The unique hetero-oligomeric nature of the subunits in the catalytic cooperativity of the yeast Cct chaperonin complex

Abstract

The structural and functional organization of the Cct complex was addressed by genetic analyses of subunit interactions and catalytic cooperativity among five of the eight different essential subunits, Cct1p–Cct8p, in the yeast Saccharomyces cerevisiae. The cct1–1, cct2–3, and cct3–1 alleles, containing mutations at the conserved putative ATP-binding motif, GDGTT, are cold-sensitive, whereas single and multiple replacements of the corresponding motif in Cct6p are well tolerated by the cell. We demonstrated herein that cct6–3 (L19S), but not the parolog cct1–5 (R26I), specifically suppresses the cct1–1, cct2–3, and cct3–1 alleles, and that this suppression can be modulated by mutations in a putative phosphorylation motif, RXS, and the putative ATP-binding pocket of Cct6p. Our results suggest that the Cct ring is comprised of a single hetero-oligomer containing eight subunits of differential functional hierarchy, in which catalytic cooperativity of ATP-binding/hydrolysis takes place in a sequential manner different from the concerted cooperativity proposed for GroEL.

The well-studied chaperonin, GroEL, in Escherichia coli is a double toroid multimeric ATPase protein complex with seven identical subunits in each ring. In contrast, eight different but closely related subunits, Cct1p–Cct8p, have been identified in the eukaryotic chaperonin complex, Cct. Two plausible models exist for the structural composition of the Cct complex. The simplest model is that only one single functional species of a Cct complex exists whose rings are comprised of a stoichiometrical array of all eight different subunits. Evidence for this include biochemical copurification of Cct1p–Cct8p in an equimolar amount (1–4) and genetic interactions among the Cct1p–Cct4p subunits of the yeast Saccharomyces cerevisiae Cct complex (5, 6). Alternatively, the Cct complex can be a composite of numerous essential species as a result of the stacking of two different homo-oligomeric rings in various combinations. This is suggested by some unpublished results on the isolation of homo-oligomeric Cct complexes (7, 8), and on the in vitro assembly of the β subunits of the dimeric thermosome into homo-oligomeric rings as reported by Waldmann et al. (9). Defining the composition of the Cct rings is of primary importance toward the understanding of the evolutionary significance and functional requirements of the eight different subunits.

We have undertaken an analysis of residues sensitive to replacements in the yeast Cct6p chaperonin, including those that, by analogy to GroEL, are implicated in peptide binding, allosteric transition, and inter-ring/intra-ring subunit-to-subunit interactions (10). The study revealed that replacement of the absolutely conserved GDGTT, putative ATP-binding motif, with a stretch of alanine residues in the putative equatorial domain of Cct6p caused little functional impairment, and suggested the presence of a functional hierarchy among the Cct subunits for the binding/hydrolysis of ATP (10). In this study, we established the interaction of the GDGTT motif with three other putative ATP-binding motifs of Cct6p by a double-mutant analysis. In addition, catalytic interactions of putative ATP-binding pockets in various Cct subunits were demonstrated through suppressor analyses of cct1–1, cct2–3, and cct3–1 cold-sensitive mutants by cct6 mutants, and the modulation of the suppression by additional replacements in Cct6p. These results suggest that the Cct complex is comprised of a single hetero-octamer containing all subunits, and that catalytic cooperativity of ATP binding/hydrolysis takes place in a manner different from the concerted cooperativity proposed for GroEL.

MATERIALS AND METHODS

Genetic Nomenclature and General Methods.

CCT6⁺ denotes the wild-type allele encoding the wild-type Cct6 protein (or Cct6p). Mutant alleles are designated cct6 followed by the allele number, e.g., cct6–1, cct6–10, etc.

General methods used in the construction of plasmids, restriction enzyme digests, separation of plasmid DNA and restriction fragments on agarose gels, ligation of DNA fragments, the isolation of plasmid DNA (11), transformation of E. coli (12), and conditions for the PCR (13) have been described. The cultivation, manipulation (14), and transformation (15) of yeast strains followed standard procedures.

Construction and Testing cct6 Mutants.

The cct6 mutants were constructed and tested as described by Lin et al. (10). The plasmid pAB1058 [CCT6⁺ LEU2⁺ CEN6 ARSH4] was used for oligonucleotide-directed mutagenesis (10), using the method described by Kunkel et al. (16). Mutations were confirmed by corresponding changes in restriction sites or by direct DNA sequencing (17). The plasmid pAB990 [CCT6⁺ URA3⁺ CEN4 ARS1] was used in the plasmid shuffle experiment (10), using the method described by Boeke et al. (18). The yeast strain B-9017 (cct6-Δ1::TRP1 MATa ura3–52 trp1-Δ63 leu2-Δ1 GAL2⁺ pAB990) was transformed either with pAB1058 [CCT6⁺ LEU2⁺ CEN6 ARSH4] to yield the normal control strain, B-9018 (cct6-Δ1::TRP1 MATa ura3–52 trp1-Δ63 leu2-Δ1 GAL2⁺ pAB990 pAB1058), or with mutant plasmids created by oligonucleotide-directed mutagenesis.

The functionality of each mutant cct6 allele was tested by the plasmid shuffling technique (18). Standard yeast extract/peptone/dextrose (YPD) and synthetic media for the growth of yeast were as described by Sherman (14). FOA medium containing 1.0 mg/ml of 5-fluoroorotic acid (5-FOA) and 0.05 mg/ml of uracil was used to select against a functional URA3⁺ gene (18). Independent transformants were plated and restreaked on synthetic minimal medium. To determine the essentiality of each allele, cells were spotted onto FOA plates at 30°C to evict the URA3⁺ plasmid, pAB990, which contains the extrachromosomal wild-type CCT6⁺ allele. Nonviable alleles can be easily identified by the absence of growth. Strains containing viable alleles were picked and streaked on YPD plates. These strains were further spotted onto YPD, TBZ (thiabendazole), and NaCl media and incubated at 37°C, 30°C, or 15°C. TBZ medium, containing YPD plus 20 μg/ml of a microtubule inhibitory drug, thiabendazole, was used to detect alleles conferring microtubule biosynthetic defects (19). NaCl medium, containing YPD plus 0.9 M NaCl, was used to detect alleles affecting actin biosynthesis that are hypersensitive to hypertonic conditions (20), or to detect osmotic remedial alleles (10).

Suppression or complementation analyses were carried out with the following conditional lethal mutants, using the multi-copy or single-copy plasmids listed in Table 1: cct1–1, cct1–2, cct1–3, cct2–1, cct2–2, cct2–3, cct2–4, cct3–1, and cct4–1 (see Fig. 2 for amino acid residue alterations). The sources of cct alleles are as follows: cct1–1 (tcp1–1) [MATα ura3–52 trp1–7 leu2–3,112 cct1::LEU2 (cct1–1 TRP1 YCpM538)]; cct1–2 (tcp1–2) [MATα ura3–52 trp1–7 leu2–3,112 cct1::LEU2 (cct1–2 TRP1 YCpM538])]; cct1–3 (tcp1–3) [MATα ura3–52 trp1–7 leu2–3,112 cct1::LEU2 (cct1–3 TRP1 YCpM538)] (21). cct2–1 (bin3–1) (MATa ura3–52 his4–539 lys2–801 cct2–1); cct2–2 (bin3–2) (MATa ura3–52 his4–539 leu2–3, 112 cct2–2); cct2–3 (bin3–3) (MATa ura3–52 cct2–3); cct2–4 (bin3–4) (MATa ura3–52 leu2–3, 112 cct2–4); cct3–1 (bin2–1) (MATa ura3–52 his4–539 cct3–1) (5); cct4–1 (anc2–1) (MATa ura3–52 leu2–3, 112 lys2–801 cct4–1) (6).

Table 1.

Complementation analysis of the various conditional lethal cct1–cct4 alleles

Suppressor				cct alleles tested for suppression*
Allele†	Alteration	Phenotype‡	Plasmid	cct 1-1	cct 2-3	cct 3-1	cct1-2, cct1-3, cct2-1, cct2-2, cct2-4, cct4-1
CCT6+	None	A	§	0	0	0	0
cct6-Δ	(Deficient)	D	§	0	0	0	0
cct6-3	L19S	A	§	2	1	2	0
			¶	1	2	1	0
cct6-5	G38S	A	§	0	ND	ND	ND
cct6-101	L19S G38S	A	§	0	ND	ND	ND
cct6-14	G59R	B	§	0	ND	ND	ND
cct6-102	L19S G59R	B	§	0	ND	ND	ND
cct6-19	G90E	B	§	0	ND	ND	ND
cct6-103	L19S G90E	C	§	0	ND	ND	ND
cct6-32	R155K	A	§	0	ND	ND	ND
cct6-104	L19S R155K	A	§	2	2	2	0
cct6-33	S157A	A	§	0	ND	ND	ND
cct6-105	L19S S157A	A	§	4	2	4	0
			¶	2	3	3	0
cct6-34	S157E	B	§	0	ND	ND	ND
cct6-106	L19S S157E	B	§	1	0	1	0
cct6-94	G414E	C	§	0	ND	ND	ND
cct6-107	L19S G414E	D	§	0	ND	ND	ND
cct1-5	R26I	A	§	0	0	0	0

The cct6-1–cct6-100 mutants are from Lin et al. (10), whereas the cct6-101–cct6-112 mutants are from this study. ND, not determined. 

^*See Materials and Methods for the sources of various cct alleles and refer to Fig. 1 for amino acid residue alterations; 0-4 denotes the relative growth of the strains, where 0 denotes no growth, and 4 normal or near growth. 

^†The following other cct6 alleles did not suppress cct1-1: cct6-5 (G38S), cct6-6 (G38S G59R), cct6-7 (G38S G90E), cct6-8 (G38S G414E), cct6-10 (G38S G59R G414E), cct6-14 (G59R), cct6-15 (G59R G90E), cct6-16 (G59R G414E), cct6-19 (G90E), cct6-33 (S157A), cct6-39 (A186S), cct6-40 (A186S D187R), cct6-41 (A186S C376S), cct6-42 (A186S T377S), cct6-43 (A816S D187R C376S), cct6-44 (A186S D187R T377S), cct6-45 (A186S D187R C376S T377S), cct6-46 (A186S C376S T377S), cct6-47 (D187R), cct6-48 (D187R C376S), cct6-49 (D187R T377S), cct6-50 (D190G), cct6-57 (H221G), cct6-79 (K311V), cct6-82 (E326S), cct6-86 (C376S), cct6-87 (C376S T377S), cct6-88 (T377S), cct6-92 (D408L), and cct6-98 (G438L). 

^‡Property of each allele during vegetative growth is based on the following criteria: A, normal to near normal; B, minor defect; C, major defect, or conditional lethal; D, lethal (10). 

^§The multi-copy plasmids used in this study contain various cct1 or cct6 alleles in YEp32 [URA3 2 μ]. 

^¶The single-copy plasmid used in this study contain various cct1 or cct6 alleles in PRS306 [URA3 CEN6 ARSH4]. 

Figure 2.

The conserved putative ATP-binding/hydrolysis motifs from various chaperonins. The L19S (cct6–3) mutation of Cct6p suppresses, but does not bypass, the conditional lethal alleles, cct1–1, cct2–3, and cct3–1, all of which have mutations within the putative ATP-binding site, GDGTTT/S; however, L19S does not suppress cct1–2 and cct1–3, which have replacements in other conserved ATP-binding/hydrolysis motifs. Double mutations involving any two motifs in Cct6p result in synthetic lethality, whereas leucine-19 interacts with two but not others of these motifs, suggesting that these four motifs possess both common and distinct functions, to account for the specificity of suppression by L19S. The changes of cct2–1, cct2–2, cct2–3, cct2–4, and cct3–1 mutations were determined by direct DNA sequencing of the PCR products of the genomic DNA using the methods of Sanger et al. (17).

RESULTS

We have addressed the question concerning the structure–function relationships of the Cct complex by a suppression analysis of the cct1–1 cold-sensitive mutant allele from S. cerevisiae that contains a D96E mutation in the putative ATP-binding motif, GDGTTT/S, of Cct1p (ref. 21; Fig. 2). This mutant allele confers both tubulin and actin biosynthetic defect phenotypes at the restricted temperature (21).

Although no biochemical data for this mutant is available, mutation of the corresponding conserved aspartic acid residue (D87) of GroEL (Fig. 2) impairs ATP binding, substrate binding, and ATPase activity (22, 23). The D87K and D87N mutations of GroEL abolished ATPase activity almost completely and affected ATP affinity by one-third, as shown by Fenton et al. (23). Furthermore, Weiss and Goloubinoff (22) suggested a loss of cooperativity of ATP binding/hydrolysis in the D87E mutant of GroEL, although its initial rates of ATP binding were not reported.

The cooperativity of ATP hydrolysis by GroEL was first demonstrated by Gray and Fersht (24). It has been suggested that the cooperativity of ATP binding and hydrolysis occur at two levels: (i) an intra-ring “MWC or concerted” positive cooperativity and (ii) an inter-ring “KNF or sequential” negative cooperativity (25–27). Interestingly, the R13G A126V mutant of GroEL affects the “sequential” inter-ring negative cooperativity without affecting the “concerted” intra-ring positive cooperativity of ATP binding, or the inhibitory effect of GroES cofactor on the extent of ATP hydrolysis of GroEL (28).

Other factors that seem to affect ATPase activity of GroEL include phosphorylation. It has been shown that, under heat shock, a small fraction of GroEL (3–5%) undergoes phosphorylation, which enhances the binding of GroEL to unfolded substrates in vitro (29, 30). Although the physiological relevance of phosphorylation remains to be elucidated, the phosphorylated, but not the unphosphorylated, form of GroEL can be dissociated from the unfolded substrates in an ATP-dependent fashion that does not require GroES (29).

The arginine-13 residue has been postulated to be involved for the inter-ring negative cooperativity of ATP binding by GroEL (28). The corresponding residue of GroEL arginine-13 is leucine-19 in Cct6p from S. cerevisiae (Fig. 3; Table 1). In light of the seemingly opposing effects of these mutants on the cooperativity of ATP binding in GroEL, we have created the cct6–3 allele, which has a L19S replacement, and have investigated the potential suppression of the cct1–1 allele by cct6–3. Notably, the cct6–3 is functional during vegetative growth (phenotype A) (ref. 10). Strikingly, the growth of the cct1–1 CCT6⁺ strain was enhanced in the presence of high-copy number plasmid containing cct6–3 (data not shown; Table 1), and the sensitivity to the microtubular inhibitory drug, TBZ, was reduced (data not shown). Thus, we established that cct6–3 is a dominant suppressor of cct1–1. Furthermore, tetrad analysis showed that CCT6⁺/cct6–3 cct1-Δ spores were not viable, demonstrating that the cct6–3 allele was not acting as a bypass suppressor by replacing the Cct1p function. (Tetrad analysis was carried out in the following genetic background: MATα/MATa ura3–52/ura3–52 trp1–7/trp1–7 leu2–3, 112/leu2–3, 112 CCT1⁺/cct1::LEU2 CCT6⁺/CCT6⁺ [cct6–3 URA3 YEp352]). In addition, 42 other cct6 alleles, including CCT6⁺, were ineffective in the suppression of cct1–1 (Table 1), indicating that the cct6–3 suppressor was highly specific (Table 1).

Figure 3.

An alignment of amino acid sequences of the GroEL, TF55, and S. cerevisiae Cct1p–Cct8p subunits. Amino acid positions are indicated on the top row for Cct6p and GroEL (shown in parentheses). The leucine-19 residue of Cct6p and its parologous residues from other chaperonins, including arginine-26 of Cct1p and arginine-13 of GroEL, are denoted by a box. The putative cAMP kinase phosphorylation motif, RXS, for Cct6p is shown in reverse type.

To test for the range of suppressibility, the high-copy number cct6–3 plasmids were inserted in the other conditional lethal cct strains, cct1–2, cct1–3, cct2–1, cct2–2, cct2–3, cct2–4, cct3–1, and cct4–1. Significantly, cct6–3 also suppressed, but did not bypass, the cct2–3 and cct3–1 cold-sensitive lethal mutations, but not other conditional lethal mutations, cct1–2, cct1–3, cct2–1, cct2–2, cct2–4, and cct4–1 (Table 1). DNA sequencing of these various cct mutants indicated that all suppressible sites reside either immediately adjacent to or within the putative ATP-binding motif, GDGTTT/S, but not elsewhere, including three other conserved loop structural motifs near the ATP-binding pocket (e.g., cct1–2 and cct1–3; Fig. 2). Although the sample number was small, this result suggested that whereas LGPVG, LTKDG, GDGTT, and GAG motifs together form the ATP-binding pocket, based on the crystal structure of GroEL (31, 32), only a specific function encoded at the GDGTT motifs common to Cct subunits appeared to be restored by the L19S mutation. More importantly, suppression without bypassing the requirements for these cct mutants suggests that numerous essential species of Cct rings, such as homo-oligomeric rings of Cct subunits, each of which is responsible for the folding of a specific spectrum of substrates, could not exist in the cell. This is because the rings containing cct1–1, cct2–3, or cct3–1, but lacking cct6–3, would not be functional at 15°C in vivo. For example, the complexes with different portions of various subunits, including stacking of two different homo-oligomeric rings in various combinations, would be comprised of Cct assemblages such as [Cct6–3p Cct1–1p], [Cct6–3p Cct2⁺p], [Cct1–1p Cct2⁺p], or [Cct1–1p Cct3⁺p], etc., subunits, in which [Cct1–1p Cct2⁺p] and [Cct1–1p Cct3⁺p]), etc., lacking Cct6–3p would be responsible for the conditional lethality of the cell due to the cct1–1 defect. Therefore, for the suppression by Cct6–3p to occur, it appears that all other Cct subunits need to be physically associated with Cct6–3p. Because there are equimolar amounts of Cct subunits, this physical linkage among Cct6p and Cct1p-Cct3p has to reside within a single [Cct1p Cct2p Cct3p Cct6p, etc.] complex, but not different [Cct1p Cct6p, etc.], [Cct2p Cct6p, etc.], and [Cct3p Cct6p, etc.] complexes, because it would require severalfold more Cct6p than other Cct subunits (1–3). Together with the synthetic lethal effects among the cct1 through cct4 mutants (5, 6), our data suggest the presence of a single species of [Cct1p Cct2p Cct3p Cct4p Cct6p, etc.] complex. In other words, a single species of the Cct complex containing a stoichiometrical array of all eight subunits should be responsible for the folding of all potential substrates.

The physiological role of phosphorylation of GroEL has not been well defined. Furthermore, no conclusive evidence on phosphorylation of mammalian Cct subunits has been reported. However, because overexpression of Cct6p suppressed a tor2 mutation in S. cerevisiae (33), and because we demonstrated here that homo-oligomeric Cct complexes are not expected to be functional, we suggest an interesting possibility that Cct6p directly interacts with Tor2p and Sit4p in the kinase/phosphatase signal transduction pathway, leading to actin hyperpolarization (33, 34). Therefore, the potential effect of phosphorylation of the serine residue from a putative cAMP kinase-dependent phosphorylation motif, RXS (refs. 35 and 36; Fig. 3), on the ATPase activity was investigated by suppression analysis (Table 1). Notably, this motif is not present in GroEL, but appears to be present in all Cct subunits, and is conserved among orthologous Cct6p across species (7), although the location of such a motif for each subunit varies (10). Remarkably, the suppressibility of cct1–1, cct2–3, and cct3–1 by cct6–3 was enhanced by the copresence of S157A mutation (cct6–105) within the RXS motif (refs. 35 and 36; Table 1). In contrast to the enhancement by S157A, the presence of another S157E mutation (cct6–106), mimicking constitutive phosphorylation at the RXS motif, resulted in a weaker suppressibility of cct1–1, cct2–3, and cct3–1, than that by L19S (cct6–3) alone (Table 1). Furthermore, a second R155K mutation, known to reduce the efficiency of phosphorylation of the RXS motif by several order of magnitude in vitro (35), enhanced the suppressibility of cct2–3 by L19S, although no appreciable amount of such enhancement was noticed in cct1–1 and cct3–1. Neither the S157A or the R155K mutation had a suppressing effect by itself, and the weakening of suppressibility by S157E was not due to a loss of cct6–106 function (Table 1). Therefore, our data suggests the critical nature of this region (residues 155–157) from the putative intermediate domain of Cct6p (10) in the modulation of ATPase activity of Cct complex. It remains to be established biochemically whether this effect is indeed linked to phosphorylation/dephosphorylation regulation of Cct6p.

How the conserved putative ATP-binding pockets from various Cct subunits interact during the cooperative ATP binding/hydrolysis within the Cct hetero-octamer was also investigated genetically. Although the conditional lethal cct1–1, cct2–3, and cct3–1 alleles contain mutations at the GDGTT putative ATP-binding motif, a similar single replacement within or a total replacement of the homologous motif in Cct6p is well tolerated by the cell, suggesting that Cct6p might be a noncatalytic subunit (10). Alternatively, Cct6p was catalytically involved in the cooperative ATP binding/hydrolysis, but was more amenable than other Cct subunits to mutations in the active site due to an heightened threshold (10). To this end, double-mutant analyses involving two of the four putative ATP-binding loops of Cct6p in all possible combinations were carried out by the plasmid shuffle technique. The phenotypes of the strains having double replacements were determined. Strikingly, double mutations involving any two of the four putative ATP binding loops were all lethal, phenotype D (cct6–6 [G38S G59R], cct6–7 [G38S G90E], cct6–8 [G38S G414E], cct6–9 [G38S G416E], cct6–15 [G59R G90E], cct6–16 [G59R G414E], cct6–17 [G59R G416E], cct6–22 [G90E G414E], and cct6–23 [G90E G416E]). The interactions among the ATP binding loops appeared to be specific, as no negative synergistic effects were observed between the S157A mutation from the RXS motif, and any of the G38S, G59R, G90E, or G416E mutations from the putative ATP binding motifs within the putative equatorial domain of Cct6p (cct6–109 [G38S S157A, phenotype B], cct6–110 [G59R S157A, phenotype B], cct6–111 [G90E S157A, phenotype B], and cct6–112 [S157A G416E, phenotype C]). This supported a heightened threshold of Cct6p to mutations within the catalytic sites. To address whether or not the putative active site in Cct6p interacted with those from other subunits, tests for suppression of cct1–1 by double-mutant cct6 alleles, involving L19S and a second mutation from any of the conserved glycine residues within the four putative ATP binding motifs, were performed (Table 1). Remarkably, all second site mutations that affected the conserved glycine residues of the putative ATP binding pocket, G38S, G59R, G90E, and G414E (cct6–101 [L19S G38S, phenotype A], cct6–102 [L19S G59R, phenotype B], cct6–103 [L19S G90E, phenotype C], and cct6–107 [L19S G414E, phenotype D], respectively), abolished the suppressibility of cct1–1 by L19S (Table 1). This observation suggested that Cct6p interacted with other subunits catalytically in the cooperative ATP binding/hydrolysis.

Notably, the loss of suppression by G38S and G59R could not be explained by a loss of cct6–101 and cct6–102 functions (see above). On the other hand, the cct6–103 (L19S G90E) and cct6–107 (L19S G414E) alleles were nonfunctional. Therefore, the differential interactions of these glycine residues with the leucine-19 residue from the N terminus of Cct6p further supports the notion that the LGPVG, LTKDG, GDGTT, and GAG motifs (Fig. 2) might possess unique properties. This might be the basis for why L19S of Cct6p extrageneticly suppressed mutations in the parologous putative ATP-binding site, GDGTTT/S, of cct1–1, cct2–3, and cct3–1, but not mutations in the LGPKG (e.g., cct1–3) and GAG (e.g., cct1–2) motifs (Fig. 2). Additionally, the suppression did not appear to occur entirely via a direct restitution of cct1–1, etc., defects by the L19S mutation of Cct6p as this L19S replacement exacerbated the defects caused by mutations from the parologous motif in Cct6p, i.e., L19S G90E (cct6–103). Although the basis of the negative synergism between L19S and G90E or G416E remains to be established biochemically, both L19S and G90E could affect ATP hydrolysis as inferred from the GroEL R13G A126V mutant (31, 32), thereby accounting for the synthetic lethal phenotype of the cell.

The growth of cct1–1, cct2–3, and cct3–1 at 15°C with pRS315 single-copy plasmids carrying cct6–3 (L19S) or cct6–105 (L19S S157A) alleles was further tested. Remarkably, the single-copy plasmid was stronger than the multi-copy plasmid in the suppression of cct2–3 by the cct6–3 and cct6–105 alleles (Table 1). Furthermore, the reverse is true for cct1–1 and cct3–1. Because the Cct complex is composed of two rings, this suggested that either one of the rings, cct2–3 CCT6⁺ or cct2–3 cct6–3, might be functionally or structurally more favorable than the other. Alternatively, Cct complex with symmetrical cct6–3 cct2–3 rings might be structurally or functionally less stable. In either case, the cct6–3 (L19S) allele apparently affected the inter-ring communication, because if cct6–3 exerted its suppressing effect strictly via an intra-ring mechanism, we would expect a high-copy number to be a stronger suppressor for cct2–3, just as for cct1–1 and cct3–1. Regardless of the mechanism of suppression by cct6–3, it appeared that cct1–1, cct2–3, and cct3–1 possessed differential sensitivity to the inter-ring allosteric adjustment conferred by cct6–3 and the modulatory effects by mutations in residues 155–157, RSS. This is consistent with the notion that there existed a functional hierarchy among various Cct subunits as suggested recently (8, 10). This idea was further supported by the finding that the cct1–5 (R26I) allele, which has a mutation parologous to leucine-19 of Cct6p and arginine-13 of GroEL (Fig. 3), failed to suppress cct2–3 and cct3–1, despite its ability to complement the cct1–1 defect (Table 1).

DISCUSSION

In this study, we have excluded the presence of multiple essential species of Cct complex, and established the unique hetero-oligomeric nature of the Cct rings based on our suppression analyses of several mutant alleles of Cct1p, Cct2p, Cct3p, and Cct6p. The suppression shown in this study was specific, involving the N terminus of Cct6p, but not that of Cct1p, and acted exclusively on the GDGTT motif, but not three other conserved putative ATP-binding motifs. In addition, we have demonstrated the potential modulatory effects on cooperative ATP binding/hydrolysis by the putative phosphorylation motif, RXS, from the hypothetical intermediate domain of Cct6p. This study suggests that all Cct subunits participated structurally in the formation of a single species of hetero-oligomeric ring complex, but not multiple essential species, such as by stacking of homo-oligomeric rings. An important functional implication would be that this single species would be responsible for the folding of all potential substrates, and that substrate specificity could not be a result of formation of multiple different assemblages by various Cct subunits. Rather, these subunits participated in the catalytic cooperativity of ATP binding/hydrolysis in an apparently nonidentical fashion.

It is important to note that all of the cct1–1 etc. mutations suppressed by cct6–3 are recessive, despite the fact that they all contain mutations within the actual ATP-binding site. Without biochemical data, it is not known whether these recessive alleles may affect both the structure and function of the protein. Recent biochemical evidence indicated that Mg²⁺–ATP promote the assembly of monomeric forms of α and β subunits of thermosome, and a similar mechanism also appears to exist in GroEL chaperonin (9). It is possible that subunits with GDGTT mutations promoted instability of the Cct complex due to their decreased affinity for ATP. However, because multiple mutated subunits are suppressed, the structural instability of these mutants, if any, must be reversed via long-range allosteric changes within the whole Cct complex. Therefore, while the mechanism could be long ranged, it is also specific as the parologous cct1–5 (R26I) mutation fails to suppress these mutants.

The following model can explain the observations from this and other genetic studies (refs. 5, 6, and 10; Fig. 1):

Figure 1.

(a) Model depicting a highly sequential order of positive cooperative ATP binding within the stoichiometrical array of Cct subunits in the ring in the R state (denoted by an asterisk), and negative cooperative ATP binding in the opposite ring in the T state, based on the KNF model (see text). (b) Proposed sequence of cooperative ATP binding/hydrolysis among Cct1p, Cct2p, Cct3p, and Cct6p based on the KNF model. In the Cct complex containing normal Cct6⁺p, cooperative ATP binding occurs in the order of Cct1p, Cct3p, Cct2p, and Cct6p. Cct6p is placed the downstream-most position due to its high tolerance to mutations at the putative ATP-binding loop, GDGTT. In contrast, in Cct complex containing Cct6–3p, cooperative ATP binding can occur in the order of Cct6–3p, Cct1p, Cct3p, and Cct2p, due to an increased T to R state transition in Cct6–3p. This results from a change of the relative positions of Cct6–3p with respect to Cct1–1p, Cct2–3p, and Cct3–1p, thereby displacing these mutants further downstream in the course of sequential cooperative ATP binding/hydrolysis. As a result, mutations at the GDGTT motifs in these mutant subunits become more tolerable to the cell. Double arrows between subunits indicate unknown numbers of intervening subunits. Single arrow indicates two adjacent subunits, as is proposed between Cct2p and Cct6p.

(i) All Cct subunits participate in the cooperative ATP binding/hydrolysis within the hetero-octameric rings of Cct complex. This conclusion is based on the abolition of the cct6–3 suppressing effects on cct1–1, cct2–3, and cct3–1 by second-site mutations in the ATP pocket of Cct6p. The involvement of various ATP-binding pockets along the same pathway has been suggested by the synthetic lethality between the cct1–1 and cct3–1 mutants (5, 6). In addition, each of the four ATP-binding loops of Cct6p appear to contribute to ATP binding/hydrolysis, based on the synthetic lethality between any two of the four loops in Cct6p.

(ii) ATP binding occurs sequentially in a defined order, due to the intrinsic differences of these subunits in their energy states. This is because some parologous mutations of Cct subunits within the ATP-binding pocket confer more severe growth phenotypes than others (5, 6, 10); cct1–5 and cct6–3 differ in the ability to suppress cct1–1, cct2–3, and cct3–1, and cct1–1, cct2–3 and cct3–1 differ in their response to single and high-copy suppression. In another words, the homologous putative ATP-binding pockets of various Cct subunits appear to be nonidentical in importance, consistent with the notion that there exists a functional hierarchy among these subunits in the process of ATP binding/hydrolysis within the Cct complex (10). Whereas the bases of this functional hierarchy and the differences of Cct1p and Cct6p in suppression remain unknown, it is apparently inconsistent with the “concerted or MWC” model, which presumes strict molecular symmetry among all subunits in either the tense state T or the relaxed state R, as proposed for GroEL (25). The MWC model assumes that all subunits exist at the same time in either the T or R state at the same time (26). The R state has higher affinity for the ligand than the T state. The affinity for the ligand is always the same among subunits, due to the presence of molecular symmetry among subunits, and the stereo-specific active sites of the subunits. In the case of the allosteric binding of ATP by the Cct hetero-octamer, the MWC model would predict that each nonidentical subunit of Cct should be constrained symmetrically in the T state with respect to one another in the absence of ATP, and a concerted transition into a symmetrical Cct hetero-octamer in the high-affinity state for ATP should occur. In addition, each subunit should be equally capable of transiting from the T state to the R state (i.e., stochastically), and in either state should have equal affinity for ATP. In another words, we would anticipate the same detrimental effects of mutations within the ATP-binding sites for all Cct subunits, and suppressing effects of cct2–3 and cct3–1 by the cct6–3 (L19S) and the cct1–5 (R26I) mutations.

An alternative model to explain cooperative ligand binding by oligomeric proteins is the “simple sequential model” or “KNF model” proposed by Koshland et al. (27). In contrast to the MWC model, this model can explain both positive and negative cooperativity of ligand binding by oligomeric proteins, although it was used exclusively to explain the inter-ring negative cooperative ATP binding by GroEL (25). This model assumes that the affinity for the ligand progressively changes as increasing number of subunits are occupied by the ligand, because each binding event is associated with a separate conformational adjustment of the enzyme. In addition, this model takes into consideration the geometric arrangements of subunits (27). Based on the sequential or KNF model, ATP binding to Cct subunits occurs progressively, and intermediates comprised of certain Cct subunits in their T states and others in their R states could exist until all subunits in Cct complex allosterically transit to R state and bind ATP.

A unique sequential binding of ligands could exist in hetero-oligomers. This type of allosteric behavior could be influenced by the geometry of the molecule or by the effect of nonidentical subunits, and is perhaps best illustrated by the x-ray crystallographic finding that there are negligible diagonal α–α and β–β interactions within hemoglobin hetero-tetramers during cooperative binding to oxygen (37). Notably, our results are the most consistent with sequential and progressive recruitment of hetero-oligomeric Cct subunits in a defined order by allosteric shifts into the R state (Fig. 1). In such a modified sequential model, the nonidentical Cct subunits formed a specific geometric array and cooperative ATP occurred in a specific order due to differences in the coefficient of T → R transition for each subunit, in the affinity for ATP, in the free energy change for intra-subunit allosteric adjustment upon ATP binding to the active site, and in the free energy barrier at each interface for subunit-to-subunit allosteric adjustment. As a result, increasing numbers of ATP-binding sites from various subunits are recruited, and the affinity for ATP progressively increases. As such, subunits residing later in the process of cooperative ATP binding/hydrolysis would be less sensitive to mutations within a single loop of the ATP-binding pocket than those residing earlier in the process, thereby explaining the functional hierarchy observed for Cct subunits. According to this model, Cct6p is assigned downstream to Cct1p, Cct2p, and Cct3p in ATP binding/hydrolysis, based on our genetic evidences (Fig. 1b).

(iii) Conversely, because of inter-ring negative cooperativity of ATP binding, as suggested by the studies of the GroEL R13G V126A mutant (28), the T → R transition of each subunit within the ring could be further modified by the inter-ring negative interactions. In “normal” state, only one Cct ring undergoes sequential positive cooperative ATP binding among the hetero-octamer. However, because of likely disruption of inter-ring negative regulation of T → R transition due to cct6–3, the equilibrium is shifted toward the R state for that subunit. As a result, in Cct complex containing Cct6–3p, cooperative ATP binding can begin in Cct6–3p and proceed to Cct1p, Cct3p, or Cct2p due to an increased T to R state transition in Cct6–3p (Fig. 1b). This results in a change of the relative positions of Cct6–3p with respect to Cct1–1p, Cct2–3p, and Cct3–1p, thereby displacing these mutants further downstream in the course of sequential cooperative ATP binding/hydrolysis. Consequently, mutations at the GDGTT motifs in these mutant subunits become more tolerable to the cell. On the other hand, cct1–5 (R26I; parologous to L19S of cct6–3) failed to suppress cct2–3 and cct3–1, suggesting that Cct1p should reside upstream, but not downstream, of both Cct2p and Cct3p, because enhancement from T to R state in Cct1–5p would not alter its relative positions with respect to Cct2–3p and Cct3–1p. Notably, the suppression of cct1–1 and cct3–1 by cct6–3 showed a positive gene dosage effect, suggesting a low efficiency of the novel pathway in cooperative ATP binding initiated by Cct6–3p. In contrast, single-copy cct6–3 resulted in stronger suppression of cct2–3 than high-copy cct6–3, suggesting the presence of interference of another higher efficiency pathway by the lower efficiency novel pathway. This “higher efficiency” pathway, presumably also via an enhanced T to R transition in Cct6–3p but at a lower gene dosage, could well be a continuation of the authentic pathway of cooperative ATP binding/hydrolysis otherwise blocked by the cct2–3 mutation. This would put Cct2p in an exclusive position immediately upstream to Cct6p in the sequence of cooperative ATP binding among Cct subunits (Fig. 1b). Thus, the proposed sequence of cooperative ATP binding/hydrolysis of the normal (CCT6⁺) complex would be Cct1p → → Cct3p → → Cct2p → Cct6p, where the single arrow (→) indicates adjacent subunits and the double arrows (→ →) indicate an unknown number of intervening subunits. On the other hand, the sequence of cooperative ATP binding/hydrolysis of the cct6–3 cct3–1 complex would be Cct6–3p → → Cct1p → → Cct3–1p → → Cct2p, whereas that of the cct6–3 cct2–3 complex would be Cct1p → → Cct3p → → Cct2–3p → Cct6–3p (Fig. 1b).

In summary, a modified KNF model is used to explain the functional hierarchy among Cct subunits in putative ATP binding/hydrolysis, and the differential capacity of parologous cct6–3 and cct1–1 in the suppression of cct1–1, cct2–3, and cct3–1, all of which have mutations at the conserved ATP-binding/hydrolysis motif, GDGTT. By using this model, cooperative ATP binding/hydrolysis in the Cct complex occurs in the order of Cct1p, Cct3p, Cct2p, and immediately afterward, Cct6p. Biochemical studies should reveal the mechanisms of suppression by cct6–3, the cooperative ATP binding/hydrolysis within the Cct rings, and the significance of the RSS motif (residues 155–157) in Cct6p in the modulation of the ATPase activity within Cct complex.

Note

The sequential order of Cct1p, Cct3p, Cct2p, and Cct6p in the binding of ATP proposed in this paper is consistent with the findings of Liou and Willison (38), who determined the subunit orientation in the Cct complex by biochemical analysis of partially assembled complexes.