Fluorescence detection of symmetric GroEL14(GroES7)2 heterooligomers involved in protein release during the chaperonin cycle.

The GroEL14 chaperonin from Escherichia coli was labeled with 5-((((2-iodoacetyl)amino)ethyl)amino)naphthalene-1-sulfonic acid (I-AEDANS), a hydrophobic probe whose fluorescent emission is sensitive to structural changes within the protein. Increasing concentrations of ATP or adenylyl imidodiphosphate but not ADP caused two successive GroES7-dependent changes in the fluorescence intensity of AEDANS-GroEL14, corresponding to the sequential binding of two GroES7 heptamers and the formation of two types of chaperonin heterooligomers, GroEL14GroES7 and GroEL14(GroES7)2. The binding of thermally denatured malate dehydrogenase (MDH) caused a specific increase in fluorescence intensity of AEDANS-GroEL14 that allowed the direct measurement in solution at equilibrium of ATP- and GroES7-dependent protein release from the chaperonin. Structure/function analysis during the generation of ATP from ADP indicated the following sequence of events: 1) ADP-stabilized MDH-GroEL14GroES7 particles bind newly formed ATP. 2) MDH-GroEL14GroES7 particles bind a second GroES7. 3) MDH-GroEL14(GroES7)2 particles productively release MDH. 4) Released MDH completes folding. Therefore, the symmetrical GroEL14(GroES7)2 heterooligomer is an intermediate after the formation of which the protein substrate is productively released during the chaperonin-mediated protein folding cycle.

In Escherichia coli cells, chaperonins GroEL 14 and GroES 7 are implicated in the folding of proteins and in the molecular response to cellular stress (Goloubinoff et al., 1989a;Cheng et al., 1989;Frydman et al., 1994;Martin et al., 1992;Horwich et al., 1993). Under stringent in vitro conditions (Schmidt et al., 1994a), chaperonin-assisted protein-refolding requires GroEL 14 , GroES 7 , K ϩ , Mg 2ϩ , or Mn 2ϩ ions and ATP hydrolysis (Goloubinoff et al., 1989b;Viitanen et al., 1990;Diamant et al., 1995b). The mechanism by which chaperonins prevent the irreversible aggregation of stress-destabilized proteins and assist them to recover a functional three-dimensional structure remains largely unclear (Martin et al., 1991(Martin et al., , 1993Engel et al., 1995;Azem et al., 1994bAzem et al., , 1995. One approach to this question is to address the relationship between the chaperonin protein folding activity and the structure of the various GroEL-GroES heterooligomers. However, for lack of a direct method to assess the structure of chaperonin heterooligomers in active protein folding solutions, models for the chaperonin mechanism have been designed on the sole basis of indirect structural information, obtained under conditions that may not guarantee preservation of the biochemical equilibrium (Langer et al., 1992;Martin et al., 1993;Azem et al., 1994b;Chen et al., 1994;Azem et al., 1995).
Negative-stain electron microscopy (EM) 1 indicated that in the presence of ADP or low ATP or AMP-PNP concentrations only one GroES 7 co-chaperonin can bind a GroEL 14 core particle and form an asymmetric GroEL 14 GroES 7 complex (Langer et al., 1992;Ishii et al., 1992;Azem et al., 1994b). In contrast, in the presence of physiological pH, Mg 2ϩ , and ATP concentrations or high concentrations of AMP-PNP but not of ADP, two GroES 7 co-chaperonins can bind the GroEL 14 cylinder and form a symmetric GroEL 14 (GroES 7 ) 2 particle (Azem et al., 1994b). Variable percentages of symmetric GroEL 14 (GroES 7 ) 2 particle were described under various conditions of pH, low protein, and high Mg 2ϩ concentrations (Schmidt et al., 1994b;Llorca et al., 1994;Harris et al., 1994;Engel et al., 1995;Llorca et al., 1996).
Although the asymmetric GroEL 14 GroES 7 particle has been considered to be the only heterooligomer of functional significance to the protein folding cycle (Langer et al., 1992;Martin et al., 1993;Engel et al., 1995;Hayer-Hartl et al., 1995), solutions populated by over 90% symmetrical GroEL 14 (GroES 7 ) 2 particles were observed by EM and SDS gels of cross-linked chaperonins under conditions that also supported maximal rates of protein folding (Azem et al., 1994b). Moreover, the rate and increased efficiency of the protein refolding reaction was found proportional to the amount of GroEL 14 (GroES 7 ) 2 particles in a chaperonin solution (Azem et al., 1995;Diamant et al., 1995b). Whereas the binding of a nonnative protein can possibly increase the amount of GroEL 14 (GroES 7 ) 2 particles detected by EM (Llorca et al., 1996), kinetic analysis of thermophilic chaperonins indicated that the release of a bound protein from asymmetric GroEL 14 GroES 7 particles depends upon the binding of a second GroES 7 heptamer (Todd et al., 1995).
However, the validity of structure/function correlations depends on whether EM and SDS gels of cross-linked chaperonins reflect the true distribution between the various heterooligomers in active chaperonin solutions at equilibrium. The possible artifactual recruitment of a second GroES 7 by GroEL 14 GroES 7 during glutaraldehyde cross-linking (Azem et al., 1994b;Diamant et al., 1995b) has been been ruled out by nearly identical results obtained from EM analysis of crosslinked/noncross-linked chaperonins (Azem et al., 1995). However, it remains to be established whether in chaperonin samples that were not previously stabilized by cross-linking (Schmidt et al., 1994b;Llorca et al., 1994;Harris et al., 1994) or incompletely stabilized by partial cross-linking in the presence of Tris buffer (Engel et al., 1995;Llorca et al., 1996) the highly labile GroEL 14 (GroES 7 ) 2 was not dissociated into GroEL 14 -GroES 7 by procedures involving dilutions, freezing, gel filtration, and fixation for EM.
In contrast to EM and cross-linking, steady-state fluorescence analysis of GroEL-bound fluorescent probes can provide a means of direct co-measurement of the chaperonin structure and protein folding activity in the same solution, at equilibrium. Specific changes in the fluorescence spectrum and intensity of pyrenyl-or AEDANS-labeled GroEL 14 molecules have been demonstrated in the presence of ATP, ADP, AMP-PNP, and GroES 7 as well as upon shifting of the temperature (Jackson et al., 1993;Burston et al., 1995;Hansen and Gafni, 1994). Because neither pyrenyl-GroEL 14 nor AEDANS-GroEL 14 molecules were impaired in their ability to hydrolyze ATP, fluorescence changes were interpreted as resulting from specific structural changes in the GroEL 14 molecule, caused by specific interactions with nucleotides and GroES 7 (Jackson et al., 1993;Hansen and Gafni, 1994;Burston et al., 1995). We show here that the binding of nucleotides, divalent ions, GroES 7 , and a nonnative protein cause distinct changes in the steady-state fluorescence intensity of AEDANS-labeled GroEL 14 . Fluorescence changes in the presence of ATP demonstrated that two forms of GroEL-GroES heterooligomers coexist in active protein folding solutions at equilibrium. Experiments using AE-DANS-GroEL 14 fluorescence showed that during the protein folding cycle, productive protein release preferentially occurs after the formation of a protein-bound GroEL 14 (GroES 7 ) 2 intermediate.

EXPERIMENTAL PROCEDURES
Proteins-GroEL 14 was purified to homogeneity as in Azem et al. (1994a). GroES 7 was purified as in Todd et al. (1993) with minor modifications. Protein concentrations were determined by the Bradford protein assay (Bio-Rad) with GroEL and GroES standard solutions, whose respective concentrations were determined by total amino acid analysis. For the sake of clarity, in this work all chaperonin concentrations were expressed in terms of the individual GroES and GroEL 10.3and 57.3-kDa protomers and not of the GroES 7 and GroEL 14 oligomers. Pig heart mitochondrial malate dehydrogenase (MDH) was from Boehringer Mannheim; hexokinase and pyruvate kinase were from Sigma.
AEDANS Labeling-GroEL 14 (140 M) in 50 mM triethanolanime, pH 7.5, and 20 mM MgOAc was incubated with 3 mM I-AEDANS (Molecular Probes Europe BV) in the dark at 37°C for 60 min. AE-DANS-labeled GroEL 14 was then separated on a Superose 6B gel filtration column (Pharmacia Biotech Inc.) in 50 mM triethanolanime, pH 7.5. Under these conditions, about five AEDANS molecules were found covalently bound to each GroEL monomer, possibly to nonessential cysteine and lysine residues (Hudson and Weber, 1973). Regardless of their identity, the target residues are not essential to the chaperonin function, because after a 22-min incubation at 47°C, AEDANS-GroEL 14 was found equally as able as unlabeled GroEL 14 to prevent the thermal aggregation and promote the subsequent reactivation at 25°C of thermally denatured MDH in a strict GroES 7 -and ATP-dependent manner (not shown). AEDANS labeling did not impair the ability of GroEL 14 to form both asymmetric GroEL 14 GroES 7 and symmetric GroEL 14 -(GroES 7 ) 2 particles, as detected by glutaraldehyde cross-linking (see Figs. 4B and 6).
The spectral properties of AEDANS-GroEL 14 were as described by Hansen and Gafni (1994). Because changes in the fluorescence intensity were more sensitive than shifts in the emission spectra, fluorescence intensity was used to monitor the structural changes in GroEL 14 in the presence of nucleotides, divalent ions, GroES 7 , native, and nonnative MDH. In storage, the chaperonin activity and the fluorescence emission of AEDANS-GroEL 14 remained unchanged after 2 weeks of incubation in the dark at 4°C as a 20 -30 M solution in the presence of 50 mM triethanolanime, pH 7.5, and 2 mM MgOAc.
Fluorescence-The fluorescence of AEDANS-GroEL 14 was measured at 25°C in a Perkin-Elmer luminescence spectrometer LS50B. Excitation was at 340 nm. Detection at 477 nm was performed in a 3-ml four-sided quartz cell in the presence of active stirring. With the exception of the kinetic experiments described in Figs. 4 and 5, the steadystate fluorescence of each condition was first equilibrated for 3 min in the dark and then measured for 50 s and averaged.
Enzymatic Assay-The activity of mMDH was assayed at 25°C in 150 mM potassium phosphate buffer, pH 7.5, 10 mM dithiothreitol, 0.5 mM oxaloacetate, and 0.28 mM NADH (Sigma). The time-dependent oxidation of NADH by mMDH was monitored at 340 nm (as in Diamant et al., 1995b).
Protein Cross-linking-Chaperonin heterooligomers were crosslinked as described in Azem et al. (1994bAzem et al. ( , 1995. AEDANS-GroEL 14 and GroES 7 was cross-linked in the presence of 22 mM glutaraldehyde, first for 30 s at 25°C and then for 7 min at 37°C. SDS gel electrophoresis of cross-linked proteins was carried out in tubes containing 3.3% polyacrylamide in phosphate buffer according to Azem et al. (1994aAzem et al. ( , 1994bAzem et al. ( , 1995. Gels were stained by Coomassie Brilliant Blue R-250 (Sigma).

RESULTS
The Effect of Nucleotides-In the presence of an ATP regeneration system, increasing concentrations of ATP caused a decrease in the intensity of the steady-state fluorescence of AEDANS-GroEL 14 . This effect was specific, because 1 mM ATP caused less than 1% change in the fluorescence intensity of free I-AEDANS or of the same amount of AEDANS-labeled malate dehydrogenase as controls (not shown). When AEDANS-GroEL 14 alone was exposed to increasing concentrations of ATP, the change in the fluorescence intensity was monophasic. A linear representation of the data in Fig. 1A clearly shows that the fluorescence signal saturates around 14.3% ( Fig. 1A) with an EC 50 of 56 M.
When AEDANS-GroEL 14 in the presence of a molar excess of GroES 7 was exposed to increasing concentrations of ATP, the decrease in the fluorescence intensity was sharper and biphasic with a first EC 50 of 5.5 M and a second approximate EC 50 of 60 Ϯ 20 M (Fig. 1, A and B). When, in a control experiment, GroES 7 was first chemically modified by a pretreatment with glutaraldehyde (as in Fig. 4B), repurified, and then used as in Fig. 1A, the ATP-dependent decrease in the fluorescence intensity of GroEL 14 was monophasic, as if GroES 7 was absent from the solution (not shown). Thus, the two-step ATP-dependent transition in the fluorescence intensity of GroEL 14 is specific to the presence of unmodified GroES 7 in the solution.
Similarly to ATP, increasing concentrations of the ATP analog AMP-PNP caused a monophasic change in the fluorescence intensity of AEDANS-GroEL 14 alone, with an EC 50 value of about 100 M. In the presence of GroES 7 , increasing AMP-PNP concentrations above 160 M caused a biphasic change in the GroEL 14 fluorescence with a first EC 50 of about 315 M and a second EC 50 above 2 mM (Fig. 1, C and D).
In contrast to ATP and AMP-PNP, increasing concentrations of ADP caused a monophasic change in the fluorescence intensity of AEDANS-GroEL 14 , both in the absence and the presence of GroES 7 , with similar EC 50 values of 136 and 113 M, respectively (Fig. 1E). Thus, in the presence of GroES 7 , ATP, and AMP-PNP share the same ability to cause a biphasic change, as opposed to ADP, which causes a monophasic change in the fluorescence intensity of AEDANS-GroEL 14 (Fig. 1F).
The ATP-dependent two-step effect of GroES 7 on the fluorescence of AEDANS-GroEL 14 was compared with the ATP-dependent formation of GroEL 14 GroES 7 and GroEL 14 (GroES 7 ) 2 particles in chaperonin solutions, as previously measured under the same conditions by EM and SDS gels of cross-linked chaperonins (Azem et al. (1995); see Fig. 6A). In Fig. 1B, a clear correlation was observed between the ATP-dependent formation of asymmetric GroEL 14 GroES 7 particles and the first ATP-dependent fluorescence transition with EC 50 values of 4.5 and 5.5 M, respectively. Moreover, the chaperonin solution was saturated with asymmetric GroEL 14 GroES 7 particles in the presence of the same ATP concentration (20 M), which also saturated the first GroES-dependent fluorescence transition. Finally, the ATP-dependent formation of symmetric GroEL 14 (GroES 7 ) 2 particles correlated with the second ATPdependent fluorescence transition, displaying approximate EC 50 values of 62 M and 60 Ϯ 20 M, respectively (Fig. 1B). Thus, changes in the fluorescence intensity of AEDANS-GroEL 14 reflect the distribution of the two chaperonin heterooligomers in solution.
The Effect of GroES 7 -In the presence of a saturating amount of ADP when only one GroES 7 can asymmetrically bind GroEL 14 (Azem et al., 1994b;Schmidt et al., 1994b), increasing concentrations of GroES 7 decreased the fluorescence intensity of AEDANS-GroEL 14 with an EC 50 of 0.91 (GroES/GroEL molar ratio) and a maximal fluorescence change of 7.1% from the GroES-less control (Fig. 2). In contrast, in the presence of saturating amounts of ATP and Mg 2ϩ , where two GroES 7 can symmetrically bind GroEL 14 (Azem et al. (1994b(Azem et al. ( , 1995; see Fig. 6A), increasing concentrations of GroES 7 decreased the fluorescence intensity of AEDANS-GroEL 14 with an EC 50 of 0.77 and a maximal fluorescence change of 14.1% from the GroES-less control (Fig. 2). Thus, in the presence of ATP, the net effect of GroES 7 on the fluorescence intensity of AEDANS-GroEL 14 was twice as high as that in the presence of ADP, confirming that fluorescence can distinguish between the binding of a single GroES 7 to GroEL 14 and the binding of a second GroES 7 to GroEL 14 GroES 7 .
Mn 2ϩ ions increase the affinity of GroES for GroEL (Diamant et al., 1995b). The presence of 2 mM Mn 2ϩ , in addition to the preexisting 20 mM Mg 2ϩ and ATP, increased the effect of GroES 7 on the GroEL 14 -fluorescence and reduced the EC 50 from 0.77 to 0.51 (Fig. 2). Thus, when the GroES/GroEL molar ratio was 0.51 and the chaperonin solution was accordingly saturated with asymmetric GroEL 14 GroES 7 particles (Diamant et al., 1995b), the GroES 7 -dependent fluorescence signal was half-saturated. This demonstrates that GroES 7 does not only interact with GroEL 14 to form GroEL 14 GroES 7 particles but also interacts to the same extent with GroEL 14 GroES 7 to form symmetrical GroEL 14 (GroES 7 ) 2 particles.
The Effect of Mg 2ϩ Ions-In the presence of saturating amounts of GroES 7 and ATP, increasing concentrations of Mg 2ϩ decreased the fluorescence intensity of GroEL 14 (Fig. 3A), as expected from the cumulative effects of free and ATP-bound Mg 2ϩ ions on the structure and activity of GroEL 14 , as well as on the affinity of GroES 7 for both GroEL 14 and GroEL 14 GroES 7 particles (Azem et al., 1994a;Diamant et al., 1995aDiamant et al., , 1995b. However, at variance with previous evidence from EM (Engel et al., 1995), estimated physiological amounts of 2.5 mM of free Mg 2ϩ ions (Alatossava et al., 1985), sufficed to produce the fluorescence signal of a chaperonin solution saturated with GroEL 14 (GroES 7 ) 2 particles (Fig. 3B).
The Effect of ADP Conversion into ATP-Chemical crosslinking has shown that even in the presence of a saturating concentration of ATP, ADP can affect the stability of GroEL 14 (GroES 7 ) 2 particles and displace the equilibrium in favor of asymmetric GroEL 14 GroES 7 particles (Azem et al., 1995). In the absence of GroES 7 , the slow conversion of ADP into ATP caused an immediate change in GroEL 14 fluorescence (Fig. 4A), as expected from the higher affinity of the newly formed ATP, replacing ADP in the active sites of GroEL 14 . When after 10 min an excess of GroES 7 was added, the decreased fluorescence reached 25% (not shown), which is the signal from a solution saturated with ATP-stabilized GroEL 14 (GroES 7 ) 2 particles (see Fig. 1A). In contrast, when GroES 7 was present, the same rate of ATP synthesis did not cause the same immediate decrease in the fluorescence intensity of GroEL 14 . Only after a delay of about 3 min did the fluorescence signal shift from that of a majority of ADP-GroEL 14 GroES 7 particles into that of a majority of ATP-GroEL 14 (GroES 7 ) 2 particles. The established higher affinity of ADP for GroEL 14 GroES 7 , as compared with GroEL 14 particles (Todd et al., 1994) and the consequent delay in the binding of ATP and of the second GroES 7 can explain the observed delay in the fluorescence change (Fig. 4A). It should be noted that because after 10 min the maximal decrease in fluorescence was less than 25%, a minority of GroEL 14 GroES 7 particles were likely to be also present in the chaperonin solution. This was confirmed by limited glutaraldehyde cross-linking (Fig. 4B). In the absence of nucleotide, GroES 7 did not bind GroEL 14 , as revealed by the presence of the final GroEL 14 cross-linking species and of a single transient GroEL 7 species, without a transient GroEL 7 GroES 7 species (Fig. 4B, lane 1). In contrast, 40 s after the addition of ADP, the chaperonin solution was populated by a mixture of asymmetric GroEL 14 GroES 7 and GroEL 14 particles (Fig. 4B, lane 2). The asymmetric GroEL 14 GroES 7 particle steadily remained the major species in the solution during at least 3 min after the initiation of ATP synthesis, as demonstrated by the equal amounts of GroEL 7 GroES 7 and GroEL 7 transient species (Fig. 4B, lanes  3-6). Beyond 3 min, symmetric GroEL 14 (GroES 7 ) 2 particles started to accumulate at the expense of asymmetric GroEL 14 GroES 7 particles, as reflected by the formation of GroEL 7 GroES 7 species in excess, and at the expense of the GroEL 7 species (Fig. 4B, lanes 7-9). Thus, the two successive fluorescence changes, first slow and then rapid, during ATP synthesis in the presence of ADP and GroES 7 (Fig. 4A) coincided with a maintained maximal level of asymmetric GroEL 14 GroES 7 particles, followed by the formation of symmetric GroEL 14 (GroES 7 ) 2 particles at the expense of the GroEL 14 GroES 7 particles in the solution (Fig. 4B, lanes 7-9).
The Effect of a Nonnative Protein-The fluorescence intensity of GroEL 14 after exposure to 47°C for 22 min and then to 25°C was the same as that of GroEL 14 incubated for 22 min at 25°C (not shown), indicating that the heat treatment did not irreversibly affect the structure of the chaperonin. However, when native MDH was added to GroEL 14 at 47°C, the steadystate fluorescence intensity of GroEL 14 was increased within 5 min by 4.6% (Fig. 5A, inset). After 22 min at 47°C and cooling to 25°C, the steady-state fluorescence intensity of GroEL 14 was increased by ϳ7%, as compared with a corresponding treatment at a constant 25°C (Fig. 5A) or with the same treatment at 47°C in absence of MDH (not shown). The residual MDH activity, which was less than 1.5%, did not increase during more than 60 min after the heat treatment, as long as both GroES 7 and ATP were not supplied (not shown), indicating that unbound MDH was irreversibly aggregated. The addition of a saturating concentration of ADP did not reduce the net fluorescence signal for bound MDH. The subsequent addition of GroES 7 did reduce by ϳ18% the signal for bound MDH; however, no MDH reactivation was observed in the solution. Only after the initiation of ATP synthesis by pyruvate kinase and a delay of about 2-3 min did the net fluorescence signal of bound MDH start to subside with a t1 ⁄2 of 2.5 min (Fig. 5B). The fact that the same delay was necessary for GroEL 14 (GroES 7 ) 2 formation as for protein release suggests that the formation of a transient protein-bound GroEL 14 (GroES 7 ) 2 ternary complex is rate-determining for the protein folding cycle. Also after the  Fig. 1 for 200 s after the addition of 1 mM ADP in the absence (E) or the presence of 14 M GroES 7 (q). Then, to convert the ADP into ATP, a limiting amount of pyruvate kinase (PK, 2.5 ng/ml) was added, and the change in fluorescence was measured for 10 min. B, time-dependent analysis of the formation of ATP-GroEL 14 (GroES 7 ) 2 particles from ADP-GroEL 14 GroES 7 particles as in A. Chaperonins were crosslinked with glutaraldehyde, before (lane 1) or 40 s after addition of ADP (lane 2) or 40, 80, 120, 180, 260, 420, or 600 s after the addition of pyruvate kinase (lanes 3-9, respectively). Cross-linked samples were separated by SDS gel electrophoresis as in Azem et al. (1995). same delay, in parallel to the protein release, albeit at a slower rate, MDH activity was recovered (t1 ⁄2 of 9 min) (Fig. 5B). Thus, consistent with similar observations using RubisCO and Rhodanese (Todd et al., 1994;Weissman et al., 1994), the difference between the rate of protein release and that of protein folding indicate that the final steps of folding and assembly of a protein such as MDH, occur after it has been completely released from the chaperonin.
Cross-linking and fluorescence confirmed that transient MDH-GroEL 14 (GroES 7 ) 2 complexes form prior to the productive release of the protein. Hence, after 1 min of incubation at 37°C with increasing concentrations of ATP, both GroES 7 bind the MDH-GroEL 14 complex equally well as the free GroEL 14 core particle (Fig. 6, A and B). In parallel, fluorescence revealed that within that time, all the nonnative MDH still remained associated to the chaperonin (not shown), even when ATP was saturating (0.25 mM). Thus, the transient protein-GroEL 14 (GroES 7 ) 2 complex can accumulate and become the major species in the solution, prior to the initiation of productive protein release. DISCUSSION We have shown that the interaction of Mg 2ϩ ions, nucleotides, GroES 7 , and a nonnative protein caused distinct and specific changes in the quantum yield of the steady-state fluorescence of AEDANS-GroEL 14 . Thus, in the presence of increasing concentrations of ATP or AMP-PNP, the interaction of GroES 7 caused a biphasic change, whereas in the presence of ADP, GroES 7 caused a monophasic change in the AEDANS-GroEL 14 fluorescence. Similarly, SDS gels and EM of crosslinked chaperonins have previously shown that in the presence of ATP or AMP-PNP, two GroES 7 can bind both ends of GroEL 14 , whereas in the presence of ADP, GroES 7 can bind only one end of the GroEL 14 core chaperonin (Azem et al., 1994b;Schmidt et al., 1994b;Azem et al., 1995). Data from fluorescence experiments not only concurred qualitatively but also quantitatively with the cross-linking and EM data. Hence, the two fluorescence changes occurred at the same ATP concentrations previously reported for the successive formation of asymmetrical GroEL 14 GroES 7 as of symmetrical GroEL 14 (GroES 7 ) 2 particles in the solution (Figs. 1C and 6A;Azem et al. (1995)). We conclude from this precise match that the three methods faithfully reflect the steady-state equilibrium between GroEL 14 GroES 7 and GroEL 14 (GroES 7 ) 2 particles in the functional chaperonin solutions. Moreover, all three methods describe chaperonin solutions populated by a majority of symmetrical GroEL 14 (GroES 7 ) 2 particles under conditions that also support maximal rates of protein folding (Azem et al., 1994(Azem et al., , 1995Diamant et al., 1995b). Remarkably, these conditions of 2.5 mM Mg 2ϩ and pH 7.5 contradict a previous suggestion that symmetrical GroEL 14 (GroES 7 ) 2 particles can form in significant amounts only under nonphysiological pH 8.0 and high (50 mM) Mg 2ϩ concentrations (Engel et al., 1995).
Symmetrical GroEL 14 (GroES 7 ) 2 particles often remained undetected by EM and even by fluorescence analysis. Thus, in the presence of low Mg 2ϩ concentrations, extensive chaperonin dilutions, the absence of an ATP regeneration system, or without performing a prior thorough cross-linking of GroES to GroEL, GroEL 14 (GroES 7 ) 2 particles were not reported (Langer et al., 1992;Martin et al., 1993;Ishii et al., 1992Ishii et al., , 1994. The lack of a specific fluorescence signal for GroEL 14 (GroES 7 ) 2 particles in a similar fluorescence analysis of pyrenyl-GroEL 14 can now be attributed to suboptimal conditions, such as low ATP Then, at 37°C, ATP was added (0, 10, 20, 40, 80, 120, 180, or 250 M in lanes 1-8, respectively). After 1 min, during which GroES was allowed to bind, chaperonins were cross-linked with glutaraldehyde and separated on SDS-polyacrylamide gel electrophoresis as described in the legend to Fig. 4B. concentrations (10 -40 M), the absence of an ATP regeneration system, and nonsaturating amounts of GroES (Burston et al., 1995). More recently, varying minor amounts of symmetrical GroEL 14 (GroES 7 ) 2 particles have been reported by EM, particularity under the stabilizing effect of unphysiologically high concentrations of Mg 2ϩ (30 -50 mM) (Schmidt et al., 1994b;Llorca et al., 1994;Harris et al., 1994;Engel et al., 1995;Llorca et al., 1996). In contrast, when GroES was fully cross-linked to GroEL prior to EM preparations, as opposed to partially chemically modified in Tris buffer (Engel et al., 1995;Llorca et al., 1996), fully active chaperonin solutions were found saturated with symmetrical GroEL 14 (GroES 7 ) 2 particles (Azem et al., 1994b), whose relative amounts precisely correlated with the rate and increased efficiency of the protein folding reaction (Diamant et al., 1995b;Azem et al., 1995).
The Role of the Symmetric Complex-Although gel filtration, equilibrium dialysis, EM, and surface plasmon resonance analysis produced conflicting evidence suggesting that the binding of a second GroES 7 and of a nonnative protein are mutually exclusive (Engel et al., 1995;Hayer-Hartl et al., 1995), fluorescence now confirmed that a strong correlation exists between protein folding activity and the presence of symmetric GroEL 14 (GroES 7 ) 2 particles in the solution (Azem et al., 1995;Diamant et al., 1995b). Moreover, fully corroborating previous evidence using cross-linking, EM, gel filtration, and kinetic analysis (Azem et al., 1994b(Azem et al., , 1995Diamant et al., 1995b;Llorca et al., 1996;Todd et al., 1995), fluorescence, and crosslinking (Fig. 6) showed here that transient MDH-GroEL 14 (GroES 7 ) 2 complexes can form, accumulate, and even become the major species of an optimally active protein folding solution of chaperonins.
How Can Proteins Be Released from GroEL14(GroES7)2 Particles?-Protein release has been previously described from ADP-stabilized asymmetric GroEL 14 GroES 7 particles upon activation with ATP (Langer et al., 1992;Martin et al., 1993;Todd et al., 1994;Weissman et al., 1995). Remarkably, kinetic studies showed that productive protein release preferentially occurs from GroEL 14 GroES 7 particles in which the protein is bound in a cis conformation, i.e., on the same GroEL 7 toroid as the bound GroES 7 (Weissman et al., 1995). It should be noted that structurally speaking, a GroEL 14 (GroES 7 ) 2 particle may contain as many as two protein binding sites in a cis configuration (Todd et al., 1995). Although it is not clear if the cis-bound protein is released directly from the asymmetric GroEL 14 GroES 7 particle or from a transiently forming GroEL 14 (GroES 7 ) 2 particle, this nevertheless demonstrates that during the folding cycle, a bound protein must reside on the same side as GroES 7 before it can be productively released (Bochkareva and Girshovich, 1992;Weissman et al., 1995;Todd et al., 1995).
Similar to the cis MDH-GroEL 14 GroES 7 complex (Weissman et al., 1995(Weissman et al., , 1996, we show here that MDH-GroEL 14 (GroES 7 ) 2 complex can serve as an efficient species from which productive protein folding and release can occur. Although the sharp decrease in the chaperonin affinity for a bound protein appears to be a consequence of the binding of the second GroES 7 , it is likely that the minority of MDH-GroEL 14 (GroES 7 ) 2 molecules that release MDH at a given time do so after of the dissociation of the capping GroES 7 from the complex (Weissman et al., 1996).
A model for the protein folding cycle of chaperonins is proposed in Fig. 7. Due to the high affinity of GroES 7 for the GroEL 14 core oligomer (Azem et al., 1994b) and the consequent stability of the GroEL 14 GroES 7 heterooligomer, active exchange of GroES 7 during protein folding is more likely to occur between the labile GroEL 14 (GroES 7 ) 2 and stable GroEL 14 -GroES 7 particles than between the stable GroEL 14 GroES 7 and GroEL 14 particles. For this reason, we expect GroEL 14 to be a rare, nonobligatory, intermediate of the protein folding cycle. Whereas the nonnative protein may preferentially bind a highaffinity asymmetrical GroEL 14 GroES 7 particles (Todd et al., 1994;Engel et al., 1995;Weissman et al., 1995), it may dissociate after it has become capped by GroES 7 within a protein-GroEL 14 (GroES 7 ) 2 ternary complex. ATP hydrolysis and the dissociation of GroES 7 from GroEL 14 (GroES 7 ) 2 has been suggested to be rate-determining for the chaperonin ATPase cycle and, consequently, of the protein folding cycle (Todd et al., 1994) (thick arrow). This could account for the observed accumulation of protein-GroEL 14 (GroES 7 ) 2 ternary complexes, despite their transient nature, prior to the protein release step. If incompletely folded, the released protein can rebind the minority of GroEL 14 GroES 7 species that was formed and undergo another folding cycle. ADP must leave the chaperonin before ATP and GroES 7 rebind to the MDH-GroEL 14 GroES 7 species. Given the cellular concentrations and ratios of GroEL and GroES (Lorimer, 1996) and the concentrations ATP and Mg 2ϩ (Alatossava et al., 1985), it is likely that the symmetrical GroEL 14 (GroES 7 ) 2 particle plays an essential role in chaperonin-mediated protein folding in vivo as well as in vitro.