Energy-dependent proteases ensure the timely removal of unwanted proteins in a highly selective fashion. In Caulobacter crescentus, protein degradation by the ClpXP protease is critical for cell cycle progression; however, only a handful of substrates are currently known. Here, we use a trapping approach to identify putative substrates of the ClpP associated proteases in C. crescentus. Biochemical validation of several of these targets reveals specific protease recognition motifs and suggests a need for ClpXP-specific degradation beyond degradation of known cell cycle regulators. We focus on a particular instance of regulated proteolysis in Caulobacter by exploring the role of ClpXP in degrading the stalk synthesis transcription factor TacA. We show that TacA degradation is controlled during the cell cycle dependent on the ClpXP regulator CpdR and that stabilization of TacA increases degradation of another ClpXP substrate, CtrA, while restoring deficiencies associated with prolific CpdR activity. Together, our work reveals a number of new validated ClpXP substrates, clarifies rules of protease substrate selection, and demonstrates how regulated protein degradation is critical for Caulobacter development and cell cycle progression.
The proteome is maintained by the synthesis and removal of cellular proteins at specific times and places in response to internal and external signals (Goley et al., 2009; Sauer and Baker, 2011). Proteolysis is a robust process that ensures the elimination of proteins when they are no longer needed; however, the irreversible nature of protein degradation demands stringent selectivity in substrate choice. In bacteria, regulated protein degradation is often governed by energy-dependent proteases, also known as AAA+ proteases. One member of this family is ClpXP, an oligomeric machine formed from the ClpX unfoldase and ClpP peptidase. ClpP alone is incapable of degrading folded proteins and requires associated unfoldases, such as ClpX, that engage substrates via binding of short peptide sequences (recognition motifs). Targets are actively unfolded and translocated into the ClpP chamber where they are processed to peptides. The ClpP system has been best characterized in Escherichia coli where the Clp family proteases regulate many pathways such as protein quality control, exit from stationary phase and response to damaging agents (Sauer and Baker, 2011).
Caulobacter crescentus hereafter referred to as either Caulobacter, or C. crescentus, undergoes an obligate morphological differentiation from nonreplicative swarmer cells to replication competent stalked cells every cell cycle. This remodelling depends on proper coordination of regulatory proteolysis and synthesis. For example, ClpX and ClpP are essential in Caulobacter (Jenal and Fuchs, 1998) and regulate proteins important for cell development, cell cycle progression (Domian et al., 1997; Iniesta et al., 2006; Biondi et al., 2006a; Radhakrishnan et al., 2010), chemotaxis (Tsai and Alley, 2001; Potocka et al., 2002) and replication (Gorbatyuk and Marczynski, 2005; Lesley and Shapiro, 2008). Although several ClpXP substrates have been identified in Caulobacter, expression of mutant non-degradable alleles of these proteins or blocking degradation of these substrates does not mirror the many effects that arise when ClpXP is lost (Domian et al., 1997; Iniesta et al., 2006; Radhakrishnan et al., 2010). Thus the ClpXP-dependent changes in development and cell cycle progression must rely on degradation of other as yet unknown substrates.
Here, we present a list of potential substrates of the ClpP protease system in Caulobacter that includes known substrates, those documented from other bacteria, and previously unknown candidates. In vitro biochemical experiments validate a number of these targets and confirm the preservation of particular ClpXP recognition motifs across species. Next, we identify a role for the zinc-binding N-terminal domain of ClpX for Caulobacter viability and endogenous substrate selection. Finally, we focus on the stalked cell regulator TacA, a candidate substrate found in our trap. We show that TacA is proteolysed by ClpXP in vitro and that TacA is selectively degraded during cell cycle progression dependent on the ClpXP regulator CpdR. Interestingly, stabilization of TacA also affects the degradation of CtrA, another CpdR-dependent ClpXP substrate, suggesting a linkage between TacA, CpdR and CtrA. Finally, we find that prolific TacA degradation contributes to the stalkless phenotype of cells lacking the PleC phosphatase as stabilization of TacA partially rescues stalked cell formation in this mutant.
Construction of a ClpP trap in C. crescentus
We identified novel substrates of ClpXP in C. crescentus using the strategy outlined in Fig. 1 similar to that performed previously in E. coli (Flynn et al., 2003). Briefly, we mutated the active-site serine of ClpP and appended a C-terminal tandem affinity tag to generate ClpPtrap (Fig. S1). Substrate proteins are recognized by the native active ClpX or ClpA unfoldases and translocated into the inactive chamber of ClpPtrap (Katayama et al., 1988; Gottesman et al., 1993; Flynn et al., 2003). The substrate-containing ClpPtrap complexes are then rapidly purified through successive rounds of affinity chromography. Our initial efforts in wild-type cells were unsuccessful, likely due to incorporation of active subunits in the ClpPtrap oligomer; however, depletion of the chromosomally encoded ClpP while expressing ClpPtrap improved the recovery of candidates (Fig. 1). Known substrates of ClpXP, such as the master regulator CtrA, were specifically enriched in the trapped substrate pool (Figs 1 and S1) and subsequent mass spectrometry identified 127 unique proteins in the eluted fraction (Fig. 1; Table S1). We pruned this list with high stringency requirements (see Supplemental information) to generate a candidate list of 32 proteins (Table 1). We note that likely ClpP substrates (such as LexA and others shown below) fail to meet these more stringent requirements, but are clearly identified in our trap (Tables S1 and S2); therefore, these lists should not be considered exhaustive, but reflect a high confidence subset of ClpP substrates.
Table 1. Subset (32/127) trapped candidates that satisfy high stringency requirements of > 5 unique peptides in replicate trapping experiments, removing proteins that co-purified with active ClpP, and eliminating proteins that are overly abundant in whole-cell extracts (see Supplemental information)
Candidates are listed by CC numbers from CB15 (GenBank ID AE005673) or CCNA numbers from NA1000 (GenBank ID CP001340). Where known, gene names are also listed. References are: (1) proteins known to be unstable during cell cycle progression (Grunenfelder et al., 2001); (2) genes known to be transcriptionally regulated during cell cycle (Laub et al., 2000); (3) proteins known to be tagged by the trans-translation pathway (Hong et al., 2007); (4) proteins that end in either AA or SA at their extreme C-terminus; (5) putative ClpP substrates identified in other bacteria (Flynn et al., 2003; Neher et al., 2006; Gerth et al., 2008); and (6) proteins whose levels decrease during carbon starvation (Britos et al., 2011). An expanded version of this table with all ClpP candidates (127) can be found in Supplemental information (Tables S1 and S2).
2, 5, 6
1, 2, 5, 6
1, 2, 4, 6
The ClpP candidate pool shown in Tables 1 and S1 is distributed across a wide range of functional categories. Mass spectrometry results identified two previously known Caulobacter ClpP substrates [CtrA and SspB (Domian et al., 1997; Jenal and Fuchs, 1998; Chien et al., 2007b)] and Western blot analysis showed the capture of other known ClpXP substrates such as PdeA (Abel et al., 2011; Rood et al., 2012) that were not detected by mass spectrometry (Fig. S1). Interestingly, a number of the candidates found by mass spectrometry had been identified in other proteomic surveys as potential ClpP regulated targets. For example, several are annotated in Caulobacter as substrates of the trans-translation pathway (Hong et al., 2007), wherein polypeptides are co-translationally appended with degradation tags that target them to either ClpXP or ClpAP. It is possible that some of these candidates are only targets of ClpP when tagged by the ssrA peptide. We also find conservation of ClpP substrates among widely divergent organisms as several candidates are known to be degraded in a ClpP-dependent fashion in either E. coli or Bacillus subtilis (Flynn et al., 2003; Kock et al., 2004; Neher et al., 2006; Gerth et al., 2008). Finally, we identify a number of substrates previously shown to be unstable in Caulobacter whose regulation can now be assigned to ClpP (Grunenfelder et al., 2001; Britos et al., 2011).
Recognition motifs for ClpXP are preserved across species
We next used a biochemical approach to validate putative ClpP substrates. We initially focused on candidates with known ClpXP recognition motifs classified by previous studies in E. coli by recombinantly expressing, purifying and biochemically validating several ‘C-terminal class I’ family substrates (distinguished by the presence of Ala–Ala at the extreme C-terminus) (Flynn et al., 2003). Of the six candidates of this class that we tested, four (FlaF, IbpA, CtrA, CC2882) were recognized by ClpXP in vitro (Chien et al., 2007b) (Figs 2 and S2). Furthermore, the ClpXP-dependent degradation of CC2323, which contains SA rather than AA, agrees with the known tolerance of this motif (Fig. S2A and B). Mutating the C-terminal residues of FlaF and IbpA to Asp-Asp eliminates recognition, revealing the importance of the C-terminal dipeptide for specific recognition by ClpXP (Fig. 2A and B). The validated ClpXP substrates do not share any obvious common functional or sequence features that distinguish them from non-degraded substrates (DnaK and CC0321). Thus, our data show, as would be predicted, that while conserved protease recognition motifs can dictate targeting by a protease, the simple presence of a particular motif does not ensure that a protein will be directly recognized by a particular protease.
The N-terminal domain of ClpX is critical for its in vivo function
We identified CC0360, a putative ornithine decarboxylase, as a likely ClpP substrate using our most stringent requirements (Table 1) and found that it was rapidly degraded by ClpXP in vitro (Fig. 3A). The lack of a C-terminal Ala–Ala motif in CC0360 (Fig. 2A) prompted us to explore how ClpXP was recognizing this substrate. The N-terminal zinc-binding domain of ClpX is unique to the ClpX family of unfoldases and is required for recognition of some substrates, such as MuA, lambdaO and FtsZ for the E. coli ClpX (Wojtyra et al., 2003; Abdelhakim et al., 2008; Camberg et al., 2009); but dispensable for degradation of others, such as CtrA and ssrA-tagged proteins for the C. crescentus ClpX (Figs 3A and B and S3) (Chien et al., 2007a,b). As in E. coli, the N-terminal domain of the C. crescentus ClpX is also critical for adaptor binding, as seen in the case of SspB (Chien et al., 2007a). Interestingly, we found that CC0360 was exclusively degraded by the full-length ClpXP complex and not by a version of ClpX lacking the N-terminal domain (ΔNClpXP; Fig. 3A). A similar, but less pronounced effect was seen with CC2882 (Fig. S3C), while other substrates, such as FlaF was recognized by ΔNClpXP with rates similar to full-length ClpXP (Fig. S3D). Because ClpX is essential in Caulobacter, we tested the importance of the N-terminal domain of ClpX in vivo and found that ΔNClpX could not support viability (Fig. 3C). We interpret this to mean that at least one important role for the full-length ClpXP in vivo is the degradation of substrates which rely on the N-terminal domain of ClpX (such as CC0360). This, of course, does not rule out other crucial roles for the N-terminal domain of ClpX, such as adaptor binding or localization.
TacA is degraded in a regulated fashion by ClpXP during the cell cycle
Stalk synthesis is a programmed developmental step in the life cycle of Caulobacter and is controlled transcriptionally by the σ54-dependent response regulator TacA (Marques et al., 1997; Biondi et al., 2006b; Radhakrishnan et al., 2008). TacA was identified in our trapping study and we found that the purified protein was degraded in vitro by ClpXP (Fig. 4A). ClpXP recognizes the TacA via its C-terminal residues (ending in MKEAG-cooh) as mutation of the C-terminal dipeptide AG to DD prevents degradation (Fig. 4A). An active, epitope-tagged variant of TacA (Biondi et al., 2006b) was degraded in vivo and stabilized by the same mutations that eliminate ClpXP recognition in vitro (Figs 4B and S4B). The regulator CpdR is required for the polar localization of ClpXP during the cell cycle (Iniesta et al., 2006) and is critical for the in vivo degradation of all known ClpXP substrates to date (Iniesta et al., 2006; Biondi et al., 2006a; Abel et al., 2011). As predicted from these observations, and in accordance with its ability to be degraded by ClpXP in vitro (Fig. 4A), we saw that the stability of TacA was dependent on CpdR in vivo (Fig. 4B), further supporting its identification as a true ClpXP substrate. Stabilization of TacA results in smaller colonies when cells are inoculated into low-percentage agar media, suggesting that TacA degradation may impact normal growth, differentiation or motility (Fig. S4D and E).
A constitutively expressed M2-TacA was degraded at the G1–S transition in a CpdR-dependent fashion (Figs 4C and S4C),similar to that seen for CtrA (Domian et al., 1997; Iniesta et al., 2006). Interestingly, cells expressing the non-degradable M2-TacA-DD show a delay in CtrA accumulation when compared with cells expressing M2-TacA (Fig. 4C). Because CtrA positively regulates its own expression late in the cell cycle (Domian et al., 1999), this result suggests that stabilized TacA results in lower CtrA activity. Our data support a post-translational role for this lower CtrA activity as CtrA is degraded more rapidly in cells expressing the non-proteolysed M2-TacA-DD compared with cells expressing the properly degraded M2-TacA (Fig. 4D).
Stabilization of TacA restores stalk formation in pleC mutants
Our results suggest proteolysis of TacA is integrated with changes in CpdR activity which in turn are dependent on the changing status of DivK phosphorylation during the cell cycle (Curtis and Brun, 2010). DivK phosphorylation by DivJ is countered by the PleC phosphatase and cells depleted of PleC have reduced CckA kinase activity resulting in lower levels of phosphorylated CpdR (depicted schematically in Fig. 5A) (Tsokos et al., 2011). Based on these observations, we hypothesized that the defects in stalk formation of pleC mutant cells (Sommer and Newton, 1989) may arise in part from the rapid turnover of TacA driven by persistent dephosphorylation of CpdR. If our model is correct, then expression of a non-degradable TacA should restore stalk formation in ΔpleC cells. We tested our hypothesis by replacing the endogenous allele of TacA in ΔpleC cells with a non-degradable TacA-DD variant. Because stalked and swarmer cells are difficult to distinguish by size alone, we counted the number of predivisional cells in a mixed population with visible stalks (as illustrated in Fig. 5B). Over 90% of wild-type predivisional cells had visible stalks, while 2–3% of predivisional ΔpleC cells expressing normal TacA have stalks (Fig. 5C). Expressing a stabilized variant of TacA in ΔpleC cells increased the fraction of stalked predivisional cells fivefold (Fig. 5C). Consistent with these findings, stalk elongation in ΔpleC strains during phosphate limitation is also increased when TacA is stabilized (Fig. S5). Collectively, our results show how CpdR activity, which is normally regulated during cell cycle, can affect the developmental programme of stalk formation via degradation of the novel ClpXP substrate TacA.
Regulated proteolysis is essential for biological processes in all organisms and ClpP associated proteases contribute to a large fraction of protein degradation in bacteria (Sauer and Baker, 2011). In C. crescentus, control of proteolysis during cell cycle is critical for timing of development, DNA replication and cell division (Curtis and Brun, 2010). By using a ClpP trapping approach, we now report a number of potential ClpP candidates that include known substrates (such as the master regulator CtrA), factors known to be ClpP substrates in other bacteria (such as LexA and FtsZ), and several novel ClpXP-dependent proteolysis targets (such as TacA and IbpA). Some of these validated Caulobacter ClpXP substrates are targeted for degradation in other bacteria by other energy-dependent proteases [e.g. IbpA is degraded by Lon in E. coli (Bissonnette et al., 2010)]. Interestingly, CC0360, a putative ornithine decarboxylase, is degraded rapidly by full-length ClpX from Caulobacter and the eukaryotic ornithine decarboxylase is a rapidly degraded ubiquitin independent substrate of the 26S proteasome (Murakami et al., 1992). Biological networks rely on rapidly changing protein levels for effective signalling and the response in a protein's concentration given changes in its synthesis rate is more rapid when the half-life of that protein is short (Rosenfeld et al., 2002). Thus, our work suggests that some proteins may be rapidly degraded regardless of the organism or proteolytic pathway involved because there is a universal need for their dynamic responses.
ClpXP mediated degradation of protein substrates often requires small peptide motifs that are directly recognized by the protease (Sauer and Baker, 2011) and a classification of these motifs has been proposed (Flynn et al., 2003). We tested the conservation of these protease recognition motifs in our candidate pool and found that while motifs can direct degradation of some substrates, the simple presence of a motif is insufficient to predict direct recognition by ClpXP. We also found that, unlike E. coli, Caulobacter viability requires a function unique to the full-length ClpX protein. Although a variant of ClpX lacking its N-terminal zinc-binding domain can readily degrade substrates such as ssrA-tagged proteins and CtrA, this variant is incapable of supporting viability in Caulobacter and poorly degrades some of the validated ClpXP substrates presented in this work. Together, these biochemical observations demonstrate the utility and limitations of recognition motifs for substrate recognition and reveal a crucial role for the N-terminal domain of the Caulobacter ClpXP in substrate recognition and in vivo function.
The characteristic swarmer to stalked cell transition in Caulobacter requires a substantial reorganization of the proteome and we identified a number of cell stage-specific factors in our trapping studies including the σ54-dependent response regulator TacA, several chemotaxis proteins (CheR, CheW, CheD and McpJ) and flagellar regulators (FlaF and FliM). Our results show that TacA is directly recognized by ClpXP in vitro but degradation in vivo requires CpdR. CpdR is needed for degradation of all known ClpXP substrates in vivo and dephosphorylation of CpdR increases degradation of these substrates (Iniesta et al., 2006; Biondi et al., 2006a; Abel et al., 2011). Phosphorylation of CpdR occurs via the CckA/ChpT phosphotransfer pathway (Biondi et al., 2006a) and CckA is repressed by phosphorylated DivK (Tsokos et al., 2011). The PleC phosphatase dephosphorylates DivK, and as a consequence, loss of pleC results in more dephosphorylated CpdR (Tsokos et al., 2011) and presumably would increase degradation of ClpXP substrates, such as TacA. Because ΔpleC cells lack stalks (Sommer and Newton, 1989) and as TacA is critical for stalk formation (Biondi et al., 2006b), it is tempting to speculate that prolific degradation of TacA is responsible for this absence of stalked cells. In support of this model, we found that stabilization of TacA partially restored stalked cell formation (Fig. 5), although the lack of complete restoration implicates the role of additional factors as well.
Why does stabilization of TacA cause increased degradation of CtrA? TacA influences asymmetric cell division in Caulobacter in part through synthesis of SpmX which is needed for the proper accumulation and activation of the histidine kinase DivJ that phosphorylates DivK (Radhakrishnan et al., 2008). As mentioned above, DivK promotes CpdR and CtrA dephosphorylation by inhibiting the kinase activity of CckA (Tsokos et al., 2011), unleashing CpdR's ability to activate ClpXP (Iniesta et al., 2006; Biondi et al., 2006a; Abel et al., 2011) but blocking CtrA's activity as a transcription factor (Quon et al., 1996; Biondi et al., 2006a). One intriguing possibility is that TacA promotes CpdR dephosphorylation, perhaps through SpmX, effectively stimulating its own degradation and that of CtrA. In this type of negative feedback loop, increasing TacA through stabilization may result in prolific CpdR activation that could underlie the increased degradation of CtrA.
Plasmids and strain construction
Standard techniques were used to manipulate C. crescentus. All C. crescentus strains were grown at 30°C in PYE with the appropriate antibiotics and sugars (Skerker et al., 2005) and can be found in Table S3. Briefly, we generated the proteolytic inactive ClpPtrap by mutating the active-site serine to alanine (S98A) and appending a tandem affinity tag (His6-TEV-M2FLAG) to the C-terminus of ClpP. Low-copy plasmids expressing ClpPtrap or wild-type affinity tagged ClpP were transformed into a strain whose sole, genomic copy of ClpP was driven by the xylose promoter (Jenal and Fuchs, 1998) to generate trapping or control strains. ClpP substrate candidates were cloned with appropriate primers either into pET23SUMO plasmids for recombinant protein expression in E. coli (Wang et al., 2007) or into pENTR plasmids for Gateway-based cloning into expression plasmids for either E. coli or C. crescentus (Skerker et al., 2005). ΔNClpX (removing residues 1–58) and full-length ClpX were expressed in ClpX depletion strains (Jenal and Fuchs, 1998) using low-copy plasmids downstream of the clpX promoter (Osteras et al., 1999). Caulobacter strains harbouring alleles of TacA-DD were generated by transforming wild type or ΔpdeA:tet (Skerker et al., 2005) with pENTR TacA-DD. Because these promoter-less plasmids cannot replicate in Caulobacter, antibiotic resistance is gained only if they are integrated. Subsequent screening by PCR verified that plasmids were inserted at the correct locus downstream of the native tacA promoter.
Purification of ClpP trap and candidate identification
Detailed procedures for ClpPtrap expression and purification are given in the Supplemental information. Briefly, sequential affinity columns were used to purify ClpPtrap after depletion of endogenous ClpP, preliminary validation of the trap was confirmed by Western analysis against the known substrate CtrA. Trapped substrates were separated by 15% SDS-PAGE, gel slices (∼ 1–5 mm) were excised, digested with trypsin and analysed by mass spectrometry (Proteomics and Mass Spectrometry Facility, University of Massachusetts, Worcester).
Protein purification and degradation assays
Detailed information is given in Supplemental information. Untagged C. crescentus ClpX and ΔNClpX were purified using a similar protocol as that described for E. coli ClpX (Flynn et al., 2003). Recombinant his-tagged C. crescentus ClpP was purified as described (Chien et al., 2007b). Candidates from the ClpP trap were expressed either as his-tagged constructs (Skerker et al., 2005) or as his-SUMO-tagged constructs (Wang et al., 2007) and purified accordingly. Degradation reactions were performed in H-buffer (20 mM HEPES, 100 mM KCl, 10 mM MgCl2, 10% glycerol, pH 7.5; and a creatine kinase-based ATP regeneration mix) at 30°C (Chien et al., 2007b). In a typical time-course, aliquots at specific times were added to SDS-PAGE loading dye to quench the reaction and snap frozen on dry ice. Samples were heated at 95°C for 5 min or 65°C for 10 min and resolved by 12% or 15% SDS/PAGE gels. Coomassie-stained gels were quantified by measuring intensity of substrate bands using ImageJ (NIH).
In vivo stability and synchrony experiments
For asynchronous in vivo degradation assays, cells with plasmids expressing epitope tagged TacA from xylose promoters were grown in PYE with xylose (0.1%) and spectinomycin (25 μg μl−1) overnight. In the morning the cultures were back diluted into the same media, grown until exponential growth, after which cells were harvested, washed twice in fresh PYE medium and resuspended in PYE-glucose to initiate the time-course. Protein levels were assayed by Western blotting using antibodies against the M2-FLAG epitope (Sigma-Aldrich). For synchronized growth, Caulobacter were grown to an OD600 of 0.3–0.5 in PYE medium with the appropriate antibiotic, swarmer cells were isolated by Percoll density centrifugation, and cells were released into the same media. For CtrA stability experiments, swarmer cells were isolated and resuspended in pre-warmed PYE media containing 30 μg ml−1 chloramphenicol to shut off synthesis.
We thank U. Jenal, J. Gober and Y. Brun for providing reagents and strains; and L. Francis, M. Laub for comments. N.B. and P.C. were funded by NIH Grant R00GM084157. P.S. was also supported by a HHMI Summer Research Fellowship for Undergraduates. We especially thank K. Green and the Proteomics and Mass Spectrometry Facility at the University of Massachusetts, Worcester Medical School.