Advertisement

TRiC/CCT cooperates with different upstream chaperones in the folding of distinct protein classes

Katja Siegers, Bettina Bölter, Juliane P. Schwarz, Ulrike M. K. Böttcher, Suranjana Guha, F. Ulrich Hartl

Author Affiliations

  1. Katja Siegers*,1,
  2. Bettina Bölter1,
  3. Juliane P. Schwarz1,
  4. Ulrike M. K. Böttcher1,
  5. Suranjana Guha1 and
  6. F. Ulrich Hartl1
  1. 1 Max Planck Institute of Biochemistry, Department of Cellular Biochemistry, D‐82152, Martinsried, Germany
  1. *Corresponding author. E-mail: siegers{at}biochem.mpg.de

Abstract

The role in protein folding of the eukaryotic chaperonin TRiC/CCT is only partially understood. Here, we show that a group of WD40 β‐propeller proteins in the yeast cytosol interact transiently with TRiC upon synthesis and require the chaperonin to reach their native state. TRiC cooperates in the folding of these proteins with the ribosome‐associated heat shock protein (Hsp)70 chaperones Ssb1/2p. In contrast, newly synthesized actin and tubulins, the major known client proteins of TRiC, are independent of Ssb1/2p and instead use the co‐chaperone GimC/prefoldin for efficient transfer to the chaperonin. GimC can replace Ssb1/2p in the folding of WD40 substrates such as Cdc55p, but combined deletion of SSB and GIM genes results in loss of viability. These findings expand the substrate range of the eukaryotic chaperonin by a structurally defined class of proteins and demonstrate an essential role for upstream chaperones in TRiC‐assisted folding.

Introduction

A substantial fraction of newly synthesized polypeptides depends on molecular chaperones for efficient folding (Bukau et al., 2000; Frydman, 2001; Hartl and Hayer‐Hartl, 2002). Chaperones involved in de novo protein folding in the cytosol either interact co‐translationally with nascent chains or function downstream of translation. Among the first class of components, the heat shock protein (Hsp)70s (and their Hsp40 cofactors) in bacteria and eukarya bind and release, in an ATP‐dependent manner, hydrophobic segments exposed by nascent chains, and are thought to prevent a wide range of polypeptides from misfolding and aggregation during translation (Beckmann et al., 1990; Nelson et al., 1992; Pfund et al., 1998; Deuerling et al., 1999; Teter et al., 1999). In contrast, the ATP‐dependent chaperonins of the GroEL‐ and TRiC/CCT‐type engage their substrates either post‐translationally or late in translation, and promote the folding of a subset of cytosolic proteins, ∼10% by mass (Ewalt et al., 1997; Thulasiraman et al., 1999). They form large cylindrical complexes with a central cavity that provides a sequestered space in which the folding of complete proteins (or protein domains) may occur unimpaired by aggregation (Bukau and Horwich, 1998; Hartl and Hayer‐Hartl, 2002; Meyer et al., 2003).

Despite considerable progress in the mechanistic understanding of individual components, how nascent chain‐binding chaperones and the chaperonins cooperate in protein folding and how their functions are coordinated with protein synthesis have remained largely unexplored. One difficulty in defining the pathways of de novo folding in the cytosol results from a high degree of functional redundancy among chaperone components. For example, both eukaryotic and prokaryotic cells contain chaperones that interact directly with ribosomes, such as Trigger factor in bacteria and the Hsp70s of the Ssb‐type in yeast. These factors are probably important in functionally coupling translation with the action of the protein folding machinery, however, their individual deletion does not result in significant defects in protein folding.

The chaperone machinery in the eukaryotic cytosol has been studied in most detail in the yeast Saccharomyces cerevisiae (Bukau et al., 2000). Like other eukaryotic cells, yeasts do not have Trigger factor but contain a number of other ribosome‐associated factors with putative chaperone function, including NAC (nascent chain associated complex), RAC (ribosome associated complex) and Ssb1/2p. NAC, a heterodimer of α‐ and β‐subunits (Wiedmann et al., 1988), interacts with many nascent polypeptides and has been assigned a role in protein sorting to mitochondria and the ER (Funfschilling and Rospert, 1999; Wiedmann and Prehn, 1999). NAC may also fulfil a Trigger factor‐like function; however, deletion of the genes encoding NAC subunits does not significantly affect the growth of yeast (George et al., 1998; Reimann et al., 1999). RAC, a complex of the Hsp70 homologue Ssz1p and the Hsp40 homologue zuotin (Gautschi et al., 2001), binds to ribosomes and was reported to mediate the interaction of the Ssb‐type Hsp70s with nascent chains (Gautschi et al., 2002). The Ssbs are highly abundant nascent chain‐binding chaperones and exist as both ribosome‐associated and free populations (Nelson et al., 1992; Pfund et al., 1998). They are functionally distinct from the Ssa‐type Hsp70s, which act post‐translationally in protein folding/assembly and sorting (Bukau et al., 2000; Frydman, 2001). Again, neither the Ssb proteins nor RAC are essential for viability (Nelson et al., 1992; Gautschi et al., 2001).

In contrast, the function of the eukaryotic chaperonin TRiC (also known as CCT) is indispensable. TRiC forms an ∼900 kDa complex consisting of two rings with eight orthologous ∼60 kDa subunits each (Frydman, 2001; Valpuesta et al., 2002). While originally thought to be a chaperone highly specialized for the folding of actin and tubulins, recent work suggests that TRiC participates in the folding and assembly of a wider range of substrates, including G‐α‐transducin (Farr et al., 1997) and the von Hippel‐Lindau tumor suppressor protein (Feldman et al., 1999; Melville et al., 2003). Interestingly, a recent proteomic analysis of protein complexes in yeast identified a number of WD40 repeat proteins as interaction partners of TRiC‐subunits (Ho et al., 2002; Valpuesta et al., 2002), suggesting that these proteins may also be chaperonin substrates.

TRiC cooperates with the co‐chaperone GimC (also referred to as prefoldin), a complex of six distinct but structurally related proteins of 13–23 kDa (Geissler et al., 1998; Vainberg et al., 1998). Prefoldin binds to nascent chains in a cell‐free translation extract (Hansen et al., 1999) and is thought to target substrate proteins to TRiC. Loss of GimC function in yeast is not lethal but results in a substantially reduced efficiency of actin and tubulin folding (Siegers et al., 1999).

We have undertaken a combined genetic and biochemical approach in S.cerevisiae to define better the contribution of TRiC to protein folding and to understand its functional cooperation with upstream chaperones. We show that a group of WD40 domain proteins are in fact substrates of TRiC and require the chaperonin to reach their native state. The folding of these proteins is GimC‐independent and instead involves a functional cooperation between TRiC and the Ssb‐type Hsp70s. Strikingly, while GimC can replace Ssb1/2p in the folding of WD40 substrates, combined deletion of the SSB and GIM genes results in loss of viability, presumably as a result of a defect in TRiC‐assisted protein folding. These findings demonstrate the essential nature of the cooperation between nascent chain‐binding chaperones and the chaperonin.

Results

Ssb1/2p, GimC and TRiC have related chaperone functions

In Escherichia coli, deletion of the gene encoding DnaK in a strain lacking the ribosome‐associated Trigger factor is lethal under appropriate growth conditions (Deuerling et al., 1999; Teter et al., 1999), indicating that these chaperones provide overlapping functions in de novo protein folding. To investigate the possible cooperation of different chaperone systems in protein folding in the eukaryotic cytosol, we screened a variety of chaperone mutants in yeast for synthetic lethality.

We found that deletion of both SSB genes (SSB1 and SSB2) strongly reduced the viability of three conditionally lethal alleles of the α‐subunit of the chaperonin TRiC under permissive conditions (Table I). Deletion of various subunits of GimC has shown similar effects on these TRiC mutants (Siegers et al., 1999). Interestingly, deletion of SSB1/2 in yeast cells lacking all six GIM genes resulted in a severe growth defect and eventual loss of viability (Figure 1A). In contrast, loss of GimC function in strains lacking RAC or NAC subunits did not cause a synthetic growth defect (data not shown).

Figure 1.

Ssb1/2p, GimC and TRiC have related chaperone functions. (A) The effect of simultaneous loss of Ssb1/2p and GimC function was determined by a plasmid shuffle experiment. Cells lacking all six GIM genes encoding the subunits of GimC were transformed with a URA3‐based plasmid carrying SSB2, and subsequently deleted for one (SSB1) or both chromosomal copies (SSB1, SSB2) of the SSB gene. The resulting strains, as well as wild‐type, Δssb1Δssb2 and Δgim1‐6 cells, also transformed with the URA3‐based SSB2 plasmid, were tested for growth on plates in the absence or presence of 5′‐FOA at 30°C. Counter‐selection for the URA3‐based SSB2 plasmid on 5′‐FOA plates shows that yeast can tolerate deletion of both SSB genes or deletion of GIM1‐6 but their combined deletion results in loss of viability. (B) Lysates from wild‐type yeast cells were fractionated on sucrose density gradients. Lysates were treated with cycloheximide (CHX) to stabilize polysomes (left) or with RNase to disrupt polysomes (right). Gradient fractions were analysed by western blotting with antibodies against Ssb, GimC, TRiC and the ribosomal protein L3. The position of polysomes in the gradient was determined by monitoring A260. The top of the gradient is shown on the left, and the bottom is on the right. (C) Association of TRiC and GimC with polysomes is not affected by the absence of Ssb1/2p. Cell lysate from the SSB1/2 deletion strain was analysed as in (B). (D) Co‐immunoprecipitation of TRiC and Ssb1/2p. Immunoprecipitation of TRiC from ATP‐depleted cell lysates was carried out with TRiC‐specific antibodies and protein A–Sepharose‐coupled pre‐immune serum as a control. Western blots of the total cell extracts (TE; in this and subsequent figures, 5 μg for detection of Ssb and Ssa, and 10 μg in all other cases) and immunoprecipitates from lysates containing 200 μg total protein (IP/control) were analysed with antibodies against TRiC, Ssb, GimC and Ssa.

View this table:
Table 1. Genetic interactions of SSB and GIM with TCP1 genes

The Ssb‐type Hsp70s bind newly synthesized polypeptide chains on translating ribosomes. Since loss of Ssb1/2p function drastically affects the viability of GimC and TRiC mutants, we analysed the association of these chaperones with translating ribosomes. As reported previously, sucrose gradient fractionation of cell lysates revealed the association of 35–40% of total Ssb1/2p with translating ribosomes (Nelson et al., 1992; Pfund et al., 1998). In addition, substantial amounts of GimC (15–20% of total) and TRiC (10–15% of total) were also found to cofractionate with polysomes (Figure 1B), extending previous findings from in vitro translation extracts to the in vivo situation (Hansen et al., 1999; McCallum et al., 2000). The association of these chaperones with translating ribosomes was abolished by treatment of the lysates with RNase A, which causes dissociation of polysome complexes and ribosome release of nascent chains.

Interestingly, antibodies against TRiC co‐precipitated both Ssb1/2p and GimC when ATP was depleted, whereas an interaction of the Ssa‐type Hsp70s with TRiC was not detected (Figure 1D). Together these results suggest that Ssb1/2p and GimC both function in TRiC‐assisted folding.

Actin folding requires GimC and TRiC but not Ssb1/2p

The association of GimC and TRiC with translating ribosomes was undiminished in a SSB1/2 deletion strain (Figure 1C), excluding a simple model in which Ssb1/2p recruits GimC/TRiC to nascent chains. Nevertheless, immunoprecipitation of actin and α‐tubulin in ATP‐depleted lysates of wild‐type cells resulted in the co‐precipitation of Ssb1/2p, GimC and TRiC. From the amount of co‐precipitated chaperones it could be calculated that actin and α‐tubulin each occupy ∼15–20% of total TRiC, suggesting that ∼50–60% of the chaperonin capacity is devoted to the folding of actin and α‐ and β‐tubulin. The amount of TRiC and GimC associated with actin or tubulin was independent of the presence of Ssb1/2p (Figure 2A, middle panel), consistent with the results of the polysome gradient analysis above (Figure 1C). However, deletion of the GIM genes resulted in an ∼60% reduction in actin binding to TRiC (Figure 2A, right panel), demonstrating the proposed role of GimC in actin/tubulin delivery to the chaperonin (Hansen et al., 1999). Thus, although both GimC and Ssb1/2p interact with actin and tubulin, only GimC is required for the recruitment of TRiC to these substrates. Ssa‐type Hsp70 proteins were not found in association with actin or tubulin, independent of the presence of Ssb1/2p or GimC (data not shown).

Figure 2.

GimC, TRiC and the Ssb‐type Hsp70s interact with actin and tubulin, but Ssb1/2p is not required for actin folding. (A) GimC, TRiC and the Ssb‐type Hsp70s interact with actin and α‐tubulin. ATP‐depleted cell lysates were subjected to immunoprecipitations with polyclonal actin‐ and α‐tubulin‐specific antibodies. Western blot analyses of the total cell extracts (TE) and immunoprecipitates (IP) were carried out with antibodies against actin, α‐tubulin, Ssb, GimC and TRiC as in Figure 1D. (B) Ssb protein is effectively depleted in conditional ssb mutants upon incubation under non‐permissive conditions. A conditional ssb mutant strain was generated by deleting the chromosomal SSB1 and SSB2 genes and reintroducing a GALS‐promoter‐controlled variant of SSB1 with a TEV‐protease cleavage site between the ATPase and the peptide‐binding domain of Ssb. Cell extracts of wild‐type (lane 1), Δssb1 Δssb2 (lane 2) and conditional ssb mutant cells (lanes 3–8) were prepared under the conditions indicated and analysed by western blotting with anti‐Ssb antibodies. In comparison with growth on galactose (lanes 3–5), incubation on glucose medium for 12 h drastically reduces the cellular levels of Ssb1p in the conditional mutant strain (lane 6). Upon induction of expression of the TEV‐protease by removal of tetracycline, the Ssb protein is cleaved (lanes 4 and 5). In extracts from cells grown with glucose as carbon source, no Ssb protein is detectable after 0.5 h induction of TEV‐protease (lanes 7 and 8). Note that the Ssb‐specific antibodies are directed against the C‐terminus of the protein. (C) Loss of Ssb function does not affect actin folding. Actin folding in pulse–chase‐labelled wild‐type, Δssb1 Δssb2, Δgim2 Δgim5 and gimC/ssbcond mutant cells under non‐permissive conditions was assayed as described previously (Siegers et al., 1999). (D) Phalloidin staining of the actin cytoskeleton. Cells from the yeast strains indicated were analysed by fluorescence microscopy. Bar represents 5 μm. Exponentially growing cells were shifted to non‐permissive conditions (glucose, absence of tetracycline for gimC/ssbcond and 23°C for all other strains examined) and incubated for 6 h prior to microscopic analysis.

It remained possible that in the absence of Ssb1/2p, GimC and TRiC interact with their substrates in a manner non‐productive for subsequent folding. We therefore measured the efficiency of actin folding in the respective chaperone mutant backgrounds. Cells were pulse‐labelled with [35S]methionine/cysteine, followed by a chase with unlabelled amino acids. At the different time‐points cells were lysed and folded actin precipitated with immobilized DNase I, which specifically binds actin in its native state. Deletion of SSB1/2 had no effect on the efficiency or kinetics of actin folding (Figure 2C), whereas loss of GimC function upon deletion of GIM1‐6 or GIM2/GIM5 resulted in a substantial reduction in folding yield and kinetics (Figure 2C and data not shown) (Siegers et al., 1999). GIM2 and GIM5 encode the two central subunits of GimC that are required for the assembly of the functional complex (Siegert et al., 2000; and unpublished observations). Visualization of the actin cytoskeleton by rhodamin–phalloidin staining demonstrated that wild‐type and Δssb1 Δssb2 cells have normal actin distribution with cortical actin patches merely in the bud, and actin cables reaching from the bud into the cell body of the mother cell. In contrast, the Δgim2 Δgim5 mutant has dislocalized actin patches and no cables are visible (Figure 2D).

To address the possibility that Ssb1/2p can functionally replace GimC in actin folding, a strain was generated in which Ssb function can be removed in the background of the GIM2 and GIM5 deletion. We first constructed a Δssb1 Δssb2 strain expressing a modified SSB1 gene under control of the GALS promoter, producing an Ssb1 protein with a TEV‐protease cleavage site in the flexible hinge region between the ATPase and peptide‐binding domains (Figure 2B). Incubation on glucose medium for 12 h greatly reduced the level of Ssb1p (Figure 2B, lane 6), and induction of TEV‐protease from a tetracycline controllable promoter resulted in virtually complete removal of the residual Ssb1 protein (Figure 2B, lanes 7 and 8). In a second step, GIM2 and GIM5 were deleted in the conditional Δssb1/2 strain background. As expected, incubation of the resulting strain, termed gimC/ssbcond, under non‐permissive conditions resulted in loss of viability after 8–12 h (approximately two further cell divisions). However, removal of Ssb in the GimC mutant background did not further reduce the efficiency of actin folding (Figure 2C) nor did it aggravate the impairment of the actin cytoskeleton (Figure 2D). Likewise, loss of Ssb function did not increase the sensitivity of the GimC mutant strain towards the microtubule destabilizing drug benomyl (data not shown), suggesting that tubulin folding is also independent of the Ssb chaperones. Elevated levels of the Ssa‐type Hsp70 or the cytosolic Hsp40 protein Ydj1p did not rescue the gimC/ssbcond strain under non‐permissive conditions, and neither overexpression of SSB, SSA nor YDJ1 alleviated the phenotype of the Δgim2 Δgim5 mutant strain (data not shown).

Based on these results, the folding of actin and tubulins is independent of Ssb1/2p but requires the specific cooperation of TRiC with GimC/prefoldin. Why then does the loss of Ssb1/2p function in TRiC or GimC mutants result in loss of cell viability? In addressing this question we tested the hypothesis that Ssb1/2p may cooperate with TRiC or TRiC/GimC in the folding of proteins other than actin or tubulins.

Ssb1/2p and TRiC interact with a group of WD40 repeat proteins

Microscopic analysis of gimC/ssbcond cells under non‐permissive conditions revealed a pronounced growth defect with abnormal budding morphology, elongated buds, and failure of cytokinesis and cell separation (Figure 3A). In contrast, cells deleted either only for the two SSB genes or GIM2 and GIM5 did not exhibit this phenotype. However, a similar morphology as in the gimC/ssbcond cells was seen in the TRiC mutant tcp1‐3 at the non‐permissive temperature (Figure 3A). Essentially the same phenotype has been reported previously for several yeast mutants defective in cell division cycle control genes, including CDC55 (Healy et al., 1991; Figure 3A). Interestingly, an interaction of Cdc55p with TRiC‐subunits had been noted in a recent proteomic analysis of protein complexes in yeast (Ho et al., 2002). Cdc55p is a regulatory subunit of a PP2A protein phosphatase, which functions in the cell cycle checkpoint that inhibits exit from mitosis in response to spindle or kinetochore damage. Defects in this checkpoint result in continued cell division and rapid loss of viability when microtubules are destabilized, for example by treatment with the microtubule depolymerizing drug nocodazole (Minshull et al., 1996; Yang et al., 2000).

Figure 3.

Cdc55p interacts with Ssb1/2p and TRiC. (A) Cells from the yeast strains indicated were grown as described in Figure 2D and analysed by phase contrast microscopy. Bar represents 5 μm. (B) Cdc55p specifically interacts with the Ssb‐type Hsp70s and TRiC, but not with GimC and the Ssa‐type Hsp70s. Cell lysates from cells expressing C‐terminally epitope‐tagged Cdc55p (CDC55–3HA, CDC55–3myc) and from wild type were subjected to immunoprecipitation with anti‐HA or anti‐c‐myc antibodies. Western blot analyses of the total cell extracts (TE) and immunoprecipitates (IP) were performed with antibodies against the HA or c‐myc epitope, TRiC, Ssb, GimC and Ssa as in Figure 1D.

In order to test whether Cdc55p is in fact a TRiC substrate, we constructed haploid yeast strains expressing C‐terminally haemagglutinin (HA)‐ or c‐myc‐tagged Cdc55p. The epitope‐tagged Cdc55p variants supported wild‐type growth rates (data not shown) and the respective strains did not exhibit any of the morphological characteristics described for mutants defective in Cdc55p (Figure 3A). Co‐immunoprecipitations with anti‐HA or anti‐c‐myc antibodies from ATP‐depleted lysates revealed that a small fraction of total cellular Cdc55p (up to 3%) interacts specifically with TRiC and the Ssb1/2 proteins, consistent with the possibility that only newly synthesized, not yet folded molecules are chaperone bound. In contrast, GimC and the other cytosolic Hsp70 chaperones encoded by the SSA genes did not detectably associate with Cdc55p (Figure 3B). HA‐tagged Erg5p, used as a control protein, interacted with neither Ssb1/2p, GimC nor TRiC (Figure 4B). Remarkably, as the co‐immunoprecipitation experiments were carried out with C‐terminally tagged Cdc55p, the Ssb proteins maintain an interaction with full‐length Cdc55p after its release from the ribosome. Re‐addition of ATP to the lysates prior to immunoprecipitation destabilized the interaction of Cdc55p with Ssb1/2p and TRiC (data not shown). Together these results suggest that Cdc55p is a TRiC substrate that uses Ssb1/2p for efficient transfer from the ribosome to the chaperonin. In addition to binding nascent chains on ribosomes, the Ssb chaperones may cooperate with TRiC in the post‐translational folding or assembly of certain proteins.

Figure 4.

Ssb1/2p and TRiC interact with a group of WD40 repeat proteins. (A) Members of the WD40 repeat protein family were tested for their interaction with chaperones. ATP‐depleted lysates from strains STE4‐3HA, PEX7‐3HA, PRP46–3HA, SEC27–3HA, CDC20–3HA and CDC4‐3HA, expressing C‐terminally epitope‐tagged versions of the respective proteins, were subjected to immunoprecipitations with anti‐HA antibodies. Lysates from cells lacking the HA‐epitope were used as a control. Total extracts (TE) and immunoprecipitates (α‐HA‐IP) were analysed by western blotting for co‐immunoprecipitation of Ssb1/2p, GimC and TRiC as in Figure 1D. (B) TRiC and the Ssb‐type Hsp70 proteins interact with the Cdc55p partner protein Sit4p, but not with Erg5p. Lysates from yeast cells expressing epitope‐tagged Sit4p and Erg5p were subjected to immunoprecipitation with anti‐HA antibodies. Western blots of the total cell extracts (TE) and immuno precipitates (IP) were analysed as described above.

Figure 5.

Transient interaction of newly synthesized WD40 proteins with TRiC. (A) Transit of newly synthesized Cdc55p through TRiC and accumulation of non‐chaperone‐bound, newly synthesized Cdc55p. Yeast spheroplasts from cells expressing HA‐tagged Cdc55p were labelled with 100 μCi/ml [35S]methionine/cysteine for 110 s at 30°C. Translation and incorporation of radioactivity into nascent chains was stopped by the addition of cyclo heximide followed by a chase during which samples were withdrawn at the time‐points indicated. Cells from individual time‐points were lysed rapidly in the presence of EDTA to stop the ATP‐dependent activity of the chaperonin, and TRiC was immunoprecipitated as described previously (Siegers et al., 1999). Bound proteins were eluted from the antibody–protein‐G–Sepharose beads by incubation in the presence of SDS. The eluates were diluted 20‐fold in lysis buffer and subjected to a second immunoprecipitation with anti‐HA antibodies. The TRiC‐depleted supernatants from the first immunoprecipitation were also subjected to a second immunoprecipitation with anti‐HA antibodies to follow the accumulation of free radiolabelled Cdc55–3HAp. The anti‐HA‐precipitates were separated by SDS–PAGE and analysed by autoradiography and western blotting. Time courses of Cdc55–3HAp transit through TRiC and the accumulation of free protein are also shown. (B) Transit of newly synthesized Pex7p through TRiC and accumulation of non‐chaperone‐bound, newly synthesized Pex7p was analysed as described for Cdc55p in (A).

Taking into consideration that the cellular abundance of Cdc55p is very low (see Figures 3B and 6B), it seemed possible that the observed interaction between Ssb1/2p and TRiC (Figure 1D) could be mediated by a larger group of substrate proteins in transfer from Ssb1/2p to the chaperonin. Cdc55p belongs to the family of WD40 repeat proteins. These proteins contain four or more copies of the WD‐repeat, a weakly conserved sequence motif of ∼31 amino acids ending with conserved tryptophan and aspartate residues (Smith et al., 1999). The common fold of these proteins was defined by the structure of β‐transducin (Ste4p in yeast) as a β‐propeller (Sondek et al., 1996). In β‐transducin the propeller has seven blades, each consisting of a WD40 repeat folded into a small antiparallel β‐sheet.

Figure 6.

Chaperone dependence of Cdc55p biogenesis. (A) Cdc55p activity was assayed by measuring the amount of tyrosine‐phosphorylated Cdc28p in lysates of wild‐type cells and chaperone mutants. All strains analysed carried HA‐tagged Cdc28p, which was immunoprecipitated with anti‐HA antibodies. The immunoprecipitates were analysed by western blotting with rabbit polyclonal anti‐phosphotyrosine antibodies (Zymed Laboratories) before stripping the blots and reprobing them with anti‐HA antibodies as a control for equal loading. (B) GimC interacts with Cdc55p in the absence of Ssb1/2p and levels of Cdc55p are strongly reduced in cells lacking Ssb and GimC function. Cdc55–3HA was immunoprecipitated from lysates of wild‐type, Δssb1/2 and gimC/ssbcond cells after incubation for 6 h at non‐ permissive growth conditions. Western blots of the total cell extracts (TE) and immunoprecipitates (IP) were analysed with antibodies against the HA‐epitope, GimC and Ssa. Note that at this exposure Cdc55p in TE samples is not visible.

In addition to Cdc55p, we tested a number of otherwise unrelated WD40 repeat proteins (Ste4p, Pex7p, Prp46p, Sec27p, Cdc20p and Cdc4p) for their possible interaction with Ssb1/2p and TRiC. In these proteins of ∼40–100 kDa, the WD40 propeller is combined with other domains and functions as a protein‐interaction module. ATP‐depleted lysates from yeast strains expressing C‐terminally HA‐tagged versions of the various proteins under their endogenous promoters were subjected to immunoprecipitation with anti‐HA antibodies, and the precipitates analysed for co‐precipitation of chaperones. Apart from Cdc4p, for which we could only detect an interaction with Ssb1/2p, all the other WD40 repeat proteins were found to interact with both Ssb1/2p and TRiC (Figure 4A). However, none of them showed a detectable interaction with GimC, in contrast to actin and tubulin (Figure 2A). About 1–5% of the total cellular amounts of the WD40 proteins was recovered in association with TRiC. Additionally, TRiC was also observed to interact with Sit4p, the catalytic subunit of a PP2A phosphatase (Figure 4B). Sit4p and Cdc55p must assemble into a complex with Tpd3p in order to form a functional enzyme (Nickels and Broach, 1996). In contrast, Erg5p, a known substrate of the Hsp90 chaperone system (Goasduff and Cederbaum, 2000), interacted with neither Ssb1/2p nor TRiC (Figure 4B).

Pulse–chase labelling experiments with yeast strains expressing HA‐tagged Cdc55p and Pex7p were performed to investigate the kinetics of the interaction of TRiC with newly synthesized WD40 proteins. Spheroplasts were labelled with [35S]methionine/cysteine and chased in the presence of cycloheximide (CHX) to block further protein synthesis. At different time points cells were lysed rapidly in the presence of EDTA to inhibit the TRiC ATPase and TRiC was immunoprecipitated (Figure 5A and B). To detect the low‐abundance WD40 substrates bound to TRiC, the immunoprecipitates were subjected to a second immunoprecipitation with anti‐HA antibodies, followed by autoradiography or sensitive western blotting with anti‐HA antibodies.

Both Cdc55p and Pex7p efficiently bound to TRiC immediately upon synthesis. During the chase, the newly synthesized proteins dissociated rapidly from TRiC with an apparent half‐time of 1–2 min (Figure 5A and B, left panel), very similar to the rate of TRiC‐mediated actin folding (compare with Figure 2C). Analysis of the supernatants of the TRiC precipitations revealed the inverse kinetics for the production of non‐TRiC‐bound, newly synthesized Cdc55p and Pex7p (Figure 5A and B, right panel). Remarkably, almost no free radiolabelled Cdc55p or Pex7p was detectable in TRiC‐depleted cell lysates from early time‐points of the experiment (Figure 5A and B, right panel), indicating that the WD40 proteins interact essentially quantitatively with the chaperonin upon de novo synthesis. The rapid dissociation from TRiC indicates that the WD40 substrates have folded into a structure no longer recognized by the chaperonin. It is possible that in this state Cdc55p and Pex7p have already assembled with their respective partner proteins, and an additional role of TRiC in mediating assembly cannot be ruled out.

TRiC cooperates with Ssb1/2p or GimC in the production of functional Cdc55p

In order to establish the chaperone requirement for Cdc55p folding/assembly directly, the functional activity of Cdc55p was determined in chaperone mutant strains. Cdc55p activity was assayed based on the observation that cdc55 mutants with perturbed phosphatase activity accumulate tyrosine‐phosphorylated Cdc28p detectable with anti‐phosphotyrosine antibodies (Minshull et al., 1996). As expected, immunoprecipitations revealed that an increased amount of Cdc28p was phosphorylated in Δcdc55 cells compared with wild type (Figure 6A). Consistent with our finding that Cdc55p is a substrate of TRiC, the extent of Cdc28p‐phosphorylation was also increased in the tcp1‐3 mutant strain already at the permissive temperature. In contrast, other point mutations in TRiC subunits did not detectably affect the function of Cdc55p under permissive conditions (Figure 6A). Loss of either Ssb or GimC function alone did not result in a significant increase of Cdc28p phosphorylation (data not shown). This is consistent with the observation that, in contrast to the tcp1‐3 mutant, ssb and gim mutants do not show the morphological changes that characterize the cdc55 deletion strain (Figure 3A). Interestingly, however, when the mutations in TRiC were combined with a deletion of the two SSB genes, tyrosine‐phosphorylated Cdc28p was readily detectable in all TRiC mutants under permissive conditions. These results are consistent with the possibility that TRiC cooperates with the Ssb‐type Hsp70s in Cdc55p folding.

Normal Cdc55p levels were maintained in the SSB1/2 deletion strain (Figure 6B) and the association of Cdc55p with TRiC was not measurably reduced in the absence of Ssb1/2p (data not shown), suggesting that GimC may replace Ssb1/2p in TRiC‐mediated Cdc55p folding. In contrast to wild‐type cells, an association of Cdc55p with GimC was indeed detectable in the Δssb1/2 strain, whereas an interaction with the Ssa‐type Hsp70 proteins was not observed (Figure 6B). Notably, the total cellular amount of Cdc55p decreased by >50% within 6 h of Ssb1/2p depletion in the gimC/ssbcond strain (Figure 6B). This effect is most likely due to misfolding of newly synthesized Cdc55p and turnover of pre‐existent, folded protein. Thus, GimC can replace Ssb1/2p in the folding of Cdc55p, and presumably other TRiC‐dependent WD40 substrates, and this functional redundancy plausibly explains why yeast can tolerate the individual loss of Ssb1/2p or GimC, but not the loss of both functions.

Discussion

Our experiments provide functional evidence that a group of WD40 domain proteins constitutes a structurally defined class of substrates of the cytosolic chaperonin TRiC/CCT. In the folding of these proteins, TRiC cooperates with the Hsp70 chaperones of the Ssb‐type rather than with GimC/prefoldin, the co‐chaperone required for efficient folding of actin and tubulins (Figure 7). However, GimC can substitute for Ssb1/2p in the folding of Cdc55p, and presumably other TRiC‐dependent WD40 substrates, whereas the Ssb proteins are unable to replace GimC in actin and tubulin folding. Thus, distinct classes of TRiC substrates use different, but partially overlapping chaperone pathways for productive delivery to the chaperonin, with possible additional roles for Ssb1/2p and GimC in TRiC‐assisted folding or assembly.

Figure 7.

Model for the chaperone pathways of different classes of TRiC substrates. The chaperone pathway for the folding of actin and tubulins (left) differs from that of WD40 domain proteins such as Cdc55p (right) in that GimC/prefoldin is required only for actin/tubulin folding and does not interact with the WD40 substrates. The latter rely on the cooperation of Ssb1/2p with TRiC for efficient folding. Ssb1/2p is likely to initiate its interaction early in translation but appears to cooperate closely with TRiC in post‐translational completion of folding and/or assembly. Whereas GimC can replace Ssb1/2p in WD40 protein folding, the Ssb chaperones are unable to substitute for GimC in actin/tubulin folding.

WD40 proteins as a structurally defined class of TRiC substrates

In recent years, the number of different proteins identified as substrates of TRiC/CCT has increased steadily, modifying the original view of the eukaryotic chaperonin as a specialized folding machine devoted solely to actin and tubulins. Based on our analysis, in the yeast S.cerevisiae these abundant cytoskeletal proteins occupy ∼50–60% of the total capacity of TRiC, leaving the possibility that numerous additional, newly synthesized proteins of lower abundance could also interact with the chaperonin. We now show that these proteins include a group of WD40 domain proteins, thus functionally validating data from a recent interaction proteome analysis (Ho et al., 2002).

In contrast to actin and tubulins, which are structurally unrelated, the WD40 proteins constitute the first class of chaperonin substrates that are defined by a common fold, the WD40 β‐propeller (Smith et al., 1999). Why many proteins containing this domain are TRiC‐dependent remains to be determined, but we note that the WD40 propeller represents a complex folding unit with a high degree of structural cooperativity. These proteins may be particularly aggregation‐prone, due to their high β‐sheet content. Interestingly, WD40 domain proteins are found in all eukaryotes but are largely absent in bacteria and archaea (Neer et al., 1994). Saccharomyces cerevisiae is thought to contain >90 of these proteins (>1% of the genome) (Valpuesta et al., 2002). The abundant appearance of WD40 proteins in eukaryotes goes along with a dramatic increase in subunit complexity of TRiC (eight different subunits) that occurred during its evolution from simpler archaeal ancestors with only one to three different subunits. This increase in subunit complexity may reflect an adaptation of the chaperonin to the folding of distinct groups of substrates as structurally diverse as WD40 proteins and actin/tubulins.

Our observation that Cdc4p does not detectably interact with TRiC suggests that only a subset of WD40 domain proteins are TRiC dependent. WD40 proteins are typically part of larger complexes and it is possible that the chaperonin supports these assembly reactions beyond folding per se, as shown for other non‐WD40 substrates (Feldman et al., 1999; Guenther et al., 2002). For example, TRiC also interacts with Sit4p, the catalytic subunit of a PPA2 phosphatase, which must assemble with the WD40 protein Cdc55p and with Tpd3p to form a functional enzyme (Nickels and Broach, 1996). Similarly, the WD40 protein Cdc20p is a component of the anaphase‐promoting complex, and the Ste4p homologue in higher eukaryotes, β‐transducin, interacts with G‐α‐transducin, a protein previously identified as a TRiC substrate (Farr et al., 1997).

Chaperone pathways of TRiC‐assisted folding

Another central finding of this study concerns the differential cooperation of TRiC with the Hsp70 chaperones, Ssb1/2p and with GimC/prefoldin in the folding of WD40 domain proteins and actin/tubulin, respectively (Figure 7). Functional evidence for a role of Ssb1/2p in folding had been missing so far. Considering that the Ssb proteins are highly abundant ribosome‐associated chaperones (Pfund et al., 1998), it would appear that the interaction with Ssb1/2p provides the default pathway for delivery of polypeptides to TRiC in a conformational state competent for folding. TRiC may contact newly synthesized polypeptides either co‐translationally (Frydman et al., 1994; McCallum et al., 2000; and this study) or after their release from the ribosome. Significantly, our observation that the Ssb1/2 proteins interact with the full‐length forms of WD40 substrates suggests that beyond stabilizing nascent chains on ribosomes, these chaperones may also function in certain post‐translational folding or assembly reactions.

In contrast to Ssb1/2p, GimC appears to have a special role in the delivery of newly synthesized actin and tubulin to TRiC, and this function cannot be performed by either the Ssb‐ or Ssa‐type Hsp70s. Direct binding of GimC/prefoldin to TRiC (Vainberg et al., 1998; Martin‐Benito et al., 2002) may facilitate the delivery of actin and tubulins in a defined orientation relative to the subunit topology of the chaperonin ring (Llorca et al., 1999).

Ssb1/2p can be replaced by GimC in the folding of WD40 proteins such as Cdc55p, and simultaneous loss of Ssb and GimC function results in a severe synthetic growth defect with eventual loss of viability. Because it is unlikely that the lesser abundant GimC can take over the general function of Ssb1/2p as a nascent chain‐binding chaperone, it is plausible that this synthetic growth defect is caused by a combined deficiency in the folding of several TRiC substrates. It may be important in this regard that a defect in Cdc55p results in more rapid loss of cellular viability when microtubules are destabilized (Minshull et al., 1996; Yang et al., 2000), as is the case in GimC and TRiC mutant strains (Geissler et al., 1998). Notably, the abundant Ssa‐type Hsp70s cannot substitute for the TRiC‐related function of Ssb1/2p and are predominantly involved in post‐translational processes such as protein transport. A specific association of Ssa proteins with nascent polypeptide chains has not yet been observed, raising the question as to which chaperone component(s) may act in protecting nascent chains in the absence of Ssb1/2p. In mammalian cells, the constitutively expressed Hsc70 binds to nascent chains and may functionally cooperate with TRiC (Feldman et al., 1999).

Our finding of a partial functional overlap between Ssb1/2p and GimC in the yeast cytosol is reminiscent of the reported redundancy of Trigger factor and DnaK functions in protein folding in the bacterial cytosol (Deuerling et al., 1999; Teter et al., 1999). In comparing these two situations, Ssb1/2p would be functionally analogous to Trigger factor, as both components interact directly with the ribosome and bind short nascent chains, whereas both DnaK and GimC act later in translation and have no affinity for the ribosome. Taking these considerations further, it seems plausible that in certain archaea that lack Hsp70 altogether (Ruepp et al., 2001), the simpler and more abundant archaeal version of GimC/prefoldin may function more generally in stabilizing nascent polypeptides.

Materials and methods

Yeast strains and genetic methods

Basic yeast genetic methods and media were as described previously (Guthrie and Fink, 1991). TCP1, SSB1/2, EGD1, EGD2, BTT1, SSZ1, ZUO1, CDC55 or GIM genes were deleted in strain YPH499 (Sikorski and Hieter, 1989) using PCR‐amplified kanMX4 (Wach et al., 1994), HIS3MX6 (Wach et al., 1997) or cre‐loxP (Delneri et al., 2000) cassettes. Chromosomally integrated gene fusions to the 3′ end of various genes with the HA or c‐myc epitope were generated by homologous recombination of PCR‐amplified cassettes in strain YPH499 (Knop et al., 1999).

Radiolabelling of yeast cells and actin folding

Pulse–chase radiolabelling of spheroplasts was performed as described previously (Siegers et al., 1999).

Growth analysis of mutant strains

Growth of wild‐type YPH499 cells and mutants was determined on appropriate selective media plates at 30°C. Genetic interactions of various chaperone mutants were determined by plasmid shuffle experiments. The mutant strains indicated were transformed with a URA3‐based plasmid carrying a wild‐type copy of the gene of interest and subsequently deleted for the chromosomal copy of the respective open reading frame. The resulting strain was transformed with a non‐URA3‐based plasmid encoding a wild‐type or mutant copy of the gene of interest or with an empty control plasmid and assayed for growth on plates containing 5′‐fluororotic acid (5′‐FOA) at 30°C. 5′‐FOA only allows growth of cells that have lost the URA3‐based plasmid encoding the wild‐type copy of the respective gene.

Phalloidin staining

Actin was visualized with rhodamin–phalloidin (Molecular Probes) as described previously (Adams and Pringle, 1991).

Immunoprecipitation and western blotting

Specific antibodies or pre‐immune sera were cross‐linked to protein A– or protein G–Sepharose 4 Fast Flow (Amersham), respectively, according to standard procedures. Cells were grown to mid‐log phase, harvested by centrifugation (3000 g, 5 min), washed twice in immunoprecipitation buffer [1 × PBS (10 mM Na2HPO4, 1.8 mM KH2PO4, 2.7 mM KCl, 140 mM NaCl) pH 7.4, 5 mM EDTA, protease inhibitors] and resuspended in 0.5 ml of the same buffer lacking EDTA but containing 5 U/ml apyrase or 5 mM MgATP. Cytosolic extracts were prepared by agitating the cells with glass beads for 6 × 30 s at 4°C. For phosphotyrosine detection, lysates were prepared in buffer containing 1 mM Na3VO4 and 50 mM NaF. Lysates were cleared by centrifugation (20 000 g, 4°C, 15 min). Supernatants were incubated in the presence of 1% Tween‐20 with antibodies coupled to protein A–/G–Sepharose beads for 2 h at 4°C on a rotating platform. Beads were washed extensively with lysis buffer containing 1% Tween‐20 prior to SDS–PAGE. For western blots, 10 μg of the total cell extracts (5 μg for western blots with anti‐Ssb or ‐Ssa antibodies) were analysed and 200 μg protein was used for each immunoprecipitation. Immunoprecipitations of GimC from cell lysates were performed as described previously (Geissler et al., 1998). Western blotting was performed using enhanced chemiluminescence (ECL; Amersham).

Polysome gradients

Fractionation of ribosomes from yeast lysates was carried out as described previously (Nelson et al., 1992), with the following alterations. All lysates were centrifuged through 5 ml 20–50% continuous sucrose density gradients. Each gradient was loaded with an equal volume of lysate (150 μl, adjusted to 20 OD260) and subjected to centrifugation in an SW55Ti rotor (Beckmann) at 54 000 r.p.m. for 75 min at 4°C.

Acknowledgements

We thank Drs E.A.Craig and E.Schiebel for the generous gift of antibodies and yeast strains, and Dr J.C.Young for critically reading the manuscript. This work was supported by the Deutsche Forschungsgemeinschaft.

References