The C Terminus of ς³² Is Not Essential for Degradation by FtsH

Abstract

A key step in the regulation of heat shock genes in Escherichia coli is the stress-dependent degradation of the heat shock promoter-specific ς³² subunit of RNA polymerase by the AAA protease, FtsH. Previous studies implicated the C termini of protein substrates, including ς³², as degradation signals for AAA proteases. We investigated the role of the C terminus of ς³² in FtsH-dependent degradation by analysis of C-terminally truncated ς³² mutant proteins. Deletion of the 5, 11, 15, and 21 C-terminal residues of ς³² did not affect degradation in vivo or in vitro. Furthermore, a peptide comprising the C-terminal 21 residues of ς³² was not degraded by FtsH in vitro and thus did not serve as a recognition sequence for the protease, while an unrelated peptide of similar length was efficiently degraded. The truncated ς³² mutant proteins remained capable of associating with DnaK and DnaJ in vitro but showed intermediate (5-amino-acid deletion) and strong (11-, 15-, and 21-amino-acid deletions) defects in association with RNA polymerase in vitro and biological activity in vivo. These results indicate an important role for the C terminus of ς³² in RNA polymerase binding but no essential role for FtsH-dependent degradation and association of chaperones.

The expression of heat shock genes in Escherichia coli is positively controlled at the transcriptional level by the product of the rpoH gene, the ς³² subunit of RNA polymerase (RNAP) (5, 8, 30). Stress treatment of the cells, such as a sudden temperature upshift, induces transient heat shock gene expression until the cells have adapted to the applied stress. This heat shock response is mediated by increases in the translation of rpoH, stabilization of ς³², and activation of ς³² by sequestration of the DnaK chaperone and its DnaJ cochaperone from a complex with ς³².

Under steady-state growth conditions, ς³² is an extremely unstable protein with half-lives at 30 and 42°C of less than 1 and 4 min, respectively (17, 23, 26). Heat shock treatment of the cells by transfer from 30 to 42°C increases the half-life of ς³² transiently to approximately 10 min (23). Degradation is mediated mainly by FtsH (HflB), an ATP-dependent metalloprotease associated with the inner membrane (10, 25, 27, 28). The interaction of ς³² with the RNAP core enzyme prevents degradation, indicating that FtsH and RNAP compete for binding to ς³² (28). The DnaK chaperone system plays an active but mechanistically unclear role in ς³² degradation, since ς³² is stabilized in dnaK and dnaJ mutant backgrounds (22, 24, 25, 28).

A conceptually interesting question with respect to ς³² degradation is how this biologically active, and hence seemingly folded, protein can be subject to such efficient proteolysis. Either ς³² carries specific degradation signals within its polypeptide, or it is thermodynamically unstable despite its biological activity. Earlier work showed that the in vivo half-life of fusions between N-terminal fragments of ς³² and β-galactosidase increased when a stretch of 23 residues (R122 to Q144), located between conserved regions 2 and 3 of ς³² and termed region C, is deleted or replaced by another reading frame (17). However, subsequent analysis of ς³² mutants altered in region C showed that this region is not essential for degradation but instead plays a role in ς³² binding to RNAP (1). Another study identified a role for the C terminus of ς³² in degradation by FtsH (3). In this previous study, rpoH genes with truncated 3′ ends were cloned into an expression vector such that fusion proteins between C-terminally truncated ς³² proteins (by 20 residues) and six amino acids from the vector sequence were generated. These fusion proteins were more stable than wild-type ς³² in E. coli cells and in vitro. However, this approach did not exclude the possibility that the vector-encoded extra residues at the C terminus artificially stabilize the fusion proteins.

In this study we investigated further the role of the C terminus of ς³² in the degradation process. Our analysis of C-terminally truncated ς³² proteins shows that the authentic C-terminal residues are not essential for degradation.

MATERIALS AND METHODS

Strains, plasmids, and media.

Cells of strains C600 (thr-1 leuB6 thi-1 lacY supE44 rfbD1 fhuA21) and BB2019 [GW1000 recA441 sulA11 Δ(argF-lac)U169 supC(Ts) rpoH165(Am) pDMI,1] (6) were grown aerobically at 30 or 42°C in Luria broth (LB) or in M9 minimal medium supplemented with glucose (0.2%; M9-Glu), thiamine (20 μg/ml), and appropriate amino acids (50 μg/ml). Growth media were further supplemented with isopropyl-β-d-thiogalactopyranoside (IPTG; 1 mM), kanamycin (20 μg/ml), and ampicillin (50 μg/ml) when required. The wild-type rpoH gene cloned into plasmid pUHE21-2fdΔ12 (7) was used as template for construction of 3′-truncated rpoH alleles. These alleles (rpoH-Δ5aa, rpoH-Δ11aa, rpoH-Δ15aa, and rpoH-Δ21aa) were generated by PCR using appropriate oligonucleotides with HindIII and PstI restriction sites for cloning of the coding sequences into pUHE21-2fdΔ12. For production of hexahistidine-tagged ς³² variants, wild-type and mutant alleles of rpoH were subcloned by inserting the HindIII-PstI fragments into pUHE212-1 (to encode N-His₆-ς³²). For immunoblotting and pulse-chase experiments, the rpoH alleles encoding wild-type ς³², N-His₆-ς³², C-His₆-ς³² (7), and C-terminally truncated ς³² proteins were subcloned by inserting the respective EcoRI-HindIII fragments into pFN476 (24). Plasmid pDMI,1 carrying lacI^q and encoding kanamycin resistance (14) was used to provide cells with a Lac repressor.

Overproduction of ς³² proteins and ex vivo degradation assay.

C600 cells carrying rpoH expression plasmids and pDMI,1 were grown at 30°C in 20 ml of LB containing ampicillin and kanamycin. IPTG (1 mM final concentration) was added to exponential-phase cultures (optical density at 600 nm [OD₆₀₀], 0.5 to 0.7) for 1 h, followed by harvesting and washing of the cells in buffer A (50 mM Tris-acetate [pH 8.0], 5 mM magnesium acetate, 2 mM β-mercaptoethanol, 50 mM KCl, 5% glycerol). The cell pellet was dissolved (OD₆₀₀, 50) in buffer A containing 1 mM phenylmethylsulfonyl fluoride and sonicated. The lysate was transferred to Eppendorf tubes and centrifuged at 2,500 × g for 2 min, and the supernatant was subjected to ultracentrifugation at 100,000 × g for 2 h. The protein content of the soluble cytoplasmic fraction was quantified by Bradford assay using bovine serum albumin as a standard. A 5-μg portion of the cytoplasmic fraction was mixed with FtsH reaction buffer (50 mM Tris-acetate [pH 8.0], 5 mM magnesium acetate, 2 mM β-mercaptoethanol, 50 mM KCl) to reach a final volume of 15 μl. Degradation assays were started by adding 1 μl of 100 mM ATP and 4 μl of 10 μM FtsH activated by the addition of Zn²⁺ (27) or buffer without FtsH as a control and incubated at 42°C for 1 h.

In vitro degradation of ς³².

Wild-type and C-terminally truncated N-His₆-ς³² proteins were purified and radiolabeled with N-succinimidyl[2,3-³H]propionate (Amersham) as previously described (7, 28). These proteins (1 μM each) were incubated with purified FtsH (2 μM) and tested for degradation as described previously (1, 28). After precipitation with trichloroacetic acid (TCA; 10%) followed by centrifugation (15,000 rpm, 3 min), the radioactive peptides generated by proteolysis were quantified in the supernatant by liquid scintillation counting. For assaying degradation of peptides, the final volume of the reaction was 60 μl. The ς³²-derived peptides Q132-Q151-C (QRKLFFNLRKTKQRLGWFNQC) and A264-A284-C (AERVRQLEKNAMKKLRAAIEAC) (50 μM each) were mixed and incubated with FtsH (1). At various time points, aliquots of 18 μl were mixed with 92 μl of 0.5% trifluoroacetic acid to stop the reaction. Products were analyzed by reverse-phase chromatography using a 5-to-80% acetonitrile gradient in 0.1% trifluoroacetic acid.

Analysis of protein interactions.

DnaK and RNAP core enzyme were purified as described previously (4, 15). Association of N-His₆-ς³² or C-terminally truncated N-His₆-ς³² with DnaK, DnaJ, and RNAP core was determined by gel filtration using a Superdex 200 column essentially as described previously (1, 7). To determine association of ς³² with RNAP core, N-His₆-ς³² (1 μM) was incubated with RNAP core (1.5 μM) for 10 min at 30°C in transcription buffer (20 μl, final volume). To determine association of ς³² with DnaK, DnaK (5 μM) was incubated for 2 h at 30°C in transcription buffer to disfavor oligomerization, mixed with N-His₆-ς³² (1 μM) in a final volume of 20 μl, and further incubated for 30 min at 30°C. These mixtures were placed on ice, adjusted to 100 μl by addition of transcription buffer, and loaded on a Superdex 200 column at 4°C. Labeled N-His₆-ς³² was detected in the elution fractions by liquid scintillation counting.

In vivo stability of ς³².

For determination of ς³² stability in vivo, C600 cells carrying rpoH expression plasmids were grown at 30°C in M9-Glu until mid-log phase. One milliliter of culture was labeled for 1 min with 70 μCi of [³⁵S]methionine followed by chase with unlabeled methionine (200 μg/ml, final concentration). After 30 s (time for synthesis of ς³²), aliquots of 200 μl were collected at various times, mixed with TCA (10%, vol/vol), and incubated for 15 min on ice. After centrifugation for 15 min at 14,000 rpm, the pellets were washed with acetone and resuspended in 50 mM Tris-HCl (pH 8.0), 1% sodium dodecyl sulfate (SDS), and 1 mM EDTA. Samples were subjected to immunoprecipitation using ς³²-specific rabbit antiserum as described previously (25, 29).

SDS-PAGE, immunoblotting, and quantifications.

Polyacrylamide gel electrophoresis (PAGE) was carried out as described by Laemmli (13) using SDS–12% polyacrylamide gels and staining with Coomassie brilliant blue. Immunoblotting was carried out according to standard procedures, using rabbit antisera specific for the relevant proteins as primary antibodies, and blots were developed with a Vistra ECF fluorescence Western blotting kit (Amersham) or 5-bromo-4-chloro-3-indolylphosphate–nitroblue tetrazolium color detection using alkaline phosphatase-conjugated anti-rabbit immunoglobulin G as the secondary antibody (Vector Laboratories, Inc.). Stained gels and developed immunoblots were scanned using a fluoroimager (FLA-2000) and quantified using MacBAS software (Fuji Film Co.).

RESULTS

Construction of truncated ς³² mutant proteins.

To investigate whether C-terminal residues of ς³² constitute a destabilizing element, we generated ς³² mutant proteins with truncations of their C termini by 5, 11, 15, and 21 residues (ς³²-Δ5aa, ς³²-Δ11aa, ς³²-Δ15aa, and ς³²-Δ21aa) (Fig. 1). The truncations were chosen such that the new termini are at two positions within (ς³²-Δ11aa and ς³²-Δ15aa), N-terminal to (ς³²-Δ21aa), or C-terminal to (ς³²-Δ5aa) an α-helix predicted by a secondary-structure-predicting algorithm (Fig. 1). PCR fragments of the appropriately truncated rpoH genes were cloned into the expression vector pUHE21-2fdΔ12 (7) such that a stop codon immediately follows the last codon, thus ensuring that no vector-encoded amino acids are fused to the ς³² mutant proteins. The C termini of the truncated ς³² mutant proteins do not resemble the consensus sequence of destabilizing C termini, as described by Sauer and coworkers (21), are not particularly hydrophobic, and do not constitute predicted DnaK binding sites (19).

FIG. 1.

Mutational alteration of ς³². The locations of conserved regions 1 to 4 and details of the C-terminal region and its predicted secondary structure (using the PhD program) are shown. α-helices are shown as cylinders. The end points of the C-terminal truncations of the ς³² mutant proteins are indicated.

In vivo activity of truncated ς³² mutant proteins.

To determine the in vivo activity of the truncated ς³² mutant proteins, we first tested the ability of pUHE21-2fdΔ12-borne rpoH mutant alleles to complement the temperature-sensitive growth of rpoH165(Am) mutant cells on LB agar plates. The rpoH165(Am) mutant cells carry a temperature-sensitive amber suppressor mutation which is active and supports growth at 30°C. Only the rpoH-Δ5aa mutant allele allowed complementation of growth at 42°C to an extent similar to that allowed by wild-type rpoH (data not shown). Even in the absence of IPTG, a condition in which read-through expression of the rpoH mutant alleles produced ς³² levels only approximately fivefold above wild-type levels, ς³²-Δ5aa complemented the growth defects of rpoH(Am) mutants at 42°C (data not shown).

We then determined the cellular levels of the ς³² mutant proteins after IPTG-induced overproduction in rpoH165(Am) mutant cells. We observed that wild-type ς³² was poorly overproduced, while ς³²-Δ5aa was overproduced to intermediate levels and ς³²-Δ11aa, ς³²-Δ15aa, and ς³²-Δ21aa were overproduced to high levels (Fig. 2).

FIG. 2.

In vivo activity of C-terminally truncated ς³² proteins. Cells of strain BB2019 [rpoH165(Am)] were transformed either with pUHE212-1 expressing wild-type or 3′-truncated rpoH alleles (producing N-His₆-ς³² proteins) or with pUHE212-1 control vector. Cultures of these cells were induced with IPTG, followed by SDS-PAGE of equal amounts of total protein and staining of the gels with Coomassie brilliant blue (top) or immunoblotting using DnaK- and DnaJ-specific sera (bottom). Lanes: 1, vector control; 2, N-His₆-ς³²; 3, N-His₆-ς³²-Δ5aa; 4, N-His₆-ς³²-Δ11aa; 5, N-His₆-ς³²-Δ15aa; 6, N-His₆-ς³²-Δ21aa.

Overproduction of wild-type ς³² led to strong increases in DnaK and DnaJ levels, overproduction of ς³²-Δ5aa led to intermediate increases, and overproduction of the other ς³² mutant proteins did not lead to increases in DnaK or DnaJ levels. The changes in DnaK and DnaJ levels were the result of the biological activity of ς³² in transcribing heat shock genes, including dnaK and dnaJ, and of the loss of this activity in the case of the ς³² mutant proteins with C-terminal truncations. Thus, through its ability to promote production of heat shock proteins, including FtsH and DnaK, the active ς³² triggers its own degradation and hence prevents stronger overproduction. Together, these results indicate that deletion of the ultimate C-terminal five residues (280 to 284) do not lead to complete loss of activity, whereas the segment between residues 264 and 279 is essential for the activity of ς³².

In vivo stability of truncated ς³² mutant proteins.

We determined the in vivo half-lives of two of the C-terminally truncated ς³² mutant proteins (ς³²-Δ11aa and ς³²-Δ15aa) by pulse-chase experiments followed by immunoprecipitation of ς³². Such experiments were complicated by our finding that most of the C-terminally truncated ς³² mutant proteins lack activity in vivo. Consequently, the synthesis of heat shock proteins is lower in cells producing the inactive mutant proteins from plasmids than in cells producing wild-type ς³² from plasmids (Fig. 2). These heat shock proteins include DnaK and DnaJ, which are limiting for degradation of ς³² (24, 25, 28). In the cells expressing the inactive ς³² mutant proteins, ς³² should be less subject to chaperone-mediated degradation. When the ς³² proteins are overexpressed, the efficiency by which ς³² mutant proteins are degraded in vivo will therefore be influenced by their biological activity. This complication led us to perform the ς³² half-life determinations in cells of strain C600 expressing low levels of plasmid-encoded ς³² proteins. Cells were transformed with plasmids expressing wild-type or mutant rpoH alleles under transcriptional control of a promoter specific for T7 polymerase. Since C600 cells lack T7 polymerase, only very little read-through from the T7 promoter occurs, just enough to produce approximately wild-type levels of ς³² (Fig. 3). This fact allowed us to perform the half-life determinations of the ς³² proteins under almost physiological conditions.

FIG. 3.

Cellular levels of plasmid-encoded ς³² mutant proteins. Cells of strain C600 transformed with plasmid pFN476 expressing the appropriate rpoH allele were grown at 30°C in LB medium. Aliquots of exponential-phase cultures (OD_600, 0.5 to 0.7) were collected, and equal amounts of total protein were analyzed by SDS-PAGE. Levels of ς³² were determined by immunoblotting.

Only minor differences existed between the half-life of wild-type ς³² (approximately 60 to 100 s) and those of ς³²-Δ11aa and ς³²-Δ15aa (Fig. 4). This result indicates that the C-terminal truncations do not affect significantly the half-life of ς³² in vivo provided that the levels of DnaK and DnaJ are kept constant.

FIG. 4.

In vivo stability of ς³² mutant proteins at 30°C. Cells of strain C600 which harbor pFN476 expressing rpoH-Δ11aa, rpoH-Δ15aa, or wild-type rpoH were grown at 30°C, pulse-labeled with [³⁵S]methionine, and chased with unlabeled methionine. Aliquots were taken at the indicated times, followed by immunoprecipitation of ς³² and quantification of the precipitated proteins relative to the largest value. Closed circles, wild-type ς³²; closed triangles, ς³²-Δ11aa; open triangles, ς³²-Δ15aa.

In vitro degradation of truncated ς³² mutant proteins by FtsH.

We investigated the susceptibility of N-His₆-tagged wild-type and C-terminally truncated ς³² mutant proteins to degradation by FtsH in vitro. These degradation assays were performed only at 42°C, since FtsH is known to be poorly active at 30°C when tested in vitro in the presence of detergent (12). We first determined whether exogenously added FtsH is capable of degrading the ς³² proteins present in extracts of C600 cells after IPTG-induced overproduction. The rationale for this experiment was that for these ex vivo degradation assays we could use authentic ς³² mutant proteins without histidine tags, ruling out any contribution by the tag. All truncated and wild-type ς³² proteins were degraded by FtsH in presence of ATP to similar extents (Fig. 5A).

FIG. 5.

(A) FtsH-dependent degradation of overproduced ς³² in cytoplasmic fractions. Cells of strain C600 which harbor plasmid pUHE21-2fdΔ12 expressing wild-type or truncated rpoH alleles mutant ς³² or the control plasmid were grown at 30°C with or without (lanes 1 and 7) IPTG. The extracts of cells grown without IPTG served as a control to ensure that the endogenous chaperones, whose cellular concentrations increase upon IPTG-induced production of plasmid-encoded wild-type ς³², do not interfere with the degradation of ς³² in the extracts. Cytosolic extracts were prepared and assayed for degradation by exogenously added FtsH. Lane M, molecular weight marker. All other lanes show extracts of cells producing plasmid-encoded wild-type ς³² supplemented with 2 μM purified wild-type ς³² (since wild-type ς³² could not be overproduced to large amounts) (lanes 1, 2, 7, and 8) or plasmid-encoded mutants ς³²-Δ5aa (lanes 3 and 9), ς³²-Δ11aa (lanes 4 and 10), ς³²-Δ15aa (lane 5 and 11), and ς³²-Δ21aa (lanes 6 and 12). The extracts were incubated with (+FtsH) or without (−FtsH) added protease. (B) In vitro degradation of ς³² mutant proteins by FtsH. ³H-labeled N-His₆-ς³² proteins were incubated with FtsH in the presence of 5 mM ATP, followed by TCA precipitation at the indicated times. The curves represent the percentage of radioactivity in the supernatants which contain the proteolytic fragments. Closed circles, N-His-ς³²; open circles, N-His₆-ς³²-Δ5aa; closed triangles, N-His₆-ς³²-Δ11aa; open squares, N-His₆-ς³²-Δ21aa.

We then determined with purified components the efficiency by which FtsH degrades the N-His₆-tagged ς³² mutant proteins. All ς³² proteins were soluble after overproduction in E. coli cells and had wild type-like elution profiles during nickel-nitrilotriacetic acid and ion-exchange chromatography. Furthermore, they were indistinguishable from wild-type protein with respect to the proteolysis pattern obtained by partial proteinase K and trypsin digestion (data not shown). Therefore, there is no indication of changes in their overall tertiary structures. All C-terminally truncated ς³² mutant proteins were degraded by FtsH in the presence of ATP, with kinetics similar to that of wild-type ς³² (Fig. 5B). Thus, the C-terminal truncations did not affect the efficiency of ς³² degradation by FtsH in vitro.

C-terminal residues of ς³² do not constitute a substrate for FtsH.

As an independent experimental approach to investigate the possibility that the C-terminal residues of ς³² constitute a destabilizing element for FtsH-dependent degradation, we tested whether the C-terminal sequences themselves provide a substrate motif for FtsH. We designed a peptide comprising the C-terminal 21 residues of ς³² and an additional cysteine at the C-terminal end (AERVRQLEKNAMKKLRAAIEAC) and tested whether it is a substrate for FtsH. The length of this 22-mer peptide was well above the minimal length of approximately 15 residues required for FtsH to degrade peptides (T. Tomoyasu and B. Bukau, unpublished results). As a positive control, we included a peptide of the same length and carrying a cysteine at the C terminus, derived from region C of ς³² (QRKLFFNLRKTKQRLGWFNQC). This peptide has been shown previously to be a good substrate peptide for FtsH (1).

In one experiment, both peptides were mixed at a 1:1 molar ratio and incubated with FtsH and ATP. Aliquots taken at 0, 10, and 20 min were analyzed for peptide degradation by high-pressure liquid chromatography. While the region C peptide was efficiently degraded with a half time of approximately 10 min, the C-terminal peptide was stable within the duration of the experiment (Fig. 6). This result was verified in experiments in which each peptide was tested individually for degradation by FtsH (data not shown). These results show that at the peptide level, the 22 residues of the C terminus of ς³² do not provide a recognition site for degradation by FtsH.

FIG. 6.

FtsH-mediated degradation of peptides derived from ς³². Peptides from region C (ς³²-Q132-Q151-C) and the C terminus (ς³²-A264-A284-C) of ς³² were mixed and incubated with FtsH. Degradation of the peptides at various times after mixing is shown as high-pressure liquid chromatography–reverse-phase chromatography data.

DnaK and DnaJ bind to truncated ς³² mutant proteins in vitro.

We tested whether DnaK binding to ς³² is impaired by the C-terminal truncations. ³H-labeled wild-type ς³² and mutant N-His₆-ς³² were incubated with DnaK followed by gel filtration to separate DnaK-ς³² complexes from free ς³² (Fig. 7). Under the conditions used, approximately 75% of wild-type [³H]ς³² was recovered in complex with DnaK (eluting in fractions 12 to 17). The ³H-labeled C-terminally truncated N-His₆-ς³² mutant proteins (ς³²-Δ5aa, ς³²-Δ11aa, and ς³²-Δ21aa) showed similar efficiencies of complex formation. Earlier work established that the interaction of DnaK with ς³² does not occur through the histidine tag (6). Furthermore, no defects in chaperone binding were observed when the N-His₆-ς³² mutant proteins were incubated with DnaK together with DnaJ in the presence of ATP (data not shown). These data indicate that the authentic C terminus of ς³² is not essential for interaction with DnaK and DnaJ.

FIG. 7.

Binding of ς³² mutant proteins to DnaK. ³H-labeled N-His₆-ς³², N-His₆-ς³²-Δ5aa, N-His₆-ς³²-Δ11aa, and N-His₆-ς³²-Δ21aa were incubated with DnaK, followed by gel filtration of the reaction product. Labeled protein was quantified in the elution fractions. Open circles, wild-type ς³²; closed circles, ς³²-Δ5aa; open squares, ς³²-Δ11aa; closed squares, ς³²-Δ21aa.

RNAP binding to truncated ς³² mutant proteins in vitro.

We considered the possibility that the truncated C-terminal segments of ς³² are involved in the interaction with the RNAP core enzyme. We determined the efficiency of association of ³H-labeled N-His₆-ς³² with RNAP by gel filtration. The relative amounts of all truncated ς³² proteins (ς³²-Δ5aa, ς³²-Δ11aa, ς³²-Δ15aa, and ς³²-Δ21aa) recovered in association with RNAP (eluting in fractions 8 to 15) were lower than that of wild-type ς³², with ς³²-Δ5aa showing the highest affinity for RNAP among all truncated proteins (Fig. 8). These results indicate a role for the C terminus of ς³² in the binding to RNAP.

FIG. 8.

Binding of ς³² mutant proteins to RNAP. ³H-labeled N-His₆-ς³² proteins were incubated with RNAP, followed by gel filtration. The amount of labeled protein was quantified. Open circles, wild-type ς³²; closed circles, ς³²-Δ5aa; open squares, ς³²-Δ11aa; closed squares, ς³²-Δ21aa.

Involvement of N-terminal segments of ς³² in degradation by FtsH in vitro.

Given our finding that the C terminus of ς³² is not an essential degradation signal, we considered that perhaps N-terminal sequences play such a role. However, N-terminal sequences of ς³² are not amenable to mutational analysis, since the efficiency of translation of rpoH mRNA drops dramatically upon mutational alteration of the downstream box located at the 5′ end of the coding sequence (18).

We therefore restricted our efforts to an analysis of fragments of authentic ς³², N-His₆-ς³², and C-His₆-ς³² which we observed to accumulate during FtsH-dependent proteolysis in vitro. Although these fragments may be dead-end products, as they did not chase into smaller peptides, they may provide a tool for dissecting the degradation process. The immunologically detectable fragments, ranging from approximately 10 to 20 kDa, were identical for authentic ς³² and N-His₆-ς³², which indicates that the N-terminal His₆ extension of the N-His₆-ς³² protein is missing in these fragments (Fig. 9). In contrast, the C-His₆-ς³² fragments showed a similar pattern but had higher molecular weights than the ς³² and N-His₆-ς³² fragments, which indicates that the C-terminal extension of the C-His₆-ς³² protein was still attached. Together these findings suggest that the degradation of ς³² proceeds through cuts in the N-terminal segment of ς³², although they do not indicate whether the N terminus of ς³² is involved in the recognition by FtsH.

FIG. 9.

Determination of in vivo degradation intermediates of ς³². Cells of strain C600 which harbor plasmids expressing IPTG-regulated rpoH alleles encoding N-His-ς³², C-His-ς³², or authentic ς³² were grown at 30°C in LB medium. Aliquots of exponential-phase cultures (OD₆₀₀, 0.5 to 0.7) were collected, and equal amounts of total protein were analyzed by SDS-PAGE. The degradation intermediates were determined by immunoblotting using ς³²-specific antisera.

DISCUSSION

The aim of this study was to investigate the role of the C terminus of ς³² in the degradation by FtsH. C termini frequently constitute the degradation determinants of naturally unstable proteins, which target them to proteolysis by AAA proteases (9, 21). Such a role has been also proposed for the C terminus of ς³² on the basis of an analysis of C-terminal variants of ς³² (3).

We show here that a series of C-terminally truncated ς³² proteins are as unstable as wild-type ς³² in vivo, in cell extracts supplemented with FtsH and ATP, and in a purified ATP-dependent degradation system with FtsH and ς³² as the sole protein components. Furthermore, we show that the C-terminal truncations do not compromise the ability of ς³² to associate with DnaK in vitro. The C-terminal truncations were chosen such that the local predicted secondary structures are considered and that the novel C termini do not constitute predicted DnaK binding sites (16, 19) and do not resemble the consensus sequence of protease targeting sites (9). Taken together, our results do not provide any evidence for an essential role of C-terminal sequences of ς³² in the FtsH-dependent degradation process or in chaperone binding. Our findings agree well with the results of a recent study of truncated ς³² proteins and hybrids of stable Bradyrhizobium japonicum and unstable E. coli proteins (2). That study mapped a region of 85 residues, located between residues 36 to 122 of ς³², as being responsible for degradation of ς³².

We do not know the reason for the discrepancy with a previous study claiming an essential role for the C terminus of ς³² in degradation (3). It is possible that the additional six residues encoded by the vector which were fused to the C-terminally truncated ς³² proteins in the former study generated fusion proteins which became stabilized for unknown reasons. Our results are, however, in agreement with the findings that (i) class I fusions between N-terminal segments of ς³² and β-galactosidase exhibit normal shutoff of the heat shock response and ς³² instability (17) and (ii) blocking the N terminus, but not the C terminus, of ς³² by fusion with green fluorescent protein resulted in stabilization of the fusion protein, indicating the importance of sequences at or near the N terminus of ς³² in its degradation by FtsH (T. Tatsuta and T. Ogura, unpublished results). At this point we cannot exclude an involvement of internal sequences of ς³² in degradation. Furthermore, our conclusion is supported by our finding that fragments of ς³² proteins which were generated by FtsH-dependent degradation in vitro lack N-terminal segments but not C-terminal segments of ς³². Although this finding should still be interpreted with caution, it suggests important roles for N-terminal segments of ς³² in the degradation process. Further genetic and biochemical dissection is required to identify such a role.

Our analysis provided evidence for a role of the C-terminal sequences of ς³² in association with RNAP. Only the mutant protein lacking five C-terminal residues remained partially proficient in binding to RNAP and in in vivo activity, whereas all mutant proteins with longer truncations showed no significant RNAP binding or in vivo activity. The molecular basis for this role of the C terminus of ς³² in the association with RNAP is unclear. However, on the basis of several criteria (solubility and wild type-like partial proteolysis pattern), the C-terminally truncated ς³² proteins are not perturbed in their overall structure. Thus, either local conformational changes induced by the truncations or the lack of C-terminal residues directly involved in the association process may account for the observed loss in affinity. In this respect it is interesting that many regions within the polypeptide chain of ς³², including the regulatory region C and the conserved regions 2.1, 2.2, 3, and 4, have been implicated in the binding of RNAP (1, 3, 11, 20, 31). Structural analysis seems essential to solve this obvious complexity of the interaction between ς³² and RNAP.