Partial Occlusion of Both Cavities of the Eukaryotic Chaperonin with Antibody Has No Effect upon the Rates of β-Actin or α-Tubulin Folding

The eukaryotic chaperonin containing T-complex polypeptide 1 (CCT) is required in vivo for the production of native actin and tubulin. It is a 900-kDa oligomer formed from two back-to-back rings, each containing eight different subunits surrounding a central cavity in which interactions with substrates are thought to occur. Here, we show that a monoclonal antibody recognizing the C terminus of the CCTα subunit can bind inside, and partially occlude, both cavities of apo-CCT. Rabbit reticulocyte lysate was programmed to synthesize β-actin and α-tubulin in the presence and absence of anti-CCTα antibody. The binding of the antibody inside the cavity and its occupancy of a large part of it does not prevent the folding of β-actin and α-tubulin by CCT, despite the fact that all the CCT in the in vitro translation reactions was continuously bound by two antibody molecules. Furthermore, no differences in the protease susceptibility of actin bound to CCT in the presence and absence of the monoclonal antibody were detected, indicating that the antibody molecules do not perturb the conformation of actin folding intermediates substantially. These data indicate that complete sequestration of substrate by CCT may not be required for productive folding, suggesting that there are differences in its folding mechanism compared with the Group I chaperonins.

Chaperonins are ATP-dependent protein folding machines composed of two back-to-back rings, each containing seven, eight, or nine polypeptides of ϳ60 kDa. Each ring encloses a cavity that is capped by the co-chaperonin ring of Group I chaperonins, such as GroES in the case of GroEL (1), or access to the cavity may be closed off by an in-built lid, which is the helical protrusion of the thermosome, a Group II chaperonin (2,3). Two extreme views can be taken for the existence of cavities in chaperonins. Either they act primarily as folding cages, which sequester folding polypeptide chains, or they exist as a consequence of the allosteric mechanism of action of chaperonin rings, which may require the specific positioning of subunits in relation to one another. The rings of the Group II chaperonin containing T-complex polypeptide 1 (CCT) 1 are composed of eight different subunit species (4), with one copy of each subunit occupying a fixed position in each ring (5).
CCT subunits are divergent from each other in their apical, putative substrate binding domains (6). Unlike the Group I chaperonins, which fold a broad range of proteins, the folding substrates of CCT are predominantly limited to actins and tubulins (7), although G-protein ␣-transducin has recently been identified as a substrate for CCT in vivo (8). The complex arrangement of divergent CCT subunits and the limited number of CCT substrates are highly suggestive of specific, sequencedependent interactions occurring between CCT and its substrates, rather than the more general interactions that presumably occur between GroEL (which contains only one type of subunit) and non-native proteins (5).
In this study we have utilized a monoclonal antibody (mAb), 23C, which recognizes the extreme C terminus of the CCT␣ subunit (9,10). Analysis of the crystal structure of the Group I chaperonin, GroEL (11), and of the Group II chaperonin, the thermosome (3), has shown that the C termini lie within the central cavity. Therefore, an antibody such as 23C would be expected to occupy at least part of the central cavity of CCT. This has been confirmed here by using electron microscopy to analyze negatively stained CCT-23C complexes. The images also support the biochemical model in which CCT contains a single copy of CCT␣ in each ring, although they do not prove formally that this is the case. The ability of CCT to produce native actin and tubulin when bound by two molecules of the 23C antibody has been investigated in functional in vitro translational experiments using rabbit reticulocyte lysate programmed with both ␤-actin and ␣-tubulin.

EXPERIMENTAL PROCEDURES
Electron Microscopy and Image Processing-CCT-23C complexes were obtained by incubation of CCT with excess 23C, which yields a mixture of single bound and double bound CCT-23C complexes. Lower antibody/CCT ratios were used for electron microscopy compared to the biochemical experiments to reduce the background on the grid due to unbound free antibody. Afterward, CCT-23C complexes were incubated with 10 mM MgCl 2 and 5 mM ATP to induce the appearance of a larger percentage of side views and stained with 1% uranyl acetate. Front and side views of the CCT-23C complexes were directly recorded using a GATAN ssCCD camera attached to a JEOL 1200EX II microscope. The particles were centered using a synthetic mask and aligned using a free-pattern algorithm (12,13). The average images of the front views of the CCT-23C complexes and of the side views of CCT bound to one or two 23C antibodies were generated with 423, 371, and 71 particles, respectively.
In Vitro Translation-In vitro translations of ␤-actin and ␣-tubulin (10) were carried out by priming TNT TM rabbit reticulocyte lysate (Promega, Madison, WI) with Bluescript SKIIϩ plasmids encoding full-* This work was supported by the Cancer Research Campaign. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed: were taken and further translation was stopped by addition of EDTA (pH 8.0) to 6 mM. Samples were stored on ice prior to loading on a 6% native polyacrylamide gel. Native polyacrylamide gel electrophoresis (PAGE) was carried out according to Liou and Willison (5). Upon cotranslation, to quantitate more accurately the 35 S-protein bands corresponding to native ␤-actin and ␣-tubulin monomers by phosphor imaging, monoclonal antibodies to ␤-actin and ␣-tubulin were added to samples of the reticulocyte mixture after translation had been stopped by the addition of EDTA. The formation of actin-or tubulin-antibody complexes results in a less diffuse migration of the [ 35 S]␤-actin and [ 35 S]␣-tubulin polypeptides during electrophoresis but still allows for discrimination of the respective proteins. Samples of reticulocyte lysate reaction (2 l) were mixed with 6 l of phosphate-buffered saline (PBS), 1.1 g of anti-␤-actin mAb (A5441, Sigma), and 1.5 g of anti-␣-tubulin mAb (T5168, Sigma) and incubated on ice for 30 min prior to analysis by native PAGE and autoradiography.

RESULTS
One objective of these experiments was to address the question of the degree of homogeneity of CCT. Is "core" CCT composed of two identical rings containing one copy of each of the eight CCT subunits as previously proposed (5)? Biochemical analysis of the CCT/antibody interaction showed that the migration of rabbit CCT in native polyacrylamide gel bandshift assays can be substantially retarded by addition of mAb 23C, which binds specifically to the CCT␣ subunit, and that replacement of one CCT␣ subunit by a mutant CCT␣ subunit unable to bind mAb 23C produced an intermediately migrating complex (10). Liou et al.'s interpretation of these results was that the CCT 16-mer contains two copies of the CCT␣ subunit, one in each ring. Furthermore, because all CCT complexes were shifted by mAb 23C in these experiments, Liou et al. (10) suggested that all CCT complexes contain two copies of the CCT␣ subunit. Now we have examined CCT-mAb 23C complexes by negative-stain electron microscopy and image processing. CCT-antibody complexes are composed of CCT and either one (Fig. 1B) or two (Fig. 1C) antibody molecules. Side views of double antibody-bound CCT only showed one antibody bound to each ring (Fig. 1C), which is consistent with there being only one CCT␣ subunit present in each ring. However, if two CCT␣ subunits were directly adjacent in each of the rings, it is possible that steric hindrance would result in only one antibody molecule binding. The disposition of antibody in the double bound complexes also shows that the CCT␣ subunits do not contact each other across the rings; in fact, the two CCT␣ subunits are opposite, or nearly opposite, each other in the opposing rings. Clearly, the antibody binds within the central cavity of CCT and partially occludes it, suggesting that the C terminus of CCT␣ is located within the cavity. Although the C termini of the chaperonin subunits in the structures of GroEL (11) and the thermosome (3) are disordered, the structures show that they are also located within the central cavity. A further indication that 23C is fairly deeply embedded in the cavity of CCT is that, although 23C is a bivalent IgG antibody, it is not able to cross-link multiple CCT complexes (10). Nevertheless, if an anti-rat secondary antibody is added to the CCT-23C complex, all the CCT becomes cross-linked and trapped in the well of a native gel (data not shown), demonstrating that the Fc portion of the 23C IgG is protruding, as can be seen clearly in the side view images of 23C-decorated complexes (Fig. 1, B and C).
We have already shown that 23C can bind CCT complexes that contain bound substrates such as ␣-tubulin (10). Therefore, we wondered whether 23C might interfere with processing of substrates by CCT in rabbit reticulocyte lysate in vitro translation reactions, and we have analyzed ␤-actin and ␣-tubulin folding in this system. Various experiments were carried out to optimize the amount of 23C to avoid inhibitory effects of the addition of exogenous protein to this biochemically rather labile system. A final concentration of 60 g ml Ϫ1 23C was well tolerated (data not shown), and this concentration was in excess for complete CCT binding. An example of a stacking gel is shown ( Fig. 2A) to confirm that little protein denaturation/ aggregation is occurring in this system. Individual translation reactions were programmed with both ␤-actin and ␣-tubulin to act as internal controls for each other for any possible effects of 23C on folding of substrates. The results are clear and slightly surprising; there are no inhibitory effects of 23C on the rate of production or yield of either ␤-actin or ␣-tubulin over a 90-min time course (Fig. 2, A-G). The lag phase seen during the production of ␣-tubulin monomers may be a result of the co-factor binding of ␣-tubulin after release from CCT (14) and has been observed previously in our native gel system (Fig. 2) (10). Although the total level of native actin produced in the presence and absence of 23C appears unchanged, it would appear that some difference occurs between the distribution of actin in FIG. 1. Processed two-dimensional images of CCT-23C antibody complexes. CCT was prepared from mouse testis as described previously (5). mAb 23C is a rat IgG 2c immunoglobulin recognizing an epitope, Leu 554 , Asp 555 , Asp 556 , located at the C terminus of CCT␣/TCP-1 (9, 10). In the experiment shown, 136 g of mouse testis CCT was incubated with 100 g of 23C in a reaction volume of 195 l overnight at 4°C. CCT-23C complexes (retention time of 35 min) were fractionated from unbound 23C antibodies (retention time of 50 min) by high pressure liquid chromatography using a Superose-6 gel filtration column (flow rate, 0.5 ml/min) in 50 mM Tris-HCl buffer, pH 7.5. A, average top view; B, average side view with one antibody bound; C, average side view with two antibodies bound. the two closely migrating actin bands shown in Fig. 2. However, upon further analysis no difference could be detected by SDS-PAGE Ϯ 23C (Fig. 3A), and both preparations of ␤-actin folded in the presence and absence of 23C bound to DNase I (Fig. 3B), showing that the ␤-actin produced is native. The susceptibility of ␤-actin to cleavage by trypsin while bound to CCT and CCT-23C complexes was analyzed by semi-native diagonal electrophoresis (5). The ␤-actin bound to both CCT and CCT-23C consists of two closely migrating species corresponding to native ␤-actin (Fig. 4, A and C, asterisks) and three smaller fragments, known to be the products of initiations at internal methionines during translation 3 (Fig. 4, A and B). Upon trypsin treatment a further three ␤-actin fragments were detected bound to CCT (Fig. 4C, arrowheads), whereas no native G-actin was digested under these conditions, consistent with the notion that ␤-actin is in a different conformation to native G-actin when bound to CCT (Fig. 4, C and D). This suggests that the presence of the antibody does not alter the binding or conformation of ␤-actin sufficiently to result in changes in protease sensitivity when ␤-actin is bound to CCT-23C complexes. The amount of 23C added, in this proteolysis experiment only, was sufficient to produce a partial shift of CCT (Fig. 4, B and D). This made it possible to confirm that the trypsin susceptibility of the ␤-actin when bound to either CCT or CCT-23C is the same within a single sample. DISCUSSION The cavity of the thermosome has a volume of 130,000 Å 3 (3), and we have recently shown the striking similarity in volume of the thermosome and the asymmetric folding cavity of CCT (15). Approximately 20% of the volume of the CCT cavity is occupied by antibody (Fig. 1A) as measured by the signal of the twodimensional projection of the density. An IgG antibody has a total volume estimated at 100,000 Å 3 , and each Fab comprises around 30% of the molecule. Because it can be seen that only one combining site of the 23C mAb enters the CCT cavity and that the Fc portion protrudes, one can estimate that 30,000/ 130,000, i.e. 23% of the CCT cavity would be occluded, in good agreement with the experiment. Although we do not know the conformation that a substrate such as actin (375 amino acid residues) adopts when it is enclosed inside the CCT cavity, one can estimate the volume of a globular protein of 375 residues to be 29,000 Å 3 and conclude that such a molecule would occupy 22% of the cavity. However, it should be noted that native actin is not globular and its maximum dimensions will only just fit in 3  A recent study on the maximum size of proteins able to occupy and complete folding in the GroEL cavity underneath GroES showed that a 54-kDa fused dimer of green fluorescence protein could fold, but an 82-kDa green fluorescence protein trimer could not (16). CCT seems to function normally with its cavity occupied at the upper limit of the GroEL substrate size range, because ␤-actin ϩ 23C and ␣-tubulin ϩ 23C have a combined molecular mass of 70-and 80-kDa, respectively. Furthermore, Weissman et al. (17) showed that the substrate is protease-resistant in preformed polypeptide-GroEL-GroES complexes, although we found ␤-actin bound to CCT to be highly sensitive to proteolysis.
It can be seen from the negatively stained side views that the CCT rings bound by 23C are distorted; this may reflect the flexibility of the helical protrusions of the apical domains. Perhaps, because of this inherent flexibility (2, 3), the apical domains are still able to undergo the movements required (15) to close off the CCT cavity in the presence of antibody without perturbing the folding cycle. Apart from being smaller, the antibody-occluded folding cavity could reasonably be expected to have a different structure, with some of its internal surfaces completely obscured. If all the eight apical domains have equivalent functions, one would predict the folding rates of the two substrates would be slowed down by the presence of 23C. Therefore, as suggested by Liou and Willison (5), perhaps not all the subunits are involved in substrate interactions. Nevertheless, both actin and tubulin fold normally in such smaller cavities, and these experiments point to differences between the folding mechanisms of CCT and GroEL. A, human ␤-actin was translated in rabbit reticulocyte lysate in the presence and absence of 60 g ml Ϫ1 23C mAb. Translation was stopped after 90 min by the addition of EDTA to 6 mM. Samples (4 l) were prepared for SDS-PAGE and resolved on a 12% polyacrylamide gel, and the production of 35 S-labeled ␤-actin was analyzed by autoradiography. The positions of molecular standards and their sizes (in kDa) are indicated on the left. B, 1 l of stopped lysate was incubated in the presence and absence of 10 g of DNase I with PBS added to 10 l for 30 min at 4°C. 1.5 l of loading buffer was added, and the entire sample was resolved on a 6% polyacrylamide native gel followed by autoradiography. The positions of CCT, native ␤-actin, ␤-actin/DNase I complex, and a ␤-actin complex formed during later stages of translation (but disrupted in the presence of DNase I) are indicated.
FIG. 4. Limited proteolysis of ␤-actin bound to CCT and CCT-23C complexes. Rabbit reticulocyte lysate (50 l) was programmed with 0.9 g of cDNA encoding full-length human ␤-actin in the presence and absence of the 23C mAb. After translation for 20 min at 30°C, 1-l samples were incubated in the presence and absence of 10 ng of trypsin for 15 min at room temperature. 6 l of PBS and 1 l of sample buffer were added, and the entire sample was resolved on a 6% polyacrylamide native gel. Lanes were excised, rotated 90°, and run on a 12% SDS-PAGE gel. A, ␤-actin translated in the absence of 23C; B, ␤-actin translated in the presence of 23C; C, ␤-actin translated in the absence of 23C followed by trypsin treatment; D, ␤-actin translated in the presence of 23C with subsequent trypsin treatment. The positions of unshifted (CCT(u)) and antibody-shifted (CCT(s)) CCT and the position of native ␤-actin in the first gel dimension are indicated.