Binding and degradation of heterodimeric substrates by ClpAP and ClpXP.

ClpA and ClpX function both as molecular chaperones and as the regulatory components of ClpAP and ClpXP proteases, respectively. ClpA and ClpX bind substrate proteins through specific recognition signals, catalyze ATP-dependent protein unfolding of the substrate, and when in complexes with ClpP translocate the unfolded polypeptide into the cavity of the ClpP peptidase for degradation. To examine the mechanism of interaction of ClpAP with dimeric substrates, single round binding and degradation experiments were performed, revealing that ClpAP degraded both subunits of a RepA homodimer in one cycle of binding. Furthermore, ClpAP was able to degrade both protomers of a RepA heterodimer in which only one subunit contained the ClpA recognition signal. In contrast, ClpXP degraded both subunits of a dimeric substrate only when both protomers contained a recognition signal. These data suggest that ClpAP and ClpXP may recognize and bind substrates in significantly different ways.

In the presence of ATP, Clp ATPases self-assemble into oligomeric rings. The individual subunits are markedly similar in structure to the subunits of classic AAAϩ ATPases (10 -13). A Clp protease resembling the eukaryotic 26 S proteasome is formed when the oligomeric ATPase rings associate with one or both ends of a proteolytic component (14). Crystal structures of the proteolytic components ClpP and HslV reveal that the proteolytic sites are within an internal chamber formed by two stacked rings of identical peptidase subunits, resembling the 20 S proteolytic core of the proteasome (10,15,16). Access to the proteolytic chamber appears to be limited to narrow pores at either end of the stacked rings, which are not large enough to allow passage of a native globular protein. The current mechanistic model proposes that to gain entry to the proteolytic chamber, native substrates are first specifically bound and unfolded by the ATPase components flanking the proteolytic core. The unfolded polypeptide is then translocated through the small pores into the proteolytic chamber (17,18). Studies demonstrating that ClpA and ClpX bind substrates specifically and catalyze ATP-dependent protein unfolding provide support for this model (19 -21). Additionally, ClpA and ClpX translocate substrates in an ATP-dependent reaction from binding sites on the ATPase component to ClpP in a directional manner, providing further experimental evidence for this model (20 -24).
In some instances Clp proteases require specificity factors (or adaptor proteins) for recognition of particular substrates. These adaptors act by specifically facilitating the interaction of a Clp ATPase with a substrate that has low affinity for the Clp ATPase. For example, S , the E. coli stationary phase sigma factor, is not detectably degraded by ClpXP alone. However, the RssB adaptor interacts with both ClpX and S , thereby delivering S to ClpXP for degradation, although not being degraded itself (25). Two other ClpX specificity factors, SspB and UmuD, and one ClpA specificity factor, ClpS, have also been well characterized (26 -32).
Substrate recognition and binding by the ATPase component is the initial step in protein remodeling and degradation by Clp chaperones and Clp proteases, respectively. Generally, although not exclusively (33), ClpA and ClpX recognize substrates through short signals located very near either the N or C terminus of the polypeptide. Systematic analysis of more than 50 ClpX substrates identified five signal motifs ranging in size from 3 to 6 amino acids (34). Two of these motifs are C-terminal, and the other three are located at the N terminus of substrate proteins. Additionally, the specific interactions between substrate and Clp ATPase have been examined in great detail for several substrates.
One well characterized ClpA substrate is RepA, the P1 plasmid initiator protein. ClpA binds inactive RepA dimers and, in a reaction requiring ATP-dependent unfolding, converts dimers into active monomers that can then bind oriP1 DNA (3). However, when ClpA is associated with ClpP, ClpA targets RepA for degradation. The recognition signal that directs RepA to ClpA is located within the first 15 amino acids of RepA. This signal is both necessary and sufficient to target a protein fused to the peptide for unfolding by ClpA and degradation by ClpAP (35).
Another well characterized ClpA and ClpX recognition signal is SsrA. A polypeptide is tagged for degradation by the addition of an 11-amino acid peptide encoded by a small RNA, ssrA, when an mRNA lacking an in-frame stop codon stalls on the ribosome (36,37). The SsrA peptide is added to the C terminus of a nascent polypeptide chain by cotranslational switching of the ribosome from the damaged mRNA to ssrA RNA. This signal then targets the abnormal protein for degradation. In E. coli, both ClpAP and ClpXP degrade SsrA-tagged proteins (36, * 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 38,39). A study comparing SsrA-tagged proteins of different known stabilities revealed that the efficiency of degradation by ClpXP is determined largely by the presence of the tag, not by the intrinsic stability of the attached protein (21,40,41). It was shown further that when ClpXP interacts with heterodimeric substrates in which only one protomer is SsrA-tagged, only the subunit with the recognition signal is degraded (40).
The goals of the present study were to elucidate further the mechanisms of Clp-substrate interactions. We investigated the fate of individual subunits of a dimeric substrate interacting with the ClpAP protease to determine whether both protomers may be degraded during one round of binding. The interactions of ClpAP and ClpXP with heterodimeric substrates were also examined. Our findings indicate that the closely related proteases ClpAP and ClpXP may have fundamentally different substrate recognition, binding, and processing mechanisms.

EXPERIMENTAL PROCEDURES
Materials-ATP and ATP␥S 1 were obtained from Roche Applied Science. Restriction endonucleases and DNA-modifying enzymes were obtained from New England BioLabs. PCR reagents were obtained from PerkinElmer Life Sciences.
Plasmids and Strains-To construct pET-RepA(⌬25), a repA PCR fragment corresponding to RepA amino acids 26 -286, and flanked with NdeI and HindIII sites at the 5Ј-and 3Ј-end, respectively, was cloned between the NdeI and HindIII sites of pET24B (Novagen). pET-RepA(⌬25)SsrA was constructed by generating a repA PCR fragment flanked with NdeI and HindIII sites at the 5Ј-and 3Ј-ends, respectively, in which the 3Ј-oligonucleotide incorporated the coding region for the 11-amino acid SsrA tag followed by a TAA stop codon. This PCR fragment was cloned between the NdeI and HindIII sites of pET24B. pET-RepAhis was constructed by synthesizing a repA PCR fragment lacking the stop codon and flanked at the 5Ј-and 3Ј-ends by NdeI and HindIII sites, respectively. The PCR fragment was then cloned between the NdeI and HindIII sites of pET24b, such that a C-terminal His 6 tag was added to RepA. All mutations were confirmed by DNA sequencing.
Preparation of RepAhis:RepA⌬25 and RepAhis:RepA(⌬25)SsrA Heterodimers-RepAhis and either RepA(⌬25) or RepA(⌬25)SsrA were mixed together in a 1:5 ratio in 1ϫ phosphate-buffered saline, 100 mM NaCl, 10% glycerol (v/v), and 6 M guanidine HCl. After 15 min at 24°C, the denatured protein mixture was dialyzed for four h at 4°C against the same buffer but lacking guanidine HCl. Heterodimers were isolated by applying the mixture to an immobilized metal affinity chromatography column (Talon) following the manufacturer's procedure (Clontech). The columns were washed successively with 20, 40, 80, and 100 mM imidazole. The purified heterodimers, eluting with 80 and 100 mM imidazole, were stored at Ϫ70°C.   Fig. 3, 4, and 7. Trichloroacetic acid was added to 20% (w/v). The trichloroacetic acid pellets were subjected to SDS-PAGE and transferred onto nitrocellulose membranes (Invitrogen) by electroblotting. RepA and its derivatives were detected with specific rabbit antiserum using a Western blot immunodetection kit (Novex Western Breeze, Invitrogen) and the results quantified by densitometry.
Single Round Degradation Assay-Reaction mixtures (100 l) containing buffer A, 10 mM MgOAc, 2 mM ATP␥S, 0.1 mg/ml bovine serum albumin, 2.4 M ClpA, 5.5 M ClpP, and 1 M RepA-GFP or 4.6 M RepA(1-70)-GFP were incubated for 15 min at 23°C. Substrate-ClpAP complexes were isolated by gel filtration on a Sephacryl S-200 column (0.7 ϫ 17 cm) equilibrated with 20 mM Tris-HCl, pH 7.5, 100 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, 0.05 mg/ml bovine serum albumin, 5 mM MgOAc, and 0.25 mM ATP␥S. Fractions containing substrate-Cl-pAP complexes were pooled (450 l). ␣-Casein (120 g), 6 mM ATP, 10 mM MgOAc, and an ATP-regenerating system (in 26 l) were added to one aliquot of 150 l, and 26 l of buffer A was added to another aliquot. Degradation was measured by monitoring loss of GFP fluorescence at 23°C for 5 min. To measure the effectiveness of ␣-casein in inhibiting RepA-GFP degradation by ClpAP, complexes of ClpAP were generated and isolated as above, but without substrate. RepA-GFP (2.5 g, an amount equivalent to that present in the degradation reactions described above), 6 mM ATP, 10 mM MgOAc, and an ATP-regenerating system were added in the presence or absence of 120 g of ␣-casein to 150-l aliquots of ClpAP complexes. Degradation was measured by monitoring the loss of GFP fluorescence at 23°C for 5 min.

ClpAP Degrades Both Subunits of a Dimer in One Round of
Binding and Proteolysis-According to the current mechanistic model of ClpAP, a recognition signal on the substrate is bound by ClpA. After binding, ClpA unfolds the substrate in an ATPdependent reaction, then translocates it to ClpP in a second ATP-dependent reaction. The substrate is degraded upon translocation into the proteolytic chamber of ClpP. An aspect of the ClpAP mechanism that remains unclear is the fate of the individual subunits of a multimeric substrate. After binding a dimeric substrate, ClpAP may unfold and degrade one subunit while releasing the other, or both protomers of the substrate may be unfolded and degraded.
RepA is a native, dimeric substrate of ClpAP, and previous work in our laboratory demonstrated that one RepA dimer is bound by one ClpA 6 ClpP 14 complex (47). To determine whether one or both RepA protomers were degraded after binding ClpAP, GFP fusion proteins were utilized. RepA-GFP contains the full-length RepA protein with GFP fused to the C terminus, and it exists as a dimer in solution ( Fig. 1A and Ref. 33). RepA(1-70)-GFP contains the 70 N-terminal amino acids of RepA with GFP fused to the C-terminal amino acid, and it exists as a native monomer at the concentrations used here ( Fig. 1A and data not shown). Both fusion proteins are substrates for ClpAP (33,45). Complexes of substrate-ClpAP were formed in the presence of ATP␥S, a poorly hydrolyzed ATP analog, by incubating ClpAP with either [RepA-GFP] 2 or RepA(1-70)-GFP. The substrate-ClpAP complexes were then isolated by gel filtration. After the addition of a large excess of ␣-casein to compete with substrate released from ClpAP, ATP was added, and substrate degradation was measured by monitoring the decrease in GFP fluorescence. If both protomers of the dimer were degraded there would be close to a 100% de-crease in fluorescence, whereas if one protomer was degraded and one was released, there would be a 50% decrease in fluorescence.
The addition of ATP to ClpAP⅐[RepA-GFP] 2 complexes re-sulted in an 80% decrease in fluorescence (Fig. 1B). Separation of the reaction mixtures by SDS-PAGE followed by staining and densitometry confirmed that 80% of the RepA-GFP was degraded by ClpAP (data not shown). These data suggest that the binding of dimers to ClpAP results in the proteolysis of both protomers in the majority of complexes. By comparison the addition of ATP to ClpAP⅐RepA(1-70)-GFP complexes resulted in degradation of 94% of the substrate as determined by the decrease in fluorescence and by densitometry ( Fig. 1C and data not shown). This indicates that very few bound monomeric substrates escape proteolysis by ClpAP. The effectiveness of ␣-casein in competing with free RepA was determined by isolating ClpAP complexes in the absence of substrate, then adding [RepA-GFP] 2 , excess ␣-casein, and ATP together and measuring the loss of fluorescence. The presence of excess ␣-casein resulted in greater than 90% inhibition of RepA-GFP degradation (Fig. 1D), revealing that if RepA-GFP was released from ClpAP, it was unlikely to rebind and be degraded. Taken together, these results demonstrate that interactions between ClpAP and a homodimeric substrate most often lead to the proteolysis of both subunits within a single round of binding. The Presence of a Recognition Signal on One Protomer of a Heterodimer Is Sufficient to Target Both Protomers for Degradation by ClpAP-After demonstrating that ClpAP can unfold and degrade both subunits of a dimer in one round of binding, we then wanted to determine whether degradation of the dimer required the presence of a recognition signal on both protomers. RepA heterodimers were constructed in which only one subunit contained the ClpA recognition signal (Fig. 2A). Previously, we demonstrated that ClpA does not recognize a truncated version of RepA in which the 25 N-terminal residues are removed (RepA(⌬25)), although this mutant protein is a dimer and can be activated for DNA binding both by the DnaK chaperone system and by treatment with chemical denaturants (Ref. 48 and data not shown). Heterodimers containing one RepA(⌬25) protomer and one full-length, His-tagged RepA protomer were prepared by mixing RepA(⌬25) and His-tagged full-length RepA in a 5:1 ratio in denaturant and then remov- ing the denaturant. The renatured heterodimers were isolated by immobilized metal affinity chromatography (Fig. 2B, lane 3). To demonstrate that the heterodimers were stable, they were diluted 5-fold and mixed with a 5-fold molar excess of non-Histagged RepA and subjected to another metal affinity purification. SDS-PAGE showed that the RepA(⌬25) subunit coeluted with the His-tagged RepA protomer, indicating that the heterodimeric complex was stable in solution (Fig. 2B, lane 4).
We then measured degradation of the heterodimers by ClpAP by monitoring the disappearance of the substrate with time by SDS-PAGE followed by Western blot analysis and densitometry. We found that both protomers were degraded, although only one subunit contained the recognition signal (Fig. 3A). The protomer lacking the recognition signal was degraded significantly slower than the full-length RepA protomer (Fig. 3A); RepAhis and RepA(⌬25) were degraded 50% in 7.2 Ϯ 0.4 and 20.8 Ϯ 0.6 min, respectively (calculated from three separate experiments). A slower rate of degradation and a lag phase appear to contribute to the slower proteolysis of the untagged protomer. Both subunits were undetectable when the incubations with ClpAP were extended to 40 min (data not shown). With the same conditions homodimeric RepA(⌬25), which lacks the ClpA recognition signal, was not detectably degraded by ClpAP within 25 min (Fig. 3B). Under the same conditions, more than 80% of RepAhis was degraded by ClpAP (Fig. 3C).
Why the subunit lacking the recognition signal was consistently degraded more slowly than the tagged subunit is not known. We demonstrated previously that ClpAP degrades unfolded proteins lacking recognition signals (45). Therefore, one possible reason might be that the RepA(⌬25) subunit is released from ClpAP in an unfolded form and then rebound by another ClpAP molecule. This was tested for by denaturing RepA(⌬25) in guanidine HCl and then measuring degradation following the addition of the denatured substrate to reaction mixtures containing ClpAP (Fig. 4). There was no detectable degradation, implying that if the untagged subunit was released in an unfolded conformation, it was not likely to be rebound and degraded by ClpAP with these conditions. We also measured degradation of the RepAhis:RepA(⌬25) heterodimers after the isolation of ClpA-heterodimer complexes. First, ClpA and the heterodimers were incubated in the presence of ATP␥S to generate stable complexes, and then the complexes were isolated by size exclusion chromatography. ClpP was added, either with or without ATP, and degradation of the bound substrate was measured by SDS-PAGE. The results showed that both subunits of the heterodimer were degraded (Fig. 5A). When complexes of ClpA⅐[RepAhis] 2 were similarly isolated and degradation was measured, both subunits were degraded (Fig. 5B). Because RepA(⌬25) lacks a recognition signal, stable complexes between it and ClpA could not be isolated.
Taken together, these data demonstrate that the presence of a recognition signal on one protomer of a dimeric substrate is sufficient to target both protomers for degradation by ClpAP. They suggest that the untagged subunit remains associated with ClpAP until it is eventually degraded.
ClpXP Degrades Only the Protomer Containing the Recognition Signal When Interacting with Heterodimeric Substrates-In contrast to our observations for ClpAP, ClpXP was previously demonstrated to degrade only the protomer containing the recognition signal when interacting with a heterodimeric substrate (40). The ClpXP experiments utilized Arc repressor heterodimers in which only one protomer contained the SsrA recognition signal. To determine whether the contrasting observations with ClpAP reflect different mechanisms for ClpAP and ClpXP or are because of differences in the substrates utilized, we investigated the interactions between ClpXP and RepA heterodimers.
RepA is not a natural substrate for ClpXP, so an SsrA recognition signal was added to the C-terminal end of the RepA(⌬25) construct utilized in the ClpAP experiments (Fig.   6A). Without the SsrA recognition signal, ClpX was not able to activate RepA(⌬25) for DNA binding (Fig. 6B). However, ClpX was able to activate the RepA(⌬25)SsrA fusion protein for DNA binding, indicating that RepA(⌬25)SsrA exists as an inactive dimer in solution and is converted to an active monomer by the chaperone activity of ClpX (Fig. 6B). Activation of RepA(⌬25)SsrA by ClpX was similar to activation of RepAhis by ClpA (data not shown). Additionally, ClpA was able to activate RepA(⌬25)SsrA through recognition of the SsrA tag, although ClpA is unable to recognize RepA(⌬25) (data not shown). These data demonstrate that the presence of an SsrA recognition signal on RepA directs the fusion protein to ClpX To determine whether ClpXP acted similarly to ClpAP with respect to heterodimeric RepA substrates, RepAhis: RepA(⌬25)SsrA heterodimers were prepared and isolated as described above for RepAhis:RepA(⌬25). As with the ClpAP heterodimeric substrate, the RepAhis:RepA(⌬25)SsrA heterodimers were stable to subunit exchange in solution (data not shown). Incubation of ClpXP with RepA(⌬25)SsrA homodimers resulted in complete degradation of the substrate, demonstrating that SsrA-tagged RepA(⌬25) is proteolyzed efficiently by ClpXP (Fig. 7A). RepAhis homodimers, which lack the SsrA recognition signal, were not detectably degraded by ClpXP, showing that without the SsrA tag, RepA is not a substrate for ClpXP-dependent degradation (Fig. 7B). Importantly, when ClpXP was incubated with the heterodimeric substrate, only the SsrA-tagged subunit was degraded (Fig. 7C). This demonstrates that, unlike ClpAP, ClpXP requires both subunits of a dimeric substrate to contain the ClpX recognition signal for both protomers to be degraded. Taken with the ClpAP data from above, these results suggest that despite their extensive structural and functional homology, ClpAP and ClpXP process substrates in different ways.
ClpXP Releases the Nondegraded Subunit of a Heterodimeric Substrate as an Active Monomer-Unlike ClpAP, ClpXP did not degrade both subunits of a heterodimer. To determine whether the protomer that lacked the recognition signal was released as an active monomer or not, the DNA binding activity of the nondegraded subunit was measured. ClpX and ClpXP were incubated with RepAhis homodimers, RepA(⌬25)SsrA homodimers, and RepA:RepA(⌬25)SsrA heterodimers. A portion of each reaction was subjected to SDS-PAGE followed by Western blot analysis and densitometry. Another portion was used to measure DNA binding. RepAhis homodimers were neither activated by ClpX nor degraded by ClpXP because of the absence of the SsrA tag, as anticipated (Fig. 8A). The presence of the recognition signal on RepA(⌬25)SsrA homodimers resulted in activation by ClpX and proteolysis by ClpXP (Fig. 8B). In contrast, RepAhis:RepA(⌬25)SsrA heterodimers were activated for DNA binding by both ClpX and ClpXP. As expected, activation was reduced ϳ50% in the samples incubated with ClpXP compared with those incubated with ClpX because of degradation of the tagged protomer by ClpXP (Fig. 8C). Western blot analysis confirmed that ClpXP degraded the SsrAtagged subunit but not the subunit lacking SsrA (Fig. 8C). Together, these data show that when only one subunit of a dimeric substrate is degraded by ClpXP, the remaining protomer is released. DISCUSSION The data presented here shed light on the mechanism of substrate recognition by Clp ATPases and degradation by Clp proteases. The results demonstrate that ClpAP degrades both subunits of a RepA heterodimer when only one subunit contains the N-terminal recognition signal. One possible mechanism for the action of ClpAP on heterodimers is that the interaction between ClpA and the tagged subunit brings the untagged subunit into close proximity to ClpA, where it is bound through low affinity secondary recognition signals. This model is consistent with previous results from our laboratory demonstrating that although the first 15 amino acids of RepA are sufficient to target a RepA-GFP fusion protein for degradation by ClpAP, fusion proteins containing larger portions of RepA (namely the first 46 or 70 amino acids) are bound by ClpA with higher affinity (35). These observations suggest that in addition to the essential Clp recognition signal located near the N or C terminus of the substrate, the presence of secondary recognition signals in the polypeptide may increase substantially the efficiency of substrate recognition and degradation by Clp proteases.
The reason why the untagged subunit is degraded more slowly by ClpAP than the tagged subunit remains unexplained (Fig. 3A). It has been shown that protein unfolding is the rate-limiting step in degradation by Clp proteases (21). Therefore one possibility is that more cycles of ATP hydrolysis are required to unfold and translocate an untagged subunit compared with a tagged subunit, and thus more time is required. It is possible that the lag in degradation of the untagged subunit seen in Fig. 3A may be caused by preferential unfolding and translocation of the tagged subunit prior to the processing of the untagged subunit.
In contrast to ClpAP, ClpXP was only able to degrade the tagged subunit of a RepA heterodimer containing one tagged FIG. 8. The RepA subunit lacking the ClpX recognition signal is activated for DNA binding, whereas the RepA(⌬25)SsrA subunit is degraded by ClpXP. RepAhis homodimers (A), RepA(⌬25)SsrA homodimers (B), and RepAhis:RepA(⌬25)SsrA heterodimers (C) were incubated as indicated with ClpX, ClpP, and ATP as described under "Experimental Procedures." oriP1 DNA binding was measured at 0°C as described under "Experimental Procedures" by adding 10-l portions of the reactions to 10 l of buffer A containing 1 g of calf thymus DNA and 11 fmol of 3 H-labeled oriP1 plasmid DNA. Degradation was measured by trichloroacetic acid precipitating 90-l portions of the reaction mixtures and subjecting the samples to SDS-PAGE followed by Western blot analysis as described under "Experimental Procedures." Experiments were carried out three times, and the results shown for RepA activation are the averages of three experiments Ϯ S.E. For each substrate one representative Western blot is shown. and one untagged subunit. Similar results were obtained by Burton et al. (40) using another substrate, Arc repressor. In their experiments, one Arc subunit was SsrA-tagged and the other was not; ClpXP only degraded the SsrA-tagged subunit. Thus, with two different substrates, the untagged subunit was not degraded by ClpXP. Our results also show that the RepA subunit lacking the SsrA tag that was not degraded by ClpXP was activated for oriP1 DNA binding. One possible explanation is that the untagged subunit is simply not recognized or bound by ClpXP and is activated as a consequence of the unfolding and degradation of the SsrA-containing subunit. Another possible interpretation is that the subunit without the SsrA tag is unfolded by ClpX coincidentally with the unfolding of the tagged subunit, but instead of being translocated to ClpP, is released and refolds spontaneously. This scenario is consistent with previous work demonstrating that ClpXP, unlike ClpAP, fails to degrade an unfolded substrate lacking a recognition signal (20). These possibilities remain to be tested.
In summary, our results demonstrate that ClpAP and ClpXP interact differently with heterodimeric substrates and suggest that there may be a significant difference in the mechanisms of substrate recognition and binding between ClpAP and ClpXP.