Characterization and Mechanism of Action of a Reactivating Factor for Adenosylcobalamin-dependent Glycerol Dehydratase*

Adenosylcobalamin-dependent glycerol dehydratase undergoes mechanism-based inactivation by its physiological substrate glycerol. We identified two genes ( gdrAB ) of Klebsiella pneumoniae for a glycerol dehy-dratase-reactivating factor (Tobimatsu, T., Kajiura, H., Yunoki, M., Azuma, M., and Toraya, T. (1999) J. Bacteriol. 181, 4110–4113). Recombinant GdrA and GdrB proteins formed a tight complex of (GdrA) 2 (GdrB) 2 , which is a putative reactivating factor. The purified factor reactivated the glycerol-inactivated and O 2 -inactivated glyc- erol dehydratases as well as activated the enzyme-cya-nocobalamin complex in vitro in the presence of ATP, Mg 2 (cid:1) , and adenosylcobalamin. The factor mediated the exchange of the enzyme-bound, adenine-lacking cobalamins for free, adenine-containing cobalamins in the presence of ATP and Mg 2 (cid:1) through intermediate formation of apoenzyme. The factor showed extremely low ATP-hydrolyzing activity and formed a tight complex with apoenzyme in the presence of ADP. Incubation of the enzyme-cyanocobalamin complex with the reactivating factor in the presence of ADP brought about release of the enzyme-bound cobalamin. AdePeCbl was prepared as described before (21). [ (cid:2) - 32 P]ATP and [ 57 Co]CN-Cbl (spe- cific activity, 240 (cid:3) Ci/ (cid:3) g) were obtained from PerkinElmer Life Sci-ences and ICN Pharmaceuticals, respectively. Glycerol dehydratase was purified from Escherichia coli JM109 carrying pUSI2E(GD) (12) by Sepharose CL-6B and hydroxyapatite column chromatography. Glyc- erol dehydratase with a specific activity of more than 50 units/mg of protein was used in this study. All other chemicals and the enzymes used for construction of plasmids were commercial products of the highest grade available and were used without further purification. Bacterial Strain, Plasmids, and Culture Conditions—E. coli BL21(DE3) was used as a host. The expression plasmid for the glycerol

siella and Citrobacter, and other bacteria when they are grown anaerobically in a medium containing glycerol (1)(2)(3)(4)7), this enzyme undergoes mechanism-based inactivation by glycerol during catalysis (3,8,9). The glycerol-inactivated enzyme in permeabilized cells (in situ) of Klebsiella pneumoniae is rapidly reactivated by exchange of the modified coenzyme for intact AdoCbl in the presence of ATP and Mg 2ϩ (or Mn 2ϩ ) (10). The complex of enzyme with CN-Cbl is also activated in situ under the same conditions (11). Recently, we identified two open reading frames in the vicinity of the glycerol dehydratase genes (12) as the genes encoding reactivating factor for glycerol dehydratase and designated them gdrAB (13). The products of the genes formed a tight complex that was considered a putative reactivating factor.
Diol dehydratase (EC 4.2.1.28), an isofunctional enzyme, is inducibly formed when certain bacteria are grown in a medium containing 1, 2-propanediol (3). This enzyme involved mainly in the utilization of 1,2-propanediol by Enterobacteriaceae and other bacteria (3,14) also undergoes mechanism-based inactivation by glycerol (15,16). A reactivating factor for diol dehydratase (DdrA⅐DdrB complex (17)) of Klebsiella oxytoca has been well characterized (18), and the mechanism of action established in vitro (19). It is a molecular chaperon-like factor that mediates ATP-dependent release of the tightly bound, modified coenzyme from the inactivated holoenzyme. The reactivation takes place in two steps (i.e. ADP-dependent cobalamin release and ATP-dependent dissociation of apodiol dehydratase-reactivating factor complex (19)).
The amino acid sequences of GdrA and GdrB are 61 and 30% identical to those of DdrA and DdrB, respectively (17). Although functional similarity of these reactivating factors was suggested from these sequence similarities, the reactivating factor for glycerol dehydratase has not yet been investigated in vitro. In the present paper, we report the purification, characterization, and mechanism of action of the recombinant glycerol dehydratase-reactivating factor. During the preparation of this manuscript, a paper has appeared that reports characterization of a similar factor of Citrobacter freundii (20), but the mechanism of its function still remains obscure.

Materials-Crystalline
AdoCbl was a gift from Eizai Co., Ltd. (Tokyo, Japan). CN-Cbl was obtained from Glaxo Wellcome. AdePeCbl was prepared as described before (21). [␥-32 P]ATP and [ 57 Co]CN-Cbl (specific activity, 240 Ci/g) were obtained from PerkinElmer Life Sciences and ICN Pharmaceuticals, respectively. Glycerol dehydratase was purified from Escherichia coli JM109 carrying pUSI2E(GD) (12) by Sepharose CL-6B and hydroxyapatite column chromatography. Glycerol dehydratase with a specific activity of more than 50 units/mg of protein was used in this study. All other chemicals and the enzymes used for construction of plasmids were commercial products of the highest grade available and were used without further purification.
Bacterial Strain, Plasmids, and Culture Conditions-E. coli BL21(DE3) was used as a host. The expression plasmid for the glycerol dehydratase-reactivating factor was constructed as follows. The plasmid pCXV(gdrB⅐gdrA) (13) was digested completely with BglII and partially with NdeI. The resulting 2.2-kilobase pair DNA fragment was inserted into the NdeI-BamHI region of pET21a to produce pET(gdrB⅐gdrA). E. coli BL21(DE3) carrying plasmid pET(gdrB⅐gdrA) was cultured in Terrific Broth containing 50 g/ml ampicillin at 30°C. When the culture reached an A 650 of ϳ1.0, isopropyl-1-thio-␤-D-galactopyranoside was added to a concentration of 1 mM. The E. coli cells were cultivated overnight at 17°C, harvested, and stored at Ϫ80°C until use.
Purification of the Recombinant Reactivating Factor-About 15 g of wet cells were suspended in 150 ml of 0.05 M potassium phosphate buffer (pH 8) containing 2 mM EDTA and 2 mM phenylmethanesulfonyl fluoride and disrupted by sonication. To the cell extract obtained after centrifugation at 27,200 ϫ g for 30 min, ammonium sulfate was added to 5% saturation. After removal of a precipitate by centrifugation, ammonium sulfate was added to 25% saturation. The precipitate was collected, dissolved in 15 ml of 0.05 M potassium phosphate buffer (pH 8) containing 2 mM EDTA and 2 mM phenylmethanesulfonyl fluoride, and dialyzed against 5 mM potassium phosphate buffer (pH 8) containing 0.5 mM EDTA. Dialysate was applied onto a Sepharose CL-6B column (bed volume, 500 ml) that was equilibrated with 0.05 M potassium phosphate buffer (pH 8) containing 0.3 M 1,2-propanediol. The column was developed with the same buffer, and the fractions containing GdrA and GdrB were collected and concentrated by ultrafiltration through a Centriplus-10 concentrator (Amicon) and stored at Ϫ80°C.
Assays of Enzyme and Reactivating Factor-The amount of propionaldehyde formed from 1,2-propanediol was determined by the 3-methyl-2-benzothiazolinone hydrazone method (22). One unit of glycerol dehydratase is defined as the amount of enzyme activity that catalyzes the formation of 1 mol of propionaldehyde/min at 37°C with 1,2-propanediol as substrate. Activity of the glycerol dehydratase-reactivating factor was assayed with 1,2-propanediol as substrate by its capability of activating the inactive enzyme⅐CN-Cbl complex. This activity was well correlated with its capability of reactivating the glycerolinactivated holoenzyme.
Protein Assay-Protein concentration of crude enzyme and reactivating factor was determined by the method of Lowry et al. (23) with crystalline bovine serum albumin as a standard. The concentrations of purified glycerol dehydratase and its reactivating factor were determined by measuring the absorbance at 280 nm. The molar absorption coefficient at 280 nm calculated by the method of Gill and von Hippel (24) for the reactivating factor was 86,500 M Ϫ1 cm Ϫ1 .
PAGE-PAGE was performed under nondenaturing conditions as described by Davis (25) or under denaturing conditions as described by Laemmli (26). Nondenaturing PAGE of glycerol dehydratase was performed in the presence of 0.1 M 1,2-propanediol to prevent its subunits from dissociation (27). In some experiments, ATP or ADP was also added with MgCl 2 (1 mM each) and KCl (2 mM) to gels and electrode buffer. Protein was stained with Coomassie Brilliant Blue G-250. The NIH Image program, version 1.61 (National Institutes of Health) was used for densitometric analysis of gels.
ATPase Activity-The products formed by the hydrolysis of ATP with the reactivating factor were analyzed by TLC on a poly(ethyleneimine)cellulose F plate (Merck) with 2 M formic acid containing 0.5 M LiCl as a solvent system (28), as described before (19). ATP-hydrolyzing activity of the reactivating factor was assayed by the release of [ 32 P]P i from [␥-32 P]ATP, as described before (19) by a modification of the method of Schnebli and Abrams (29). An appropriate amount of the factor was incubated at 37°C for 30 min with 3 mM [␥-32 P]ATP (ϳ3.1 ϫ 10 6 dpm) in 20 mM potassium phosphate buffer (pH 8) containing 3 mM MgCl 2 , in a total volume of 100 l. After termination of the reaction by adding 0.9 ml of ice-cold suspension of 6% (w/v) charcoal in 50 mM NaH 2 PO 4 and mixing vigorously for 10 min, the charcoal was removed by centrifugation. The amount of radioactivity in 0.5 ml of the supernatant was determined by liquid scintillation counting, and ATPase activity was obtained by subtracting the radioactivity of a minus reactivating factor control.

Purification and Subunit Structure of the Recombinant
Reactivating Factor-The M r of the two proteins (64,000 and 12,000) overexpressed in E. coli BL21(DE3) carrying pET(gdrB⅐gdrA) corresponded to the predicted molecular weights of GdrA (63,594) and GdrB (11,994), respectively. They were co-purified through ammonium sulfate fractionation and Sepharose CL-6B column chromatography (Fig. 1A) and mi-grated as a single band when electrophoresed under nondenaturing conditions (Fig. 1B). The sequences of N-terminal six amino acids of the large and small subunits were confirmed by Edman sequencing to be identical with those predicted from the nucleotide sequences of gdrAB (13,17). Densitometric analysis indicated that the molar ratio of the large and small polypeptides was ϳ1:1. The apparent molecular weight of the complex determined by FPLC with a calibrated Superose 6 column (HR 10/30) was ϳ135,000 (data not shown), suggesting that the subunit structure of the putative reactivating factor is most likely (GdrA) 2 (GdrB) 2 .
Reactivation of Inactivated Holoenzymes by the Reactivating Factor-Reactivation of the glycerol-inactivated hologlycerol dehydratase in vitro was monitored by the recovery of 1,2propanediol-dehydrating activity. In the presence of the purified reactivating factor, AdoCbl, ATP, and Mg 2ϩ , the glycerolinactivated holoenzyme restored its catalytic activity but not at all without the factor nor in the absence of ATP and Mg 2ϩ ( Fig.  2A). Fig. 2B showed that the holoenzyme inactivated by O 2 in the absence of substrate (9, 30) also underwent rapid reactivation by the factor in the presence of AdoCbl, ATP, and Mg 2ϩ without a significant lag period. Thus, it is evident that the putative reactivating factor actually functions as a reactivating factor for both glycerol-inactivated and O 2 -inactivated hologlycerol dehydratases in vitro.
Activation of the Enzyme-Cobalamin Complex by the Reactivating Factor-The inactive complex of glycerol dehydratase with CN-Cbl, an adenine-lacking cobalamin, can be considered as a model of the inactivated holoenzyme. Fig. 2C indicated that the enzyme⅐CN-Cbl complex was also activated rapidly in vitro by the factor in the presence of free AdoCbl, ATP, and Mg 2ϩ . In contrast, the complex of enzyme with AdePeCbl, an adenine-containing cobalamin, did not undergo significant activation (Fig. 2D). AdePeCbl is an inactive analog of AdoCbl, in which the ribose moiety is replaced by a pentamethylene group (21) and forms a very tight complex with diol dehydratase (22).
Evidence for Cobalamin Exchange-The absolute requirement of both reactivation of glycerol-inactivated and O 2 -inactivated holoenzymes and activation of the enzyme⅐CN-Cbl complex for free AdoCbl strongly suggested that the reactivation and the activation take place by exchange of the enzyme- bound, modified coenzyme and CN-Cbl, respectively, for free intact AdoCbl. CN-Cbl and AdePeCbl were used as models of the modified and intact coenzymes, respectively (18). When the enzyme⅐CN-Cbl complex was incubated with or without the reactivating factor in the presence of AdePeCbl, ATP, and Mg 2ϩ , followed by dialysis to remove unbound cobalamins, the spectrum of the dialysate indicated that the enzyme-bound CN-Cbl was replaced by AdePeCbl in the presence of factor (Fig. 3A). This exchange did not occur without the factor. The reverse was not the case even in the presence of factor, ATP, and Mg 2ϩ (Fig. 3B). Thus, it was suggested that the reactivating factor mediates the exchange of the enzyme-bound, adenine-lacking cobalamin for free, adenine-containing cobalamin.
To examine the possibility that only an upper ligand undergoes exchange under the conditions, the enzyme⅐[ 57 Co]CN-Cbl complex was incubated with and without the factor in the presence of AdePeCbl, ATP, and Mg 2ϩ , followed by ultrafiltration to remove unbound cobalamins. One of the typical results is shown in Table I. The 57 Co radioactivity bound to glycerol dehydratase was almost completely lost in the presence of factor, ADP, and Mg 2ϩ , whereas the amount of radioactivity corresponding to more than 1 mol of CN-Cbl/mol of enzyme was retained in the protein fraction in the absence of factor. In the presence of ATP, a half of [ 57 Co]CN-Cbl bound to enzyme was lost by the factor. This result, together with the spectroscopic observation, indicated that the entire molecule of enzymebound CN-Cbl underwent the exchange for free AdePeCbl or release in the presence of reactivating factor. It is evident that ADP is more effective than ATP for this exchange or release.
Evidence for Intermediary Formation of Apoglycerol Dehydratase in the Reactivation-Glycerol-inactivated holoenzyme and the enzyme⅐aqCbl⅐5Ј-deoxyadenosine complex were incubated with the reactivating factor in the presence of ATP, Mg 2ϩ , and K 2 SO 3 and then dialyzed to remove unbound cobal-  amin. SO 3 2Ϫ was added to prevent the dissociated aqCbl from reassociation with the enzyme by converting it to sulfitocobalamin (31). The spectrum of the dialysate from the glycerolinactivated holoenzyme indicated that, although a small amount of cob(II)alamin remained bound, a major part of the enzyme-bound cobalamin was released in the presence of factor but not in its absence (Fig. 3C). Catalytic activity was restored by adding AdoCbl to the dialysate (data not shown). Fig. 3D showed that most of the enzyme-bound aqCbl dissociated from the inactive enzyme⅐aqCbl⅐5Ј-deoxyadenosine complex, a model of the inactivated holoenzyme. It was thus concluded that apoenzyme is formed as an intermediate in the reactivating factor-mediated cobalamin exchange.
ATP-hydrolyzing Activity of the Reactivating Factor-When ATP was incubated with the reactivating factor, time-dependent decrease of ATP and formation of ADP were observed by TLC (data not shown). Release of [ 32 P]P i was observed upon incubation of [␥-32 P]ATP with the factor, indicating that the factor hydrolyzes ATP to ADP and P i . k cat was 1.5 min Ϫ1 (mean, n ϭ 5). The time required for hydrolysis of 1 mol of ATP/mol of factor was calculated to be ϳ40 s, indicating that this factor is an ATPase with extremely low activity. Glycerol dehydratase itself did not hydrolyze ATP at all.
The ATP-hydrolyzing activity (k cat ) of the factor during glycerol dehydration (reactivating condition) was 0.9 min Ϫ1 , which was significantly lower than that during 1,2-propanediol dehydration (essentially nonreactivating condition) (1.6 min Ϫ1 ) or in the absence of AdoCbl (1.5 min Ϫ1 ). The molar ratio of glycerol dehydratase to reactivating factor was ϳ1:3 in this experiment. When the rate of ATP hydrolysis by the factor was measured with various ratios of the enzyme to the factor, ATP-hydrolyzing activity decreased until the molar ratio became to ϳ2:1 (Fig. 4). It was thus suggested that the ATPase activity of the reactivating factor is inhibited by its binding to glycerol dehydratase.
Complex Formation between Glycerol Dehydratase and the Reactivating Factor-Complex formation between glycerol dehydratase and its reactivating factor was analyzed by nondenaturing PAGE (Fig. 5). When apoenzyme was incubated with the factor in the presence of ADP, a new major band appeared (Fig. 5C). When this band was developed in the second dimension by SDS-PAGE, all of the ␣, ␤, and ␥ subunits of glycerol dehydratase and the GdrA and GdrB proteins were detected (data not shown), indicating that this band is a complex between the enzyme and the factor. Formation of the apoenzymefactor complex was much less in the absence of nucleotides (Fig. 5A) and not detectable at all in the presence of ATP (Fig.  5B). Therefore, the ADP-and ATP-bound reactivating factors are high and low affinity forms for the apoenzyme, respectively. When the factor was incubated with enzyme-cobalamin complexes, the enzyme-factor complex was not formed at all from either enzyme⅐AdePeCbl or enzyme⅐CN-Cbl complex in the presence of ATP (Fig. 5B) or in its absence (Fig. 5A). In the presence of ADP, only a small amount of the enzyme-factor complex was formed from the enzyme⅐AdePeCbl complex, a "nonactivable" complex, whereas a large amount of the enzyme-factor complex was formed from the enzyme⅐CN-Cbl complex, an "activable" complex (Fig. 5C). The second band of the complex was observed in these cases. It may have an assembly different from the first one, although their exact molar compositions were not determined.
Release of Enzyme-bound Cobalamin upon Enzyme⅐Factor Complex Formation-When the enzyme⅐CN-Cbl and enzyme⅐AdePeCbl complexes were incubated with and without the reactivating factor in the presence and absence of ATP or ADP, followed by ultrafiltration to remove unbound cobalamins, the spectrum of the protein fraction indicated that most of the enzyme-bound CN-Cbl was released from the enzyme by the factor in the presence of ADP (Fig. 6A). The release was not observed at all without the factor, indicating that the ADP form of the factor causes release of the enzyme-bound CN-Cbl. ATP was also effective for the release of CN-Cbl but to a lesser extent. This suggests that ATP exerts its effect after being hydrolyzed to ADP. In contrast, the factor-mediated release of the enzyme-bound AdePeCbl was not observed with either ADP or ATP (Fig. 6B). It seems likely that the enzyme-bound, adenine-lacking cobalamins are released from the enzyme through the complex formation with the reactivating factor.
Inhibition of Apoenzyme by the Reactivating Factor and Reversal by ATP-The effect of enzyme-factor complex formation on catalytic activity was investigated with added coenzyme. When apoglycerol dehydratase was incubated with the reactivating factor in the presence of ADP, it lost enzyme activity in a time-dependent manner (Fig. 7C). ADP alone did not inhibit the activity (data not shown). The inhibition was observed in the absence of ADP as well but to a lesser degree (Fig. 7B). In contrast, inhibition was not observed at all in the presence of ATP. It is likely that the extent of inhibition of enzyme activity by the factor depends on formation of the enzyme-factor complex. Thus, it was concluded that the apoenzyme-factor complex is unable to be reconstituted into active holoenzyme with added AdoCbl. When ATP was added together with AdoCbl, it completely reversed the inhibition (Fig. 7), probably by facilitating dissociation of the inactive apoenzyme-factor complex into free apoenzyme and the factor. Such an effect of ATP on the dissociation of the complex was confirmed by nondenaturing PAGE (data not shown) and seems reasonable, because the ATP-bound reactivating factor is the low affinity form for glycerol dehydratase. DISCUSSION Glycerol dehydratase as well as diol dehydratase undergoes rapid inactivation by glycerol during catalysis (3,8,9,15,16), although glycerol is a physiological substrate for these enzymes (3)(4)(5)(6). This enigma was solved by our discovery of the reactivation systems in the bacteria that produce these enzymes (10,11). The subunit structure, function, and mechanism of action of the putative reactivating factor was unraveled with diol dehydratase for the first time (18,19) and now with glycerol dehydratase. Recently, we obtained evidence for the presence of a specific reactivating factor for ethanolamine ammonia-lyase as well. 2 Thus, it seems likely that other AdoCbl-dependent enzymes also have their own reactivating factors. This prediction would be quite reasonable, because all of the AdoCbl-dependent enzymes catalyze reactions by a radical mechanism and therefore tend to undergo inactivation during catalysis or even in the absence of substrate (32).
The reactivating factor for diol dehydratase was shown to cross-reactivate the glycerol dehydratase in permeabilized E. coli cells (33). This fact was supported by the finding that the reactivating factor for glycerol dehydratase was very similar to that for diol dehydratase in the subunit structure and the functional role. However, it should be noted that the glycerol dehydratase-reactivating factor was not capable of cross-reactivating the inactivated holoenzyme of diol dehydratase (20,33). This indicates that there must be differences between them as well.
The reactivating factor showed low but distinct ATPase activity. Although the hydrolysis of ATP was absolutely required for reactivation of the inactivated holoenzymes by the reactivating factor (10), the ATP hydrolysis was not directly linked to the reactivation of the glycerol-inactivated holoenzyme. Such a phenomenon is seen with diol dehydratase-reactivating factor (19) as well and is not unusual in nature (34,35). The inhibition of ATPase activity of the reactivating factor by binding to apoenzyme under reactivating conditions was observed here for the first time. This may be suitable for cellular economy of energy. Fig. 8A shows the two-step mechanism that we propose for the reactivation of glycerol-inactivated hologlycerol dehydratase. The O 2 -inactivated holoenzyme and the enzyme⅐CN-Cbl complex can be (re)activated by the same mechanism. First, the factor-bound ATP is hydrolyzed to ADP, which induces the conformational change of the factor from a low affinity form to a high affinity form. Complexation of the ADP form of the factor with inactivated holoenzymes results in release of the modified coenzyme, because the apoenzyme gets energeti-2 K. Mori and T. Toraya, manuscript in preparation. cally more stabilized by complex formation with the ADP form of factor than with adenine-lacking cobalamins. The resulting apoenzyme-factor (ADP form) complex is inactive even when assayed with added AdoCbl, but the exchange of the factorbound ADP for free ATP induces the conformational change of the factor back to the low affinity form and results in dissociation of the complex into free apoenzyme and the ATP form of the factor. The free apoenzyme can be reconstituted to catalytically active holoenzyme with added AdoCbl, and the reactivating factor returns to the reactivation cycle by hydrolysis of ATP. The rate of ATP hydrolysis under reactivating conditions (0.9 min Ϫ1 ) is in the same range as the rate of mechanismbased inactivation of the holoenzyme by glycerol (0.35 min Ϫ1 ) (36). Therefore, the inactivation of the enzyme and its reactivation are not directly coupled but roughly synchronized. ATP plays dual roles in the reactivation as the precursor of ADP for cobalamin release and as the allosteric effector for dissociation of the enzyme-factor complex.
The following two assumptions are important for interpreting discrimination of the adenine-lacking cobalamin from adenine-containing cobalamin (Fig. 8B). (i) There is an equilibrium between enzyme-cobalamin complex plus reactivating factor (ADP form) and a cobalamin-enzyme-factor (ADP form) ternary complex, and (ii) release of cobalamin occurs from this ternary complex. If the enzyme-bound cobalamin is an adenine-lacking cobalamin, the equilibrium toward the right would be favored, and the cobalamin would be finally released from the enzyme. The enzyme would become more stabilized upon complex formation with the ADP form of factor than with an adeninelacking cobalamin. In contrast, if the enzyme-bound cobalamin is an adenine-containing cobalamin, the equilibrium would shift toward the left, and the cobalamin would be virtually unreleased. This difference could be explained by a larger binding energy with an adenine-containing cobalamin than that with an adenine-lacking cobalamin upon binding to the enzyme. In the case of diol dehydratase, it was shown from the K D value for adenine (37) that the presence of adenine moiety may decrease K eq by a magnitude of 1.7 ϫ 10 4 (19).
All of the data presented in this paper indicate that the complex of GdrA and GdrB serves as the reactivating factor for glycerol dehydratase by a mechanism similar to that of diol dehydratase-reactivating factor. Both of these factors bind to the target proteins and induce their conformational change through tight complex formation with them. This results in release of the modified coenzyme, and the reactivating factors themselves do not become a constituent of the final products. Thus, they meet the criteria of molecular chaperones. Their extremely low ATPase activity is not unusual for molecular chaperones (35). Three regions of the deduced amino acid sequence of the GdrA and DdrA subunits of the factors show high fragmentary similarity with those of the DnaK protein and the other Hsp70 group of molecular chaperones (17,38). These regions constitute the ADP-binding site of human Hsp70 (39,40). Therefore, the ATP/ADP-switching mechanism might be conserved between these reactivating factors for glycerol and diol dehydratases and the Hsp70 family of molecular chaperones.