Role of the Chaperonin CCT/TRiC Complex in G Protein βγ-Dimer Assembly*

Gβγ dimer formation occurs early in the assembly of heterotrimeric G proteins. On nondenaturing (native) gels, in vitro translated, 35S-labeled Gγ subunits traveled primarily according to their pI and apparently were not associated with other proteins. In contrast, in vitro translated, 35S-labeled Gβ subunits traveled at a high apparent molecular mass (∼700 kDa) and co-migrated with the chaperonin CCT complex (also called TRiC). Different FLAG-Gβ isoforms coprecipitated CCT/TRiC to a variable extent, and this correlated with the ability of the different Gβ subunits to efficiently form dimers with Gγ. When translated Gγ was added to translated Gβ, a new band of low apparent molecular mass (∼50 kDa) was observed, which was labeled by either 35S-labeled Gβ or Gγ, indicating that it is a dimer. Formation of the Gβγ dimer was ATP-dependent and inhibited by either adenosine 5′-O-(thiotriphosphate) or aluminum fluoride in the presence of Mg2+. This inhibition led to increased association of Gβ with CCT/TRiC. Although Gγ did not bind CCT/TRiC, addition of Gγ to previously synthesized Gβ caused its release from the CCT/TRiC complex. We conclude that the chaperonin CCT/TRiC complex binds to and folds Gβ subunits and that CCT/TRiC mediates Gβγ dimer formation by an ATP-dependent reaction.

The human genome codes for nearly 700 G protein-coupled receptors (1,2), which signal through heterotrimeric G proteins and mediate the effects of a vast array of hormones, neurotransmitters, sensory signals, and drugs (3,4). G protein-coupled receptors transmit signals to downstream effectors by catalytically activating target G proteins (5,6). Theoretically, at least, the G protein targets of the G protein-coupled receptors are as diverse as the receptors through their combinatorial formation from 16 G␣, 5 G␤, and 12 G␥ gene products (7). Signaling specificity of heterotrimers is affected both by the biochemistry of their protein-protein interactions and by cellular targeting mechanisms that direct placement of receptors, G proteins, and second messenger targets (7,8). An important but poorly understood process that affects all aspects of the signaling specificity of the heterotrimeric G proteins is the synthesis and assembly of G protein heterotrimers.
G␤␥ dimer formation is an early event in G protein assembly (9,10). Biochemically, the G␤␥ complex is a very stable dimer (11) that reversibly interacts with the G␣ subunit, dependent on the associated guanine nucleotide (5). Given the apparent stability of the G␤␥ dimer, its assembly is thought of as an all-ornone event; however, recent work indicates that the efficiency of dimer assembly in vitro is related to isoform-specific interactions of G␤ (12). G␤ is a member of a large family of WD repeat proteins (13) that contain 4 -16 repeats of a core of ϳ40 amino acid. G␤ has seven repeats that form a closed, toroidshaped, seven-bladed, ␤-propeller structure (14,15). Folding of G␤ into this structure requires G␥ (16) and the participation of as yet unidentified components of in vitro translation mixtures (17,18). One candidate for mediating G␤ folding and G␤␥ dimer formation is the chaperonin CCT/TRiC complex (18 -20).
65-80% of newly translated proteins fold spontaneously. In eukaryotes, at least three systems of molecular chaperones help fold the other 20 -35% of cellular proteins: the heat shock proteins Hsp40 and Hsp70, the Hsp90 system, and the chaperonins (19,21,22). Chaperonin complexes are found in prokaryotes (Group I) and eukaryotes (Group II). Group I is the GroEL⅐GroES complex, in which GroEL is formed by two rings of seven identical 57-kDa monomers that nonspecifically recognize proteins with exposed hydrophobic surfaces. The eukaryotic Group II chaperonins are called CCT (chaperonin containing TCP-1) or TRiC (TCP-1 Ring complex); TCP-1 is tailless complex polypeptide-1 (22). Eukaryotic CCT/TRiC is considerably more complex than GroEL, with eight different subunits that share 30% sequence identity, and it recognizes specific but poorly defined sequences in target proteins (19,23,24). CCT/TRiC was originally characterized in the folding of actin and tubulin, but is now recognized to be involved in the folding of many different proteins (19,21,22), a subset of which includes some WD repeat proteins (19,25), although it is not involved in the folding of all WD repeat proteins (26).
Isolated CCT/TRiC subunits were identified as interacting partners of the yeast G␤ homolog Ste4 in a high throughput screen of yeast genome coding regions (27), but this did not extend to functional characterization of these interactions or to association with the CCT/TRiC complex. More recently, small interfering RNA to TCP-1␣ was shown to decrease G␤␥ levels in cells (20), indicating the possibility that some cellular component involved in G␤␥ dimer formation is dependent on the function of CCT/TRiC, and this was proposed as a site of regulation of G␤␥ dimer formation. Despite these observations, there is no direct evidence that CCT/TRiC does or even could participate either in folding of G␤ or in G␤␥ dimer formation. There is not even any information about whether eukaryotic G␤ or G␤␥ interacts with CCT/TRiC. Here, we have studied the mechanism of G␤␥ dimer formation in rabbit reticulocyte lysate and describe the role of CCT/TRiC both in G␤ folding and in G␤␥ dimer formation.

MATERIALS AND METHODS
Vectors-All G␤ and FLAG-G␥ constructs were as described previously (12). All FLAG-G␤ constructs were obtained from the Guthrie cDNA Resource Center (available at www.cdna.org).
In Vitro Translation and Transcription-In vitro translation of subunits was carried out using the TNT Quick-Coupled Transcription/Translation System (manual no. TM045, Promega Corp.) as described previously (12). Any additions to the translation mixture were added before DNA and lysate.
G␤␥ Dimerization Assay by Immunoprecipitation-Quantitation of synthesized protein and the dimerization reaction were done as described previously (12).
Immunoprecipitation of Synthesized Subunits-Immunoprecipitation of the FLAG-tagged subunits from translations and/or dimerizations was carried out using agarose-conjugated anti-FLAG antibody (catalog no. A-2220, Sigma) with 5-20 l of beads/sample. Beads were washed twice with 50 mM Tris (pH 7.4), 100 mM NaCl (Tris-buffered saline (TBS) 2 ), 1 mg/ml bovine serum albumin, and 0.1% C 12 E 10 (polyoxyethylene 10-lauryl ether) (TBSBC). The beads were blocked against nonspecific binding by incubation for 1 h at room temperature in 10 volumes of TBSBC with 5% reticulocyte lysate and then washed twice with 50 mM Tris (pH 7.4), 100 mM NaCl, and 0.1% C 12 E 10 (TBSC). For immunoprecipitation, the sample was diluted 10 -20-fold with TBSC; 5-20 l of beads were added; and the samples were incubated for 1 h at room temperature. The samples were centrifuged, and the beads were resuspended in TBSC and transferred to a new tube to reduce nonspecific binding of G␤. The samples were washed a second time. Electrophoresis sample buffer was added to the beads to elute the bound proteins.
Electrophoresis-Electrophoresis was carried out by the Laemmli procedure using Criterion Tris-HCl precast gels (Bio-Rad). Native (nondenaturing) gel electrophoresis was carried out following a procedure modified from that of Hansen et al. (28) also using Criterion gels, but SDS and ␤-mercaptoethanol were omitted from both sample and electrophoresis buffers. The gels were run at 4°C at 0.4 A with an initial voltage of 140 V for 20 -30 min and then 180 -200 V for the remainder of the run. For autoradiography, the gel was washed twice for 15 min with 50% methanol and 10% acetic acid and then fixed for 5 min with 7% acetic acid, 7% methanol, and 1% glycerol before drying. Dried gels were exposed to film overnight at Ϫ80°C or to a phosphor storage screen for 3-5 days at room temperature. The screen was imaged and analyzed using a GE Healthcare Storm instrument and ImageQuant software. Molecular mass standards for nondenaturing gels were from Amersham Biosciences (high molecular mass calibration kit) and included thyroglobulin (669 kDa), ferritin (440 kDa), catalase (232 kDa), lactate dehydrogenase (140 kDa), and bovine serum albumin (67 kDa).
Proteomics Analysis of Samples Recovered after SDS-PAGE-FLAG-tagged G␤ 1 was expressed in vitro as described previously (16) and immunoprecipitated with anti-FLAG beads. Immunoprecipitated protein was eluted from the beads with SDS sample buffer without ␤-mercaptoethanol and was run on 4 -20% Criterion gels. Gels were stained with Coomassie Blue and destained with 10% acetic acid. Bands were excised with a scalpel and further processed in a PerkinElmer Life Sciences MultiPROBE II HT EX liquid-handling robot equipped with a Millipore Montage In-Gel Digest ZP manifold. Gel slices were sequentially equilibrated with 25 mM ammonium bicarbonate and 5% acetonitrile, then with 25 mM ammonium bicarbonate and 50% acetonitrile, and finally with 100% acetonitrile. Protein was digested by reconstituting dehydrated gel slices with 15 l of 10 g/ml trypsin in 25 mM ammonium bicarbonate and incubating for 3 h at 37°C. Recovered peptides were analyzed by MALDI-TOF/TOF on an Applied Biosystems 4700 proteomics analyzer. Mass spectra were generated from 3000 -5000 laser shots, and the 10 most intense protein peaks were sequenced by tandem mass spectrometry (MS/MS). The lower and upper limits of mass detection were 1000 and 3000 Da, respectively. Data were analyzed with an ABI GPS Explorer 3.5 running Mas-cot Version 1.9.05 for peak identification (31). The minimum signal-to-noise ratio was 10.0, and assignments were based upon a precursor tolerance of 1.0 Da, although most assigned peaks had a tolerance of Ͻ0.2 Da. Two data sets were analyzed: one of 4981 records containing all NCBI non-redundant rabbit (Oryctolagus cuniculus) proteins and the other of 163,861 records containing all NCBI non-redundant rodent proteins. Significant matches were identified based upon a Mascot protein score with Ͼ95% confidence interval for identification of the protein contained in the library. For the rabbit data base, this included all identified peaks with protein scores Ͼ44, whereas for the rodent data base, this included peaks with protein scores Ͼ63. 3 -G␤ 1 , G␤ 2 , G␤ 3 , G␤ 4 , G␤ 5 , and FLAG-G␥ 3 were synthesized in the reticulocyte lysate in the presence of [ 35 S]methionine ( Fig.  1A), and quantitated as described (12). G␤ and FLAG-G␥ 3 were added together at a 2:1 stoichiometry, allowed to form dimers for 2 h at 37°C, and then immunoprecipitated with anti-FLAG beads. As shown in Fig. 1B, G␤ 1 , G␤ 2 , and G␤ 4 were easily co-immunoprecipitated with FLAG-G␥ 3 , indicating that G␤ 1 ␥ 3 , G␤ 2 ␥ 3 , and G␤ 4 ␥ 3 dimers readily form. In contrast, no G␤ 5 was detectable in immunoprecipitates of FLAG-G␥ 3 coincubated with this G␤, indicating that G␤ 5 forms dimers very poorly with G␥ 3 . In immunoprecipitates with G␤ 3 , much less G␤ 3 was present (relative to G␥ 3 ) than was seen with G␤ 1 , G␤ 2 , or G␤ 4 , but G␤ 3 was clearly present as compared with G␤ 5 or the control lanes. These results agree with our recent characterization of G␤␥ dimer association (12), which showed that G␤ 1 and G␤ 4 readily form dimers with all G␥ isoforms; that G␤ 2 selectively forms dimers at good efficiency with most G␥ isoforms, but not G␥ 1 , G␥ 8c , G␥ 11 , or G␥ 13 ; that G␤ 5 does not form dimers in the assay with any G␥ isoforms; and that G␤ 3 selectively binds G␥ isoforms, but at a low level even in the best of cases.

Analysis of G␤ and G␥ Subunits on Nondenaturing Gels-
The inference from our previous work (12) and from the data in Fig. 1, as well as from analogous experiments with mutants of conserved Asp residues in G␤ (18), was that other constituents of the reticulocyte lysate interact with G␤ isoforms and are involved in G␤ subunit dimerization with G␥. Nondenaturing (native) gel analysis was used to identify constituents of the reticulocyte lysate that interact with either G␤ or G␥. Interpretation of results with nondenaturing gels is more complex than with SDS-polyacrylamide gels because protein migration is determined by multiple properties, including molecular mass, pI, and shape. However, native gels preserve many interactions that would be disrupted with SDS and/or ␤-mercaptoethanol, and so migration is also affected by formation of protein complexes. Taking all of these factors into account, globular proteins (or complexes) of similar pI will still travel on nondenaturing gels according to their molecular mass (or log molecular mass). This was (approximately) true, for example, for standards used in the characterizing the mobility of G␤, G␥, and G␤␥ dimers (Fig. 2), where we used a set of globular proteins with pI values of 5.4 -6.8. The G␤␥ dimer deviates somewhat, but not drastically, from being a globular protein (14), and it has a pI dominated by the G␤ component predicted to be between 5.6 and 6.0 for the different isoforms. Thus, our expectation was that the dimer, and perhaps the folded G␤ protein alone, would travel with an apparent molecular mass on nondenaturing gels only slightly greater than its predicted 45 kDa. On the other hand, G␥ alone is small (8 kDa) and of heterogeneous pI among its isoforms and appears to have defined structure based primarily upon its association with G␤ (14). Thus, the behavior of G␥ on nondenaturing gels was less predictable.
When in vitro translated, 35 S-labeled G␥ isoforms were analyzed on native gels ( Fig. 2A), they did not travel with similar mobility, as they would on SDS-polyacrylamide gels. In fact, they appeared to segregate primarily according to their pI, as shown in the accompanying graph (Fig. 2B). Those G␥ subunits with a pI near 7, i.e. G␥ 2 , G␥ 3 , G␥ 4 , G␥ 8c , G␥ 8olf , and G␥ 10 , migrated in the middle of the gel. Those with a pI Ͻ7, i.e. G␥ 1 , G␥ 11 , and G␥ 13 , migrated more rapidly on the gel. And those with a pI Ͼ7, i.e. G␥ 5 , G␥ 7 , and G␥ 12 , either did not enter the gel (i.e. G␥ 5 ) or migrated near the top. There were also two nonspecific bands that were present in all of the lanes, including the pcDNA control lane ( Fig. 2A, arrows). For some G␥ isoforms, for example, G␥ 8c , G␥ 8o , and G␥ 11 , two bands were seen. G␥ subunits have post-translational modifications, including by farnesylation or geranylgeranylation at the C terminus and by N-terminal processing that can include removal of methionine and/or acetylation of the N terminus. Variable post-translational modification at these sites, as we have described recently for brain G␥ subunits (32), may affect the hydrophobicity and/or shape of the G␥ subunits such that their migration on nondenaturing gels is affected.
The behavior of G␤ isoforms on native gels ( Fig. 2C) was quite different from that of the G␥ isoforms ( Fig. 2A). For all the in vitro transcribed G␤ isoforms, 35 S-labeled protein accumulated in two primary locations (Fig. 2C): at the top of the native gel, presumably as denatured, aggregated protein of a size too great to enter the gel, and as a sharp, well delineated band at FIGURE 1. Dimerization of G␤ 1 , G␤ 2 , G␤ 3 , G␤ 4 , and G␤ 5 with FLAG-G␥ 3 . A, 35 S-labeled G␤ 1 , G␤ 2 , G␤ 3 , G␤ 4 , and G␤ 5 and FLAG-G␥ 3 proteins were expressed in the reticulocyte lysate. Equal aliquots of the total translation mixture were run on a 10 -20% gradient gel. A PhosphorImager image of the dried gel is shown. The approximate positions of molecular mass standards (in kilodaltons) are shown on the left. B, G␤ 1 , G␤ 2 , G␤ 3 , G␤ 4 , and G␤ 5 were incubated with FLAG-G␥ 3 at a 2:1 molar ratio, and FLAG-G␥ 3 and any associated G␤ subunits were immunoprecipitated (lanes 6 -11). Controls included G␤ subunits alone without FLAG-G␥ 3 (lanes 1-5) and FLAG-G␥ 3 alone (lane 11). Samples were analyzed as described for A, but on an 8 -16% gradient gel. The approximate positions of molecular mass standards (in kilodaltons) are shown on the right.
Ͼ669 kDa (band a). This band is specific to translation of G␤ and was not found in the pcDNA control or after translation of G␥ (Fig. 2C). There was also some labeled material near the 67-kDa marker that was diffuse for G␤ 1 , G␤ 2 , G␤ 3 , and G␤ 4 ; sharper for G␤ 5 (band f); and prominent but diffuse for G␤ 5L (band e). This material migrated at a molecular mass slightly larger than expected for G␤ subunits, perhaps representing incompletely folded protein. There were also two nonspecific bands (bands b and d), also seen with G␥ ( Fig. 2A), and a band of variable intensity at ϳ140 kDa (band c), which might be a dimer of the material at ϳ60 -70 kDa or G␤ complexed with other specific proteins. Most interesting in these samples are the bands at Ͼ669 kDa. Because G␤ was the only significantly labeled protein on this gel (see Fig. 1A, for example), the best explanation for band a is that it is G␤ migrating at a very high molecular mass in association with a multiprotein complex. The fact that this is a sharp band supports this idea because denatured and/or aggregated protein either would not enter the gel (as seen at the top) or would have heterogeneous mobility and produce a protein smear, as would incompletely or randomly folded protein.
Characterization of the Interaction of G␤ with the Chaperonin CCT/TRiC Complex-The apparent size of the interacting partner on native gels and previous reports that some WD repeat proteins interact with or are folded by CCT/TRiC (19,25,27) suggested that one candidate for the binding partner affecting the mobility of G␤ subunits on native gels is the chaperonin CCT/TRiC complex. The intact CCT/TRiC protein, with an approximate molecular mass of ϳ800 kDa, migrates very slowly on a nondenaturing gel (28,33). This is shown in Fig. 3A, where samples of the reticulocyte lysate with or without expressed proteins (FLAG-G␤ 1 and FLAG-G␥ 3 ) were run on a native gel and immunoblotted with an antibody specific for the ␣-subunit of CCT/TRiC (TCP-1␣). All these samples contained a high molecular mass band recognized by anti-TCP-1␣ antibody at a high molecular mass similar to that seen in the G␤ subunit lanes (Fig. 2). Thus, G␤ synthesized in the reticulocyte lysate migrates on native gels in a region also containing a high molecular mass complex of the CCT/TRiC subunits. To more definitively evaluate whether G␤ and CCT/TRiC co-migrate on nondenaturing gels, we stained the immunoblot with antibodies tagged with infrared dyes that are amenable to double staining of immunoblots (Fig. 4). FLAG-G␤ 1 , FLAG-G␤ 2 , or the pcDNA control was translated in the reticulocyte lysate and then separated on either nondenaturing (Fig. 4A) or denaturing (SDS) (Fig. 4B) gels. Proteins transferred to nitrocellulose were The pI values for the G␥ subunits were predicted using the Compute pI/Mw tool available on the ExPASy proteomics server (available at us.expasy.org/tools/pi_tool.html). C, native (nondenaturing) gel analysis of G␤ subunits. G␤ subunits or FLAG-G␥ 3 was expressed in the reticulocyte lysate and labeled with [ 35 S]methionine. Samples were run on a nondenaturing 4 -15% gradient gel, and the gel was dried and autoradiographed. Band a, G␤ of high apparent molecular mass; band b, nonspecific band present in all samples; band c, band seen with G␤ 1 , G␤ 2 , G␤ 3 , and G␤ 4 ; band d, nonspecific band present all samples; band e, band specific to G␤ 5L ; band f, band specific to G␤ 5 . The gray arrow indicates the top of the gel. Note that the position of TCP-1␣ on the gel, traveling slightly slower than the 669-kDa standard, is similar to the position of the G␤ band labeled a in Fig. 2C. B, G␤ 1 co-immunoprecipitates with CCT/TRiC. Unlabeled FLAG-G␤ 1 and unlabeled FLAG-G␥ 3 were separately expressed and immunoprecipitated (IP) using anti-FLAG beads. Duplicate samples were separated by SDS-PAGE on an 8 -16% gel, transferred, and immunoblotted with rat monoclonal anti-TCP-1␣ antibody (left) or rabbit anti-G␤ subunit antibody BC1 (right). Lanes 1, FLAG-G␤ 1 ; lanes 2, FLAG-G␥ 3 ; lanes 3, pcDNA3. The approximate positions of molecular mass standards (in kilodaltons) are indicated on the right. The secondary antibody for TCP-1␣ is anti-rat, whereas the secondary antibody for BC1 is anti-rabbit. (The anti-rat secondary antibody used in the TCP-1␣ immunoblot recognized mouse heavy and light chains (indicated with gray arrows on the left) that co-eluted from the anti-FLAG beads (containing a mouse monoclonal antibody).) Note that TCP-1␣ co-immunoprecipitated only with FLAG-G␤ 1 (lanes 1), but not with FLAG-G␥ 3 (lanes 2) or with anti-FLAG beads from a mock translation (lanes 3).
blotted simultaneously with anti-TCP-1 and anti-FLAG antibodies and visualized with secondary antibodies tagged with different infrared-emitting dyes (green for TCP-1 and red for FLAG in Fig. 4). On SDS-polyacrylamide gels, G␤ and TCP-1 traveled independently of one another and at their expected monomer molecular masses (Fig. 4B). On nondenaturing gels, however, G␤ and TCP-1 traveled at a high molecular mass that could be co-localized as a merged signal of the two tagged antibodies. These data demonstrate that G␤ and TCP-1 co-migrate on nondenaturing gels, suggesting their association in a complex.
To determine whether G␤ and the CCT/TRiC subunits are actually physically associated, we evaluated whether the subunits co-immunoprecipitate. Unlabeled FLAG-G␤ 1 , FLAG-G␥ 3 , or the pcDNA control was expressed in lysate and subjected to immunoprecipitation with anti-FLAG beads (Fig. 3B). Only the FLAG-G␤ 1 immunoprecipitates contained TCP-1␣ immunoreactivity. Notably, the FLAG-G␥ 3 and pcDNA control samples contained only the heavy and light chains from the anti-FLAG antibody and nothing in the position of TCP-1␣ or G␤. Thus, FLAG-G␤, but not FLAG-G␥, binds to TCP-1␣. To more generally evaluate the specificity of association of CCT/ TRiC with G␤ and to confirm that the immunoreactivity observed in Fig. 3 was in fact TCP-1␣, we stained SDS gels of immunoprecipitates with Coomassie Blue (Fig. 5A). In comparing control (pcDNA3) and FLAG-␤1 samples, we observed several protein bands between 50 and 60 kDa associated with the FLAG-␤1 sample (Fig. 5B). The specificity of this complex of proteins for association with FLAG-G␤ was striking given that reticulocyte lysate is essentially intact cytosol containing a large amount of hemoglobin (Fig. 5C). To identify the constituents of this complex, we used mass spectrometric (MS) peptide map- FLAG-G␤ 1 and FLAG-G␤ 2 were individually synthesized with rabbit reticulocyte lysate in the presence of unlabeled methionine. Samples of the translated proteins or translation mixture incubated with pcDNA3 with no insert (control (C )) were run on 4 -20% gradient gels under native (A) and denaturing (SDS) (B) conditions, transferred, and immunoblotted simultaneously with anti-FLAG (rabbit) and anti-TCP-1 (goat) antibodies. Secondary antibodies were tagged with infrared-emitting labels: secondary antibody to FLAG (donkey anti-rabbit IgG) emitting at 700 nm and colored red and secondary antibody to TCP-1␣ (donkey anti-goat IgG) emitting at 800 nm and colored green. Immunoblots were scanned in an Odyssey infrared imager (LI-COR). In each case, the identical immunoblot is shown with its red-emitting signal (anti-FLAG antibody) and its green-emitting signal (anti-TCP-1 antibody) and as the simultaneous signal, with yellow indicating co-localization of the two signals. The approximate positions of molecular mass markers (in kilodaltons) are shown on the left. Note that G␤ and TCP-1 traveled at their subunit molecular masses in B, but co-migrated at high apparent molecular mass in A, an apparent molecular mass that is compatible with that of the intact CCT/TRiC complex. Antibody heavy chain (Ab(HC)) and light chain (Ab(LC)) bands common to both lanes are labeled on the right. Bands labeled TCP are unique to the G␤ 1 lane. The assignment of these bands as TCP subunits is based upon the data in Table 1. The position of FLAG-G␤ 1 is indicated on the right. This was a weak band commensurate with the low level of protein produced in the reticulocyte lysate (estimated to be 50 -100 ng in this sample). Stoichiometric amounts of CCT/TRiC to G␤ 1 should generate a ratio of TCP bands to a FLAG-G␤ band of ϳ30:1 based upon equivalent staining with Coomassie Blue. B, enlarged image of unique bands labeled TCP in the Coomassie Bluestained lane containing FLAG-G␤. The approximate positions of the five gel slices analyzed by MALDI-TOF/TOF after trypsin digestion are labeled 1-5, the data from which are summarized in Table 1. C, Coomassie Blue-stained gel of 1 l of rabbit reticulocyte lysate run between lanes containing molecular mass standards and demonstrating the specificity of the immunoprecipitation shown in A. JULY 21, 2006 • VOLUME 281 • NUMBER 29 ping and MS/MS sequencing of trypsin digests of the bands excised from the gel ( Table 1). The results were analyzed using the Mascot search algorithm (31) to evaluate which proteins in a library of 163,000 rodent proteins (NCBI Protein Database) could account for the MS fragments and MS/MS-sequenced peptides identified. For all five bands, the only proteins with significant protein scores were TCP subunits or predicted proteins from genome data bases that are TCP homologs. For the highest scoring matches, we obtained both MS fingerprinting data and MS/MS sequencing of multiple peptides. The only peptide identified in these experiments that was not a TCP subunit or homolog was an IgG heavy chain fragment. This is an expected contaminant in this molecular mass range from the immunoprecipitation procedure.

G␤␥ Assembly on CCT/TRiC
CCT/TRiC is thought to be a dynamic complex in cells, existing both as an intact complex and as individual subunits (34). To confirm the interaction of G␤ with CCT/TRiC and to evaluate whether different G␤ subunits interact selectively with TCP subunits or with the CCT/TRiC complex, we immunoprecipitated FLAG-G␤ constructs from the reticulocyte lysate and immunoblotted for all eight TCP subunits (Fig. 6). We characterized the level of each TCP isoform in the reticulocyte lysate as well as any TCP immunoprecipitated from a mock translation with pcDNA3 or with translated FLAG-G␥ 2 as controls (Fig. 6A). All of the anti-TCP isoform antibodies recognized bands in the lysate, except anti-TCP-1␦ antibody (data not shown). TCP-1␦ was probably present in the lysate, but the antibody (raised against a human sequence) may not recognize rabbit protein. Nevertheless, seven of the eight isoforms were readily identifiable. TCP isoforms were present in immunoprecipitates for all of the G␤ constructs, but none of the seven isoforms assayed was present in the pcDNA3 mock translation or in the immunoprecipitates of FLAG-G␥ 2 (Fig. 6). Samples (Fig. 6A) were normalized for FLAG-G␤ expression using anti-FLAG immunoblotting, and the level of each TCP isoform was compared for each of the G␤ isoforms (Fig. 6B). These results indicate that G␤, but not G␥, associates with the CCT/TRiC complex after synthesis and that G␤ associates with the complex and not just with isolated TCP subunits. This is a compelling explanation for the behavior of the G␤ isoforms on native gels (Fig. 2). In addition, the analysis of the relative amounts of TCP isoforms recovered with each G␤ subunit varied selectively with the G␤ isoform in the order of G␤ 5 Ͻ G␤ 3s Ͻ G␤ 3 Ͻ G␤ 2 Յ G␤ 1 Յ G␤ 4 (Fig. 6, B and C). Interestingly, this corresponds closely to our recent results describing the efficiency of specific G␤ isoforms forming functional G␤␥ dimers (12) and suggests that there is a relationship between the G␤ interaction with CCT/TRiC and dimer formation.
Assay of G␤␥ Dimer Assembly on a Nondenaturing Gel-The results with native gels (Fig. 2) suggested an approach for studying G␤␥ dimer formation, including the role of intermediates in this process, such as G␤ association with CCT/TRiC. To further investigate any relationship between the various G␤ and G␥ bands observed on native gels (Fig. 2) and the mobility of G␤␥ dimers, 35 S-labeled G␤ was combined with unlabeled G␥ 1 or G␥ 3 , or unlabeled G␤ was combined with 35 S-labeled G␥ 1 or G␥ 3 ; and after dimerization, samples were run on a nondenaturing gel (Fig. 7). Lanes 1-6 correspond to labeled G␤ 1 , and lanes 7-14 correspond to labeled FLAG-G␥. G␤ 1 without G␥ (lanes 1 and 2) had one specific band near the top of the gel at ϳ669 kDa. This band was seen only in samples containing labeled G␤ 1 (lanes 1-6) and not in samples containing labeled G␥ (lanes 7-14). New bands, not seen in the G␤-or G␥-only lanes, appeared in all lanes containing labeled G␤ 1 with FLAG-

MS fingerprinting and MS/MS sequencing identification of proteins that immunoprecipitate with FLAG-G␤ synthesized in reticulocyte lysates
FLAG-G␤ 1 was synthesized in the reticulocyte lysate and immunoprecipitated with anti-FLAG antibody. Control samples were translated in the presence of pcDNA3 lacking a functional insert. A representative Coomassie Blue-stained gel is shown in Fig. 5. The bands labeled 1-5 in Fig. 5B were excised from the gel with a scalpel and digested with trypsin as described under "Materials and Methods." Extracts from the trypsin digestion were analyzed by MALDI-TOF/TOF on an Applied Biosystems 4700 proteomics analyzer. The 10 most intense peaks in each spectrum were sequenced by MS/MS. Data were analyzed using ABI GPS Explorer 3.5 software implementing the Mascot search algorithm (31). a Bands were recovered from the Coomassie Blue-stained gel shown in Fig. 5. b Mascot was used to match MS and MS/MS data to protein sequences in NCBI non-redundant protein data bases composed of all rabbit proteins (4981 records) or all rodent proteins (163,861 records). The data reported are from matches to the rodent data base and indicate the rank of the match for which further information is given. Mascot analysis of the rabbit data base generated only a few low significance hits, probably because this data base contains only 11 TCP files, eight of which contain partial sequences of fewer than 30 residues, two of which contain Ͻ50% of the predicted sequence, and one of which contain an intact TCP sequence. c This indicates how many of the 10 highest scoring hits had sequences related to the one for which information is given. In the case of TCP isoforms, sometimes multiple isoforms were identified in a single sample, but we indicate only a single one here. The number in parentheses is the quantity of those hits whose identification had a protein score confidence interval Ͼ95%. d The NCBI accession number is for the protein indicated. e The species is for the identified protein. f A brief description of the protein from the NCBI record is given. g Listed is the predicted molecular mass of the identified protein. h The number of mass peaks in the MS record within 1 Da of the predicted mass of a trypsin fragment from the identified protein is given. i The values indicate the number of the 10 most intense mass peaks that were sequenced by MS/MS and that generated a sequence compatible with the identified protein.
j The values indicate the Mascot total protein score for the identified protein (31). k The values indicate the Mascot total protein score confidence interval (C.I.) for the identified protein (31). G␥ 1 or with FLAG-G␥ 3 (lanes 3-6); these bands represent the dimers G␤ 1 /FLAG-G␥ 1 and G␤ 1 /FLAG-G␥ 3 . The apparent size difference observed for the two dimers can be explained by the pI differences in the two G␥ subunits (see also Fig. 2B). The conclusion that the new bands seen with 35 S-labeled G␤ and unlabeled G␥ in lanes 3-6 are G␤␥ dimers is corroborated by the observation that bands at the identical positions were seen for G␤ 1 /FLAG-G␥ 1 and G␤ 1 /FLAG-G␥ 3 , respectively, in lanes 11-14, where the labeled partner was G␥ instead of G␤ 1 . The bands corresponding to G␤␥ dimers were absent in the lanes containing labeled G␥ without G␤ (lanes 7 and 8 for FLAG-G␥ 1 and lanes 9 and 10 for FLAG-G␥ 3 ) (immediately under the faster migrating nonspecific band) (Fig. 7B). Fig. 7B shows that the bands for FLAG-G␥ 3 and the G␤ 1 /FLAG-G␥ 3 dimer are different. Significantly, 35 S-labeled G␥ was not seen in association with the CCT/TRiC band when expressed either alone or with G␤ and forming a G␤␥ dimer (lanes 7-14). These experiments demonstrate that G␤␥ dimer formation can be observed on native gels and provide additional evidence for the stability of these dimers because they remain associated during electrophoresis.
To evaluate the relationship between G␤ and G␥ synthesis and dimer assembly, G␤ 1 , either alone or in combination with FLAG-G␥ 3 cDNA, was expressed in lysate with [ 35 S]methionine. Over the course of 2 h, samples were removed at the time points (minutes) shown in Fig.  8. Fig. 8A shows the time course of G␤ 1 subunit synthesis without FLAG-G␥ 1 . G␤ 1 could be seen after 10 min, migrating at the high molecular mass of the CCT/TRiC complex. When cDNAs for both subunits were included (Fig. 8B), FLAG-G␥ 1 appeared sooner than G␤ 1 , as early as 4 min, and the G␤ 1 / CCT/TRiC band appeared at 10 min, as for G␤ 1 alone (Fig. 8A). The presence of G␥ cDNA or G␥ protein (Fig. 8B) did not affect the timing of synthesis of G␤, although it did affect its distribution by decreasing the amount of both smeared radioactivity on the gel (presumably misfolded G␤ protein) and radioactivity at the top of the gel (aggregated protein). The G␤ 1 /FLAG-G␥ 1 dimer appeared by 30 min, after the appearance of both FLAG-G␥ 1 and the G␤ 1 /CCT/TRiC complex.

Role of CCT/TRiC in G␤␥ Dimer
Assembly-The results shown in Fig. 8 indicate that dimer formation follows G␤ interaction with CCT/TRiC and is compatible with a role of CCT/TRiC in G␤␥ dimer assembly. CCT/ TRiC has an integral ATPase activity that has been linked to closing of the CCT/TRiC folding chamber and facilitating protein folding as well as releasing the folded substrate from the complex (21,22,26). This ATPase activity has also been associated with complex formation in the assembly of specific multimeric structures (35)(36)(37). The functions of this ATPase activity can be blocked by addition of adenine nucleotide analogs such as ATP␥S and ADP␤S or by chelation of Mg 2ϩ . In addition, the "closed" state of the CCT/TRiC complex is thought to correspond to ADP⅐P i , the mid-hydrolysis state of ATP. ADP⅐P i FIGURE 6. Interaction of G␤ isoforms with the CCT/TRiC complex. A, FLAG-G␤ isoforms were translated in the reticulocyte lysate with unlabeled methionine and immunoprecipitated (IP) with anti-FLAG beads. The immunoprecipitates were run on an SDS-polyacrylamide gel, transferred, and separately immunoblotted (IB) with antibodies specific for the chaperonin TCP subunits ␣, ␤, ␥, ⑀, , , and . Each blot was subsequently blotted with anti-FLAG antibody to visualize FLAG-G␤. For the TCP immunoblots, only the region of the gels corresponding to 50 -65 kDa is shown; the expected molecular masses of the TCP subunits are all ϳ60 kDa. For the FLAG immunoblots, only the region of the blot corresponding to 30 -50 kDa is shown. Arrows indicate the positions of the individual TCP subunits. B, immunoblots for each chaperonin subunit (TCP) and its corresponding anti-FLAG blot were imaged with a Bio-Rad Fluor-S MAX MultiImager, and the signal was quantitated. The data were normalized to reflect the relative amount of a given chaperonin subunit for the amount of each G␤ subunit present. Each experiment was done a minimum of three times, and the normalized data were averaged. Data are graphed as mean Ϯ S.E. C, shown is a summary of the G␤ isoform interaction with the CCT/TRiC complex.
can be mimicked by ADP and by Al(NO 3 ) 3 , MgCl 2 , and NaF, whereby the fluoride coordinates with the aluminum in a manner that mimics the ␥-phosphate of ATP. Previous studies have not evaluated whether G␤␥ dimer formation requires ATP independently of G␤ or G␥ synthesis. One prediction of a role of CCT/TRiC in dimer formation would be that this process is ATP-dependent; if it were not, this would suggest that the role of CCT/TRiC is primarily in the folding of G␤, rather than participating in dimer formation directly. We therefore tested whether G␤␥ dimer formation is ATP-dependent. G␤ 1 , G␤ 2 , FLAG-G␥ 1 , and FLAG-G␥ 3 were individually expressed with unlabeled methionine or [ 35 S]methionine for 2 h at 37°C and then incubated in combinations containing one unlabeled and one labeled subunit. From the time course in Fig.  8, it is apparent that any requirement for folding of G␤, either by CCT/TRiC or other chaperones, can be completed in 60 min or less because the G␤␥ dimer will have already appeared by then. In addition, in other studies, we found that productive dimer formation did not proceed beyond 2 h (data not shown). Before combining the subunits, ATP; Al(NO 3 ) 3 , MgCl 2 , and NaF; ATP␥S; or ADP␤S was added to the expressed G␤ subunit (Fig. 9). Fig. 9A shows control results with 5 mM ATP. Incuba-tion of FLAG-G␥ 3 with either G␤ 1 or G␤ 2 resulted in generation of new 35 S-labeled G␤ bands and a slight change in the mobility of 35 S-labeled FLAG-G␥ 3 . Incubation of FLAG-G␥ 1 with G␤ 1 resulted in a new band that could be labeled with either 35 S-labeled G␤ or 35 S-labeled FLAG-G␥ 1 . In contrast and as expected from previous work (see "Discussion" in Ref. 12), FLAG-G␥ 1 did not form a dimer with G␤ 2 . During dimerization in the presence of Al(NO 3 ) 3 , MgCl 2 , and NaF (Fig. 9B); ATP␥S (Fig. 9C); or ADP␤S (Fig. 9D), the radioactivity associated with the dimer was abolished or greatly reduced. Conversely, the amount of 35 S-labeled G␤ migrating with CCT/ TRiC was increased under these same conditions. This suggests that the G␤ involved in dimer formation came from that associated with CCT/TRiC and that this complex is the relevant target for the effect of ATP analogs on the dimerization process. The nucleotide analogs did not promote G␥ association with CCT/TRiC, and G␤ 2 ␥ 1 , a dimer that does not normally occur, was not seen either as dimer or as an intermediate of that dimer. Additional experiments showed that chelation of Mg 2ϩ with EDTA also abolished dimer formation (data not shown).
The data in Fig. 9 suggest that CCT/TRiC mediates G␤␥ dimer formation via an ATP-dependent process that involves an interaction with G␤, but not G␥. We tested this conclusion by evaluating whether addition of G␥ affects G␤ association with CCT/TRiC. FLAG-G␤ 1 and G␥ 2 were independently synthesized with unlabeled methionine, and then either excess lysate (5-fold) containing G␥ 2 or mock-translated lysate (with pcDNA3) was added to lysate containing FLAG-G␤ 1 . After 90 min, FLAG-G␤ 1 was immunoprecipitated, and samples were blotted for TCP-1␥ and FLAG-G␤ 1 (Fig. 10). In the absence of G␥ 2 , TCP-1␥ immunostaining was more prominent than FLAG-G␤ 1 staining (Fig. 10B). Immunostaining of FLAG-G␤ 1 in these samples was about equal in intensity with the control (Fig. 10A) and represents immunoprecipitation of 20 -30% of the synthesized FLAG protein. This is compatible with the data in Fig. 8, suggesting that, in the absence of G␥, G␤ protein tends FIGURE 7. G␤␥ dimer formation assayed on a nondenaturing gel. To determine whether nondenaturing gels could be used to monitor G␤␥ dimer formation, G␤ or G␥ was independently labeled with [ 35 S]methionine, and their mobility on 4 -15% nondenaturing gels was evaluated with or without preincubation with the non-radioactively labeled complementary subunit. A: lanes 1-6, G␤ 1 was synthesized in the presence of [ 35 S]methionine and combined with G␥ 1 or G␥ 3 that was synthesized in the presence of unlabeled methionine. 35 S-Labeled G␤ 1 alone (lanes 1 and 2) traveled primarily at high molecular mass. If preincubated with either unlabeled G␥ 1 (lanes 3 and 4) or unlabeled G␥ 3 (lanes 5 and 6), 35 S-labeled G␤ 1 had increased mobility dependent on with which G␥ isoform it was preincubated (arrows). Lanes 7-14, G␥ 1 (lanes 7 and 8) or G␥ 3 (lanes 9 and 10) was synthesized in the presence [ 35 S]methionine; their mobility on the nondenaturing gel was isoformspecific. When unlabeled G␤ 1 was preincubated with labeled G␥ 1 (lanes 11 and 12) or with labeled G␥ 3 (lanes 13 and 14), a new band of altered mobility was observed at a similar position as observed for the 35 S-labeled G␤ 1 samples in the presence of unlabeled G␥ (lanes 3-6). Dimer-specific bands are indicated by arrows. B: an enlargement of the area in A shows G␥ 3 alone and G␤ 1 ␥ 3 -specific bands (arrows). FIGURE 8. Time course of dimer formation between G␤ 1 and FLAG-G␥ 1 on a nondenaturing gel. 35 S-Labeled G␤ 1 was expressed in the reticulocyte lysate alone (A) or with FLAG-G␥ 1 to allow dimer formation (B). Samples were taken at the times indicated (minutes after addition of both cDNAs) and run on a native gel. A, G␤ 1 at apparent high molecular mass (black arrow) complexed with CCT/TRiC was first seen at ϳ10 min and increased in amount for times up to 120 min. Increasing background radioactivity with time suggests accumulation of randomly folded and/or denatured protein with variable mobility. B, FLAG-G␥ 1 (gray arrow) was seen at ϳ4 min. G␤␥ 1 as a complex with CCT/TRiC (upper black arrow) was seen at 10 min, as when synthesized alone (as in A). The G␤ 1 /FLAG-G␥ 1 dimer (lower black arrow) appeared by 30 min, after the appearance of both FLAG-G␥ 1 and the high molecular mass G␤ 1 , and continued to accumulate at times up to 120 min.
to be misfolded or aggregated and less accessible to anti-FLAG antibody. If synthesized FLAG-G␤ 1 was incubated with lysate containing G␥ 2 , the amount of FLAG-G␤ 1 recovered in the immunoprecipitation increased to 70 -80% of the total (Fig. 10,  compare A and B), whereas the amount of TCP-1␥ associated with FLAG-G␤ 1 decreased both in amount relative to FLAG-G␤ 1 and in absolute amount. This result, particularly the absolute decrease in the amount of TCP-1␥ associated with FLAG-G␤ 1 , implies that G␥ and CCT/TRiC compete for binding of G␤ during the process of G␤␥ dimer formation.

DISCUSSION
The data reported here indicate that G␤ isoforms bind the chaperonin CCT/TRiC complex, which is known to be involved in protein folding (19,21,38) and to be a mediator of assembly for selected protein complexes (35,36). G␤ co-migrated with CCT/TRiC on nondenaturing gels (Figs. 2-4), and FLAG-G␤ co-immunoprecipitated CCT/TRiC (Figs. 2, 5, 6, and 10 and Table 1). These data indicate that G␤ associates with the CCT/ TRiC complex and not just with individual TCP subunits (Figs. 2 and 6). In contrast, CCT/TRiC did not associate with G␥ isoforms either on nondenaturing gels (Fig. 2) or by co-immunoprecipitation (Fig. 6). Most chaperones are major cellular constituents found in all cells. Despite this, analysis by MS fingerprinting and MS/MS sequencing (Table 1) of FLAG-G␤ immunoprecipitated from Coomassie Blue-stained gels (Fig. 5) identified only TCP subunits. This observation was confirmed by immunoblotting for other major chaperone proteins, including Hsp40, Hsp70, and Hsp90. 3 This does not negate the involvement of other proteins or cofactors, but it does suggest a primary role for CCT/TRiC in the post-translational events leading to dimer formation.
We report here substantial evidence for the role of CCT/TRiC in the folding of G␤ and as a major determinant or intermediate of G␤␥ dimer formation. These include the following. 1) G␤ specifically binds CCT/TRiC (Figs. 5 and 6), a complex involved in protein folding (19,21,38) and protein complex formation (35,36), shortly after its synthesis (Fig. 8); there is a lag of several minutes before dimer formation, as would be expected if folding of G␤ on CCT/TRiC is a requirement for production of functional G␤. 2) G␤ association with CCT/TRiC is isoform-specific (Fig. 6), and the specificity of these interactions correlates closely with the ability of different G␤ isoforms to form dimers in the same rabbit reticulocyte lysate system as described here (Fig. 1) and previously (12). The selectivity of these interactions argues against nonspecific association of unfolded or misfolded, newly synthesized G␤ with CCT/TRiC and is consistent with observations that the protein substrates of this complex interact with CCT/TRiC through specific sequences (19,23,24). 3) We show here for the first time that G␤␥ dimer formation is ATP-dependent and that this is a requirement separate from synthesis or folding of either G␤ or G␥ (Fig. 9). This is both a prediction and a test of the involvement of CCT/TRiC in dimer formation because nearly all of its activities, including folding, release of folded peptide, and complex formation, have been shown to be ATP-dependent (21,22,26,(35)(36)(37). That the target of this inhibition is CCT/TRiC is supported strongly by the observation that inhibition of ATPase activity leads to increased association of labeled G␤ association with the CCT/TRiC complex (Fig. 9). 4) Addition of G␥ to previously synthesized G␤ under conditions allowing dimer formation (Fig. 1) reduces G␤ association with CCT/ TRiC (Fig. 10), indicating that G␤ bound to CCT/TRiC is capable of functionally interacting with G␥ and that exposure of the complex to G␥ results in release of G␤ from CCT/TRiC.
Our results are summarized in the model shown in Fig. 11. Shortly after synthesis, G␤ associates with CCT/TRiC. This would most likely be mediated through specific sites of interaction on G␤, as is found for most well characterized substrates (19,21,38), and may require partial folding of the protein, per-  haps mediated by other chaperones. The specificity of this interaction would explain the variable association of different G␤ isoforms with CCT/TRiC (Fig. 6). ATP-dependent events would then be required to mediate folding of G␤ to a state competent to bind G␥ subunits. There is evidence that, for many substrates, the folding process on CCT/TRiC proceeds through rounds of binding and ATP-dependent release, ultimately leading to properly folded native protein (39). Release of properly folded protein in the presence of G␥ subunits would allow binding of the two partners and prevent rebinding of G␤ (as a G␤␥ dimer) to CCT/TRiC. This is essentially what we observed, i.e. when G␥ is added to presynthesized G␤, it results in dissociation of G␤ from CCT/TRiC (Fig. 10). This result is comparable with three other cases in which CCT/TRiC mediates complex formation: association of the von Hippel-Lindau protein with its partner elongin B/C (33,35), complex formation between histone deacetylase-3 and its binding partner SMRT (36), and formation of a cyclin E⅐CDK2 complex (37). Although the details differ somewhat between these various reactions, in all three cases, one component is folded on CCT/ TRiC and is removed from the complex in an ATP-dependent reaction by its binding partner, which itself does not appear to interact with the CCT/TRiC complex.
This work explains a number of past observations about G␤ folding and G␤␥ dimer assembly that include the following findings: G␤ does not fold correctly without G␥ (16,40); prokaryotes do not have the ability to assemble G␤␥ dimers (17); some in vitro systems, such as the plant wheat germ system, also lack the ability to form dimers (17); cellular requirements for dimer formation are not completely or necessarily met in in vitro assembly systems (12,18); and there are structural requirements associated with G␤ subunits that determine their ability to efficiently form dimers (12,18). It has been reported that the yeast G␤ homolog Ste4 associates with the TCP-1␣ subunit in a mass screening of interacting partners in yeast (19,27). Our results extend this observation to animal cells and to an association with the CCT/TRiC complex rather than its isolated TCP subunits. Notably, our results complement the work of Humrich et al. (20), which showed that small interfering RNA to TCP-1␣ decreases G␤␥ levels in human embryonic kidney 293 cells. Our studies here indicate that the requirement for TCP-1␣ is for folding G␤ and not for production of some other intermediate required for dimer formation. The previous experiments (20) are also important because they indicate that the mechanism we have described here using rabbit reticulocyte lysate is likely the same mechanism used by other cells.
CCT/TRiC was originally characterized in its folding of actin and tubulin (19,21,41,42) and then for other proteins, including the G␣ subunit of transducin (39). More recently, it has been characterized in the folding of several proteins that are assembled into complexes through a CCT/TRiC-dependent mechanism (33,(35)(36)(37). The folding of these proteins, as well as any associated complex formation, can require the participation of other proteins or chaperones, and there are multiple types of co-chaperones and mechanisms associated with these FLAG-G␤ 1 and G␥ 2 were separately synthesized in rabbit reticulocyte lysate for 2 h at 30°C. Duplicate samples of FLAG-G␤ 1 (10 l of lysate) were then combined with either the control (pcDNA3) or G␥ 2 -containing lysate (50 l) and incubated for 90 min to allow dimer formation. Samples were immunoprecipitated with anti-FLAG beads, run on a 4 -20% gradient gel, transferred to nitrocellulose, and immunostained first with antibody to TCP-1␥ and then with antibody to FLAG. A, immunoblot of 2 l of the starting lysate containing FLAG-G␤ 1 after sequential staining with anti-TCP-1␥ antibody and then anti-FLAG antibody. B, immunoblot of FLAG-immunoprecipitated (IP) samples after sequential staining with anti-TCP-1␥ antibody and then anti-FLAG antibody. Ab(HC) and Ab(LC), antibody heavy and light chains, respectively. C, lanes from same gels as in A and B containing 1 l of the final incubation mixture before immunoprecipitation. Only the TCP-1␥ band is shown, demonstrating that equivalent samples were loaded on the gel. G␤ associates with CCT/TRiC shortly after synthesis. Folding of proteins by CCT/TRiC is known to be ATP-dependent, as is release of the folded protein from the complex. We have not here found evidence of association of G␥ with CCT/TRiC or with other proteins, but its presence either during or after G␤ synthesis promotes release of G␤ from CCT/TRiC and formation of the G␤␥ dimer. This process is also ATP-dependent. This model does not exclude the participation of other chaperone proteins or cofactors such as PhLP, which may be a co-chaperone or a regulator of G␤ folding and G␤␥ assembly (20,46).
processes. In addition to the requirement for the CCT/TRiC complex shown here, Hsp90 (43) and phosducin-like protein (PhLP) (20, 44 -46) may participate in G␤ folding, the regulation of its folding, or the assembly of G␤␥ dimers. In the immunoprecipitation of FLAG-G␤ or FLAG-G␥, we did not identify any associated Hsp40, Hsp70, or Hsp90, 3 but such interactions may be transient or unstable. The behavior of G␥ subunits on native gels (Fig. 2) seems to be largely dictated by their pI and, in contrast to G␤, not by their interaction with other proteins. Some G␥ proteins do migrate as doublets on native gels (Fig. 2), which could indicate such interactions, but could also be explained by variable post-translational modification (32). Likewise, although the mobility of G␤ isoforms on native gels can be explained by their formation of a complex with CCT/ TRiC (or by accumulation of denatured, aggregated protein on the top of the gel as in Fig. 2), there are other bands, varying in position or intensity, associated with the different isoforms. These could represent complexes with other proteins (not CCT/TRiC) involved in folding or dimer assembly.
The G␤-binding protein PhLP has recently been reported to have a role in G␤ synthesis and G␤␥ dimer assembly (20,46). One previous study indicated that PhLP binds both G␤ and CCT/TRiC and that small interfering RNA knockdown of PhLP decreases G␤ synthesis and G␤␥ dimer assembly (46). That work led to the novel conclusion that PhLP is a molecular chaperone for G␤␥ dimer assembly. In contrast, another study suggested that PhLP overexpression negatively regulates G␤ synthesis and G␤␥ dimer assembly by an inferred interaction with CCT/TRiC, based upon small interfering RNA knockdown of TCP-1␣ (a CCT/TRiC subunit), decreasing G␤/G␤␥ production (20). It is not entirely clear how to reconcile those two studies with one another. It seems probable that PhLP is one of the components (co-chaperones) of the CCT/TRiC complex involved in folding G␤ and assembling the G␤␥ dimer. One possibility, perhaps compatible with both previous studies (20,46), is that different splice variants or phosphorylated forms of PhLP reciprocally regulate these processes. It will be particularly important to work out the sequence of events involved and the relationship of PhLP to the structure of the CCT/TRiC complex. PhLP may be a permanent component of a subset of CCT/TRiC complexes, specifically targeting them to G␤, or it may be a transient component interacting with G␤ or G␤␥ at a specific stage of folding and assembly. It could be an active participant in the folding and dimer assembly process, or it could be a regulatory component. Whatever these roles are, our data indicate that the CCT/TRiC complex provides the scaffold for the events mediated by PhLP and that CCT/TRiC itself must be an active participant in the processes because PhLP is not know to possess ATPase activity, which appears to be an integral part of the dimer assembly process.
Recent studies suggest that G protein subunit isoforms extensively regulate each other's expression in cells at both the protein and mRNA levels (47). G␤␥ dimer expression is regulated both by controlled proteolysis through the proteasome (20,48,49) and possibly by regulation of dimer formation (20,44). Various phosducin-like proteins have been shown to be either negative regulators of CCT/TRiC function (44,45), suppressing G␤␥ dimer expression (20), or co-chaperones in the folding of G␤ (46). The results reported here and the model in Fig. 11 provide a mechanism to explain these observations and suggest possible associated physiological consequences. In contrast to the other known protein dimers assembled by CCT/ TRiC, G␤␥ dimers represent a combinatorially complex family of proteins with at least 60 possible dimer combinations (7). A large number of these pairs, but not all, are biochemically compatible (Ref. 12 and references therein). One possible consequence of the model in Fig. 11 would be that the G␤ protein on CCT/TRiC provides an uncommitted reservoir of protein capable of responding to changing cellular requirements for different G␤␥ dimers, dependent upon production of the small and highly variable G␥ subunit. Thus, this mechanism of G␤␥ dimer production (Fig. 11) may provide a "proofreading" step whereby G␤ remains in reserve on CCT/TRiC until a correct G␥ subunit is provided to stabilize its release from the complex. Through this, the folding of G␤ by CCT/TRiC allows an additional level of regulation of the heterotrimeric G proteins based upon their inherent specificity for dimer assembly.