In vivo and in vitro function of GroEL mutants with impaired allosteric properties.

Escherichia coli cells that produce only plasmid-encoded wild-type or mutant GroEL were generated by bacteriophage P1 transduction. Effects of mutations that affect the allosteric properties of GroEL were characterized in vivo. Cells containing only GroEL(R197A), which has reduced intra-ring positive cooperativity and inter-ring negative cooperativity in ATP binding, grow poorly upon a temperature shift from 25 to 42 degrees C. This strain supports the growth of phages T4 and T5 but not phage lambda and produces light at 28 degrees C when transformed with a second plasmid containing the lux operon. In contrast, cells containing only GroEL(R13G, A126V) which lacks negative cooperativity between rings but has intact intra-ring positive cooperativity grow normally and support phage growth but do not produce light at 28 degrees C. In vitro refolding of luciferase in the presence of this mutant is found to be less efficient compared with wild-type GroEL or other mutants tested. Our results show that allostery in GroEL is important in vivo in a manner that depends on the physiological conditions and is protein substrate specific.

The Escherichia coli GroE system facilitates protein folding in vivo and in vitro in an ATP-dependent manner (for recent reviews see, for example, Refs. [1][2][3][4]. It is composed of GroEL, an oligomer of 14 identical subunits that form two heptameric rings, stacked back-to-back, with 7-fold symmetry and a cavity at each end (5), and its helper-protein GroES which is a sevenmembered ring of identical subunits (6). GroEL has 14 ATPbinding sites and a weak K ϩ -dependent (7) ATPase activity. It undergoes ATP-induced conformational changes (8) that are reflected in binding of ATP with intra-ring positive cooperativity (9 -11) and inter-ring negative cooperativity (12,13).
Coupling between protein folding and allostery in the GroE system has recently been demonstrated in vitro (14). The importance of the allosteric properties of GroEL for its function in vivo remains, however, unclear. It has been questioned due to (i) the fact that the allosteric transitions of GroEL take place in vitro at subphysiological (micromolar) concentrations of ATP (13), and (ii) the finding that the apical domain of GroEL, which is devoid of ATPase activity, is active in vivo (15). More recently, it has been demonstrated that the oligomeric structure of GroEL is required for biological activity because of the need for an intact cavity (16,17). Evidence for the importance of the oligomeric structure of GroEL for activity in vivo owing to the requirement for proper allosteric communication within and between rings has, however, not been reported. We decided to begin addressing this issue by generating E. coli strains that express only plasmid-derived GroEL which is either wild type (as a control) or mutant with modified allosteric properties. The following GroEL mutants with different altered allosteric properties were chosen for analysis as follows: (i) GroEL(K4E) with disrupted inter-subunit contacts (18,19); (ii) GroEL(R13G, A126V) with intact positive cooperativity and disrupted negative cooperativity (20); (iii) GroEL(R197A) with strongly diminished positive cooperativity and weakened negative cooperativity (12); (iv) GroEL(E409A, R501A) with increased positive and slightly weakened negative cooperativity (21), and (v) GroEL(R501A) with weakened positive cooperativity and disrupted negative cooperativity (21). To date, there have been very few studies on the in vivo consequences of mutations in GroEL known to modify its properties in vitro (see, for example, Ref. 22). Herein, we show that such mutations affect the function of GroEL in vivo (in cells that lack background chromosomal wild-type GroEL) and demonstrate that the effects depend on the physiological conditions and are protein substrate-specific.

EXPERIMENTAL PROCEDURES
Materials-Molecular biology reagents were purchased from Roche Molecular Biochemicals unless otherwise stated. The synthetic autoinducer N-(3-oxyhexanoyl) homoserine lactone was kindly provided by A. Eberhard, Ithaca College, NY. All other reagents were obtained from Sigma or Aldrich. GroEL and GroES were purified as described (23).
Subcloning-The groEL gene was amplified from the pOA plasmid (24) using Pwo DNA polymerase and the following oligonucleotides: FOR, 5Ј-GCG AAT TCA TCC GCG CAC GAC ACT G-3Ј; BACK, 5Ј-TGT AAA ACG ACG GCC AGT-3Ј. The FOR oligonucleotide is complementary to the region between GroEL and GroES except for the first 8 nucleotides that contain an EcoRI restriction site (underlined). The BACK oligonucleotide is complementary to the region of the plasmid flanking the 3Ј end of the gene. The reactions were carried out using Pwo DNA polymerase in 10 mM Tris-HCl buffer (pH 8.85 at 20°C) containing 25 mM KCl, 5 mM (NH 4 ) 2 SO 4 , 2 mM MgSO 4 , 0.8 mM of each dNTP, 140 pmol of each primer, and 0.5 g of template DNA (containing the wild-type or mutant groEL gene). The PCR 1 cycle consisted of 3 min at 94°C followed by 30 cycles of 1 min at 94°C, 1 min at 58°C, and 2 min at 72°C and, finally, 10 additional min at 72°C. The 1.8-kilobase pair PCR products that contained the groEL gene were purified, digested by HindIII and EcoRI, heated for 5 min at 45°C, repurified, and ligated overnight on ice to the plasmid pBAD30 (25) which was previously digested with the same enzymes. The groEL genes subcloned into the pBAD vector were fully sequenced. The pBAD vector contains the arabinose P BAD promoter which is induced by L-arabinose and repressed by glucose. This vector confers ampicillin resistance and has a pA-* This work was supported by the Israel Science Foundation administered by The Israel Academy of Sciences and Humanities and the MINERVA Foundation, Germany. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
A 12-kilobase pair SalI restriction fragment of the plasmid pBTK5 (26) containing the complete lux operon of Vibrio fischeri (27) was subcloned into pALTER-1 (Promega) previously digested with the same enzyme. This vector confers tetracycline resistance and has a ColE1 origin of replication that is compatible with the pBADEL vectors.
Creation of E. coli Strains with a Deletion of the Chromosomal groEL Gene-The chromosomal groEL gene of E. coli TG1 cells containing pBADEL(mutant type) was replaced with the nptII gene that confers kanamycin resistance by P1 transduction using AI90/pBADEL as a donor strain (28). The AI90/pBADEL strain is able to grow because the plasmid pBADEL expresses wild-type GroEL under control of the arabinose promoter (25). The transduction was performed by adding 0.1 ml of P1 phage (10 9 pfu/ml) to 0.1 ml of TG1/pBADEL(mutant type) cells that were grown overnight in 5 ml of LB medium containing 50 g/ml ampicillin and then spun down in a microcentrifuge (6000 rpm for 10 min), resuspended in 5 ml of 0.1 M MgSO 4 and 5 mM CaCl 2 , and gently shaken at 37°C for 15 min. Following incubation for 20 min at 37°C, 0.2 ml of 0.9 M trisodium citrate were added, and the cells were then spun down as above. The cells were gently resuspended in 1 ml of LB medium, shaken for 3 h at 37°C, and then centrifuged (6000 rpm for 10 min). The supernatant was removed completely, and the cells were resuspended in 100 l of LB medium and plated on LB plates containing 10 g/ml kanamycin, 50 g/ml ampicillin, and 0.2% L-arabinose. The colonies that appeared were streaked on LB plates containing 50 g/ml kanamycin, 50 g/ml ampicillin, and 0.2% L-arabinose. The replacement of the groEL gene by the nptII gene was further confirmed by the ability of the cells to grow in the presence of L-arabinose but not glucose and by PCR as described (28). By using this methodology, we have created strains that contain only pBADEL-encoded wild-type GroEL or one of the five following mutants: (i) Lys-4 3 Glu; (ii) Arg-197 3 Ala; (iii) Arg-501 3 Ala; (iv) Glu-409 3 Ala; Arg-501 3 Ala; and (v) Arg-13 3 Gly; Ala-126 3 Val. The source of GroES is endogenous under control of the original GroE promoter. The strains are designated TG1⌬EL/pBADEL(mutant type).
Cell Growth-Overnight cultures grown at 25°C were diluted 1:100 in 10 ml of LB medium containing 50 g/ml ampicillin, 50 g/ml kanamycin, and 0.02% arabinose in 100-ml flasks. The temperature was shifted to 42°C after 2.5 h. Cell density was determined at different time intervals by measuring the optical density at 560 nm.
Phage Infection-TG1⌬EL/pBADEL cells containing only plasmidencoded wild-type or mutant GroEL were tested for their ability to support the growth of phages (b 2 CI), T5, and T4(D 0 ). The different strains were grown at 37°C in 2 ml of LB medium containing 50 g/ml ampicillin, 50 g/ml kanamycin, and 0.2% arabinose until stationary phase. Cells (300 l) were mixed with 3 ml of molten soft agar (0.8% Bacto-agar) at 48°C without or with arabinose at a final concentration of 0.2% and spread on LB plates containing 50 g/ml ampicillin and 50 g/ml kanamycin to create a lawn. 2 l of phages (2 ϫ 10 9 pfu/ml), T5 (4 ϫ 10 10 pfu/ml), and T4 (2 ϫ 10 9 pfu/ml) were spotted onto the lawn. After overnight incubation at 25 or 37°C clear areas appeared in some cases indicating phage growth.
Measurements of Bioluminescence-The different TG1⌬EL/pBADEL strains containing the lux operon of V. fischeri were grown overnight with shaking at 30°C in NZCYM broth containing 50 g/ml kanamycin, 50 g/ml ampicillin, 5 g/ml tetracycline, and unless otherwise stated, 0.02% arabinose. The overnight cultures were then transferred to 20 or 28°C, and growth was continued in the presence or absence of 50 ng/ml of the synthetic autoinducer N-(3-oxyhexanoyl) homoserine lactone. At different times, 100 l of the cultures were removed, and their luminescence was measured using a Microtox model 2055 luminometer. The different TG1⌬EL/pBADEL strains were also streaked on NZCYM plates containing the appropriate antibiotics and incubated for 20 h at 30°C. Luminescence started to develop upon transfer to 28°C.
GroE-assisted Reactivation of V. fischeri Luciferase ␣ and ␤ Subunits-Lyophilized luciferase (Sigma) was dissolved in buffer A containing 100 mM Hepes, 100 mM KCl, 100 mM MgCl 2 , and 4 mM dithiothreitol (pH 7.6). The luciferase was denatured by a 4-fold dilution with a solution containing final concentrations of 6 M urea, 10 mM HCl, and 4 mM dithiothreitol followed by incubation at 25°C for 30 min. Fractions were snap-frozen and stored at Ϫ80°C in siliconized test-tubes. A solution (94 l) containing wild-type or mutant GroEL (9.1 M) with or without 9.1 M GroES in buffer A with 10 mM dithiothreitol was incubated at 25°C for 2 min before adding it to 2 l of 38 mg/ml unfolded luciferase that was preincubated similarly in a siliconized test-tube. ATP at a final concentration of 20 mM was added after 5 min, and incubation at 25°C was continued for an additional 10 min before transfer to 18°C. Aliquots from the reactivation mixture were removed at different times, and luciferase activity was measured as described (29) with some modifications. A mixture of 10 l of 1 mg/ml FMN and 300 l of buffer A containing 0.05% bovine serum albumin was prepared. The FMN was reduced by adding 10 l of 0.1 mg/ml sodium dithionite. An aliquot of 10 l from the reactivation mixture was then immediately added followed by rapid addition of 300 l of 0.005% (v/v) n-decyl aldehyde that had been freshly sonicated. The solution was vortexed for 5 s, and luminescence was measured using a Lumac 3M luminometer.

RESULTS
Cell Growth-The E. coli TG1⌬EL/pBADEL cell strains containing only plasmid-encoded wild-type or mutant GroEL were tested for their ability to grow at different temperatures. No differences in the growth of the various strains were observed at 25°C in the presence of 0.02 or 0.2% arabinose. Upon a temperature shift to 37 (data not shown) or 42°C (Fig. 1), TG1⌬EL/pBADEL(R197A) cells were found to reach a lower yield. The TG1⌬EL/pBADEL strains expressing wild-type GroEL or mutants other than GroEL(R197A) were found to grow similarly.
Infection with Phages , T4, and T5-The six TG1⌬EL/ pBADEL strains containing only plasmid-encoded wild-type or mutant GroEL were tested for their ability to support the growth of phages (b 2 CI), T5, and T4(D 0 ) under different conditions (Table I). All six strains were found to support the growth of phages T4 and T5 at 25°C in the presence of 0.002 or 0.2% arabinose which induces expression of GroEL (25). Growth of phage under these conditions was found to be supported by all the strains except TG1⌬EL/pBADEL(R197A). All six strains were found to support the growth of phages T4 and T5 at 37°C in the presence of 0.2% arabinose. Support of growth of phage is observed under these conditions in the case of TG1⌬EL/pBADEL(wild-type) and all the other strains except TG1⌬EL/pBADEL(R197A) (Fig. 2). All the strains were found to support the growth of phages T4, T5, and at 37°C in the presence of 0.002% arabinose except the TG1⌬EL/ pBADEL(R197A) strain which is not viable under these conditions (Fig. 2).
Effects of Mutations in GroEL on Bioluminescence-The different TG1⌬EL/pBADEL cell strains containing the full lux operon of V. fischeri were streaked on an NZCYM plate containing 0.02% arabinose and the appropriate antibiotics and FIG. 1. Effects of temperature shift from 25 to 42°C on cell growth. TG1⌬EL/pBADEL strains were grown overnight at 25°C and diluted 1:100 in 10 ml of LB medium containing 50 g/ml ampicillin, 50 g/ml kanamycin, and 0.02% arabinose in 100-ml flasks. The temperature was shifted to 42°C after 2.5 h as indicated by the arrow. Cell density was determined at different times by measuring the optical density at 560 nm. incubated for 20 h at 30°C. Upon transfer to 28°C, luminescence started to develop with kinetics and intensity which were found to be GroEL mutant-specific. After 2 h at 28°C, all the strains produced light, but the amount of light produced by the TG1⌬EL/pBADEL strains expressing GroEL(R501A) and GroEL(R13G, A126V) was found to be significantly lower (Fig.  3, left). It may be seen in Fig. 3 that the amount of light produced is not necessarily proportional to the cell density. TG1⌬EL/pBADEL(R501A), for example, grew well but produced little light, whereas TG1⌬EL/pBADEL(R197A) grew poorly but produced much more light (Fig. 3, right). In order to investigate this issue further, we followed the development of luminescence by each of the TG1⌬EL/pBADEL strains at different temperatures and initial cell densities (Fig. 4). All the strains grew similarly during the course of the experiments (not shown) and produced light at 20°C at a high cell density (Fig. 4B). The maximal light production of strains containing either GroEL(R501A) or GroEL(R13G, A126V) was found to be less than that of the other strains in agreement with results shown in Fig. 3. These two strains and also the strain containing GroEL(E409A, R501A) did not produce any measurable light at 28°C, whereas the strains containing GroEL(wild-type), GroEL(K4E), or GroEL(R197A) produced light at 28°C but less than at 20°C. The TG1⌬EL/pBADEL(R197A) strain produced light at 28°C at a low cell density, whereas at high cell densities it did not produce any light (Fig. 4, C and D) in agreement with the results shown in Fig. 3. The strains containing GroEL(wild-type), GroEL(K4E), or GroEL(R197A) produce more light at 20°C than the three strains containing either GroEL(R501A), GroEL(R13G, A126V), or GroEL(E409A, R501A) which at 28°C do not produce any light. All the strains were found to produce light at 28°C if the synthetic autoinducer was added, but the light production by the TG1⌬EL/ pBADEL(R501A) and TG1⌬EL/pBADEL(R13G, A126V) strains was very low (data not shown).
Effects of Expression Level of GroEL on Luminescence-The development of luminescence by the TG1⌬EL/pBADEL strains containing wild-type GroEL, GroEL(R197A), and GroEL(R501A) was followed at 28°C in the presence of different concentrations of arabinose, which induces expression of GroEL (25). In the presence of 0.01% arabinose, the TG1⌬EL/pBADEL(wild-type) and TG1⌬EL/pBADEL(R197A) strains produced light, whereas the TG1⌬EL/pBADEL(R501A) strain did not (not shown). In the presence of 0.02% arabinose, only the TG1⌬EL/pBADEL(wildtype) strain was found to produce light (Fig. 4D). In the presence of 0.2% arabinose none of these three strains produced light (not shown).
Effects of Mutations in GroEL on Reactivation of Denatured V. fischeri Luciferase ␣ and ␤ Subunits-Luciferase ␣ and ␤ subunits that had been denatured in 6 M urea and 10 mM HCl were found to fold spontaneously to a small extent upon dilution into folding buffer (Fig. 5A). In the presence of wild-type GroEL (Fig. 5A) or the different mutants, folding was completely arrested indicating that denatured ␣ and/or ␤ luciferase subunits bind tightly to GroEL. Reactivation was found to require the full GroE system and was arrested also in the presence of GroEL and ATP except in the case of the GroEL(R13G, A126V) mutant (Fig. 5B). In the presence of wild-type GroEL, GroES, and ATP, the yield of reactivation was increased about 4-fold relative to spontaneous folding (Fig.  5A). The yield of reactivation by GroEL(R13G, A126V), in the presence of GroES and ATP, was about 50% that of wild-type GroEL (Fig. 5C). The yields of reactivation by the other mutants in the presence of GroES and ATP were found to be similar to that of wild-type GroEL. DISCUSSION Bacteriophage P1 transduction was used to generate E. coli TG1 strains that express only plasmid-derived wild-type GroEL or mutants with various modified and well characterized allosteric properties. The effects of these mutations on the function of GroEL in vivo were studied using the following two assays: (i) propagation of phages , T4, and T5 and (ii) the bioluminescence of cells containing the full lux operon of V. fischeri. The GroE system was first identified by genetic studies of bacteriophage growth (30). Bacteriophages (31,32) and T5 (33) employ the host GroE system for the folding of their own proteins. Bacteriophage T4 uses host GroEL but its own co-chaperonin, Gp31, for the folding of its major capsid protein, Gp23 (34,35). Luminescence in V. fischeri cells and in E. coli cells that contain the lux genes requires the product of the luxR gene which activates transcription of the lux operon upon binding to an autoinducer (27). The GroE system is believed to facilitate the folding in vivo of the LuxR protein (36,37) and possibly also the luciferase ␣ (LuxA) and ␤ (LuxB) subunits (38). The lux operon also contains luxC, luxD, and luxE which code for enzymes required for synthesis of the long chain aldehyde luciferase substrate, luxI which codes for an enzyme involved in autoinducer synthesis and luxG which codes for a  2. Support of growth of phages , T4(D 0 ), and T5. The TG1⌬EL/pBADEL(wild-type) and TG1⌬EL/pBADEL(R197A) strains were grown in the presence of 0.2% arabinose until stationary phase and spread on LB plates with a final concentration of 0.2 or 0.002% arabinose. 2 l of phages (2 ϫ 10 9 pfu/ml), T5 (4 ϫ 10 10 pfu/ml), and T4 (2 ϫ 10 9 pfu/ml) were spotted onto the lawn. After overnight incubation at 37°C, clear areas appeared in certain cases thus indicating phage growth. In Vivo Function of GroEL Mutants 37954 protein with unknown function (27).
The functional consequences in vivo of mutations that alter the allosteric properties of GroEL are found in this study to be protein substrate-specific. Cells containing GroEL(R197A), for example, grow poorly at 37 (not shown) and 42°C (Fig. 1) and do not support growth of phage but do support growth of phages T4 and T5 (Fig. 2) and the folding of the lux operon gene products and other proteins that may be required for light production (Figs. 3 and 4). Cells containing GroEL(R501A) or GroEL(R13G, A126V), on the other hand, support the growth of phages , T4, and T5 (not shown) but not the folding at 28°C of one or more of the proteins required for bioluminescence (Figs.  3 and 4). The need for specific allosteric properties in GroEL therefore depends on the nature of the protein substrates.
The requirement for the GroE system for folding in vivo can be circumvented by changing conditions in a substrate-specific manner. All the TG1⌬EL/pBADEL strains examined in this study produce light at 20°C (Fig. 4B), thus suggesting that unassisted folding of the proteins required for bioluminescence is more efficient at this temperature as, for example, observed in the case of ribulose-bisphosphate carboxylase/oxygenase (7). All the TG1⌬EL/pBADEL strains examined in this study also produce light at 28°C when inducer is added (not shown), perhaps because the active conformation of LuxR is stabilized in its presence (27). Light production by cells containing GroEL(R501A) or GroEL(R13G, A126V) at 28°C is, however, very low in the presence of inducer. Interestingly, the TG1⌬EL/ pBADEL(R197A) strain which grows poorly at 37 and 42°C is the only one that produces relatively more light at a low cell density, perhaps because folding of proteins which inhibit luminescence, such as LexA (27), is not facilitated by GroEL(R197A). The effects in vivo of mutations in GroEL are also found to depend on its level of expression. If the GroEL mutant has low affinity for unfolded protein substrates then increasing its concentration may reduce the effect of the mutation in vivo. This may explain why the TG1⌬EL/pBADEL(R197A) strain is not viable at 37°C when the concentration of arabinose is less than 0.002% (Fig. 2). If, however, the GroEL mutant binds unfolded proteins but does not release them (i.e. it is a "trap" mutant), then lowering its concentration may diminish the effect of the mutation. For example, all the TG1⌬EL/pBADEL strains produce less light at 28°C in the presence of 0.2% arabinose than in the presence of 0.02% arabinose (not shown). Changes in the expression level of GroEL also alter the GroEL/GroES ratio in the cell which, although physiologically important, does not affect the conclusions in this study since the amount of GroEL in the different strains was determined to be the same (not shown).
An understanding of how the mutations in GroEL affect its function in vivo requires identification of the relevant substrate protein(s) whose misfolding leads to the observed phenotype and establishing the mechanism by which the mutation causes the misfolding. Here, we concentrated on trying to understand the reasons for differences in bioluminescence of the different TG1⌬EL/pBADEL strains containing the lux operon. We initially focused on LuxR as the substrate of GroEL because of reports in the literature that GroE facilitates the folding in vivo of the LuxR protein (36,37). GroEL was found to bind denatured LuxR and release it in an ATP-and GroES-dependent manner, but no differences in binding or release of LuxR by the different mutants were observed (not shown) in agreement with the small effect of the autoinducer on light production by the TG1⌬EL/pBADEL(R501A) and TG1⌬EL/pBADEL(R13G, A126V) strains. Next we analyzed GroE-assisted reactivation of denatured ␣ and ␤ luciferase subunits. The GroE system was previously shown to facilitate the in vitro folding of bacterial luciferase from Vibrio harveyi (38). The extent of reactivation of ␣ and ␤ luciferase subunits by wild-type GroEL and the various mutants, in the presence of GroES and ATP, was found to be similar except in the case of GroEL(R13G, A126V) where the yield was about 50% that of the others (Fig. 5C). GroEL(R13G, A126V) lacks negative cooperativity between rings but has intact intra-ring positive cooperativity and a k cat of ATP hydrolysis similar to that of wild-type GroEL (20). This mutant was also found to differ from wild-type GroEL and the other mutants in being able to release bound luciferase in the presence of ATP alone (Fig. 5B). It was recently shown that the ATP-bound conformation of GroEL(A126V) is similar to the GroES-bound conformation of wild-type GroEL (39) thus explaining why ATP by itself can trigger release of substrates FIG. 5. GroE-assisted reactivation of V. fischeri luciferase. Unfolded ␣ and ␤ subunits of luciferase were transferred to folding conditions in the absence of GroEL or in the presence of either wild-type GroEL, wild-type GroEL and ATP, or wild-type GroEL with GroES and ATP, and the recovery of enzyme activity was measured as a function of time (A). Similar experiments were carried out in the presence of GroEL (wild-type or the various mutants) and ATP (B) and in the presence of GroEL (wild-type or the various mutants), GroES, and ATP (C). The data in C are normalized relative to wild-type GroEL. bound to GroEL(R13G, A126V). ATP-triggered release of luciferase subunits in a conformation not yet committed to fold may contribute to the lower yield of folding in the presence of GroEL(R13G, A126V) relative to wild-type GroEL. The GroEL(A126V) mutant was also found to form symmetric 2:1 GroES-GroEL complexes (39) in which substrate-binding sites are blocked. These properties of GroEL(A126V) may explain why cells containing GroEL(R13G, A126V) do not produce light at 28°C. We do not yet know which proteins required for bioluminescence fail to fold in the presence of GroEL(R501A) and also which proteins involved in cell growth and phage propagation fail to fold in the presence of GroEL(R197A) at 25 and 37°C.
Our results suggest that mutations that perturb specific steps in the reaction cycle of GroEL are likely to be relatively more damaging. For example, cells containing only GroEL-(D398A), which is defective in ATP hydrolysis, are not viable (not shown), whereas cells containing GroEL(K4E), which tends to dissociate into monomers (18,19), exhibit a normal phenotype. This phenotype may be due, in part, to assembly into oligomeric structures in the presence of physiological concentrations of ATP (data not shown). GroEL(K4E) and also minichaperones (15) may retain some chaperoning function owing to mass action without trapping substrate proteins.
In summary, our results indicate that allosteric communication in GroEL is important for the in vivo folding of a subset of substrates under certain physiological conditions. The poor growth of TG1⌬EL/pBADEL(R197A) at 42°C and the formation of inclusion bodies in these cells (not shown) suggests that intact positive cooperativity may be of particular importance under stress conditions. Disrupted positive cooperativity reduces the shift in equilibrium under stress conditions toward the protein acceptor state of GroEL. Hartl and co-workers (40) recently identified a set of in vivo substrates of GroEL. Our results suggest that a "universal" set of protein substrates does not exist and that the set of substrates that interact with GroEL in vivo depends on the physiological conditions.