract and Brock's salts) in a 160-liter fermenter as described previously (23). Cells were lysed by raising the pH to 7.5 with 0.1 N NaOH in the presence of 0.25% Triton X-100 (v/v) in 50 mM Tris/HCl, pH 7.5, 10 mM 2-mercaptoethanol, and 1 mM EDTA. Soluble crude cell extract was chromatographed on a FastQ FPLC column, and proteins were eluted with 20 mM Tris/HCl, pH 7.5, 1 mM DTT, and 0 -500 mM NaCl gradient. Fractions containing chaperonin were concentrated on YM100 membrane from Amicon, and chaperonin was further purified by gel permeation chromatography on a Sephacryl 300 column in 20 mM Tris/HCl pH 7.5 buffer containing 1 mM DTT and 250 mM NaCl. A protein peak that eluted close to void volume comprised the archaeosome. Finally, chaperonin was separated on a high resolution MonoQ (16/10) column using FPLC and eluted with 20 mM Tris/HCl pH 7.5 buffer, 1 mM DTT, and 150 -400 mM NaCl gradient. Chaperonin was at least 95% pure as judged by gel electrophoresis and staining with silver. Protein sequence was confirmed by partial sequencing of a chymotrypsin-resistant 30-kDa protein fragment (W. M. Keck Foundation, Biotechnology Resource Laboratory, Yale University). Pure chaperonin was shown to prevent heat-induced aggregation of S. shibatae glutamate dehydrogenase and stimulate folding of guanidinium/HCl denatured glutamate dehydrogenase under conditions described previously (22).
Archaeosome subunits were purified using the following procedure. S. shibatae crude cell extract was chromatographed on a FastQ FPLC column, and proteins were eluted with 20 mM Tris/HCl pH 7.5 buffer, 1 mM DTT, and 0 -500 mM NaCl gradient. Fractions containing ␣ and ␤ subunits were concentrated on YM30 membrane from Amicon and were further purified by gel permeation chromatography on a Sephacryl 300 column in 20 mM Tris/HCl pH 7.5 buffer, 1 mM DTT, and 250 mM NaCl. A protein peak that eluted close to two void volumes contained archaeosome subunits. Finally, ␣ and ␤ subunits were separated on a high resolution MonoQ (16/10) column using FPLC and eluted with 20 mM Tris/HCl, pH 7.5, 1 mM DTT, and 0 -150 mM NaCl gradient. Under these conditions ␣ and ␤ subunits can be separated from each other. Alternatively, pure archaeosome was dissociated to subunits by a 1-h incubation at 75°C with 2 mM ATP, and ␣ and ␤ subunits were separated on a high resolution MonoQ (16/10) column using FPLC, as described above. Subunits were at least 90% pure as judged by gel electrophoresis and staining with silver. Thermus aquaticus chaperonin was purified using a procedure similar to that described for the archaeosomes.
Glutamate dehydrogenase from S. shibatae (DSM strain 5389) was purified using the modified procedure of Robb et al. (25), in which Sepharose CL-6B column was replaced with Sephacryl 300 column and Phenyl-Sepharose CL-6B column was replaced with MonoQ (16/10). The purity and the identity of the protein were confirmed by sequencing of the N terminus (W. M. Keck Foundation, Biotechnology Resource Laboratory, Yale University) and by activity assay (22).
Native Gel Electrophoresis-Chaperonin complexes were separated on 4, 6, 10, and 6 -10% gradient polyacrylamide (75/1) gels that were prerun at 25 V/cm at constant power for 2 h in 11 mM Tris/phosphate buffer, pH 7.5, at 4°C. Gels were run at 25 V/cm with buffer recirculation at constant power for 2 h at 4°C in a gel system from Hoefer. Gels were stained for 5 min with Coomassie Brilliant Blue to visualize proteins, vacuum dried, and autoradiographed. Archaeosome closed and open complexes were purified in milligram scale by native gel electrophoresis using preparative cell model 491 from Bio-Rad using conditions described above and following the manufacturer's protocol. Fractions containing relevant complexes were concentrated on Centricon YM100 from Amicon and stored at room temperature.
Electron Micrograph Images-Specimens were applied to a 400 mesh glow discharged carbon grid and stained with 1% uranyl acetate. In order to obtain the best contrast, an ultra thin (ϳ10 Å) carbon film was laid over a holey carbon film to provide as thin a substrate as possible. The clarity of particles lying on the thin film was considerably greater than that of particles lying on the thicker carbon film. Electron micrographs were obtained using a Philips CM10 electron microscope at 100 kV and at an electron optical magnification of 52,000. Micrographs were digitized on an Optronics P1000 rotating microdensitometer set on the 3 optical density range and at a pixel size of 25 microns (corresponding to 4.8 Å/pixel). The optical density range averaged from 1.4 to 2.8. Computations were carried out on a VAX 4000/90 work station and displayed on a Raster technologies 1/25 display system. Groups of 5 to 12 single molecules were low passed filtered to reduce high frequency noise. One particle was chosen as a reference image, and the others were rotationally and translationaly aligned with it by maximizing the cross-correlation coefficient between them. Rotational alignments were accomplished in one degree increments, and translational alignment were accomplished in increments of 0.5 pixels. Rotational alignment was carried out by first transforming images from (x, y) space to (R, ) space. Aligned images were added in order of decreasing correlation coefficients, which ranged from 0.7 to 0.5. Each image in a group was used as a reference, and the set with the highest correlation coefficients was chosen.
Circular Dichroism Spectroscopy-Circular dichroism measurements were recorded at Brookhaven National Laboratory NSLS beam line U9B at 25°C. Native polyacrylamide gel-purified chaperonin closed and open complexes were diluted 10-fold into spectrally pure 50 mM phosphate buffer, pH 7.5, to protein concentration 0.6 mg/ml (5.8 M). Spectra were recorded in 0.5-ml quartz cuvette with a 5-mm optical path at 2-nm intervals and 10-s time constant and were processed and scaled using U9B/CD/ORIGIN software and 50 mM phosphate buffer blanks were subtracted from corresponding samples.
Archaeosome Reconstitution Experiments-The archaeosome complex was reconstituted from free subunits using the following protocol. A mixture of purified ␣ (5 M) and ␤ (5 M) subunits in 20 mM Tris/HCl buffer pH 7.5, 1 mM DTT, 100 mM NaCl were diluted 20-fold with 50 mM spectrally pure sodium phosphate in spectrally pure water, pH 7.5, and concentrated on YM30 Centricon from Amicon at 20°C to subunit concentrations exceeding 10 M. Reconstituted complexes appear stable at room temperature for several months in this buffer system.

Composition and Properties of Archaeosome Complexes-
Trent et al. (23) reported that the native TF55 protein migrates as a double band in native polyacrylamide gel electrophoresis (PAGE). We have found that pure S. shibatae archaeosome separates into two major components on native PAGE (Fig. 1, lane 1). The protein complexes representing the two major components, the slower migrating top band (TB) and the faster migrating bottom band (BB), were purified to homogeneity using preparative native PAGE and were characterized further. Two-dimensional gel electrophoresis of the native archaeosome shows that it is composed of two polypeptides that separate into several charge variants (Fig. 1, 2-DE panel). Similar heterogeneity has been observed for other thermophilic proteins (26), and is most likely a result of deamidation at Asn and Gln side chains at high temperature (27). It is also possible that some of the heterogeneity can be attributed to phosphorylation, which has been reported for the Sulfolobus solfataricus chaperonin (18).
The two polypeptides are sequence-related and have molecular masses of 59.8 and 59.9 kDa, respectively, as derived from amino acid sequences of their genes (19,23). The 59.8 kDa protein (␤) is identical to the TF55 protein (23), and the 59.9 kDa protein (␣) will be characterized in detail elsewhere (19). Pure top and bottom complexes were examined by two-dimensional gel electrophoresis and showed identical protein composition ( Fig. 1, 2-DE panel). In both complexes, ␣ and ␤ subunits were present in roughly stoichiometric amounts (19 and below). Both complexes are free of nucleotides, as indicated by their UV spectra (not shown).
Using electron microscopy and circular dichroism (CD), we examined the possibility that conformational differences in the chaperonin complex could account for the observed difference in gel mobility. Electron micrographs showed that the pure top band complex forms a nine-fold, double-ring structure that has an electron dense core, i.e. it stains preferentially in the center with uranyl acetate (Fig. 1, micrographs panel). This complex appears identical to the TF55 structures published earlier (23). We describe top band as an "open" complex because it appears to contain an open cavity when viewed down the nine-fold axis of the particle (Fig. 1, symmetry panel). The side view shows the characteristic four-band striation pattern reported for other hsp60 chaperonins having 7/2, 8/2, and 9/2 symmetries (3,20,21,23). In contrast, the faster moving BB complex showed no obvious symmetry on electron micrographs and forms a "closed" complex with a poorly defined central cavity (Fig. 1,  micrographs panel), although it is similar to the open complex in size and appearance both in top and side views. CD spectra of the purified closed and open complexes show quite remarkable differences (Fig. 2). The archaeosome closed complex show ellipticity typical of proteins with high ␣-helix contents. The CD spectrum of the open complex is similar but the minimum at 224 nm shifts to 220 nm perhaps suggesting lower ␣-helical content.
Relative Amount of Closed and Open Complexes Varies During Heat Shock-In S. shibatae the relative amounts of closed and open complexes are affected by growth temperature (Fig.  3A). Under normal growth conditions (75°C) the closed complex is more abundant. Under stress conditions (heat shock temperature Ͼ85°C), more open complex is detected. Under lethal conditions (Ͼ90°C), the open complex disappears, and only the more temperature-stable closed complex persists. Under both normal and heat shock conditions S. shibatae cells also contain free subunits (data not shown).
Trent et al. (23) reported previously that TF55 protein binds guanidine/HCl-denatured dihydrofolate reductase and unspecified E. coli proteins. We found that both the open and closed complexes bind heat-denatured 32 P-␣-S1-casein (Fig. 3B). The binding of casein is temperature-dependent and occurs only above 50°C. The association of casein with the archaeosome complexes is weak, as compared with the binding of 32 P-␣-S1casein to thermophilic GroEL-like chaperonin from T. aquaticus ( Fig. 3C and Ref. 28), and the ternary complex appears to be formed transiently. Similar results have been reported recently for binding of ␤-actin to c-cpn60 cytosolic chaperonin of eukaryota (29). C-cpn60 chaperonin binds ␤-actin ten times more weakly than E. coli GroEL, but in contrast to GroEL c-cpn60 supports effective folding of ␤-actin.
Specific Dissociation of Archaeosome Complexes-Like other chaperonins, the archaeosome shows ATPase activity (18,22,23). We investigated the effect of ATP and its analogs on the equilibrium between closed and open complexes. Purified closed complex of archaeosome was incubated with ATP, ADP, or ATP␥S (a slowly hydrolyzable analog of ATP) (Fig. 4A). In some experiments, small amounts (5%) of TFE were added; TFE and other alcohols have been reported to affect protein conformation by weakening hydrophobic interactions and disrupting oligomeric structures (30).
In the presence of ATP, ADP or ATP␥S, the closed complex can be partly converted to the open complex (Fig. 4A, lanes 4, 5,  8, and 9). The open complex appears to be stabilized by the addition of ADP or ATP␥S (Fig. 4A, lanes 4, 5, 8, and 9). The binding of ATP and subsequent hydrolysis of the phosphoester bond lead to complex dissociation (Fig. 4A, lanes 6 and 7). As expected, the addition of 5% TFE further destabilizes the complexes, causing dissociation of the complex to subunits (Fig. 4A,  lanes 3 and 7). More complex dissociates in the presence of ATP.
The observed effect of ATP is very specific, because the closed complex resists treatment with 7 M urea, and both complexes are stable for weeks at ambient temperatures with or without bound nucleotide (data not shown and Ref. 18).
Reconstitution of Archaeosome from Free Subunits-We purified ␣ and ␤ subunits of the archaeosome to homogeneity directly from cell extracts of S. shibatae as described under "Materials and Methods." These subunits can be reconstituted into the full-size complex by incubating mixtures of subunits at concentrations greater than 10 M at 20°C in phosphate buffer and at neutral pH (Fig. 4B, left panel). Small amounts of intermediate-size complexes that behave like a single ring were also observed on overloaded gels. Electron micrographs of the reconstituted archaeosome display normal bitoroidal structures (Fig. 4B, left panel) that bind 32 P-␣-S1-casein (data not shown). Neither ATP nor peptides are required for the reconstitution, but both affect it (see below).
To study the composition of the archaeosome, purified ␣ and ␤ subunits were 32 P-labeled, and the relative composition of the reconstituted archaeosome was determined. When a stoichiometric mixture of ␣ and ␤ subunits was reconstituted with a small amount of 32 P-labeled ␣ subunit, 18.4 Ϯ 3.2% of 32 P-label was found in the reconstituted chaperonin. Under identical conditions 15.0 Ϯ 3.3% 32 P-labeled ␤ subunit reconstituted into the chaperonin complex. This result suggests an apparent 1:1 stoichiometry for ␣ and ␤ subunits in archaeosome. Each subunit alone, however, can also reconstitute a high molecular weight complex; these complexes have gel mobilities slightly different from that of the heteroligomeric archaeosome (19). ␤ subunits form a complex both in the presence and absence of ATP, whereas ␣ subunits only assemble in the absence of ATP. In the presence of both ␣ and ␤ subunits, only the stable ␣-␤ heteroligomeric complex is formed with and without ATP (19). Hence, the bitoroidal archaeosome seems to be asymmetric in respect to ATP-dependent ring stability.
When ␣ and ␤ subunits reconstitute in the presence of heatdenatured 32 P-␣-casein, the 32 P-casein is found mainly bound to subunits and to an intermediate complex, whereas only small amounts are associated with the open and closed forms of full-size archaeosome (Fig. 4B, right panel). Similar weaker binding of denatured proteins to TCP-1-like chaperonin was reported by Tian et al. (29). We evaluated the relative affinity of ␣ and ␤ subunits for 32 P-␣-casein in titration experiments (Fig. 4C). Both subunits bind 32 P-casein, but the ␤ subunit appears to bind casein 5-10 times more strongly than the ␣ subunit. The subunits-32 P-casein complex migrated in gels slower than free subunits but faster than the archaeosome complex, suggesting that this complex may represent an intermediate. As judged by its gel mobility, it could represent a single ring composed of identical subunits with bound ␣-casein. Single rings of thermophilic chaperonin have been reported recently (31), and the equilibria between chaperonin, single rings, and subunits have been observed for three different chaperonins including GroEL (32). DISCUSSION Extensive efforts are underway in many laboratories to fully characterize the eukaryotic chaperonin TCP1/TRiC. In contrast to bacterial GroEL, where the underlying mechanisms of protein binding and folding have been studied in great depth, less is known about the mechanism by which the eukaryotic TCP1 chaperonin folds proteins. The most obvious difference is the presumed multisubunit complexity of the TCP1 chaperonin. Archaebacterial chaperonins show extensive sequence similarity to eukaryotic cytosolic chaperonins (TCP1) and represent a simpler model with which to study folding in eukaryotic cells because the archaeal chaperonin complex is composed of just two subunits (18,19).
We have shown that the S. shibatae chaperonin exists in vitro and presumably also in vivo as two distinct complexes that can be separated by native PAGE. We believe that these two complexes can be detected because the complex is frozen in these two states by lowering the temperature from 75°C to room temperature. Two distinct conformations of TCP1 chaperonin from mouse testis have been observed by Hynes et al. (16) using specific antibodies. These authors suggested that binding or hydrolysis of ATP acts as a switch between two conformational forms of chaperonin. Knapp et al. (18) observed similar complexes of S. solfataricus chaperonin on electron micrographs. Guagliardi et al. (22) reported that the chaperonin from S. solfataricus (termed Ssocpn) in the presence of Mg-ATP undergoes a large conformational rearrangement (as observed by change in tryptophan fluorescence). Conformational changes in E. coli GroEL (33) and complexes of GroEL with bound ATP, GroES, and protein-substrate (3) have been reported. The magnitude of the conformational changes in the archaeosome structure can be pertinent to the structure of chaperonin from thermophilic archaebacteria obtained with electron microscopy by Phipps et al. (21). This structure shows a large mass of protein blocking the entrance to the chaperonin central cavity. These EM reconstructions could represent the structure of closed complex described here or complex with bound protein substitute.
The CD spectra of purified complexes showed remarkable differences, implying that the open complex is structurally altered (Fig. 2). Both the electron microscopy and the CD results suggest that there is a major conformational difference between the two complexes. The structural changes in the open complex that allow uranyl acetate to bind within the central cavity also increase the effective cross-section of the complex, which reduces its mobility in native polyacrylamide gels. The change in the CD spectrum reflects more extensive structural rearrangement than just domain movement. If the domain organization of archaeosome is similar to that of E. coli GroEL (as suggested by limited but significant sequence similarity), then our data would imply conformational changes in the apical region. This domain was proposed by Horwich and coworkers (11) to be involved in protein binding and folding.
We have shown here that the conformational changes are related to conversion of closed complex to open complex and dissociation to subunits. The equilibrium between three states, open and closed complexes and free subunits, is affected in vitro by temperature and by Mg-ATP and its derivatives, ADP and ATP␥S. It appears that in vivo under mild heat shock, the amount of open complex increases, suggesting a response to a stress. Under lethal heat shock conditions only the closed complex remains. Our data suggest that this dissociation is controlled by ATP hydrolysis. A large concentration of free subunits may provide an advantage to the cell by capturing unfolded polypeptides under heat shock conditions and arresting protein aggregation. The full role of free subunits in the protein folding cycle has yet to be established, and an hsp70like function of the ␣ and ␤ subunits cannot be excluded. A similar function in heat shock response has been attributed to yeast hsp104 protein, which is believed to assist protein solu- The archaeosome was incubated for 15 min with 32 P-␣-S1-casein at 75°C, cooled to 4°C, and loaded onto 4% polyacrylamide gels. Lane 1, free 32 P-␣-S1-casein at 0.5 M. Lanes 2-6 contain 32 P-␣-S1-casein at 0.5 M and increasing concentration of the archaeosome (as indicated on the figure). The amounts of open and closed complex were approximately the same at the start of incubation. Specific radioactivity of 32 P-␣-S1casein is 25.2 kcpm/pmol. C, binding of 32 P-␣-S1-casein to GroEL-like chaperonin from thermophilic bacteria T. aquaticus is shown as a control (28). Binding and gel conditions are like in B. Lane 1, free 32 P-␣-S1-casein at 0.5 M; lane 2, contain 32 P-␣-S1-casein at 0.5 and 1 M chaperonin from T. aquaticus. bilization rather than folding. Parsell et al. (34) recently reported that yeast hsp104 can rescue proteins from aggregates once they have formed. Strikingly, it has been noted earlier that the stability of hsp104 hexamer in vitro, similar to the archaeosome, is ATP-dependent (35). Thus the dissociation of the archaeosome to subunits could be an important part of a functional cycle that links protein-mediated thermotolerance with protein folding.
Several protein folding cycles have been proposed for bacterial chaperonin (7-9, 13, 14). These cycles postulate the forma-tion of specific bi-, ter-, and quaternary complexes that facilitate protein folding and chaperonin regeneration. The cycle that we propose in Fig. 5 includes a change in conformation and in oligomerization state. The cell maintains all the chaperonin components (closed complex, open complex, and subunits) in equilibrium. Both complexes and subunits appear to bind denatured proteins. Our data imply that in the archaea, unfolded protein enters the cycle by binding to subunits, proceeds through an intermediate that is composed of individual subunits, and continues to the double-ring complex. We believe P-␣-S1-casein at 0.5 M; lane 2, stoichiometric amounts of unlabeled ␣ and ␤ at 10 M reconstituted with 32 P-␣-S1-casein at 0.5 M. C, binding 32 P-␣-S1-casein to ␣ and ␤ subunits. Left panel, lane 1, free 32 P-␣-S1-casein at 0.5 M; lanes 2-6, 32 P-␣-S1-casein at 0.5 M incubated at 75°C with increasing concentrations of ␤ subunits (as indicated on the figure); lanes 7-11, 32 P-␣-S1-casein at 0.5 M incubated at 75°C with increasing concentrations of ␣ subunit (as indicated on the figure). Complexes were separated on 10% native PAGE as described under "Materials and Methods" and in the legend to Fig. 3. Specific radioactivity of radiolabeled proteins was: 32 P-␣-S1-casein, 25.2 kcpm/pmol; ␣ subunit, 3.3 kcpm/pmol; and ␤ subunit, 2.0 kcpm/pmol. that the folded protein is released when the open complex dissociates into subunits (Fig. 5).
There is a direct analogy between the archaeosome open complex and the GroEL-GroES complex as high energy states and the free subunits and GroEL as the low energy states (7). The dissociation of the archaeosome to subunits is the ultimate relaxation of the high energy state. The closed complex appears to represent an intermediate energy state. We suggest that as previously reported for GroEL (7), the thermodynamic barriers separating protein-bound and free archaeosome states are overcome by ATP hydrolysis. The dissociation of bound protein is most likely accomplished by a change in the binding affinities of the chaperonin for a non-native protein. The extreme way to achieve this is to break up the structure of the complex into its subunits. Clearly the chaperonin complex and free subunits must present different interactive surfaces for unfolded proteins.
We propose that, as an unfolded protein assumes its native structure, the archaeosome undergoes conformational changes. The high entropy of the unfolded protein is assimilated by the chaperonin as the protein folds, the archaeosome acting as an "entropic sink." After ATP hydrolysis, the ternary complex dissociates, releasing folded protein and subunits (Fig. 5). Free subunits reassemble into complexes, completing the cycle. It is likely that the archaeosome folds proteins in a quite different way than the GroEL-like chaperonins. In fact, Cowan and co-workers (29) showed recently that a distinct set of folding intermediates is released from different chaperonins. The hyperthermophilic archaebacteria must protect and fold proteins that are already quite thermostable, and therefore archaeo-some may require higher energy to overcome thermodynamic barrier in folding of these proteins. It is likely that proteins and chaperonins coevolved to optimize folding requirements of the cell. This is reflected in the properties of chaperonin and the dynamics and degree of structural changes during chaperoninmediated protein folding.