Clp ATPases and ClpP proteolytic complexes regulate vital biological processes in low GC, Gram-positive bacteria

Authors


*E-mail hi@life.kudk; Tel. (+45) 3528 2773; Fax (+45) 3528 2755.

Summary

Clp proteolytic complexes consisting of a proteolytic core flanked by Clp ATPases are widely conserved in bacteria, and their biological roles have received considerable interest. In particular, mutants in the clp genes in the low-GC-content Gram-positive phyla Bacillales and Lactobacillales display a diverse range of phenotypic changes including general stress sensitivity, aberrant cell morphology, failure to initiate developmental programs, and for pathogens, severely attenuated virulence. Extensive research dedicated to unravelling the molecular mechanisms underlying these complex phenotypes has led to fascinating new insights that will be covered by this review. First, Clp ATPases and ClpP-containing proteolytic complexes play indispensable roles in cellular protein quality control systems by refolding or degrading damaged proteins in both stressed and non-stressed cells. Secondly, ClpP proteases and the chaperone activity of Clp ATPases are important for controlling stability and activity of central transcriptional regulators, thereby exerting tremendous impact on cell physiology. Targets include major stress regulators like Spx (oxidative stress), the antisigma factor RsiW (alkaline stress) and HdiR (DNA damage) in addition to regulators of developmental programs like ComK (competence development), σH and Sda (sporulation). Thus, Clp proteins are central in co-ordinating developmental decisions and stress response in low GC Gram-positive bacteria.

The bipartite ClpP proteolytic complexes and the Clp chaperones

The highly conserved ATP-dependent ClpP proteases are two-component proteases consisting of separately encoded ATPase and peptidase subunits (Fig. 1A). Extensive structural analysis of the ClpP proteases from Escherichia coli has formed the paradigm for the way we understand the function of these proteins (reviewed by Sauer et al., 2004). The central proteolytic core consists of 14 ClpP serine peptidase subunits stacked in two heptameric rings, forming an internal chamber in which the active sites are sequestered from the cytoplasm (Wang et al., 1997). Narrow axial pores that exclude native proteins and all but the smallest peptides control access to this proteolytic chamber. To gain proteolytic activity, the ClpP multimer associates with one or two hexameric rings of Clp ATPases, forming the ClpP-containing proteolytic complex (designated the ClpP protease). In association with the proteolytic core, the Clp ATPases are responsible for the recognition, unfolding and translocation of substrates into the ClpP degradation chamber (Zolkiewski, 2006). This overall structure and organization bears resemblance to the 26S proteasome of eukaryotic cells (Kessel et al., 1995).

Figure 1.

A. Structural organization of the ClpP proteolytic complex. Details are described in the text.
B. The ClpATPase subfamilies. The ClpATPases contain either one or two nucleotide binding domains (AAA-1, AAA-2) and the length of the spacing between these domains, as well as the presence of specific signature sequences (not indicated in the figure), form the basis for the subfamily classification ClpA, ClpB, ClpC, ClpE and ClpL (Schirmer et al., 1996; Ingmer et al., 1999; Porankiewicz et al., 1999). Functional domains include the P domain required for binding to ClpP (Kim et al., 2001), the Zn binding domain involved in dimerization (Wojtyra et al., 2003) and the N1 and N2 domains proposed to be involved in protein binding (Barnett et al., 2005). In addition, a domain (UVR) resembling the interaction domain between the nucleotide excision repair proteins, UvrB and UvrC was identified in several ClpATPases (Ingmer et al., 1999).

The Clp ATPases constitute a family of closely related proteins that carry one or two nucleotide binding domains typical of the AAA+ superfamily of ATPases (Schirmer et al., 1996; Neuwald et al., 1999). Clp ATPases are divided into subfamilies on the basis of the presence of specific signature sequences and the number and spacing of the nucleotide binding sites (Fig. 1B). A role in protein folding, activation, or disaggregation seems to be shared by all Clp ATPases (Zolkiewski, 2006). However, only a subgroup of the Clp ATPases can interact with ClpP, a property that is associated with the presence of a ClpP recognition tripeptide found in ClpA, ClpC and ClpE but not in ClpB and ClpL (Kim et al., 2001). The specific number and types of Clp ATPases vary even between closely related genera but as a general rule, ClpX and ClpC are present in all low GC Gram-positive bacteria while ClpA is absent. In Bacillus subtilis cells the number of Clp ATPases have been determined to 1400 ClpX hexamers, 250 ClpC hexamers and 100 ClpE hexamers in comparison to 1200 tetradecameric ClpP during exponential growth at 37°C. As each tetradecameric ClpP can bind two hexameric Clp ATPases these data suggest that the Clp ATPases do not compete for the proteolytic ClpP subunit (Gerth et al., 2004).

Importantly, recent studies revealed major functional differences between the ClpC ATPase from B. subtilis and the Clp ATPases described in E. coli. For example, ClpC, in contrast to other characterized Clp ATPases, has low intrinsic ATPase activity and depends on cofactors to gain chaperone activity (Schlothauer et al., 2003; Kirstein et al., 2006). MecA, the best-characterized ClpC cofactor, facilitates ClpC oligomerization into a complex containing both ClpC and MecA (Kirstein et al., 2006). Formation of this complex is a prerequisite for all ClpC mediated activities, including ATPase activity, substrate recognition and interaction with ClpP (Kirstein et al., 2006). In E. coli adaptor proteins are required only to modulate substrate recognition, not for basic functions of their ClpATPase partner (see review by Dougan et al., 2002). Based on this important difference we propose to use the term cofactor instead of adaptor protein for MecA and related proteins.

Regulation of ClpP and the Clp ATPases

Production of ClpP and most Clp ATPases is strongly increased in response to heat-shock and other stress conditions. In low GC Gram-positive bacteria, transcription control of the clp genes is primarily mediated by binding of the CtsR repressor to a heptanucleotide repeat A/GGTCAAA/T located in the σA promoter region of target genes (Derréet al., 1999; Nair et al., 2000). However, in some organisms the HrcA repressor as well as the alternative sigma factor, σB are also involved as summarized in Fig. 2 (Krüger and Hecker, 1998; Derréet al., 1999; 2000; Varmanen et al., 2000; Chastanet et al., 2001; Chastanet and Msadek, 2003; Grandvalet et al., 2005).

Figure 2.

Regulation of clp gene expression by CtsR, HrcA and σB. In low GC firmicutes the composition of CtsR (boxed) and HrcA (circle) and overlapping CtsR and HrcA (dotted) regulons show considerable variation. In Bacilli the CtsR and HrcA regulons are distinct, whereas the HrcA regulon is embedded within the CtsR regulon in staphylococci. Alternative sigma factor σB plays a minor role (dashed line, B) in expression of clpP and clpC in B. subtilis and L. monocytogenes, whereas clpL of S. aureus is solely transcribed by σB. In Lactobacillales the CtsR and HrcA regulons coexist and partially overlap in several members of Streptococcaceae. The clp gene regulation in the other members of this order are predominantly controlled by either CtsR or HrcA. (Model adapted from Chastanet and Msadek, 2003.)

The central question of how the repressor activity of CtsR is relieved in response to heat shock has been addressed in B. subtilis, where detailed in vitro studies have demonstrated that DNA binding by CtsR is modulated by a complex titrating mechanism involving McsB, McsA and ClpC, all encoded by the ctsR operon (Krüger et al., 2001; Kirstein et al., 2005). According to the present model illustrated in Fig. 3, McsA, McsB and ClpC form a ternary complex in which the kinase activity of McsB is inhibited during non-stress conditions (Kirstein and Turgay, 2005; Kirstein et al., 2005). Upon protein denaturation, a hallmark of most stress conditions and notably of heat stress, ClpC is titrated away from the ternary complex (Kirstein and Turgay, 2005). In the absence of ClpC, the kinase activity of McsB is stimulated by McsA and the resulting phosphorylation of McsB enhances its affinity for CtsR, leading to the formation of a new ternary complex consisting of phosphorylated McsA, McsB and CtsR. The formation of this complex and the phosphorylation of CtsR result in sequestration of CtsR from its target promoters, leading to derepression of the CtsR regulon. Interestingly, the phosphorylation of CtsR also seems to be implicated in tagging it for degradation by ClpCP and ClpEP during stress conditions (Kirstein et al., 2005; Miethke et al., 2006). CtsR is part of the CtsR regulon and its synthesis is greatly stimulated in response to stress. Therefore, rapid inactivation of newly synthesized CtsR by phosphorylation and subsequent degradation is probably required to sustain derepression of the CtsR regulon until the problems with non-native proteins are overcome (Kirstein and Turgay, 2005). Notably, clpX appears not to be part of the CtsR regulon in any organism examined so far. Based on this observation and the lack of heat induction of ClpX we speculate that the primary cellular role of ClpX is not linked to handling stress-damaged proteins in low GC Gram-positive bacteria (Frees et al., 2003; Gerth et al., 2004).

Figure 3.

In response to stress the repressor activity of CtsR is inhibited by McsA/McsB-mediated phosphorylation and subsequent degradation by ClpCP. Details of the model are described in the text. Long arrow indicates high level gene expression, broken arrow indicates low level gene expression and P indicates phosphorylation.

Clp proteins are indispensable in protein quality control during both stress and non-stress conditions

The implication of Clp proteins in protein quality control in the stressed cell

Unfolding and subsequent aggregation of denatured proteins is the hallmark of heat shock and most other stress conditions and ClpP proteases clearly play an essential role in removing heat-damaged proteins from the low GC Gram-positive bacteria. This notion was first demonstrated indirectly by showing that bacterial mutants lacking ClpP were restricted for growth at high temperatures in a wide range of bacteria including B. subtilis, Lactococcus lactis, Listeria monocytogenes, Staphylococcus aureus and Streptococci (Msadek et al., 1998; Frees and Ingmer, 1999; Gaillot et al., 2000; Lemos and Burne, 2002; Robertson et al., 2002, Frees et al., 2003; Nair et al., 2003). Directly, it was observed that inactivation of ClpP decreased in vivo degradation of non-native model substrates (puromycyl-peptides) to less than 30% of the wild-type level in L. lactis, L. monocytogenes and in B. subtilis (Frees and Ingmer, 1999; Gaillot et al., 2000; Krüger et al., 2000). To recognize and degrade non-native proteins, ClpP must associate with a Clp ATPase partner and current knowledge supports that ClpC is the principal ClpATPase responsible for recruiting ClpP to degrade non-native proteins in B. subtilis and S. aureus (Krüger et al., 2000; Frees et al., 2004; Kock et al., 2004a). Similarly, ClpE appears to play a role in protein quality control during heat stress whereas the function of ClpX varies between organisms (Nair et al., 1999; Chastanet et al., 2001; Miethke et al., 2006). For example a B. subtilis clpX mutant is heat sensitive (Gerth et al., 1998) while deletion of clpX improved heat tolerance of S. aureus (Frees et al., 2003; 2004). ClpX is essential in lactococci and streptococci but the molecular basis behind this observation is unknown (Robertson et al., 2003; K. Savijoki, unpubl. results).

Finally, ClpB is important for growth at very high temperatures as well as for induced thermotolerance in L. monocytogenes and S. aureus (Chastanet et al., 2004; Frees et al., 2004). The absence of a ClpP interaction domain in ClpB suggests that ClpB primarily functions as a chaperone and that the unique role of ClpB in solubilizing protein aggregates described in E. coli is conserved in Gram-positive bacteria (Mogk et al., 1999). Interestingly, ClpL also lacks the ClpP interaction domain and the phenotypic resemblance between clpL and clpB mutants in S. aureus indicates that ClpL and ClpB may have a similar function (Frees et al., 2004).

Protein quality control in the non-stressed cell

Measurements of overall protein degradation rates in B. subtilis growing under non-stress conditions revealed that cellular proteins were largely stable during exponential growth but overall protein degradation occurred at a relatively high rate upon entry into stationary phase (Kock et al., 2004a). Importantly, this growth phase-dependent breakdown of cellular proteins did not occur in cells lacking ClpP, suggesting that ClpP is the major determinant of bulk protein turnover in B. subtilis even under non-stress conditions. The importance of ClpP for protein quality control in non-stressed cells was further emphasized by the finding that the absence of ClpP caused 20–30% of newly synthesized proteins to aggregate (Kock et al., 2004a). In accordance, electron-dense particles representing insoluble protein aggregates were observed in non-stressed Bacillus cells lacking ClpP activity (Krüger et al., 2000). An example of a conserved prokaryotic ClpP-dependent protein quality mechanism operating in the non-stressed cell is the cotranslationally tagging of proteins synthesized from damaged mRNAs by an 11 amino acid peptide (AANDENYALAA) that will destine the tagged proteins for degradation by the ClpP proteases (Keiler et al., 1996; Wiegert and Schumann, 2001). Hence, cotranslationally tagged proteins may represent one group of newly synthesized proteins that end up in aggregates in cells devoid of ClpP.

ClpP proteolytic complexes control specific stress responses by regulated proteolysis

Recent data emphasize that ClpP mediated proteolysis also plays a more specific role in cellular stress responses by controlling stability of key regulatory proteins. The degradation of CtsR during stress conditions represents an example of regulated proteolysis where we have a good understanding of how specific signals are transmitted into proteolysis of a regulatory protein. In this section three other highly interesting examples of regulated proteolysis will be described. The first is HdiR (heat and DNA damage-inducible regulator), recently identified in L. lactis as a transcriptional repressor that upon either DNA damage or heat shock, undergoes autocatalytic cleavage ( Fig. 4A, Savijoki et al., 2003). The N-terminal cleavage product of HdiR retains DNA binding activity, and expression of target genes is induced only when HdiR is degraded by the ClpXP protease (Savijoki et al., 2003; K. Savijoki, unpublished). Structurally and functionally, HdiR resembles the SOS response regulator LexA and as a lexA homologue is absent in streptococcus genomes, HdiR might perform a LexA-like function in these organisms.

Figure 4.

Examples of mechanisms underlying regulated proteolysis by ClpP proteases.
A. Degradation of the transcription factor HdiR by ClpXP in response to heat stress or DNA damage.
B. Regulated proteolysis of the antisigma factor RsiW by ClpXP/ClpEP in response to alkaline stress.
C. Regulated proteolysis of the competence regulator, ComK, by ClpCP.
D. Degradation of the antisigma factor SpoIIAB by ClpCP during sporulation. See text for details of the depicted models.

Another biologically important example, where initial cleavage uncovers a tag that targets the cleavage product for degradation by ClpXP, was recently described by Zellmeier et al. (2006). This study convincingly demonstrated that the B. subtilisextracytoplasmic alternative sigma factor (ECF sigma factor) σw is activated by stepwise degradation of the associated antisigma factor RsiW in response to alkaline stress (Fig. 4B). The first cleavage step reveals a C-terminal -LAA signature motif demonstrated to be required for recognition and degradation, mainly by ClpXP but also by ClpEP (Table 1). The degradation of RsiW releases σw and promotes induction of σw-dependent genes (Zellmeier et al., 2006). Interestingly, this mechanism for activating ECF σ factors might be widely conserved, as it is essentially identical to the mechanism described for activation of σE in E. coli, despite the fact that the antisigma factors are only distantly related on the sequence level (Zellmeier et al., 2006).

Table 1.  Motifs and cofactors directing ClpP proteolysis.
Target proteinClp degradation complexDegradation tagCofactor requirement for degradationOrganismReference
  1. ND, not determined.

SpoIIABClpCP-LCN at C-terminal. Stabilized by -EEN and -LCNEDoes not require MecAB. subtilisPan and Losick (2003)
SdaClpXP in vivo but not in vitro-VSS at C-terminal. Stabilized by VDDNDB. subtilisRuvolo et al. (2006)
SpxClpXP in vivo and in vitro-LAN at C-terminal. Stabilized by -LDDDoes not require MecA or YpbH in vivoB. subtilisNakano et al. (2002; 2003a,b)
SpxClpCP in vitroNDMecA or YpbHB. subtilisNakano et al. (2002)
MurAA NDDoes not require MecA or YpbH in vivoB. subtilisKock et al. (2004b)
SsrA tagged proteinsClpXP-AA at C-terminal. Stabilized by -DDNDB. subtilisWiegert and Schumann (2001)
ComKClpCPNDMecAB. subtilisTurgay et al. (1997); Turgay et al. (1998)
RsiWClpXP (ClpEP)-AA in C-terminal of processed protein Stabilized by -DDNDB. subtilisZellmeier et al. (2006)

Many phenotypes associated with mutations in clpP and clpX in B. subtilis can be alleviated by the concomitant inactivation of spx (suppressor of clpP and clpX) (Nakano et al., 2001). Spx is a global transcriptional regulator of oxidative stress in several Gram-positive bacteria, and it is a substrate of the ClpXP protease –Table 1 (Nakano et al., 2002; 2003a, Pamp et al., 2006). Spx interacts directly with the α subunit of the RNA polymerase (RNAP) to control global transcription initiation, either negatively or positively, by a unique mechanism not involving initial contact with DNA (Nakano et al., 2003a,b; 2005). In the absence of ClpXP, accumulating Spx binds to the RNAP α subunit thereby blocking interaction between the RNAP and transcriptional activators such as the response regulators ComA and ResD. As ComA controls genes required for developing genetic competence (the ability to internalize exogenous DNA) while ResD induces genes involved in adaptation to oxygen limitation, this explains why clpP and clpX mutants are defective in developing genetic competence and in growth under oxygen restricted conditions in B. subtilis (Nakano et al., 2003a; Zuber, 2004).

Homologues of Spx are highly conserved among low GC Gram-positive bacteria (Zuber, 2004). Interestingly, the spx homologue trmA was previously identified as the site of mutations that could alleviate the heat-sensitive phenotype of clpP and recA mutant in L. lactis (Duwat et al., 1999; Frees et al., 2001). L. lactis encodes 7 Spx paralogues and we hypothesize that these proteins represent RNAP binding global regulators whose activity is controlled by ClpP mediated proteolysis. The signals that govern ClpXP regulated proteolysis of Spx have not yet been identified.

Finally, a recently published DNA-microarray study comparing overall transcription in wild-type S. aureus and its clpP mutant revealed that ClpP influences expression of multiple central regulons, including the PerR, Fur, MntR and LexA regulons (Michel et al., 2006). Thus, the number of regulators whose activity is controlled by ClpP mediated proteolysis might be significantly underestimated at present.

Clp proteins in cellular differentiation

Clp proteins controlling developmental programs

In response to starvation for essential nutrients B. subtilis initiates several developmental programs including sporulation and competence development. The ClpXP mediated control of the cellular level of Spx represents an example of how ClpXP regulates development of competence, however, Clp protein control multiple key switches in these programs. Central to competence development is the transcriptional activator, ComK that is required for expression of genes encoding the DNA uptake apparatus. Importantly, ComK also positively regulates its own expression and therefore only small fluctuations in the level of ComK can initiate a positive feed-back loop leading to the development of competence as recently discussed by Dubnau and Losick (2006). Accordingly, production of ComK is carefully controlled at multiple levels and any ComK molecules that escape this tight control are via binding to MecA directed to ClpCP for degradation (Fig. 4C, Msadek et al., 1994; Kong and Dubnau, 1994; Turgay et al., 1997; 1998). The ComK pathway, and consequently competence, is induced when synthesis of ComS is initiated as a result of signalling by a quorum-sensing two component regulatory pathway. Subsequently, ComS displaces ComK from the MecA-ClpC ternary complex and the released ComK then activates transcription of its own gene, thereby initiating a burst of competence gene expression (Turgay et al., 1997).

In response to more severe starvation B. subtilis initiates sporulation. Timely and spatial expression of transcription factors is essential for successful sporulation and both chaperone and proteolytic activities of the Clp proteins are required at multiple levels. ClpX is required at the onset of sporulation to activate the alternative sigma factor σH (Spo0H) and once it has completed its tasks σH is removed in a process involving ClpC, most likely as part of a ClpCP complex (Nanamiya et al., 1998; Liu et al., 1999; Liu and Zuber, 2000). During sporulation, the activity of the sporulation sigma factor σF is confined to the forespore by ClpCP (Pan et al., 2001). At the transition from the vegetative cell to the forespore σF is activated by being released from the anti-sigma factor SpoIIAB in a process involving the anti-antisigma factor SpoIIAA (Fig. 4D). Dissociation of σF from SpoIIAB uncovers a ClpC recognition motif leading to degradation by ClpCP (Pan et al., 2001; Table 1).

Finally, ClpP affects the co-ordination of DNA replication and initiation of sporulation by degrading Sda, a small protein whose synthesis is induced in response to perturbations in chromosome replication or elongation. Sda specifically inhibits the histidine kinases required to initiate spore development, and recovery from this state requires the removal of Sda by the ClpXP protease (Table 1, Ruvolo et al., 2006).

Clp proteins affecting cell division and transition between growth phases

Inactivation of Clp proteins often leads to aberrant morphology such as filamentation, impaired cell septation and separation, abnormal cell wall structure and irregular cell division (Gerth et al., 1998; Msadek et al., 1998; Nair et al., 1999; 2003). Collectively, these data suggest that Clp ATPases and ClpP proteolytic complexes are required for basic biological processes fundamental to cell integrity. Recent reports have shed some light on the processes that might underlie these observations. As described above, massive ClpP-dependent degradation of proteins produced during exponential growth occurs in the transition to stationary phase in B. subtilis (Kock et al., 2004a). One of the ClpCP-degraded substrates is MurAA (Table 1), which catalyses carboxyvinyl transfer from phosphoenolpyruvate to UDP-N-acetyglucosamine, the first committed step in peptidoglycan biosynthesis (Kock et al., 2004b). Notably, cells overproducing MurAA display the same filamentous growth as clpP mutant cells, suggesting that this clpP phenotype is at least in part be due to failure to degrade MurAA (Kock et al., 2004b). Interestingly, an independent role of ClpX in Bacillus cell division was recently demonstrated by showing that ClpX functions as an inhibitor of the cell division protein FtsZ, possibly by hindering FtsZ polymerization that is required for septum formation by sequestering unassembled FtsZ protein (Weart et al., 2005).

Clp proteins affecting virulence of pathogens

Intriguingly, inactivation of ClpP or Clp ATPases significantly reduces virulence of the important pathogens, S. aureus, L. monocytogenes and Streptococcus pneumoniae (Gaillot et al., 2000; Frees et al., 2003; Ibrahim et al., 2005; reviewed by Butler et al., 2006).

The entry of bacterial pathogens into host organisms constitutes a dramatic environmental change and the apparently conserved role of Clp proteins in virulence could be a consequence of their involvement in handling stress imposed damage of bacterial proteins. As an example, S. aureus cells lacking the ClpB chaperone are unable to replicate intracellularly in bovine cells (Frees et al., 2004). As ClpB is needed for growth only under conditions generating massive protein aggregation, and does not affect production of any known virulence factors, this result supports the idea that the Clp protein quality control systems are required in the host. However, Clp chaperones and proteases also contribute to virulence by controlling synthesis of major virulence factors. For example, ClpC is required for transcription of inlA, inlB and actA encoding proteins important for host cell invasion of L. monocytogenes; while the absence of ClpP reduces activity, but not production of listeriolysin O, a major virulence factor implicated in phagosome lysis (Nair et al., 2000; Gaillot et al., 2000; 2001). Similarly, S. aureus ClpXP is required for transcription of a number of important virulence genes encoding haemolysins, extracellular proteases surface adhesins and others (Frees et al., 2003; 2005a; Michel et al., 2006). Interestingly, the regulatory activity of ClpXP seems to be tightly linked to the quorum sensing Agr two-component system, the best-characterized virulence regulator in this organism (reviewed by Novick, 2003). Induction of the agr operon in response to cell density leads to the production of the 514 nt small regulatory RNA, RNAIII, that is the actual effecter of virulence gene regulation. How RNAIII regulates the transcription of more than 50 genes remains puzzling but recently it was shown that RNAIII inhibits translation of rot encoding a transcriptional regulator of virulence genes (Geisinger et al., 2006). We have proposed that RNAIII may additionally bind directly to regulatory proteins thereby inducing a conformational change that destine the proteins for degradation by ClpXP (Frees et al., 2005a). Noteworthy, ClpX independent of ClpP is essential for transcription of spa, encoding the important Staphylococcal surface protein, Protein A (Frees et al., 2003; 2005a). At present we believe that ClpX is required either for expression, folding or dimerization of Rot that is a transcriptional activator of spa transcription (Frees et al., 2005a; Oscarsson et al., 2005).

The conserved role of Clp proteins in virulence may also be exploited therapeutically. Interestingly, S. aureus ClpP was recently shown to be the target of a new promising class of antibiotics, the acyldepsipeptides (Brotz-Oesterhelt et al., 2005). This compound apparently enables ClpP to degrade proteins in the absence of an associating Clp ATPase, leading to uncontrolled proteolysis that eventually kills the cells.

Concluding remarks and future perspectives

Bacteria can adapt to a wide range of growth conditions. Traditionally, studies have focused on changes in transcription, but it has become clear that regulated proteolysis represent an equally important aspect of growth adaptation. In E. coli four classes of ATP-dependent proteases [ClpAP/XP, ClpYQ (HslVU), Lon and FtsH] have overlapping substrate specificities, perhaps explaining why inactivation of individual proteases causes only modest phenotypic changes under the conditions tested (reviewed by Gottesman, 2003). In striking contrast the wide range of phenotypes conferred by inactivation of ClpP in the low GC Gram-positive bacteria suggest that the ClpP proteolytic proteases are the major proteases both for eliminating misfolded proteins and for controlling the activity of central regulatory proteins in these organisms. In streptococci homologues of Lon and ClpYQ are missing whereas in S. aureus ClpYQ is required for growth at very high temperature, suggesting that ClpYQ mediated proteolysis becomes essential only when heat-induced protein denaturation exceeds the capacity of the ClpP complex (Frees et al., 2005b). Thus, the ATP-dependent proteases appear to be dedicated to specialized tasks in Gram-positive bacteria. Additional important mechanistic differences to the E. coli paradigm was demonstrated by the finding that ClpC in contrast to other characterized ClpATPases has low intrinsic ATPase activity and depends on cofactors for all basic activities. One major task of future studies will be to understand the role of cofactors in relation to selection of substrates. A breakthrough in understanding the general rules governing substrate recognition by the ClpXP protease was recently obtained in E. coli, where specifically one large class of substrate proteins carry the C-terminal -LAA signature that is part of the SsrA tag (Flynn et al., 2003). Similar motifs have been shown to be required for ClpP-dependent proteolysis in Gram-positive bacteria suggesting that recognition of substrate proteins follow the general rules established in E. coli (Table 1).

A very interesting aspect of regulated proteolysis is the conditional degradation that occurs only in response to specific cellular signals. Examples of factors modulating degradation are shown in Figs 3 and 4. One challenge of the future will be to unravel the complex networks controlling regulated proteolysis. In the case of competence development proteolysis is regulated in response to environmental signals perceived and transmitted via two-component signal transduction systems. Alternatively, cellular processes such as DNA damage-repair may communicate with ClpATPases via the UVR domain (Fig. 1B), which resembles the UvrB–UvrC interaction domain. This communication event is functionally similar to the interaction that takes place between the DNA damage repair proteins and the 26S proteasome in eukaryotes (Sweder and Madura, 2002). Studies aimed at elucidating the mechanisms underlying regulated proteolysis, and specifically the ways external signals are communicated to the proteases, will be central for future work in this field.

Acknowledgements

We are thankful for the support of the Danish Research Council to D. Frees and for the support of the Academy of Finland to P. Varmanen (project no. 211165). We also gratefully acknowledge the helpful comments provided by M.T. Cohn, L. Jelsbak, L. E. Thomsen and M. Bojesen.

Ancillary