ClpAP and ClpXP degrade proteins with tags located in the interior of the primary sequence

Abstract

Clp/Hsp100 ATPases comprise a large family of ATP-dependent chaperones, some of which are regulatory components of two-component proteases. Substrate specificity resides in the Clp protein and the current thinking is that Clp proteins recognize motifs located near one or the other end of the substrate. We tested whether or not ClpA and ClpX can recognize tags when they are located in the interior of the primary sequence of the substrate. A protein with an NH₂-terminal ClpA recognition tag, plasmid P1 RepA, was fused to the COOH terminus of green fluorescent protein (GFP). GFP is not recognized by ClpA or ClpX and is not degraded by ClpAP or ClpXP. We found that ClpA binds and unfolds the fusion protein and ClpAP degrades the protein. Both the GFP and RepA portions of the fusion protein are degraded. A protein with a COOH-terminal ClpX tag, MuA, was fused to the NH₂ terminus of GFP. ClpXP degrades MuA-GFP, however, the rate is 10-fold slower than that of GFP-MuA. The MuA portion but not the GFP portion of MuA-GFP is degraded. Thus, a substrate with an internal ClpA recognition motif can be unfolded by ClpA and degraded by ClpAP. Similarly, although less efficiently, ClpXP degrades a substrate with an internal ClpX recognition motif. We also found that ClpA recognizes the NH₂-terminal 15 aa RepA tag, when it is fused to the COOH terminus of GFP. Moreover, ClpA recognizes the RepA tag in either the authentic or inverse orientation.

Clp/Hsp100 ATPases comprise a large family of homologous ATPases with ATP-dependent molecular chaperone activity. They participate in many cellular functions, including DNA replication, tolerance to heat stress, control of gene expression, and protein degradation (1, 2). Several Clp proteins, including Escherichia coli ClpA, ClpB, and ClpX, and yeast Hsp104, act in protein remodeling reactions in vitro such as disassembly of complexes and the reactivation and disaggregation of denatured proteins (3–7). In addition, some Clp proteins play a direct role in protein degradation by associating with a proteolytic component to form ATP-dependent proteases. For example, ClpA, ClpX, and HslU of E. coli are the ATPase components of ClpAP, ClpXP, and HslUV, respectively (8, 9).

Structural studies have shown that Clp ATPases self-assemble into oligomeric rings in the presence of ATP or nonhydrolyzable ATP analogs (10, 11). When associated with a proteolytic component, the ATPase rings are located at either or both ends of the proteolytic core forming a structure resembling the eukaryotic 26S proteasome (10, 12). The crystal structures of ClpP and HslV show that the proteolytic active sites are in an internal chamber formed by two stacked rings of identical subunits (13, 14). Access to the proteolytic chamber appears to be through narrow pores at either end of the stacked rings. However, the pores, which measure about 10 Å in ClpP (13), are not wide enough to allow the passage of globular proteins.

The pathway of proteolysis suggested by the structural and biochemical data is that protein substrates are specifically recognized and unfolded by the Clp ATPase component and then threaded through the apical pore of the protease component and into the proteolytic chamber (2, 13, 15). Aspects of this model have been verified. ClpA and ClpX bind substrates specifically and mediate ATP-dependent protein unfolding of tagged substrates (16–19). When ClpP is present, the unfolded substrates are translocated from ClpA or ClpX into the cavity of ClpP in an ATP-dependent manner (19–21). In support of the threading model of substrate translocation from the Clp ATPase to the protease, recent results show that ClpAP and ClpXP degrade substrates processively starting from the degradation signal (22, 23).

The initial step in both protein remodeling by Clp chaperones and degradation by Clp proteases is substrate recognition and binding by the Clp protein. For the substrates studied to date, Clp ATPases recognize either the NH₂- or COOH-terminal region of the substrate. For several substrates, including MuA (4), Mu vir repressor (24), and SsrA-tagged polypeptides (25–27), sites in the COOH-terminal sequences of the substrates are primarily responsible for recognition by ClpX and/or ClpA. For others, including certain β-galactosidase fusion proteins bearing hydrophobic NH₂-terminal amino acids (28, 29), RepA (17, 30), HemA (31), UmuD′ (32), and λO (33), recognition is through sites near the NH₂ terminus. Although several specific recognition signals for ClpA and ClpX have been identified, general sequence rules governing substrate recognition by either protein are not yet apparent.

We examined whether substrate recognition tags could be moved from their naturally occurring position near one end of a substrate to another location without destroying the ability of the tags to promote remodeling by ClpA and degradation by ClpAP or ClpXP. We found that ClpA recognizes, unfolds, and translocates to ClpP for degradation a substrate with a recognition tag located internally in the primary sequence of the substrate. ClpXP also degrades a substrate with an internal tag, however, for the ClpA- and ClpX-specific substrates tested, degradation was much more robust with ClpAP than ClpXP.

Experimental Procedures

Materials.

ATP and γ-thio-ATP (ATP[γS]) were from Roche Applied Science. Restriction endonucleases and DNA-modifying enzymes were from New England Biolabs and PCR reagents were from Perkin–Elmer Life Sciences.

Plasmids and Strains.

To generate pBAD-RepA-GFP a repA PCR products containing 5′ NheI and ribosome binding sites and a 3′ KpnI site was cut and ligated into NheI- and KpnI-digested pBAD-GFP (GFP, green fluorescent protein) (17). To generate pET-GFP-RepA and pET-GFP-RepAΔ50 the internal NdeI site in gfpuv was first removed by PCR mutagenesis with pGFPuv (CLONTECH) as the starting template. gfpuv(NdeIΔ) was then PCR-amplified and the stop codon was removed with NdeI containing 5′ and 3′ oligonucleotides. The product was digested with NdeI and ligated into NdeI-digested pET-RepAH₆ and pET-RepAΔ50H₆. pET-RepAH₆ and pET-RepAΔ50H₆ were constructed by generating repA PCR fragments coding for RepA (amino acids 1–286) and RepAΔ50 (amino acids 51–286) that contain 5′ and 3′ NdeI and HindIII sites, respectively. The products were digested with NdeI and HindIII and ligated into similarly digested pET24b (Novagen). To construct pBAD-GFP-RepA(1–15) and pBAD-GFP-RepA(15–1) the 3′ terminal SacI site in gfpuv was used to insert annealed oligonucleotides to yield protein fusions containing RepA amino acids 1–15 fused to the COOH terminus of GFPuv in the authentic order of 1–15 and the inverted order of 15–1.

To construct pET-EGFP-L-MuA, an enhanced GFP (EGFP) PCR fragment (without a stop codon) was generated by using pEGFP (CLONTECH) as a template and NdeI-site-containing 5′ and 3′ oligonucleotides. The egfp PCR fragment and a muA PCR fragment flanked by BamHI sites were individually cloned into the NdeI site and the BamHI site, respectively, in pET-3a (Novagen). To generate pET-MuA-EGFP and pET-MuAΔ48-EGFP, an egfp PCR fragment was first cloned between the NdeI and BamHI sites in pET-3c. The resulting plasmid was digested by NdeI and ligated with a muA PCR fragment of appropriate length flanked by NdeI sites.

All fusion sequences were verified by DNA sequencing.

Proteins and DNA.

P1 RepA (34), ClpA (35), ClpX (36), and ClpP (35) were purified as described. GFP, RepA-GFP, RepA(1–15)-GFP, GFP-RepA(1–15), and GFP-RepA(15–1), were isolated from SG22215 cells harboring plasmids expressing those proteins as described (17).

GFP-RepAH₆ and GFP-RepAΔ50H₆ were isolated from BL21(DE3) cells harboring pET-RepAH₆ and pET-RepAΔ50H₆, respectively, after overnight induction with 0.1 mM isopropyl β-d-thiogalactoside (IPTG) at 23°C. The fusion proteins were purified by immobilized-metal-affinity chromatography using Talon resin (CLONTECH) as recommended by the manufacturer.

MuA-GFP, GFP-L-MuA, and MuAΔ48-GFP fusion proteins were isolated from BL21(DE3) cells carrying the plasmids expressing the gene fusions. Cells were grown in LB-broth containing 100 mg/ml ampicillin at 27°C to OD₆₀₀ of 1. IPTG was added to 0.4 mM, and 4 h after induction the cells were harvested by centrifugation at 4°C. Purification of the fusion proteins was carried out as described for MuA (37). Final protein solutions were dialyzed against 25 mM Hepes, pH 7.6/500 mM KCl/0.1 mM EDTA/1 mM DTT/10% glycerol (vol/vol).

GroEL(D87K) was prepared as described (38). The plasmid used for the expression of GroEL(D87K) was kindly provided by Arthur Horwich (Yale University).

RepA was ³H-labeled in vitro by using N-succinimidyl [2,3-³H]propionate (20). Protein concentrations are expressed as molar amounts of RepA, RepA-GFP, GFP-RepA, and GFP-RepAΔ50 dimers, ClpA and ClpX hexamers, ClpP tetradecamers, and GFP-L-MuA, MuA-GFP, MuAΔ50-GFP, RepA(1–15)-GFP, GFP-RepA(1–15), and GFP-RepA(15–1) monomers.

[³H]oriP1 plasmid DNA (3,800 cpm/fmol) was prepared as described (39).

RepA Activation Assay.

Reaction mixtures (20 μl) contained buffer A [20 mM Tris⋅HCl, pH 7.5/100 mM KCl/5 mM DTT/0.1 mM EDTA/10% glycerol (vol/vol)], 1 mM ATP, 10 mM MgCl_2, 50 μg/ml BSA, 0.005% Triton X-100, 0.07 pmol of ClpA, and RepA as indicated. After 10 min at 23°C the mixtures were chilled to 0°C. Calf thymus DNA (1 μg) and 14 fmol of [³H]oriP1 plasmid DNA were added. After 5 min at 0°C, the mixtures were filtered through nitrocellulose filters and the retained radioactivity was measured.

Degradation Assays.

Reaction mixtures were assembled in 100 μl of buffer A/20 mM MgCl₂/5 mM ATP/20 mM creatine phosphate/6 μg of creatine kinase/ClpA or ClpX, ClpP, and a RepA or MuA fusion derivative as indicated in the figure legends. The decrease in relative fluorescence was measured at 23°C as a function of time by using a Perkin–Elmer LS50B luminescence spectrophotometer equipped with a well plate reader. Excitation and emission wavelengths were 395 nm and 510 nm for GFP-containing fusions and 475 nm and 510 nm for EGFP-containing fusions.

Alternatively, reaction mixtures were assembled as above and, after incubation at 23°C for the times indicated, trichloroacetic acid (TCA) was added to 20% (wt/vol). The TCA pellets were analyzed by SDS/PAGE and the amount of substrate was quantitated by densitometry.

RepA-GFP Unfolding Assays.

GFP fusion proteins were incubated with 200 pmol of ClpA in 100-μl reaction mixtures containing buffer A, 100 μg/ml BSA, 0.005% Triton X-100, 5 mM ATP, 20 mM creatine phosphate, and 6 μg of creatine kinase. Protein unfolding was measured as described above for the fluorescent proteolysis assay but ClpP was omitted. Where indicated, mixtures contained 2.5 μM GroEL(D87K).

Results

ClpA Recognizes the Same Tag When It Is at the NH₂ or COOH Terminus of a Protein and When It Is in Its Authentic or Inverse Orientation.

We wanted to know whether a ClpA-recognition motif could function as a tag when moved from its normal location at one end of a substrate to the other end. We were also curious as to whether the tag could be recognized when in the inverse orientation. P1 RepA was chosen for these studies because it is a well-characterized substrate for remodeling by ClpA and for degradation by ClpAP (3, 17, 20). GFP was chosen because ClpA does not recognize it (16, 17). We previously showed that the RepA tag is located between amino acids 10 and 15 (30). Moreover, when the NH₂-terminal 15 aa of RepA are placed at the NH₂ terminus of GFP, the tag is sufficient to target the fusion protein for unfolding by ClpA and degradation by ClpAP (30). We constructed fusion proteins in which the NH₂-terminal 15-aa tag of RepA was joined to the COOH terminus of GFP in the authentic, GFP-RepA(1–15), and inverse, GFP-RepA(15–1) orientation, to compare with the fusion protein in which the NH₂-terminal 15 aa of RepA were fused to the NH₂ terminus of GFP, RepA(1–15)-GFP (Fig. 1A).

Fig 1.

Unfolding of GFP fusion proteins containing NH₂- or COOH-terminal RepA tags. (A) RepA-tagged GFP substrates. (B) Steady-state unfolding of RepA(1–15)-GFP (blue circles), GFP-RepA(1–15) (red triangles), and GFP-RepA(15–1) (black squares) in the presence of ClpA (filled symbols) and absence of ClpA (open symbols). Unfolding reactions were carried out as described in Experimental Procedures using 1 μM substrate and 2 μM ClpA, where indicated. Fluorescence was monitored over time. (C) Unfolding in the presence of GroEL(D87K) of RepA(1–15)-GFP (blue circles), GFP-RepA(1–15) (red triangles), and GFP-RepA(15–1) (black squares) in the presence (filled symbols) or absence (open symbols) of ClpA. Reactions were performed as in B with the exception that 2.5 μM GroEL(D87K) was added to the mixtures.

We tested whether ClpA unfolds the GFP fusion proteins. The Horwich laboratory showed that protein unfolding by ClpA could be detected by monitoring the decrease in fluorescence of a tagged GFP fusion protein after incubation with ClpA and ATP in the presence of GroEL(D87K), a mutant form of GroEL (38), which binds and sequesters unfolded proteins (16). When GFP-RepA(1–15), GFP-RepA(15–1), and RepA(1–15)-GFP were incubated with ClpA and ATP, the fluorescence intensity of all three substrates decreased by 10–20%, indicating steady-state unfolding and refolding (Fig. 1B). There was no change in the fluorescence of these substrates without ClpA (Fig. 1B) or of GFP with or without ClpA (data not shown; refs 16, 17). When GroEL(D87K) was added to ClpA unfolding reaction mixtures to trap unfolded proteins released by ClpA, the fluorescence of GFP-RepA(1–15) decreased 90%, GFP-RepA(15–1) decreased 60%, and RepA(1–15)-GFP decreased 20% (Fig. 1C). More unfolded GFP was trapped by GroEL(D87K) when the tag was located at the COOH terminus compared with the NH₂ terminus of GFP. These results show that the tag targets GFP for unfolding by ClpA when located at either end of the protein and in either orientation.

We then measured degradation of the NH₂- and COOH-terminally tagged GFP fusion proteins by ClpAP. About 90% of GFP-RepA(1–15) was degraded in 10 min as measured by a decrease of fluorescence. GFP-RepA(15–1) and RepA(1–15)-GFP were degraded more slowly; 50% of the fluorescence was lost in 20 min (Fig. 2A). The fluorescence of GFP was unchanged by incubation with ClpAP (data not shown; refs. 16 and 17). Degradation was confirmed by quantitating the disappearance of the proteins by SDS/PAGE. After a 20-min incubation with ClpAP, using conditions shown in Fig. 2A, 87% of the GFP-RepA(1–15), 47% of the GFP-RepA(15–1), and 43% of the RepA(1–15)-GFP were degraded. We observed that GFP containing a COOH-terminal 10-aa extension that was not known to be a ClpA recognition tag was degraded about 10% after a 20-min incubation with ClpAP under the same conditions as used in Fig. 2A (data not shown).

Fig 2.

Degradation of GFP fusion proteins containing NH₂- or COOH-terminal RepA tags by ClpAP and ClpXP. (A) Degradation of RepA(1–15)-GFP (blue circles), GFP-RepA(1–15) (red triangles), and GFP-RepA(15–1) (black squares) by ClpAP. Degradation of GFP substrates was measured as described in Experimental Procedures, using 1 μM substrate in the presence of 1 μM ClpA and 1 μM ClpP (filled symbols) or absence of ClpAP (open symbols). Fluorescence was monitored over time. (B) Degradation of GFP-RepA(1–15) (filled red triangles), GFP-RepA(15–1) (filled black squares), and GFP-SsrA (filled green circles) by ClpXP. Degradation was measured as described in Experimental Procedures, using 1 μM substrate, 0.5 μM ClpX, and 0.5 μM ClpP.

Thus, the same tag can promote degradation of an otherwise unrecognized substrate, GFP, independent of the position of the tag at the NH₂ or COOH terminus of the protein. Both the unfolding and the degradation experiments suggest that ClpA is more effective in promoting the unfolding of GFP when the tag is at the COOH terminus of GFP than when it is at the NH₂ terminus of GFP. The observation that the tag is recognized when in the inverse orientation shows that the interaction of ClpA with the tag sequence does not require a specific orientation, although the fusion containing the authentically oriented tag was more efficiently unfolded by ClpA and degraded by ClpAP.

We next tested whether the COOH-terminally tagged GFP substrates were specifically recognized by ClpAP and not ClpXP. When ClpXP was incubated with GFP-RepA(1–15) or GFP-RepA(15–1), there was no detectable loss of fluorescence (Fig. 2B) or disappearance of the substrates by SDS/PAGE (data not shown). With the same conditions, ClpXP degraded GFP-SsrA, as previously shown (Fig. 2B; refs. 18 and 19). Thus, the RepA tag is specific for ClpAP regardless of its location or orientation.

ClpA Recognizes Tags Located in the Interior of the Primary Sequence of a Protein and ClpAP Degrades Substrates with Internal Tags.

To address the question of whether ClpA recognizes a substrate in which the recognition signal has been moved from its natural location near an end to a position within the protein, we constructed several more fusion proteins that joined RepA to GFP. The RepA tag was placed in the interior of the primary sequence of the protein by fusing the NH₂ terminus of RepA to the COOH terminus of GFP (GFP-RepA). As a control, a RepA derivative lacking the first 50 aa of RepA was joined to the COOH terminus of GFP (GFP-RepAΔ50). Both of these fusion proteins contained a COOH-terminal His₆ tag to facilitate protein purification. For comparison, the RepA tag was left at its natural NH₂-terminal location by connecting the COOH terminus of RepA to GFP (RepA-GFP) (Fig. 3A).

Fig 3.

Complex formation between ClpA and GFP-RepA. (A) RepA and GFP fusion proteins. (B and C) Gel filtration of GFP-RepA with ClpA (B) or alone (C). GFP-RepA (100 pmol) was incubated alone or with 400 pmol of ClpA for 10 min at 23°C in 100-μl mixtures of S-200 buffer [20 mM Tris⋅HCl, pH 7.5/150 mM KCl/1 mM EDTA/1 mM DTT/5% glycerol (vol/vol)/50 μg/ml BSA/10 mM MgCl₂/0.005% Triton X-100] containing 2 mM ATP[γS]. Reaction mixtures were then applied to a 0.7 × 15 cm Sephacryl S-200 column equilibrated with S-200 buffer containing 0.5 mM ATP[γS]. The relative fluorescence of a portion of each fraction was measured. (D and E) Gel filtration of GFP-RepAΔ50 with ClpA (D) or alone (E).

Previous experiments showed that ClpA forms stable complexes with RepA and RepA(1–70)-GFP but does not associate with GFP (16, 17, 40). To determine whether ClpA could recognize the substrate with the internal tag, we incubated GFP-RepA with ClpA and ATP[γS] under conditions that support stable complex formation between ClpA and RepA and then analyzed the mixtures by gel filtration. Approximately 75% of the GFP-RepA appeared in the excluded volume of the column with ClpA, showing that GFP-RepA, like RepA, has a high affinity for ClpA (Fig. 3B). Because fluorescence was used to measure the elution position of the fusion protein, the result also shows that the GFP portion of the fusion protein was not unfolded simply by binding ClpA. In the absence of ClpA, GFP-RepA eluted significantly later, at the position expected for dimeric GFP-RepA (Fig. 3C). In contrast, only about 20% of the GFP-RepAΔ50 eluted with ClpA on gel filtration, indicating that the substrate lacking the NH₂-terminal RepA tag has a lower affinity for ClpA (Fig. 3D). When chromatographed alone, GFP-RepAΔ50 eluted at the position expected for the dimeric form (Fig. 3E). These results demonstrate that ClpA interacts with a substrate even when the recognition motif is not near the end of a protein.

We measured steady-state unfolding of the substrates (shown in Fig. 3A) by monitoring fluorescence in the presence of ClpA and ATP. We found that when either GFP-RepA or RepA-GFP was incubated with ClpA and ATP, the fluorescence intensity decreased 60–70% (Fig. 4A). It was observed that with time, fluorescence is regained because of ATP depletion and spontaneous refolding of the GFP fusion protein (30). After 60 min, Rep-GFP regained 75% of its original fluorescence, suggesting that the GFP fusion protein refolded into its native conformation (Fig. 4A). GFP-RepA regained fluorescence slowly, indicating that the GFP portion of the protein was not released from ClpA or did not spontaneously refold in the native conformation during the time of the experiment. When GFP-RepAΔ50 was incubated with ClpA and ATP, the fluorescence decreased about 15% (Fig. 4A). The addition of GroEL(D87K), which bound the unfolded GFP-RepAΔ50, resulted in a small decrease in fluorescence (data not shown), indicating that the protein was poorly unfolded by ClpA. These results suggest that ClpA is able to promote ATP-dependent unfolding of the GFP portion of a fusion protein in which the ClpA recognition signal is located internally between GFP and RepA.

Fig 4.

Unfolding and activation of GFP-RepA. (A) Steady-state unfolding of GFP-RepA (red triangles), RepA-GFP (black circles), and GFP-RepAΔ50 (blue squares) was measured as described in Experimental Procedures using 300 nM substrate and 2 μM ClpA. Fluorescence was monitored over time. (B) Activation of the DNA binding activity of RepA (green squares), RepA-GFP (black circles), and GFP-RepA (red triangles) was measured after incubation with ClpA (filled symbols) or without (open symbols) as described in Experimental Procedures. Results are the means (±SEM) of three independent experiments.

Having found that ClpA unfolded GFP-RepA, we thought it likely that ClpA might also activate the latent DNA-binding activity of GFP-RepA, as it does wild-type RepA. In the absence of ClpA, GFP-RepA and RepA-GFP, like RepA, were inactive in oriP1-specific DNA binding (Fig. 4B). Incubation with ClpA and ATP activated oriP1 DNA binding by all three proteins, although GFP-RepA and RepA-GFP were activated to a lesser extent than wild-type RepA (Fig. 4B). GFP-RepAΔ50 was not tested in this assay because deletion of the first 50 aa impairs DNA binding by RepA (30). These results demonstrate that ClpA is able to remodel RepA, which has an internal ClpA recognition signal.

We measured degradation of the internally tagged fusion proteins by ClpAP. GFP-RepA, RepA-GFP, and GFP-RepAΔ50 were incubated with ClpAP and ATP and degradation was measured both by a decrease in fluorescence and by the disappearance of the substrate as analyzed by SDS/PAGE. The fluorescent intensity of both RepA-GFP and GFP-RepA decreased at similar rates and after 30 min, there was a 40–50% decrease (Fig. 5A). A similar rate of degradation of RepA-GFP and GFP-RepA was observed by quantitating the disappearance of the substrate proteins by SDS/PAGE (Fig. 5B). Degradation products were not seen by the gel analysis, indicating that the GFP and RepA portions of both RepA-GFP and GFP-RepA were degraded by ClpAP to small polypeptides as expected (35). GFP-RepAΔ50 was also degraded, but at a 5-fold slower rate than GFP-RepA, consistent with the complex formation and unfolding results (Figs. 3D and 4A). It is possible that the COOH-terminal His₆ tag contributes to the recognition of GFP-RepAΔ50. Other experiments showed that the rate of degradation of RepAΔ50, without a His₆ tag, was about 20% the rate of wild-type RepA (data not shown), suggesting that RepAΔ50 is recognized by ClpAP, but has a lower affinity than the wild type. Previous work suggested that secondary ClpA recognition motifs exist in RepA between amino acids 15 and 70 (30) and it is possible that the residual degradation of RepAΔ50 and GFP-RepAΔ50 may reflect the presence of alternative sites.

Fig 5.

ClpAP degradation of GFP fusion proteins containing NH₂-terminal and internal recognition tags. (A) RepA-GFP (black circles), GFP-RepA (red triangles), and GFP-RepAΔ50 (blue squares) (30 pmol) were incubated with 10 pmol of ClpA and 23 pmol of ClpP as described in Experimental Procedures. Fluorescence was monitored over time. (B) Reactions were as in A, but at the times indicated TCA was added to 20% (wt/vol). TCA pellets were analyzed by SDS/PAGE and substrate degradation was quantitated by densitometry.

Taken together, the results with the GFP-RepA fusion protein demonstrate that ClpA can recognize a substrate-binding motif when it is embedded in the interior of the amino acid sequence of a protein. Moreover, the results show that a substrate with an internal tag can be unfolded and remodeled by ClpA and degraded by ClpAP.

ClpXP Degrades a Substrate with an Internal Tag.

To find out whether ClpAP has the unique ability among Clp proteases to degrade substrates with internal tags, we tested whether ClpXP could degrade an internally tagged substrate. MuA transposase was used for these experiments because it is degraded by ClpXP (4, 36). The ClpX recognition motif resides in the COOH-terminal 10 aa. A fusion protein was constructed with GFP COOH-terminal to MuA, positioning the tag between the MuA and GFP portions of the protein. Two control fusion proteins were made, one in which a MuA derivative lacking the COOH-terminal 48 aa was fused to the NH₂ terminus of GFP, MuAΔ48-GFP, and one with MuA COOH-terminal to GFP, GFP-MuA, placing the tag at its natural COOH-terminal location (Fig. 6A). This protein and several others that were tested had a 14-aa linker between the MuA and GFP portions of the fusion. However, no differences were seen in the ability of the fusion proteins with and without linkers to serve as substrates (data not shown).

Fig 6.

ClpXP degradation of GFP fusion proteins containing COOH-terminal and internal recognition tags. (A) GFP and MuA fusion proteins. (B) GFP-MuA (black circles), MuA-GFP (red triangles), and MuAΔ48-GFP (blue squares) (15 pmol) were incubated with 32 pmol of ClpX and 35 pmol of ClpP as described in Experimental Procedures. Fluorescence was monitored over time. (C) Degradation reactions were carried out as in B but contained 20 μg/ml BSA and were in 20 μl. At the times indicated, TCA was added to 20% (wt/vol) and the TCA pellets were analyzed by SDS/PAGE. Degradation was quantitated by densitometry. Results are the means (±SEM) of three (for the measurement of GFP-MuA degradation) or five (for the measurement of MuA-GFP and MuAΔ48-GFP) independent experiments. (D) Degradation reactions were carried out as in C, but after SDS/PAGE the proteins were transferred to poly(vinylidene difluoride) (PVDF) membranes and Western blot analysis was performed using GFP antibody (CLONTECH).

When we tested whether ClpXP could degrade MuA-GFP and MuAΔ48GFP by measuring fluorescence, there was no detectable change in fluorescence (Fig. 6B). In contrast, ClpXP degraded GFP-MuA efficiently as indicated by a rapid 90% decrease in fluorescence (Fig. 6B). We then measured degradation by quantitating the disappearance of the substrate band by SDS/PAGE and found that the results were somewhat different. With this assay, MuA-GFP was degraded. However, it was degraded 10-fold more slowly than GFP-MuA (Fig. 6C). There was also detectable degradation of MuAΔ48GFP, although consistently 2 to 4-fold less than seen with MuA-GFP (Fig. 6C). With all three substrates, degradation required ClpX, ClpP, and ATP. Thus, although the amount of degradation was very small, ClpXP maintained some preference for the internally tagged substrate over the untagged substrate.

We observed that as the MuA-GFP and MuAΔ48GFP bands disappeared with time of incubation with ClpXP, a product about 3 kDa larger than GFP appeared. Western blot analysis using GFP antibody confirmed that the degradation products contained GFP (Fig. 6D and data not shown). This observation explains the apparent discrepancy between the lack of decrease in fluorescence and the disappearance of the substrate by SDS/PAGE analysis and is in contrast to the results with ClpAP in which no intermediate degradation products of RepA-GFP or GFP-RepA were detected by PAGE.

The results with ClpXP demonstrate that ClpXP, like ClpAP, degrades a substrate whose recognition signal is located in the interior of the primary sequence of the protein. However, ClpAP degrades GFP-RepA much more efficiently than ClpXP degrades MuA-GFP.

Discussion

The results presented here shed light on the mechanism of substrate recognition and commencement of unfolding by Clp ATPases and degradation by Clp proteases. We have shown that ClpA binds and unfolds a substrate whose recognition tag is located in the interior of the primary sequence of the substrate and translocates the substrate to ClpP for degradation. ClpXP also degrades a substrate with an internal ClpX recognition signal; however, ClpXP is much less efficient at this process than ClpAP with the two substrates tested. It has been assumed that substrate recognition motifs must be near an end to initiate unfolding and translocation of the protein from the Clp ATPase through the small pore leading into the proteolytic chamber of the proteolytic component. Consistent with this notion, the substrates studied to date (fewer than 10) possess recognition signals near one end.

The threading model of substrate translocation is supported by a recent study by Horwich and coworkers (23). Their results with ClpAP demonstrated that the tagged end of a COOH-terminally SsrA-tagged substrate enters ClpP first, as measured by the fluorescent anisotropy of substrates labeled with fluorescein near one or the other terminus and by fluorescent resonance energy transfer between donor fluorophores in the ClpP cavity and the fluorescein-labeled substrates. In another recent study Matouschek and coworkers (22) observed that both ClpAP and ClpXP degrade the protein domain nearest to the recognition tag first. In these experiments, a ClpA tag was fused to the NH₂ terminus of two fusion proteins, dihydrofolate reductase (DHFR)-barnase and barnase-DHFR, and a ClpX tag was fused to the COOH terminus of the same two fusion proteins. When DHFR was placed next to the ClpA or ClpX recognition signal, methotrexate protected both DHFR and barnase from degradation by ClpAP and ClpXP, respectively. When barnase was next to the tag, methotrexate protected only the DHFR domain from degradation. Both of these studies show that degradation proceeds from the position of the tag.

For internally tagged substrates it is not known whether an untagged end or a loop in an unfolded polypeptide chain enters ClpP first. Studies by Burton et al. (41) demonstrated that the pore through which unfolded proteins enter the ClpP proteolytic chamber is large enough to accommodate simultaneous passage of two polypeptide chains. They observed that ClpXP is able to degrade disulfide-crosslinked SsrA-tagged Arc repressor dimers. Studies by Hoskins et al. (17) showed that untagged proteins, which are not bound by ClpA in their native form, can be degraded by ClpAP when the ATP-dependent unfolding step is bypassed by unfolding the substrate with chemical denaturants. Thus, the dispensability of tags for degradation after the unfolding step demonstrates that it is not essential for a tagged end to enter ClpP first for a protein to be degraded. Preliminary data suggest that the COOH-terminal extension of GFP-RepA provided by the histidine tag may play a role in efficient unfolding and degradation of the protein (unpublished observations). However, the His₆ extension itself is insufficient to serve as an effective recognition signal. Further work is necessary to determine the mechanism of degradation of internally tagged proteins.

The observation that ClpXP did not degrade MuA-GFP to completion, yielding GFP plus about 3 kDa of the COOH-terminal portion of MuA, is consistent with the observations of Lee et al. (22). They found that when ClpAP and ClpXP encountered an unfolding-resistant domain, such as DHFR stabilized by methotrexate, an undegraded tail of about 30 aa remained attached to the folded domain. They suggest that 30 aa may be the length of protein necessary to reach from the substrate translocation site on the Clp ATPase to the proteolytic cavity of ClpP. The peptide tail attached to GFP that remains after the degradation of MuA-GFP by ClpXP very likely includes the 10-aa ClpX recognition signal.

The present study shows that ClpA recognizes the 15-aa RepA tag when it is moved from the NH₂- to the COOH terminus of GFP. Moreover, the COOH-terminally tagged GFP-RepA(1–15) is degraded at a significantly faster rate than the NH₂-terminally tagged RepA(1–15)-GFP. It is possible that Clp proteins translocate substrates more efficiently in the COOH-to-NH₂ direction than in the reverse direction. However, if there is a directional preference, it must be slight because ClpA and ClpX degrade both NH₂- and COOH-terminally tagged substrates. To date, the ability of Clp proteins to translocate substrates more efficiently in one direction or the other has not been distinguished from the inherent ability of a substrate protein to be unfolded more efficiently from one end or the other. The present results suggest that it is likely that the structure of GFP is more easily unfolded by ClpA when the tag is located on the COOH terminus than when it is on the NH₂ terminus, because the rate-limiting step in degradation by ClpAP and ClpXP is unfolding (18, 19). A similar suggestion emerges from the experiments with ClpXP showing that when the MuA tag is COOH-terminal to GFP, the GFP portion of the fusion protein is degraded and when the MuA tag is NH₂-terminal to GFP, the GFP portion is not degraded.

Our working model is that substrate recognition by ClpA involves a relatively unstructured short amino acid segment containing a set of preferred amino acids, rather than a segment with a unique three-dimensional structure. After the substrate is bound, protein unfolding commences, even when the primary recognition motif is located in the interior of the primary sequence of the substrate. The question of whether substrates are unfolded in a directional or concerted fashion remains to be answered. Once unfolded, the substrate is translocated to the cavity of ClpP, beginning either from an untagged end or, less likely, from an internal loop in an unfolded polypeptide chain.

Abbreviations

• TCA, trichloroacetic acid

• GFP, green fluorescent protein

• ATP[γS], γ-thio-ATP