The ClpP Peptidase Is the Major Determinant of Bulk Protein Turnover in Bacillus subtilis

Abstract

Measurements of overall protein degradation rates in wild-type and clpP mutant Bacillus subtilis cells revealed that stress- or starvation-induced bulk protein turnover depends virtually exclusively on the ClpP peptidase. ClpP is also essential for intracellular protein quality control, and in its absence newly synthesized proteins were highly prone to aggregation even at 37°C. Proteomic comparisons between the wild type and a ΔclpP mutant showed that the absence of ClpP leads to severe perturbations of “normal” physiology, complicating the detection of ClpP substrates. A pulse-chase two-dimensional gel approach was therefore used to compare wild-type and clpP mutant cultures that had been radiolabeled in mid-exponential phase, by quantifying changes in relative spot intensities with time. The results showed that overall proteolysis is biased toward proteins with vegetative functions which are no longer required (or are required at lower levels) in the nongrowing state. The identified substrate candidates for ClpP-dependent degradation include metabolic enzymes and aminoacyl-tRNA synthetases. Some substrate candidates catalyze the first committed step of certain biosynthetic pathways. Our data suggest that ClpP-dependent proteolysis spans a broad physiological spectrum, with regulatory processing of key metabolic components and regulatory proteins on the one side and general bulk protein breakdown at the transition from growing to nongrowing phases on the other.

Adaptation to an ever-changing environment has been the fundament of evolutionary success for virtually all species. The capability of encountering times of nutrient deprivation or physical stress in a manner that ensures survival is particularly crucial for microorganisms, whose large surface-to-volume ratio renders them widely susceptible to environmental alterations. Consequently, considerable effort has been, and continues to be, spent to elucidate the complex adaptational networks that have evolved to meet the demands of successfully coping with a multitude of possible limitations. The primary focus of these efforts is to discover typical changes in transcriptional and/or translational output in response to the environmental factor in question.

An instructive example of this approach is the dissection of the heat stress stimulon of the soil-dwelling gram-positive model organism Bacillus subtilis. On the basis of globally acting transcriptional regulators, five classes of heat-inducible proteins have been distinguished so far. Class I is controlled by the HrcA repressor (30, 35), class II by the alternative RNA polymerase sigma factor SigmaB (3), class III by the CtsR repressor (6), and class V by the CssR response regulator (5). Class IV comprises all those heat-inducible genes whose expression has not yet been linked to a specific regulator. Not surprisingly, chaperones (such as the GroE and the DnaK machineries) and protease components (several HSP100-like Clp ATPases and the ClpP peptidase) that counteract the negative effects of protein denaturation, either by assisting in refolding or by the degradation of irreversibly damaged proteins, form an essential part of the heat stress stimulon (31).

In addition to specific changes in transcriptional and translational efficiency, protein breakdown may also represent a strategy for achieving an appropriate reallocation of cellular resources. This concept forms the background of our interest in the physiological relevance of the Clp protease and prompted us in the present study to assess its importance for overall protein degradation in B. subtilis and to search for novel substrates of Clp-dependent proteolysis by means of two-dimensional polyacrylamide gel electrophoresis (2D-PAGE). Sequence data for an increasing number of genomes have revealed Clp proteins, originally isolated as an ATP-dependent caseinolytic protease in Escherichia coli (15), to be highly conserved and widely distributed throughout the eubacterial and eukaryotic domains (12, 20). Functional Clp proteases have a bipartite molecular architecture. A central proteolytic chamber is formed by two heptameric rings of the ClpP peptidase. Proteolytic processing of substrate polypeptides can occur only after they have been unfolded and translocated into the central cavity by an attached Clp ATPase. Clp ATPases form hexameric rings that can dock to either one or both apical sides of the ClpP double ring, giving rise to a symmetry mismatch (11, 13, 16). Clp ATPases may also alter the conformation of proteins without subsequent translocation to the ClpP tetradecamer, and such a genuine chaperone function has been reported for individual substrates as well as for the entirety of heat-aggregated abnormal proteins (21, 33).

Much work has been aimed at unraveling the molecular interactions and their underlying determinants both in terms of the catalytic mechanism of the Clp protease and in terms of the recognition of substrates (reviewed in references 9-11, 14, and 34). A considerable number of Clp targets and target candidates have been discovered in different organisms, but the parameters that actually determine recognition and subsequent proteolysis are poorly understood. This study underpins the high importance of Clp-dependent proteolysis by demonstrating that in B. subtilis, general protein turnover depends virtually exclusively on the ClpP peptidase. Moreover, we propose novel Clp substrate candidates, extending the understanding of the physiological role of Clp-mediated protein degradation.

MATERIALS AND METHODS

Pulse-chase radiolabeling.

Wild-type B. subtilis 168 (1) and B. subtilis 168 ΔclpP (QB4916) (22) cells were grown at 37°C in Belitsky minimal medium which had been supplemented with 0.01% (wt/vol) yeast extract to allow growth and labeling of the clpP deletion mutant. Overnight cultures were diluted to an optical density at 500 nm of approximately 0.05, and when exponentially growing cultures reached an optical density at 500 nm of ∼0.4, 6.75 μCi of l-[³⁵S]methionine/ml was added. Ten minutes after addition of the label, further incorporation of radioactive methionine was stopped by addition of 4.7 mM nonradioactive methionine (an approximately 200,000-fold excess). Immediately after the stopping, one parallel culture was transferred to 48°C, and the time zero (t₀) sample was taken. Cultures were then subjected to the indicated temperature regimen, and further samples of constant volume were drawn at 30, 60, 120, 240, 360, and 480 min and at approximately 20 h after the chase. Cells were centrifuged (4°C, 6,800 × g, 10 min), washed twice with TE buffer (10 mM Tris, 1 mM EDTA [pH 7.5]), resuspended in 300 μl of disruption buffer (TE with 1.4 mM phenylmethylsulfonyl fluoride), and disrupted by intermittent sonication (three to four cycles at 55 W for 1 min, with breaks of 1 min).

To measure the incorporation of radiolabel into protein in the crude extracts, two aliquots (5 μl) were pipetted onto filter paper disks and subjected to trichloroacetic acid (TCA) precipitation. The remainder of the crude extracts were then centrifuged (4°C, 17,900 × g, 30 min) to sediment cell debris and aggregated material. Two aliquots (5 μl) of the supernatant (i.e., the soluble fraction) were TCA precipitated. The insoluble pellet was washed twice with TE, resuspended in 100 μl of sodium dodecyl sulfate boiling buffer (1% sodium dodecyl sulfate, 0.375 M Tris HCl [pH 8.8], 50 mM dithiothreitol, 25% glycerol), and boiled for 10 min at 95°C. As with the crude extract and the soluble fraction, two aliquots (5 μl) of this pellet fraction were also subjected to precipitation with 10% TCA on ice. After two washes with 5% TCA and one with 96% ethanol, the amount of radioactivity in the duplicate sets of dried filter disks was then measured as counts per minute on a Packard Tricarb 2900 TR liquid scintillation counter.

2D-PAGE.

To obtain nonlabeled protein extracts, samples from the respective bacterial cultures were processed as described in the preceding section. The concentration of soluble proteins was measured by the Roti-Nanoquant protein assay. For gels to be stained with colloidal Coomassie solution (see Fig. 2), 400 μg of soluble protein was separated, while for radiolabeled gels, 100 μg was used. The appropriate protein extract volume was transferred to a reaction tube, vacuum dried, and then resuspended in 380 μl of rehydration solution {8 M urea, 2 M thiourea, 1% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate [CHAPS], 0.5% Pharmalyte [pH 3 to 10]} with vigorous shaking. The samples were briefly spun at room temperature (17,900 × g, 2 min) prior to loading of the supernatant onto ready-made IPG strips (pI range, 4 to 7) according to the manufacturer's recommendations (Amersham Biosciences). After completion of the first dimension, the second-dimension separation was performed as described previously (4).

FIG. 2.

(A) Growth curves of wild-type and ΔclpP cultures in Belitsky minimal medium from which samples were taken for 2D-PAGE 100 min after the cultures had been shifted from 37 to 48°C in mid-exponential phase. (B) Dual-channel comparison of the colloidal Coomassie solution-stained 2D gels of the resulting wild-type (blue) and clpP mutant (orange) extracts. The labeled spots were identified by MALDI-TOF peptide mass fingerprinting and are listed in Table 1.

For colloidal Coomassie staining, gels were fixed (40% ethanol, 10% acetic acid) for 2 h, washed twice with distilled water (for 10 min each time), and imbued with colloidal Coomassie solution overnight with slow shaking. The solution was prepared directly before use; for each gel to be stained, 4 ml of a Coomassie brilliant blue stock solution (5%, wt/vol), 20 g of ammonium sulfate, and 2.4 ml of 85% phosphoric acid were dissolved in distilled water to a final volume of 200 ml before 50 ml of methanol was added. After staining, the gels were washed with distilled water (twice for 10 min each time, once for 20 min) and scanned with a Hewlett-Packard Scan Jet 6000 at a resolution of 200 dpi.

The gels that had been loaded with radioactive proteins were fixed (50% ethanol, 12% acetic acid) for 2 h and subsequently washed with distilled water (three times for 20 min each time) and with 2% (vol/vol) glycerol (20 min). The dried gels were exposed to storage phosphor screens (Molecular Dynamics) and scanned with a PhosphorImager SI (Molecular Dynamics) at a resolution of 200 μm.

MALDI-TOF peptide mass fingerprinting.

Protein spots were excised from colloidal Coomassie-stained 2D gels by using a spot cutter (Proteome Works; Bio-Rad) with a 2-mm-diameter picker head. Cut spots were transferred to 96-well microtiter plates. Tryptic digestion with subsequent spotting on a matrix-assisted laser desorption ionization (MALDI) target plate was conducted automatically with an Ettan Spot Handling workstation (Amersham Biosciences) by using the following protocol. The gel pieces were washed twice with 100 μl of a solution of 50% acetonitrile (CH₃CN) and 50% 50 mM ammonium bicarbonate (NH₄HCO₃) for 30 min and once with 100 μl of 75% acetonitrile for 10 min. After gel pieces were dried at 37°C for 17 min, 10 μl of a trypsin solution containing 20 ng of trypsin (Promega)/μl was added and incubated at 37°C for 120 min. For fragment extraction, gel pieces were covered with 60 μl of 0.1% trifluoroacetic acid (TFA) in 50% acetonitrile and incubated at 40°C for 30 min. The peptide-containing supernatant was transferred to a new microtiter plate, and the extraction was repeated with 40 μl of the same solution. The supernatants were completely dried at 40°C for 220 min. The residue was dissolved in 2.2 μl of 0.5% TFA in 50% acetonitrile, and 0.4 μl of this solution was directly spotted onto the MALDI target plate. Then 0.4 μl of a saturated α-cyano-4-hydroxycinnamic acid solution in 70% acetonitrile was added and mingled with the sample by aspirating the mixture five times. The samples were allowed to dry on the target plate for 10 to 15 min before they were subjected to MALDI-time-of-flight (TOF) analysis.

MALDI-TOF measurements were carried out with a Proteome Analyzer 4700 (Applied Biosystems). The spectra were recorded in a mass range from 900 to 3,700 Da with a focus mass of 2,000 Da. For 1 main spectrum, 25 subspectra with 100 shots per subspectrum were accumulated by using a random search pattern. If the autolytic fragment of trypsin with the monoisotopic m/z ratio of 2211.104 reached a signal-to-noise ratio of at least 10, an internal calibration was automatically performed as one-point calibration using this peak. The standard mass deviation was less than 0.15 Da. If the automatic mode failed (in less than 1% of cases), calibration was accomplished manually. After calibration, peak lists were created for a signal-to-noise ratio of at least 6 by using the “peak to mascot” script of the 4700 Explorer software. Sample spot identification was then completed by submitting the peak lists to the Mascot search engine (Matrix Science) with a specific B. subtilis sequence database.

Gel comparison and quantitation.

In order to properly compare the spot patterns of different gels, Delta2D software (version 3.1.2; Decodon) was used to warp gel images, produce overlay pictures (as in Fig. 2B), and quantify spot intensities. The analyzed pulse-chase gel series consisted of gels loaded with proteins from the soluble fractions of the experiments visualized in Fig. 1A and B (neglecting the overnight samples), i.e., a “wild-type 37°C” and a “ΔclpP 37°C” series. Quantitation data sets for each of the two pulse-chase gel series were generated by warping the consecutive gels of one series to either the wild-type or the ΔclpP 37°C t₀ gel, resulting in two warped match sets. Prior to this step, the t₀ gels had been thoroughly compared with a comprehensive B. subtilis master gel (4), so that names of previously identified proteins could be assigned to the majority of spots. For each of the match sets, the software created an artifical fused union gel in which the outlines of all congruent, colocalized spots were merged into single outlines. These were then transferred to the constituent gels and served as a basis for the subsequent quantitation of relative spot intensities. Quantitation data were exported to Excel software (Microsoft), in which the sets of spot intensity values were normalized to the t₀ value and used to calculate linear regression curves. On the basis of positive or negative slopes of the regression curves, every spot could be considered either stabilized or destabilized in the wild-type and ΔclpP gel series.

FIG. 1.

Measurements of overall protein degradation in the wild type and the clpP mutant. Cells were radiolabeled for 10 min with l-[³⁵S]methionine during exponential growth at 37°C. A ∼200,000-fold excess of nonradioactive methionine was added to suppress further incorporation of radioactivity. Directly after this labeling stop, bacterial cultures were subjected to different temperature regimens: they were either kept at 37°C (A and B), shifted to 48°C (C and D), shifted to 54°C (E and F), or shifted to 54°C for 60 min and then downshifted again to 37°C (G and H). Samples of constant volume were drawn at successive time points according to a chase scheme of 0, 30, 60, 120, 240, 360, and 480 min and a last overnight sample (after approximately 20 h of chase). After disruption of cells, radioactivity was measured before (total protein) (open columns) and after centrifugation in the supernatant (soluble fraction) (solid columns) and the pellet (pellet fraction) (shaded columns) by scintillation counting.

RESULTS

Global protein degradation starts at entry into stationary phase and under severe heat stress.

A pulse-chase approach was used to assess the in vivo stability of l-[³⁵S]methionine-labeled proteins over a long time span and with different temperature regimens in the wild-type and clpP mutant strains. Taking into account the reported formation of protein aggregates during heat stress (18), incorporation of radioactivity was tracked not only in the soluble protein fraction but also in the insoluble fraction that pelleted during centrifugation of crude cell extracts (Fig. 1). The hypothesis that the cells may undergo fundamental processes of adaptation and redistribution of their metabolic resources during growth that would be detectable only in stationary phase impelled us to use an extended chase time of approximately 20 h. The persistence of radiolabel during exponential growth of the wild-type strain at 37°C (Fig. 1A) or 48°C (Fig. 1C) indicated that proteins were largely stable at both temperatures. Upon entry into stationary phase, however, this stabilization was no longer evident, and overall protein degradation began to occur at a relatively high rate. This rate was slightly higher at 48°C than at 37°C (Fig. 1A and C) and implied an average half-life of ∼8 h; the amount of the radioactivity initially incorporated was reduced to 36% after 8 h and to 17% after 22 h of chase. Shifting wild-type cells to 54°C, a temperature that is very near the upper limit for growth of B. subtilis, resulted in different degradation kinetics with respect to the onset of proteolysis (Fig. 1E). Under these harsh conditions, cell growth stopped instantaneously. Concomitantly, protein breakdown was triggered without delay and proceeded at a rather constant rate, with 45% of the initial radioactivity remaining after 8 h of chase.

Global protein degradation is markedly diminished in the absence of ClpP.

A similar analysis of the clpP deletion mutant yielded completely contrary results. Figures 1B, D, and F reveal a significant overall stabilization of proteins during the chase. Only the last time points for samples at 37 and 48°C (∼21 h of chase) contained considerably smaller amounts of radiolabeled proteins. At 54°C (Fig. 1F), both the optical density of the culture and the extent of radioactivity incorporated remained constant in the clpP mutant.

Deletion of clpP renders bulk protein prone to aggregation.

Strikingly, the proportion of radiolabeled proteins that failed to attain their native conformation in the soluble fraction and ended up as aggregates in the pellet fraction was much higher in the clpP mutant than in the wild type. Even the samples from the first time points, which were all drawn at 37°C directly before any temperature shift, showed a high tendency of newly synthesized proteins to aggregate in the clpP mutant. Approximately 20% (Fig. 1B and D) to 30% (Fig. 1F) of the radioactivity incorporated was promptly routed to the pellet fraction in these cultures. This proportion increased even further at later time points, and the effect was most pronounced under the extreme heat stress of 54°C, where the ratio of soluble to pellet fraction radioactivity gradually was completely reversed (Fig. 1F). In clpP mutant samples, only approximately 25% of the radioactivity measured was detected in soluble proteins at the latest time points; the remaining 75% was found in the pellet fraction. This tendency for proteins to aggregate could also be observed in the wild type, but here it appeared to be linked to the degree and duration of heat stress. At 37°C, virtually all incorporated radioactivity remained soluble until mid-stationary phase, when low levels (5 to 10%) of pellet-borne radioactivity became detectable (Fig. 1A). This proportion eventually increased to 40% in the 20-h sample. Subjecting the cells to 48°C accelerated this process but did not lead to a predominance of the pellet fraction (Fig. 1C). However, when cells were shifted to 54°C, the pellet fraction already accounted for ∼20% of the total radioactivity at the first time point, 30 min after the upshift. The high proportion of aggregated proteins increased further in samples from later time points.

ClpP is crucial for disaggregation.

These results support the idea that aggregate formation is a constant challenge to protein integrity under severe and prolonged heat stress or in the absence of ClpP. Against this background, we tested if the wild type and the clpP mutant were able to disaggregate previously formed protein aggregates by upshifting cultures to 54°C for 1 h before returning them to 37°C (Fig. 1G and H). While the wild-type strain was capable of reducing the pellet fraction formed during the 60 min of heat stress, the clpP mutant possessed this ability only to a limited extent (Fig. 1G and H). The upshift-downshift procedure had no apparent effect on the overall degradation rates of the strains. Degradation in the wild type was induced by the drastic temperature shift and then continued irrespective of the downshift to 37°C.

2D-PAGE exposes indirect effects of clpP deletion.

Given the fundamental role that the ClpP peptidase plays as the major component of intracellular protein turnover in B. subtilis, we endeavored to find novel ClpP substrates by comparing 2D gel spot patterns of extracts from the wild type and from the clpP mutant. Wild-type and clpP mutant cells were shifted from 37 to 48°C in mid-exponential phase and were harvested 100 min after the upshift in early-stationary phase (Fig. 2A). Comparison of the resulting 2D gels using the dual-channel imaging technique (2) revealed many differences (Fig. 2B). On the one hand, a large group of proteins showed up as being downregulated in the clpP mutant, while on the other hand many spots were also present in larger amounts in this strain. Potential ClpP substrates should belong to the latter group, and we therefore tried to identify those protein spots by MALDI-TOF peptide mass fingerprinting. The results are summarized in Table 1, where the 62 proteins identified are grouped according to function. The list does indeed contain bona fide ClpP substrates, including MecA (32) and MurAA (17). Degradation of ClpE and ClpX has also been suggested to be dependent on ClpP (8). However, the presence of Clp ATPases (in parallel experiments with extracts from stationary-phase cell cultures, the GroE and DnaK chaperones were also more abundant in the clpP mutant [data not shown]), 17 proteins known to be sigma B regulated (28), and 21 Spx-affected proteins (23) points at the major shortcoming of this approach: indirect and compensatory effects caused by the absence of ClpP may account for many of the observed differences in the spot patterns.

TABLE 1.

List of proteins that accumulate in the clpP mutant relative to the wild type^a

Name	Description	Categoryb	Comments
MurAA	UDP-N-acetylglucosamine 1-carboxyvinyltransferase	1.1	ClpP substrate (17)
MecA	Negative regulator of competence	1.10	ClpP substrate (32); Spx-dependently inducedc
NfrA	FMN-containing NADPH-linked nitro/flavin reductase	1.4	Spx-dependently induced
TrxA	Thioredoxin	1.4	Sigma B dependentd; Spx-dependently induced
TrxB	Thioredoxin reductase	1.4	Spx-dependently induced
YfmJ	Unknown; similar to quinone oxidoreductase	1.4
YqiG	Unknown; similar to NADH-dependent flavin oxidoreductase	1.4	Spx-dependently induced
YqjM	Unknown; similar to NADH-dependent flavin oxidoreductase	1.4	Spx-dependently induced
YvaB	Unknown; similar to NAD(P)H dehydrogenase (quinone)	1.4
YwrO	Unknown; similar to NAD(P)H oxidoreductase	1.4	Sigma B dependent; Spx-dependently induced
GidA	Glucose-inhibited division protein	1.7
RapC	Response regulator phosphatase	1.8
YhdN	Unknown; similar to aldo/keto reductase	2.1.1	Sigma B dependent
YhxB	Unknown; similar to phosphomannomutase	2.1.1	Sigma B dependent; Spx-dependently induced
YktC	Unknown; similar to inositol monophosphatase	2.1.1
YqkF	Unknown; similar to oxidoreductase	2.1.1
YugJ	Unknown; similar to NADH-dependent butanol dehydrogenase	2.1.1	Spx-dependently induced
PdhC	Pyruvate dehydrogenase (E2 subunit)	2.1.2	Cotranscribed with pdhD
PdhD	Pyruvate dehydrogenase (E3 subunit)	2.1.2	Cotranscribed with pdhC
Tkt	Transketolase	2.1.2
YqjI	Unknown; similar to 6-phosphogluconate dehydrogenase (pentose phosphate)	2.1.2
CysH	Phosphoadenosine phosphosulfate reductase	2.2
YjbG	Unknown; similar to oligoendopeptidase	2.2	Spx-dependently induced
YmfH	Unknown; similar to processing protease	2.2	Spx-dependently induced
YpwA	Unknown; similar to carboxypeptidase	2.2	Spx-dependently induced
YrhB	Unknown; similar to cystathionine gamma-synthase	2.2
Cmk	Cytidylate kinase	2.3
GuaC	GMP reductase	2.3
PurA	Adenylosuccinate synthetase	2.3
PurH	Phosphoribosylaminoimidazole carboxy formyl formyltransferase	2.3	Cotranscribed with purL, purM, purS
PurL	Phosphoribosylformylglycinamidine synthetase II	2.3	Cotranscribed with purH, purM, purS
PurM	Phosphoribosylaminoimidazole synthetase	2.3	Cotranscribed with purH, purL, purS
PurS	Required for phosphoribosylformylglycinamidine synthetase activity	2.3	Cotranscribed with purH, purL, purM
YdbM	Similar to butyryl-CoA dehydrogenase	2.4
YhfJ	Unknown; similar to lipoate-protein ligase	2.4	Cotranscribed with yhfK; Spx-dependently induced
YoxD	Unknown; similar to 3-oxoacyl-acyl carrier protein reductase	2.4	Spx-dependently induced
RecA	Involved in recombination and DNA repair (LexA autocleavage)	3.3
SigB	RNA polymerase general stress sigma factor	3.5.1	Sigma B dependent
AbrB	Transcriptional pleiotropic regulator of transition state genes	3.5.2
GreA	Transcription elongation factor	3.5.3	Spx-dependently induced
GatB	Glutamyl-tRNAGln amidotransferase (subunit B)	3.7.2
PpiB	Peptidyl-prolyl isomerase	3.8
ClpC	ATP-dependent Clp protease ATP-binding subunit	4.1	CtsR controllede (cotranscribed with mcsB); Sigma B dependent
ClpE	ATP-dependent Clp protease ATP-binding subunit	4.1	CtsR controlled
ClpX	ATP-dependent Clp protease ATP-binding subunit	4.1
GsiB	General stress protein	4.1	Sigma B dependent
GspA	General stress protein	4.1	Sigma B dependent
McsB	Modulator of CtsR repression	4.1	CtsR controlled (cotranscribed with clpC); sigma B dependent
YraA	Unknown; similar to general stress protein	4.1	Spx-dependently induced
KatE	Catalase 2	4.2	Sigma B dependent
MsrA	Peptidyl methionine sulfoxide reductase	4.2	Spx-dependently induced
SodA	Superoxide dismutase	4.2	Sigma B dependent; Spx-dependently induced
Tpx	Probable thiol peroxidase	4.2	Spx-dependently induced
YceH	Unknown; similar to toxic anion resistance protein	4.2	Sigma B dependent
YdbD	Unknown; similar to manganese-containing catalase	4.2	Sigma B dependent
YfkM	Unknown; similar to unknown proteins	5.2	Sigma B dependent
YhfK	Unknown; similar to unknown proteins	5.2	Cotranscribed with yhfJ; Spx-dependently induced
YvaA	Unknown; similar to unknown proteins	5.2	Sigma B dependent
YwfI	Unknown; similar to unknown proteins	5.2
YdaE	Unknown	6	Sigma B dependent
YflT	Unknown	6	Sigma B dependent
YuaE	Unknown	6	Spx-dependently induced

^aWild-type and clpP mutant cells were shifted from 37 to 48°C in mid-exponential phase and were harvested 100 min after the upshift in early-stationary phase (see Fig. 2A). Respective spots were excised from the ΔclpP gel of Fig. 2B and identified by MALDI-TOF peptide mass fingerprinting.

^bClassification and numbering of functional categories were adopted from the SubtiList database (http://genolist.pasteur.fr/SubtiList/).

^cSpx-dependent induction as reported in reference 23.

^dSigma B-dependent transcription as described in reference 28.

^eControlled by CtsR as reported in reference 6.

To exclude these indirect effects and to better differentiate between secondary effects of clpP deletion and true stabilization due to blocked proteolysis, we subjected the soluble fraction samples from Fig. 1A and B to 2D-PAGE. In these samples, potential ClpP substrates are labeled protein spots that decrease in intensity during the chase in the wild-type strain but maintain their levels in the clpP mutant.

2D-PAGE pulse-chase series reveal novel ClpP substrate candidates.

Quantitation regression curves were calculated for every detected spot from the wild-type and ΔclpP pulse-chase 2D gel series. It was then possible to pinpoint substrate candidates for Clp-dependent proteolysis and to group them according to the functional category to which they belong. Several proteins with “housekeeping” functions, such as enzymes of amino acid (GlmS, IlvA, IlvB, LeuA, LeuD, MetE, YjbG), carbohydrate (OdhA, Pgm, PycA), or nucleotide (PnpA, Xpt) metabolism and aminoacyl-tRNA synthetases (GlyS, IleS, MetS, ProS), appeared as favored putative ClpP substrates under the chosen conditions. Figure 3 shows graphical representations of the “exemplary” gel spot intensity data sets of GlmS and IlvB that particularly suggest that these two proteins are ClpP targets.

FIG. 3.

Spot volume intensity graphs of GlmS (top) and IlvB (bottom) with clearly decreasing values in the wild type (squares) and comparatively higher values in the ΔclpP mutant (circles).

DISCUSSION

Overall measurements of remaining radioactivity in radiolabeled proteins at different time points of growth suggest that general protein breakdown is carried out virtually exclusively by the ClpP peptidase (Fig. 1). Interestingly, the onset and rate of ClpP-mediated general proteolysis are not determined only by the mere amount of ClpP but also appear to be regulated by additional factors. When wild-type cells were shifted from 37 to 48°C, a temperature known to elicit the heat stress stimulon, the degradation kinetics, starting upon entry into stationary phase, were similar to those of a culture that had been kept at 37°C (Fig. 1). Apparently, under nonstress and mild heat stress conditions, bulk protein gets considerably degraded only once a transition phase-related signal has been generated, even if ClpP is present in greater amounts.

Entry into stationary phase is not the only condition under which proteolysis was induced, as can be seen from the immediate onset of degradation after an upshift from 37 to 54°C (Fig. 1E). At this high temperature, the extent of heat denaturation probably exceeded the capacity of the quality control mechanisms to hold cellular proteins in their native conformations.

In essence, our data imply two different “modes” of Clp-mediated proteolysis: on the one hand, severe heat stress with concomitant protein aggregation induces protein degradation, while on the other hand, entry into stationary phase without significant protein aggregate formation also yields the same result.

It was intriguing to note the high degree of protein aggregation in the clpP mutant even at the nonstress temperature (Fig. 1B). Without ClpP, the structural integrity of bulk protein is impaired and at the same time degradation is blocked, resulting in dead-end protein aggregates. Somehow the efficiency of (re)folding as performed by the DnaK and GroE chaperones or the solitary Clp ATPases seems to be linked to the presence of the ClpP peptidase. Recent in vitro data established a disaggregation capability of the B. subtilis ClpC ATPase that could readily be rerouted toward degradation when ClpP was added (29). Notably, the presence of the MecA adaptor protein was an indispensable prerequisite for both activities.

Against the background of extensive overall protein degradation in the wild type and stabilization in the clpP mutant, we set out to identify novel candidates for substrates of the ClpP peptidase by means of 2D-PAGE. Initial comparisons between the spot patterns of wild-type and ΔclpP extracts suggested that indirect effects are likely to account for the elevated (as well as for reduced) amounts of many proteins in the clpP mutant (Fig. 2 and Table 1). The most prominent putative indirect effects were the compensatory induction of Clp ATPases and class I chaperones, a continued activation of sigma B-dependent gene expression, and a similarly locked-on transcription of Spx-induced genes. Spx is an “anti-alpha” factor induced under thiol-specific oxidative stress (23) and degraded by ClpXP (24). Since the spx gene (formerly yjbD) is also sigma B-dependently transcribed (28), it is conceivable that there is a sequential triggering and amplification of the indirect effects. First, activation of sigma B would also entail induced transcription of spx. Second, the Spx protein cannot be degraded in the absence of ClpP (24), which in turn results in a severe perturbation of physiologically “normal” transcription levels.

To mask out the pleomorphic indirect effects of clpP disruption and to specifically address degradation candidates, the soluble fractions from pulse-chase experiments with the wild type and the ΔclpP mutant performed at 37°C were separated by 2D-PAGE, and the spots on the resulting autoradiographs were matched, quantified, and identified by thorough comparison with an amended comprehensive B. subtilis master gel (4). This strategy allowed the detection of proteins that are degraded at an above-average rate in the wild type while remaining stable in the clpP mutant. Except for YjbG, none of the proteins detected showed up as increased in the clpP mutant in the plain nonradioactive gel-to-gel comparison (Fig. 2 and Table 1); this finding hints at the high impact of indirect effects on the intracellular proteome in the absence of ClpP. The majority of the substrate candidates derived from the pulse-chase experiments have “housekeeping” functions and comprise proteins involved in the metabolism of amino acids and related molecules (GlmS, IlvA, IlvB, LeuA, LeuD, MetE, YjbG), proteins of carbohydrate metabolism (OdhA, Pgm, PycA), proteins involved in nucleotide metabolism (PnpA, Xpt), and aminoacyl-tRNA synthetases (GlyS, IleS, MetS, ProS). Currently, we are analyzing additional pulse-chase gel series of glucose-starved wild-type and clpP, clpC, clpE, and clpX mutant strains in order to assay the degradation of substrate candidates under defined limiting conditions and to determine the participating ATPase component(s).

Generally, this approach addresses only above-average degradation rates; the real impact of protein breakdown will certainly be much broader. At the same time, monitoring the autoradiographs allows no inference of the actual protein amounts; required proteins could have been resynthesized. Moreover, protein aggregation in the clpP mutant might simulate degradation (i.e., disappearance or reduction of spot intensities in gels of the soluble fraction).

Despite these caveats, our data do supply promising leads in the search for novel ClpP substrates of B. subtilis. Among others, the degradation patterns of spots identified with the GlmS and IlvB proteins suggest that they are ClpP-dependently degraded (Fig. 3). Interestingly, both GlmS (l-glutamine-d-fructose-6-phosphate amidotransferase) and IlvB (the large subunit of the acetolactate synthase) are degraded throughout the chase and are involved in the first committed steps of certain biosynthetic pathways: acetolactate synthase catalyzes the first reaction in the anabolism of the branched-chain amino acids isoleucine, valine, and leucine, whereas GlmS transfers an amino group to fructose-6-phosphate and thus constitutes the very beginning of hexosamine metabolism. Hexosamines, in the form of UDP-N-acetylglucosamine, are primarily channeled into peptidoglycan biosynthesis. Strikingly, the enzyme responsible for the first committed step of this pathway, UDP-N-acetylglucosamine 1-carboxyvinyltransferase (MurAA), is also a bona fide ClpP substrate (17). Although further work will be needed to conclusively establish that GlmS and IlvB are genuine targets of ClpP, our data suggest a regulatory role for ClpP in the proteolysis of metabolic enzymes, in addition to its reported role in the degradation of regulatory proteins such as ComK and its cognate adaptor MecA (32), CtsR (19), SpoIIAB (26), or Spx (25).

Likewise, it would be of prime interest to know more about the actual determinants of substrate recognition under different conditions. We were unable to detect specific amino acid sequence patterns among the 16 most promising substrate candidates that might serve as targeting signals. Conceivably, substrate recognition may rely on a multiplicity of adaptor proteins that have different recognition sites in their own right (7). In B. subtilis, only two adaptors for Clp-mediated proteolysis are known so far: MecA (32) and YpbH (25). Since their absence is not sufficient to block the in vivo degradation of SpoIIAB (27) or MurAA (17), it is very likely that as yet unidentified adaptors exist. The finding that elevated amounts of Clp proteins, as observable under mild heat stress conditions, were not sufficient for triggering bulk protein breakdown supports the basic idea that correct targeting could be ensured by a potentially significantly underestimated number of adaptor proteins.