|Accueil | Aperçu | Revues | Soumettre | Nous Contacter | English|
In Caulobacter crescentus, progression through the cell cycle is governed by the periodic activation and inactivation of the master regulator CtrA. Two phosphorelays, each initiating with the histidine kinase CckA, promote CtrA activation by driving its phosphorylation and by inactivating its proteolysis. Here, we examined whether the CckA phosphorelays also influence the downregulation of CtrA. We demonstrate that CckA is bifunctional, capable of acting as either a kinase or phosphatase to drive the activation or inactivation, respectively, of CtrA. By identifying mutations that uncouple these two activities, we show that CckA's phosphatase activity is important for downregulating CtrA prior to DNA replication initiation in vivo but that other phosphatases may exist. Our results demonstrate that cell cycle transitions in Caulobacter require and are likely driven by the toggling of CckA between its kinase and phosphatase states. More generally, our results emphasize how the bifunctional nature of histidine kinases can help switch cells between mutually exclusive states.
Caulobacter crescentus is a tractable model system for understanding the molecular mechanisms underlying cell cycle progression and the establishment of cellular asymmetry in bacteria. Each cell division for Caulobacter produces two morphologically different daughter cells, a swarmer cell and a stalked cell, which also differ in their ability to initiate DNA replication. A stalked cell can immediately initiate DNA replication following cell division, whereas a swarmer cell cannot initiate until after differentiating into a stalked cell. The swarmer-to-stalked cell transition thus coincides with a G1-S cell cycle transition. DNA replication occurs once and only once per cell cycle, resulting in distinguishable G1, S, and G2 phases.
Progression through the Caulobacter cell cycle requires the precise temporal and spatial coordination of both morphological and cell cycle events. Previous genetic screens have uncovered numerous two-component signal transduction genes that help to regulate these events (10, 11, 17, 24, 25, 29, 33, 34, 42). Two-component signaling pathways are typically comprised of a sensor histidine kinase that, upon activation, autophosphorylates and subsequently transfers its phosphoryl group to a cognate response regulator, which can then effect changes in cellular physiology (35). One common variation of this signaling paradigm is called a phosphorelay (3). Such pathways also initiate with the autophosphorylation of a histidine kinase and subsequent phosphotransfer to a response regulator, but these steps often occur intramolecularly within a hybrid histidine kinase. The phosphoryl group on the receiver domain of a hybrid kinase is then passed to a histidine phosphotransferase, which subsequently phosphorylates a soluble response regulator to effect an output response. Relative to canonical two-component pathways, phosphorelays provide additional points of control and enable signal integration; they are often involved in regulating key cell fate decisions in processes such as sporulation, cell cycle transitions, and quorum sensing (1, 3, 8).
The master regulator of the Caulobacter cell cycle is CtrA, an essential response regulator that directly activates the expression of at least 70 genes (19, 29). CtrA also regulates DNA replication by binding to and silencing the origin of replication (30). Progression through the Caulobacter cell cycle thus requires the precise control of CtrA activity. CtrA must be abundant and active thoughout most of the cell cycle to drive gene expression and to silence the origin but must be temporarily inactivated in stalked cells prior to S phase to permit the initiation of DNA replication (see Fig. Fig.88).
CtrA is regulated on at least three levels: transcription, proteolysis, and phosphorylation (4, 5). During G1, CtrA is phosphorylated and proteolytically stable. At the G1-S transition, CtrA is dephosphorylated and degraded, thereby freeing the origin of replication to fire. After DNA replication initiates, ctrA is transcribed and the newly synthesized CtrA is again phosphorylated and protected from proteolysis. Following septation of the predivisional cell, CtrA remains phosphorylated and stable in the swarmer cell but is dephosphorylated and degraded in the stalked cell to permit DNA replication initiation. Cells that constitutively transcribe ctrA are viable and display only a mild phenotype, indicating that regulated phosphorylation and proteolysis alone can ensure the periodicity of CtrA activity (4). Cells producing nondegradable, constitutively active CtrA arrest in G1 because CtrA activity cannot be eliminated (4).
The regulation of CtrA activity involves two phosphorelays. Each initiates with CckA, a hybrid histidine kinase, and ChpT, a histidine phosphotransferase. After receiving a phosphoryl group from CckA, ChpT can act as the phosphodonor for either CtrA or the single-domain response regulator CpdR (1). Phosphorylation of CpdR prevents it from triggering CtrA proteolysis (1, 14). Unphosphorylated CpdR triggers CtrA degradation by somehow influencing the polar localization of the protease ClpXP (14), although why the protease must be localized is unclear.
The downregulation of CtrA prior to DNA replication involves the dephosphorylation of CtrA and CpdR such that CtrA is both dephosphorylated and, ultimately, degraded. These events coincide with the time in the cell cycle when CckA's kinase activity is lowest (16). As the phosphoryl groups on phosphorylated CtrA (CtrA~P) and CpdR~P are relatively stable, at least in vitro (1), phosphatases are likely critical to eliminating CtrA activity prior to S phase. For some phosphorelays, inactivation of the top-level kinase leads to a siphoning of phosphoryl groups from the terminal regulator back to the hybrid kinase's receiver domain. The bifunctional hybrid kinase may then act as a phosphatase, stimulating hydrolysis and loss of the phosphoryl group (7, 8). For other phosphorelays, there are separate and dedicated phosphatases (23, 26, 27).
Here, we demonstrate that CckA is bifunctional and can act as both a kinase and a phosphatase, such that inactivation of CckA as a kinase stimulates the dephosphorylation of CtrA~P and CpdR~P. We provide evidence that CckA's phosphatase activity contributes to the downregulation of CtrA in vivo but that other phosphatases may exist. Our results indicate that the periodic toggling of CckA between kinase and phosphatase states is crucial to cell cycle progression in Caulobacter.
Escherichia coli and C. crescentus strains were grown as described previously (33). Strains, plasmids, and primers used in this study are listed in Table S1 in the supplemental material. All plasmids were introduced into C. crescentus by electroporation. PCR amplification of genes and promoters from CB15N genomic DNA was done under previously described conditions (33). For Gateway-based cloning, PCR amplicons of CB15N genes (primer sequences are listed in Table S1 in the supplemental material) were first cloned into the pENTR/D-TOPO vector according to the manufacturer's protocol and sequence verified with M13F and M13R primers or primers within the gene. All site-directed mutagenesis was performed using the following PCR mixture: 75 ng pENTR clone, 50 μM of each deoxynucleoside triphosphate, 100 nM of each primer, 1× Pfu Turbo buffer, 1.25 U Pfu Turbo polymerase (Strategene), 2% dimethyl sulfoxide, and 60 mM betaine. For each reaction, 17 cycles of the following sequence were run: 94°C for 1 min, 55°C for 1 min, and 68°C for 15 min when using pENTR clones or 68°C for 45 min when using other plasmids as templates. pENTR clones were then recombined into destination vectors following the manufacturer's protocols (Invitrogen, Carlsbad, CA).
To construct strain ML1054, chpT was amplified from the chromosome using primers alt_ChpT_fw and ChpT_rev to create pENTR:chpT. This pENTR clone was recombined into the destination vector pLXM-DEST and then transformed into the wild type followed by transduction of a marked chpT deletion from ML780.
To construct strains ML1491 to ML1499, a pENTR clone of the cckA gene (pENTR:PcckA-cckA), including 158 bp upstream of the translational start that presumably encompasses the cckA promoter, was amplified from CB15N genomic DNA using primers PcckA-cckA-fw and PcckA-cckA-rev. This pENTR clone was recombined into the destination vector pMR20-DEST to produce a low-copy-number plasmid harboring a full-length copy of cckA under the control of its native promoter (pMR20-PcckA-cckA). The plasmid pMR20-PcckA-cckA was then transformed into CB15N, followed by Cr30-based transduction of a gentamicin-marked cckA deletion from strain LS3382 (16). To generate cckA point mutants, site-directed mutagenesis was performed on pENTR:PcckA-cckA using primers listed in Table S1 in the supplemental material. These pENTR plasmids were sequence verified and then recombined into the pMR20 destination vector prior to transformation and transduction of the marked cckA deletion.
Strains expressing mutant or wild-type cckA and overexpressing mutant ctrA (ML1567, ML1571, ML1572, ML1576, ML1578, ML1583, ML1585, and ML1587) were made by transforming ML1491 and ML1497 with the following plasmids: pJS14, pJS14-Pxyl-ctrA, pJS14-Pxyl-ctrA(D51E), pJS14-Pxyl-ctrAΔ3Ω, and pJS14-Pxyl-ctrA(D51E)Δ3Ω (4).
To construct strain ML1073, full-length cckA was amplified from CB15N genomic DNA with forward primer CckA_full_fw, which adds an NdeI site at the 5′ end of the gene, and reverse primer CckA_full_rev, which adds a SalI site at the 3′ end. Both pML83 and the PCR product containing cckA were digested with NdeI and SalI and ligated to form plasmid pML83-Pxyl-cckA, which was then electroporated into a pleC::Tn5 strain (41). ML1709 was constructed similarly, but pML83-Pxyl-cckA was modified by site-directed mutagenesis PCR with primers V366P_fw and V366P_rev before being electroporated into the pleC::Tn5 strain.
To construct a strain overexpressing full-length cckA (ML1688), we first made pENTR:Pxyl-cckA by using primers Pxyl_fw and CckA_full_rev to amplify a fragment containing a xylose-inducible promoter and cckA from the plasmid pML83:Pxyl-cckA. This pENTR clone was then recombined into the destination vector pJS14-DEST. To construct a strain overexpressing full-length cckA(H322A) (ML1738), we used primers H322A_fw and H322A_rev for site-directed mutagenesis of the plasmid pJS14:Pxyl-cckA from ML1688.
To construct strains overexpressing pieces of cckA containing no transmembrane domain (ML1689) or only the receiver domain (ML1692), we generated PCR products from CB15N genomic DNA using the following primers: HK7_fw and RR53_rev (ML1689) or RR53_fw and RR53_rev (ML1692). pENTR clones containing these PCR fragments were then recombined into pHXM2-DEST using the Gateway cloning method. To construct ML1690 and ML1691, we performed site-directed mutagenesis of pENTR:cckA-HK-RD with primers D623A_fw and D623A_rev (ML1690) or V366P_fw and V366P_rev (ML1691) before recombination into pHXM2-DEST.
To construct strain ML1681, the last 519 codons (without the stop codon) of cckA were amplified by PCR with primers CckA_GFP_fw and CckA_GFP_rev. The reverse primer removed the stop codon, added two nucleotides to keep it in frame with the downstream GFP fusion, and contained an EcoRI site. The forward primer contained a KpnI site. The cckA PCR product was cloned in frame with the egfp gene in pGFP-c4 (37), using the KpnI and EcoRI restriction sites. The coding region was sequence verified, and the plasmid was recombined into CB15N by electroporation to generate chromosomally encoded CckA-GFP.
To create pJS14-DEST and pMR20-DEST for Gateway cloning, the RfA Gateway cassette was blunt cloned into an EcoRV site in pJS14 and pMR20. To create pHXM2-DEST, the SacI-KpnI fragment containing a xylose-inducible promoter and M2 tag (Pxyl-M2) was digested out of pHXM-DEST and then cloned into pJS14.
Differential interference contrast microscopy was performed on mid-exponential-phase cells after they were fixed in phosphate-buffered saline with 0.5% paraformaldehyde.
Mixed populations of wild-type cells grown in M2G were synchronized using Percoll density centrifugation as previously described (1). Cell samples were taken every 20 min for 140 min, resolved on a 12% sodium dodecyl sulfate-polyacrylamide gel, transferred to polyvinylidene difluoride transfer membrane (Pierce), and probed with anti-ChpT serum at a 1:10,000 dilution. Polyclonal rabbit antisera (Covance) were generated using His6-ChpT and His6-CtrA.
Single colonies were inoculated into 5- to 10-ml liquid cultures from plates and grown overnight at 30°C under appropriate antibiotic selection but were always maintained at an optical density at 600 nm (OD600) of less than 0.7. Cultures were then diluted to an OD600 of 0.005 to 0.01 and grown to an OD600 of ~0.2 to 0.4 before processing. All strains were grown in peptone-yeast extract (PYE) medium, except for strains overexpressing ctrA alleles (see Fig. Fig.6),6), which were grown in M2G. Strains overexpressing cckA or ctrA were induced by the addition of 0.3% xylose to the culture medium or maintained in 0.2% glucose and processed after 4 or 8 h. After 8 h of induction, rifampin (rifampicin) (20 μg/ml) was added to strains overexpressing ctrA, which were grown for 3 more hours to allow for completion of DNA replication. Cells were fixed in 70% ethyl alcohol overnight at 4°C and stored at 4°C for up to a week. They were spun at 6,000 rpm for 4 min, resuspended in 1 ml 50 mM sodium citrate, and incubated for 4 h at 50°C with 2 μg/ml RNase to allow complete RNA digestion. After digestion, cells were incubated in 2.5 μM Sytox green nucleic acid stain (Invitrogen) for 15 min at room temperature before being analyzed by flow cytometry using an Epics C analyzer (Beckman-Coulter). For quantification of the flow cytometry data shown in Fig. Fig.6,6, we gated 1N DNA content peaks, using the same gate for all samples. The percentages shown in the bar graph were obtained by dividing the gated number of cells with 1N DNA content by the total number of cells, which were gated to exclude cellular debris on the far left of the flow cytometry profiles.
All protein purifications were done as reported previously using either a maltose-binding protein (MBP)-His6 or thioredoxin (TRX)-His6 tag (33). Primers CC3470_HPT_for and ChpT_Hbox_rev were used to amplify the H-box-containing N terminus of ChpT for constructing the plasmid pENTR:chpTΔC. Primers phoR_fw and phoR_rev were used to amplify the last 415 amino acids of PhoR (CC0289) for constructing the plasmid pENTR:phoR. Primers EnvZ_T247R_fw and EnvZ_T247R_rev were used for site-directed mutagenesis of the plasmid pENTR:envZ to create the plasmid pENTR:envZ(T247R). The creation of other pENTR clones for protein purification has been described previously (33).
First, 10 μM TRX-His6-CtrA was incubated with 0.2 μM PhoR with 5 μCi [γ-32P]ATP (~6,000 Ci/mmol, Amersham Biosciences) in storage buffer supplemented with 2 mM dithiothreitol and 5 mM MgCl2. Reaction mixtures were incubated at 30°C for 60 min and then depleted of any remaining ATP by the addition of 1.5 U hexokinase (Roche) and 5 mM d-glucose for 5 min at room temperature. Reaction mixtures were then washed in 10-kDa Nanosep columns four times with 10× reaction mixture volumes of HKEDG buffer (10 mM HEPES-KOH, pH 8.0, 50 mM KCl, 10% glycerol, 0.1 mM EDTA, 1 mM dithiothreitol added fresh), with a final resuspension in the original reaction mixture volume of HKEDG buffer. A similar procedure was used to prepare CpdR~P, except that EnvZ(T247R) was used instead of PhoR. These preparations of CtrA~P or CpdR~P were then incubated with upstream components as indicated in figure panels, each at a final concentration of 5 μM. Phosphatase reactions were supplemented with 5 mM MgCl2 and incubated at 30°C before being stopped at time points indicated in the figures by the addition of 3.5 μl 4× sample buffer (500 mM Tris, pH 6.8, 8% sodium dodecyl sulfate, 40% glycerol, 400 mM beta-mercaptoethanol). Samples were heated at 30°C for 2 min before being loaded onto 10% Tris-HCl gels (Bio-Rad) for electrophoresis at room temperature for 40 min at 150 V. Gels were exposed to phosphor screens overnight at −80°C and then scanned using a Storm 86 imaging system (Amersham Biosciences).
Histidine kinase constructs at 5 μM were incubated with 0.5 mM ATP and 5 μCi [γ-32P]ATP in HKEDG buffer supplemented with 5 mM MgCl2 at 30°C for 60 min. Reactions were stopped at time points indicated in the figures by the addition of 4× sample buffer, and mixtures analyzed as described above for phosphatase reaction mixtures.
His6-ChpT or TRX-His6-ChpT was added to autophosphorylation reaction mixtures to a final concentration of 12.5 μM, and reaction mixtures incubated at 30°C before being stopped at time points indicated in the figures and processed as described above.
CckA, unlike CtrA, is present throughout the cell cycle but is only active at certain stages of the cell cycle (16, 17). To test whether the abundance of ChpT is cell cycle regulated and, hence, a possible means of controlling the timing of CtrA activity, we generated polyclonal antibodies for ChpT. Immunoblotting with crude serum revealed a single major band in wild-type lysates that was absent in lysates from a chpT depletion strain and was the correct approximate size (see Fig. S1A in the supplemental material). To examine the cell cycle abundance of ChpT, we synchronized a population of wild-type cells and isolated samples every 20 min. Immunoblotting of these samples demonstrated that ChpT was present throughout the cell cycle, in contrast to CtrA, which showed a characteristic cell cycle dependence (Fig. (Fig.1).1). These results suggest that phosphate flux from CckA to CtrA is probably not regulated by changes in ChpT abundance.
Our ChpT antiserum also recognized purified His6-ChpT, although the molecular weight of this purified ChpT appeared slightly higher than that found in wild-type lysates (see Fig. S1A in the supplemental material). This difference could not be accounted for by the epitope tag, suggesting that the translational start site for chpT might have been erroneous in the original annotation of the C. crescentus genome (22). The chpT open reading frame contains methionines at positions 19 and 29 (relative to the originally annotated protein), each of which could serve as the bona fide translational start site. Alignment of chpT orthologs from several alphaproteobacteria indicated that the first 28 amino acids of C. crescentus ChpT were not conserved (see Fig. S2 in the supplemental material). We were able to complement the lethality of a chromosomal deletion of chpT with a plasmid expressing a version of chpT lacking the first 28 codons of the original annotation (see Fig. S1B in the supplemental material). This result strongly suggests that C. crescentus chpT encodes a protein of only 225 amino acids with a molecular mass of 23.4 kDa.
To verify that the smaller version of ChpT is capable of shuttling phosphate from CckA to CtrA and CpdR, we reconstituted the two cell cycle phosphorelays (CckA-ChpT-CtrA and CckA-ChpT-CpdR) using a purified version of the smaller ChpT, hereafter referred to simply as ChpT (see Fig. S1C in the supplemental material). Indeed, this shorter version of ChpT was able to efficiently shuttle phosphate from the receiver domain of CckA to either CtrA or CpdR.
The reconstituted cell cycle phosphorelays shown in Figure S1 in the supplemental material and those reported previously (1) involved a split version of CckA in which the histidine kinase (CckA-HK) and receiver domains (CckA-RD) were purified as separate polypeptides. Here, we wanted to examine the phosphotransfer behavior of a CckA construct containing both the kinase and receiver domains, as occurs in vivo. This construct, called CckA-HK-RD, lacking only the transmembrane domains, autophosphorylated and was an efficient phosphodonor for ChpT (Fig. (Fig.2A),2A), which then transferred the phosphoryl group to either CtrA or CpdR, as with the split version of CckA (1). These data confirm that CckA initiates two phosphorelays, culminating in the phosphorylation of CtrA and CpdR.
Phosphorelays are often reversible, such that phosphoryl groups can flow either up or down the pathway according to the principles of mass action (7-9, 40). In some cases, the histidine kinase involved can be bifunctional, acting to stimulate dephosphorylation of its cognate response regulator or, in the case of a hybrid kinase, its receiver domain. These bifunctional kinases can thus drive the rapid dephosphorylation of the terminal response regulator when they are not stimulated to autophosphorylate. To test whether CckA is bifunctional, we isolated radiolabeled CtrA~P and CpdR~P by phosphorylating each regulator for extended periods of time with the heterologous kinases PhoR (a histidine kinase from C. crescentus) and EnvZ(T247R) (a histidine kinase from E. coli that does not harbor significant phosphatase activity), respectively. The phosphorylated response regulators were then purified away from unreacted, radiolabeled ATP. This purification step was not 100% efficient, and each preparation of CtrA~P or CpdR~P retains some radiolabeled ATP that runs at a position similar to that of inorganic phosphate at the bottom of each gel in Fig. 2B and C.
Incubation of each regulator in buffer alone demonstrated that their aspartyl-phosphates are both relatively stable against autodephosphorylation in vitro, showing only a minor production of radiolabeled inorganic phosphate after 60 min; the band at the bottom of the second to fourth lanes in Fig. 2B and C increases in intensity only marginally relative to that in the first lane. In contrast, incubation of CtrA~P with ChpT and CckA-HK-RD led to a significant depletion of radiolabel from CtrA~P within 10 min, with nearly complete depletion in 60 min (Fig. (Fig.2B,2B, 8th to 10th lanes). The loss of radiolabel from CtrA~P also coincided with the appearance of radiolabeled inorganic phosphate, suggesting active dephosphorylation and not just partitioning of the phosphoryl groups among phosphorelay components. Incubation of CpdR~P with ChpT and CckA-HK-RD also led to a decrease in radiolabeled CpdR~P and an increase in inorganic phosphate (Fig. (Fig.2C,2C, 8th to 10th lanes), although not as much as with CtrA.
Notably, the dephosphorylation of CtrA and CpdR occurred at much higher rates when the kinase and receiver domains of CckA were fused as a single polypeptide. Incubation of the radiolabeled response regulators with ChpT and the split version of CckA (CckA-HK and CckA-RD) did not lead to a significant production of inorganic phosphate (Fig. 2B and C, fifth to seventh lanes). In these cases, phosphoryl groups did flow up the phosphorelay, as indicated by the appearance of radiolabeled ChpT and CckA-RD and the depletion of radiolabeled CtrA and CpdR. However, the levels of inorganic phosphate did not increase significantly, indicating that CckA-RD must be tethered to CckA-HK for efficient dephosphorylation. Taken together, our data suggest that the cell cycle phosphorelays can run in reverse and that CckA is bifunctional such that it can stimulate the dephosphorylation of its own receiver domain. Together, these two mechanisms, phosphorelay reversal and the phosphatase activity of CckA on its own receiver domain, can indirectly drive the dephosphorylation of CtrA~P and CpdR~P.
To assess whether CckA and phosphorelay reversal contribute to the dephosphorylation of CtrA or CpdR in vivo, we sought to identify mutations in cckA that uncouple its kinase and phosphatase activities to yield CckA with kinase-only (K+ P−) or phosphatase-only (K− P+) activity. To this end, we generated 10 mutant alleles of cckA based on mutations that render E. coli EnvZ either K+ P− or K− P+ (Fig. (Fig.3A)3A) (2, 6, 12, 21, 31, 39). We also made alanine mutations at the site of histidine autophosphorylation (H322) and at the site of aspartate phosphorylation in the receiver domain (D623), for a total of 12 mutations. We first introduced these mutations into our CckA-HK-RD construct and tested their abilities to autophosphorylate and phosphotransfer to ChpT in vitro (Fig. (Fig.3B).3B). Four of the mutant kinases (harboring mutations G318T, G319E, V366P, and D623A) retained clear, detectable levels of autophosphorylation, and each construct could phosphotransfer to ChpT, except for D623A. The G318T and G319E mutations each led to significantly higher levels of autophosphorylated CckA-HK-RD and higher levels of ChpT~P than in wild-type CckA-HK-RD. The V366P mutation, however, produced levels of CckA autophosphorylation and ChpT~P comparable to those seen with wild-type CckA-HK-RD.
Next, we tested whether any of the mutant kinase constructs that autophosphorylated could efficiently drive the dephosphorylation of CtrA~P via phosphorelay reversal and hydrolysis of phosphorylated CckA-RD (Fig. (Fig.3C).3C). Each mutant construct that retained kinase activity was added to ChpT and CtrA~P and then incubated for 30 min at 30°C. For the H322A, G318T, G319E, and V366P mutants, the radiolabeled phosphoryl groups flowed in reverse, as seen by the appearance of radiolabeled bands corresponding to ChpT and CckA. For CckA(D623A), phosphoryl groups partitioned between CtrA and ChpT but could not transfer back to CckA. CckA(D623A) lacks the aspartate phosphorylation site within the receiver domain and, therefore, cannot participate in phosphotransfer with ChpT. The dephosphorylation of CckA's receiver domain by its kinase domain was assessed by examining the production of inorganic phosphate and the coincident depletion of radiolabel from all other bands. The only mutant with phosphatase activity comparable to that of wild-type CckA was that harboring the substitution H322A.
These in vitro data indicate that the V366P mutation produces a version of CckA that retains kinase activity but lacks significant phosphatase activity (K+ P−), while the H322A mutation produces a version lacking kinase but not phosphatase activity (K− P+). To better characterize these two mutants, we analyzed time courses of CtrA~P dephosphorylation (Fig. 4A to C). CckA-HK-RD and CckA-HK-RD(H322A) each showed a depletion of radiolabel from the phosphorelay components along with an increase in inorganic phosphate. In contrast, the constructs harboring D623A and V366P showed little to no depletion of phosphorelay components and no significant production of inorganic phosphate. These data support the characterization of V366P as a K+ P− mutant of CckA with kinase activity comparable to that of wild-type CckA.
To test whether the phosphatase activity of CckA is important for cell cycle progression and viability, we tested whether the mutant alleles of cckA we created could complement a cckA chromosomal deletion. For these experiments, we placed a full-length copy of each mutant allele of cckA, driven by the native cckA promoter, on the low-copy-number plasmid pMR20. Each plasmid was transformed into the wild type, followed by transduction of a gentamicin-marked cckA deletion onto the chromosome. As expected, transduction of the ΔcckA mutation into a strain harboring the wild-type copy of cckA yielded thousands of colonies, while transduction into a strain harboring an empty vector yielded none. Transduction of ΔcckA into a strain containing cckA(D623A) also produced no colonies, consistent with the notion that phosphorylation of the receiver domain is essential for viability. Unexpectedly, we recovered hundreds of colonies when transducing the ΔcckA mutation into a strain containing cckA(H322A). However, sequencing of the plasmids in several of these colonies revealed that the mutation had reverted in each case, likely via recombination with the chromosomal copy of cckA prior to transduction. Reversion did not occur with the plasmid harboring cckA(D623A), probably because the D623A mutation is toward the end of the pMR20-cckA coding region and does not have sufficiently long regions of homology to efficiently drive recombination. Because we were unable to produce the cckA(H322A) ΔcckA strain, we conclude that H322, like D623, is essential for CckA function.
As with the H322A mutation, transduction of the ΔcckA mutation into a strain expressing cckA(G318T) yielded abundant colonies, but plasmid sequencing from multiple colonies indicated reversion to wild-type CckA. In vitro, CckA(G318T) had shown significantly increased kinase activity relative to that of wild-type CckA-HK-RD and no detectable phosphatase activity (Fig. (Fig.3).3). The inability of cckA(G318T) to complement a cckA deletion suggests that an imbalance in CckA activities is lethal. We cannot, however, say whether the lethality results from a lack of phosphatase activity, excessive kinase activity, or both.
For the G319E and V366P mutants, we successfully constructed and sequence verified strains in which the chromosomal copy of cckA was deleted and the mutant allele of cckA was carried on a plasmid. The strain expressing cckA(G319E) grew more slowly than a strain expressing wild-type cckA and exhibited severe cellular filamentation (Fig. (Fig.5).5). These cells formed long, relatively straight filaments reminiscent of the morphology of a strain overproducing CtrA(D51E)Δ3Ω, a nonproteolyzable version of CtrA that mimics the phosphorylated state and induces a G1 arrest (4). Indeed, the pMR20-cckA(G319E) + ΔcckA strain showed a significant increase in cells with one chromosome (Fig. (Fig.5).5). Our in vitro studies showed that CckA(G319E) exhibits a substantial increase in autophosphorylation relative to that of wild-type CckA. Taken together, these data suggest that the G319E mutation renders CckA hyperactive as a kinase, resulting in constitutive phosphorylation of CtrA and CpdR and, consequently, a G1 arrest.
The K+ P− mutation V366P did not lead to a severe cell cycle phenotype (Fig. (Fig.5),5), suggesting that the phosphatase activity of CckA is either not strictly essential for viability or that V366P does not completely eliminate phosphatase activity in vivo. However, even if CckA phosphatase activity is not strictly essential, CckA could still be an important phosphatase in vivo for either CtrA or CpdR. To further test this possibility, we sought to examine whether the phenotype of a strain expressing cckA(V366P) as the only copy of cckA was exacerbated by the synthesis of CtrA(D51E) or CtrAΔ3Ω. For example, if CckA is a key phosphatase for CtrA, cells producing both CtrAΔ3Ω and a K+ P− version of CckA may exhibit a G1 arrest phenotype, as with cells producing CtrA(D51E)Δ3Ω. For these experiments, we transformed the pMR20-cckA(V366P) + ΔcckA strain with medium-copy-number plasmids carrying ctrA, ctrA(D51E), ctrAΔ3Ω, or ctrA(D51E)Δ3Ω under the control of a xylose-inducible promoter. For comparison, we transformed the pMR20-cckA ΔcckA strain with the same set of plasmids. Each strain was grown in the presence of xylose to mid-exponential phase, and the chromosome content measured by flow cytometry (Fig. (Fig.6).6). The strains synthesizing CtrA(D51E) or CtrAΔ3Ω each showed a small but reproducible increase in G1-phase cells when combined with cckA(V366P) compared to the number when combined with cckA. No difference was seen between the strains synthesizing CtrA(D51E)Δ3Ω, indicating that CckA(V366P) mediates its cell cycle effect through the CckA-ChpT phosphorelays and not through other pathways. These data further suggest that CckA participates in the dephosphorylation of both CtrA~P and CpdR~P in vivo. However, the fact that CckA(V366P) does not yield a G1 arrest suggests that other phosphatases for CtrA and CpdR may exist. Or, as noted above, the V366P mutation may be an imperfect K+ P− allele that retains sufficient phosphatase activity in vivo to permit the dephosphorylation of CtrA~P and CpdR~P prior to DNA replication initiation.
To further test whether phosphorelay reversal and CckA phosphatase activity can drive the dephosphorylation of CtrA~P and CpdR~P in vivo, we examined the effect of overexpressing cckA. We hypothesized that overproducing CckA should siphon phosphoryl groups back through the phosphorelay driving the dephosphorylation of CtrA and CpdR, leading to a decrease in CtrA activity. To test this prediction, we placed a full-length copy of cckA on the plasmid pJS14 under the control of a xylose-inducible promoter. After growth in xylose for 4 h, this strain exhibited mild cellular filamentation and some accumulation of chromosomes, consistent with a downregulation of CtrA (Fig. (Fig.7A).7A). Overproducing a version of CckA lacking its transmembrane domains, CckAΔTM, produced more-severe filamentation and led to excessive accumulation of chromosomal DNA (Fig. (Fig.7A),7A), consistent with an even more significant downregulation of CtrA. This cellular filamentation and accumulation of chromosomes depended on back transfer to the CckA receiver domain, as overproducing CckAΔTM(D623A) did not severely disrupt the cell cycle (Fig. (Fig.7A).7A). However, back transfer alone was insufficient and CtrA downregulation also depended on the phosphatase activity of CckA, as overproducing the CckA receiver domain alone (CckA-RD) or a version of CckA lacking phosphatase activity, CckAΔTM(V366P), did not lead to cellular filamentation or chromosome accumulation (Fig. (Fig.7A7A).
The more-severe phenotype of overproducing CckAΔTM relative to the phenotype of full-length CckA may indicate that CckA in the membrane can adopt either a kinase or phosphatase state while a cytoplasmic fragment functions primarily as a phosphatase. Consistent with this hypothesis, we found that overproducing a full-length version of CckA(H322A), which can only function as a phosphatase, produced a more-severe phenotype than overproducing wild-type full-length CckA (Fig. (Fig.7A);7A); CckA(H322A) may also have a dominant negative effect by forming inactive heterodimers with the chromosomally expressed CckA.
Taken together, these data support a model in which the direction and flow of phosphoryl groups through the cell cycle phosphorelays in vivo is dictated by both mass action equilibrium and the kinase/phosphatase balance of CckA. When CckA is stimulated to autophosphorylate, the net result is an accumulation of phosphoryl groups on CtrA and CpdR. Conversely, when CckA is not activated as an autokinase, phosphoryl groups can flow back to the CckA receiver domain, where the kinase domain stimulates their hydrolysis.
As noted above, overproducing a full-length version of CckA did not yield a severe cell cycle phenotype, in contrast to the case of overproducing CckAΔTM, indicating that full-length CckA may retain a balance of kinase and phosphatase activities. If so, the overexpression of full-length CckA should, in principle, be exacerbated by mutations in other genes that regulate the activity of CckA. Our previous studies indicated that the response regulator DivK is a negative regulator of CckA (1). DivK phosphorylation is controlled by the reciprocal actions of a cognate histidine kinase, DivJ, and a cognate phosphatase, PleC (10, 41, 42). We therefore tested the effect of overproducing full-length CckA in either a divJ or a pleC mutant background. While CckA overproduction did not have a strong effect in the divJ mutant (data not shown), it appeared to be strongly synthetic with the pleC mutant (Fig. (Fig.7B).7B). CckA overproduction and the pleC mutation each yield a relatively mild phenotype on their own; however, the combination produced cells that were extremely filamentous and that accumulated multiple chromosomes, consistent with a significant drop in CtrA~P (Fig. (Fig.7B).7B). This severe cell cycle phenotype was completely dependent on the phosphatase activity of CckA, as overproducing full-length CckA(V366P) in a pleC mutant background had little to no effect on cells (Fig. (Fig.7B).7B). These results indicate that cckA likely lies genetically downstream of pleC and further support a model in which phosphorylated DivK downregulates CtrA by influencing the kinase/phosphatase balance of CckA.
In addition to changing from kinase to phosphatase during the cell cycle, CckA also dynamically changes its subcellular localization. CckA, which is present thoughout the cell cycle, was first reported to be polarly localized only in predivisional cells (17), with a second study indicating that CckA is also polarly localized in swarmer cells (1). Here, to further characterize CckA's polar localization and identify the source of this difference, we fused full-length cckA to gfp and integrated this construct on the chromosome as the only copy of cckA. The fusion used here includes the last two amino acids, both alanines, of CckA that had been removed in fusing cckA to gfp in both of the previous studies. By following a synchronous population of swarmer cells isolated from an exponential-phase culture (see Fig. S3 in the supplemental material), we found that CckA-GFP was delocalized in nearly all swarmer cells and remained delocalized upon differentiation into stalked cells. CckA-GFP then localized to the nascent swarmer pole in late stalked and early predivisional cells before localizing bipolarly in late predivisional cells. CckA-GFP was delocalized in daughter swarmer cells following cell division. In the daughter stalked cells, the pattern was variable, with CckA-GFP delocalized in some cells but retained at the stalked pole in most (>75%) cells, in contrast to both of the previous studies showing, at least in the small number of cells examined, that CckA-GFP is delocalized following cell division. Finally, we found that CckA-GFP localization in the initial synchronized population of swarmer cells was strongly dependent on the density of the culture used for synchronization. As cells progressed through early exponential phase and into late exponential phase, an increasing percentage of swarmer cells showed polarly localized CckA (see Fig. S4 in the supplemental material), indicating that the localization of CckA-GFP in swarmer cells is dependent on culture conditions but is not typically localized in early exponential phase. The overall pattern of subcellular localization observed here for CckA-GFP is in accord with that described by the C. Jacobs-Wagner group (personal communication). Also, we note that a similar pattern of CckA-GFP localization during synchronous cell cycle progression was seen with a strain expressing CckA-GFP from the low-copy-number plasmid pMR20 (data not shown). Whether the subcellular localization of CckA affects its activity as a kinase or phosphatase or vice versa is not yet clear and will likely require the identification of polar factors that directly influence CckA.
The Caulobacter cell cycle is ultimately driven by the periodic rise and fall in activity of the master regulator CtrA (Fig. (Fig.8A).8A). Our previous work identified two phosphorelays (1) that collaborate to activate CtrA by promoting its phosphorylation and proteolytic stabilization, the latter via CpdR phosphorylation. Conversely, the downregulation of CtrA depends critically on the dephosphorylation of CtrA and CpdR, but the mechanisms involved have been unknown previously. Here, we demonstrated that CckA, when not active as a kinase, can stimulate the dephosphorylation of CtrA and CpdR to help drive the initiation of DNA replication. We showed that phosphoryl groups can be transferred from CtrA~P and CpdR~P, via ChpT, back to the CckA receiver domain where the bifunctional CckA can stimulate hydrolysis (Fig. (Fig.8B8B).
As with phosphorelays in other organisms (7-9, 40), the direction of flow through the cell cycle phosphorelays in C. crescentus appears to be dictated by mass action. Hence, when CckA is not active as a kinase to drive CtrA and CpdR phosphorylation, the flow of phosphate can reverse. Overexpressing full-length cckA, however, resulted in a relatively minor cell cycle phenotype, likely because the CckA produced retains a balance of kinase and phosphatase activities. In contrast, overproducing a version of CckA lacking the transmembrane domains, CckAΔTM, led to a severe disruption of the cell cycle and downregulation of CtrA activity, as evidenced by chromosome accumulation. The more-severe effect of overproducing CckAΔTM relative to the effect of overproducing full-length CckA may indicate that CckA must associate with other factors in the membrane to autophosphorylate. This downregulation requires both the reversed flow of phosphoryl groups and their active elimination by CckA phosphatase activity (Fig. (Fig.7A).7A). The latter requirement is supported by the observation that overexpressing CckAΔTM(V366P) did not disrupt cell cycle progression. In wild-type cells, CckA and ChpT are present at much lower levels (E. G. Biondi and M. T. Laub, unpublished data) than CtrA, which is estimated to be present at ~20,000 molecules per cell (18). Such stoichiometries imply that redistribution alone could only ever deplete a small fraction of the phosphate on CtrA without CckA participating as a phosphatase.
Phosphorelay reversal and CckA phosphatase activity together constitute one mechanism for inactivating CtrA prior to S phase. Using a K+ P− mutant of CckA, CckA(V366P), we demonstrated that the phosphatase activity of CckA contributes to the downregulation of CtrA and CpdR in vivo. However, cells producing CckA(V366P) are still viable and able to initiate DNA replication, indicating that other phosphatases likely exist. If other phosphatases do exist, they may be difficult to identify owing to redundancy with CckA's phosphatase activity, either of which may be sufficient for survival. Moreover, aspartyl-phosphatases do not comprise a single, paralogous family and typically show little to no sequence homology with one another, making their identification difficult (23, 26, 27, 36, 43). Alternatively, no other phosphatases may exist if the phosphoryl groups on CtrA~P and CpdR~P are intrinsically labile, as with CheY and other response regulators (28, 32, 38). However, our data suggest that the aspartyl-phosphates on CtrA and CpdR are relatively stable (Fig. 2B and C), indicating that active dephosphorylation is probably necessary and tightly regulated. Finally, as noted earlier, CckA could be the only phosphatase if the V366P mutation does not completely eliminate phosphatase activity. Our in vitro studies did not indicate any significant phosphatase activity for CckA(V366P), but the in vitro conditions may not perfectly reflect in vivo conditions.
How does the V366P mutation produce a kinase-positive and phosphatase-negative version of CckA? Notably, valine-366 in CckA is predicted, based on alignment to EnvZ, to lie at the C-terminal end of α-helix-2 in the dimerization and histidine phosphotransfer (DHp) domain, near the linker that connects the DHp and catalytic and ATPase (CA) domains. It is thus tempting to speculate that a proline at this position (V366P in CckA, which was based on the previously reported L288P in EnvZ ) may interfere with kinase/phosphatase balance by affecting domain-domain interactions. Recent structural studies of a full-length histidine kinase provided evidence that modulating DHp-CA domain interactions significantly influences the kinase/phosphatase balance of bifunctional histidine kinases (20). It will be interesting to see whether mutations equivalent to V366P in CckA and L288P in EnvZ can produce K+ P− versions of other bifunctional histidine kinases.
In sum, our results indicate that CckA switches between a kinase state and a phosphatase state to help drive the changes in CtrA's activity that are crucial for proper cell cycle progression. In vivo measurements of CckA phosphorylation indicated that CckA kinase activity is detectable in swarmer cells, drops to its lowest levels in stalked cells, and then accumulates again to maximal levels in predivisional cells (16). CtrA and CpdR phosphorylation levels change in a similar fashion during the cell cycle (4, 14, 16), consistent with a model in which changes in CckA's kinase activity are translated into changes in CtrA's activity (Fig. (Fig.8A8A).
What then regulates CckA's activity? The essential single-domain response regulator DivK plays a key role. A cold-sensitive divK mutant [divK(Cs)] is unable to downregulate CtrA and consequently arrests with a single chromosome (13), as seen with cells overproducing CtrA(D51E)Δ3Ω (4) or as seen here with cells overproducing CckA(G319E), a version of CckA with a high level of kinase activity. While DivK could control CtrA phosphorylation and degradation independently, a simpler model is that DivK regulates CckA, either directly or indirectly switching CckA from the kinase to the phosphatase state. Consistent with this model, CckA phosphorylation levels per cell were found to increase in a divK(Cs) mutant (1). Although the increase was only fourfold, it should be noted that this measurement compared the divK(Cs) mutant to a mixed population of wild-type cells which includes predivisional cells, where CckA is most active. The divK(Cs) strain, however, is arrested at the G1-S transition when CckA kinase activity is normally at its lowest; in fact, the unabated activity of CckA as a kinase in the divK(Cs) strain may be responsible for its G1 arrest phenotype. We also found here that cckA overexpression exhibits a strong synthetic interaction with pleC, which encodes a key phosphatase of DivK. This synthetic interaction was dependent on CckA's ability to act as a phosphatase, as overexpressing the phosphatase-deficient CckA(V366P) in a pleC mutant did not cause cellular filamentation or chromosomal accumulation (Fig. (Fig.7B).7B). In a pleC mutant, DivK~P levels are elevated (41), and our results suggest that this increase may bias CckA toward the phosphatase state when overproduced, leading to the downregulation of CtrA and a severe cell cycle phenotype (Fig. (Fig.7B).7B). If DivK functioned independently of CckA to regulate CtrA, the overexpression of cckA in a pleC background may have resulted in an additive, and consequently less-severe, effect on the cell cycle.
DivK also affects CckA localization, with CckA-GFP present at the stalked pole but absent from the opposite pole in divK(Cs) mutants (1). However, this may be a secondary effect of DivK's effect on CckA activity and the consequent G1 arrest. Whether the localization of CckA influences its activity as a kinase or phosphatase or vice versa is not yet clear. CckA is most active as a kinase in predivisional cells when it is localized to the nascent swarmer pole and least active in stalked cells where it is either delocalized or only at the stalked pole. This may suggest that CckA receives an activation signal at the nascent swarmer pole or a repressing signal at the stalked pole. However, CckA also has moderate kinase activity in exponential-phase swarmer cells when it is typically delocalized. A better understanding of the role of subcellular localization in modulating the kinase and phosphatase states of CckA will require the identification of factors that directly activate or repress CckA.
The model that DivK negatively regulates CckA is consistent with recent data suggesting that CpdR phosphorylation levels may increase after prolonged depletion of DivK (15). This observation could indicate that DivK functions in a second pathway to specifically stimulate CpdR dephosphorylation. Alternatively, or perhaps in addition, the depletion of DivK may simply lead CckA to remain in a kinase rather than a phosphatase state; this would lead to increased phosphorylation of CpdR (and CtrA) and, ultimately, the G1 arrest phenotype characteristic of divK loss-of-function mutants. Further, divK mutants can be rescued if cpdR is replaced by a mutant allele that cannot be phosphorylated (15), and divK lethality was previously shown to be suppressed by other mutations that diminish CtrA activity (42). We thus favor a model in which DivK helps switch CckA, either directly or indirectly, from acting predominantly as a kinase to predominantly as a phosphatase and where an inability to switch (in either direction) is lethal (Fig. (Fig.8).8). The switching of CckA from kinase to phosphatase likely depends on the phosphorylation of DivK by DivJ, its cognate kinase. DivJ is preferentially inherited by stalked cells and accumulates in stalked cells following the swarmer-to-stalked transition (41), presumably helping to temporally restrict the downregulation of CckA kinase activity and the dephosphorylation of CtrA to stalked cells.
In sum, our results emphasize the critical role played by CckA in controlling cell cycle oscillations and cellular asymmetry in Caulobacter. Although CtrA is also regulated transcriptionally, constitutive expression of ctrA does not significantly disrupt or delay cell cycle progression, indicating that proteolysis and phosphorylation are likely the dominant modes of regulation. CckA controls both of these processes. In turn, a complex network of regulatory molecules, including DivJ, PleC, and DivK, appear to regulate CckA activity, helping to toggle it between kinase and phosphatase states at the appropriate stages of the cell cycle.
We thank members of the Laub laboratory for critical reading of the manuscript.
This work was supported by an NIH grant (5R01GM082899) to M.T.L. M.T.L. is an Early Career Scientist at the Howard Hughes Medical Institute.
Published ahead of print on 25 September 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.