Concurrent Chaperone and Protease Activities of ClpAP and the Requirement for the N-terminal ClpA ATP Binding Site for Chaperone Activity*

ClpA, a member of the Clp/Hsp100 family of ATPases, is both an ATP-dependent molecular chaperone and the regulatory component of ClpAP protease. We demonstrate that chaperone and protease activities occur concurrently in ClpAP complexes during a single round of RepA binding to ClpAP and ATP-dependent release. This result was substantiated with a ClpA mutant, ClpA(K220V), carrying an amino acid substitution in the N-terminal ATP binding site. ClpA(K220V) is unable to activate RepA, but the presence of ClpP or chemically inactivated ClpP restores its ability to activate RepA. The presence of ClpP simultaneously facilitates degradation of RepA. ClpP must remain bound to ClpA(K220V) for these effects, indicating that both chaperone and proteolytic activities of the mutant complex occur concurrently. ClpA(K220V) itself is able to form stable complexes with RepA in the presence of a poorly hydrolyzed ATP analog, adenosine 5′-O-(thiotriphosphate), and to release RepA upon exchange of adenosine 5′-O-(thiotriphosphate) with ATP. However, the released RepA is inactive in DNA binding, indicating that the N-terminal ATP binding site is essential for the chaperone activity of ClpA. Taken together, these results suggest that substrates bound to the complex of the proteolytic and ATPase components can be partitioned between release/reactivation and translocation/degradation.

The Clp/Hsp100 ATPase family of proteins are highly homologous and ubiquitous in nature (1). All members contain one or two very conserved ATP binding domains and function as ATPdependent molecular chaperones and/or as regulatory components of energy-dependent proteases (1,2). ClpA of Escherichia coli is a well studied example. It carries out the chaperone function of DnaJ and DnaK in the activation of the latent DNA binding activity of RepA, the plasmid P1 DNA replication initiator protein, by converting inactive dimers to active monomers (3,4). ClpA also protects proteins, including firefly luciferase and RepA, from irreversible heat inactivation in vitro (3). In addition, ClpA targets specific proteins for degradation by ClpP, a peptidase that alone is unable to degrade proteins, in an ATP-dependent reaction (5,6). ClpX, another E. coli Clp protein with chaperone activity, disassembles MuA transposase-DNA intermediates during the replication and transpo-sition of bacteriophage Mu DNA (7)(8)(9), disaggregates O aggregates in vitro (10), and activates the plasmid RK2 DNA replication initiator protein, TrfA, by dissociating dimers into monomers (11). ClpX is an alternate regulatory component for ClpP and targets the degradation of specific proteins, including MuA transposase and O (9,12,13). E. coli has two other Clp proteins: HslU (ClpY), the ATPase component of the HslUV (ClpYQ) protease (14,15), and ClpB, an ATPase involved in thermotolerance (16,17). The Saccharomyces cerevisiae homologue of ClpB, Hsp104, acts as a chaperone to dissolve protein aggregates formed during exposure to high heat (18 -20), to reactivate heat-denatured luciferase and RNA splicing enzymes (21), and to alter the conformation of a prion-like protein, [psi ϩ ] (22).
The Clp proteins self-assemble into oligomeric rings in the presence of ATP or nonhydrolyzable ATP analogs and many form stable complexes with the corresponding proteolytic component (23)(24)(25)(26)(27). For example, ClpA forms hexamers that bind at one or both ends of two stacked heptameric rings of ClpP (23,28). The crystal structure of ClpP shows that the proteolytic active sites are located within the chamber formed by the junction of two heptameric rings, resembling the structure of the eukaryotic 20 S proteasome (29). There are small axial pores at either end of ClpP, only large enough to allow short polypeptides or unfolded proteins to enter the proteolytic chamber. The archaebacterial and eukaryotic 26 S proteasomes have topology similar to that of ClpAP, suggesting a common mechanism of action despite little sequence similarity between the analogous components.
The contribution of each of the two ATP binding domains of ClpA to oligomerization, ATP hydrolysis, and proteolysis has been addressed by studies of ClpA proteins with mutations at the conserved lysine in one or the other of the ATP binding sites (30,31). The N-terminal site is involved in the assembly of ClpA hexamers, and the C-terminal site is involved in ATP hydrolysis and degradation of proteins. A functional C-terminal site is not required for propeptide degradation, a reaction whose nucleotide requirement is met by nonhydrolyzable ATP analogs (31).
From studies of the mechanism of the chaperone activity of ClpA and the proteolytic activity of ClpAP, the pathways of protein remodeling and degradation have emerged. During RepA activation, a stoichiometric complex of ClpA hexamers and RepA dimers forms in a reaction requiring nonhydrolyzable ATP analog (4). Upon exchange of the nonhydrolyzable ATP analog with ATP, RepA dimers are converted to monomers that can bind with high affinity to RepA-specific DNA binding sites. A single cycle of RepA binding to ClpA and ATP-dependent release from ClpA is sufficient to activate RepA. For protein degradation, ClpA⅐ClpP substrate complexes assemble in the presence of a nonhydrolyzable ATP * 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.
¶ To whom correspondence should be addressed: Bldg. 37 analog, by either ClpA-substrate complexes binding ClpP or ClpA⅐ClpP complexes binding substrates (32). Then in a reaction dependent on ATP hydrolysis, ClpA translocates specific substrates to ClpP for subsequent degradation. Degradation of some substrates, including RepA and ␣-casein, can be accomplished in a single round of substrate binding to ClpAP and ATP-dependent release of products that are identical to the steady state products.
The current model for the role of ATPase components or domains in ATP-dependent proteolysis proposes that the chaperone activity functions early in the pathway of degradation by unfolding or remodeling specific substrates and irreversibly damaged proteins to allow substrate translocation to the proteolytic chamber for proteolysis (1,33). The model also proposes that the ATPase components function in regulating proteolysis by remodeling and reactivating less severely damaged proteins and some specific substrates. Thus one prediction of this model is that the chaperone activity is manifest when the ATPase component is assembled in the proteolytic complex. A second prediction is that the chaperone activity is an essential intermediate step during proteolysis. The results presented in this study provide support for both of these predictions.
We demonstrate that ClpA, while in a complex with ClpP, is able to carry out concurrently chaperone and proteolytic activities. We also found, by characterizing ClpA derivatives with mutations in each of the two ATP binding sites, that the Nterminal site is essential for the chaperone activity.

EXPERIMENTAL PROCEDURES
Materials-ATP, ATP␥S, 1 and Triton X-100 were obtained from Roche Molecular Biochemicals. [␥-32 P]ATP was from ICN Pharmaceuticals, Inc. Sephacryl S-100 and S-200 high resolution gel filtration media were obtained from Amersham Pharmacia Biotech.
Gel Filtration Chromatography of RepA⅐ClpA Complexes-To isolate RepA⅐ClpA complexes, reaction mixtures (70 l) containing 500 pmol of [ 3 H]RepA, 100 pmol of ClpA or ClpA(K220V), and 1 mM ATP␥S in Buffer A containing 5% (v/v) glycerol were incubated for 15 min at 23°C. The reaction mixtures were separately loaded onto a Sephacryl S-200 HR column (0.7 cm x 7 cm) equilibrated with Buffer A containing 5% glycerol and 1 mM ATP␥S. Fractions (120 l) were collected, protein was measured by the Bio-Rad assay, and RepA was determined by measuring radioactivity by scintillation counting. To isolate RepA released from complexes, RepA⅐ClpA(wt) and RepA⅐ClpA(K220V) complexes were formed as described above with 60 pmol of [ 3 H]RepA and 120 pmol of ClpA or ClpA(K220V) and subjected to Sephacryl S-200 HR column chromatography as above. Fractions in the excluded volume were pooled. Next, 65 l of the pooled material was incubated in a 70-l reaction mixture with 10 mM ATP for 10 min at 23°C and applied onto a Sephacryl S-100 HR column (0.7 cm x 7 cm) equilibrated with Buffer A containing 5% glycerol and 1 mM ATP. Fractions (125 l) were collected, and radioactivity in each fraction was measured.

ClpA(K220V) Lacks Chaperone Activity but Degrades RepA in Conjunction with
ClpP-Our first evidence to support the hypothesis that the chaperone and proteolytic activities of ClpAP are concurrent came from the study of a ClpA mutant. This mutant, ClpA(K220V), has a valine substitution for the conserved lysine in the N-terminal ATP binding domain. Using activation of specific DNA binding by plasmid P1 RepA as a measure of chaperone activity, we found that ClpA(K220V) retained Ͻ5% of the wild-type ClpA ability to activate RepA for DNA binding (Fig. 1A).
Surprisingly, when we examined this chaperone defective ClpA mutant for its ability to promote RepA degradation by ClpP, we found that ClpA(K220V)P degraded RepA (Fig. 1B). The rate of RepA degradation by ClpA(K220V)P was slightly faster than by ClpA(wt)P. Consistent with its ability to degrade RepA, ClpA(K220V)P degraded ␣-casein at 60% the rate of wild-type ClpAP (Table III).
Because ClpA(K220V) was able to function in protein degradation, we wanted to test the efficiency of this mutant in translocating substrates to ClpP. We used a protocol that measures transfer of substrates from ClpA binding sites to ClpP without degradation (32). Using proteolytically inactive variants of ClpP, obtained by either chemical modification or sitedirected mutagenesis, [ 3 H]RepA⅐ClpA(K220V)P complexes were formed in the presence of ATP␥S. ATP was added in a 10-fold molar excess to ATP␥S to facilitate transfer of RepA to the inactive ClpP. The reaction mixtures were treated with 1 M NaCl to dissociate ClpA from ClpP, and then ClpP antibody was added to immunoprecipitate ClpP and any associated [ 3 H]RepA. Last, radioactivity in the immunoprecipitates was measured. When ClpA(K220V) was used in this assay, we found that RepA was transferred to inactive ClpP (Fig. 2, columns 1 and 2). ATP was required for the translocation (Fig.  2, column 3), and the extent of the reaction was similar to that seen with wild-type ClpA (Fig. 2, columns 1 and 2).
ClpA(K220V)P Is Able to Concurrently Activate and Degrade RepA-The observation that ClpA(K220V) was defective in chaperone activity but functional in proteolysis with ClpP was in apparent contradiction to the proposal that substrate unfolding or remodeling by the ATPase component is a prerequisite step in the pathway of degradation. The fact that ClpA(K220V) has the ability to transfer substrates to ClpP implied that ClpA(K220V) also has the ability to unfold substrates. The simplest explanation for this seeming discrepancy was that the presence of ClpP restored the chaperone activity of ClpA(K220V). To test this, we assayed RepA activation by ClpA(K220V) in the presence of ClpP. RepA⅐ClpA(K220V)P complexes were assembled in the presence of ATP␥S, ␣-casein was added to limit the reaction to a single cycle of RepA binding, and last, excess ATP was added. Under these conditions, we found that ClpA(K220V) was able to activate RepA in the presence of ClpP but not in its absence (Fig. 3A). ClpA(K220V) also activated RepA when chemically inactivated ClpP, Clp-P(in), was substituted for wild-type ClpP. In control experiments with wild-type ClpA, RepA was activated by ClpA alone, ClpAP, or ClpAP(in) (Fig. 3B). Thus, these results show that the chaperone defect in ClpA(K220V) is reversed by the interaction of ClpA(K220V) with ClpP.
To determine whether ClpP was necessary throughout the activation reaction or only to promote a functional conformation in RepA⅐ClpA(K220V) complexes, we assembled [ 3 H]RepA⅐ClpA(K220V)P complexes in the presence of ATP␥S. Next the complexes were treated with 1 M NaCl, a treatment known to dissociate ClpA⅐ClpP but not ClpA⅐RepA (32). After a short incubation, ATP was added in a large excess over ATP␥S, and activated RepA was measured. The results showed that ClpA(K220V), unlike ClpA, was not able to activate RepA after dissociation from ClpP (Fig. 4A). Control experiments showed that the NaCl treatment did not disrupt RepA⅐ClpA(K220V) or RepA⅐ClpA complexes (data not shown). RepA degradation from both RepA⅐ClpA(K220V)P and RepA⅐ClpAP complexes was abolished by NaCl treatment as expected, because ClpP was dissociated by the treatment (Fig. 4B). When NaCl was added after ␣-casein and ATP, ClpA(K220V)P complexes activated and degraded RepA, indicating that functional RepA⅐ClpA(K220V)P complexes were formed under these conditions (data not shown).
Thus, the initially surprising result that ClpA(K220V) lacks chaperone activity but has proteolytic activity with ClpP is explained by the ability of ClpP to overcome the chaperone defect of ClpA(K220V), very likely by stabilizing a conformation of ClpA that is essential for the chaperone activity. Thus, these results support the proposal that protein unfolding or remodeling needed for presentation of substrates to the proteolytic component occurs by a mechanism similar to that of protein remodeling involved in the chaperone activity.

Concurrent Chaperone and Protease Activities of Wild-type
ClpAP-Because ClpA(K220V) alone is defective in chaperone activity, the experiments presented in Fig. 3A demonstrate concurrent activation and degradation by ClpA(K220V)P. A similar conclusion could not be drawn from the results with ClpA(wt)P, because the possibility existed that activation was performed by RepA⅐ClpA, and degradation was performed by RepA⅐ClpAP. To determine whether ClpA(wt)P could simultaneously activate and degrade RepA, we first formed [ 3 H]RepA⅐ClpAP complexes using a 10-fold molar excess of ClpP to ClpA to ensure all of the ClpA was associated with ClpP. Then reaction mixtures were diluted 1:500 in the presence of excess ␣-casein. Last, ATP was added in excess of ATP␥S, and RepA activation and degradation were measured. We discovered that ClpAP is able to both activate and degrade RepA following just one round of RepA binding to ClpAP and ATP-dependent release (Fig. 5, column 1). The lack of activation and degradation when all components were added after dilution to the second reaction (Fig. 5, column 2) confirmed that single round binding conditions had been achieved. Similar amounts of activation and degradation were seen when the ratio of ClpP to ClpA was 5:1 or 1:1 (data not shown), indicating that all of the ClpA was complexed with ClpP. Thus the chaperone activity of wild-type ClpA can function in the proteolytic ClpAP complex and is not exclusively an activity of free ClpA.
Characterization of ClpA(K220V)-We further characterized ClpA(K220V) with the expectation that by understanding the step in the chaperone pathway that was blocked, we would better understand the mechanism of action of ClpA. To determine whether ClpA(K220V) could assemble into hexamers, we incubated ClpA(K220V) with or without ATP␥S and analyzed the samples by size exclusion column chromatography. ClpA(K220V), like ClpA, eluted as expected for a hexamer of 84-kDa protomers in the presence of nucleotide (Fig. 6). Elec-tron microscopy of ClpA(K220V) showed a preponderance of hexameric rings in the presence of ATP␥S. 2 Thus, ClpA(K220V) is not blocked in assembly. Measurement of ATPase activity showed that ClpA(K220V) hydrolyzed ATP at about 60% that of the rate of wild-type ClpA and was stimulated 2-fold by ClpP(in) ( Table I).
Because it is known that ATP promotes the release of RepA from RepA⅐ClpA complexes in a form active for DNA binding (4), we wanted to know if ClpA(K220V) was defective in its ability to release RepA. We isolated [ 3 H]RepA⅐ClpA(K220V)⅐ATP␥S complexes, incubated them with an excess of ATP over ATP␥S, and resolved free RepA from ClpA(K220V)⅐RepA complexes by gel filtration. We found that RepA was freed from ClpA(K220V) and eluted in the partially included volume (Table II). To see whether RepA released from RepA⅐ClpA(K220V) complexes was in its active or inactive form, we incubated a portion of the addition of ClpA (8 fmol of oriP1 DNA bound). As a control, RepA from RepA⅐ClpA(wt) complexes was active without the addition of ClpA (12 fmol of oriP1 DNA bound). Therefore, ClpA(K220V), like wild-type ClpA, is able to bind RepA and to release RepA in an ATP-dependent reaction. However, the RepA released from ClpA(K220V) is inactive, demonstrating that the defect in ClpA(K220V) is its inability to carry out the remodeling step. Finally, RepA does not appear to be modified while in complex with ClpA(K220V) because it can be activated by wild-type ClpA.
Characterization of the Chaperone and Protease Activities of Other ClpA Mutants-Two other ClpA mutants with different amino acids substituted for K220 were tested for chaperone activity. ClpA(K220R) was deficient in RepA activation, but the defect was less severe than that of ClpA(K220V) (Fig. 8A). RepA was degraded by ClpA(K220R)P at about 25% that of the rate of wild-type ClpAP (Fig. 8B). ClpA(K220Q), which is defective in assembly, ATP hydrolysis, and ␣-casein degradation (30) lacked chaperone activity (Fig. 8A) and did not promote RepA degradation with ClpP (Fig. 8B).
We also tested ClpA proteins mutated in the C-terminal ATP binding site. ClpA(K501R) and ClpA(K501Q) retained 60 -70% of the wild-type chaperone activity (Fig. 8C), although they had Ͻ10% of the wild-type ATPase activity (30). With ClpP, these two mutants degraded RepA at 15-25% of the wild-type rate (Fig. 8D). Thus a functional N-terminal ATP binding site, but not a C-terminal site, is essential for chaperone activity. DISCUSSION The results presented here point out an important new element in the model of energy-dependent proteolysis. They demonstrate that both chaperone and protease activities can occur concurrently in ClpAP complexes. Thus the fate of a substrate  ATPase activity was measured as described (30) in 50-l reaction mixtures with 4 mM [ 32 P]ATP, 0.5 g of ClpA, and 2 g of chemically inactivated ClpP as indicated.

ClpA ATPase Activity
ϪClpP(in) ϩClpP(in) mol/min/mg ClpA⅐RepA complexes RepA⅐ClpA and RepA⅐ClpA(K220V) complexes were formed in the presence of ATP␥S. The complexes were separated from free RepA by Sephacryl S200 gel filtration, incubated with ATP, and then applied onto a Sephacryl S100 column as described under "Experimental Procedures." Radioactivity in the column fractions was measured to assess RepA released from the complexes. Results are expressed as % of total radioactivity recovered from the S100 column (40% from ClpA⅐RepA and 30% from ClpA(K220V)⅐RepA complexes). In both experiments, about 20% of counts recovered from S100 columns were distributed elsewhere and not associated with protein peaks.
Complex applied to S100 column to be degraded instead of refolded is not simply a consequence of a proteolytic component associating with an ATPase component. Our working model for ClpAP is shown in Fig. 9. First, RepA⅐ClpAP complexes assemble in a reaction requiring ATP␥S. Exchange of the bound ATP␥S with ATP leads to substrate unfolding and remodeling by ClpA, and subsequently, some RepA is released as active monomers, and some is translocated to ClpP for degradation. In the absence of ClpP, RepA binds ClpA in a reaction requiring ATP␥S. Following ATP exchange, RepA is unfolded, remodeled, and released as active monomers. Although it is known that one ClpA 6 ClpP 14 interacts with one RepA dimer, 3 it remains to be determined if one RepA monomer is degraded and one is activated from each RepA⅐ClpAP complex. In our working model, the role of Clp ATPases in proteolysis is 3-fold: (i) recognizing and binding substrates (10,32,38), (ii) unfolding substrates, and (iii) translocating substrates to the proteolytic chamber of ClpP (32). Protein unfolding as a step in the pathway of degradation has yet to be directly demonstrated but has been postulated based on the finding that Clp ATPases have chaperone activity (3,10) and on the structural analysis of ClpP showing that the axial pores leading to the ClpP proteolytic chamber are only large enough to allow passage of a ␤ strand or an ␣ helix (29). Very likely, the chaperone activities of Clp ATPases also involve unfolded substrate intermediates. Thus, RepA may be released in an unfolded state that spontaneously refolds into an active monomer. Consistent with this, RepA is rapidly activated by treatment with guanidine⅐HCl followed by dilution of the denaturant. 4 The observation that a ClpA mutant, which is by itself devoid of chaperone activity, is able to facilitate proteolysis by ClpP initially appeared to contradict the working model. However, our finding that ClpP overcomes the chaperone defect of the mutant ClpA supports the proposal that substrate unfolding needed for presentation of substrates to the proteolytic component occurs by a mechanism similar to that of protein remodeling involved in the chaperone activity.
Cryo-electron microscopy has shown that the ClpAP complex contains two internal aqueous chambers in addition to the proteolytic cavity in ClpP. One chamber is present between the a Assays were carried out as described under "Experimental Procedures." b Data from Singh and Maurizi (30). c NT, not tested. two ATPase domains of ClpA, and another is formed by the junction of the proximal ATPase domain and ClpP (28). Electron micrographs of RepA⅐ClpAP complexes indicate that large conformational changes occur within the complex when ATP is added. 5 These changes appear to increase the sizes of the internal chambers and to widen the narrow channels connecting the ClpA chambers to the ClpP chamber. Such changes are consistent with our current working model for the mechanism of ClpAP and suggest a means by which unfolded substrates can be translocated to the proteolytic chamber without exposure to the surrounding medium.
It has recently been shown that ClpX is able to function as a chaperone while associated with ClpP (39). However, the results with ClpXP differ from those presented here with ClpAP in that the substrate for the chaperone activity is distinct from the substrate for proteolytic activity. Jones et al. (39) found that saturating levels of ClpP do not inhibit the ability of ClpX to perform its chaperone function of promoting MuA-dependent Mu transposition in vitro, although free MuA transposase is degraded by ClpXP. In fact, ClpP stimulates Mu transposition 2-3-fold without degrading MuA in the transpososome. In this case it is likely that the two substrates, MuA-DNA and free MuA transposase, are recognized differently by ClpXP and are consequently remodeled or degraded, respectively. Altogether, the studies from ClpAP and ClpXP suggest that concurrent chaperone and proteolytic activities may be a general function of proteolytic systems.
Our results showing that substitution of a valine for lysine at residue 220 of ClpA eliminates the ability of ClpA to remodel RepA imply that ATP binding in the first domain is necessary for the conformational changes of ClpA associated with protein modeling. The restoration of ClpA(K220V) chaperone activity and the increase in ATPase activity in the presence of ClpP suggest that interactions between ClpA(K220V) and ClpP lead to critical conformational changes in ClpA(K220V) that overcome the barrier in RepA activation. The elimination of ClpA(K220V) chaperone activity upon dissociation of ClpP from RepA⅐ClpA(K220V)P complexes by NaCl treatment further suggests that ClpP facilitates and stabilizes the formation of a conformation necessary for chaperone activity. Additional evidence for ClpP-mediated structural changes in ClpA as a general aspect of ClpAP interactions is provided by the observation that ClpP is able to stimulate the ATPase activities of other ClpA mutants in the first ATP binding site (30,31). Table III summarizes the properties ClpA derivatives containing mutations in the N-terminal and C-terminal ATP binding sites. Previous work had shown that mutations in the first ATP binding domain can interfere with ClpA hexamer assembly, and mutations in the second domain can impair ATPase and protease activities (30,31). The present results show that the first ATP binding site of ClpA is necessary for RepA activation, and the second site contributes only minimally. Ultimately, optimal chaperone activity of ClpA and proteolytic activity of ClpAP require some contribution from both ATP binding domains. Similar studies of Hsp104 derivatives containing mutations in the two nucleotide binding sites demonstrated that the first domain of Hsp104 contributes to the ATP hydrolytic activity, and the second domain contributes to assembly (40). This reversal of the roles does not indicate any novel distinction between ClpA and Hsp104. As seen for ClpA, although each domain of Hsp104 appears to have distinct ac-tivity, both domains have overlapping functions that are critical for optimal activity.
Altogether, our work demonstrates the versatility of proteases to function simultaneously in substrate remodeling and degradation. The present work also shows the conformational changes induced by ATP binding or hydrolysis in the N-terminal ATP binding domain constitute a crucial step toward ClpA chaperone activity.