Two peptide sequences can function cooperatively to facilitate binding and unfolding by ClpA and degradation by ClpAP

Abstract

Clp/Hsp100 proteins comprise a large family of AAA⁺ ATPases. Some Clp proteins function alone as molecular chaperones, whereas others act in conjunction with peptidases, forming ATP-dependent proteasome-like compartmentalized proteases. Protein degradation by Clp proteases is regulated primarily by substrate recognition by the Clp ATPase component. The ClpA and ClpX ATPases of Escherichia coli generally recognize short amino acid sequences that are located near the N or C terminus of a substrate. However, both ClpAP and ClpXP are able to degrade proteins in which the end containing the recognition signal is fused to GFP such that the signal is in the interior of the primary sequence of the substrate. Here, we tested whether the internal ClpA recognition signal was the sole element required for targeting the substrate to ClpA. The results show that, in the absence of a high-affinity peptide recognition signal at the terminus, two elements are important for recognition of GFP-RepA fusion proteins by ClpA. One element is the natural ClpA recognition signal located at the junction of GFP and RepA in the fusion protein. The second element is the C-terminal peptide of the fusion protein. Together, these two elements facilitate binding and unfolding by ClpA and degradation by ClpAP. The internal site appears to function similarly to Clp adaptor proteins but, in this case, is covalently attached to the polypeptide containing the terminal tag and both the “adaptor” and “substrate” modules are degraded.

Clp/Hsp100 molecular chaperones are members of the AAA⁺ family of ATPases that mediate ATP-dependent protein unfolding reactions. Some Clp proteins function as regulatory components of two-component ATP-dependent proteases, controlling substrate recognition (1–6). In prokaryotes, peptide sequences in proteins target the proteins to proteases, whereas in eukaryotes, protein ubiquitination targets proteins to the proteasome (3, 7).

Proteolysis by Clp proteases requires substrates first to be bound by the Clp ATPase component. Bound substrates are unfolded by the ATPase and then translocated through a narrow pore into the proteolytic chamber of the peptidase component in ATP-dependent reactions (3, 5). Many aspects of this model have been demonstrated. Both Escherichia coli ClpA and ClpX bind specific substrates and catalyze ATP-dependent protein unfolding (8, 9). Biochemical and electron microscopic data show that unfolded substrates are translocated processively starting from the recognition signal into the cavity of ClpP from ClpA and ClpX in ATP-dependent reactions (10–14). For some substrates, adaptor proteins modulate degradation, either enhancing or inhibiting specific recognition (5). For example, the RssB and SspB adaptor proteins deliver specific substrates, which are poorly recognized by ClpX, to ClpXP for degradation (15–19).

Protein substrates of ClpAP and ClpXP generally contain recognition signals of ≈10 aa, which are located either at the N or C terminus. One well studied signal is SsrA, an 11-aa peptide encoded by ssrA tmRNA. The SsrA peptide is added cotranslationally to the C terminus of polypeptides that become stalled on ribosomes (20). It interacts with ClpA and ClpX and targets the polypeptides for degradation by ClpAP and ClpXP (21). Although both ClpA and ClpX recognize SsrA-tagged proteins, other substrates are preferentially recognized by either ClpA or ClpX. Hundreds of ClpXP substrates have been identified; in contrast, there are relatively few known substrates for ClpAP (1, 5, 22).

One well characterized ClpAP substrate is P1 RepA. Deletion analysis revealed that the ClpA recognition signal is in the first 15 aa of RepA (23). Moreover, the N-terminal 15-aa peptide of RepA, when fused to the N or C terminus of GFP, is both necessary and sufficient to target the fusion protein for recognition and degradation by ClpAP (23, 24). In additional studies, it was found that when RepA was fused to the C terminus of GFP (thus moving the recognition sequence to the middle of the fusion protein), ClpA was able to unfold and translocate the fusion protein to ClpP for degradation (24). This observation and similar results with ClpXP revealed that recognition signals are functional when located in the interior of proteins (24).

In this study, we addressed the question of whether binding, unfolding, and degradation of the GFP-RepA fusion protein by ClpAP requires a second element in addition to the internal recognition signal. Our results show that both the internal recognition signal and a C-terminal signal are required for GFP-RepA to be efficiently recognized by ClpA. Together, these two elements facilitate binding by ClpA and degradation by ClpAP.

Results

Two Sequence Elements Are Required for Recognition of GFP-RepA by ClpA. Previously, we constructed fusion proteins of RepA and GFP to determine whether the RepA recognition signal could target a fusion protein for degradation by ClpAP when the signal was located in the interior of the primary sequence of the protein. We showed that a GFP-RepA-X₇-H₆ fusion protein, in which RepA was fused to the C terminus of GFP, was degraded by ClpAP at a rate only slightly slower than RepA-GFP, as measured by the disappearance of fluorescence of the fusion protein (Fig. 1) (24). In contrast, a GFP-RepA(Δ50)-X₇-H₆ fusion protein (which lacked the natural ClpA recognition signal because of a deletion of the first 50 aa of RepA) was degraded at a significantly slower rate (Fig. 1) (24). These experiments showed that an internally located recognition signal facilitates degradation by ClpAP.

Fig. 1.

ClpAP degradation of GFP-RepA fusion proteins. (A) RepA-GFP and GFP-RepA fusion proteins. (B) We incubated 0.24 μM ClpA and 0.34 μM ClpP with 0.2 μM RepA-GFP (black circles), GFP-RepA-X₇-H₆ (red triangles), GFP-(Δ50) RepA-X₇-H₆ (green diamonds), or GFP-RepA (blue squares), and GFP fluorescence was monitored over time as a measure of protein degradation, as described in Materials and Methods.

Whereas our previous work demonstrated the importance of the internal recognition signal, we were interested in determining how these substrates were bound and processed by ClpA. To focus on the possible importance of a free end for degradation, we constructed several more fusion proteins. In the set of experiments described above, both GFP-RepA and GFP-RepA(Δ50) contained H₆ plus a 7-aa linker at their C termini to allow for purification by metal-affinity chromatography. Because the C-terminal X₇-H₆ peptide was surface-exposed, it could be involved in substrate binding. To investigate this possibility, a GFP-RepA fusion protein that lacked the 13-aa C-terminal extension containing H₆ and the linker (referred to as GFP-RepA) was constructed and tested as a substrate for ClpAP. Interestingly, it was not detectably degraded (Fig. 1). SDS/PAGE confirmed that the full-length fusion protein remained intact (data not shown). Thus, when the high-affinity ClpA recognition signal of RepA is not located at the end of the substrate, an additional element (at the end) is required for degradation by ClpAP. These results indicate that the internal signal and terminus function together to facilitate degradation by ClpAP.

An Internal Recognition Signal Is Insufficient to Target a Protein for Remodeling by ClpA. In the absence of ClpP, the chaperone action of ClpA remodels inactive RepA dimers into active monomers that bind oriP1 DNA with high affinity (25). Previous results showed that the GFP-RepA fusion protein with the C-terminal X₇-H₆ peptide could be activated for P1 DNA binding by ClpA to the same extent as RepA-GFP (Fig. 2A) (24). However, when GFP-RepA that lacked the C-terminal peptide was incubated with ClpA and ATP, there was no detectable activation of DNA binding (Fig. 2A). To eliminate the possibility that GFP-RepA was incapable of being activated, we measured activation by the DnaK chaperone system, which is also known to activate RepA (26). We found that DnaK and DnaJ activated GFP-RepA similarly to Rep-GFP and GFP-RepA-X₇-H₆ (Fig. 2B). Thus, consistent with the degradation results, both the natural ClpA recognition signal and the C-terminal X₇-H₆ peptide are required for the GFP-RepA fusion protein to be remodeled by ClpA.

Fig. 2.

Activation of DNA binding activity of GFP-RepA fusion proteins. RepA-GFP (black circles), GFP-RepA-X₇-H₆ (red triangles), and GFP-RepA (blue squares) were incubated in the presence (filled symbols) or absence (open symbols) of ClpA (A) or DnaJ/DnaK (B). DNA binding activity was measured by using a nitrocellulose-filter binding assay, as described in Materials and Methods. Results are the means (±SEM) of three independent experiments. Open symbols overlap the x axis.

An Internal Recognition Signal Is Not Sufficient for Stable Substrate Binding by ClpA. Because GFP-RepA was neither degraded by ClpAP nor remodeled by ClpA, we tested whether or not GFP-RepA could form a stable complex with ClpA. Earlier experiments demonstrated that ClpA forms complexes with RepA but not GFP (8, 25, 27). Despite the inability of GFP-RepA to be remodeled by ClpA or degraded by ClpAP, it could possibly interact with ClpA by means of the internal recognition signal in RepA. We performed gel-filtration experiments to determine whether GFP-RepA formed a stable complex with ClpA. GFP-RepA was incubated with ClpA in the presence of ATP[γ-S], and the reaction mixture was then analyzed by Sephacryl S300 gel filtration. GFP-RepA eluted at the position expected for a protein of ≈120 kDa and at the same position as GFP-RepA in the absence of ClpA, showing that no detectable GFP-RepA was associated with ClpA (Fig. 3A). In contrast, when GFP-RepA-X₇-H₆ was incubated with ClpA and ATP[γ-S] and then subjected to gel filtration, ≈80% of the substrate eluted in the excluded volume of the column at the position where ClpA and ClpA–substrate complexes elute, demonstrating that GFP-RepA-X₇-H₆ interacts with ClpA (Fig. 3B). These results suggest that in the absence of the C-terminal X₇-H₆ peptide, the internal RepA recognition signal is not sufficient to allow stable interaction with ClpA.

Fig. 3.

Complex formation between ClpA and GFP-RepA-X₇-H₆. GFP-RepA (A) and GFP-RepA-X₇-H₆ (B) were incubated with ClpA in the presence of ATP[γ-S], and complexes were analyzed by gel filtration on a Sephacryl S-300 column, as described in Materials and Methods. Arrows indicate the elution position for ClpA-bound substrates and unbound substrates. GFP-RepA-X₇-H₆ and GFP-RepA when run alone eluted at the position indicated for unbound substrate (data not shown). Fluorescence is shown as arbitrary units (A.U.).

Addition of a His Tag Is Sufficient to Target Some Proteins for Degradation by ClpAP. To determine the essential features of the weakly recognized C-terminal His tag of the GFP-RepA(Δ50)-X₇-H₆ fusion protein, we constructed GFP fusion proteins with the following five different C-terminal extensions. (i) GFP fused to the same C-terminal peptide used above but lacking RepA (GFP-X₇-H₆); (ii) GFP with H₆ (GFP-H₆); (iii) GFP with the X₇ linker (GFP-X₇); (iv) GFP with the three C-terminal residues of X₇, followed by H₆ (GFP-X₃-H₆); and (v) GFP with a 30-aa linker ending in X₇ and followed by H₆ (GFP-X₃₀-H₆) (Fig. 4A). We then measured ClpAP degradation of the GFP derivatives by monitoring both the decrease in GFP fluorescence and the disappearance of protein by SDS/PAGE. GFP-X₇-H₆ was degraded at approximately the same rate as GFP-RepA(Δ50)-X₇-H₆, showing that the small amount of degradation detected with GFP-RepA(Δ50)-X₇-H₆ was independent of amino acid sequences in RepA(Δ50) (Fig. 4 B and C, compared with Fig. 1B). GFP-H₆ was not detectably degraded in either assay (Fig. 4 B and C). In contrast, the rate of GFP-X₇ degradation was ≈5-fold higher than that of GFP-X₇-H₆ (Fig. 4 B and C). When the linker was reduced from 7 to 3 aa, GFP-X₃-H₆, the substrate was poorly degraded, suggesting that some of the essential residues had been deleted (Fig. 4 B and C). GFP-X₃₀-H₆ was degraded at approximately half the rate of GFP-X₇ and at a slightly higher rate than GFP-X₇-H₆ (Fig. 4 B and C). Thus, the 7-aa linker peptide is recognized by ClpA, and the His tag is somewhat inhibitory. The GFP derivatives were also tested as substrates for ClpXP. By using conditions in which GFP-SsrA was degraded, GFP-X₇-H₆, GFP-H₆, GFP-X₇, GFP-X₃-H₆, and GFP-X₃₀-H₆ were not detectably degraded (Fig. 4D). Thus, as determined previously, ClpA and ClpX recognize different sequences.

Fig. 4.

Degradation of GFP fusion proteins in vitro and in vivo. (A) C-terminal peptide sequences of constructed GFP fusion proteins. (B) Protein degradation was measured by monitoring GFP fluorescence in reaction mixtures containing 0.24 μM ClpA, 0.34 μM ClpP, and 0.2 μM GFP-X₇-H₆ (pink squares), GFP-X₃-H₆ (green triangles), GFP-H₆ (blue diamonds), GFP-X₇ (light blue circles), or GFP-X₃₀-H₆ (red diamonds), as described in Materials and Methods.(C) Degradation of the above substrates by using the same reaction conditions was measured by densitometric analysis of substrate proteins after SDS/PAGE. (D) Degradation of GFP fusion proteins by ClpXP was determined by monitoring GFP fluorescence in reaction mixtures containing 0.16 μM ClpX, 0.34 μM ClpP, and 0.2 μM GFP fusion proteins, as described in B.(E) E. coli wt and ΔclpP strains harboring plasmids expressing GFP-X₇, GFP-X₃₀-H₆, GFP-X₇-H₆, GFP-H₆, and GFP-X₃-H₆ were spotted on LB agar plates, and, after 96 h of incubation, the plates were photographed by using visible light or UV transillumination, as described in Materials and Methods.

We also tested whether the GFP derivatives described above were degraded in vivo. Plasmids carrying the GFP derivatives were transformed into WT and ΔclpP cells. The cells were grown in liquid media and then spotted on LB plates without inducer. These growth conditions were chosen specifically to allow a very low level of expression of the substrate. None of the GFP fusion proteins were detrimental to cell growth, as evidenced by the similarity in appearance of the WT and ΔclpP strains carrying the various GFP-expressing plasmids after 96 h of growth (Fig. 4E). Visualization of the spots at that time by UV transillumination showed that ΔclpP cells expressing GFP-X₇, GFP-X₃₀-H₆, and GFP-X₇-H₆ appeared significantly more fluorescent than WT cells expressing the same fusion proteins, suggesting that the GFP derivatives were degraded in a ClpP-dependent reaction (Fig. 4E). The difference in fluorescence became more pronounced with increasing time of incubation between 24 and 96 h (data not shown). By comparison, little difference in fluorescence was observed between WT and ΔclpP cells expressing GFP-H₆ or GFP-X₃-H₆ after 96 h of incubation, suggesting that the substrates were not degraded (Fig. 4E). These results demonstrate that the GFP derivatives that are degraded in vitro are degraded in vivo.

ClpAP was able to degrade GFP with a linker and His tag to some extent without the addition of any further recognition signals. It is possible that residues in GFP, although not sufficient alone to target GFP for degradation by ClpAP, are used in combination with the C-terminal residues of GFP. This possibility is supported by the observation that GFP can interact with the N-terminal domain of ClpA (28). To determine whether His-tagged proteins in general are substrates for degradation by ClpAP, we tested two other His-tagged proteins. CbpA, an E. coli DnaJ homolog and DNA binding protein, was not detectably degraded by ClpAP, as measured by the disappearance of protein with time by SDS/PAGE (Fig. 5 A and B). However, a CbpA derivative containing X₇-H₆ fused to its C terminus (CbpA-X₇-H₆) was degraded by ClpAP (Fig. 5 A and B). Similarly, CbpM, a modulator of CbpA, was slowly degraded by ClpAP but CbpM-X₇-H₆ was rapidly degraded (Fig. 5 C and D). Another His-tagged protein, CspE, containing a different linker and an N-terminal His tag, was not detectably degraded (data not shown). Thus, a His tag and linker are sufficient to target some proteins for degradation. Moreover, the rates of degradation of His-tagged CbpA and CbpM were significantly higher than those of His-tagged GFP and GFP-RepA.

Fig. 5.

ClpAP degradation of other His-tagged proteins. (A) We incubated 0.24 μM ClpA and 0.34 μM ClpP with 0.8 μM CbpA or CbpA-X₇-H₆, and after incubation at 23°C for the indicated times, trichloroacetic acid (TCA) was added to 15% (wt/vol). The TCA pellets were analyzed by SDS/PAGE. (B) The CbpA (open squares) and CbpA-X₇-H₆ (filled squares) protein bands in A were quantified by densitometry. (C) We subjected 7.0 μM CbpM and CbpM-X₇-H₆ to ClpAP degradation, as described above for CbpA and the TCA pellets that were analyzed by SDS/PAGE. (D) CbpM (open circles) and CbpM-X₇-H₆ (filled circles) bands were quantified by densitometry.

Because His-tagged proteins were degraded in vitro and in vivo by ClpAP, we next sought to determine whether His-tagged proteins were degraded in vivo under conditions that are generally used for the overexpression of proteins. WT and ΔclpP cells carrying the plasmids coding for either GFP-X₇-H₆ or GFP-H₆ were grown in LB to an A₆₀₀ of 0.5 and induced by the addition of IPTG. Cells were collected after 2.5 h of induction. Analysis by SDS/PAGE of whole cell lysates revealed that both GFP-X₇-H₆ and GFP-H₆ accumulated to approximately equal levels in the ΔclpP and WT strains (Fig. 6). Thus, when high-level expression is obtained by growing cells in the presence of inducer, there is not appreciable degradation of the His-tagged proteins. These results suggest that the ClpP protease may be saturated and that degradation is not a serious concern when overexpressing His-tagged proteins.

Fig. 6.

Overexpressed His-tagged proteins are not detectably degraded in a ClpP-dependent reaction in vivo. WT (BL21 and DE3) and ΔclpP (SG1146a) strains expressing GFP-X₇-H₆ or GFP-H₆ were grown to midlogarithmic phase in LB broth at 37°C followed by an additional 2.5 h of growth in the absence or presence of 1 mM IPTG. Whole-cell lysates were analyzed by SDS/PAGE. An equivalent number of cells were loaded in each lane.

Discussion

The results presented here demonstrate that if the natural ClpA recognition signal of RepA is moved from an end to the interior of a fusion protein, then an additional amino acid sequence at the terminus is required for recognition by ClpA. Combined, these two signals target the substrate for unfolding by ClpA and degradation by ClpAP at rates similar to those of RepA with a naturally occurring N-terminal recognition signal. The two elements appear to be cooperative in nature, rather than additive.

Our working model (Fig. 7) for the dual recognition signal requirement is that ClpA binds with low affinity to the C-terminal His-tagged end of RepA, very likely in the vicinity of the ClpA channel. This interaction brings the internally located ClpA recognition signal of RepA close to its docking site on ClpA. Alone, the affinity of the internal recognition signal for ClpA is too low to allow the fusion protein to be stably bound, but interaction of the internal recognition signal with ClpA stabilizes the interaction of ClpA with the C terminus. In this model, the two elements function together to facilitate stable substrate binding. The two cooperating elements may also have a role in positioning the substrate for unfolding. By anchoring the substrate in two positions, the forces associated with ATP-dependent ClpA conformational changes could increase the distance between the two binding sites, forcing the substrate to unfold. This model is consistent with the recent results of Horwich and colleagues showing that the natural ClpA recognition signal in RepA interacts with the N-terminal domain of ClpA, which is distinct from the domain 2 region of ClpA where the C-terminal SsrA tag interacts with residues in a channel-facing loop (28). Our model suggests that the C-terminal peptide of GFP-RepA-X₇-H₆ binds in the channel, but this implication remains to be tested.

Fig. 7.

Model for the mechanism of GFP-RepA-X₇-H₆ recognition by ClpA. Two peptide sequences function cooperatively to target the substrate to ClpAP. See Discussion for details.

The mechanism of substrate presentation suggested here resembles that of substrates that require adaptor proteins for degradation by Clp ATPases. Adaptors are proteins that interact with a specific protein substrate and a Clp ATPase, thus facilitating the degradation of substrates that have too low affinity to be recognized as a substrate. Adaptors are not degraded during the reaction. One example of a ClpXP adaptor is RssB, which is a protein that is required for σ^S degradation (15, 16). Although neither RssB nor σ^S binds ClpX well alone, RssB interacts with σ^S and the RssB–σ^S complex binds to ClpX (17). Another example of an adaptor is SspB, which is a protein that increases the affinity of SsrA-tagged proteins and RseA for ClpX by interacting with both substrate and ClpX (18, 19). Both adaptor-mediated substrate targeting and targeting of GFP-RepA-X₇-H₆ to ClpA require two Clp recognition signals. However, in adaptor-mediated degradation one signal is on the adaptor and one is on the substrate, whereas with GFP-RepA-X₇-H₆, both signals are covalently linked in the same molecule. Thus, the RepA “adaptor domain” is degraded with the rest of the substrate. For the SspB adaptor protein, it has been shown that both the substrate and SspB are degraded by ClpXP when SspB is covalently attached to an SsrA-tagged protein through a disulfide bond (29).

Although this system of dual recognition signals has been constructed in vitro, there are several reports suggesting that a similar mechanism of recognition occurs among natural substrates as well. For example, deletion analysis of λO showed that sequences located near both the N and C termini contribute to the efficiency of degradation by ClpXP (30). Importantly, Mettert and Kiley (31) have recently demonstrated that FNR degradation by ClpXP requires two ClpX recognition motifs, one at the N terminus and the other at the C terminus. Deletion of either signal results in stabilization of FNR. Dual recognition signals may be a more general mechanism for protein recognition than previously thought, because 15 of 50 identified substrates of ClpXP contain potential recognition motifs at both their N and C termini (22). For two of these substrates, when the C terminus was fused to a stable protein, ClpXP degraded the fusion protein; for the same two substrates, ClpX bound to a peptide corresponding to the N terminus (22). It is likely that natural substrates of ClpAP exist that will similarly be found to require two recognition signals to target the substrate for degradation as a mechanism to increase affinity and/or specificity.

Materials and Methods

Plasmids and Strains. pETGFP-RepA was constructed by creating a stop codon after codon 286 of repA in pETGFP-RepA-X₇-H₆ by QuikChange (Stratagene) mutagenesis, according to the manufacturer's recommendations (24). pETGFP-X₃₀-H₆ was generated by PCR amplification of gfpuv(NdeIΔ) using NdeI incorporating 5′ and 3′ oligonucleotides (24). The resulting product, which does not contain the natural gfp stop codon, was digested with NdeI and ligated into NdeI digested pET24b (Novagen). The resulting plasmid expresses GFPuv with a 30-aa C-terminal linker, generated by the pET24b multicloning site, followed by a H₆ tag. pETGFP-X₇-H₆, pETGFP-X₃-H₆, pETGFP-H₆, and pETGFP-X₇ were each constructed by ligating annealed oligonucleotides into SacI- and HindIII-digested pETGFP-X₃₀-H₆ (gfpuv contains a 3′-terminal SacI site). These plasmids express GFPuv with differing C-terminal extensions, as shown in Fig. 4A.

pETCbpA-H₆ was constructed by generating a cbpA PCR fragment from E. coli DH5α, containing 5′ and 3′ NdeI and HindIII sites, respectively, and lacking a terminal stop codon. The PCR fragment was digested with NdeI and HindIII and ligated into similarly digested pET24b. pETCbpM-H₆ was constructed as described for pETCbpA-H₆ but by using a cbpM PCR fragment. All constructs were verified by DNA sequencing. SG1146a (BL21, λDE3, and ΔclpP) was a gift from Susan Gottesman (National Cancer Institute).

Proteins and DNA. P1 RepA (32), DnaJ (33), DnaK (34), ClpA (35), ClpX (36), ClpP (35), CbpA (37), CbpM (37), RepA-GFP (27), GFP-RepA-X₇-H₆ (24), and GFP-RepA(Δ50)-X₇-H₆ (24) were purified as described previously.

GFP-RepA, GFP-X₃₀-H₆, GFP-X₇-H₆, GFP-X₃-H₆, GFP-H₆, and GFP-X₇ were isolated from BL21(DE3) cells harboring pET-GFP-RepA, pETGFP-X₃₀-H₆, pETGFP-X₇-H₆, pETGFP-X₃-H₆, pETGFP-H₆, and pETGFP-X₇, respectively. The cells were grown at 25°C to an A₆₀₀ of 0.4 and induced overnight with 0.1 mM isopropyl-β-d-thiogalactopyranoside at 25°C. For GFP-RepA, all subsequent purification steps were similar to those for P1 RepA. GFP-X₇ was purified by organic extraction as described in ref. 27. GFP-X₃₀-H₆, GFP-X₇-H₆, GFP-X₃-H₆, and GFP-H₆ were purified by chromatography with Talon resin according to the manufacturer's instructions.

CbpA-X₇-H₆ and CbpM-X₇-H₆ were isolated from BL21(DE3) cells harboring plasmids pETCbpA-X₇-H₆ and pETCbpM-X₇-H₆, respectively. The cells were grown at 37°C to an A₆₀₀ of 0.4 and then induced for 2 h with 1.0 mM isopropyl-β-d-thiogalactopyranoside. CbpA-X₇-H₆ and CbpM-X₇-H₆ were purified by chromatography with Talon resin according to the manufacturer's instructions.

We prepared [³H]oriP1 plasmid DNA (2,740 cpm/fmol) as described in ref. 26. All proteins were >95% pure, as determined by SDS/PAGE. Protein concentrations are expressed as molar amounts of ClpA hexamers, ClpX hexamers, ClpP tetradecamers, DnaJ dimers, DnaK monomers, CbpA dimers, CbpM monomers, dimers of GFP-RepA fusion proteins, and monomers of GFP derivatives.

RepA Activation Assay. After incubation with ClpA (24) or DnaK and DnaJ (37), RepA DNA binding was measured by using a nitrocellulose-filter binding assay, as described in ref. 24.

ClpAP and ClpXP Degradation Assays. Degradation of GFP fusion proteins by ClpAP or ClpXP was measured by monitoring GFP fluorescence or by densitometric analysis of degradation products after SDS/PAGE, as described in ref. 24.

Gel Filtration. GFP-RepA-H₆ (0.5 μM) or GFP-RepA (0.5 μM) was incubated with ClpA (2 μM) in the presence of ATP[γ-S] and developed over a Sephacryl S-300 column, as described in ref. 24.

In Vivo Degradation Assay. E. coli strains BL21(DE3) and SG1146a (BL21, DE3, and ΔclpP) harboring a plasmid expressing GFP-X₃₀-H₆, GFP-X₇-H₆, GFP-X₃-H₆, GFP-H₆, or GFP-X₇ were grown in LB broth containing 25 μg/ml kanamycin to late logarithmic phase and diluted to A₆₀₀ = 0.3. We then spotted 2 μl of each culture on LB agar plates containing 25 μg/ml kanamycin and incubated them at 23°C in the dark. Plates were photographed by using visible light or UV transillumination (312 nm).