Topological control of the Caulobacter cell cycle circuitry by a polarized single-domain PAS protein

Despite the myriad of different sensory domains encoded in bacteria, only a few types are known to control the cell cycle. Here we use a forward genetic screen for Caulobacter crescentus motility mutants to identify a conserved single-domain PAS (Per-Arnt-Sim) protein (MopJ) with pleiotropic regulatory functions. MopJ promotes re-accumulation of the master cell cycle regulator CtrA after its proteolytic destruction is triggered by the DivJ kinase at the G1-S transition. MopJ and CtrA syntheses are coordinately induced in S-phase, followed by the sequestration of MopJ to cell poles in Caulobacter. Polarization requires Caulobacter DivJ and the PopZ polar organizer. MopJ interacts with DivJ and influences the localization and activity of downstream cell cycle effectors. Because MopJ abundance is upregulated in stationary phase and by the alarmone (p)ppGpp, conserved systemic signals acting on the cell cycle and growth phase control are genetically integrated through this conserved single PAS-domain protein.

C ellular motility is responsive to external signals such as nutritional changes, but it is also regulated by cues induced systemically during each cell division cycle 1,2 . This latter characteristic has been successfully exploited in motility screens to uncover cell cycle regulators in the model bacterium Caulobacter crescentus (herein Caulobacter), an aquatic a-Proteobacterium that is easily synchronized by density-gradient centrifugation owing to a cell cycle-regulated capsule [3][4][5][6] . Motility is conferred by the polar flagellum, a structure that is required for the dispersal of the swarmer cell type. Caulobacter divides asymmetrically into a swarmer cell that resides in a G1-like quiescent state and harbours the flagellum and several pili at the same cell pole, and a replicative (S-phase) cell whose old cell pole is decorated by a cylindrical extension of the cell envelope (the stalk) tipped by an adhesive holdfast (Fig. 1a) 2 . Flagellar motility along with adhesive properties (conferred by the polar pili and holdfast) and cell division are wired into the C. crescentus cell cycle circuitry at the transcriptional level by the master cell cycle regulator CtrA 7 , a DNA-binding response regulator (RR) of the OmpR family 4 whose synthesis is activated in S-phase by the transcriptional regulator GcrA 8 . Following its synthesis, CtrA activates many developmentally regulated promoters, including those of motility, pilus, holdfast and cell division genes 7,[9][10][11] . CtrA also functions as negative regulator of gene expression and, directly and/or indirectly, inhibits firing of the origin of DNA replication (Cori). Cori fires only once during the Caulobacter cell cycle and is bound by CtrA at multiple sites 12,13 .
The DNA-binding activity of RRs such as CtrA is regulated by phosphorylation at a conserved (Asp) residue, a step that is often executed directly by a histidine kinase (HK) 7 . Before the  phosphate can be accepted by the RRs, the dimeric HK transauto-phosphorylates a conserved histidine (His) in an ATPdependent manner. However, phosphorylation of RRs can also occur indirectly by way of two intermediary components in the His-Asp phosphorylation pathway, an Asp-containing receiver domain and a His-containing phosphor-transfer domain 14 .
Complex regulatory schemes emerge when such multicomponent His-Asp pathways are arranged in tandem 15 . The accumulation of phosphorylated CtrA (CtrABP) underlies such a complex His-Asp pathway topology [15][16][17] . CtrABP is present in G1-phase, degraded at the G1-S transition by the ClpXP protease 18,19 , to enable the initiation of DNA replication, and then re-synthesized and phosphorylated during S-phase (Fig. 1a). The removal of CtrABP at the G1-S transition is induced when the DivJ HK phosphorylates its substrate, the DivK receiver domain, which downregulates both the phosphorylation and halflife of CtrA via the DivL tyrosine kinase 15,16,[20][21][22][23][24] . CtrA abundance and stability are also regulated by the conserved alarmone (p)ppGpp, which tunes cell cycle progression in response to nutritional signals [25][26][27][28][29] . Superimposed on the temporal events is the spatial regulation of these His-Asp pathway components (Fig. 1a) 30 . At the G1-S transition, DivJ is recruited to the stalked pole by its localization factor SpmX, an event that also is required for optimal DivJ kinase activity in vivo 3,31,32 . DivK is enriched at the poles as well, enhancing kinase activity 33 , but a significant fraction of DivK is also dispersed 17,34 . DivK activity is antagonized by the phosphatase PleC that is recruited to the pole opposite the stalk by its localization factor PodJ 32, [35][36][37] . A pleiotropic effector of protein polarization in C. crescentus is the polar organizing protein Z that is thought to self-assemble into a polar matrix at both cell poles to sequester proteins like SpmX, DivJ, DivK and others [38][39][40][41] . Although the DivJ and SpmX complex is unipolar, DivK localization is bipolar and promoted by SpmX and DivJ activity 17,38 , indicating that DivK localization opposite the stalk is regulated by another factor. The sequestration of DivK to the stalked pole is governed by DivJ, even in the absence of kinase activity 17 .
Here, we report a screen for motility mutants that unearthed a conserved and cell cycle-regulated single-domain PAS (Per-Arnt-Sim) 42 protein (MopJ) that promotes CtrA accumulation during exponential growth and in stationary phase. MopJ is localized to the cell poles where it acts on downstream cell cycle signalling proteins, whereas upstream cell cycle regulators promote MopJ polarization. Our work unveils MopJ as a modulator of the spatio-temporal circuitry controlling bacterial cell cycle progression and as a regulatory node through which stationary phase and (p)ppGpp-dependent nutritional signals are integrated.

Results
MopJ is a pleiotropic regulator of motility. Our previous comprehensive transposon mutagenesis screen for strains with a motility defect on swarm (0.3%) agar that led to the identification of the spmX gene 3 also yielded one mutant strain (NS61, Fig. 1b) harbouring a himar1 insertion in the uncharacterized gene CCNA_00999 (at nucleotide position 1082252 of the Caulobacter crescentus wild-type (WT) strain NA1000 (ref. 43). This gene is predicted to encode a single-domain PAS protein (PAS_5, pfam07310, residues 15-147, Fig. 1c) that is 165 residues in length and is henceforth referred to as MopJ (motility PAS domain associated with DivJ, see below). The PAS domain, a fivestranded antiparallel b-sheet flanked by several a-helices, is generally used to bind small-molecule ligands and is structurally related to the GAF regulatory (GMP-specific and -regulated cyclic nucleotide phosphodiesterase, Adenylyl cyclase and Escherichia coli transcription factor FhlA) domain often encoded on the same polypeptide 42 . Orthologues of MopJ are also encoded in the genomes of distantly related a-Proteobacteria (Supplementary Fig. 1) such as the animal pathogen Brucella melitensis (BMEI0738), the plant pathogen Agrobacterium tumefaciens (Atu1754) and the plant symbiont Sinorhizobium meliloti (SMc01000). These proteins exhibit a similar domain organization with MopJ, that is, a single PAS_5 domain without any associated regulatory or effector domain (Fig. 1c). The himar1 transposon in NS61 lies near the middle of the mopJ gene (that is, after codon 89, Fig. 1c), presumably disrupting its function. In support of this notion, an in-frame deletion of mopJ (DmopJ) in WT cells recapitulated the motility defect observed in the mopJ::himar1 mutant on soft agar (Fig. 1b). As the motility defect of DmopJ cells can be corrected by supplying mopJ in trans on a plasmid (pMT335-mopJ) under the control of the vanillateinducible promoter P van ( Supplementary Fig. 2a), we conclude that MopJ is a hitherto uncharacterized regulator of motility in C. crescentus.
As inactivation of cell cycle regulators often results in pleiotropic defects in C. crescentus including a reduction of motility 5 caused by a change in the fraction of G1-phase cells in the population, we probed for such changes by fluorescenceactivated cell sorting (FACS) analysis of exponentially growing WT and DmopJ cells stained with the nucleic acid dye SYTOX Green (Fig. 1d). A lower G1:G2 cell ratio was observed in DmopJ cultures compared with WT, while the number of S-phase cells seemed unaffected. By contrast, stationary WT or DmopJ cultures contained few S-phase cells, confirming that the majority of cells are no longer replicating (Fig. 1e). Although stationary WT cells have an equal tendency to arrest either in G1-or G2-phase, stationary DmopJ cells are enriched in G2-phase over G1-phase. Consistent with these findings, differential interference contrast (DIC) microscopy and FACS revealed only slight increase in cell length caused by the DmopJ mutation in exponential phase (Fig. 1d,f), whereas in stationary phase many DmopJ cells are elongated and appear to divide aberrantly (Fig. 1e,g).
Taken together, our results reveal MopJ as a novel pleiotropic regulator of motility and of normal cell cycle progression in C. crescentus. Specifically, MopJ promotes the accumulation of G1-phase cells in exponential phase and of G1-and G2-phase cells in stationary phase.
MopJ acts on the CtrA regulon and is induced by (p)ppGpp. Knowing that MopJ impacts both motility and the abundance of G1 cells, we tested if MopJ affects the expression of the CtrA regulon, as this includes motility, division and G1-phase genes 9,10,12 . In addition, this master transcriptional regulator is a likely target of MopJ because it also regulates the initiation of chromosome replication 4,12 . To this end, we measured if b-galactosidase (LacZ) expression from CrA-regulated promoters is altered in exponential and stationary phase WT and DmopJ cells harbouring P pilA -, P sciP -or P fljM -lacZ promoter probe plasmids (harbouring LacZ under the control of the promoter of the pilA pilin gene, the sciP repressor gene or the fljM minor flagellin gene, Fig. 2a,b) that are directly activated by CtrA 9,10,44 versus a podJ promoter reporter ( Supplementary  Fig. 3A) that is repressed by CtrA. Although only a modest reduction in LacZ activity (14-25%, depending on the CtrAactivated promoter, see Supplementary Table 1) was discernible in exponentially growing DmopJ cells compared with WT cells (Fig. 2a), this difference was accentuated in stationary phase ( Fig. 2b and Supplementary Table 1) and even further magnified in a reporter that is indirectly activated by CtrA such as P fljK -lacZ, in which the promoter of the major flagellin gene fljK drives LacZ expression (Fig. 2a,b and Supplementary Table 1). P fljK is a target of the FlbD activator whose expression in turn is directly and positively regulated by CtrA 9,10 .
Immunoblotting using polyclonal antibodies to FljK, SciP and PilA confirmed the trends observed with the LacZ transcriptional reporters ( Fig. 2c and Supplementary Fig. 4a). (Note that the PilA protein is absent from stationary phase WT cells for reasons that are currently unknown, but is likely operating at the posttranscriptional level (compare Fig. 2b,c and Supplementary  Fig. 4a). Immunoblotting using antibodies to CtrA revealed that CtrA abundance is strongly dependent on MopJ in stationary phase ( Fig. 2d and Supplementary Fig. 4b), consistent with the exacerbated effects on CtrA and its regulon in stationary DmopJ cells compared with WT cells. Stabilizing CtrA either by masking the C-terminal recognition motifs for ClpXP in CtrA (using the ctrA::ctrA-M2 allele 45 ) or by inactivating the gene encoding the CpdR proteolytic activator of CtrA 16,20 restores CtrA abundance to near WT levels (Fig. 2e), indicating that CtrA proteolysis by the ClpXP protease contributes to downregulation of CtrA levels in stationary DmopJ cells.
Next, we raised antibodies to MopJ to probe for commensurate changes in MopJ abundance in stationary phase versus exponential phase. In support of the stationary phase defects of DmopJ cells described above, we observed that MopJ was barely detectable in lysates from exponential phase cells, but is abundant in lysates from stationary phase cells ( Fig. 2d and see Fig. 3a and Supplementary Fig. 4b). As the alarmone (p)ppGpp accumulates in stationary phase when nutrients become exhausted in bacteria 25,46 , we asked if the stationary phase induction of MopJ requires the (p)ppGpp-synthase/hydrolase SpoT 27,28,47 . To resolve this question, we conducted b-galactosidase (LacZ) measurements in cells harbouring a transcriptional reporter, in which a promoterless lacZ gene is fused to the mopJ promoter (P mopJ -lacZ). Accordingly, we measured P mopJ -lacZ activity during   exponential and stationary phase in the presence or absence of SpoT (that is, in WT and DspoT cells, respectively), observing a strong (50%) reduction of P mopJ -lacZ expression in stationary cells lacking SpoT (Fig. 2f,g). To demonstrate that (p)ppGpp induction is sufficient for the P mopJ -lacZ induction even in exponential phase cells, we induced (p)ppGpp synthesis in ARTICLE exponential phase cells using the constitutively active (p)ppGpp synthase RelA' from E. coli 47 and detected a commensurate increase in P mopJ -lacZ activity (Fig. 2h). By contrast, no induction was observed upon induction of the catalytic mutant derivative RelA'-E335Q (Fig. 2h), demonstrating that (p)ppGpp is necessary and sufficient for induction of P mopJ -lacZ.
Thus, MopJ acts on CtrA and its regulon, especially in stationary phase, and (p)ppGpp signalling by SpoT induces MopJ at the transcriptional level in stationary phase cells.
Coordinated synthesis of MopJ and CtrA in S-phase. The functional relationship between MopJ and CtrA is further reinforced by their concurrent accumulation during the cell cycle, coordinated by the S-phase regulator GcrA. Immunoblotting revealed that MopJ appears during the cell cycle coincident with or slightly ahead of CtrA and follows the accumulation of GcrA and DivJ ( Fig. 3a and Supplementary Fig. 5a,b). Separation of exponential and stationary phase WT cells into swarmer (G1-phase) and stalked/pre-divisional (S-phase) cell fractions showed MopJ to be absent from the former and present in the latter ( Fig. 3b and Supplementary Fig. 5c). The MopJ expression pattern matches that of the TipF flagellar regulator, an unstable protein that is expressed from a GcrA-activated promoter in early S-phase and is subsequently proteolysed in a manner that depends on the ClpXP protease 48 . Measuring LacZ activity in GcrA-depleted cells harbouring the P mopJ -lacZ reporter, we observed a strong reduction after 6 h of depletion relative to WT (Fig. 3c) or to GcrA-replete cells ( Supplementary Fig. 3b), indicating that GcrA also acts positively on this promoter. In support of this, previous chromatin immunoprecipitation coupled to deep-sequencing experiments revealed that GcrA binds P mopJ (Fig. 3d) efficiently in vivo and that it harbours N6-methyladenosine (m6A, Fig. 3e) marks, a hallmark of a class of promoters that fire in S-phase and that require methylation for efficient activation and binding of GcrA in vitro and in vivo 49,50 .
Interestingly, MopJ features two terminal alanine residues that resemble (class I) C-terminal recognition motifs of the ClpXP protease 51 and MopJ was previously found associated with the ClpP protease in pull-down assays 52 . We therefore investigated if ClpXP regulates MopJ abundance in vivo. To this end, we used immunoblotting to monitor the abundance of MopJ in cells from which the ClpX ATPase component of the ClpXP proteolytic machine had been depleted ( Fig. 3f) or in cells intoxicated with a dominant negative version of ClpX ( Fig. 3g) 53 . We found steadystate levels of MopJ to be elevated under these conditions. Moreover, the abundance of MopJ and CtrA, a known ClpXP substrate 18 , is elevated upon disruption of the clpX or the clpP gene in DsocB mutant cells (Fig. 3h). (Note that inactivation of clpX or clpP is lethal in WT cells, but no longer in DsocB cells, in which the gene encoding the SocB toxin has been deleted 54 ). Finally, the levels of a MopJ variant harbouring GFP fused to the C-terminus (MopJ-GFP) expressed from P mopJ no longer fluctuate during the cell cycle ( Fig. 3I and Supplementary  Fig. 5d) and confers motility to a comparable level as the untagged (WT) version ( Supplementary Fig. 2b).
In sum, a combination of GcrA-controlled synthesis and ClpXP-dependent proteolysis restricts MopJ accumulation to S-phase and ensures that re-synthesis of CtrA and its reinforcement factor MopJ (see above) is linked, thus optimally reinforcing CtrA accumulation and function. Moreover, as MopJ is also induced in stationary phase by (p)ppGpp, an adaptive (nutritional) feed exists into the cell cycle.
Bipolar localization of MopJ requires the DivJ kinase. Superimposed on the temporal regulation of MopJ is its cell cyclecontrolled polar sequestration revealed in time-lapse fluorescence microscopic imaging of live mopJ::mopJ-GFP cells (Fig. 3j). Although diffuse fluorescence was observed in G1-phase (swarmer) cells, MopJ-GFP localizes first to the stalked pole during the transition into S-phase (stalked) cells and adopted a bipolar localization until MopJ-GFP dispersed from the opposite pole at the time of cell division (Fig. 3j). MopJ-GFP localized in a similar way when expressed constitutively from the xylose-inducible P xyl promoter (Fig. 3k), compared with expression from its native promoter in S-phase, providing further evidence that the cell cycle-regulated localization occurs even when MopJ is expressed ectopically. Imaging of MopJ-GFP (Fig. 4a) or GFP-MopJ ( Supplementary Fig. 6) expressed from the xylX locus in unsynchronized C. crescentus cultures also revealed monopolar and bipolar localization patterns. Consistent with the time-lapse imaging, MopJ-GFP is bipolar in pre-divisional cells, but is dispersed from the pole opposite the stalk in deeply constricted cells (Figs 3j,k and 4a), a characteristic also observed for the DivK RR that is phosphorylated by the DivJ kinase (Fig. 1a). Further reinforcing the parallels between MopJ and DivK (bi)polarity, we observed that MopJ-YFP and DivK-CFP co-localize in C. crescentus WT cells (Fig. 4b). This prompted us to explore if MopJ and DivK rely on a common localization mechanism. Indeed, in the absence of DivJ, DivK-GFP delocalized 34 (Fig. 4c) and MopJ-GFP as well (Fig. 4a,d and Supplementary Fig. 7a,b). Moreover, a truncated version (residues 1-392) of DivJ containing the HisKA domain, but lacking the ATPase (HATPase_c) domain, is sufficient to direct MopJ-GFP and DivK-GFP to the stalked pole, but no longer to the opposite pole (Fig. 4c,d and Supplementary Fig. 7a,b). Thus, as was reported for DivK-GFP previously, MopJ-GFP requires DivJ kinase activity to be  ARTICLE sequestered to the pole opposite the stalk, but its sequestration to the stalked pole occurs independently of kinase activity 17 (Fig. 4c,d). A further truncated version of DivJ (residues 1-329) no longer supported polar localization of MopJ-GFP and DivK-GFP (Fig. 4c,d and Supplementary Fig. 7a,b), indicating that the dimerization/phosphor-acceptor (HisKA) domain and/or a determinant encoded from residues 329 to 392 is critical to recruit MopJ and DivK to the stalked pole. By contrast, expression of full-length DivJ from a comparable plasmid reinstated bipolarity upon MopJ-GFP and DivK-GFP (Fig. 4c,d and Supplementary Fig. 7a,b). As both derivatives of DivJ can still localize to the stalked pole (albeit 50% less efficient for the shorter version, Supplementary Fig. 8), this experiment confirms that residues 329-392 of DivJ encode for the localization determinants of MopJ and DivK. We conclude that DivJ integrates the bipolar localization of two dissimilar effectors, MopJ and DivK, to the stalked pole and, in early pre-divisional cells, also to the opposite pole.

+ Vector
In support of the notion that a physical interaction between MopJ and DivJ underlies the localization of MopJ to the stalked pole, co-immunoprecipitation experiments revealed that DivJ is pulled-down with MopJ-GFP from cellular lysates of a WT strain expressing MopJ-GFP from P xyl (Fig. 4e). By contrast, DivJ was not detectable in comparable pull-downs from lysates of the WT strain expressing GFP from P xyl (Fig. 4e). Furthermore, DivJ was also recovered with MopJ in tandem-affinity purification experiments 55 conducted with lysates of WT cells expressing MopJ-TAP from P van (Fig. 4f). Having shown that MopJ resides in a complex with DivJ, we then tested whether overexpression of MopJ can affect the DivJ-regulated subcellular distribution of the other DivJ-interacting effector DivK 17,34 (Fig. 5a,b). Live-cell imaging of DivK-GFP in cells overexpressing MopJ from P van revealed an enhancement of DivK-GFP polar localization and a concomitant reduction of the cytoplasmic fluorescence that correlated with the strong induction of MopJ upon the addition of vanillate (Fig. 5a,b), without affecting DivKBP levels noticeably by in vivo phosphorylation analysis (Fig. 5c). However, we found that DmopJ cells are sensitized towards elevated DivK levels from a plasmid expressing DivK under the control of P xyl , suggesting that extra DivK 17 further curbs the CtrA pathway in DmopJ cells (in which activation of the CtrA regulon is already disturbed, see above) to inhibit growth (Fig. 5d) and division (Fig. 5e) more than in WT cells. Taken together, these experiments show that MopJ binds DivJ and depends on it for its own polar localization, while promoting the polar sequestration of DivK and attenuating its activity.
Examining possible effects on downstream components, we found that MopJ overexpression not only alters the subcellular distribution of DivK, but also that of its interaction partner, the tyrosine kinase DivL 24 that localizes primarily to the pole opposite the stalk 15,22,56 (Fig. 1a). Overexpression of MopJ caused a near fivefold increase in the fraction of bipolar DivL-GFP (expressed from the native promoter at the divL locus in lieu of endogenous DivL; Fig. 5f and Supplementary Fig. 9). Conversely, in cells lacking MopJ, DivL-GFP was largely delocalized (Fig. 5g) and a divL::Tn5 mutation (in which Tn5 is inserted upstream of the region encoding the HisKA domain) did not impact MopJ-GFP localization (Supplementary Fig. 10). The previous observation that DivL is delocalized in the absence of DivJ 56 is consistent with our findings, as we showed above that DivJ is also required for the polarization of MopJ-GFP. Intriguingly, DivL localization is thought to be dependent on the onset of DNA replication 57 . The result that MopJ is required for DivL localization and MopJ expression requires GcrA is noteworthy, as it could provide an explanation for the finding that DivL polarization (and the concomitant re-accumulation of CtrA that this event regulates) requires S-phase entry 57 . GcrA expression and S-phase entry are known to be dependent on the replication initiator DnaA 58 .
In sum, MopJ is both target (via DivJ) and effector (via DivK and DivL) of the Caulobacter spatiotemporal cell cycle network and is thus perfectly positioned to reinforce CtrA with the synthesis of both proteins in S-phase and is genetically linked as expression of both proteins is directly induced by GcrA.
A second localization pathway controls polarization of MopJ. Although DivJ is necessary for the bipolar localization of MopJ, two findings indicate that it is not sufficient for bipolarity. First, MopJ is localized to the pole opposite the stalk (Fig. 4a), where no focus of DivJ is observed 32 . Second, MopJ is still polarized when DivJ is delocalized by inactivation of its localization factor SpmX (that is, in DspmX mutant cells, Fig. 4a and Supplementary  Fig. 11) 3 . Taken together, this indicates that other factors must promote polarization of MopJ. As it is also conceivable that DivJ simply targets MopJ to the membrane or fastens it there, we also explored whether the bipolar scaffolding protein PopZ that forms adhesive patches at both cell poles 26,36,38-40 might keep MopJ-GFP polarized. Live-cell imaging revealed that MopJ-GFP is dispersed in DpopZ::O mutant cells (Fig. 4a). As PopZ has also been likened to a molecular plug that prevents diffusion away from the cell pole, we predicted that an additional factor promotes MopJ-GFP localization to the pole opposite the stalk. The PodJ polarity protein is known to attract regulatory proteins of CtrA to the pole opposite the stalk 36,37,59,60 . We observed MopJ-GFP to be monopolar in DpodJ mutant cells, unable to assemble into foci at the pole opposite the stalk regardless of whether SpmX was present or not ( Fig. 4a and Supplementary  Fig. 11). This dependence of MopJ-GFP on PodJ for localization to the flagellated pole not only supports the existence of distinct recruitment mechanisms of MopJ-GFP for each cell pole in WT cells, but also provides an explanation of why DivL is no longer localized to the flagellated pole in DpodJ cells 59 . It also points to a possible explanation of how DivJ might influence MopJ localization to the pole opposite the stalk. As the expression of the gene encoding the PerP periplasmic protease that cleaves PodJ is strongly and directly upregulated by CtrA in DdivJ::O mutant cells 9,10,61 , we hypothesize that DivJ promotes the localization of MopJ at the pole opposite the stalk indirectly through PodJ (or another DivJ-dependent protein).

Discussion
A single-domain PAS protein, MopJ, is not only regulated by cell cycle cues, but also integrates nutritional and growth phase regulatory signals to exert topological control over conserved components of a bacterial cell cycle network (Fig. 6). The identification of MopJ enabled us to illuminate two important and hitherto enigmatic cell cycle events. The first conundrum was how the master regulator CtrA can re-accumulate in S-phase 4 , a time when DivK is still phosphorylated and signals the removal of CtrA 30,34 . Unless the proteolytic cascade is attenuated and/or the rate of synthesis is potentiated such that it exceeds proteolysis, CtrA accumulation should be suppressed or curbed. We found that the PAS-domain protein MopJ indeed enhances CtrA accumulation and acts on DivK and its target DivL 15 . MopJ re-accumulation is temporally linked with that of CtrA through the S-phase-specific transcriptional activator GcrA that binds the ctrAP1 and the mopJ promoter 8,49,50 , thus reinforcing CtrA re-synthesis with the coordinated synthesis of its enhancing factor MopJ in S-phase.
By discovering that MopJ acts at the same subcellular site as the CtrA His-Asp phosphorelay components DivJ, DivK and DivL, we were able to identify MopJ as an important missing link connecting DivL polarization to DivJ (Fig. 6) and likely to the onset of DNA replication. MopJ polarization requires the polar organizer PopZ and the DivJ kinase. With the previous findings that polar localization of DivJ is itself dependent on PopZ assembly at polar sites 38,40 , our work extends this spatiotemporal regulatory network by showing that MopJ associates with DivJ ( Fig. 4), enhances the sequestration of DivK to the cell poles (attenuating DivK function, Fig. 5), governs subcellular localization of DivL and facilitates CtrA accumulation and expression of its target promoters (Fig. 2). The stationary phase and (p)ppGpp-based induction of MopJ unveils a concerted feed-forward mechanism underlying signal integration during bacterial cell cycle progression, acting on an essential master transcriptional regulator in different phases of growth (Fig. 6). Why MopJ abundance increases in stationary phase remains to be determined, but our experiments suggest that this inductions occurs at the transcriptional level as the activity of the mopJ promoter is elevated in stationary phase and is inducible in nutrient replete conditions when (p)ppGpp levels are raised artificially in exponential phase (Fig. 2). As (p)ppGpp signals nutrient starvation by globally reprogramming RNA polymerase and is induced in stationary phase in many bacteria 46 , the induction of MopJ is likely mediated at the level of promoter activity. It is also possible that MopJ synthesis is further reinforced at the level of protein stability by inhibiting ClpXPmediated proteolysis (Fig. 3). It is not obvious why MopJ should be induced in stationary phase when nutrients are exhausted to enhance the cell cycle regulator CtrA. However, it is known that CtrA levels are maintained in the presence of (p)ppGpp 26,28,47 , and our results raise the possibility that this effect is (at least partially) mediated by (induced) MopJ. In light of the fact that the number of cells undergoing DNA replication is strongly diminished in stationary phase, with cells accumulating predominantly in G1 and G2 phase (Fig. 1), along with the fact that CtrA acts on replication control 4,12 , we suspect that CtrA is responsible or at least contributes to this replication arrest. In support of this, cells no longer exhibit a normal stationary phase arrest in the absence of MopJ (Fig. 1), and the abundance of CtrA is strongly reduced under these conditions (Fig. 2). Such stationary-phase-induced replication arrest may be further accentuated by other (p)ppGpp-dependent mechanisms, such as the reduction of the replication initiator DnaA, an unstable protein 62,63 , when (p)ppGpp levels are high 26,28,47 . It is also noteworthy that in contrast to this specific cell cycle control mechanism on the level of a master regulator, (p)ppGpp induces a general growth arrest and a state of antibiotic tolerance (known as persistence) in E. coli and likely other bacteria through a shutdown of macromolecular synthesis by type II toxin-antitoxin systems that are activated indirectly upon accumulation of (p)ppGpp 64 .
(p)ppGpp-induced replication arrest mechanism(s) are also operational in other a-Proteobacteria 25 , as a nutritional downshift induces a cell cycle arrest in S. meliloti 65 . Although CtrA is not known to bind the origin of replication in Sinorhizobium fredii directly 10 , CtrA may inhibit replication initiation indirectly as was recently also suggested to be the case for Caulobacter 13 and reinforcement of CtrA by induced MopJ presents an appealing and conceivable possibility. This is mechanistically distinct from the (p)ppGpp-based replication arrest at the level of DNA primase described for the Gram-positive soil bacterium Bacillus subtilis 66 , a member of the Firmicutes, clearly showing that evolution has found elegant solutions to the same problem.
Swarmer cell isolation, electroporation, biparental mating and bacteriophage fCr30-mediated generalized transduction were performed as described in ref. 5. Briefly, swarmer cells were isolated by Ludox or Percoll density-gradient centrifugation at 4°C, followed by three washes and final re-suspension in prewarmed (30°C) M2G. Electroporation was done using 1 mm gap electroporation cuvettes at 1.5 kV in an Eppendorf 2510 electroporator and 6 ml exponential phase cells that had been washed three times in sterile water. Biparental matings were done using exponential phase E. coli S17-1 donor cells and Caulobacter recipient cells washed in PYE and mixed at 1:10 ratio on a PYE plate. After 4-5 h of incubation at 30°C, the mixture of cells was plated on PYE harbouring nalidixic acid (to counter select E. coli) and the antibiotic that the conjugated plasmid confers resistance to. Generalized transductions were done by mixing 50 ml ultraviolet-inactivated transducing lysate with 500 ml exponential phase recipient cells, incubation for 2 h, followed by plating on PYE containing antibiotic to elect for the transduced DNA.
Bacterial strains, plasmids and oligonucleotides. Bacterial strains, plasmids and oligonucleotides used in this study are listed and described in Supplementary Tables 3-5, respectively.
Co-immunoprecipitation. When the culture (50 ml) reached OD 660nm ¼ 0.4-0.6 in the presence of xylose, cells were harvested by centrifugation at 6,000g for 10 min. The pellet was then washed in 10 ml of dilution/wash buffer (10 mM Tris-HCl (pH 7.5); 150 mM NaCl; 0.5 mM EDTA) and lysed for 15 min at room temperature in 1 ml of lysis buffer (dilution/wash buffer þ 0.5% NP40, 10 mM MgCl 2 , two protease inhibitor tablets (for 50 ml of buffer; Complete EDTA-free, Roche), 1 Â Ready-Lyse lysozyme (Epicentre), 50 U of DNase I )Roche)). Cellular debris was removed by centrifugation at 7,000g for 15 min at 4°C. The supernatant was incubated for 1 h (or overnight) at 4°C with GFP-Trap_A beads (ChromoTek GmbH) previously washed three times with 500 ml of dilution/wash buffer. The sample was then centrifuged at 2,500g for 2 min at 4°C and the supernatant was removed. The beads were washed three times with 500 ml of dilution/wash buffer and finally resuspended in 2 Â SDS sample buffer (50 mM Tris-HCl (pH 6.8), 2% SDS, 10% glycerol, 1% b-mercaptoethanol, 12.5 mM EDTA, 0.02% Bromophenol Blue), heated to 95°C for 10 min and stored at À 20°C.
Tandem affinity purification (TAP). The TAP procedure was based on that described by ref. 55 pH 7.4, 50 mM NaCl, 1 mM EDTA) and lysed for 15 min at room temperature in 10 ml of buffer II (buffer I þ 0.5% n-dodecyl-b-D-maltoside, 10 mM MgCl 2 , two protease inhibitor tablets (for 50 ml of buffer II; Complete EDTA-free, Roche), 1 Â Ready-Lyse lysozyme (Epicentre), 500 U of DNase I (Roche)). Cellular debris was removed by centrifugation at 7,000g for 20 min at 4°C. The supernatant was incubated for 2 h at 4°C with IgG Sepharose beads (GE Healthcare Biosciences) that had been washed once with IPP150 buffer (10 mM Tris-HCl at pH 8, 150 mM NaCl, 0.1% NP40). After incubation, the beads were washed at 4°C three times with 10 ml of IPP150 buffer and once with 10 ml of TEV cleavage buffer (10 mM Tris-HCl at pH 8, 150 mM NaCl, 0.1% NP40, 0.5 mM EDTA, 1 mM DTT). The beads were then incubated overnight at 4°C with 1 ml of TEV solution (TEV cleavage buffer with 100 U of TEV protease per millilitre (Promega)) to release the tagged complex. CaCl 2 (3 mM) was then added to the solution. The sample with 3 ml of calmodulin-binding buffer (10 mM b-mercaptoethanol, 10 mM Tris-HCl at pH 8, 150 mM NaCl, 1 mM magnesium acetate, 1 mM imidazole, 2 mM CaCl 2 , 0.1% NP40) was incubated for 1 h at 4°C with calmodulin beads (GE Healthcare Biosciences) that previously had been washed once with calmodulin-binding buffer. After incubation, the beads were washed three times with 10 ml of calmodulin-binding buffer and eluted five times with 200 ml IPP150 calmodulin elution buffer (calmodulin-binding buffer substituted with 2 mM EGTA instead of CaCl 2 ). The eluates were then concentrated using Amicon Ultra-4 spin columns (Ambion).
b-Galactosidase assays. b-Galactosidase assays were performed at 30°C as described previously 3  MopJ purification and production of antibodies. MopJ-short protein, lacking the first N-terminal 45 residues, was expressed from pET28a in E. coli Rosetta (DE3)/ pLysS (Novagen) and purified under native conditions using Ni 2 þ chelate chromatography. A 5-ml overnight culture was diluted into 1 l of pre-warmed LB. When cells reached OD 660nm ¼ 0.3-0.4, 1 mM isopropyl-b-D-thiogalactoside was added to the culture and growth continued. After 3 h, cells were pelleted and resuspended in 25 ml of lysis buffer (10 mM Tris HCl (pH 8), 0.1 M NaCl, 1.0 mM b-mercaptoethanol, 5% glycerol, 0.5 mM imidazole Triton X-100 0.02%). Cells were sonicated (Sonifier Cell Disruptor B-30; Branson Sonic Power. Co.) on ice using 12 bursts of 20 s at output level 5.5. After centrifugation at 4,300g for 20 min, the supernatant was loaded onto a column containing 5 ml of Ni-NTA agarose resin pre-equilibrated with lysis buffer. Column was rinsed with lysis buffer, 400 mM NaCl and 10 mM imidazole, both prepared in lysis buffer. Fractions were collected (in 300 mM Imidazole buffer, prepared in lysis buffer) and used to immunize New Zealand white rabbits (Josman LLC).
Microscopy. PYE or M2G cultivated cells in exponential growth phase were immobilized using a thin layer of 1% agarose. For time-lapse experiments, synchronized cells were immobilized using a thin layer of 1% agarose in M2G supplemented with 0.4% PYE. Fluorescence and contrast microscopy images were taken with an Alpha Plan-Apochromatic Â 100/1.46 DIC(UV) VIS-IR oil objective on an Axio Imager M2 microscope (Zeiss) with acquisition at 535 nm (enhanced green fluorescent protein), 580 nm (yellow fluorescent protein (YFP)) and 480 nm (cyan fluorescent protein (CFP); Visitron Systems GmbH) and a Photometrics Evolve camera (Photometrics) controlled through Metamorph V7.5 (Universal Imaging). Images were processed using Metamorph V7.5.
Fluorescence-activated cell sorting. FACS experiments were performed as described previously 50 . Cells in exponential growth phase (OD 660nm ¼ 0.3-0.6) or in stationary phase (diluted to obtain an OD 660nm ¼ 0.3-0.6), cultivated in M2G, were fixed in ice-cold 70% ethanol solution. Fixed cells were re-suspended in FACS staining buffer, pH 7.2 (10 mM Tris-HCl, 1 mM EDTA, 50 mM NaCitrate, 0.01% Triton X-100) and then treated with RNase A (Roche) at 0.1 mg ml À 1 for 30 min at room temperature. Cells were stained in FACS staining buffer containing 0.5 mM of SYTOX Green nucleic acid stain solution (Invitrogen) and then analysed using a BD Accuri C6 flow cytometer instrument (BD Biosciences). Flow cytometry data were acquired and analysed using the CFlow Plus V1.0.264.15 software (Accuri Cytometers Inc.). 20,000 cells were analysed from each biological sample. The forward scattering (FSC-A) and Green fluorescence (FL1-A) parameters were used to estimate cell sizes and cell chromosome contents, respectively. Experimental values represent the averages of three independent experiments. Relative chromosome number was directly estimated from the FL1-A value of NA1000 cells treated with 20 mg ml À 1 Rifampicin for 3 h at 30°C, done in ref. 50. Rifampicin treatment of cells blocks the initiation of chromosomal replication, but allows ongoing rounds of replication to finish.
In vivo phosphorylation-immunoprecipitation experiments. In vivo phosphorylation measurements by 32 P labelling of WT cultures harbouring pMT335 or pMT335-mopJ after induction with vanillate (50 mM), followed by immunoprecipitation with antibodies to DivK as described in the study by Radhakrishnan et al. 3 . Briefly, a freshly grown colony was picked from a PYE plate, washed with M5G medium lacking phosphate and was cultivated overnight in M5G with 0.05 mM phosphate to an optical density of 0.3 at 660 nm. One millilitre of culture was labelled for 4 min at 28°C using 30 mCi of g-[32P]ATP. Upon lysis, proteins were immunoprecipitated with 3 ml of anti-DivK antiserum and Protein A agarose (Roche, CH) and the precipitates were resolved by SDS-polyacrylamide gel electrophoresis and radiolabelled DivK was quantified.
Strain constructions NA1000 DmopJ. Deletions were introduced using SacB-based counterselection using 3% sucrose. Briefly, pNPTS138-DmopJ-KO was first introduced into NA1000 (WT) by intergeneric conjugation and then plated on PYE harbouring kanamycin (to select for recombinants) and nalidixic acid to counter select E. coli donor cells 5 . A single homologous recombination event at the CCNA_00999 locus of kanamycin-resistant colonies was verified by PCR. The resulting strain was grown to stationary phase in PYE medium lacking kanamycin. Cells were then plated on PYE supplemented with 3% sucrose and incubated at 30°C. Single colonies were picked and transferred in parallel onto plain PYE plates and PYE plates containing kanamycin. Kanamycin-sensitive cells, which had lost the integrated plasmid due to a second recombination event, leaving a deleted version of mopJ behind (DmopJ), were then identified for disruption of the mopJ locus by PCR.
NA1000 DspmXDpodJ. pNPTS138-DspmX-KO 3 was first introduced into NA1000 DpodJ 36 by intergeneric conjugation and then plated on PYE harbouring kanamycin (to select for recombinants) and nalidixic acid to counter select E. coli donor cells. A single homologous recombination event at the CCNA_02255 locus of kanamycin-resistant colonies was verified by PCR. The resulting strain was treated as above.
Strains harbouring pMT335 or pMT335-derived plasmids. pMT335 or pMT335-mopJ was introduced by electroporation and then plated on PYE plates harbouring gentamycin.  19 allele was introduced into NA1000 DsocB as described in ref. 54 by fCr30-mediated transduction followed by selection on plates harbouring spectinomycin.
Strains harbouring pMR10 or derivatives. Plasmids were introduced by electroporation, followed by selection of transformants on PYE plates harbouring tetracycline (pMR20) or kanamycin (pMR10).
Strains harbouring xylX::P xyl -mopJ-GFP. pCWR282 (pXGFP4-mopJ) was introduced into C. crescentus by electroporation and then plated on PYE harbouring kanamycin (to select for recombinants). A single homologous recombination event at the xyl locus of kanamycin-resistant colonies was verified by PCR.
NA1000 mopJ::P mopJ -mopJ-GFP. pGFP4-mopJ was introduced into NA1000 DmopJ by electroporation and then plated on PYE harbouring kanamycin (to select for recombinants). A single homologous recombination event at the mopJ locus of kanamycin-resistant colonies was verified by PCR.
NA1000 xylX::P xyl -GFP-mopJ. pXGFP4-C1-GFP-mopJ was introduced into NA1000 by electroporation and then plated on PYE harbouring kanamycin. A single homologous recombination event at the xyl locus of kanamycin-resistant colonies was verified by PCR.

Plasmid constructions
pCWR296 (pNPTS138-DmopJ-KO). The plasmid construct used for mopJ (CCNA_00999) deletion was made by PCR amplification of two fragments. The first, a 568-bp fragment (nt 1081464-1082022, NA 1000 genome coordinates, flanked by an EcoRI site at the 5 0 end and a BamHI at the 3 0 end) was amplified using primers delmopJ_1-BamHI and delmopJ_1-EcoRI (sequences for all the primers used in this work can be found in Supplementary Table 5). The second, a 447-bp fragment (nt 1082446-1082880, flanked by a BamHI site at the 5 0 end and a HindIII site at the 3 0 end) was amplified using primers delmopJ_2-HindIII and delmopJ_2-BamHI. These two fragments were first digested with appropriate restriction enzymes and then triple ligated into pNTPS138 (M.R.K. Alley, Imperial College London, unpublished) that had been previously restricted with HindIII and EcoRI. This construct deletes 423 nt of the mopJ-coding sequence (nt 1082023-1082445, or codons 14-154 of the annotated CCNA_00999 coding sequence).
pMT335-divJ329. The divJ329-coding sequence (short divJ form without sequence encoding for dimerization and kinase domains, nt 1220499-1221485) was PCR amplified from the NA1000 strain using the divJ-NdeI and divJ329-MunI primers. This fragment was digested with NdeI/MunI and cloned into NdeI/EcoRI-digested pMT335.
pXGFP4-divJ329. The divJ329-coding sequence (short divJ form without sequence encoding for dimerization and kinase domains, nt 1220499-1221485) was PCR amplified from the NA1000 strain using the divJ-NdeI and divJ329fusion-MunI primers. This fragment was digested with NdeI/MunI and cloned into NdeI/EcoRIdigested pXGFP4.
pGFP4-mopJ. The mopJ-coding sequence without stop codon and the upstream region of 640 bp (nt 1081990-1082478) were PCR amplified from the NA1000 strain using the mopJ-NheI and mopJfusion-BamHI primers. This fragment was digested with NheI/BamHI and cloned into pXGFP4, previously digested with NheI/BamHI to remove the PxylX locus.