Catapult mechanism renders the chaperone action of hsp70 unidirectional1

Article No. mb981815

J. Mol. Biol. (1998) 279, 833±840

Catapult Mechanism Renders the Chaperone Action of Hsp70 Unidirectional Serge M. Gisler, Ezra V. Pierpaoli and Philipp Christen* Biochemisches Institut UniversitaÈt ZuÈrich Winterthurerstr. 190 CH-8057, ZuÈrich, Switzerland

Molecular chaperones of the Hsp70 type promote the folding and membrane translocation of proteins. The interaction of Hsp70s with polypeptides is linked to ATP binding and hydrolysis. We formed complexes of seven different ¯uorescence-labeled peptides with DnaK, the Hsp70 homolog of Escherichia coli, and determined the rate of peptide release under two different sets of conditions. (1) Upon addition of ATP to nucleotide-free peptide
DnaK complexes, all tested peptides were released with similar rate constants (2.2 sÿ1 to 6.7 sÿ1). (2) In the binding equilibrium of peptide and ATP-liganded DnaK, the dissociation followed one or two-step reactions, depending on the amino acid sequence of the peptide. For the monophasic reactions, the dissociation rate constants diverged by four orders of magnitude from 0.0004 sÿ1 to 5.7 sÿ1; for the biphasic reactions, the rate constants of the second, slower isomerization step were in the range from 0.3 sÿ1 to 0.0005 sÿ1. The release of the different peptides in case (1) is 1.4 to 14,000 times faster than in case (2). Apparently, binding of ATP induces a transient state of the chaperone which ejects target peptides before the ®nal state of ATP-liganded DnaK is reached. This ``catapult'' mechanism provides the chaperone cycle with a mode of peptide release that does not correspond with the reverse of peptide binding. By allowing the conformation of the outgoing polypeptide to differ from that of the incoming polypeptide, a futile cycle with respect to conformational work exerted on the target protein is obviated. # 1998 Academic Press Limited

*Corresponding author

Keywords: Hsp70; DnaK; DnaJ; GrpE; chaperone cycle

Introduction DnaK and its two co-chaperones, DnaJ and GrpE, prevent proteins from aggregation and disentangle misfolded polypeptide chains (Hartl, 1996). DnaK has been found to occur in two conformations: the ATP-liganded, low-af®nity T state that rapidly binds and releases peptides and the ADP
Pi-liganded, high-af®nity R state that slowly interacts with peptides (Schmid et al., 1994; Pierpaoli et al., 1997). The chaperone cycle can be subdivided into three steps: (1) interaction of DnaK
ATP (T state) with a hydrophobic target sequence; (2) DnaJ-triggered conversion of peptide
DnaK
ATP to peptide
DnaK
ADP
Pi (R state); (3) GrpE-facilitated ADP/ATP exchange and S.M.G. and E.V.P. have contributed equally to this work. Abbreviations used: NEM, N-ethylmaleimide; acrylodan (a), 6-acryloyl-2-dimethylaminonaphthalene. 0022±2836/98/240833±08 $25.00/0

release of the polypeptide due to the binding of ATP to DnaK (Palleros et al., 1993; McCarty et al., 1995; Theyssen et al., 1996; Pierpaoli et al., 1997). Recently, we identi®ed the slow, DnaJ-triggered conformational change (T!R) as the power stroke of the DnaK/DnaJ/GrpE system underlying its chaperone effects (Pierpaoli et al., 1997). Here we measured the rate of peptide release triggered by the addition of ATP to nucleotide-free DnaK
peptide complexes or to ADP
DnaK
peptide complexes in the presence of the ADP/ATP exchange factor GrpE. The rates of peptide release under these conditions proved, depending on the amino acid sequence of the peptide, more than four orders of magnitude faster than the rates of dissociation of the peptide
DnaK
ATP complex under equilibrium conditions. The data indicate a forced, ATP-induced release of the peptide ligand in the chaperone cycle which allows the conformation of the released peptide to differ from that of the incoming peptide. # 1998 Academic Press Limited

834

Catapult Mechanism in Hsp70 s Chaperones

Results and Discussion Peptide release from nucleotide-free peptide
DnaK complexes upon addition of ATP We measured the rates of dissociation of seven different ¯uorescence-labeled target peptides (Table 1) from nucleotide-free peptide
DnaK complexes upon binding of ATP:

corresponding to equation (2) and (3), respectively (Figure 2 and Table 2): k‡1

! peptide ‡ DnaK
ATP ÿÿÿ ÿ peptide
DnaK
ATP kÿ1

…2†

R!T conformational change

peptide
DnaK ‡ ATP ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ ÿ! peptide ‡ DnaK
ATP

…1†

The reactions were initiated in a stopped-¯ow device by rapidly mixing ATP with preformed complexes of the labeled peptides and nucleotidefree DnaK, the conformation of which corresponds to that of ADP-liganded, i.e. R-state DnaK (Theyssen et al., 1996; Pierpaoli et al., 1997). All peptides were released with fast rates (Figure 1) ranging from 2.2 sÿ1 to 6.7 sÿ1 (Table 3, below). At the ATP concentration used (1 mM), the release of the peptides may be considered practically irreversible (Kd(ATP)7 nM; Theyssen et al., 1996). To correlate peptide release with the ATPinduced conformational alterations of DnaK, we recorded the changes in intrinsic ¯uorescence of nucleotide-free DnaK upon mixing with ATP. In the absence of peptides, the structural changes of DnaK followed two-step kinetics (Figure 1, inset) with kobs1 ˆ 12.6(0.4) sÿ1 and kobs2 ˆ 0.56(0.003) sÿ1. These results agree with those recently proposed by Slepenkov & Witt (1998). In the presence of the unlabeled peptides p6 or p4, similar rates were observed (see the legend to Figure 1). Obviously, the second ATP-induced conformational change of DnaK is too slow to account for the fast peptide release with koff values from 2.2 sÿ1 to 6.7 sÿ1. We therefore conclude that peptide dissociation is linked to the ®rst conformational change brought about by ATP. The discrepancy with the report of Theyssen et al. (1996) who found peptide release from DnaK to be coupled to the second step of ATP binding remains unexplained for the time being. Peptide association and release in the binding equilibrium of peptide and DnaK
ATP In the presence of ATP, complex formation between peptide and DnaK proved to be, depending on the peptide, single or double-step processes Table 1. Peptides used in this study Designationa a-pp a-p4 a-p5 a-p6 a-NR a-(LS)4 a-ps32

Sequence a-CALLQSRLLLSAPRRAAATARA a-CALLQSRLLS a-CLLLSAPRR a-CARSLLLSS a-NRLLLTG a-LSLSLSLS a-QRKLFFNLRKTKQ

a The peptides were ¯uorescence-labeled with acrylodan (Pierpaoli et al., 1997).

Figure 1. ATP-induced release of the acrylodan-labeled peptide a-p6 from its complex with DnaK. DnaK (2 mM) and a-p6 (100 nM) in assay buffer (25 mM HepesNaOH, 100 mM KCl (pH 7.0)) were incubated for four hours at 25 C and mixed in a stopped-¯ow device with an equal volume of assay buffer containing 2 mM ATP and 10 mM MgCl2. For details of ¯uorescence measurements, see Materials and Methods. Similar results were obtained for the ATP-induced release of peptides a-pp, a-p4, a-p5, a-NR, a-(LS)4 and a-ps32 (Table 3). All reaction traces could be ®tted to a single-exponential decay function. Inset: conformational changes of DnaK upon binding of ATP in the absence and presence of unlabeled peptide p6 with the free cysteine residue blocked with NEM. DnaK alone or DnaK plus saturating concentrations of p6 were incubated for ten minutes and four hours, respectively, in assay buffer at 25 C to allow formation of the p6
DnaK complex. The reactions were initiated by mixing in a stopped-¯ow apparatus equal volumes of DnaK or p6
DnaK with assay buffer containing 2 mM ATP and 10 mM MgCl2. The ®nal concentrations of p6 and DnaK were 20 mM and 1 mM, respectively. Both reactions could be described with double-exponential decay functions (kobs1 ˆ 12.6(0.4) sÿ1 and kobs2 ˆ 0.56(0.003) sÿ1 in the absence of p6; kobs1ˆ13.5(0.1) sÿ1 and kobs2 ˆ 0.67(0.03) sÿ1 in the presence of p6). The apparent difference in rates in the absence and presence of p6 is due to different amplitudes of the two phases under the two conditions, the ®rst phase contributing 35% and 94% of the total amplitude in the absence and presence of peptide p6, respectively. An almost coinciding curve was obtained for the reaction performed with 25 mM peptide p4 and 1 mM DnaK (kobs1 ˆ 15.2(0.1) sÿ1, kobs2 ˆ 0.88(0.04) sÿ1).

835

Catapult Mechanism in Hsp70 s Chaperones

k‡1

! peptide ‡ DnaK
ATP ÿÿÿ ÿ ‰peptide
DnaK
ATPŠ

GrpE-triggered peptide release in the complete DnaK chaperone system

kÿ1 k‡2

ÿÿÿ ! ÿ ‰peptide
DnaK
ATPŠ kÿ2

…3† The two-step reaction of equation (3) might arise from a process in which the peptide in the ®rst, faster step forms a loose encounter complex with DnaK. In the second, slower step, the peptide might become locked into a higher-af®nity complex after rearrangement of the peptide, DnaK or both (Hiromi, 1979; Schmid et al., 1994; Takeda & McKay, 1996). To determine the association and dissociation rate constants k‡1 and kÿ1, and the isomerization rate constants k‡2 and kÿ2, complex formation was performed at a constant concentration of peptides (100 nM) and varying concentrations of DnaK (1 mM to 22 mM). The dissociation rate constants kÿ1 in the case of a one-step reaction (equation (2)) and the isomerization rate constant kÿ2 in the case of a two-step reaction (equation (3)) depended markedly on the nature of the peptide (Table 2). The release of the tested peptides during the ATP-triggered conformational change according to equation (1) proved to be 1.4 to 14,000 times faster than these dissociation reactions in the equilibria of equations (2) and (3).

In the complete chaperone system that includes the co-chaperones DnaJ and GrpE, the peptide is released from the peptide
DnaK
ADP
Pi complex consequent to GrpE-facilitated ADP/ATP exchange:

…4†

Complexes of R-state DnaK with ¯uorescent peptides were preformed in the presence of DnaJ and ATP, and the release of the peptides after addition of GrpE was monitored with stopped-¯ow instrumentation. We used equimolar concentrations of DnaK and GrpE. GrpE acts as a dimer (Harrison et al., 1997) whereas DnaK is generally assumed to act as a monomer. Recently, GrpE has been shown to act in a catalytic manner, i.e. the full effect of GrpE is already attained at GrpE concentrations substoichiometric to those of DnaK (Pierpaoli et al., 1998). The reactions of all peptides were fast monophasic processes (Figure 3), with rate constants varying only from 0.9 sÿ1 to 2.1 sÿ1 (Table 3). The release of peptides during the GrpE-induced con-

Figure 2. Complex formation of DnaK with acrylodan-labeled peptides in the presence of ATP. A, The reactions with peptide a-p4, a-(LS)4 and a-NR were followed with a stopped-¯ow device. Equal volumes of [DnaK ‡ ATP] and [a-peptide ‡ ATP] were mixed to start the reaction. Protein and peptide solutions were prepared in assay buffer containing 1 mM ATP and 5 mM MgCl2. The ®nal concentrations were 100 nM a-p4 and 20 mM DnaK for the reaction (a-p4 ‡ DnaK
ATP), 100 nM a-NR and 20 mM DnaK for the reaction (a-NR ‡ DnaK
ATP) and 200 nM a-(LS)4 and 4 mM DnaK for the reaction (a-(LS)4 ‡ DnaK
ATP). B, Peptide a-p6 or a-ps32 was added to DnaK in assay buffer containing 1 mM ATP and 5 mM MgCl2, and the ¯uorescence was recorded with a spectro¯uorimeter. The ®nal concentrations in the reactions (a-p6 ‡ DnaK
ATP) and (a-ps32 ‡ DnaK
ATP) were 200 nM a-p6, 2 mM DnaK and 100 nM a-ps32, 10 mM DnaK, respectively. The reaction trace for a-p6 and a-(LS)4 was ®tted to a single-exponential function, that for a-p4, a-NR and a-ps32 to a double-exponential function. The ®nal values were normalized to be in approximately the same range. In the case of peptide a-ps32, the slower isomerization step results in a decrease in ¯uorescence. For details of kinetic ¯uorescence measurements, see Materials and Methods.

2.3(0.16) 7.9(0.17) 6.8(0.7) 0.001(0.000003) 5.9(0.16) 0.04(0.003) 0.017(0.0003)

a-ppg a-p4 a-p5h a-p6 a-NR a-(LS)4 a-ps32

0.001(0.00005)

0.02(0.005)

0.3(0.04)

kobs2 of 2nd phasea (sÿ1) 100 78 100 100 55 100 100

1st ampl. (% of total
F)

ÿ155i

45

22

2nd ampl. (% of total
F) 450,000 55,000 1,115,000 44 104,000 96 210

k‡1b (Mÿ1 sÿ1) 1.8 7.2 5.7 0.0004 4.5 0.04 0.013

kÿ1b (sÿ1)

0.001

0.016

0.8

k‡2c (sÿ1)

0.0005

0.018

0.3

kÿ2c (sÿ1)

62

43.3

131

Kÿ1d (mM)

0.48

1.2

0.38

Kÿ2e

2.2[4] 36.1(7) 5.1 8.3(2.1) [9.1] 23.8(3.8) 107(41) [417] 100(80)

Kdf (mM)

For details on the determination of k‡1, kÿ1, k‡2, kÿ2, Kÿ1, Kÿ2 and Kd, see Materials and Methods. a Final concentrations were: a-p4, a-p6, a-NR, a-(LS)4, a-ps32, 100 nM; DnaK, 10 mM. b The rate constants k‡1 and kÿ1 were determined by plotting kobs1 as a function of the concentration of DnaK (1 to 22 mM) at a constant concentration of labeled peptide (100 nM). c The rates k‡2 and kÿ2 were calculated from kobs2, Kÿ1 and Kd. d The equilibrium dissociation constant for the encounter complex Kÿ1 was calculated from the ratio kÿ1/k‡1. e Kÿ2 was determined from Kÿ1 and Kd. f The dissociation equilibrium constant Kd was determined by titration of acrylodan-labeled peptides with increasing concentrations of DnaK. The value in brackets corresponds to the dissociation equilibrium constant calculated from the ratio kÿ1/k‡1 in the case of single-exponential formation of the peptide
DnaK complex. g From Schmid et al. (1994). h From Pierpaoli et al. (1997). i In the case of peptide a-ps32, the ¯uorescence signal decreased in the second phase of complex formation (Figure 2B).

kobs1 of 1st phasea (sÿ1)

Peptide

Table 2. Kinetic rate constants and dissociation equilibrium constants of DnaK and target peptides in the presence of ATP

837

Catapult Mechanism in Hsp70 s Chaperones

version of DnaK to its T state was 1.3 to 5.5 times slower than the release of peptides from nucleotide-free DnaK upon addition of ATP. For ATP to become effective, GrpE must ®rst bind to the ATPase domain of R-state DnaK and open the domain, thereby facilitating the exchange of bound ADP against ATP (Buchberger et al., 1994; Harrison et al., 1997; Packschies et al., 1997). At the GrpE concentration used (Figure 3), ADP/ATP exchange apparently is somewhat slower than the binding of ATP to nucleotide-free DnaK (Packschies et al., 1997; Slepenkov & Witt, 1998). The non-coincidence of the rates of peptide release triggered by ATP (equations (1) and (4)) and the rates of the dissociation of peptide
DnaK
ATP (equilibrium situation; equations (2) and (3)) led to the following conclusions: (1) in the chaperone cycle, peptide release occurs during the

Figure 3. The release of acrylodan-labeled peptides from a-peptide
DnaK
ADP
Pi complexes induced by GrpEmediated ADP/ATP exchange. The reactions were monitored by following the decrease in acrylodan ¯uorescence. The two syringes of the stopped-¯ow device contained DnaK ‡ DnaJ ‡ a-peptide and GrpE, respectively, in assay buffer with 1 mM ATP and 5 mM MgCl2. Before mixing equal volumes of the reactants, the contents of the ®rst syringe were incubated for 15 minutes at 25 C to preform the a-peptide
DnaK
ADP
Pi complex. Final concentrations were: a-p4, a-p6, a-NR, aps32, 50 nM each; a-(LS)4, 100 nM; DnaK, DnaJ and GrpE, 1 mM each, with the exception of the experiment with a-(LS)4, in which the concentrations of DnaK, DnaJ, and GrpE were 2 mM. Data points from 0 to 0.1 seconds were omitted. All reactions followed single-exponential decay functions (continuous lines). All peptides used bind not only to DnaK but also to DnaJ (S.M.G., E.V.P., J.-J. SchoÈnfeld & P.C., unpublished results). Binding to DnaJ is at least one order of magnitude slower than the release from DnaK and results in a relatively weak ¯uorescence signal. Peptide binding to DnaJ may thus be assumed to interfere only insigni®cantly with the measurement of the rate of peptide release from DnaK.

GrpE-triggered conformational transition R!T rather than from the T state of DnaK, and (2) the conformation of DnaK from which the peptide is released during this conversion differs from that of the T state (Figure 4). The release of peptides during the ATP-induced R!T conversion is, depending on the peptide, accelerated up to 14,000-fold in comparison to the dissociation from T-state DnaK. Apparently, DnaK passes through an intermediate state T* in which hydrophobic interactions and hydrogen bonds to the target peptide are weakened. The closely similar koff values for all peptides suggest that the binding interactions between target peptides and DnaK (Zhu et al., 1996) in the T* conformation might be

Table 3. Rate constants for the ATP or ATP/GrpE-induced release of peptides from DnaK and for the dissociation of T-state DnaK
peptide complexes Peptide

Addition of ATPa koff (sÿ1)

Addition of ATP‡GrpEb koff (sÿ1)

Peptide
DnaK
ATP complexesc kÿ1, kÿ2 (sÿ1)

a-pp a-p4 a-p5 a-p6 a-NR a-(LS)4 a-ps32

2.6( 0.26) 5.5(1.3) 6.7(1.2) 5.6(0.035) 5.4(0.0057) 2.4(0.9) 2.2(0.01)h

NDd 1.0(0.05) 1.3(0.02)e 1.5(0.01) 2.1(0.01) 1.9(0.1) 0.9(0.004)

1.8 0.3 5.7 0.0004 0.018 0.04 0.0005

a ATP was rapidly mixed with preformed a-peptide
DnaK complex and the reactions monitored as described in the legend to Figure 1. Final concentrations were: a-pp, a-p4, a-p5, a-NR (50 nM), DnaK (1 mM); ATP (1 mM) and a-(LS)4 (100 nM), DnaK (2 mM); ATP (1 mM). Measurements in the presence of a 150-fold molar excess of unlabeled peptide (NEM-blocked p4 and p6, p5 with Nterminal alanine instead of cysteine, and NR) gave on average 1.15-fold higher rates. b Changes in acrylodan ¯uorescence were followed with a stopped-¯ow device. For details, see the legend to Figure 3. The conversion of the transient conformational intermediate T* (see Figure 4) back to the R state of DnaK may be neglected because of the high ATP:ADP ratio. The apparent rate constants may thus be assumed to correspond to the dissociation rate constants koff. c Values are taken from Table 2. For peptides a-pp, a-p5, a-p6, a-(LS)4, the indicated values refer to kÿ1 for the single-step reaction, for peptides a-p4 and a-NR they refer to the isomerization rate kÿ2 of the two-step reaction. d ND, not determined. e Final concentrations were: a-p5, 50 nM; DnaK, DnaJ and GrpE, 1 mM each.

838

Catapult Mechanism in Hsp70 s Chaperones

reduced to interactions common to all peptides, conceivably main-chain hydrogen bonds. Access of water molecules into the opened peptide-binding cleft might render these hydrogen bonds energetically less effective for binding and result in ejection of the peptide segment.

Conclusion The proposed ``catapult'' mechanism (Figure 4) provides the chaperone cycle with an alternative pathway for peptide release (equation (4)) that does not correspond to the reverse of peptide binding (equations (2) and (3)). An analogous release mechanism for target proteins has also been postulated, though not directly observed for GroE, the Hsp60 homolog of Escherichia coli (Ranson et al., 1997). Apparently, the unidirectionality of the cycle, imposed by ATP hydrolysis, pertains also to peptide binding and release. The conformation of the outgoing polypeptide, the product of the chaperone cycle, may differ from that of the incoming target polypeptide, thus obviating a futile cycle with respect to target polypeptide conformation. The catapult mechanism may thus allow to maintain the result of conformational work performed during the chaperone cycle on the target peptide (Pierpaoli et al., 1997), e.g. a disentangled structure. In addition, this release mechanism prevents blockage of Hsp70 by polypeptides that would only slowly dissociate from T-state DnaK (equations (2) and (3)).

Materials and Methods Reagents ATP-Na2, (purity598%) was purchased from Fluka. NEM was from Fluka and acrylodan was from Molecular Probes, Eugene, OR, USA. Peptides NR (NRLLLTG) and ps32 (QRKLFFNLRKTKQ) were bought from ANAWA Wangen, Switzerland (purity>80%). Peptides p4 (CALLQSRLLS), p6 (CARSLLLSS) and (LS)4 (LSLSLSLS) were synthesized by Dr S. Klauser in our Institute with an ABI 430 A Peptide Synthesizer (Applied Biosystems) with the orthogonal Fmoc protection strategy. Proteins DnaK was obtained from a yeast expression plasmid and puri®ed as described (Schmid et al., 1994). The nucleotide content was <0.1 mol per mol DnaK (Feifel et al., 1996). GrpE and DnaJ were prepared as reported (SchoÈnfeld et al., 1995a,b). Concentration of nucleotidefree DnaK was determined photometrically with a molar absorption coef®cient of e280ˆ14,500 Mÿ1 cmÿ1 (Helleburst et al., 1990). The concentrations of DnaJ and GrpE were determined by quantitative amino acid analysis. All concentrations of DnaK and the co-chaperones refer to their protomers. The stock solution of DnaJ was 120 mM in 50 mM Tris-HCl (pH 7.7), 100 mM NaCl, that of GrpE was 240 mM in 50 mM Tris-HCl (pH 7.7). The preparations of DnaK and both co-chaperones were more than 95% pure

Figure 4. Hypothetical model of Hsp70 action for the bacterial DnaK/DnaJ/GrpE molecular chaperone system. (1) The fast-binding and releasing ATP-liganded state of DnaK (T state) binds an entangled hydrophobic segment of a misfolded protein or an aggregated polypeptide. (2) The co-chaperone DnaJ triggers the hydrolysis of DnaK-bound ATP (Liberek et al., 1991). This slow conformational change represents the power stroke of the system (Pierpaoli et al., 1997). The entangled polypeptide segment is stretched and accommodated into the peptide binding cavity of DnaK. In the resulting ADP-state of DnaK (R state), the peptide has adopted an extended conformation (Landry et al., 1992; Zhu et al., 1996). (3) GrpE facilitates the exchange of ADP and Pi against ATP. Upon binding of ATP, DnaK reaches its transient T* state in which the lid of the peptide-binding site is open (Zhu et al., 1996), and interactions between peptide and DnaK are reduced in essence to hydrogen bonds of the main-chain. Hydrophobic contacts might be decreased due to a narrowing of the hydrophobic pocket in the peptide-binding cleft (Zhu et al., 1996). (4) Passage through the intermediate T* state results in ejection of the disentangled polypeptide segment.

as estimated by SDS-PAGE. All three proteins were stored at ÿ80 C. Labeling of peptides with acrylodan The same procedure as described by Schmid et al. (1994) and Pierpaoli et al. (1997) was applied to label all peptides with acrylodan. Peptides a-NR, a-(LS)4 and a-ps32 without cysteine were labeled at their a-amino group by the same procedure. The concentrations of the peptide stock solutions, determined by quantitative amino acid analysis, were in the range of 6 to 240 mM. For blocking of the free cysteine residue of peptides p4 and p6 with NEM, p4 (6.4 mM or p6 (7 mM) were reacted for two hours at room temperature with a 30-fold molar excess of dithiothreitol in 3 ml 50 mM Tris-HCl (pH 8.0) to reduce possibly oxidized cysteine residues. The reduced peptides were separated from excess dithiothreitol by size exclusion chromatography (G-10; Pharmacia) in 1 M acetic acid. Peptide containing fractions, detected at 226 nm, were lyophilized and stored at ÿ20 C in an argon atmosphere to prevent reoxidation. For the blocking of cysteine residues, NEM (90 mM) was dissolved in 3.5 ml 50 mM Hepes-NaOH

839

Catapult Mechanism in Hsp70 s Chaperones (pH 7.0) and peptide (p4; 1.6 mM; p6, 1.8 mM) was added and incubated for one hour at room temperature. The samples were applied onto a C8-reverse-phase HPLC-column (Applied Biosystems), equilibrated with 10% (v/v) acetonitrile, 0.1% (v/v) tri¯uoroacetic acid to separate the NEM-labeled peptides from unmodi®ed and doubly conjugated peptides (alkylated at both cysteine and the a-amino group) in a linear gradient up to 80% acetonitrile, 0.085% tri¯uoroacetic acid. The ef¯uent was monitored at 225 nm. NEM-p4 and NEM-p6 eluted at 33% and 27% acetonitrile, respectively. Fractions containing the labeled peptide were pooled, evaporated to dryness and taken up in 0.5 ml of 30% acetonitrile. The identity of both products was veri®ed by mass spectrometry. The concentrations, determined by quantitative amino acid analysis, were 2.1 mM and 3.5 mM, respectively. Fluorescence measurements All experiments were performed at 25  C in assay buffer (25 mM Hepes-NaOH, 100 mM KCl (pH 7.0)), or, if explicitely stated, in assay buffer containing 0.5 to 2 mM ATP and 5 mM MgCl2. In all experiments, the samples were equilibrated for at least ®ve minutes at 25 C prior to measurement. Slow kinetic measurements Slow kinetic experiments and steady-state experiments were performed either with a SPEX FLUOROLOG spectro¯uorimeter or with a Perkin-Elmer LS-50B spectro¯uorimeter. To measure complex formation of acrylodan-labeled peptides with DnaK, the excitation wavelength was set at 370 nm (band pass 4.6 nm) and the emission at 500 nm (band pass 2.25 nm) was recorded. The spectra of acrylodan-labeled peptides or of the peptide
DnaK complexes were scanned from 430 to 610 nm. The solutions were mixed with the tip of the pipette or a glass rod. With some peptides, adsorption of the peptide to the stirrer and the cuvette proved a problem which was substantially reduced by using a magnetic stirrer and a 1 cm0.4 cm cuvette (800 ml, from Helma, Switzerland) with a special stirrer compartment. Because adsorption effects increased with longer mixing times, stirring was started few seconds before adding the protein solution and then continued for only about ten seconds. Adsorption was further minimized by covering the magnetic stirrer with glass instead of te¯on and by pre-incubating the magnetic stirrer in the peptide solution for ten minutes. Reactions were initiated by adding the protein or the peptide solution with a microliter glass syringe (Hamilton, Switzerland) to the magnetically stirred peptide or protein solution. The dead time of this mixing system is approximately three seconds. The ¯uorescence spectra were integrated with the software provided by SPEX. Data were analyzed with Sigma Plot (Jandel Scienti®c). Fast kinetic measurements Rapid changes in ¯uorescence were recorded with a SF-61 stopped-¯ow spectro¯uorimeter (HI-TECH Scienti®c, Salisbury) or a SX17 MV stopped-¯ow spectro¯uorimeter (Applied Biophysics), both with a dead time of 1 ms. Reactions were initiated by mixing equal volumes (70 ml). The width of the entrance slit of the

monochromator was set at 2 mm (Trp) or 4 to 5 mm (acrylodan) and that of the exitation slit at 3 mm (Trp) or 5 mm (acrylodan). The sample was excited at 290 nm (Trp) or 370 nm (acrylodan). Emission was measured with a WG 320 nm or a GG 455 nm cut-off ®lter. Determination of dissociation equilibrium constants Kd and evaluation of kinetic measurements The dissociation equilibrium constants Kd of acrylodan-labeled peptides and DnaK were determined by titration of 50 or 100 nM peptide with increasing concentrations of DnaK or DnaJ in the range from 25 nM to 20 mM. Kd values were calculated by plotting the differences of the areas of the ¯uorescence spectra from 440 nm to 600 nm, i.e.
A ˆ (Apeptide
DnaK complex ÿ Apeptide)  dilution factor, as a function of the total protein concentration. To attain values for Kd and
Amax, the points were ®tted to the equation:
A ˆ PL 
Amax =Pt ˆ ‰
Amax =…2  Pt †Š  ‰…Kd ‡ Lt ‡ Pt † ÿ ……Kd ‡ Lt ‡ Pt †2 ÿ 4  Pt  Lt †0:5 Š; where PL denotes the concentration of a-peptide
DnaK complexes, Pt the total protein concentration (DnaK) and Lt the total concentration of the peptide ligand. Kd and
Amax were chosen as parameters for the curve ®tting. Kinetic measurements were described by either a single or a double-exponential equation. The program provides for each evaluation the corresponding asymptotic standard deviation of the calculated parameters. In the case of stopped-¯ow experiments, the average of at least four measurements was analyzed with the software supplied by HI-TECH or Applied Biophysics. Determination of the reaction rate constants k‡1, kÿ1, k‡2, kÿ2 and the dissociation equilibrium constants Kÿ1, Kÿ2 To determine the rate constants k‡1 and kÿ1 for complex formation between DnaK and peptides (see equation (2)) in the presence of ATP, the apparent rate constants kobs were measured with 100 nM labeled peptide and 1 mM to 22 mM DnaK. As expected for a second-order reaction, the kobs values increased linearly, dependent on the concentration of DnaK. The association rate constants k‡1 and the dissociation rate constants kÿ1 were determined from the slope and the intercept of the plot of kobs1 as a function of DnaK concentration, respectively. From k‡1 and kÿ1 the dissociation equilibrium constant Kÿ1ˆkÿ1/k‡1 was calculated, which corresponds to Kd in the case of one-step reactions. In the case of twostep reactions (see equation (3)), Kÿ1 for the encounter complex was in the range from 40 mM to 130 mM (Table 3); therefore, the apparent rate constant kobs2 did not signi®cantly increase in the concentration range used for DnaK. The rates k‡2 and kÿ2 for the two-step reactions were thus directly calculated from kobs2, Pt, Kÿ1 and Ks according to the following equations (Hiromi, 1979): K‡2 ˆ …Kÿ1 ÿ Kd †=Kd ; kÿ2 ˆ kobs2  …Kÿ1 ‡ Pt †=…K‡2  Pt ‡ Kÿ1 ‡ Pt †; where Pt ˆ ‰DnaKŠ k‡2 ˆ K‡2  kÿ2

840

Catapult Mechanism in Hsp70 s Chaperones

In the case of the one-step reaction with peptide a-p5, k‡1 and kÿ1 were calculated from kobs1 and Kd with the equations k‡1 ˆ kobs1/([DnaK] ‡ Kd) and kÿ1 ˆ Kd  k‡1. Protein concentrations were at least tenfold higher than the concentrations of the peptide to ensure pseudo-®rstorder conditions.

Acknowledgments We are grateful to Heinz Gehring for helpful comments on the manuscript. DnaJ and GrpE were a kind gift from Hans-Joachim SchoÈnfeld, Hoffmann-La Roche, Basel. We thank Bastian Feifel for nucleotide-free DnaK; Peter Hunziker, Ragna Sack and Peter Gehrig for quantitative amino acid analyses as well as for mass spectra; Bastiaan van Wieringen for technical assistance. This work was supported in part by the Swiss National Science Foundation (grant no. 31-45940.95), the EMDOStiftung, ZuÈrich, the Fonds fuÈr medizinische Forschung der UniversitaÈt ZuÈrich and the Ernst-GoÈhner Stiftung, ZuÈrich.

References Buchberger, A., SchroÈder, H., BuÈttner, M., Valencia, A. & Bukau, B. (1994). A conserved loop in the ATPase domain of the DnaK chaperone is essential for stable binding of GrpE. Nature Struct. Biol. 1, 95 ± 101. Feifel, B., Sandmeier, E., SchoÈnfeld, H.-J. & Christen, P. (1996). Potassium ions and the molecular-chaperone activity of DnaK. Eur. J. Biochem. 237, 318 ±321. Harrison, C. J., Hayer-Hartl, M., Liberto, M. D., Hartl, F.-U. & Kuriyan, J. (1997). Crystal structure of the nucleotide exchange factor GrpE bound to the ATPase domain of the molecular chaperone DnaK. Science, 276, 431± 435. Hartl, F. U. (1996). Molecular chaperones in cellular protein folding. Nature, 381, 571± 580. Helleburst, H., UhleÂn, M. & Enfors, S. O. (1990). Interaction between heat shock protein DnaK and recombinant staphylococcal protein A. J. Bacteriol. 172, 5030± 5034. Hiromi, K. (1979). Kinetics of Fast Enzyme Reactions, Halsted Press, New York. Landry, S. J., Jordan, R., McMacken, R. & Gierasch, L. M. (1992). Different conformations for the same polypeptide bound to chaperones DnaK and GroEL. Nature, 355, 455± 457. Liberek, K., Marszalek, J., Ang, D., Georgopoulos, C. & Zylicz, M. (1991). Escherichia coli DnaJ and GrpE heat shock proteins jointly stimulate ATPase activity of DnaK. Proc. Natl Acad. Sci. USA, 88, 2874± 2878.

McCarty, J. S., Buchberger, A., Reinstein, J. & Bukau, B. (1995). The role of ATP in the functional cycle of the DnaK chaperone system. J. Mol. Biol. 249, 126± 137. Packschies, L., Theyssen, H., Buchberger, A., Bukau, B., Goody, R. S. & Reinstein, J. (1997). GrpE accelerates nucleotide exchange of the molecular chaperone DnaK with an associative displacement mechanism. Biochemistry, 36, 3417± 3422. Palleros, D. R., Reid, K. L., Shi, L., Welch, W. J. & Fink, A. L. (1993). ATP-induced protein-Hsp70 complex dissociation requires K‡ but not ATP hydrolysis. Nature, 365, 664± 666. Pierpaoli, E. V., Sandmeier, E., Baici, A., SchoÈnfeld, H.J., Gisler, S. & Christen, P. (1997). The power stroke of the DnaK/DnaJ/GrpE molecular chaperone system. J. Mol. Biol. 269, 757± 768. Pierpaoli, E. V., Sandmeier, E., SchoÈnfeld, H.-J. & Christen, P. (1998). Control of the DnaK chaperone cycle by substoichiometric concentrations of the cochaperones DnaJ and GrpE. J. Biol. Chem. 273, 6643± 6649. Ranson, N. A., Burston, S. G. & Clarke, A. R. (1997). Binding, encapsulation and ejection: substrate dynamics during a chaperonin-assisted folding reaction. J. Mol. Biol. 266, 656± 664. Schmid, D., Baici, A., Gehring, H. & Christen, P. (1994). Kinetics of molecular chaperone action. Science, 263, 971± 973. SchoÈnfeld, H.-J., Schmidt, D., SchroÈder, H. & Bukau, B. (1995a). The DnaK chaperone system of Escherichia coli: quartenary structures and interactions of the DnaK and GrpE components. J. Biol. Chem. 270, 2183± 2189. SchoÈnfeld, H.-J., Schmidt, D. & Zulauf, M. (1995b). Investigation of the molecular chaperone DnaJ by analytical ultracentrifugation. Progr. Colloid. Polym. Sci. 99, 7 ±10. Slepenkov, S. V. & Witt, S. N. (1998). Kinetics of the reactions of the Escherichia coli molecular chaperone DnaK with ATP: evidence that a three-step reaction precedes ATP hydrolysis. Biochemistry, 37, 1015± 1024. Takeda, S. & McKay, D. B. (1996). Kinetics of peptide binding to the bovine 70 kDa heat shock cognate protein, a molecular chaperone. Biochemistry, 35, 4636± 4644. Theyssen, H., Schuster, H.-P., Packschies, L., Bukau, B. & Reinstein, J. (1996). The second step of ATP binding to DnaK induces peptide release. J. Mol. Biol. 263, 657± 670. Zhu, X., Zhao, X., Burkholder, W. F., Gragerov, A., Ogata, C. M., Gottesman, M. E. & Hendrickson, W. A. (1996). Structural analysis of substrate binding by the molecular chaperone DnaK. Science, 272, 1606± 1614.

Edited by A. R. Fersht (Received 30 October 1997; received in revised form 19 March 1998; accepted 27 March 1998)