Recognition, Targeting, and Hydrolysis of the λ O Replication Protein by the ClpP/ClpX Protease*

It has previously been established that sequences at the C termini of polypeptide substrates are critical for efficient hydrolysis by the ClpP/ClpX ATP-dependent protease. We report for the bacteriophage λ O replication protein, however, that N-terminal sequences play the most critical role in facilitating proteolysis by ClpP/ClpX. The N-terminal portion of λ O is degraded at a rate comparable with that of wild type O protein, whereas the C-terminal domain of O is hydrolyzed at least 10-fold more slowly. Consistent with these results, deletion of the first 18 amino acids of λ O blocks degradation of the N-terminal domain, whereas proteolysis of the O C-terminal domain is only slightly diminished as a result of deletion of the C-terminal 15 amino acids. We demonstrate that ClpX retains its capacity to bind to the N-terminal domain following removal of the first 18 amino acids of O. However, ClpX cannot efficiently promote the ATP-dependent binding of this truncated O polypeptide to ClpP, the catalytic subunit of the ClpP/ClpX protease. Based on our results with λ O protein, we suggest that two distinct structural elements may be required in substrate polypeptides to enable efficient hydrolysis by the ClpP/ClpX protease: (i) a ClpX-binding site, which may be located remotely from substrate termini, and (ii) a proper N- or C-terminal sequence, whose exposure on the substrate surface may be induced by the binding of ClpX.

The Hsp100 family of protein ATPases was originally identified in Escherichia coli through the roles of individual AT-Pases as regulatory components of certain ATP-dependent serine proteases, such as ClpP (for reviews see Refs. [1][2][3]. For example, the ClpA and ClpX family members were isolated as factors that enabled the ClpP catalytic component to hydrolyze certain polypeptide substrates in an ATP-dependent manner (4 -7). On its own, the ClpP protease degrades only very small peptides, and such hydrolysis does not depend on ATP (8,9). In the presence of ATP␥S, 1 a poorly hydrolyzable ATP analogue, ClpP/ClpA-mediated hydrolysis of peptides of moderate length is stimulated. This suggests that interactions between ClpA and ClpP, promoted by nucleotide binding to ClpA, bring about an activation of proteolysis by ClpP. Nevertheless, for both the ClpP/ClpA (9) and ClpP/ClpX (6) proteases, ATP hydrolysis per se is required to achieve processive degradation of larger polypeptides.
Hsp104, a Saccharomyces cerevisiae homologue of E. coli ClpB, has been shown to be essential for cellular thermotolerance (10). Interestingly, overexpression of the molecular chaperone Hsp70 largely reverses the temperature sensitivity exhibited by the hsp104 mutant (11). These and other findings support the idea that members of the Clp family of ATPases can perform certain chaperone functions in the absence of their partner protease subunit (12). This hypothesis has been confirmed by recent experimental findings, namely: (i) that Hsp104 participates in the disaggregation of partially damaged or aggregated protein structures (13,14) as well as modulates the conformational transition between normal and altered forms of prion-like factors (15), (ii) that bacterial ClpA protein substitutes in vitro for the DnaK/DnaJ/GrpE chaperone machine in activation of the phage P1 RepA replication protein for binding to its recognition sites in the P1 replication origin (16,17), and (iii) that ClpX protein enhances the binding of the O replication initiator to the replication origin (18) and promotes the binding of the RK2 plasmid-encoded TrfA protein to the oriRK2 sequence (19). Yet perhaps the most convincing evidence for the notion that Clp ATPases have intrinsic chaperone activity comes from investigations of the bacteriophage Mu DNA replication system. Initially it was discovered that deletion of the E. coli clpX gene (but not the clpP gene) resulted in a total inactivation of Mu DNA replication in vivo (20). Later, studies of the mechanism of Mu DNA replication in a reconstituted multiprotein system showed that ClpX acts at a step prior to initiation of DNA replication to promote the disassembly of an extremely stable MuA transposase nucleoprotein complex (21)(22)(23). It is suggested that ClpX-mediated release of MuA transposase from a transposition intermediate triggers initiation of Mu DNA replication by providing access of host replication proteins to Mu DNA ends (24).
Both the protease and chaperone activities of the Clp ATPases depend on ATP hydrolysis. In the presence of ATP, the ClpA or Hsp104 proteins individually oligomerize to give rise to a hexameric, ring-like structure (25)(26)(27). Electron microscopic studies have shown that ClpA or ClpX rings bind to the 7-fold symmetric ClpP component, which itself is arranged in a barrel-like double-ring structure. These interactions produce an overall structure that highly resembles the eukaryotic 26 S proteosome (28 -30).
It has been suggested that ClpA recognizes relatively un-folded proteins, whereas ClpX is apparently much more specific. For example, ClpX is known to recognize the O (6, 7), P1 Phd (31), Mu Repvir (32), MuA transposase (21), s (33), and UmuDЈ (34) proteins, each of which seemingly exists in a native, properly folded state. These findings support the hypothesis that the Clp ATPases are specificity factors that are capable of recruiting different protein substrates and, subsequently, transferring them to the ClpP subunit for proteolysis (6,7). It is still not certain which sequences or protein structures in substrate proteins are recognized by the Clp/Hsp100 family (2,35). For several protein substrates, e.g. MuA transposase (36), mutants of bacteriophage Mu repressor (32), or SsrA-tagged polypeptides (37), it has been shown that C-terminal sequences are primarily responsible for their targeting to the ClpP/ClpX protease. Recently, the presence of PDZ-like domains has been detected in ClpX and other Clp/Hsp100 family members. These structural motifs have been suggested to mediate the binding specificity of these proteins to different protein substrates (36). Furthermore, it was postulated that the initiating event during substrate selection by ClpX is the formation of a complex between ClpX and exposed C-terminal residues of a substrate protein (36). We decided to test the universality of this proposed mechanism using wild type and deletion mutant forms of the O replication initiator protein. Wild type O protein, a highly effective substrate of the ClpP/ClpX protease, is composed of Nand C-terminal structural domains, which are connected by a relatively unstructured linker region (38). Contrary to published results with MuA transposase, mutants of Mu repressor, and SsrA-tagged polypeptide substrates, which demonstrated the importance of C-terminal sequences for hydrolysis by the ClpP/ClpX protease, we provide evidence that it is the N-terminal portion of the O replication protein that is primarily required for its efficient hydrolysis by this ATP-dependent protease. In addition, we also demonstrate that stable binding of substrate polypeptides to ClpX is not sufficient to ensure proteolysis by ClpP/ClpX. We suggest that in addition to a ClpXbinding site, special N-terminal or C-terminal sequences must be present in the ClpP/ClpX substrate polypeptide.  (19 -110) from plasmid pRLM151) have been described (38). Plasmids that overexpress O(150 -299) (plasmid pRLM216) and O(19 -299) (plasmid pRLM312) will be described in detail elsewhere. In brief, plasmid pRLM77 contains a 1.5-kilobase pair AluI fragment harboring the O gene inserted into the multiple cloning site present on expression plasmid pRLM76 (39). Plasmid pRLM84 was constructed from plasmid pRLM77 by digesting the plasmid at the unique EcoRI site, located within the O gene, filling in the cohesive end with the Klenow fragment of DNA polymerase I, and religating. This creates a 4-base pair duplication that results in an in-frame chain termination codon. Plasmid pRLM186 was constructed by digesting plasmid pRLM77 with EcoRI, treating the linearized DNA with Bal31 nuclease, and religating. Plasmids that overproduce truncated O proteins were identified and sequenced so that the end points of each internal deletion in the O gene could be defined. A plasmid producing an O protein, deleted for amino acid residues 127-180, was named pRLM186. Plasmids pRLM146, pRLM147, pRLM150, pRLM151, pRLM216, and pRLM311 each contain a segment of the O gene generated by PCR-mediated amplification of DNA sequences carried on the dv-like plasmid pRLM4 (40). The primers used at one end for PCR amplification carried in order (5Ј to 3Ј) were a BamHI restriction site, a strong ribosome-binding site from gene 10 of phage T7, a properly spaced ATG initiation codon, followed by the desired O coding sequence. The primers used for synthesis of the complementary strand contained in order (5Ј to 3Ј) were XbaI, PstI, and SalI restriction sites, two tandem chain termination anti-codons, and a sequence complementary to the codons for the desired new C terminus. The PCR-amplified DNA preparations were digested with both BamHI and PstI or with both BamHI and SalI. These DNA fragments were subsequently inserted by ligation between the BamHI and PstI polylinker sites present on plasmid pRLM76 (pRLM146, pRLM147, pRLM150, and pRLM151) or between the BamHI and SalI polylinker sites present on plasmid pRLM156 (41) (pRLM216 and pRLM311). The expression vector for O (19 -299), pRLM312, was constructed by ligating the appropriate BamHI/SalI fragment of pRLM311 between the BamHI and SalI sites of pET21(ϩ) (Novagen). All plasmid sequences derived from PCR amplification were verified by DNA sequence analysis using the dideoxynucleoside triphosphate chain termination method.

Plasmids and Mutagenesis-Construction
For mutagenesis of a region encompassing the putative ClpX consensus recognition site in the C-terminal region of O, the 488 base pairs BamHI-PstI fragment of pRLM216 was cloned into BamHI-PstI pGEM3Zf(Ϫ) vector (Promega) to give plasmid pMZ2. This plasmid was subsequently transformed into the bacterial strain CJ236 dut Ϫ ung Ϫ , and single-stranded, uracil-containing DNA was produced by mobilization with helper phage VCSM13 ( (44), the supernatant (120 ml) was mixed with ammonium sulfate (0.3 g/ml supernatant) and stirred for 30 min at 4°C. Following centrifugation at 70,000 ϫ g for 30 min at 4°C, the pellet was resuspended and dialyzed against buffer A (50 mM Tris/HCl, pH 7.4, 0.1 M KCl, 0.1 mM EDTA, 10% (v/v) glycerol, 5 mM dithiothreitol, 0.02% Triton X-100) and applied to a Pharmacia Q-Sepharose (2.5 ϫ 14 cm) column that had been equilibrated with buffer A. The void volume fractions were pooled (60 ml) and applied to a Whatman P11 phosphocellulose column (1.5 ϫ 8 cm) that was equilibrated with buffer A. Bound proteins were eluted with a linear gradient (120 ml) of 0.1-0.7 M KCl in buffer A. Fractions containing significant quantities of O protein were pooled (35 ml) and applied to a hydroxylapatite column (1.5 ϫ 5 cm) that was equilibrated with buffer A. Bound protein was eluted with a linear gradient (120 ml) of 0.1-0.7 M KCl in buffer A. The fractions containing pure protein were pooled, frozen, and stored at Ϫ80°C prior to use.
The O(1-110) N-terminal fragment of O was purified from the membrane fraction of the crude lysate. After cell lysis (40 g of cells) with lysozyme and centrifugation of the lysate at 70,000 ϫ g for 60 min at 4°C, the membrane-containing pellet was resuspended in 50 ml of buffer B (50 mM Tris/HCl, pH 7.4, 0.2 M KCl, 1 mM EDTA, 5 mM dithiothreitol, 10% (v/v) glycerol, 0.04% Triton X-100, 4 M urea) and sonicated. Following gentle overnight shaking, the suspension was centrifuged at 70,000 ϫ g for 30 min at 4°C. The supernatant (70 ml) was dialyzed against 2 liters of buffer C (50 mM Tris, pH 7.4, 0.2 M KCl, 1 mM EDTA, 5 mM dithiothreitol, 10% (v/v) glycerol, 0.04% Triton X-100) and applied to a P11 phosphocellulose column (2.5 ϫ 14 cm) equilibrated with buffer C. The column was washed with several column volumes of buffer C, and the bound proteins were subsequently eluted with a linear gradient (400 ml) of KCl (0.2-0.8 M in buffer C). Fractions containing significant amounts of the N-terminal domain of O protein were pooled, dialyzed against buffer C, and applied to a 6-ml Resource MonoS column (Pharmacia) that was equilibrated with buffer C. The bound protein was eluted with a 50-ml KCl gradient (0.2-0.5 M in buffer C) and was stored at Ϫ80°C.
O(19 -110) protein was purified as described above for O(1-110), except that a hydroxylapatite column (1.2 ϫ 4.8 cm) was substituted for the Resource MonoS column during the final purification step. A linear gradient (50 ml) from 0.1 to 0.7 M KCl in buffer C was used to elute bound protein from the column. Fractions containing homogeneous O (19 -110) were frozen in liquid nitrogen and stored at Ϫ80°C. The N-terminal sequences of O and all purified O deletion mutant proteins were determined using an ABI 473A Applied Biosystems Microsequencer system.
Protease Assays-The standard protease assay (150 l) was carried out in 20 mM HEPES, pH 7.6, 10 mM MgAc 2 , 0.5% Brij, and 10 mM ATP. Each assay included 5 g of ClpP, 5 g of ClpX, and 5 g of the appropriate substrate polypeptide. The reaction mixture was incubated at 30°C. At desired times, 25-l portions of the reaction mixture were withdrawn and processed by 12.5% SDS-polyacrylamide gel electrophoresis.
Single-stranded DNA Replication Assay-single-stranded DNA replication assays were performed as described (45,46) with modifications as follows. M13mp9 viral DNA (165 ng) was used as the singlestranded DNA template, and reaction mixtures also contained 90 ng of E. coli GrpE protein. DNA synthesis was measured after a 20-min incubation at 30°C.

RESULTS
The N-terminal Portion of O Is Important for the ClpP/ ClpX-dependent Hydrolysis of O-It has previously been shown that the ClpX and ClpA ATPases recognize specific protein substrates, leading to selective ClpP-dependent proteolysis (6,18). Subsequent studies led to the hypothesis that sequences at the C termini of substrates of the ClpP/ClpX and ClpP/ClpA proteases are required for targeting the substrate polypeptide to the ClpX or ClpA ATPases (36,37). We decided to investigate whether this C-terminal substrate recognition mechanism applies uniformly to all physiological substrates of ClpX. We tested wild type O replication protein, which is known to be a particularly good substrate of the ClpP/ClpX protease both in vivo and in vitro, and various truncated derivatives of O (Fig. 1) for their sensitivity to hydrolysis by the ClpP/ClpX protease. We were surprised to find that the C-  (Table I) when the O(150 -299) protein replaces wild type O protein in the single-stranded replication reaction (45,46). The N-terminal origin recognition domain alone is incapable of supporting this single-stranded DNA replication reaction (Table I).
To exclude the possibility that the central portion of the O protein provides the structural motif that permits binding to ClpX, we tested a O mutant with a large central deletion, O(1-126/181-299). This O deletion mutant protein retains significant functional activity in an in vitro DNA replication system (38). O(1-126/181-299) is almost as sensitive to ClpP/ ClpX-mediated proteolysis as wild type O (Fig. 3). In addition, the deletion of the C-terminal 15 amino acids from the O(150 -299) polypeptide, resulting in an O(150 -284) truncated fragment, only slightly alters its sensitivity to proteolysis (Fig.  3). This result indicates that the O protein behaves differently than the MuA transposase or a related fusion protein, Arc-MuA, where deletion of the last 8 amino acids from their C termini almost completely stabilized MuA against ClpP/ClpX-

FIG. 2. Kinetics of ClpP/ClpX-mediated hydrolysis of wild type O and of O derivatives composed of either the N-or C-terminal domain.
Proteolysis was carried out as described under "Materials and Methods." dependent proteolysis (36). It was noted previously that the aspartic acid at position 287 of O is a part of the LDL/A consensus sequence present in all known ClpX protein substrates (2). To test this putative recognition sequence, we introduced a point mutation, D287A, in the O(150 -299) polypeptide. However, contrary to expectation, this sequence alteration did not substantially affect the rate of degradation of the O C-terminal domain (Fig. 3).
The smallest O protein fragment tested here that was efficiently hydrolyzed by the ClpP/ClpX protease is the O(1-110) N-terminal polypeptide (Fig. 3). We discovered, however, that further deletion of the first 18 amino acids (MTNTAKILNF-GRGNFAGQ) from this truncated substrate, resulting in O(19 -110), renders it almost completely resistant to protease action (Fig. 3). Yet O(19 -110) retains the capacity to form a homodimer that binds specifically to the O recognition sites in ori (38). We presume, therefore, that a significant proportion of O(1-110) is folded correctly. We performed the same experiments with O(1-139) and O(19 -139) and again found that deletion of the first 18 amino acids at the N terminus causes a severe inhibition of ClpP/ClpX-dependent proteolysis (Fig. 3). Interestingly, the O(19 -299) protein lacking the first 18 amino acids is hydrolyzed but at a much slower rate than the wild type O, with kinetics comparable with those for degradation of the C-terminal domain of O (i.e. O(150 -299); Fig. 3).
Even though our protein truncation experiments suggest that the very N-terminal portion of the O protein is largely responsible for targeting this phage replication initiator to the ClpP/ClpX protease, we observed that O fragments comprised solely of the C-terminal domain are moderately sensitive to this ATP-dependent protease (Fig. 3). This result indicates the presence of one or more "weaker" protease recognition sites within the C-terminal half of the O polypeptide.
Binding of O Deletion Mutant Proteins to the ClpX ATPase-We used a modified ELISA technique to determine the relative binding affinity of ClpX for the various O protein constructs used in this work (Fig. 4). For most substrates that are sensitive to the ClpP/ClpX protease, there is a direct correlation between the apparent affinity of the substrate for ClpX and the rate of substrate hydrolysis. The single exception, O(19 -299), has a slightly higher affinity for ClpX than O(150 -299) (Fig. 4) yet yields similar rates of degradation (Fig. 3). Nevertheless, it is also apparent from these results that ClpX is capable of binding with moderate affinity to polypeptides that are inert as substrates for ClpP/ClpX. For example, O (19 -139) and O(19 -110) each form stable complexes with ClpX yet are totally resistant to proteolysis. We conclude that stable binding of ClpX to a polypeptide is not sufficient to ensure hydrolysis of the polypeptide by ClpP/ClpX. O fragments comprising the N-terminal 110 amino acids showed the highest affinity for ClpX in the ELISA assay. Removal of the first 18 amino acids at the N terminus led to a significant drop in the affinities of several O derivatives (O (19 -110), O (19 -139), and O(19 -299)) for ClpX (Fig. 4). This result suggests the possibility that there may be two or more interaction sites for ClpX in the N-terminal region of O, one of which must be comprised in part by residues 1-18. We cannot, however, rule out an alternative possibility that the N-terminal region of O contains a single high affinity site for ClpX binding that is formed by amino acid residues on both sides of residue 19. Role of ClpX in the "Presentation" of O to ClpP-To investigate the role of ClpX in delivering the O substrate to the ClpP subunit of the protease, we made use of the nucleotide   (50 l, at a protein concentration of 10 g/ml in phosphate-buffered saline buffer) or bovine serum albumin was fixed onto ELISA plate wells as described previously (18) and incubated for 1 h at room temperature. Increasing amounts of ClpX in buffer B were then added. Following a 30-min incubation at room temperature, the ELISA plates were washed once with buffer B and three times with phosphate-buffered saline buffer supplemented with bovine serum albumin. Following these steps, 100 l of a 1:10,000 dilution of anti-ClpX serum was added, and the plates were incubated for an additional 2 h at room temperature. The amount of ClpX protein stably retained in each well was determined by use of a TMB peroxidase EIA substrate kit (Bio-Rad). Absorbance at 490 nm was measured using a microplate reader. Each indicated absorbance value represents the average value obtained in four independent experiments. analogue ATP␥S, which was previously shown to block the ClpP/ClpX-dependent hydrolysis of O (18). Using an ELISA assay, we determined the minimal requirements for ClpP to enter into a protein complex with O or O deletion mutant proteins (Fig. 5). Our results indicate that O does not enter into a complex with the ClpP protein unless both ClpX and ATP␥S are present. The fact that the association of O with ClpP depends on ClpX suggests that at least in the presence of ATP␥S, a O-ClpX-ClpP intermediate complex is formed. Interestingly the O(19 -139) truncated protein, which binds moderately well to ClpX (Fig. 4) but is not hydrolyzed by ClpP/ClpX protease (Fig. 3), forms only a relatively weak complex with ClpP, even when both ClpX and ATP␥S are present (Fig. 5). This result may indicate that the first 18 amino acids of O protein are required for efficient ClpX-dependent presentation of protein substrate to ClpP protease. DISCUSSION In a recent publication, Levchenko et al. (36) suggested that the ClpX protein recognizes specific sequences located at or near the C terminus of its substrate proteins. Their results indicated that such sequences may be displayed in a relatively disordered conformation. Experiments performed on MuA transposase, selected mutants of Mu repressor, and SsrAtagged protein have implicated C-terminal residues of the substrate, mainly nonpolar amino acids, as being absolutely essential for ClpX recognition. As pointed out by Levchenko et al. (36), a second group of ClpX substrates is composed of proteins that contain predominantly polar amino acids near the C terminus, with arginine at the terminal position. In this respect, the O replication protein, whose 10 C-terminal residues (residues 290 -299) are -NTDWIYGVDL, apparently does not fit neatly into either of the known categories of ClpX substrates.
In this report, we present evidence that certain of the predictions for the first class of ClpX substrates) made by This study also revealed that for most O deletion mutant proteins there is a close correlation between their relative affinity for ClpX and their rate of hydrolysis by the ClpP/ClpX protease. However, neither O(19 -139) nor O(19 -110) obeys this correlation. Both bind ClpX with moderate affinity, yet are nearly completely resistant to ClpP/ClpX-mediated proteolysis. We infer, therefore, that binding of ClpX to a potential substrate polypeptide is not sufficient to ensure hydrolysis of the polypeptide by the ATP-dependent ClpP/ClpX protease.
Although  (19 -139). Wild type or mutant O protein (50 l, at a protein concentration of 10 g/ml in phosphate-buffered saline buffer) was fixed onto the wells of an ELISA plate as described previously (18). Subsequently, increasing amounts of ClpP (in the presence of 100 ng of ClpX in buffer B) were added in the presence (open squares) or absence (closed squares) of 1 mM ATP␥S. In a control experiment, increasing amounts of ClpP in the presence of ATP␥S (but in the absence of ClpX) were added (open circles, A). After a 30-min incubation at room temperature, the ELISA plates were washed once with buffer B and three times with phosphatebuffered saline supplemented with bovine serum albumin. Next, anti-ClpP serum (100 l of a 1:8,000 dilution) was added, and the plates were incubated for an additional 2 h at room temperature. The amount of ClpP protein retained in individual wells was determined using a TMB peroxidase EIA substrate kit (Bio-Rad). The absorbance at 490 nm was measured using a microplate reader. In a control experiment, when bovine serum albumin instead of O was used, complexes containing ClpP were not detected (data not shown).
O determines the rate of ClpP/ClpX-dependent proteolysis.
It is likely that relatively weak ClpX-binding sites are exposed on the surface of certain protein substrates of the ClpP/ ClpX protease. In this regard, it is important to stress that such ClpX substrates as MuA transposase and Phd are hydrolyzed by the ClpP/ClpX protease much less efficiently than O. 2,3 These ClpX substrates are hydrolyzed at rates roughly comparable with or even lower than that of the O(150 -299). Thus, it is certainly possible that some proteins recognized by ClpX may possess multiple sequences or structural elements with varying degrees of affinity for ClpX.
Our results show that it is possible to identify proteins that bind to ClpX but are not hydrolyzed by the ClpP/ClpX protease. This implies that certain protein substrates that are hydrolyzed by the ClpP/ClpX protease, such as O, possess at least two structural elements that define its vulnerability to the protease. One element is presumably a ClpX recognition motif, whereas the other element apparently is composed, at least in part, of a specific sequence at the N terminus or C terminus of the polypeptide. The latter sequence may play a role in efficient delivery of a ClpX-bound polypeptide to the active site of the ClpP protease, perhaps by enabling efficient unfolding of the substrate. As shown in this paper, one of the ClpX recognition sites in O is located internally, somewhere within the 19 -110-amino acid fragment of O. The demonstration that ClpX recognition sites are not necessarily located at polypeptide termini agrees with a previous genetic analysis of the stationary phase sigma factor ( s ), a study concluding that ClpX recognizes an internal amino acid sequence in this substrate protein (33). Regardless of which portion of a protein is primarily recognized by ClpX, it is likely that the final outcome, in term of proteolysis, will depend not only on the presence of an intrinsic high affinity ClpX recognition sequence but also on the availability of a special C-or N-terminal sequence to make contact with ClpX and/or ClpP during delivery of the substrate polypeptide to the ClpP catalytic subunit. This precise situation was encountered previously by Laachouch et al. (32). They fused the C-terminal 7 amino acids of the ClpX-sensitive Mu repressor vir3061 protein to the C termini of CcdA and CcdB proteins encoded by plasmid F. Fusion of this sequence to CcdB did not endow a sensitivity to ClpP/ClpX, whereas fusion of the same sequence to CcdA did engender a partial sensitivity to this ATP-dependent protease.
It was proposed previously that ClpX could participate in regulatory mechanisms that "decide" whether a particular protein substrate is either repaired or destroyed (1,2,12). This decision could be based on the affinity of ClpX for the various sites present on a given protein substrate as well as on the accessibility of appropriate C-or N-terminal amino acid sequences. The results presented here suggest at least three possible outcomes for proteins that have ClpX recognition sites. First, it is possible to find polypeptides (such as O(19 -139)) that bind to ClpX but are resistant to hydrolysis by the ClpP/ ClpX protease. In such cases, ClpX is restricted to act only as a molecular chaperone. Second, if the protein instead also contains an exposed C-or N-terminal sequence that aids ClpX binding and/or function (such as O and O (1-139)), then such polypeptides may be immediately targeted to ClpP and hydrolyzed. Third, a native protein could contain both a ClpX-binding site(s), located distantly from polypeptide termini, as well as a "proper" C-or N-terminal sequence for ClpX that is not exposed on the surface of the protein. In this situation, ClpX action would be restricted to binding, but such binding events may in turn result in partial unfolding of the substrate protein.
In the event that the special C-or N-terminal sequence of the substrate polypeptide is still not accessible once all ClpX-induced structural rearrangements have been completed, the substrate could dissociate from ClpX and have an opportunity to refold. Repeated cycles of ClpX binding, followed by substrate unfolding, dissociation, and refolding could result in the repair of a partially denatured protein substrate. Thus, ClpX would act as a molecular chaperone in this instance. On the other hand, if the initial binding of ClpX and unfolding of the substrate results in the formation of a stable complex between ClpX and the special sequence at the N or C terminus, then the refolding reaction may be largely precluded. In this case, ClpX would rapidly act to "deliver" the partially unfolded protein substrate to the ClpP protease for degradation.
The extreme lability of the O protein in vivo was first reported 25 years ago (47). Later reports demonstrated that the O protein has a chemical half-life of just 1.5 min in -infected cells (48 -50) and that an ATP-dependent protease mediated this rapid turnover (50,51). In light of the identification of the responsible protease as the ClpP/ClpX protease (6, 7) and of recent findings that O protein bound to the replication origin is, both in vivo (52) and in vitro (53), stabilized against proteolytic attack, it is likely that the studies reported here will provide useful insights into the O turnover process in vivo. For example, the rapid degradation of free intracellular O protein suggests that the various ClpX recognition motifs in O identified in the present study are freely accessible to ClpX in vivo. In contrast, the resistance of O protein assembled at ori (the O-some) to ClpX/ClpP attack makes it likely that one or more of the ClpX recognition motifs in the O-some nucleoprotein structure are inaccessible to ClpX. It is also conceivable that proteinprotein and protein-DNA interactions within the O-some stabilize O protein against a critical ClpX-mediated unfolding step that is required for initiation of proteolysis.