INTRODUCTION

GroEL₁₄ is an Escherichia coli chaperonin that facilitates the proper folding of proteins in an ATP-dependent manner (1-3). As determined by x-ray crystallography, GroEL₁₄ seems to be a hollow cylinder of 14 identical 57-kDa subunits consisting of 2 heptamer rings stacked back-to-back with a dyad symmetry (GroEL₇·GroEL₇) (4). GroEL₁₄ functionally cooperates with co-chaperonin GroES₇ (5), a single heptameric ring of 10-kDa subunits that binds one end, or two ends in some cases, of the GroEL₁₄ cylinder (6-8). From a thermophilic bacterium, Thermus thermophilus, GroEL homolog Tcpn60¹ (Thermus chaperonin 60) is purified as a complex with GroES homolog Tcpn10. The complex, termed Thermus holo-chaperonin (T.holo-cpn), is composed of two heptameric rings of Tcpn60 and a single ring of Tcpn10₇ (9-12). In contrast to chaperonins from E. coli and T. thermophilus, several members of the chaperonins, including those from Thermoanaerobacter brockii (13, 14) and mitochondria (15, 16), are purified as a single heptameric ring. In addition, purified chaperonin from Paracoccus denitrificans is a mixture of a large number of tetradecamers and a small number of heptamers (10). The physiological meaning of such a divergence in the quarternary structure of chaperonin has not been understood.

Recently, we found that when T.holo-cpn is incubated with ATP and K⁺, it splits into two parts at the equator plane between the two rings of Tcpn60₇, producing cone-shaped particles (Tcpn60₇·Tcpn10₇) and ring-shaped particles probably corresponding to Tcpn60₇ (17). Then we observed that the products of the split reaction can reassociate to form T.holo-cpn under the appropriate conditions (18). In contrast to the split reaction, Todd et al. (14) reported that the single-ring T. brockii cpn60₇ dimerizes to form a double-ring structure in the presence of T. brockii cpn10₇ and ATP. These results raise the question of whether GroEL₁₄ also undergoes the tetradecamer-heptamer transition.

Here, we report that when GroEL₁₄ and T.holo-cpn are incubated with ATP and KCl, hybrid chaperonins such as GroEL₇·Tcpn60₇·Tcpn10₇ are formed as a result of the heptamer exchange reaction. This suggests that ATP/K⁺-dependent transient dissociation of a tetradecamer into heptamers occurs not only in Thermus chaperonin but also in GroEL.

EXPERIMENTAL PROCEDURES

Isopropylmalate dehydrogenase (IPMDH) from T. thermophilus strain HB8 was a kind gift from Dr. T. Oshima and his colleagues (Tokyo University of Pharmacy and Life Science, Hachioji, Japan) (19). (2R*,3S*)-3-Isopropylmalic acid, a substrate of IPMDH, was purchased from Wako Pure Chemical Corp. T.holo-cpn was purified as described previously (9, 20). Tcpn60₁₄ expressed in E. coli was purified using modifications of procedures described previously (12). The lysate of E. coli cells containing expressed Tcpn60₁₄ was heated at 70 °C for 20 min. The supernatant containing Tcpn60₁₄ was recovered by centrifugation and applied to a hydrophobic interaction chromatography (Butyl-Toyopearl). The fractions containing Tcpn60₁₄ were pooled and concentrated by ammonium sulfate precipitation. The concentrated protein solution was applied to a gel permeation HPLC column (G3000SWXL; Tosoh) equilibrated with 25 mM Tris-HCl, pH 7.0, 100 mM Na₂SO₄, and 20% (v/v) methanol. The fractions containing Tcpn60₁₄ were further purified by a DEAE-5PW HPLC column. Tcpn10₇ expressed in E. coli was purified as described previously (12). GroEL₁₄ and GroES₇ were purified as described previously (21) from the lysate of E. coli cells bearing the multicopy plasmid pACYC 184 carrying groES-groEL genes, which was a kind gift from Dr. K. Ito (Kyoto University, Kyoto, Japan) (22). A mutant GroELAEX₁₄ (C138S, C458S, C519S, D83C, and K327C) was purified as described previously (21). Purified chaperonins were stored as a suspension in 65% ammonium sulfate at 4 °C.

T.holo-cpn or Tcpn60₁₄ (5 µg) was mixed (final volume, 10 µl) with GroEL₁₄ (5 µg) and incubated at 37 °C for 10 min in Buffer A (25 mM Tris-HCl, pH 7.5, 300 mM KCl, and 5 mM MgCl₂) containing 1 mM ATP, unless otherwise indicated. The sample solutions were applied to nondenaturing polyacrylamide gel electrophoresis (6% acrylamide), and electrophoresis was continued for about double the duration of the period required for the leading dye (bromphenol blue) to reach the front of the gel. The protein bands were stained by Coomassie Brilliant Blue. Only the regions of the chaperonin protein bands are shown in the figures.

The ammonium sulfate precipitate of the mixture of parent and hybrid chaperonins was solubilized in a minimum volume of Buffer B (25 mM Tris-HCl, pH 7.5, and 5 mM MgCl₂) and applied to a Sephadex G-25 (Pharmacia) column to remove the excessive salts. The eluted protein solution was applied to a DEAE-5PW (Tosoh) column equilibrated with Buffer B and eluted with a 0-1.0 M NaCl gradient at 1 ml/min. The chromatography was monitored by absorbance at 280 nm.

ATPase activities were assayed by measuring the amount of produced inorganic phosphate (23). Typically, the reaction was started by the addition of ATP (final concentration, 1 mM) to Buffer A containing 0.88 µM² GroEL₁₄ or Tcpn60₁₄ and, when indicated, 1.3 µM GroES₇ or Tcpn10₇. The assay solution was preincubated for 10 min at 37 °C before the addition of ATP. The reactions were terminated by the addition of perchloric acid after incubations at 37 °C for 5, 10, 15, and 20 min. The solution was treated with a malachite green reagent, and the absorbance at 630 nm was measured. One unit of activity is defined as the activity that hydrolyzes 1 µmol of ATP/min.

IPMDH (16.2 µM) denatured in 6.4 M guanidine HCl was diluted 25-fold at 37 °C in Buffer A containing the components indicated in the figure legends. After a 20-min incubation at 37 °C, an aliquot was withdrawn, and the reactivated IPMDH activity was determined as described previously (9).

Protein concentration was determined by the method of Bradford with bovine serum albumin as a standard (24). Proteins were analyzed by polyacrylamide gel electrophoresis either on a 10% polyacrylamide gel in the presence of SDS (SDS-PAGE) or on 6% polyacrylamide gels without SDS (native PAGE) (25). To obtain higher resolution on the native PAGE, electrophoresis was continued for about double the duration of the period required for the leading dye (bromphenol blue) to reach the front of the gel. Gels were stained by Coomassie Brilliant Blue R-250.

RESULTS

After we found the T.holo-cpn split (17), we attempted to identify the heptameric state of GroEL under various conditions but had no success. If GroEL₁₄ splits only transiently in its ATPase cycle, we could not detect the heptamer GroEL by the usual methods. Then we used Thermus chaperonin as a trap for GroEL₇, that is, we incubated GroEL₁₄ with ATP/K⁺ in the presence of T.holo-cpn and examined whether the hybrid chaperonin containing GroEL₇ and Tcpn60₇ was formed. We took advantage of the different electrophoretic mobility of GroEL₁₄ and T.holo-cpn in native PAGE (Fig. 1A, lanes 1-5) to identify the hybrid. As shown in Fig. 1A, lane 6, two closely moving bands appeared between T.holo-cpn and GroEL₁₄ when they were incubated with ATP/K⁺. The NH₂-terminal amino acid sequences of the two bands confirmed that the upper band contained GroEL, Tcpn60, and Tcpn10 and that the lower band contained GroEL and Tcpn60 (data not shown). Yields of phenylthiohydantoin derivatives from GroEL, Tcpn60, and Tcpn10 (upper band) were generally close to each other, indicating that the upper and lower bands corresponded to GroEL₇·Tcpn60₇·Tcpn10₇ and GroEL₇·Tcpn60₇, respectively. As described later, an analysis of isolated hybrid chaperonins supported the structures described above. For these hybrids to be formed, heptamer exchange reactions should occur, that is, both T.holo-cpn and GroEL₁₄ should split into heptamers that then rebind each other with a random combination into tetradecamers. Hereafter, we term the hybrid chaperonins as follows: GroEL₇·Tcpn60₇·Tcpn10₇, Hybrid (EL-60-10); and GroEL₇·Tcpn60₇, Hybrid (EL-60).³

The conditions required for the formation of hybrid chaperonins were the same as those required for the split reaction of T.holo-cpn (17); formation was absolutely dependent on ATP and K⁺, and other combinations such as adenosine 5-(, -imino) triphosphate + K⁺ (Fig. 1A, lane 7), adenosine 5-O-(thiotriphosphate) + K⁺ (data not shown), ADP + K⁺ (lane 8), and ATP + Na⁺ (lane 10) were not effective in generating hybrids. When free Mg²⁺ was removed by trans-1,2-diaminocyclohexanetetraacetic acid, no hybrid was formed (lane 9). The relatively high concentrations of ATP and K⁺ were necessary. The hybrid formation was half-maximal at ~50 µM ATP and ~50 mM K⁺ and saturated at ~300 µM ATP and ~300 mM K⁺ (Fig. 1, B and C). The addition of excess Tcpn10₇ to the reaction mixture resulted in an increased yield of hybrid chaperonins with a simultaneous decrease of T.holo-cpn (Fig. 1D, lane 2), but the addition of GroES₇ had only little, if any (lane 3), effect. The hybrids were formed rapidly. In 1 min (Fig. 1E, lane 5), hybrids with an amount similar to that formed in 10 min (lane 6) were detected, and a significant amount of hybrids was formed even in 20 s (lane 4). That is as fast as a single ATPase turnover in the GroEL catalytic cycle (6).

The ATPase activity of Thermus chaperonin at 37 °C, the temperature at which hybrid formation was observed, is very low (see Fig. 5); hydrolysis of a single ATP molecule by Tcpn60₁₄ takes ~15 s. Nevertheless, hybrid chaperonin was formed within 20 s. This result means that only the hydrolysis of a single ATP by Tcpn60₁₄ is sufficient to form the hybrid chaperonins. This rapid formation of hybrid might be related to the observation that the formation of a tetradecamer from T. brockii cpn60₇ is also very rapid, occurring before all the cpn60 subunits could hydrolyze ATP (14). Although the real reason why the hybrid was formed in such a short period is not known, one of the possible explanations is that the initial single turnover by one cpn60 in the tetradecamer might induce a quarternary structural change in the double ring of cpn60₇ in the presence of a high concentration of K⁺.

To know whether cpn10 is required for hybrid formation or not, next we used Tcpn60₁₄ instead of T.holo-cpn as one of the parent chaperonins. Tcpn60₁₄ was isolated from recombinant E. coli (12), and we confirmed that Tcpn60₁₄ also split in the presence of ATP/K⁺.⁴ The hybrid chaperonin was formed between Tcpn60₁₄ and GroEL₁₄ in the presence of ATP/K⁺ (Fig. 2, lane 6). Unlike the experiment using T.holo-cpn, only a single band appeared between the parent chaperonins. As expected, this band was indeed Hybrid (EL-60), because NH₂-terminal amino acid sequencing showed that the band contained an almost equal amount of Tcpn60 and GroEL. It is likely that in the experiments described in Fig. 1, Hybrid (EL-60) was formed at first, and Hybrid (EL-60-10) was generated next by attaching Tcpn10₇ to Hybrid (EL-60).

In spite of the fact that the condition required for hybrid formation as described above was the same condition required for the split reaction of Thermus chaperonin (17, 18), there was no direct evidence of a requirement for ATP hydrolysis by GroEL₁₄ for hybrid formation. To address the question, we used a GroEL mutant called GroELAEX instead of the wild-type GroEL as a parent chaperonin (21). GroELAEX is a mutant in which apical and equatorial domains in the same GroEL subunit can be cross-linked in a reversible manner (apical-equatorial cross (X)-link) (21). In the presence of a reducing reagent, GroELAEX₁₄ retains normal functional activity as a chaperonin. In contrast, oxidized GroELAEX₁₄, which is locked in a "closed" conformation by an interdomain disulfide bond, can bind but not hydrolyze ATP (21). Therefore, the requirement for ATP hydrolysis of GroEL₁₄ in the hybrid formation would be tested in the presence or absence of a reducing reagent. Note that the Thermus chaperonins have no cysteine residue (12). Just like wild-type GroEL₁₄, the hybrid chaperonin was formed from GroELAEX₁₄ and Tcpn60₁₄ in an ATP/K⁺-dependent manner under reducing conditions (Fig. 3, lane 6). However, the hybrid was not formed under the oxidizing condition (lane 2), whereas the ATPase activity of GroELAEX₁₄ was completely blocked (21). Wild-type GroEL₁₄ was able to form the hybrid with Tcpn60₁₄, irrespective of reducing or oxidizing conditions (lanes 4 and 8). The inability of oxidized GroELAEX₁₄ to form hybrid chaperonin indicates that ATP binding is not sufficient for hybrid formation and that the occurrence of ATP hydrolysis on GroEL₁₄ is essential for hybrid formation.

The hybrid between GroEL₁₄ and Thermus chaperonins was separated from the parent chaperonins with anion-exchange HPLC (Fig. 4, A and B). The hybrid chaperonin fraction contained Hybrid (EL-60-10) and Hybrid (EL-60) (see Fig. 7A, lane 5). In a similar manner, Hybrid (EL-60) was also purified (see Fig. 7A, lane 4). The relative staining intensities of the GroEL band and the Tcpn60 band in SDS-PAGE (Fig. 4, inset, lanes 1 and 2) were almost the same, again confirming that the hybrid chaperonins consisted of equal molar amounts of each chaperonin subunit. The molecular sizes of the isolated hybrid chaperonins were the same or very close to those of the parent chaperonins, because they were eluted from a gel-permeation HPLC column at the same retention time as that of GroEL₁₄ (data not shown). When hybrid chaperonins formed from GroEL₁₄ and T.holo-cpn were examined by electron micrograph, two kinds of particles, GroEL₁₄-like rectangular particles and bullet-shaped particles similar to T.holo-cpn, were observed (data not shown). Hybrid (EL-60) hydrolyzed ATP at 0.09 unit/mg¹ at 37 °C (Fig. 5). Because T.holo-cpn and Tcpn60₁₄ hydrolyzed ATP very slowly at 37 °C (~0.002 unit/mg¹), the ATPase activity of Hybrid (EL-60) would be mainly attributed to hydrolysis by the GroEL₇ ring moiety in the hybrid. The hybrids produced from GroEL₁₄ and T.holo-cpn, a mixture of Hybrid (EL-60-10) and Hybrid (EL-60), exhibited ATPase activity at 0.08 unit/mg¹. As reported for GroEL₁₄ (e.g., Ref. 26), the addition of GroES₇ or Tcpn10₇ inhibited the ATPase activity of GroEL₁₄ (35-40% inhibition). The ATPase activity of the hybrid chaperonins was also inhibited by GroES₇ and Tcpn10₇ (55-60% inhibition for Hybrid (EL-60), and ~30% inhibition for the mixture of Hybrid (EL-60-10) and Hybrid (EL-60)).

We examined the effect of hybrid chaperonins on the folding of IPMDH from T. thermophilus under a condition in which the yield of spontaneous folding was only ~10% (Fig. 6). The following experiments were carried out in the presence of ATP. Under this condition, GroEL₁₄ alone hardly promoted reactivation (less than 10% reactivation of IPMDH activity), and GroES₇ was required for effective GroEL-promoted folding. Tcpn10₇ was as effective as GroES₇ in this GroEL-promoted folding assay. In contrast to GroEL₁₄, Tcpn60₁₄ alone was able to promote folding of IPMDH (~35%). Further addition of Tcpn10₇ increased the yield of reactivation about twice, whereas the effect of GroES₇ was only marginal. Similar to GroEL₁₄, Hybrid (EL-60) alone had almost no effect on folding (less than 10%), and the addition of GroES₇ or Tcpn10₇ was required for effective folding (80-90%). In the case of the mixture of Hybrid (EL-60-10) and Hybrid (EL-60), the folding of IPMDH was promoted moderately (~35%) even in the absence of GroES₇ or Tcpn10₇, probably due to the endogenous presence of Tcpn10₇. The inclusion of GroES₇ or Tcpn10₇ in the solution caused additional promotion of folding (50-60%). Thus, it is clear that both Hybrid (EL-60) and Hybrid (EL-60-10) are active in promoting protein folding.

The isolated hybrid chaperonins were very stable. After storage at 4 °C for 3 weeks, about 90% of the hybrid chaperonins were still in the hybrid forms (data not shown). To investigate whether the two heptamer rings of hybrid chaperonins could reexchange each other in the presence of ATP/K⁺, we incubated the hybrid chaperonins with ATP/K⁺ and analyzed them by native PAGE. As shown in Fig. 7B, regeneration of the parent chaperonins, GroEL₁₄ and T.holo-cpn (or Tcpn60₁₄), was not observed, irrespective of whether Hybrid (EL-60) or Hybrid (EL-60-10) was used as a starting hybrid chaperonin. Further addition of either GroES₇ or Tcpn10₇ did not change the result (data not shown). This result was unexpected, because if the hybrid chaperonin split in the presence of ATP/K⁺, as observed for Thermus chaperonin, parent chaperonins should be regenerated more or less as a result of random reassociation of heptamers. Then we examined the effects of heat (70 °C for 10 min) or proteinase K treatment on the stability of the hybrid chaperonins. As shown in Fig. 7, C and D, Thermus chaperonin was resistant to both treatments under the conditions tested, whereas GroEL was destroyed. After the isolated hybrid chaperonins were incubated at 70 °C for 10 min, the hybrid chaperonins disappeared completely, and Thermus chaperonin but not GroEL₁₄ appeared (Fig. 7C, lanes 4 and 5). Treatment with proteinase K also gave the same results (Fig. 7D, lanes 4 and 5). These results indicate that only when the GroEL₇ moiety of the hybrid is denatured or proteolyzed, survived heptamer rings of Tcpn60 can reassociate to form Tcpn60₁₄ or T.holo-cpn.

DISCUSSION

In this report, we demonstrated ATP/K⁺-dependent formation of the hybrid between GroEL₁₄ and Thermus chaperonin. This is a result of the heptamer exchange reaction between both chaperonins. One can argue the possibility that the hybrids are made up from three stacked heptamers (GroEL₁₄·Tcpn60₇), that is, that GroEL₁₄ simply binds Tcpn60₇ as a substrate protein. This possibility was excluded from the analysis of molecular size with gel-permeation HPLC, the observation of molecular shape by electron micrograph, and the estimation of the molar ratio of GroEL:T.cpn60 from phenylthiohydantoin derivatives recovered in Edman degradation and from the staining intensity of the bands in SDS-PAGE (Fig. 4, inset). The random incorporation of each cpn60 monomer into the tetradecamers is also very unlikely. If it really happened, numerous protein bands corresponding to the complexes with various combinations of parent chaperonin monomers should have appeared between the bands of parent chaperonins in native PAGE. However, only a single band appeared between parent chaperonins when the hybrid was formed from GroEL₁₄ and Tcpn60₁₄ (Fig. 2). In addition, under the experiment conditions, bands of monomeric GroEL and monomeric Tcpn60 with meaningful staining intensity were not observed in native PAGE (data not shown). Therefore, there is little possibility, if any, that dissociation into monomers occurs before formation of the hybrid.

Hybrid formation is not restricted to the combination of Thermus chaperonins and GroEL and is also observed between Tcpn60₁₄ and chaperonin from P. denitrificans (10) in the presence of ATP and K⁺.⁴ In addition, Burston et al. (27) reported that the hybrid between wild-type GroEL₁₄ and mutant GroEL₁₄, called MR1 (mixed ring), is formed at 42 °C in the presence of ATP and K⁺. These observations suggest that hybrid formation is not an exceptional event but a rather common reaction among chaperonins from several species. Because the conditions for hybrid formation are nearly physiological, hybrids can be formed even in a living cell. Indeed, when we expressed Tcpn60₁₄ in E. coli cells, a part of the chaperonin was purified as a form of hybrid chaperonin that was separated from Tcpn60₁₄ by anion-exchange HPLC (see "Experimental Procedures").⁴ Furthermore, the observation that single-ring T. brockii cpn60 dimerizes to a tetradecamer in the presence of both adenine nucleotides and cpn10 (14) also implies that there is a heptamer exchange in the bacterium.

Unless GroEL₁₄ and Tcpn60₁₄ split into GroEL₇ and Tcpn60₇, hybrid formation is impossible. The ATP/K⁺-dependent split of Thermus chaperonin has been established, and Tcpn60₇ can be readily detected and isolated (17). A stable heptameric form of cpn60 has also been isolated from T. brockii (13, 14) and from mitochondria (15, 16). A possible heptameric form of GroEL has been reported by two groups. Mendoza et al. (28) suggested that a GroEL species, possibly heptamers, is detected in the presence of 3 M urea after the addition of unfolded rhodanese. Mizobata and Kawata (29) observed a GroEL species exhibiting decreased light scattering in the presence of less than 1 M guanidine HCl, and they speculated that the species is a heptamer of GroEL (29). We also observed a decrease in light scattering of GroEL₁₄ in response to the addition of ATP.⁴ The fact that the hybrid is formed between wild-type GroEL₁₄ and mutant MR1-GroEL₁₄ under the appropriate conditions implies that not only mutant MR1-GroEL₁₄ but also wild-type GroEL₁₄ splits into heptamers (27). Therefore, GroEL₁₄ most likely splits, although the final conclusion on the presence of GroEL₇ should be reserved until success in isolating the heptameric form of GroEL has been achieved.

Once the hybrid is formed, it is very stable. Parent chaperonins are not regenerated from hybrid chaperonin even in the presence of ATP/K⁺ (Fig. 7). If the split into heptamers is an obligatory step in the functional reaction cycle of the chaperonin, parent chaperonin should be formed as a result of a heptamer exchange reaction. Two explanations are possible: (a) split reaction occurs only once before the chaperonin starts the first turnover of the functional cycle; or (b) split reaction occurs in each of the reaction cycles, but reassociation always happens to heterologous combinations (GroEL₇-Tcpn60₇) rather than homologous combinations (GroEL₇-GroEL₇ or Tcpn60₇-Tcpn60₇), thus producing the hybrid again. However, the latter possibility is unlikely, if not impossible, because the parent GroEL₁₄ is not regenerated from the hybrid GroEL₇·MR1-GroEL₇ in the presence of ATP/K⁺ (27), and it is not easy to assume that wild-type GroEL₇ has a much higher affinity to mutant MR1-GroEL₇ than it does to wild-type GroEL₇.

A requirement for high concentrations of ATP and K⁺ is also contradictory to the notion that the split is one of obligatory steps in the chaperonin functional cycle. Steady-state ATPase activity of GroEL₁₄ is inversely dependent on K⁺; it is saturated at ~5 mM K⁺ in the presence of 50 µM ATP and at ~300 mM K⁺ in the presence of 2 µM ATP (26, 30). On the contrary, hybrids were formed only when concentrations of both K⁺ and ATP were high, and the yield of hybrids was saturated at ~300 mM K⁺ and ~300 µM ATP (Fig. 1, C and D). If either one of the concentrations was reduced, the yield of hybrids decreased, and no hybrid formation was observed at 5 mM K⁺/1 mM ATP or at 300 mM K⁺/2 µM ATP. Therefore, the requirement of K⁺ for hybrid formation is a different phenomenon than the K⁺ requirement for steady-state ATPase activity. Although the occurrence of the heptamer exchange reaction has been established, understanding of its functional and physiological significance awaits further study.