Proteolytic inactivation of key regulatory proteins is essential in eukaryotic cell‐cycle control. We have identified a protease in the eubacterium Caulobacter crescentus that is indispensable for viability and cell‐cycle progression, indicating that proteolysis is also involved in controlling the bacterial cell cycle. Mutants of Caulobacter that lack the ATP‐dependent serine protease ClpXP are arrested in the cell cycle before the initiation of chromosome replication and are blocked in the cell division process. ClpXP is composed of two types of polypeptides, the ClpX ATPase and the ClpP peptidase. Site‐directed mutagenesis of the catalytically active serine residue of ClpP confirmed that the proteolytic activity of ClpXP is essential. Analysis of mutants lacking ClpX or ClpP revealed that both proteins are required in vivo for the cell‐cycle‐dependent degradation of the regulatory protein CtrA. CtrA is a member of the response regulator family of two‐component signal transduction systems and controls multiple cell‐cycle processes in Caulobacter. In particular, CtrA negatively controls DNA replication and our findings suggest that specific degradation of the CtrA protein by the ClpXP protease contributes to G1‐to‐S transition in this organism.
The cell cycle is an ordered series of events involving sequential chromosomal duplication, and chromosomal segregation followed by division of the cell. In eukaryotes, a conserved class of regulatory proteins [cyclin‐dependent kinases (CDKs)] is responsible for the ordered progression of the cell through the periodic events of mitotic division (Nigg, 1995). Periodicity of the cell cycle also requires essential degradation processes (King et al., 1996). In particular, the G1‐to‐S transition (Hoyt, 1997) as well as onset of anaphase and exit from mitosis (Hoyt, 1997) are associated with proteolysis of key regulators like cyclins or CDK inhibitors and structural proteins. The existence of checkpoint control and similar feedback mechanisms in bacteria (i.e. SOS system) suggests common strategies in the cell‐cycle control of prokaryotes and eukaryotes (Murray, 1992; Walker, 1996). This has been underlined by recent findings obtained in a eubacterium, Caulobacter crescentus, which suggest that specific degradation events also play a key role in the bacterial cell cycle (Alley et al., 1993; Jenal and Shapiro, 1996; Domian et al., 1997).
During its biphasic life cycle, C.crescentus undergoes an asymmetric cell division that gives rise to progeny cells with different developmental programs (Figure 1). The sessile stalked cell (ST) is able to immediately initiate chromosome replication and subsequent cell division. The swarmer cell (SW) is motile and chemotactically active, but unable to replicate its chromosome and divide. To acquire its replicative capacity, the SW cell (G1 phase) must first differentiate into a ST cell (start of S phase), a process during which the flagellum is ejected and the chemotaxis apparatus is lost. Thus, cell differentiation is an integral part of the Caulobacter cell cycle.
As in most eukaryotic cells, the periods of replication and chromosome segregation do not overlap in C.crescentus. Chromosome replication is initiated once per cell cycle and is followed sequentially by a segregation period, septum formation and finally cell division. While little data is available on how these processes are controlled and integrated, recent studies have revealed several proteins involved in C.crescentus differentiation and cell‐cycle progression that are under spatial and temporal control involving both synthesis and proteolytic decay (Alley et al., 1993; Jenal and Shapiro, 1996; Quon et al., 1996; Wright et al., 1996; Domian et al., 1997).
One of these proteins, CtrA, is a member of the response regulator superfamily. CtrA modulates transcription of genes important for flagellar biogenesis, DNA methylation and cell division during the cell cycle (Quon et al., 1996; Wright et al., 1996; Kelly et al., 1998). Importantly, CtrA negatively controls initiation of chromosome replication in the SW cell (Quon et al., 1998). To restrict this inhibitory action to the motile SW cell, Caulobacter employs multiple mechanisms to control the activity and abundance of CtrA (Quon et al., 1996; Domian et al., 1997): (i) transcription of the ctrA gene occurs only in the PD cell; (ii) activation of CtrA by phosphorylation occurs in the early PD cell after initiation of DNA replication; and (iii) specific proteolysis clears CtrA from the ST compartment of the late PD cell and from the SW cell as it develops into a ST cell. Temporally and spatially appropriate degradation of CtrA is crucial for the G1‐to‐S transition, thereby coupling the cell cycle to chromosome replication.
Four families of ATP‐dependent proteases have been characterized in bacteria: Clp (ClpAP, ClpXP), HslUV (ClpYQ), AAA (FtsH) and the Lon family (Lon). The Clp proteins are assembled into structures similar to the eukaryotic proteasome (Yoshimura et al., 1993; Kessel et al., 1995, 1996; Rohrwild et al., 1996, 1997; Bochtler et al., 1997; Groll et al., 1997; Wang et al., 1997). A central cavity is formed by multimeric rings of either ClpP or ClpQ stacked on top of each other, with ATP‐dependent chaperones (ATPases; ClpA, ClpX, ClpY) regulating entry to the channel at both ends. The proteolytic activity resides in the ClpP or ClpQ subunits, while it is hypothesized that access to the channel depends on specific recognition and unfolding of a substrate protein by the ATPase. One of these ATPases, the Escherichia coli ClpX protein, interacts specifically with disordered C‐terminal peptides of its substrate proteins (Laachouch et al., 1996; Levchenko et al., 1997a,b). The C‐terminus of the Caulobacter CtrA protein is required for cell‐cycle‐dependent proteolysis (Domian et al., 1997) and shares sequence homology with the C‐terminal E.coli peptides that are specifically recognized by ClpX. We have recently isolated the genes encoding the proteolytic and regulatory subunits of the C.crescentus ClpXP protease (Figure 2A; M.Osteras, A.Stotz, S.Schmid Nuoffer and U.Jenal, manuscript in preparation). Here we show that ClpXP is responsible for the cell‐cycle‐dependent degradation of the CtrA protein. The ClpXP protease is strictly required for growth, survival and normal cell division of C.crescentus. Loss of either ClpP or ClpX results in cell‐cycle arrest. Thus, the role of Caulobacter ClpXP in the control of cell‐cycle progression in this prokaryote is similar to that of its structurally conserved eukaryotic counterpart, the proteasome.
The ClpXP protease is essential in Caulobacter
We have previously isolated the genes encoding the subunits of the C.crescentus ClpXP protease (Figure 2A). To analyze the function of the ClpXP protease in Caulobacter, we attempted to generate null mutations in clpP, the gene for the proteolytic subunit, and clpX, the gene for the regulatory subunit, by replacing the chromosomal genes with copies disrupted by an interposon carrying a spectinomycin‐resistance cassette (Figure 2A). The disrupted genes were introduced into the chromosome by homologous recombination on a plasmid that also contained a kanamycin resistance marker (nptI) and the sacB gene from Bacillus subtilis conferring sucrose sensitivity (Figure 2A, 1. rec.). Colonies that had lost the plasmid by a subsequent second recombination step (Figure 2A, 2. rec.) were isolated on the basis of their resistance to sucrose and analyzed for the presence or absence of the interposon in the clp locus. Approximately 600 sucrose‐resistant colonies were analyzed both for clpP and clpX but none of them contained a chromosomal null mutant (Figure 2B and C). Most of the colonies tested (512 out of 614 for clpP and 550 out of 600 for clpX; Figure 2) had reverted to the wild‐type genotype. Since this was a strong indication that both clp genes are essential for growth of C.crescentus, we asked if chromosomal mutations could be obtained in the presence of an additional, plasmid‐encoded, functional copy of clpP or clpX. Two constructs, pUJ148 and pUJ158, that contained either clpP alone or clpP with flanking regions, allowed the isolation of chromosomal clpP mutants when present in strain LS2037 (Figure 2B). Similarly, in the presence of plasmids pUJ145 or pUJ170, carrying either the entire clp region or clpX alone, chromosomal clpX mutants were readily obtained from strain UJ127 (Figure 2C). This suggested that both clpP and clpX are essential genes in C.crescentus and that the proteolytic activity of ClpXP is strictly required for growth or survival of Caulobacter cells.
To confirm that the proteolytic activity of ClpP is essential, we changed the codon for the active site serine residue of clpP (S107) (Maurizi et al., 1990) to an alanine codon and tested the ability of the clpPS107A allele to rescue a chromosomal clpP mutation. Out of 200 sucrose‐resistant colonies tested for strain LS2037/pTF33, none had a mutated clpP copy in the chromosome (Figure 2B). Since chromosomal clpP mutants could easily be isolated in the presence of plasmid pUJ158 (Figure 2B), which differs from construct pTF33 only in the clpP codon for Ser107, we conclude that the essential nature of ClpP is based on its proteolytic activity.
To study the phenotype of clpP and clpX mutants, strains were constructed with a single functional copy of either clpX or clpP under the control of the xylose‐dependent promoter, PxylX (Meisenzahl et al., 1997). Translational fusions between either clpP or clpX and the first six codons of xylX were created (pUJ174 and pUJ175) and introduced by homologous recombination into the chromosomal xylX locus of strains LS2037 and UJ127, respectively (see Materials and methods). A sucrose‐selection step (see above) was carried out with the resulting strains in order to isolate mutants with a disrupted clpP or clpX copy at the chromosomal clp locus and the corresponding wild‐type gene integrated at the xylX locus. Because these mutants should be strictly dependent on xylose for their growth, the sucrose selection was done on plates containing both sucrose and xylose. Since addition of glucose was shown to inhibit PxylX activity (Meisenzahl et al., 1997) we also performed a control experiment where the same selection was carried out on plates containing sucrose and glucose. Loss of the functional copy of the clpP or clpX gene at the clp locus was only possible when xylose was present in the selective medium. On medium containing glucose all sucrose‐resistant colonies either had the wild‐type clp locus restored by recombination or contained an inactivated sacB gene (data not shown).
ClpP and ClpX are required for normal cell division, growth and viability in C.crescentus
The growth of both strains isolated, UJ199 (NA1000 clpP::Ω xylX::clpP) and UJ200 (NA1000 clpX::Ω xylX::clpX), was strictly dependent on xylose (Figure 3A). The few colonies of strains UJ199 and UJ200 which were able to grow on PYE complex medium containing glucose (PYEG; Figure 3A) represent spontaneous mutations that either relax the xylose dependence of PxylX activity or suppress the strict requirement of the cells for ClpP and ClpX, respectively (U.Jenal, unpublished). Strains UJ199 and UJ200 are tetracycline‐resistant due to a resistance gene carried by the inserted constructs pUJ174 and pUJ175, respectively. In order to propagate plasmids with a tetracycline‐marker in these strains, we also generated identical clpP and clpX mutants that had a kanamycin‐resistance marker by exchanging plasmids pUJ174 and pUJ175 in strains UJ199 and UJ200 with plasmids pUJ167 and pUJ168 (see Materials and methods). The resulting strains UJ270 (clpP mutant) and UJ271 (clpX mutant) were also xylose‐dependent showing no growth on PYEG plates (data not shown). Introduction of plasmid pTF3 carrying a functional clpP gene abolished xylose dependence of strain UJ270; thus the interposon insertion in clpP had no polar effect on the expression of the downstream gene clpX (Figure 3A). Introduction of the clpPS107A allele on plasmid pTF33, did not stimulate growth of strain UJ270 on PYEG (Figure 3A).
To demonstrate the absence of ClpP and ClpX in the mutant strains UJ199 and UJ200 grown under restrictive conditions, immunoblots were carried out with antibodies (see Materials and methods) prepared against either ClpP or ClpX (Figure 3B). Both antibodies reacted with a single protein band of the expected size present in wild‐type extracts (Figure 3B, lanes 1 and 5). The same proteins were present in strain UJ199 and UJ200, respectively, at approximately wild‐type concentrations when grown in the presence of xylose (Figure 3B, lanes 2 and 6), but were absent in extracts of the same mutant strains that were grown in media containing glucose (Figure 3B, lanes 3 and 7). Thus strains UJ199 and UJ200 are depleted for the ClpP or ClpX proteins when grown overnight under non‐permissive conditions. The ClpP protein was present in strain UJ200 grown under restrictive conditions (Figure 3B, lane 4) demonstrating that the absence of the ClpX protein in this strain had no effect on the level of expression of the ClpP protein. Similarly, the chromosomal clpP mutation in strain UJ199 had no effect on the expression of the ClpX protein (Figure 3B, lane 8).
Strains UJ199 and UJ200 required exogenously added xylose for continuous growth both in complex and minimal media. Upon removal of xylose, both strains showed a severe reduction in biomass accumulation and cell viability (Figure 4A and B). Biomass accumulation of UJ199 and UJ200 was measured under non‐permissive conditions after overnight cultures were shifted from a xylose‐based medium to a medium containing glucose. Both mutant strains increased their biomass exponentially during ∼4 generations with and without xylose in the growth medium. After 4 or 5 generations both mutant strains stopped accumulating biomass in glucose medium while continuing to grow normally in medium containing xylose (Figure 4A). The increase in cell numbers of both mutant strains under permissive conditions (PYEX) again reflected exponential growth kinetics (Figure 4B). In sharp contrast, the increase of viable cell counts stopped only 1 or 2 generations after the media shift and subsequently dropped drastically for both clp mutant strains grown under restrictive conditions (PYEG). After 8 or 9 generations of growth on media without xylose, the viable cell numbers of the clpP and clpX mutants had dropped 4–5 orders of magnitude compared with the same strains supplemented with xylose (Figure 4B). It is interesting to note that loss of ClpX (UJ200) had a more drastic effect on both growth and viability than depletion of ClpP (UJ199) (Figure 4). This could be due to either a higher persistence of the ClpP protein at the restrictive conditions (see below) or to a stricter requirement for the ClpX protein during growth.
ClpP antibodies and ClpX antibodies were used to monitor the concentrations of the proteins in strains UJ199 and UJ200, after removal of xylose (Figure 4C). Most of the cellular ClpP and ClpX had disappeared 6 h later (corresponding to about four cell doublings), demonstrating that there is a clear correlation between the concentrations of the ClpP and ClpX proteins and the phenotypic behavior of the xylose‐dependent mutant strains. We conclude that loss of growth and viability is a direct effect of decreasing cellular concentrations of ClpP or ClpX.
Mutations in clpP or clpX result in cell‐cycle arrest
Both clp mutant strains, UJ199 and UJ200, accumulated biomass in the absence of the inducer for up to 7 generation‐time equivalents, although a decrease of viable cell numbers was detected by 2 generations after the shift to restrictive conditions. This suggests that removal of ClpP or ClpX resulted in an inhibition of cell division long before affecting the increase of biomass. In agreement with this, the cells of strains UJ199 and UJ200 were filamentous when grown in a xylose‐free medium for 12 h. Figure 5 shows light microscopy images of strains UJ199 and UJ200 grown under permissive and restrictive conditions. In the absence of xylose both strains show a drastically elongated and filamentous morphology indicative of an inhibition of cytokinesis (Figure 5B and D). In most cells division had not initiated or had stopped at an early stage. A fraction of the cells, which apparently had undergone initiation of division, remained attached to each other via a short stalk‐like structure suggesting another role for ClpP or ClpX during a later stage of cell division. Both mutant strains showed a normal morphology when grown on complex medium containing xylose (Figure 5A and C) confirming that the abnormal cell morphology was a result of the absence of ClpP or ClpX.
The inhibition of cell division could be a direct and immediate effect of the absence of the ClpXP protease. Alternatively, ClpXP could be required for an earlier event, which is acting as a checkpoint for processes occurring later during cell division. Since later steps in the cell cycle such as cytokinesis are dependent on the completion of chromosome replication (Osley and Newton, 1977), we tested whether depleting cells for either ClpP or ClpX would result in a block in DNA replication. For this purpose cells were stained with the DNA‐specific dye YO‐PRO‐1 and the DNA content per cell was measured by flow cytometry (Winzeler and Shapiro, 1995). Control experiments confirmed that SW cells (G1 phase) contain a single chromosome (Figure 6B), and late PD cells (G2 phase) contain two chromosomes (Figure 6C). A mixed culture includes G1 and G2 cells, and actively replicating cells (ST and early PD; S phase) with between one and two chromosomes (Figures 7A). Wild‐type cells that had been treated with rifampicin for 3 h contained either one or two chromosomes but no actively replicating DNA (Figure 6D). To assay the chromosome content of the clp mutants, strains UJ199 and UJ200 were grown on PYEX and then shifted to PYEG medium for 12 h. Complete depletion for either ClpP or ClpX resulted in a clear reduction of cells containing more than one chromosome. In the ClpP‐depleted culture a dominant peak corresponding to one chromosome was observed although a significant fraction of the cells had a higher chromosome content (Figure 6F). The latter could account for cells that had not completely separated in the absence of the ClpP protein (see Discussion). In a ClpX depleted culture, G1 cells predominated (Figure 6H). Under permissive conditions the chromosome pattern of the mutant strains (Figure 6E, G) was comparable to wild‐type cells (Figure 6A).
Thus, depletion of C.crescentus cells for either ClpP or ClpX resulted in a defect in initiation of chromosome replication, most prominent in the clpX mutant. To confirm the replication initiation defect of UJ200 mutant cells, DNA synthesis was monitored by incorporation of radioactively labeled dCTP over the course of one cell cycle. When UJ200 was grown under non‐permissive conditions (glucose) for 6 h (see below), DNA synthesis was strongly reduced, as compared with cells that were allowed to proceed through the cell cycle in the presence of the inducer (data not shown).
ClpXP is required for degradation of a replication control protein during G1‐to‐S phase transition
The replication block of SW cells is released during the G1‐to‐S transition by dephosphorylation and specific degradation of the CtrA response regulator protein (Quon et al., 1998). Expression of a mutant CtrA protein that is both constitutively active and stable results in cells arrested in G1 similar to results obtained for the clpP and clpX mutant strains (Domian et al., 1997). This prompted us to test whether ClpXP is required for the cell‐cycle‐dependent degradation of the CtrA protein. Immunoblot analysis of strains UJ199 and UJ200 revealed that CtrA levels were slightly increased in cells grown on glucose as compared with cells grown on xylose (data not shown). To analyze if this was due to a lack of CtrA turnover during G1‐to‐S transition we assayed CtrA levels as a function of the cell cycle in the presence or absence of the ClpP and ClpX proteins. Cultures of strains UJ199 and UJ200 were grown on xylose‐containing medium (PYEX) and subsequently shifted to restrictive conditions (PYEG) for 6 h (4 generations). At this stage the cells had lost most of their ClpP (UJ199) or ClpX (UJ200) proteins (Figure 4C) but enough SW cells could still be isolated for the subsequent cell‐cycle experiment. Purified SW cells were resuspended in complex medium containing either xylose or glucose and allowed to proceed through one cell cycle. The timing of stalk formation and cell elongation was comparable to wild‐type cells and was independent of the medium supplement suggesting that the ClpXP protease is not required for the SW‐to‐ST cell transition. However, while most of the cells of both UJ199 and UJ200 grown in the presence of xylose initiated septation, developed motility, and had divided after 90 min, the majority of the cells grown without xylose arrested as elongated cells without initiating septation even after an extended period (data not shown). Immunoblot analysis of cell extracts from UJ199 and UJ200 using an anti‐CtrA antibody (Domian et al., 1997) revealed that both strains degraded CtrA normally when going through the cell cycle in a medium containing xylose (Figure 7A and B). However, CtrA was not degraded during the SW‐to‐ST cell transition in either strain grown without xylose. Immunoblots with antibodies against ClpP and ClpX confirmed that both proteins were absent in the synchronized SW cells but were resynthesized rapidly in the mutant strains grown in the presence of xylose (Figure 7A and B). In contrast, synchronized cells of both mutant strains that had been resuspended in glucose containing medium did not synthesize ClpP or ClpX (Figure 7A and B). Thus, there is a clear correlation between the presence of the Clp proteins and the cell‐cycle‐dependent degradation of CtrA.
The results shown in Figure 7 suggest that the ClpXP protease is directly responsible for degradation of the cell‐cycle transcriptional regulator CtrA during the G1‐to‐S transition. This is supported by the observation that overexpression of the ctrA wild‐type gene from a high copy number plasmid caused a slightly filamentous phenotype in strains UJ199 and UJ200 even when grown under permissive conditions (data not shown). In a wild‐type background, overexpression of the ctrA gene was shown to have no visible effect (Domian et al., 1997). The activity of the CtrAD51E mutant protein is independent of phosphorylation and its inactivation upon entry into S phase relies entirely on proteolysis (Domian et al., 1997). When the ctrAD51E allele (Quon et al., 1996), was overexpressed in clpP and clpX mutant strains, the cells had a very strong cell division defect at the permissive conditions (PYEX). This phenotype was identical to that of the clp mutants grown for an extended period under non‐permissive conditions (Figure 5B and D). When strains UJ199 and UJ200 containing the ctrAD51E allele were shifted to the non‐permissive conditions (PYEG), they arrested growth immediately (data not shown) indicating a strict requirement of ClpP and ClpX for CtrAD51E removal.
The degradation of CtrA by ClpXP during the cell cycle must be a tightly controlled event. Several possible mechanisms exist for degradation control. One possibility is that the cell simply restricts the expression of one or both components of the ClpXP protease to only the period of the cell cycle or the compartment of the PD cell where their activity is needed. To test if the ClpP or ClpX protein levels fluctuate during the cell cycle we monitored their concentration during the cell cycle of NA1000 wild‐type cells by immunoblot. We found that neither the ClpP nor the ClpX protein concentration changed during the course of one cell cycle (Figure 8). These results exclude a mechanism of temporal separation of substrate and protease.
We show here that ClpXP, an essential bacterial protease, is required for the cell‐cycle‐dependent turnover of CtrA, an inhibitor of G1‐to‐S transition in the eubacterium C.crescentus. Mutant C.crescentus strains that either lack the regulatory or the proteolytic subunit of the ClpXP protease are blocked in the cell division process; flow cytometry experiments indicated that the mutants arrested in G1 before initiation of chromosome replication. The involvement of ClpXP in temporally controlled degradation of a key cell‐cycle regulator and its requirement for cell‐cycle progression emphasize the crucial role of this protease in C.crescentus cell‐cycle control. The nature of this control is remarkably similar to the mechanism that regulates the G1‐to‐S transition in the budding yeast Saccharomyces cerevisiae. In eukaryotes, different types of CDK complexes trigger the transition from one phase of the cell cycle to the next (Nigg, 1995). Highly specific proteolytic events remove the cyclin proteins at stages when they are no longer required or deleterious (King et al., 1996; Hoyt, 1997). In S.cerevisiae, the Sic1 cyclin inhibits CDKs which promote S phase. Sic1 is specifically removed at the end of G1 to allow initiation of the S phase and DNA replication (Schwob et al., 1994). Specificity is conferred by a ubiquitin‐conjugating complex that targets Sic1 for degradation by the 26S proteasome (Hoyt, 1997). Similarly, CtrA inhibits the G1‐to‐S transition in Caulobacter and is also specifically targeted for degradation to allow cell‐cycle progression. Thus, as is the case for eukaryotic cyclins, oscillation of CtrA levels in Caulobacter is the result of combined transcriptional and proteolytic regulation (Quon et al., 1996; Domian et al., 1997; this paper).
CtrA belongs to the superfamily of response regulators, most members of which control transcriptional activity of target promoters in response to aspartyl phosphorylation by a cognate sensory kinase (Parkinson and Kofoid, 1992). CtrA has been shown to control the expression of cell‐cycle‐regulated genes encoding flagellar proteins, a DNA methyltransferase and cell division proteins (Quon et al., 1996; Kelly et al., 1998). CtrA is also a critical element of DNA‐replication control in C.crescentus (Quon et al., 1998).
CtrA activity is temporally and spatially controlled at several levels: synthesis, phosphorylation and degradation (Quon et al., 1996; Domian et al., 1997). Specific degradation of CtrA occurs during the G1‐to‐S transition and in the ST compartment of the late PD cell (Domian et al., 1997). This observation immediately poses an important question: how is CtrA degradation restricted to a specific compartment of the C.crescentus cell and to a distinct period of its cell cycle? The finding that ClpXP is involved in CtrA turnover prompts us to consider this problem in terms of how the protease recognizes its substrate and how this activity is controlled? With the E.coli ClpXP protease complex, the ATPase subunit is responsible for substrate recognition, disassembly and presentation to the protease active sites (reviewed in: Gottesman et al., 1997a,b). Specificity is based on selective interactions between PDZ‐like domains in ClpX and disordered C‐terminal tails of the substrate proteins (Levchenko et al., 1997a). As with some of the C‐termini of known ClpXP substrates in E.coli, the C‐terminus of CtrA is essential for degradation (Domian et al., 1997). Even though a direct interaction between CtrA and ClpXP remains to be demonstrated, we hypothesize that the CtrA C‐terminus is one part of the protein that is selectively recognized by ClpX. A more indirect role for ClpXP in CtrA abundance cannot, at this point, be ruled out completely. However, the observation that constitutive overexpression of CtrA did not alter its degradation pattern during the cell cycle (Domian et al., 1997), strongly argues against ClpXP controlling expression of CtrA.
Flow cytometry studies show that most of the ClpP‐ or ClpX‐depleted filamentous cells contain a single chromosome, suggesting that initiation of DNA replication is disrupted in these cells. Evidence for such a defect after ClpP‐ or ClpX‐depletion was also obtained from directly measuring rates of DNA synthesis during the cell cycle. Thus, CtrA stabilization at the G1‐to‐S boundary coincides with a clear reduction of DNA replication initiation. However, replication is not completely abolished in these cells. It is possible that the low levels of ClpP or ClpX observed still allow some residual replication initiation activity but are not sufficient to degrade CtrA. Stabilizing CtrA alone is not sufficient to cause a G1 arrest (Domian et al., 1997). This argues against CtrA degradation being the only essential function of the ClpXP protease. Therefore, we propose that the cell‐cycle arrest observed in ClpX‐ or ClpP‐depleted cells must be caused by the stabilization of one or several additional proteins (Figure 9). It is intriguing to speculate that members of the phosphorelay system that activates CtrA (Shapiro and Losick, 1997; Wu et al., 1998) could be degraded by ClpXP during the cell cycle in a CtrA‐like manner. In this case mutational disruption of ClpXP would not only lead to stabilization of CtrA but, due to an increased level of phosphorelay components, could also result in an increased flow of phosphate through the relay. This in turn would lead to an increased phosphorylation status of a stabilized CtrA protein, a situation that has been shown to result in a block of chromosomal replication initiation and G1 arrest, and thus could explain the cell‐cycle phenotype observed in our clp mutants. Alternatively, stabilization of other ClpXP substrates which do not affect CtrA activity could contribute to the cell‐cycle arrest.
Our observation that the block in initiation of DNA replication was more pronounced after depletion of ClpX as compared with ClpP‐depletion, could be explained by a ClpP‐independent role for ClpX in replication control. In E.coli both ClpA and ClpX have been shown to function as chaperones involved in assembly and disassembly of DNA replication and transposition complexes (Wickner et al., 1994; Levchenko et al., 1995; Wawrzynow et al., 1995; Kruklitis et al., 1996; Konieczny and Helinski, 1997). In each of these cases it has been shown that the chaperone disassembles multimeric complexes of the substrate proteins into active monomers in vitro. It is intriguing to speculate that ClpX in C.crescentus might also be involved in remodelling of protein substrates. Disaggregation of CtrA multimers bound to the Cori (origin of replication) prior to their degradation could be one such activity. It is possible that while the chaperone activity of ClpX is required for CtrA disassembly, CtrA degradation is a non‐essential consequence of that former event. A protease‐independent role for ClpX in C.crescentus is supported by the finding that C.crescentus does not tolerate increased concentrations of the ATPase, while overexpression of ClpP has no consequences for the cells (U.Jenal, unpublished). The fact that fluctuations in ClpX levels have deleterious effects on the cell suggests that the ClpX concentration needs to be tightly controlled. In agreement with this, we have found that in C.crescentus, unlike in E.coli, clpX is not under heat‐shock control, and is not transcribed from the same promoter as clpP, but rather from at least three different promoters (M.Osteras, A.Stotz, S.Schmid Nuoffer and U.Jenal, manuscript in preparation).
The activity of the ClpXP protease must be under tight temporal control during the C.crescentus cell cycle. A simple control mechanism would be to restrict the presence of the ATPase to the corresponding time window or compartment. We have found that both ClpP and ClpX levels are not under cell‐cycle control in C.crescentus arguing against such a mechanism. However, we cannot exclude the possibility that ClpP and ClpX are separated physically inside the same cell interacting only during specific stages of the cell cycle. Alternatively, ClpP and/or ClpX activity itself could be modulated via the interaction with a regulatory protein. Such a mechanism has been shown for the control of the Lon protease activity in E.coli by the bacteriophage T4 protein PinA (Hilliard et al., 1998a,b). Finally, the accessibility of CtrA could determine its temporal and spatial degradation pattern. Precedents for such a mechanism is the interaction of ClpX with the Mu transposase, MuA (Levchenko et al., 1997b), the degradation of the E.coli heat‐shock sigma factor, RpoH, by the FtsH protease (Tomoyasu et al., 1995), and the degradation of the E.coli stationary‐phase sigma factor, RpoS, by ClpXP (Zhou and Gottesman, 1998). Interestingly, turnover of RpoS in E.coli seems to be controlled by a response‐regulator protein, RssB (also called SprE and MviA) (Muffler et al., 1996; Pratt and Silhavy, 1996; Zhou and Gottesman, 1998). One of the major tasks in the future will be to determine by which of the above mechanisms the activity of the ClpXP protease is regulated during the C.crescentus cell cycle.
Materials and methods
Bacterial strains, plasmids, media, enzymes and chemicals
The strains and plasmids used in this study are shown in Table I. Escherichia coli strains were grown at 37°C in LB (Sambrook et al., 1989), supplemented with ampicillin (100 μg/ml), carbenicillin (100 μg/ml), kanamycin (30 μg/ml), tetracycline (10 μg/ml), chloramphenicol (30 μg/ml), spectinomycin (50 μg/ml) or streptomycin (30 μg/ml) as necessary. Caulobacter crescentus strains were grown in either PYE complex medium or M2 minimal glucose (Ely, 1991) and M2 minimal xylose medium (0.2% xylose), respectively, supplemented with nalidixic acid (20 μg/ml), kanamycin (20 μg/ml), tetracycline (2 μg/ml), where indicated.
A horseradish peroxidase (HRP) conjugate of goat anti‐rabbit IgG was purchased from Gibco‐BRL (Gaithersburg, MD), Renaissance Western Blot chemiluminescence reagent from DuPont NEN (Boston, MA), restriction enzymes from New England Biolabs (Beverley, MA), resin for purification of His‐tagged proteins from Novagen (Madison, WI), kits for the isolation and purification of plasmid DNA from Qiagen (Hilden, Germany), the ‘QuickChange’ kit for site‐directed mutagenesis from Stratagene (La Jolla, CA), and YO‐PRO‐1 DNA stain for flow cytometry was from Molecular Probes (Eugene, OR).
Overexpression and purification of ClpP and ClpX His‐tagged fusion proteins and generation of antibodies
A fragment of the clpP gene encoding amino acids 1–205 was generated by PCR and cloned into pET‐21a resulting in construct pETP. A fragment of the clpX gene encoding amino acids 1–417 was generated by PCR and cloned into pET‐21a resulting in construct pETX. These constructs contained protein fusions of either clpP or clpX with an N‐terminal T7 epitope tag and a C‐terminal His tag. DNA sequence analysis was performed to verify the in‐frame nature of the fusions.
To induce expression of the clp fusion genes in E.coli, BL21(DE3) was transformed with pETP and pETX. Cultures of the obtained strains were grown to an OD600 of 1.0 and then induced by adding IPTG to a final concentration of 1 mM. Three h after induction the cells were harvested by centrifugation and lyzed by sonication. The overexpressed fusion proteins were found in inclusion bodies and were concentrated and washed by consecutive centrifugation steps. Resolubilization of inclusion bodies was performed in buffer containing 6 M urea and affinity purification on Sephadex nickel columns was carried out as described earlier (Jenal et al., 1994). The purified fractions were separated on a preparative SDS–polyacrylamide gel and the bands corresponding to the ClpP and ClpX fusion proteins were cut out. The polyacrylamide was crushed, mixed with Freund's adjuvant and used to immunize rabbits. Immunization and sampling of the sera was performed by Eurogentec (Seraing, Belgium).
Construction of chromosomal clpP and clpX interposon mutations
To construct chromosomal clpP and clpX mutants both genes were independently disrupted by insertion of an omega cassette (Prentki and Krisch, 1984) and the chromosomal wild‐type genes were exchanged by the mutated copies via homologous recombination. A 2.7 kb EcoRI–BamHI fragment containing clpP and its flanking regions was cloned into the vector pNPT228 resulting in construct pUJ112. A SmaI fragment containing the omega cassette was then cloned into the blunt‐ended single XhoI site in clpP resulting in construct pUJ115. An AflII–SpeI fragment containing the clp region with the interposon was then subcloned from pUJ115 into pNPTS138, to produce pUJ116. pNPTS138 contains two selection markers for the subsequent homologous recombination step: nptI, conferring kanamycin resistance; sacB, conferring sucrose sensitivity. To isolate a disrupted clpX copy a 2.2kb ApaI–EcoRV fragment containing clpX and flanking regions was cloned into pNPTS138 resulting in construct pTF20. A SacI–XmnI internal clpX fragment was then exchanged by the SmaI fragment containing the interposon resulting in construct pTF21. pUJ116 and pTF21 were integrated into the chromosome of C.crescentus wild‐type strain NA1000 via conjugation and homologous recombination. The correct loci of integration into the chromosome were verified by Southern blot analysis. The resulting strains (LS2037, NA1000::pUJ116; UJ127, NA1000::pTF21) contained a wild‐type and a disrupted copy of either clpP or clpX. To select for the second recombination event cultures of LS2037 and UJ127 were grown on PYE overnight and aliquots were then spread on PYE plates containing 3% sucrose. Colonies that were able to grow on sucrose were then tested for loss of the plasmid encoded kanamycin resistance and for the presence of the spectinomycin/streptomycin resistance from the interposon. Sucrose‐resistant clones that retained kanamycin resistance had not lost the plasmid by homologous recombination and presumably had acquired sucrose resistance through the mutational inactivation of the sacB gene. Kanamycin‐sensitive, spectinomycin‐sensitive colonies had lost the plasmid and recombination left a wild‐type clp locus in the chromosome of these strains. Mutant strains with a disrupted chromosomal clpP or clpX copy were identified by their sensitivity to kanamycin and resistance to spectinomycin and streptomycin.
Generation of conditional clpP and clpX mutants
To generate conditional clpP and clpX mutants, strains were constructed that contained a single functional copy of either clpP or clpX under the control of the inducible xylose promoter (Meisenzahl et al., 1997). Primers ClpP5′RI (5′‐AAA GAA TTC ATG ATG TAC GAT CCG GTT) and ClpP3′Sp (5′‐AAA ACT AGT CCG GCG ACA CCG AGA TGC) were used to amplify by PCR the entire clpP gene including the downstream transcriptional terminator signals (M.Osteras, A.Stotz, S.Schmid Nuoffer and U.Jenal, manuscript in preparation). Similarly, the clpX gene was amplified using primers ClpX5′RI (5′‐AAA GAA TTC ATG ACG AAA GCC GCG AGC) and ClpX3′Sp (5′‐AAA ACT AGT GCT TCG AAA GCA CGC GCT). Both products were then cloned into the EcoRI–SpeI sites of plasmid pUJ83, resulting in constructs pUJ165 (clpP) and pUJ166 (clpX), respectively. This resulted in a translational fusion between the first six codons of the xylX gene and the complete clpP or clpX coding regions. A SpeI–HindIII fragment was then subcloned from pUJ165 (PxylX::clpP) and from pUJ166 (PxylX::clpX) into the vector pPHU281 (Hübner et al., 1993) resulting in constructs pUJ175 (clpP) and pUJ176 (clpX). pUJ175 and pUJ176 were introduced into the chromosomal xylX locus strains LS2037 and UJ127, respectively, by conjugational transfer and subsequent homologous recombination. Insertion at the correct loci of plasmids pUJ175 and pUJ176 was verified by Southern blot analysis. The resulting strains had a total of three clpP or clpX copies: strain UJ218 contained a functional clpP copy fused to PxylX, a clpP wild‐type gene and a clpP gene disrupted by an omega cassette in the clp locus. Strain UJ220 contained a functional clpX copy fused to PxylX, a clpX wild‐type gene and a clpX gene disrupted by an omega cassette in the clp locus. To isolate strains that carried a single functional clpP or clpX gene in their xylX locus, sucrose selection was carried out with strains UJ218 and UJ220. Cultures of both strains were grown overnight on PYE complex medium and then spread onto plates containing 3% sucrose and 0.1% xylose or glucose. Sucrose‐resistant colonies were then tested for their antibiotic resistance profile as described above. Colonies that were kanamycin sensitive, spectinomycin and streptomycin resistant, as well as tetracycline resistant were then assayed for their growth dependence on xylose by streaking onto PYE plates that either contained 0.1% xylose or 0.1% glucose. Xylose‐dependent expression of ClpP in strain UJ199 and of ClpX in strain UJ200 was assayed by immunoblot analysis (see below).
To construct xylose‐dependent clpP and clpX strains with a kanamycin‐resistance instead of a tetracycline‐resistance marker the clpP and clpX PCR products (see above) were cloned into the EcoRI–SpeI sites of vector pNPT228XNE resulting in constructs pUJ167 and pUJ168, respectively. pUJ167 was introduced into the xylX locus of strain UJ36 (NA1000 clpP::Ω / pUJ158) resulting in strain UJ223, and pUJ168 was introduced into the xylX locus of strain UJ222 (NA1000 clpX::Ω / pUJ170) resulting in strain UJ224. The only functional clpP copy of UJ36 is located on plasmid pUJ158 and strain UJ222 contains a single functional clpX copy on plasmid pUJ170. To remove the plasmid borne clpP and clpX copies from strains UJ223 and UJ224, respectively, pUJ158 was chased with plasmid pBBR1MCS‐Gent and pUJ170 was chased with plasmid pGML11. The resulting strains were then restreaked three times without gentamycin selection and clones that had spontaneously lost pBBR1MCS‐Gent (UJ270) or pGML11 (UJ271) were isolated. Xylose‐dependent expression of ClpP in strain UJ270 and of ClpX in strain UJ271 was assayed by immunoblot analysis (see below).
Immunoblots were performed as described previously (Jenal and Shapiro, 1996). Polyclonal sera against ClpP or ClpX were diluted 1:5000 and 1:1000, respectively, before use. Polyclonal serum that specifically recognizes CtrA (Domian et al., 1997) was diluted 1:5000 before use. Secondary antibody (goat anti‐rabbit IgG, HRP‐coupled) was used in a 1:10 000 dilution. Immunoblots were developed using the Renaissance kit from DuPont NEN, following the manufacturer's instructions.
Mixed populations of NA1000 wild‐type strain, UJ199 and UJ200 mutants were grown in PYE complex medium to an OD600 of 0.5. Caulobacter crescentus wild‐type cells were synchronized as described previously (Stephens and Shapiro, 1993) and pure populations of swarmer cells (G1) or late PD cells (G2) were harvested at an OD600 of 0.5. Mixed and pure cell populations were washed in 1 volume of staining buffer (10 mM Tris, 1 mM EDTA, 50 mM NaCitrate pH 7.2) and fixed immediately in 70% ethanol at −20°C for 1 h. Fixed cells could be stored up to 1 month at −20°C in ethanol. As an additional control, wild‐type cells were grown in PYE to an OD600 of 0.3 before rifampicin (15 μg/ml) was added to the culture for 3 h. Cells were then harvested and fixed as described above. Fixed cells were washed in 1 volume staining buffer and incubated in 1 volume staining buffer containing 0.1 mg/ml RNaseA for 30 min at room temperature. After adding 0.5 volumes staining buffer containing 1.5 μM YO‐PRO‐1 the cells were incubated in the dark for 2 h and immediately analyzed with a FACScan (Beckton Dickinson). If necessary, cells were diluted with staining buffer to an optimal cell number of 106–107/ml. For each measurement the DNA content of 5×104 cells was determined.
To analyze cell‐cycle‐dependent CtrA degradation in the mutant strains UJ199 and UJ200 under non‐permissive conditions, cultures of both strains were grown in PYE medium containing 0.1% xylose (PYEX) to an OD600 of 0.5–0.8. Cells were then washed three times in PYE medium containing 0.1% glucose (PYEG) and diluted into fresh PYEG at an OD600 of 0.05. Cultures were grown in PYEG to an OD600 of 0.5–0.6 and samples were taken periodically to monitor the decrease of ClpP and ClpX concentrations by immunoblots. After the cultures had reached an OD600 of 0.5–0.6 motility had decreased drastically. Cells were then harvested and synchronized as described previously (Stephens and Shapiro, 1993). Pure populations of swarmer cells were resuspended in either PYEG or PYEX and allowed to proceed through the cell cycle. Progression of the cell cycle was monitored microscopically and samples were collected for immunoblot analysis.
We thank C.Stephens, Y.Brun, K.Ryan and members of our laboratory for critical reading of the manuscript. We thank K.Quon, I.Domian and L.Shapiro for the gift of strains and anti‐CtrA antibody and for helpful discussions. We also thank M.Dessing for helpful assistance with the FACS analysis. This work has been supported by a Swiss National Science Foundation fellowship 31‐46764.96 to U.J.
- Copyright © 1998 European Molecular Biology Organization