Cell cycle transitions are often triggered by the proteolysis of key regulatory proteins. In Caulobacter crescentus, the G1-S transition involves the degradation of an essential DNA-binding response regulator, CtrA, by the ClpXP protease. Here, we show that another critical cell cycle regulator, SciP, is also degraded during the G1-S transition, but by the Lon protease. SciP is a small protein that binds directly to CtrA and prevents it from activating target genes during G1. We demonstrate that SciP must be degraded during the G1-S transition so that cells can properly activate CtrA-dependent genes following DNA replication initiation and the reaccumulation of CtrA. These results indicate that like CtrA, SciP levels are tightly regulated during the Caulobacter cell cycle. In addition, we show that formation of a complex between CtrA and SciP at target promoters protects both proteins from their respective proteases. Degradation of either protein thus helps trigger the destruction of the other, facilitating a cooperative disassembly of the complex. Collectively, our results indicate that ClpXP and Lon each degrade an important cell cycle regulator, helping to trigger the onset of S phase and prepare cells for the subsequent programmes of gene expression critical to polar morphogenesis and cell division.
Protein degradation is an excellent and rapid mechanism to completely remove undesired proteins from cells. In bacteria, regulated proteolysis is particularly important when cells must drastically change their proteome composition without dilution by cell division. In the α-proteobacterium Caulobacter crescentus, motile swarmer cells differentiate into sessile stalked cells at the same time they initiate DNA replication and transition from a G1 to S phase (Fig. 1A). Controlled proteolysis of many proteins is critical for this developmental and cell cycle transition and involves AAA+ proteases, oligomeric enzymes that harness ATP hydrolysis to power unfolding and degradation of specific substrates (Sauer and Baker, 2011). One of these AAA+ enzymes is the essential ClpXP protease (Jenal and Fuchs, 1998), which degrades a number of substrates including CtrA, a master cell cycle regulator. CtrA is a response regulator with a DNA-binding domain that enables it to regulate the transcription of nearly 100 genes (Quon et al., 1996; Laub et al., 2002). CtrA also binds directly to and silences the origin of replication in swarmer cells (Quon et al., 1998). During the swarmer-to-stalked cell transition, the combined action of CtrA dephosphorylation and proteolysis ensures that active CtrA is eliminated, thus enabling the onset of DNA replication and S phase (Domian et al., 1997) (Fig. 1A).
A complex circuit of regulatory proteins ensures the rapid elimination of CtrA specifically during the G1-S transition (Jenal, 2009; Tsokos and Laub, 2012). Because ClpXP is present throughout the cell cycle it was initially assumed that the protease was somehow activated to degrade CtrA at the appropriate time. However, CtrA proteolysis by ClpXP occurs in vitro at rates comparable to those observed in vivo (Chien et al., 2007). Thus, the timing of CtrA degradation in vivo may result from a relief of inhibition rather than being promoted by an activator or adaptor. Notably, in swarmer cells, immediately before being degraded, CtrA is bound to DNA, including target promoters and the origin of replication. Whether binding to DNA influences CtrA degradation is currently unclear.
Although CtrA is abundant and phosphorylated in both swarmer and predivisional cells, most CtrA-activated genes are expressed predominantly during the predivisional stage (Laub et al., 2002). These genes are not transcribed in swarmer cells because a small protein called SciP accumulates specifically in this cell type (Fig. 1A) and binds to CtrA to prevent it from activating target genes (Gora et al., 2010; Tan et al., 2010). SciP does not, however, prevent CtrA from binding DNA, so CtrA can still silence the origin of replication in swarmer cells. The elimination of SciP during the G1-S transition is likely critical to preparing cells for another round of CtrA-dependent gene expression later in the cell cycle. Consistently, cells forced to produce SciP at high levels in stalked and predivisional cells fail to activate CtrA target genes and consequently fail to divide or properly assemble polar organelles (Gora et al., 2010).
The mechanisms regulating SciP abundance during the cell cycle are largely uncharacterized. Here, we demonstrate that SciP is degraded by the AAA+ protease Lon both in vivo and in vitro, and that synthesis of a non-degradable version of SciP leads to severe cell cycle defects. Further, we find that formation of a SciP : CtrA : DNA complex helps shield SciP and CtrA from degradation by Lon and ClpXP, respectively, in swarmer cells. These results suggest that the removal of each component during the G1-S transition may help drive degradation of the other, thereby accelerating the onset of S phase. Our results further highlight the prominent role played by regulated proteolysis in driving cell differentiation processes and cell cycle transitions in bacteria.
SciP binds DNA non-specifically when complexed with CtrA
SciP does not strongly bind DNA on its own, as judged by electrophoretic mobility assays, but does form a robust complex with DNA-bound CtrA (Gora et al., 2010). Formation of this ternary complex requires DNA immediately adjacent to that bound by CtrA indicating that SciP likely contacts neighbouring DNA directly when complexed with CtrA. Similar results were obtained with three different DNA probes suggesting that the SciP-DNA interaction is likely not highly sequence specific. However, SciP was subsequently suggested to bind DNA independently through recognition of a motif with consensus TGTCGCG (Tan et al., 2010).
To test the functional significance of this motif, we examined its effect on SciP binding to a 50 bp fragment of the pilA promoter that binds SciP and CtrA together. We mutated the sequence GATCGCG, the closest match to the proposed SciP consensus, to GATTTTG (Fig. S1A). As expected, phosphorylated CtrA (CtrA∼P) alone bound both the wild-type and mutant probes, producing nearly identical shifts (Fig. S1B and C). We then added purified SciP along with CtrA∼P and observed a similar pattern of super-shifting with both probes (Fig. S1B), suggesting that the mutated sequence is not required for a SciP : CtrA : DNA complex to form on this probe. This conclusion is consistent with our previous report that SciP and CtrA form a complex on a 50 bp probe taken from the fliF promoter that does not include a close match to the proposed motif (Gora et al., 2010).
To corroborate these results in vivo, we examined a PctrA-lacZ reporter that contains the P1 and P2 promoters of ctrA (Domian et al., 1999) and two instances of the proposed SciP motif (Fig. S1D). Cells harbouring this reporter showed a significant decrease in LacZ activity when SciP was overproduced, even if the putative motif regions were mutated, individually or in combination (Fig. S1E). Thus, although SciP could have minor sequence preferences, it likely does not function through the binding of a specific motif. Instead, our findings support a model in which SciP binds both CtrA and the DNA in direct proximity to CtrA, with specificity of the ternary complex ultimately provided by CtrA. We do not, however, rule out that the proposed motif found in many CtrA-regulated genes (Tan et al., 2010) is recognized by another DNA-binding protein or by SciP in complex with a different partner under alternative conditions.
Binding to DNA and SciP stabilizes CtrA against proteolysis in vitro
Because our data indicated that SciP and CtrA form a direct, ternary complex with DNA, we wondered whether complex formation would affect the stability of each protein. CtrA is stable in swarmer cells and then rapidly degraded during the swarmer-to-stalked cell transition (Domian et al., 1997). We hypothesized that DNA binding may contribute to the stabilization of CtrA, as seen with other transcription factors (Pruteanu et al., 2007). We first tested the effect of DNA binding on CtrA stability in vitro using a 25 bp DNA fragment derived from the Caulobacter origin of replication (Cori). Whereas CtrA alone was degraded rapidly by ClpXP, the inclusion of 5 μM Cori DNA severely inhibited degradation (Fig. 2A). To assess inhibition quantitatively, we fused CtrA to GFP and measured initial rates of degradation by monitoring loss of GFP fluorescence. As expected, addition of the 25 bp Cori fragment inhibited GFP–CtrA degradation with a Ki of ∼ 550 nM (Fig. S2), in agreement with previously measured dissociation constants of CtrA and DNA (Siam and Marczynski, 2000).
Previous work demonstrated that overproducing SciP leads to a small, but significant increase in CtrA stability (Gora et al., 2010). Therefore, we tested whether SciP affects CtrA degradation in vitro. Adding SciP to CtrA in the absence of DNA had little effect on CtrA degradation (Fig. 2B). We then tested the effects of adding SciP along with a 50 bp fragment of the fliF promoter that supports formation of a CtrA : SciP : DNA complex (Gora et al., 2010). In this case we observed a dramatic stabilization of CtrA (Fig. 2B and C). Notably, these experiments were done at a concentration of DNA that is insufficient, in the absence of SciP, to completely inhibit CtrA degradation (Fig. 2B and C), indicating a synergistic effect of SciP and DNA on CtrA stability. SciP also helped to block the degradation of CtrA bound to a 50 bp pilA fragment. However, SciP did not significantly affect proteolysis of CtrA bound to a 25 bp fragment of Cori (Fig. 2C) consistent with previous results, and those above, indicating that formation of a CtrA : SciP : DNA complex requires interactions of both CtrA and SciP with DNA (Gora et al., 2010).
A point mutant of SciP, R35A, which weakens the CtrA-SciP interaction and abrogrates the ability of SciP to regulate CtrA in vivo (Gora et al., 2010), failed to stabilize CtrA against degradation in vitro when using the 50 bp fliF probe (Fig. 2D). We also tested another SciP mutant, R40A, which significantly reduces the growth defects arising from sciP overexpression, but is less deficient than SciP(R35A) in interacting with CtrA (Gora et al., 2010). With the 50 bp fliF probe, SciP(R40A) stabilized CtrA to a degree more similar to wild-type SciP than SciP(R35A) (Fig. 2D). However, with the pilA probe, which shows a more dramatic stabilization of CtrA degradation in the presence of SciP (Fig 2C), we found that SciP(R40A) was worse than wild-type SciP in protecting CtrA from ClpXP degradation (Fig. 2E). Collectively, these results are consistent with prior observations indicating that the R35A variant is almost completely deficient in binding CtrA in a yeast two-hybrid assay, while the R40A variant is compromised to a lesser extent (Gora et al., 2010). Importantly, we also found that the stabilization of CtrA by wild-type SciP requires the CtrA recognition site in the DNA as a pilA probe harbouring mutations in the CtrA binding sites failed to stabilize CtrA even in the presence of wild-type SciP (Fig. 2E). Taken together, our data indicate that SciP stabilizes the binding of CtrA to DNA, which, in turn, leads to the stabilization of CtrA against degradation by ClpXP.
Cell cycle changes in SciP abundance are controlled post-transcriptionally
Our in vitro results suggested that changes in SciP abundance may influence the degradation of CtrA in vivo, helping to stabilize CtrA in swarmer cells. Notably, SciP is maximally abundant in swarmer cells and then rapidly disappears during the G1-to-S transition, coincident with the time of CtrA degradation (Gora et al., 2010; Tan et al., 2010). We therefore sought to determine the molecular mechanisms controlling SciP accumulation.
The transcription of sciP is cell cycle-regulated with peak sciP mRNA levels observed in predivisional cells (Chen et al., 1986). To test whether regulated sciP transcription contributes to the swarmer-specific accumulation of SciP, we constructed a strain harbouring a deletion of the native sciP and constitutively expressing sciP from the chromosomal xylX locus. Synchronized swarmer cells from this strain were isolated and released into rich medium containing xylose. We found that SciP levels were highest in swarmer cells and then decreased ∼ 30 min post-synchronization, although SciP did not completely disappear, as it does in the wild type (Fig. 3A). Cells constitutively transcribing sciP also showed higher levels of SciP in predivisional cells than in the wild type.
Our results indicated that regulated sciP transcription has a modest effect on the cell cycle pattern of SciP abundance. However, SciP levels still dropped significantly during the G1-S transition suggesting that SciP may also be subject to regulated proteolysis. We therefore measured the stability of SciP using pulse-chase analysis of cells expressing sciP from a low-copy plasmid during growth in a minimal medium. Under these conditions, in which the cell cycle is ∼ 140 min, SciP had a half-life of only ∼ 23 min indicating it is a relatively unstable protein and likely being degraded during the G1-S transition when SciP levels drop (see Fig. 4A). Thus, we sought to identify the protease(s) responsible for SciP degradation.
SciP is degraded by the Lon protease
Because SciP normally disappears during the G1-to-S transition, concurrent with ClpXP-dependent proteolysis of CtrA, we first investigated whether ClpXP also degrades SciP. As both ClpP and ClpX are essential for viability, we assayed SciP levels in cells depleted of ClpP. Strain UJ199 harbours a deletion of the native, chromosomal clpP gene with a single copy of clpP under control of the Pxyl promoter at the xylX locus (Jenal and Fuchs, 1998). This strain was grown in rich medium containing glucose for 6 h (∼ four generations) to deplete ClpP. We then isolated swarmer cells and released them into glucose or xylose to repress or induce, respectively, clpP. We sampled cells every 30 min and examined both SciP and CtrA by immunoblotting (Fig. 3B). As expected, cells released into glucose failed to proteolyse CtrA, whereas cells released into xylose had almost completely eliminated CtrA after ∼ 60 min. In contrast to CtrA, SciP levels dropped significantly in both the xylose and glucose samples by ∼ 60 min, although levels remained slightly higher in glucose (Fig. 3B). These data indicate that cells lacking ClpP can still proteolyse SciP during the G1-to-S transition. These data are also consistent with our results showing that purified ClpXP degrades CtrA three times faster than SciP when both are present at the same concentration in vitro (Fig. S3).
Because ClpP is the peptidase subunit for both ClpXP and the protease ClpAP, our data argue against the involvement of either protease in SciP proteolysis. Consistently, we found that SciP was also eliminated during the G1-to-S transition in synchronized ΔclpA cells (Fig. 3C).
To test the role of HslUV, another ATP-dependent protease (Rohrwild et al., 1996), we synchronized a strain harbouring a transposon within the hslV coding region, hslV::Tn5, and monitored SciP levels by Western blotting. SciP was completely eliminated during the G1-S transition, indicating that HslUV likely does not degrade SciP (Fig. 3C).
We next measured SciP abundance in cells lacking the protease Lon (Wright et al., 1996). We synchronized Δlon cells and monitored both SciP and CtrA levels during cell cycle progression. Whereas CtrA was cleared by 30 min post-synchronization, SciP was still present after 30 min and maintained throughout the cell cycle (Fig. 3C). Further, pulse-chase analyses on mixed populations of cells indicated that SciP was significantly stabilized in Δlon cells compared with the wild type (Fig. 4A). Taken together, these data strongly suggest that SciP is degraded by Lon in vivo.
We then tested the ability of C. crescentus Lon to proteolyse SciP in vitro. Purified SciP was rapidly degraded by Lon in vitro, with a half-life of ∼ 12 min (Fig. 4B). These findings support the conclusion that SciP is a direct substrate for Lon in Caulobacter. Although SciP was stabilized most significantly during the G1-S transition in Δlon cells, levels still decreased modestly during cell cycle progression in this strain (Fig. 3C), indicating that other proteases may degrade SciP, but at lower rates.
A C-terminal tag interferes with SciP proteolysis
To test the importance of regulated SciP degradation to the pattern of SciP abundance and cell cycle progression, we sought to create a proteolytically stable version of SciP. Because energy-dependent proteases often rely on terminal sequences for substrate recognition (Domian et al., 1997; Griffith et al., 2004; Shah and Wolf, 2006; Chien et al., 2007; Rood et al., 2012), we tagged the N- and C-terminus of SciP with an M2 epitope. We transformed wild-type cells with low copy plasmids expressing sciP, M2-sciP, or sciP-M2 under the control of Pxyl. Mixed populations of cells were grown in rich medium supplemented with glucose to repress the plasmid copy of sciP, synchronized, and released into medium supplemented with xylose to induce expression. Samples were collected every 15 min for SciP immunoblot analysis (Fig. 4C). Note that we did not induce expression of the plasmid-borne copies of sciP prior to synchronization, as pre-induction decreased the yield of swarmer cells for the sciP-M2 strain during synchronization; hence, SciP levels during early time points reflect the effects of transcriptional induction and proteolysis.
For cells expressing untagged sciP from a low-copy plasmid, along with the chromosomal sciP, we found that SciP levels were highest in swarmer cells, dropped dramatically by ∼ 30–45 min post-synchronization, and then accumulated again during later stages of the cell cycle (Fig. 4C), a pattern similar to that seen when the only copy of sciP is expressed from the chromosomal xylX locus (Fig. 3A). By contrast, for cells producing N-terminally tagged SciP, we found that SciP accumulated during the early stages of the cell cycle with slightly higher levels of SciP at the 30 and 45 min time points compared with cells producing untagged SciP (Fig. 4C). A similar, but more pronounced effect was seen in cells expressing sciP-M2, with high levels of SciP-M2 detectable throughout the time-course examined (Fig. 4C), compared with cells expressing untagged sciP, and particularly when compared with wild-type cells. Our results indicate that a C-terminal M2 epitope likely stabilizes SciP by interfering with its proteolysis by Lon.
We assayed the in vivo stability of epitope-tagged SciP using pulse-chase analyses of mixed populations of cells grown in minimal medium. For cells producing M2-SciP or SciP-M2, we measured half-lives of 36 and 58 min, respectively, compared with 23 min for the untagged protein (Fig. 4A). We also found that the degradation of SciP-M2 by Lon in vitro was dramatically slowed compared with degradation of the wild-type SciP (Fig. 4B). The C-terminal tag likely did not stabilize SciP by interfering with its function as purified SciP-M2 was still capable of complexing with CtrA in vitro (Fig. S3) and SciP-M2 could still downregulate CtrA-dependent genes in vivo (see below). Together, these results support the conclusion that Lon recognizes SciP through an exposed C-terminus.
SciP and CtrA reciprocally stabilize each other
The formation of a CtrA : SciP : DNA complex can protect CtrA from degradation by ClpXP in vitro (Fig. 2), and overproducing SciP stabilizes CtrA in vivo (Gora et al., 2010). We hypothesized that SciP may similarly be protected from Lon-dependent degradation by interacting with CtrA at target promoters. Indeed, we found that SciP degradation in vitro was significantly slower in the presence of CtrA and a 50 bp fliF fragment ( Fig. 5A and B). This stabilization was dependent on complex formation as excess DNA or CtrA alone did not alter SciP degradation by Lon. Moreover, mutants of SciP that do not interact with CtrA were susceptible to Lon degradation even when both CtrA and DNA were included (Fig. 5C). We also found that SciP was not eliminated as rapidly during the G1-S transition in ΔcpdR cells in which CtrA remains stable throughout the cell cycle (Biondi et al., 2006; Iniesta et al., 2006); SciP was still abundant 30 min post-synchronization in ΔcpdR cells but not in wild-type cells (Fig. 3A and C). This result is consistent with a model in which CtrA normally stabilizes SciP; consequently, ΔcpdR cells, which do not properly clear CtrA during the G1-S transition, do not clear SciP as efficiently. Taken together, these findings suggest that SciP and CtrA reciprocally stabilize each other against degradation by Lon and ClpXP, respectively, when assembled into a DNA-bound complex. Conversely, proteolysis of each may accelerate degradation of the other during the G1-S transition, contributing to the robustness and irreversibility of this cell cycle transition.
Regulated proteolysis of SciP is necessary for proper cell cycle progression
To further investigate the phenotypic consequences of producing protease-insensitive SciP, we expressed sciP-M2 from a low-copy plasmid under the control of a xylose-inducible promoter in otherwise wild-type cells. This strain was grown in rich medium supplemented with glucose and then shifted to medium containing xylose, leading to a modest growth defect relative to a strain maintained in glucose (Fig. 6A). After 4 h, cells were also moderately filamentous and flow cytometry analysis indicated that cells had accumulated up to four chromosomes per cell, consistent with a defect in cell division (Fig. 6B). In contrast to sciP-M2, cells expressing untagged sciP or M2-sciP from the same low-copy plasmid exhibited no significant defects in growth, morphology, or chromosome content (data not shown).
The phenotypic consequences of producing SciP-M2 from a low-copy plasmid are reminiscent of those previously reported for cells producing M2-SciP from the same promoter but on a high-copy plasmid (Gora et al., 2010). To further compare the phenotypes of these two strains, we examined global changes in gene expression in each strain using DNA microarrays. Wild-type cells harbouring the low-copy sciP-M2 expression plasmid were induced with xylose for 4 h and mRNA was then isolated and compared with mRNA from identically treated wild-type cells using whole genome microarrays. We directly compared this expression profile to those obtained previously for cells expressing M2-sciP from a high-copy plasmid, and for ctrAts and cckAts strains grown at a restrictive temperature (Gora et al., 2010). For all four strains, we observed a significant downregulation of CtrA-activated genes (Fig. 6C).
Based on the DNA microarrays, we found that the expression of sciP-M2 from a low-copy vector led to ∼ threefold higher sciP mRNA relative to wild-type cells, whereas expression from a high-copy vector resulted in an ∼ 18-fold increase. However, the downregulation of CtrA target genes was comparable in magnitude in each case (Fig. 6C), consistent with our conclusions that a C-terminal M2 tag stabilizes SciP. These data support a model in which SciP is normally proteolysed at the G1-S transition and that a failure to degrade SciP during this transition (Fig. 4C) prevents the subsequent activation of CtrA-dependent genes in predivisional cells, thereby disrupting cell cycle progression.
Cell cycle regulation of SciP
SciP plays a critical role during the Caulobacter cell cycle by helping to restrict CtrA-dependent gene expression to the G1, or swarmer cell, phase (Gora et al., 2010; Tan et al., 2010). SciP is not, however, essential for viability as cells harbouring a deletion of sciP can be generated (Gora et al., 2010); moreover, resequencing of the genome of the ΔsciP strain did not identify any additional mutations relative to the parent strain CB15N (data not shown). Nevertheless, cells lacking SciP show a range of severe phenotypes, underscoring its importance as a cell cycle regulator. SciP is maximally abundant in swarmer cells and must be eliminated before cells reach the predivisional stage when CtrA is required to activate an arsenal of genes critical to morphogenesis and cell division (Laub et al., 2002). Regulated transcription and proteolysis work together to ensure SciP is abundant principally in swarmer cells. The transcription of sciP is cell cycle-regulated and directly activated by CtrA leading to maximal expression of sciP in predivisional cells (Chen et al., 1986; Gora et al., 2010; Tan et al., 2010). SciP protein does not, however, accumulate to high levels until after cell division, and then accumulates almost exclusively in the daughter swarmer cell, indicating post-transcriptional control of SciP (Gora et al., 2010). Moreover, cells constitutively expressing sciP still show cell cycle-dependent changes in SciP protein levels (Fig. 4C), suggesting other levels of regulation. Here, we demonstrated that SciP is likely rapidly degraded during the swarmer-to-stalked cell, or G1-S, transition, by the protease Lon (Fig. 7). Proteolysis allows cells to irreversibly clear SciP, priming cells for a new round of CtrA-dependent gene expression following DNA replication initiation.
Regulated proteolysis by Lon
SciP is the first direct substrate of Lon identified in Caulobacter. Genetic studies have previously implicated Lon in the degradation of the cell-cycle methyltransferase CcrM (Wright et al., 1996), but whether Lon directly degrades CcrM has not been established. Nevertheless, the Lon protease clearly plays an important role in the degradation of both CcrM and SciP, and probably other cell cycle regulators. Consistent with such a multifaceted role, cells lacking Lon exhibit a variety of cell cycle phenotypes, including moderate cellular filamentation and the accumulation of up to six chromosomes per cell, as judged by flow cytometry (Wright et al., 1996). These phenotypes are similar to those resulting from the expression of non-degradable sciP-M2 in wild-type cells suggesting that a failure to properly clear SciP at the G1-to-S transition may contribute to some aspects of the pleiotropic effects of a lon mutant.
How is Lon activity regulated? Lon is present at relatively constant levels throughout the cell cycle suggesting that its activity is not regulated through changes in abundance (Wright et al., 1996). Hence, in the case of CcrM, Lon was suggested to be constitutively active throughout the cell cycle, with an increase in ccrM transcription in predivisional cells leading to an accumulation of CcrM by exceeding the proteolytic capacity of Lon. In contrast, we found that SciP levels drop rapidly during the G1-S transition when there is no concurrent change in sciP transcription, suggesting that the specific activity of Lon could increase during this stage of the cell cycle. Polyphosphate can stimulate Lon activity in Escherichia coli (Kuroda et al., 2001), and could function similarly in Caulobacter, although a recent study suggested that polyphosphate actually slows the G1-S transition in Caulobacter (Boutte et al., 2012). However, another cofactor could, in principle, accumulate during the G1-S transition to activate Lon and stimulate SciP degradation. Alternatively, although Lon recognizes SciP directly in vitro, an adapter protein could accumulate during the G1-S transition in vivo that further improves degradation. The binding of SciP to CtrA and target promoters may also help shield SciP from degradation in swarmer cells (Fig. 5). Finally, it is also possible that differential localization of Lon during the cell cycle may contribute to changes in its activity, as suggested for ClpXP (Iniesta et al., 2006).
Lon is a AAA+ protease that is widely conserved throughout the bacterial kingdom. In E. coli, Lon is responsible for degrading unfolded proteins as well as several key regulators, including the transcription factor SoxS and the cell division inhibitor SulA (Mizusawa and Gottesman, 1983; Griffith et al., 2004; Melderen and Aertsen, 2009). Although the rules for Lon substrate specificity are still poorly defined, Lon is thought to recognize hydrophobic patches in unfolded proteins and the N- or C-terminus of the substrates SoxS and SulA respectively (Ishii et al., 2000; Shah and Wolf, 2006; Gur and Sauer, 2008). The stabilization of SciP by the addition of a C-terminal M2 epitope (DYKDDDDK) strongly suggests that the free C-terminus of SciP (RTTRIQQYR) is recognized by Lon in Caulobacter. Alternatively, addition of the M2 epitope may block degradation by masking electrostatic interactions with the C-terminus that are needed for recognition of SciP by Lon, or by forcing the C-terminus to adopt a non-native secondary structure.
Disassembly of the CtrA–SciP–DNA complex during the G1-S transition
SciP interacts directly with CtrA bound at target promoters (Fig. 7). Although SciP has a recognition helix, it binds DNA only weakly on its own, requiring CtrA to form more stable contacts with DNA. Thus, SciP likely does not recognize and bind specific sites on the chromosome like canonical transcription factors do, and instead is mainly recruited to the promoters of CtrA-regulated genes through its direct interaction with CtrA. In this sense, SciP functions as a cofactor, or regulator, of CtrA.
The formation of a complex between SciP and CtrA is critical for downregulating CtrA-activated genes in swarmer cells. This complex also promotes the stability of both proteins by protecting each from degradation by their respective protease (Figs 2 and 5). Their mutually enforced stabilization likely means that the degradation of either protein will facilitate degradation of the other (Fig. 7). Hence, the complex may be rapidly disassembled and degraded in a more switch-like fashion. Which protein gets degraded first is not yet clear. Notably though, SciP did not disappear as rapidly, or to the same extent, in ΔcpdR cells in which CtrA is not properly degraded at the G1-S transition (Fig. 3). Having SciP degradation contingent on the degradation of CtrA could be beneficial as it would help prevent a scenario where the premature removal of SciP allows CtrA to inappropriately activate gene expression in swarmer cells.
Mutual protection from degradation will, of course, only apply to CtrA and SciP proteins that are bound in a complex on DNA. The binding of CtrA and SciP is strongly promoted by their binding to DNA that contains a CtrA binding site (Fig. 2E). However, CtrA and SciP may also bind at other sites in the chromosome, even transiently, and such ternary complexes could also protect both proteins from degradation. Nevertheless, given their high concentrations in swarmer cells, not all SciP and CtrA will be complexed, and the unassociated proteins would remain susceptible to proteases. However, additional mechanisms exist that ensure CtrA is rapidly degraded only during the G1-S transition (Abel et al., 2011), and similar mechanisms could exist for SciP as well.
In sum, our results indicate that Caulobacter cells rely heavily on regulated proteolysis of a transcription factor complex to drive the G1-S transition. The degradation of CtrA by ClpXP enables a new round of DNA replication to initiate, while the degradation of SciP by Lon ultimately primes the cell for CtrA-dependent gene expression following replication initiation and the reaccumulation of CtrA.
Bacterial growth conditions
Strains used are listed in Table S1. Caulobacter strains were grown in rich (PYE) or minimal (M2G) media and supplemented with 0.2% glucose or 0.3% xylose, as indicated. ML1750 was grown in PYE supplemented with xylose, synchronized, and released into PYE supplemented with glucose or xylose. ML1990, ML1991, and ML1992 were grown in PYE supplemented with glucose and oxytetracycline, synchronized, and released into PYE with glucose or xylose, as indicated, and oxytetracyline. UJ838, UJ199, LS2382, ΔcpdR, and hslV::tn5 were grown in PYE. UJ199 was depleted and synchronized as previously reported (Jenal and Fuchs, 1998). All synchronies were performed in a 50% Percoll gradient in 50 ml Falcon tubes.
Cloning, mutagenesis, and strain construction
Primers used for strain construction are listed in Table S2. To construct the plasmid pENTR-Pxyl-sciP the Pxyl promoter was amplified from pHXM-DEST using the primers Pxyl_F and Pxyl_R, sciP was amplified from genomic DNA using the primers sciP_F and sciP_R, and the two pieces were fused together by splice overlap extension PCR. The fused construct was cloned into pENTR through TOPO cloning (Invitrogen). The plasmid pENTR-Pxyl-sciP-M2 was constructed by TOPO cloning a PCR fragment amplified from pLX-sciP using the primers Pxyl_F and sciP_M2_R. The low-copy plasmids with Pxyl-sciP (pLX-sciP) and Pxyl-sciP-M2 (pLX-sciP-M2) were constructed by LR recombination of pENTR-Pxyl-sciP and pENTR-Pxyl-sciP-M2 into the destination vector pMR20-DEST as described previously (Skerker et al., 2005). The low-copy plasmid with Pxyl-M2-sciP (pLXM-sciP) was constructed by LR recombination of pENTR-sciP into pLXM-DEST. Expression plasmids were electroporated into CB15N and Δlon electrocompetent cells.
The pBAD-lon plasmid was constructed by digesting pBAD33-lon with NdeI and SbfI and ligating in a PCR product containing C. crescentus lon amplified from CB15N genomic DNA using the primers lon_F and lon_R and digested with NdeI and SbfI.
Constructs for production of SciP variants were constructed by LR recombination of pENTR-sciP variants into pHIS-DEST (Skerker et al., 2005). A His6-tagged GFP–CtrA (pHIS-GFP-CtrA) was constructed by amplifying ctrA from genomic DNA using primers (CtrAforRI and CtrArevXhoI) to install EcoRI and XhoI sites flanking ctrA. After purification and restriction, the insert was used to replace pdeA in vector pENTR-GFP-PdeA (Rood et al., 2012) generating pENTR-GFP-CtrA. An expression construct was generated by LR recombination of pENTR-GFP-CtrA into pHIS-DEST (Skerker et al., 2005).
Plasmids harbouring the ctrA promoter were derived from pctrA290 (Domian et al., 1999). Mutations were introduced using the primers listed in Table S2 to create pctrA290-mut1, pctrA290-mut2, pctrA290-mut1 + mut2. Plasmids were transformed into ML1748 to create ML1994-ML1997.
DNA templates containing CtrA binding sites were generated by mixing equal concentrations of oligonucleotides derived from the origin of replication or from promoters of CtrA-regulated genes. Pairs of complementary oligonucleotides, listed in Table S2, were heated at 95°C for 10 min in 100 mM KCl and cooled slowly to facilitate annealing.
Cell pellets were resuspended in the equivalent of 50 μl 1× SDS sample buffer per 1 ml of OD 0.15 culture. Ten microlitre of lysate was resolved on a 12% gel (Bio-Rad) run at 130 V for 1 h at room temperature. Samples were transferred to PVDF (polyvinylidene difluoride) membrane and probed as described previously (Gora et al., 2010).
Samples were analysed as described previously (Chen et al., 2009).
SciP stability was assayed as previously described (Gora et al., 2010) with the following modifications. The various sciP alleles in strains ML1990, ML1991, ML1992 and ML1993 were induced through the addition of xylose for 30 min immediately before the pulse-chase. Samples were immunoprecipitated using 2 μl of SciP anti-sera for 2 h at 4°C.
DNA microarray analysis
Microarray analyses were conducted as previously reported using custom Agilent oligo arrays (Gora et al., 2010). Expression experiments were performed in duplicate and results averaged.
For purifications of SciP from E. coli, overnight cultures were diluted into Luria Broth with ampicillin (100 μg ml−1), grown for 2–3 h at 37°C, and then induced with 0.4 mM IPTG. After 3 h, cells were spun down at 5000 r.p.m. for 15 min. Pellets were resuspended in lysis buffer (50 mM Tris pH 8.0, 300 mM NaCl, 10 mM imidazole, 10% glycerol, 5 mM β-mercaptoethanol) and frozen at −20°C. Cells were then thawed on ice before being lysed by high pressure disruption using a Microfluidizer (Microfluidics, Newton, MA). Following clarification by centrifugation at 15 000 g, lysate was applied to a 1 ml Ni-NTA column pre-equilibrated with lysis buffer at room temperature. The column was washed with 20 vols of lysis buffer, followed by elution with 6 column volumes of elution buffer (50 mM Tris pH 8.0, 300 mM NaCl, 300 mM imidazole, 10% glycerol, and 5 mM β-mercaptoethanol). Purity was assessed by SDS-PAGE/Coomassie staining and samples containing the desired protein were pooled, concentrated, and aliquoted before being frozen at −20°C. ClpX, ClpP and CtrA were purified as previously described (Chien et al., 2007; Rood et al., 2012). Caulobacter Lon was expressed from plasmid pBAD-lon in a Δlon E. coli strain (ER2566) and purified following the protocol for purification of E. coli Lon (Gur and Sauer, 2009).
In vitro degradation assays
Protein degradation was monitored using SDS-PAGE as before (Rood et al., 2012). For ClpXP degradation reactions, the concentrations were as follows, unless otherwise noted: 0.2 μM ClpX6, 0.4 μM ClpP14, 4 mM ATP, 75 μg ml−1 creatine kinase, 5 mM creatine phosphate in H-buffer (25 mM HEPES-KOH pH 7.4, 100 mM KCl, 10% glycerol, 10 mM MgCl2, 10 mM β-mercaptoethanol). For Lon degradation reactions, concentrations of the components were the same as above, but with 0.5 μM Lon hexamer as the protease and Lon degradation buffer (25 mM Tris pH 8.0, 100 mM KCl, 10 mM MgCl2, 1 mM DTT). Proteolysis was also monitored by fluorescence, as previously described (Rood et al., 2012). Degradation of a GFP–CtrA fusion was measured via loss of fluorescence over time at 30°C. Final concentrations in these assays were: 0.2 μM ClpX6, 0.4 μM ClpP14, 4 mM ATP, 75 μg ml−1 creatine kinase, 5 mM creatine phosphate, 50 nM GFP–CtrA, and 1 μM SciP. Concentrations of DNA were varied over the ranges reported in the figures. Typical reaction volumes were 20 μl and were monitored with a Spectramax M5 (Molecular Devices) plate reader with 440 nm/510 nm excitation/emission settings and an emission cut-off set at 495 nm.
We thank J. Modell for the hslV::Tn5 strain and for assistance in whole genome sequencing, and U. Jenal for providing several strains. M. T. L. is an Early Career Scientist at the Howard Hughes Medical Institute. This work was supported by an NSF Graduate Fellowship to M. W. and NIH Grants 5R01GM082899 to M. T. L. and 5R00GM084157 to P. C.