Both the Escherichia coli Chaperone Systems, GroEL/GroES and DnaKIDnaJIGrpE, Can Reactivate Heat-treated RNA Polymerase DIFFERENT MECHANISMS FOR THE SAME ACTIVITY*

In this work we show that the GroEL (Hsp6O equivalent) chaperone protein can protect purified Esche- richia coli RNA polymerase (RNAP) holoenzyme from heat inactivation better than the DnaK (Hsp70 equiv- alent) chaperone can. In this protection reaction, the GroES protein is not essential, but its presence reduces the amount of GroEL required. GroEL and GroES can also reactivate heat-inactivated RNAP in the presence of ATP. The mutant GroEL673 protein, with or with- out GroES, is incapable of reactivating heat-inacti-vated RNAP. GroEL673 can only protect RNAP, and this protecting ability is not stimulated by GroES. The mechanism by which the DnaJ and GrpE heat shock proteins contribute to DnaK’s ability to reactivate heat-inactivated RNAP GroEL673 has

throughout evolution, their eukaryotic and prokaryotic members being at least 50% identical at the amino acid sequence level (1).
The E. coli dnaK and groEL genes were originally discovered because mutations in them block the growth of bacteriophage X (3,6). In the case of dnaK, it was subsequently shown that bacteriophage growth is blocked at the level of replication, whereas in the case of groEL, growth is blocked at the level of bacteriophage morphogenesis. In addition, through genetic analyses, the dnaK and groEL gene products have been shown to perform essential bacterial functions. With respect to groEL, this function is essential at all temperatures (7). With respect to dnaK, deletion of the gene can be tolerated, but only within a narrow temperature range, and, even then, extragenic suppressors accumulate rapidly (8). ' These genes are now known to encode products that are members of a family of proteins called "molecular chaperones." The function of these proteins, first proposed by Hightower (9) and Pelham (lo), is to protect other proteins from thermal inactivation and to reactivate protein aggregates formed under stress conditions. Given the essentiality of these genes even under nonstress conditions, the cell apparently requires this protecting activity during normal growth as well as under heat shock conditions. Other studies suggest a variety of processes in which "chaperoning" activity is required to maintain proteins in an unfolded state, preventing them from premature, incorrect interactions (e.g. nascent polypeptide folding and protein export).
The various members of the DnaK and GroEL chaperone families share a number of properties. ( a ) GroEL members possess a weak ATPase activity, and DnaK members an even weaker one; ( b ) both bind to some forms of unfolded polypeptides, perhaps as these polypeptides emerge from the ribosomes or into specialized cellular compartments (11,12); and (c) members of both families have been shown to functionally interact with other E. coli HSPs, DnaJ and GrpE in the case of DnaK and GroES in the case of GroEL (4). However, differences have been observed in the way these corresponding HSPs interact with their larger partner, DnaK or GroEL. In the case of DnaK, DnaJ and GrpE stimulate DnaK's ATPase activity up to 50-fold (13). In this reaction, DnaJ specifically accelerates the hydrolysis of bound ATP to ADP, while GrpE specifically stimulates the release of both ADP and ATP from DnaK (13). In sharp contrast, GroES has been shown to inhibit the ATPase activity of GroEL (14,15). In addition, A. Ziemienowicz, D. Skowyra, J. Zeilstra-Ryalls, 0. Fayet, C.
Georgopoulos, and M. Zylicz, unpublished observations. while DnaJ has been shown to bind to some unfolded proteins (161, no such binding has ever been demonstrated for GroES. Previously, we were able to show that DnaK can both protect RNA polymerase (RNAP) from thermal inactivation and, using energy derived from ATP hydrolysis, reactivate RNAP by dissolving the large protein aggregates formed during incubation at high temperature (17). Here we show that GroEL is also able to carry out the same functions. Demonstration of this in vitro activity provides an important correlation with a previously observed phenotype noted in both dnaK and groE temperature-sensitive (Ts-) strains, that is, a reduction in global RNA synthesis at nonpermissive temperatures (3,18). We also demonstrate that the roles of the partner HSPs, DnaJ and GrpE, and GroES, in the functioning of DnaK and GroEL, respectively, are not the same due to differences in their mechanisms of action.

Proteins
Highly purified proteins (90% or greater purity) were used. Their specific activities were as follows: DnaK, 3 X lo3 units/mg (40); DnaJ, 4 X 10' units/mg (purified as described by Zylicz et al. (22) with the modifications described by Zylicz et al. (19)); GrpE, 5 X lo6 units/mg (23). A unit of activity catalyzes the incorporation of 1 pmol of deoxynucleotides/min into trichloroacetic acid-insoluble material under the standard Xdv in vitro DNA replication assay conditions (19). Wild type RNAP enriched with u" was purified as described by Burgess and Jendrisak (24).
Purification of GroEL and GroES Proteins-E. coli TZ144 (pOF39) bacteria were cultured in 10 liter of LB medium at 37 'C to OD, = 1.2. Cells were harvested by centrifugation (8,000 rpm for 10 min at 4 "C in a Beckman J21B rotor) and washed with ice-cold T-buffer (50 mM Tris/HCl, pH 8.0, 100 mM KCl, 5 mM @mercaptoethanol). After centrifugation cells were resuspended in 10 ml of T-buffer containing 10% sucrose and frozen in liquid nitrogen. After thawing the cells in a 4-8 "C water bath, 20 ml of T-buffer containing 0.3 t d spermidine and 2 M KCl, 5 ml of lysozyme (10 mg/ml in cold distilled HaO), and enough T-buffer to reach a final volume of 200 ml were added. The mixture was incubated on ice for 1 h, shifted to a 37 "C water bath for 5 min, then returned to ice. Debris was pelleted by centrifugation (10,000 rpm for 60 min at 4 "C in a Beckman J21B rotor). To the cleared supernatant, solid ammonium sulfate was added to 30% saturation, stirred gently on ice for 30 min, then centrifuged (10,000 rpm for 160 min at 4 ' C in a Beckman J21B rotor). Ammonium sulfate was added to the supernatant to 50% saturation, stirred gently on ice for 30 min, then centrifuged as before. Precipitated proteins were dissolved in 10 ml of B-buffer (50 mM Tris/HCl, pH 8.0,10% (v/v) glycerol, 5 mM j3-mercaptoethanol,5 mM EDTA, 0.05% Triton X-100, and 1 M KC]); half of this was applied to a Sepharose CL-4B column (2.7 X 70 cm; 7 ml/h). The GroEL and GroES peak fractions, as determined by SDS-polyacrylamide gel electrophoresis, were pooled separately, and the purification was continued as described below.
GroEL-After dialysis in T-buffer, GroEL fractions were applied to a Q-Sepharose column (1.7 X 10 cm, 20 ml/h) preequilibrated with T-buffer. The column was washed with one column volume of Tbuffer, then three column volumes of 200 mM KC1 in T-buffer, followed by a 200-600 mM KC1 gradient in T-buffer (100 ml). The peak fractions were pooled. Ammonium sulfate was added to a final concentration of 0.3 g/ml, and precipitation allowed to proceed for an additional 30 min at 0 "C. Precipitated proteins were pelleted by centrifugation as described above and dissolved in a minimal volume of T-buffer prior to dialysis in T-buffer overnight. Aliquots were frozen in liquid nitrogen and stored at -70 "C. The final concentra-tion of GroEL protein was 10 mg/ml.
The same protocol was used to purify the mutant GroEL673 protein.
GroES-The peak fractions from the Sepharose CL-4B column were dialyzed in T-buffer, and applied to a Q-Sepharose column, using the same conditions as for GroEL. GroES protein was eluted from the column with a 100-500 mM KC1 gradient in T-buffer (100 ml). Fractions containing GroES were dialyzed in I-buffer (50 mM imidazole pH 6.9, 20% glycerol, 1 mM EDTA, 1 D M dithiothreitol) and loaded onto a Bio-Rex 70 column (0.7 cm X 15 cm, 4 ml/h) preequilibrated with I-buffer. The column was washed with 10 column volumes of I-buffer, followed by elution of GroES with 2 M KC]. The protein was diluted &fold with T-buffer before loading onto a Bio-Gel HTP column (0.5 X 5 cm). The column was washed with Tbuffer and GroES was eluted with P-buffer (100 mM NaPO, pH 7.5, 100 mM KC1 and 5 mM 8-mercaptoethanol). Peak fractions were combined and dialyzed in T-buffer. Aliquots were frozen in liquid nitrogen and stored at -70 "C. The final concentration of GroES protein was 4 mg/ml.

Reactivation Procedure
Purified RNAP (0.5 pl at 1.7 mg/ml) was diluted 20-fold by slowly adding 9.5 pl AB buffer (10 mM Tris-HC1 pH 8, 10 mM MgCls, 10 mM P-mercaptoethanol, 50 mM KCI, 0.1 mM EDTA, 5% (v/v) glycerol and 100 pg/ml acetylated bovine serum albumin) and inactivated by a 10-min incubation at 45 "C, as described previously (17). Following transfer of the heat-inactivated RNAP to 30 'C, 6 pl of various amounts of chaperone proteins in the presence or absence of 4 mM ATP (and 40 mM creatine phosphate and 2.5 mg/ml creatine kinase, as indicated) was added. The stock solutions of DnaK, DnaJ, GrpE, GroEL and GroES (10, 4, 2, 10, and 4 mg/ml, respectively) were diluted into AB buffer to the appropriate concentration before addition to the reactivation assay. The time of "preincubation" at 30 "C varied from 5 to 60 min (see figure legends for details). Following this preincubation, RNAP transcription was initiated as described by  Table I for details). Transcription was carried out for either 2.5 or 8 min at 30 "C and stopped as previously described (17). The 2.5-min transcription reactions were carried out in order to minimize chaperone protein action during the time of the transcription assay. Relative RNAP activity was calculated by taking as 100% the activity exhibited by non-heat-treated RNAP observed in the presence of a particular combination of chaperone proteins.

Protection of RNAP from Heat Znuctivatwn
The assay was conducted as described previously (17). Purified RNAP was diluted and heat-inactivated as described for the reactivation assay, with the only difference being that RNAP was diluted with AB buffer containing the indicated amounts of chaperone proteins, i.e. prior to heat treatment (10 min at 45 "C), the chaperone proteins were already present. Following transfer to 30 "C, the transcription assay (8 min at 30 "C) was initiated by addition of 9 pl of mixture I1 (see "Reactivation Procedure") and 6 pl of AB buffer.
Relative RNAP activity was calculated using the same assumption as for the reactivation assay.

RESULTS
Both DnaK and GroEL Protect RNAP from Heat Inactivation-We showed previously that the E. coli DnaK chaperone protein is capable of protecting RNAP from heat inactivation (17). Here we present results demonstrating that GroEL also protects RNAP from heat inactivation (Fig. 1). We found that 80% of total RNAP activity remains when heat treatment is carried out in the presence of a 101 molar ratio of native GroEL 14 mer to RNAP holoenzyme. To obtain equivalent protection by DnaK, a ratio of 1401 of DnaK monomers to RNAP is required (Ref, 17; results not shown). This suggests that GroEL may have a higher affinity for binding to RNAP than DnaK protein. RNAP was diluted 20-fold by AB buffer containing the indicated amounts of GroEL or GroEL673 proteins, incubated at 45 "C for 10 min, then transferred to 30 "C. Transcription assays were initiated as described under "Protection of RNAP from Heat Inactivation." The final concentration of RNAP in the protection assay was 0.2 PM. The symbols EL and EL 673 designate GroEL wild type and GroEL673 mutant proteins, respectively. Molar ratios were calculated on the basis of RNAP being M, 430,000 and GroEL (or GroEL673) 14 X 57,259 = M , 801,626. Relative RNAP activity was calculated by taking as 100% the activity of control unheated RNAP assayed in the presence of the various amounts of GroEL (W) and GroEL673 ( proteins. Previously we showed that nearly equivalent levels of protection are afforded by the DnaK756 mutant protein (17). The addition of purified GroEL673 mutant protein also results in protection of RNAP activity at levels close to those obtained with the wild type GroEL protein (Fig. 1). The results with the mutant proteins are not necessarily surprising. Since the DnaK and GroEL functions are essential at all temperatures, under permissive conditions the mutant proteins must retain some chaperone activity to allow E. coli growth. In the case of GroEL673, this partial functionality has been demonstrated in vivo (20).
Both genetic and biochemical studies have established that the GroEL and GroES proteins interact (3). Studies on refolding of a variety of polypeptide substrates have shown that GroES increases recovery of refolded substrates (26). Based on these observations, we asked whether GroES influences the ability of GroEL to protect RNAP from heat inactivation. Fig. 2 shows that the presence of GroES improves protection of RNAP by the wild type GroEL protein, but has no effect on protection by the GroEL673 mutant protein. The optimal molar ratio of GroES to wild type GroEL in this reaction was 1 or 2 GroES 7 mers to 1 GroEL 14 mers. The inability of GroES to enhance protection of RNAP by the GroEL673 mutant protein suggests that GroEL673 is defective in its ability to interact with GroES. In control experiments using size chromatography and native electrophoresis, we have shown that the presence of either the GroEL/S or DnaK/J/ E chaperone systems prevents formation of large RNAP aggregates (results not shown).
Wild Type DnaK and GroEL Chaperone Systems Can Reactivate Heat-treated RNAP-Given the fact that RNA synthesis is affected in both groE and dnaK mutants (3), we decided to test if the GroEL/GroES chaperone system shows the same ability as DnaK to reactivate heat-inactivated RNAP. After incubation for 10 min at 45 " C , less than 10% of the RNAP activity (measured by the transcription assay at 30 "C) remained compared to untreated RNAP. In this case, more than 90% of RNAP appeared in the void volume of the Bio-Gel A- RNAP was diluted 20-fold by AB buffer containing GroEL or GroEL673 and indicated amounts of GroES protein in the presence of 4 mM ATP and the ATP regeneration system as described under "Materials and Methods." After a 10-min incubation at 45 "C, the reaction mixtures were transferred to 30 'C, and transcription assays were initiated as described under "Materials and Methods." The final concentrations of RNAP and GroEL in the protection assay were 0.2 and 0.68 pM, respectively. The molar ratios were calculated on the basis of GroES being 7 X 10,368 = M, 72,576. Relative RNAP activity (calculated taking as 100% the activity of control unheated RNAP) assayed in the presence of varying amounts of GroES, but the same amount of GroEL 15 gel filtration column, as was previously shown (results not shown; see also Fig. 3 of Ref. 17). However, when GroEL, GroES, and ATP were added to heat-treated RNAP in a ratio of 10 GroEL 14 mers:20 GroES 7 mers:l RNAP, reactivation of RNAP activity was observed following a 10-20-min lag. After 60 min of incubation with GroEL, GroES, and ATP, more than 60% of the RNAP activity was recovered (Fig. 3).
Surprisingly, when heat-inactivated RNAP was preincubated for 10 min at 30 "C with only GroEL, followed by addition of GroES and ATP at the start of the transcription assay, no lag in the kinetics of RNAP reactivation was observed. This result suggests that the presence of GroES may interfere with the binding of GroEL to the heat-inactivated RNAP. Preincubation of RNAP with GroES, or GroES and ATP, or GroEL and ATP did not reduce this lag period (results not shown). In control experiments, we showed that GroEL, GroES, and ATP are all absolutely required for efficient reactivation of heat-inactivated RNAP ( Fig. 3; results not shown).
Athough both purified DnaK756 and GroEL673 mutant proteins can protect (Ref. 17; Fig. l), neither of the mutant proteins can function in the reactivation of heat-treated RNAP (Fig. 4). We previously showed that the wild type DnaK protein alone could carry out the reactivation reaction (17), but the DnaK756 mutant protein did not have this ability. Since, for GroEL, this activity absolutely requires GroES (Fig. 3), we asked whether GroEL673 can reactivate heat-treated RNAP in the presence of GroES. Under these conditions, the GroEL673 mutant protein showed no ability to reactivate heat-treated RNAP (Fig. 4). This result correlates with the previous observation that addition of GroES did not improve the ability of the GroEL673 mutant protein to protect RNAP from heat inactivation (see Fig. 1).
Effect of DnaJ and GrpE Proteins on DnaK-dependent Reactivation of RNAP-Previously, we showed that it was necessary to add a large excess of DnaK (156 molecules of DnaK/molecule of RNAP, (17) to restore RNAP activity,  DnaJ, and GrpE in the reactivation assay were 0.12, 0.12, and 2 p~, respectively. Relative RNAP activity was calculated taking as 100% the activity of control unheated RNAP assayed in the presence of 18.08 p~ DnaK in the presence or absence of DnaJ and GrpE protein (DnaK alone, 10,000 cpm; DnaK and DnaJ, 11,200 cpm; DnaK, DnaJ and GrpE, 14,800 cprn). (Fig. 5). Titration of DnaJ protein in this reaction clearly showed that less than one dimeric molecule of DnaJ per molecule of RNAP is sufficient to saturate the reaction (Fig.   5). Thus, in this assay system it is possible to reduce the concentration of DnaK protein needed to achieve a given level of reactivation by the addition of DnaJ.
The effect of adding GrpE to the reactivation reaction catalyzed by DnaK alone is minimal (data not shown), but there is pronounced enhancement when GrpE is added together with DnaJ to the reaction (Figs. 5 and 6). This stimulatory effect is maximal at a 1:l ratio of GrpE to DnaK (Fig.  7). As a control, neither DnaJ alone nor DnaJ together with GrpE was able to restore RNAP activity (Fig. 6). This established that the optimal reactivation by DnaK i s achieved when both DnaJ and GrpE are present.
During these studies, it became apparent that the time of preincubation of the various proteins with heat-inactivated RNAP was a critical factor in the reactivation reaction, an observation that has already been noted for the GroEL system (see Fig. 3). To more clearly resolve the kinetics of reactivation by the DnaK system, we used a brief 2.5-min transcription assay. When a large amount of DnaK protein alone was used (156 molecules of DnaK per molecule of RNAP), a lag of approximately 10 min was observed, followed by a slow reactivation reaction (Fig. 7). An 8-fold reduction in the amount of DnaK protein (19 molecules per molecule of RNAP) resulted in very little, if any, reactivation of RNAP (Fig. 7). The addition of DnaJ to this limiting amount of DnaK reduced the lag period to 5 min, and thereafter the reactivation reaction proceeded with kinetics very similar to those observed when large amounts of DnaK protein alone were used (Fig. 7).
GrpE with DnaK had very little, if any, effect on the kinetics of the RNAP reactivation reaction. However, when and DnaJ ((2°C)); DnaK (2.26 p~) , (c-".). The concentration, of RNAP, DnaJ, and GrpE used in these reactions were 0.12,0.12, and 2 pM, respectively. Following incubation at 30 'C for the indicated period of time, 2.5-min transcription assays were initiated as described under "Materials and Methods." Relative activity was calculated taking as 100% the activity of control unheated RNAP assayed in the presence of all other proteins (DnaK alone, high concentration, 1,100 cpm; DnaK alone, low concentration, 1,500 cpm; with D n d , 1,400 cpm; with GrpE, 1,600 cpm; with DnaJ and GrpE, 1,980 cpm). the same limited amount of DnaK protein was supplemented with saturating amounts of both DnaJ and GrpE proteins, the kinetics of the reactivation reaction were dramatically different. Ten minutes of preincubation were sufficient to restore more than 60% of the RNAP activity (Fig. 7).

DnaJ, and GrpE
To examine whether all three HSPs must be present during all stages of the RNAP reactivation assay, we performed a series of experiments in which different combinations of these proteins were present either during a 5-min preincubation step with heat-inactivated RNAP or were added only during the short 2.5-min transcription assay for RNAP activity. A comparison of the net effect of the various combinations is presented in Table   I, while the GrpE dependence of this reactivation reaction under these different conditions is shown in Fig. 8. These results reveal the following. With the relatively short 5-min preincubation, no significant level of DnaK-and DnaJ-dependent reactivation of RNAP was ob-    (Table I, line 6 ) ; reactivation was absolutely dependent on the presence of GrpE ( Fig. 8; Table I, line 7). Apparently the relatively short, 2.5-min transcription assay period was insufficient time to reactivate RNAP since the addition of DnaK, DnaJ, GrpE, and ATP at the start of the transcription assay resulted in no reactivation ( Fig. 8; Table I, line 4 ) . When DnaK, DnaJ, and GrpE were preincubated with heat-inactivated RNAP in the presence of ATP, only 37% of the activity was recovered ( Fig. 8; Table I, line 2). When DnaJ was preincubated alone with heat-inactivated RNAP, followed by addition of DnaK and GrpE at the start of the transcription assay, a more efficient reactivation was found to occur ( Fig. 8; Table I, line 10). Further stimulation was achieved when both DnaJ and DnaK were present (Table I, line 11). The most dramatic stimulation of the reactivation reaction was observed when DnaK, DnaJ, and GrpE were preincubated with the heat-inactivated RNAP for 5 min, in the absence of ATP (ATP was added upon initiation of the transcription assay). Under these conditions 83% of the initial RNAP activity could be recovered ( Fig. 8; Table I, line 3). This result cannot be explained by the stimulatory effect of DnaJ and GrpE on DnaK's ATPase activity (13). Rather, this result suggests that DnaJ and GrpE positively regulate the binding of DnaK to heat-inactivated RNAP.

DISCUSSION
Our results show that both DnaK and GroEL can protect RNAP from heat inactivation. Furthermore, both the DnaK/ DnaJ/GrpE and GroEL/GroES chaperone systems can reconstitute activity from heat-treated RNAP. The molecular mechanism of reactivation of heat-inactivated RNAP is still not clear. Recently, our laboratory is addressing the possibility that during heat inactivation, the u70 subunit is released from the RNAP holoenzyme, and the remaining RNAP core molecules aggregate. According to this hypothesis, the chaperone systems, GroEL/S or DnaK/J/E, may not necessarily work directly to disaggregate the aggregated cores of RNAP, but rather may promote the reassociation of d o with the RNAP cores. This would result in the disaggregation of RNAP, leading to reactivation of its enzymatic activity. Such an indirect role of the chaperone proteins on the disaggregation reaction of RNAP may explain why in other systems a direct involvement of chaperone proteins in the disaggregation of protein aggregates is hard to prove (27,28). Of course another possibility is that the "quality" of the various aggregates may be different in the various systems used, with the RNAP aggregate representing a "mild version. The RNAP reconstitution system has allowed us to compare the mechanisms by which the DnaK/DnaJ/GrpE and GroEL/GroES chaperone systems carry out the same function. We conclude that the rate-limiting step for both mechanisms is the binding of chaperone to substrate. This is based on the lag in reactivation seen when there is no preincubation of heat-treated RNAP with either DnaK or GroEL. The activity of DnaK is dramatically enhanced in the presence of DnaJ. However, the lag phase in reactivation is substantially diminished only when the third protein of the DnaK chaperone system, GrpE, is present as well. The stimulation of DnaK activity by DnaJ in this system differs from the behavior of DnaK and DnaJ in the X DNA replication system. For that system, the need for a high concentration of DnaK protein could not be overcome by DnaJ alone, but only by both DnaJ and GrpE (19,29). The observed lag phase in the case of DnaJ-, DnaK-dependent reactivation of heat-inactivated RNAP, and its disappearance when all three proteins were preincubated with heat-treated RNAP (in the absence of ATP) suggest that the GrpE protein is also involved in the binding of DnaK to RNAP. This finding is new, and sheds further light on the mechanism of DnaJ and GrpE action. It is well established that binding of DnaJ protein to different polypeptides increases the affinity of DnaK for these substrates. Examples are RepA (30), X P protein (31), and u~~.~ Recently, we have shown that, in the presence of GrpE protein, DnaK protein binds more efficiently to the XP-DnaJ complex than to XP alone (29,31). We call this the "discrimination" function of GrpE. Perhaps in the reactivation of RNAP, GrpE can change the conformation of DnaK in such a way that DnaK recognizes better those RNAP molecules that are already "bound to DnaJ.
Apparently, the rate-limiting step is the binding of DnaK to aggregated RNAP. Since ATP brings about a dramatic conformational change in DnaK that leads to its release from substrate proteins (32-34), it is not surprising that the best reactivation of RNAP was achieved when all three proteins of the DnaK chaperone system were first preincubated with heat-treated RNA, in the absence of ATP. These conditions would shift the equilibrium of the reaction towards binding of the heat shock proteins to RNAP. The subsequent addition of ATP would then induce a change of conformation of DnaK (33) and perhaps RNAP, thus favoring release of RNAP from DnaK.The data presented here suggest that the fundamental difference in the DnaK and GroE chaperone systems lies in the affinity of DnaK and GroEL for substrate polypeptides. As shown here for RNAP and for other substrates, e.g. X P protein (29,31), DnaK has an inherently low affinity for substrates which, when coupled to its promiscuity may prevent it from localizing at specific, critical sites on the RNAP aggregate. This problem may be bypassed by the presence of DnaJ and GrpE which may help DnaK to improve its affinity for certain substrates. In the GroE system, GroEL has an intrinsically high affinity for substrate (RNAP), which needs to be reduced for the release and subsequent recycling of GroEL. This reduction in affinity is brought about by GroES. A similar conclusion for the role of GroES was recently presented by Bochkareva et al. (35).
Consistent with our results, Martin et al. (36) have recently reported both in vivo and in vitro data demonstrating that the Hsp6O family of proteins is capable of preventing protein denaturation and aggregation. Similarly, Mendoza et al. (37) have shown that GroEL can prevent in vitro rhodanese enzyme from heat inactivation. Recently Hartman et al. (38) showed that substochiometric amount of GroEL and GroES prevents thermal denaturation and aggregation of mammalian mitochondrial malate dehydrogenase in vitro. In this case GroEL alone did not protect dehydrogenase activity against thermal inactivation but kept the denatured protein soluble and thereby prevented its aggregation (38).
Two recent reports have suggested a synergistic role of the DnaK and GroEL chaperone systems in preventing polypeptide misfolding and aggregation. The in vitro system of Langer et al. (16) follows the sequential in vitro binding of unfolded rhodanese by DnaK and DnaJ, and its ATPand GrpEdependent release and subsequent binding by the GroEL chaperone. In this system, GrpE appears to assist in the release of bound rhodanese from DnaK/DnaJ. Contrary to these results, GrpE can assist in both the binding of DnaK to X P (in the absence of ATP), as well as the release of X P from DnaK (in the presence of ATP hydrolysis) (31). This apparent contradiction in GrpE action can be due either to the use of K. Liberek and C. Georgopoulos, manuscript submitted for Publication. different polypeptide substrates or the ability of GrpE to act both as a "specificity" factor for DnaK binding to some polypeptides, as well as a release factor.
The in vivo report of Gragerov et al. (39) demonstrates that newly synthesized polypeptide aggregate in E. coli mutants unable to mount a proper heat shock response (deleted for rpoH, coding for u3'). This aggregation can be reversed through the overproduction of either DnaK/DnaJ or GroES/ GroEL. Alternatively, physiological levels of DnaK/DnaJ/ GroES/GroEL can also prevent protein aggregation in the rpoH mutant host. The fact that the grpE gene can be expressed from either a u3*-or do-dependent promoter' ensures the presence of GrpE protein in the rpoH-deleted host. Hence, our in vitro results with purified proteins are in complete agreement with these in vivo findings. Most likely, the demonstrated overlap in function of the DnaK and GroEL chaperone ensures that important molecules, such as RNAP, will stay functional under adverse physiological conditions, such as those encountered under heat stress, and thus assure the probability of survival.