Regulation of the activity of the bacterial histidine kinase PleC by the scaffolding protein PodJ

Scaffolding proteins can customize the response of signaling networks to support cell development and behaviors. PleC is a bifunctional histidine kinase whose signaling activity coordinates asymmetric cell division to yield a motile swarmer cell and a stalked cell in the gram-negative bacterium Caulobacter crescentus. Past studies have shown that PleC’s switch in activity from kinase to phosphatase correlates with a change in its subcellular localization pattern from diffuse to localized at the new cell pole. Here we investigated how the bacterial scaffolding protein PodJ regulates the subcellular positioning and activity of PleC. We reconstituted the PleC-PodJ signaling complex through both heterologous expressions in Escherichia coli and in vitro studies. In vitro, PodJ phase separates as a biomolecular condensate that recruits PleC and inhibits its kinase activity. We also constructed an in vivo PleC-CcaS chimeric histidine kinase reporter assay and demonstrated using this method that PodJ leverages its intrinsically disordered region to bind to PleC’s PAS sensory domain and regulate PleC-CcaS signaling. Regulation of the PleC-CcaS was most robust when PodJ was concentrated at the cell poles and was dependent on the allosteric coupling between PleC-CcaS’s PAS sensory domain and its downstream histidine kinase domain. In conclusion, our in vitro biochemical studies suggest that PodJ phase separation may be coupled to changes in PleC enzymatic function. We propose that this coupling of phase separation and allosteric regulation may be a generalizable phenomenon among enzymes associated with biomolecular condensates.

Scaffolding proteins can customize the response of signaling networks to support cell development and behaviors. PleC is a bifunctional histidine kinase whose signaling activity coordinates asymmetric cell division to yield a motile swarmer cell and a stalked cell in the gram-negative bacterium Caulobacter crescentus. Past studies have shown that PleC's switch in activity from kinase to phosphatase correlates with a change in its subcellular localization pattern from diffuse to localized at the new cell pole. Here we investigated how the bacterial scaffolding protein PodJ regulates the subcellular positioning and activity of PleC. We reconstituted the PleC-PodJ signaling complex through both heterologous expressions in Escherichia coli and in vitro studies. In vitro, PodJ phase separates as a biomolecular condensate that recruits PleC and inhibits its kinase activity. We also constructed an in vivo PleC-CcaS chimeric histidine kinase reporter assay and demonstrated using this method that PodJ leverages its intrinsically disordered region to bind to PleC's PAS sensory domain and regulate PleC-CcaS signaling. Regulation of the PleC-CcaS was most robust when PodJ was concentrated at the cell poles and was dependent on the allosteric coupling between PleC-CcaS's PAS sensory domain and its downstream histidine kinase domain. In conclusion, our in vitro biochemical studies suggest that PodJ phase separation may be coupled to changes in PleC enzymatic function. We propose that this coupling of phase separation and allosteric regulation may be a generalizable phenomenon among enzymes associated with biomolecular condensates.
Eukaryotic scaffolds add layers of regulation upon signaling pathways that include allosteric mechanisms (1), feedback regulation (2), and phase separation into distinct compartments (3). Scaffold protein phase separation occurs when a protein exhibits weak multivalent interactions with itself or other biomolecules (4). These weak multivalent interactions facilitate the formation of dense and liquid-like droplet assemblies. Given that the assemblies have diameters of hundreds of nanometers to micrometers and recruit and concentrate several proteins, they have also been coined "membraneless organelles" (5). Although phase separation has become a paradigm for spatially organizing biochemistry within eukaryotic cytoplasm (6), examples of phase separation have just recently emerged in bacteria (4,7). Phase separation has been implicated in organizing a variety of enzymes and biochemical processes in bacteria that include RNA polymerase (8), mRNA decay (9,10), cell polarity (11,12), ABC transporters (13), DNA repair (14), cell division (15,16), chromosome segregation (17), metabolic enzymes associated with day-night cycles (18) and carboxysome assembly (19).
In contrast to eukaryotic scaffolds, the mechanism of how scaffolding proteins impact bacterial signaling pathways has been less studied. Three scaffolds play roles in regulating the subcellular position of histidine kinases involved in the asymmetric cell division of the gram-negative bacterium Caulobacter crescentus: PopZ-CckA (20), SpmX-DivJ (21), and PodJ-PleC (22,23). PopZ (11) and SpmX (12) phase separate as a biomolecular condensate that recruits distinct signaling protein clients (20). At the opposite new cell pole, scaffold PodJ sequesters four distinct signaling proteins directly or indirectly: PleC, PopA, DivL, and CpaE (23). Past in vivo fluorescence recovery after photobleaching and singlemolecule tracking studies have shown that the new cell pole forms distinct compartments mediated by biomolecular condensates (11,12,20). The primary scaffold that phase separates to generate distinct compartments at each cell pole is the scaffold protein PopZ (11,12,20). A key question to consider is the roles of proteins at each of the cell pole localized biomolecular condensates. Do other proteins within these in vivo biomolecular condensates serve as a scaffold, like PopZ (11), and contribute to phase separation? Or do they serve as clients that do not impact phase separation? Here we examine the in vitro phase properties of the PodJ scaffold and consider how it impacts its signaling protein client PleC.
Coordination of C. crescentus replication, cell growth, and division requires the bifunctional cell-cycle kinase CckA to undergo a kinase-to-phosphatase switch during each cell cycle (Fig. 1). The activity changes of CckA are facilitated by the bifunctional histidine kinase PleC that also oscillates between kinase and phosphatase activity states (24,25) (Fig. 1). Several factors contribute to PleC's kinase to phosphatase switch. One factor is that unphosphorylated DivK allosterically stimulates the kinase activity of PleC (26). Second, PleC signaling activity is regulated by pilus retraction upon surface contact (27). In this model, pilus retraction leads to the accumulation of PilA monomers in the periplasm that interacts and regulates PleC activity (27). A third factor that correlates with PleC's activity is its subcellular localization pattern (Fig. 1). When PleC is localized at the new cell pole, it functions as a phosphatase (28). In contrast, when PleC is released from the cell pole in the swarmer-to-stalk transition, PleC functions as a kinase (26).
The cell pole localization pattern of PleC depends on PodJ (22,29). The PodJ spans the membrane and within the cytoplasm includes six coiled coils, an intrinsically disordered region (IDR) (Figs. 1B and S1, A and B) featured by negatively and positively charged blocks at the N and C termini, respectively, and a transmembrane (TM) anchor. Within the periplasm, PodJ contains a tetratricopeptide repeat domain and a peptidoglycan-binding domain. In C. crescentus, deletion of podJ results in delocalized PleC and downregulation of the CtrA pathway (23,29,30). Moreover, PodJ variants lacking either the cytoplasmic or the periplasmic domains also resulted in downregulation of the CtrA pathway (22) and loss of PleC cell pole localization (30). Thus, both PodJ's cytoplasmic and periplasmic domains contribute to the regulation of CtrA activity (30).
The full-length PodJ L (full length, 1-974) is expressed in the swarmer to stalk transition (Fig. 1, B and C). Then upon cell division, full-length PodJ L is proteolyzed into a shortened form PodJ S (short form, 1-702) (31,32), via a set of proteases (23,31,32). Both PodJ L and PodJ S can support cell pole localization of PleC (30). However, the proteolysis of PodJ correlates with the downregulation of the CtrA regulon (30). Therefore, directly or indirectly, the expression and proteolysis of PodJ play a central role in regulating the master regulator CtrA (22,29).
Here, we applied synthetic biology and in vitro biochemical approaches to understand PodJ's role within the new cell pole biomolecular condensate. Our results indicate that PodJ serves as a scaffold that phase separates as a biomolecular condensate in vitro that recruits PleC and inhibits PleC kinase activity. Through a complementary synthetic biology strategy, we designed and built a PleC-CcaS chimeric histidine kinase reporter assay that identified domains necessary for the PodJ stimulation of PleC. Our results suggest a model in which PodJ phase separates as a biomolecular condensate that recruits and regulates PleC signaling activity.

PodJ coiled coils contribute to cell pole localization
To understand how PodJ impacts PleC's function, we reconstituted the PodJ-PleC signaling complex in Escherichia coli. Past studies have shown that heterologous expression of PodJ in E. coli resulted in cell pole accumulation (20). Of note, the gamma-proteobacterium E. coli is divergent from the alphaproteobacterium C. crescentus and does not contain any C. crescentus polarity protein homologs (33 has been used extensively as a heterologous system for testing C. crescentus protein-protein interactions (20,34). Previous studies have also shown that an N-terminal YFP fusion to PodJ does not disrupt its regulation of the CtrA pathway in C. crescentus (29). Therefore, to reconstitute the PodJ-PleC complex, we heterologously expressed an N-terminal fluorescent fusion protein of PodJ(YFP-PodJ) in E. coli and determined its subcellular localization pattern. As shown in Fig. S2A, YFP-PodJ accumulated at the cell poles. We also found that PodJ variants lacking the periplasmic domains, transmembrane tether, IDR, or coiled coil 4 to 6 (CC4-6) also accumulated at the cell poles. In contrast, PodJ variants lacking coiled coil 1 to 3 (CC1-3) did not accumulate at the cell poles in E. coli (Fig. S2A).
In C. crescentus, Lawler et al. (30) observed that both the cytoplasmic and the periplasmic domains alone of PodJ could accumulate at the cell poles. Therefore, PodJ has at least two independent mechanisms for localizing to the cell poles. We, therefore, interrogated the role of CC1-3 in the subcellular localization of PodJ's cytoplasmic domains, YFP-PodJ(1-635). As in E. coli, we observed that YFP-PodJ(1-635) accumulated at the cell poles in C. crescentus ( Fig. 2A). In contrast, we observed that sfGFP-PodJ(250-635) was diffuse or patchy when expressed as a sole copy in C. crescentus ( Fig. 2A). This suggests that CC1-3 is critical for the ability of PodJ's cytoplasmic domains to accumulate at the cell poles.

PleC localizes to the cell pole via PodJ's IDR
With an understanding of the domains that contribute to PodJ subcellular localization ( Figs. 2A and S2A), we then interrogated if YFP-PodJ recruits PleC-mCherry to the cell poles in E. coli. The expression of PleC alone resulted in a diffuse localization pattern (Fig. S2A). In contrast, coexpression of PodJ and PleC resulted in colocalization at the cell poles in 98% of cells. To interrogate the interaction between YFP-PodJ and PleC-mCherry, we heterologously coexpressed full-length PleC-mCherry with a library of YFP-PodJ domain deletion variants in E. coli (Fig. S2A). Deletion of the periplasmic domains, the TM or CC4-6, did not affect the cell pole localization of YFP-PodJ or PleC-mCherry recruitment to the cell poles. In contrast, the YFP-PodJΔIDR variant was unable to recruit PleC-mCherry to the cell poles (Fig. S2A  This suggests that PodJ's IDR may be a site of interaction with PleC-mCherry in E. coli. We next tested if PodJ's IDR was critical for PleC localization in C. crescentus (Fig. 2B). Past work has shown that PleC-GFP localized at the new cell pole in predivisional cells (28) and was dependent on PodJ (30). Consistent with these findings, we observed that, in a podJ deletion background, PleC-mCherry exhibited a diffuse localization pattern (Fig. 2B). Expression of sfGFP-PodJΔIDR accumulated at the cell poles in C. crescentus (Fig. 2B). However, sfGFP-PodJΔIDR was unable to recruit PleC-mCherry to the cell poles (Fig. 2B). Our observation is consistent with previous PodJ domain analysis in C. crescentus (30) which indicated that a portion of the IDR of PodJ contributes to PleC's new cell pole localization in C. crescentus (23,30). Therefore, PodJ serves as a scaffold that recruits PleC as a client through its IDR.

PleC localizes at the cell pole via its tandem PAS sensor
We next asked which domains within PleC serve as the site of interaction with PodJ. PleC contains a periplasmic Cache domain, two cytoplasmic PAS domains in tandem, and a histidine kinase domain. Thus, we heterologously expressed a set of PleC-mCherry domain deletion variants and YFP-PodJ in E. coli (Fig. S2B). We observed that PleC-mCherry variants that lack the periplasmic Cache domain or the HK domains colocalized with YFP-PodJ. In contrast, PleCΔPAS-AB-mCherry displayed a diffuse localization pattern. This suggests that PleC's PAS-AB is required for recruitment to the cell pole by PodJ in E. coli.
To corroborate the importance of PAS-AB, we examined the subcellular localization of PleCΔPAS-AB-mCherry or PleC-mCherry in a podJ and pleC deletion C. crescentus background. We observed that the PodJ-PleC complex colocalized in predivisional cells at the new cell pole (Fig. 2C). In contrast, in C. crescentus strains expressing sfGFP-PodJ and PleCΔPAS-AB-mCherry, the PleC variant displayed a diffuse subcellular localization pattern. This suggests that the PleC-PAS-AB: PodJ-IDR protein-protein interaction is also critical for PleC recruitment to the cell pole in C. crescentus.
To determine if the PodJ-PleC interaction is conserved across alphaproteobacteria that encode both PleC and PodJ, we heterologously expressed PodJ and PleC homologs from Agrobacterium tumefaciens (Atu) and Hyphomonas neptunium (Hyp) individually and together in E. coli (Fig. 2D). Both YFP-AtuPodJ and YFP-HypPodJ accumulated at the cell poles in E. coli, while heterologous expression of AtuPleC-CFP or HypPleC-CFP individually resulted in a diffuse localization pattern. However, each PodJ homolog colocalized with its cognate PleC at the cell pole ( Fig. 2D) upon coexpression. Therefore, the ability of PleC and PodJ to colocalize at the cell poles in E. coli is conserved among these species. Within a representative set of 11 alphaproteobacterial species, we found that the presence of pleC's cytoplasmic PAS-A and PAS-B sensory domains correlates with the existence of podJ and pilA homologs in the genome (Fig. S3). This coconservation agrees with the requirement of PAS-AB for PleC's recruitment to the cell poles (Figs. 2, D and E and S2B).
In addition, we found that each of the studied PodJ homologs contains a putative IDR (Fig. S4). These IDRs vary in length from 242 to 261 residues and are flanked by coiled coilrich regions (Fig. S4A). Although the sequences of the IDR are nonconserved, each IDR is composed of two distinct regions (Fig. S4B). One IDR region is rich in negative charges with pI (isoelectric point) between 3.4 and 3.9. In contrast, an adjacent IDR within each PodJ homolog is rich in positive charges with pIs between 9.9 and 10.9. This conservation of charge pattern within the IDRs may play a role in PodJ function.

Design of a PleC-CcaS chimeric histidine kinase
Owing to PodJ interacting with the PleC's sensory domains (Figs. 2 and S2), we hypothesized that PodJ might also regulate PleC function. Complete reconstitution of the PleC signaling network would require the addition of PodJ, PleC, DivK, DivL, CckA, ChpT, CtrA, and a CtrA regulated promoter. We, therefore, leveraged the demonstrated technique of exchanging kinase sensory domains to construct a chimeric histidine kinase to simplify our approach (35,36). We applied this approach using the green light-sensing CcaS-CcaR system (37) to interrogate the responsiveness of PleC's sensor to PodJ expression.
We designed, built, and tested a PleC-CcaS chimeric histidine kinase library (Fig. S5). Each variant consisted of PleC's tandem PAS sensor fused to different junction sites within the histidine kinase domain of CcaS (Figs. S3, A and B and S5A). The pleC-ccaS chimera was coexpressed with its cognate response regulator ccaR. Upon phosphorylation, phosphor-CcaR activates transcription of mCherry from the cpcG2 promoter (Fig. 3B). We found that the YFP-PleC-CcaS retained its ability to colocalize with CFP-PodJ in E. coli (Fig. 3C). In the absence of PodJ stimulation, the PleC-CcaS chimera exhibited little mCherry expression. In comparison, mCherry expression was highest using chimera AB-1 upon coexpression with PodJ ( Fig. S5B). Therefore, the engineered PleC-CcaS chimera retains the on-switch function ( Fig. 3D) of the parent green light-sensing CcaS (37). The PleC-CcaS chimera AB-1 connects PAS-AB to CcaS between the D and V of the conserved DVT hinge motif (38) (Fig. S6, A-C). These results suggest that we have engineered a functional PodJ-responsive PleC-CcaS chimeric histidine kinase.

Stimulation of PleC-CcaS activity requires PodJ's IDR
To determine whether PleC-CcaS stimulation depended upon specific interactions with PodJ, we compared the effects of coexpressing the YFP-PodJ domain deletion variants with the PleC-CcaS chimera (Fig. 3D). Before the analysis, we confirmed that the YFP-PodJ variant displayed fluorescence over 10-fold greater than the empty vector (Fig. S7A). Relative to the empty vector control, the expression of PodJ stimulated the PleC-CcaS chimera and resulted in a 10-fold increase in mCherry expression (Fig. 3D). However, a PodJ variant that lacks the IDR, YFP-PodJ-ΔIDR, was unable to stimulate PleC-CcaS-mediated mCherry expression. In contrast, expression of the YFP-IDR alone led to a 3-fold activation of mCherry expression (Fig. 3D). This suggests that stimulation of the PleC-CcaS chimera is dependent upon the interaction with PodJ's IDR. However, full stimulation of PleC-CcaS requires the entire cytoplasmic domain, suggesting a role of the coiled coils in PleC-CcaS regulation.

PodJ S is unable to stimulate the PleC-CcaS chimera
Next, we asked if the PleC-CcaS chimera could be stimulated by PodJ S , the primary form in C. crescentus swarmer cells. Past experiments indicated that loss of the periplasmic domain leads to reduced expression of CtrA-mediated genes in C. crescentus (23). We observed that PodJ L and PodJ S could recruit PleC to the cell poles when heterologously expressed in E. coli (Fig. S2A), consistent with the past observation in C. crescentus (22). Therefore, the downregulation of the CtrA pathway upon PodJ proteolysis is not due to a loss of PleC-PodJ binding. Unlike the expression of full-length PodJ L we observed that expression of PodJ S was unable to stimulate mCherry expression via the PleC-CcaS chimera (Fig. 3D).
The loss of PodJ S stimulation of PleC could result from an altered PodJ transmembrane anchoring that allosterically affects the IDR-PAS conformational state. Therefore, we also examined if PodJ variants lacking the TM tether (PodJ(1-635)) could relieve steric effects on the IDR-PAS conformation. We observed that the cytoplasmic PodJ variant lacking the TM tether stimulated mCherry expression in the PleC-CcaS reporter assay (Fig. 3D). This indicates that PodJ's cytoplasmic domains alone can stimulate PleC-CcaS function. Meanwhile, it suggests that the transmembrane region may regulate the PodJ-PleC signaling complex, consistent with the reduced CtrA pathway activation of strains expressing PodJ S in C. crescentus (23,30).
We next asked how cell pole accumulation of the PodJ-PleC complex impacted PleC-CcaS signaling. The construct lacking the N-terminal coiled coils, PodJ(250-635), does not accumulate at the cell poles in E. coli (Fig. S2) or C. crescentus (Fig. S1). We found that PodJ(250-635) led to a 3-fold increase in mCherry expression compared with PodJ(1-635), which stimulated a 10-fold increase. These results indicate that cell pole accumulation of PodJ is not required for PleC-CcaS stimulation but may impact the degree of PleC-CcaS stimulation.

Stimulation of PleC-CcaS requires functional PAS domains
Given that PodJ recruits PleC through its cytoplasmic tandem PAS sensory domains, we hypothesized that the regulation of PleC-CcaS kinase domains requires sensory domain stimulation. Our sequence analysis indicated that PleC's PAS A and B domains share low similarities (47.88%) (Fig. S6A). In contrast, PAS-A homologs across alphaproteobacteria exhibit 68.3% sequence similarity, whereas PAS-B homologs exhibit 64.4% (Fig. S6D). Therefore, we suspect that each PAS domain may have a distinct function.
Previous studies have shown that the signal transmission motif D-I/V-T at the C terminus of the PAS domain allosterically relays signals from the central binding cavity to the Cterminal coiled-coil linker (39). These studies showed that mutations of the Asp to Ala or the Thr to Ala within this motif disrupted the PAS sensor's sensitivity to signal stimulation (36,38). Similarly, we hypothesized that PodJ's stimulation of PleC activity also requires PleC's PAS-A and PAS-B conserved motifs (39) (Fig. S6).
We, therefore, generated PAS-A (D433A, V435A) and PAS-B (D548A, V550A) hinge motif PleC-CcaS variants to disrupt signal flow within each PAS domain (Fig. 4A). Before analysis, we confirmed the expression of PodJ by measuring the fluorescence intensity of N-terminal YFP (Fig. S7B). The wildtype PleC-CcaS chimera displayed a 10-fold increase in mCherry expression upon PodJ induction. In contrast, PleC-CcaS PAS-A and PAS-B variants showed little to no stimulation by the expression of PodJ L (Fig. 4B). Also, compared with wildtype, the changes to the PAS domain motif appear to lock the downstream kinase into a state with a low activity, which was insensitive to PodJ expression. Thus, the reduction of stimulation by PodJ could be due to a loss of PAS domain allostery or loss of PodJ-PleC colocalization. Moreover, each PAS domain should be functional to mediate PodJ stimulation of PleC-CcaS. Regulation of the kinase PleC by the scaffold PodJ
This large oligomer size could be poised to mediate weak multivalent interactions that promote phase separation. Indeed, we observed in vitro that the sfGFP-PodJ(1-635) protein phase separates as round protein-rich droplets with a saturation concentration between 1.5 and 2.0 μM (Figs. 5C and  S8A). This C sat is less than the estimated PodJ concentration in cells from ribosome profiling measurements of 2 to 5 μM (40). Moreover, time-lapse imaging revealed that the PodJ droplets fuse upon contact and gradually relax back to a spherical droplet over 12 min, demonstrating liquid-like properties in vitro (Fig. 5D).
To determine the impact of fluorescent protein interactions, we observed that sfGFP did not form any visible droplets (Figs. 5C and S8B). In addition, SNAP-tag fused to the Nterminal PodJ(1-635) also formed droplets at the same condition with a smaller size (Fig. S8C). These results indicate that PodJ's droplet formation does not require sfGFP. However, weak fluorescent protein interactions likely mediate increases in protein droplet size.
To interrogate the biomolecular condensate material properties of PodJ, we examined the impact of 1,6-hexanediol, which is commonly used to disrupt weak hydrophobic interactions and dissolve biomolecular condensates (41). We observed that the addition of 10% 1,6-hexanediol disrupted droplet formation (Fig. S9A). Moreover, high concentrations of adenosine triphosphate (ATP) have been shown to suppress liquid-liquid phase separation assemblies through the interaction of ATP with hydrophobic or aromatic side chains that attenuate multivalent interactions of phase separating proteins (12,42,43). Similarly, we observed that the addition of 10 to 20 mM ATP or ADP led to a reduction in droplet size, partitioning ratio, and PleC-mCherry recruitment (Fig. S9, B and C). In contrast, we observed that the addition of physiological concentrations of 125 to 1000 μM ATP led to increased sfGFP-PodJ droplet size and partitioning (Fig. S9, B and D). This increase in droplet size and partitioning may suggest ATP binding to PleC, and subsequent biochemical activities may influence droplet size. In summary, the PodJ biomolecular condensate assembly properties are tunable by adding common small-molecule liquid-liquid phase separation regulators.

PodJ biomolecular condensate recruits PleC in vitro
We then considered if the in vitro sfGFP-PodJ(1-635) biomolecular condensates could recruit PleC-mCherry as a client. We observed that PleC-mCherry could readily accumulate within the PodJ-rich droplets, whereas mCherry alone was not enriched (Fig. 6A). In contrast, mCherry, mCherry + sfGFP,  PleC PAS AB-HK-mCherry, and mCherry + sfGFP-PodJ 1-635 did not form droplets under the same condition (Fig. S8B). We employed an in vitro fluorescence polarization assay to detect the interaction between PodJ's IDR and histidine kinases colocalized with PodJ at the new cell pole, including PleC PAS-AB domains, DivL, and CckA. We fluorescently labeled PodJ's IDR with a BODIPY dye and observed robust binding between PleC and PodJ's IDR. In contrast, we could not detect any interactions between PodJ's IDR and DivL and CckA (Fig. 6B). These fluorescence polarization studies indicate that PleC's tandem PAS domains interact specifically with the IDR from PodJ.
We next asked if the diminished phase separation capabilities of PodJ(250-635) impact its catalytic capabilities. We observed that the ADP production IC 50 for PodJ(250-635) was 2-fold higher than that for PodJ(1-635). Interesting, our imaging of sfGFP-PodJ(1-635) or sfGFP(250-635) indicated that low concentrations of PodJ protein do not phase separate and do not repress PleC kinase activity. In contrast, higher PodJ concentrations of both constructs led to phase separation and repression of PleC kinase activity (Fig. 6D). However, the addition of bovine serum albumin (BSA) resulted in no change in PleC ATP hydrolysis activity (Fig. 6C). Therefore, the interaction of PodJ with PleC specifically represses PleC's histidine kinase function. Although kinase activity repression typically correlates with increased phosphatase function (24,45,46), future biochemical studies should examine how PodJ impacts PleC's ATP binding, phosphatase, and phosphotransfer functions.
In vivo, PleC-CcaS chimera is inactive when expressed alone, and the addition of PodJ(1-635) stimulated kinase activity (Fig. 3D). By contrast, in vitro, PleC exhibited kinase activity in solution alone, but the addition of PodJ(1-635) represses PleC kinase activity (Fig. 6C). The differences in regulation are likely rooted in aspects of the PleC-CcaS design and the regulatory behavior of the CcaS histidine kinase. Of note, the two-parent kinases PleC and CcaS diverge in structure and regulatory mechanisms. The PodJ sensing PleC functions as an OFF switch, whereas the green light-sensing CcaS functions as an ON switch. Here we observed that the PleC-CcaS chimeric histidine kinase adopts the same ON-switch behavior as the parent CcaS histidine kinase.
In addition, past structure-function studies of PAS-linked histidine kinases have shown that alterations of the N or C terminus of PAS sensory domains can impact sensor functions as an ON switch or as an OFF switch upon signal stimulation (38,47). Other studies have also shown that the linker length between the sensory domain of CcaS and the HK can alter kinase signaling from an OFF to an ON state (48). Therefore, based on these past studies, the same modifications to PleC-CcaS versus PleC can also influence ON-switch versus OFF-switch histidine kinase function. Nevertheless, both experimental sets provide evidence that PleC's tandem sensory domain serves as a sensor for PodJ.

Discussion
Reconstitution of the PodJ-PleC complex in vitro (Figs. 5 and 6) and in E. coli (Figs. 2-4) demonstrated that PodJ recruits and regulates PleC function through a PAS-IDR interaction. Our studies demonstrated that in vitro the PodJ-PleC phase separates as a biomolecular condensate. Therefore, we propose that the multivalency of PodJ, in conjunction with the PopZ scaffold (11,12,20), promotes phase separation of the new cell pole biomolecular condensate. Within this new cell pole biomolecular condensate, PodJ recruits and regulates PleC function at the cell poles. We showed that the N-terminal coiled coils were critical for cell pole accumulation of PodJ's cytoplasmic domains in Caulobacter and E. coli (Fig. 2A).
This study leveraged a truncation of PodJ, PodJ(1-635), that lacks the transmembrane tether. This truncation accumulates at the cell pole in E. coli and C. crescentus ( Fig. 2A), recruits PleC to the cell poles (Fig. 2B), regulates PleC-CcaS function in E. coli (Fig. 3), and regulates PleC signaling in vitro. In addition, the PodJ(1-635) construct was accessible to protein purification, unlike PodJ S and PodJ L . A drawback of the PodJ(1-635) truncation is that it is not the biologically relevant and membrane-associated form of PodJ S or PodJ L. Unlike fulllength PodJ, the IDR of PodJ(1-635) is no longer directly attached to the membrane and this surface may introduce nonnative interactions that influence phase separation.
Future development of purification strategies for the fulllength membrane-associated PodJ will be needed for comparison. These studies will evaluate how PodJ's membrane tethering and its periplasmic domains impact PodJ phase separation and regulation of PleC signaling. Indeed, through PodJ truncation analysis in C. crescentus, Lawler et al. (30) have demonstrated that both the cytoplasmic and periplasmic domains of PodJ contribute to PodJ's ability to accumulate at the cell poles in C. crescentus. Moreover, additional studies are needed to conclude if full-length PodJ displays phase separation capabilities in vivo and if phase separation contributes to PodJ's ability to accumulate at the cell poles.
The discovery of PodJ's ability to directly regulate PleC function provides new data to understand how PodJ is logically connected to the cell cycle. Past work demonstrated that the cell-cycle master regulators control PodJ's expression (49) and proteolysis once per cell cycle (31,32). We observed that the shortened form of the PodJ scaffold retains the ability to recruit PleC to the cell poles. However, PodJ S no longer stimulates PleC-CcaS function. Thus, proteolytic remodeling of the PodJ within these new cell pole biomolecular condensates serves as a cell-cycle checkpoint signal to tune PleC function (Fig. 7). Upon complete degradation of PodJ S , PleC is liberated from the cell pole. In this diffuse state, PleC's activity is regulated by DivK allosteric stimulation (26) and pilus retraction (27). PleC kinase function stimulates downregulation of the CtrA pathway and the expression of new PodJ protein at the swarmer-to-stalk transition (49). This leads to stimulation of new cell pole localized PleC phosphatase function and activation of the CtrA pathway. Subsequently, robust activation of the downstream CtrA signaling pathway leads to the expression of PodJ-specific proteases (31,32) that proteolyze PodJ L into PodJ S . This proteolytic event tunes the functions of the PodJ biomolecular condensates resulting in phosphatase downregulation.
More broadly, phase separation provides an accessible compartmentalization strategy for the bacterial kingdom (7). The three scaffolding proteins, PopZ (11), SpmX (12), and PodJ (Figs. 4 and 5), phase separate as two distinct membraneless organelles at opposite ends of the cell. In addition, several other recent studies have shown that bacteria leverage phase separation to compartmentalize and regulate RNA polymerase (8), mRNA decay (9, 10), ABC transporters (13), DNA repair (14), cell division (15), chromosome segregation (17) and carboxysome assembly (19). These studies have laid the foundation to consider the extent and advantages of organizing biochemistry within biomolecular condensates in the bacterial cytoplasm.
Biomolecular condensates are generally thought to enhance enzyme reaction rates through mass action (50). Some recent studies have shown enhanced enzymatic rates, including RNA decapping (51), pyrenoid biochemistry (52) and SUMOylation (53). Here we have examined how biomolecular condensates regulate the histidine kinase activity of PleC. To determine the role of allostery, we found that a functional PAS sensor transmission motif was required for stimulation by full-length PodJ in E. coli (Fig. 4). This indicates that PodJ stimulation requires PleC-CcaS's sensor to kinase domain allosteric stimulation.
A comparison of PodJ-IDR or PodJΔCC1-3 versus fulllength PodJ stimulation of our in vivo PleC-CcaS reporter revealed a 3-fold versus 10-fold degree of stimulation. Given that neither PodJ-IDR nor PodJΔCC1-3 accumulated as a focus, this indicates that PodJ ability to bind to PleC in the dilute phase can regulate PleC-CcaS function. In contrast, the full-length PodJ variant that accumulates at the cell poles leads to an additional 3.3-fold activation of PleC-CcaS (Fig. 3). Comparison of PodJ-PleC constructs that exist solely in the dilute phase versus PodJ-PleC constructs that exist primarily in the protein-rich phase suggests that the contributions from mass action and the unique chemical environment contribute an additional 3.3-fold to PleC's activity in vivo.
Similarly, we observed that PodJ (1-635) could repress the activity of PleC in vitro under conditions where phase separation readily occurred (Fig. 6, C and D). In both cases, we suspect that there are interconnected contributions from mass action and kinase allostery. The localized high concentration of PodJ-PleC may promote an active PodJ-PleC conformation  Regulation of the kinase PleC by the scaffold PodJ in contrast to the lower PodJ-PleC concentration in the dilute phase. Similarly, it was shown that phase separation of the Saccharomyces cerevisiae RNA decapping complex Dcp1/ Dcp2 was coupled to conformational control of RNA decapping enzymatic activity (51). Two other histidine kinases, CckA (24) and DivJ (12), also show activation of enzyme function at high packing density on liposomes. In addition, both PleC and CckA require their Nterminal PAS sensory domain to mediate enzyme regulation at increased local concentration. Collectively these three studies suggest that strategies that concentrate histidine kinases, such as phase separation, may have a common effect of robustly stimulating low-copy histidine kinases (12,24). More broadly, the coupling of conformational change to the increased enzyme concentration (51) in biomolecular condensates may present a generalizable way to regulate enzyme function.
In summary, two-component systems provide bacterial cells with a customizable signaling platform. PleC has been customized for spatial control through phase separation and temporal control through sensing the cell-cycle produced and degraded signal PodJ. Moreover, PleC integrates both intraand extracellular signals (27) to coordinate growth and development. Therefore, the interplay of scaffolds, phase separation, and two-component systems provide avenues to orchestrate bacteria's complex development.

Bacterial strains
All experiments were performed using C. crescentus NA1000 (also known as CB15N) and E. coli DH5α and BL21 purchased from Promega. C. crescentus NA1000 was a kind gift from Dr Lucy Shapiro (Stanford University School of Medicine). PCR reactions and primers used for Gibson assembly are listed in Table S1. All relevant primers are given in detail in Table S2. Plasmids used in this study are listed in Table S3. More strains are listed in Table S4. Transformations were carried out as described (54). Details about plasmid and strain construction are listed in Text S1.

Plasmid cloning strategies
Fragments of target genes plasmid backbone were amplified via PCR using Phusion polymerase (Thermo Scientific). PCR reactions were performed in 50 μl reaction mixtures containing 3% (v/v) dimethyl sulfoxide, 1.3 M betaine, 0.3 μM each primer, 0.2 mM each dNTP, and 1 U Phusion High-Fidelity DNA Polymerase (Thermo Scientific). Both fragments were purified via gel extraction. Gibson assembly (55) reactions were performed in 20 μl with 100 ng backbone and typically a 1:10 backbone:insert ratio. A Gibson reaction master mix was prepared from 5x reaction buffer, T5 exonuclease (NEB), Phusion polymerase (NEB), Taq ligase (NEB) and stored as aliquots of 15 μl at −20 C. An annealing temperature of 55 C was used for all reactions, followed by 10 min at 4 C, and 10 μl of the reaction product was then transformed into chemically competent E. coli DH5a cells using the KCM transformation method. Oligonucleotide primers applied for amplification of the gene insert were designed using the j5 online program, and they featured overlaps of 26 bases to the insertion site in the plasmid (56). Oligonucleotides were synthesized by IDT, and all DNA sequencing reactions were performed by Genewiz. DNA oligos, plasmid construction methods, plasmids, and strains used in this study are listed in Tables S1-S4. Plasmids containing the CcaS-CcaR green light sensing system pJT119b was a gift from Jeffrey Tabor (Addgene plasmid # 50551; http://n2t.net/addgene:50551; RRID:Addgene_50551).

Growth and induction conditions
C. crescentus strains were grown at 28 C in peptone yeast extract. E. coli strains used for protein purifications and microscopy experiments were grown at 37 C in LB medium unless otherwise stated. When required, protein expression was induced by adding 0.002 to 0.5 mM isopropyl β-D-1thiogalactopyranoside (IPTG) or 0.5 to 10 mM arabinose in E. coli and 0.003% to 0.3% xylose or 0.005 to 0.5 mM vanillic acid in C. crescentus unless otherwise stated. The induction time for microscopy experiments is 0.5 to 2 h in E. coli and 3 to 5 h in C. crescentus.
Protein expression and purification of PodJ, PopZ, PleC, CckA, and DivL Protein expression for PodJ, PopZ, PleC, and CckA followed the same protocol described in detail below for PodJ(1-635). To purify the cytoplasmic portion of PodJ(1-635), Rosetta (DE3) containing plasmid pwz091 was grown in 6 L LB medium (30 μg/ml chloramphenicol and 100 μg/ml ampicillin) at 37 C. The culture was then induced at an A 600 of 0.4 to 0.6 with 0.5 mM IPTG overnight at 18 C. The cells were harvested and resuspended in the lysis buffer (50 mM Tris-HCl, 700 mM KCl, 20 mM Imidazole, 0.05% dextran sulfate, pH 8.0) in the presence of protease inhibitor cocktail tablets without EDTA (Roche). The cell suspension was lysed with three passes through an EmulsiFlex-C5 cell disruptor (AVESTIN, Inc), and the supernatant was collected by centrifuging at 12,000g for 30 min at 4 C. In addition, the insoluble cell debris was resuspended by the recovery buffer (50 mM Tris-HCl, 1000 mM KCl, 20 mM Imidazole, 0.05% dextran sulfate, pH 8.0), and its supernatant was collected as well as the previous centrifugation. The combined supernatants were loaded onto a 5-ml HisTrap HP column (GE Healthcare) and purified with the ÄKTA FPLC System. After washing with 10 volumes of wash buffer (50 mM Tris-HCl, 300 mM KCl, and 25 mM imidazole, pH 8.0), the protein was collected by elution from the system with elution buffer (50 mM Tris-HCl, 300 mM KCl, and 500 mM imidazole, pH 8.0). Then the protein was concentrated to a 3 ml volume using Amicon Centrifugal Filter Units, resulting in over 95% purity. Then the protein was dialyzed with a buffer containing 50 mM Tris-HCl (pH 8.0), 300 mM KCl and then aliquoted to a small volume (100 μl) and kept frozen at −80 C until use.
Labeling of SNAP-PodJ(1-635) with SNAP-Cell 505-Star SNAP-PodJ(1-635) was purified by following the protocol described above until the FPLC purification and Amicon centrifuge concentration step. Then the protein in the wash buffer was incubated with SNAP-Cell 505-Star in a 1:1.2 molar ratio at 0 C for 3 h. Next, the mixture was dialyzed against a dialysis buffer containing 50 mM Tris-HCl (pH 8.0), 300 mM KCl, then aliquoted to a small volume (20 μl) and kept frozen at −80 C until ready to use for imaging.

Fluorescence polarization assay
To label PodJ-IDR (471-635), we cloned a cysteine just after the 6X-His-tag proteins at the N terminus of each protein.
Cys-PodJ-IDR expression and purification followed the same protocol as PodJ mentioned above. These two proteins were labeled at the cysteine using thiol-reactive BODIPY FL N-(2-Aminoethyl) Maleimide (Thermo Fisher). The proteins were mixed with 10-fold excess BODIPY FL N-(2-Aminoethyl) Maleimide and allowed to react for 2 h at room temperature. The unreacted dye was quenched with mercaptoethanol (5% final concentration). The labeled proteins were purified via dialysis to remove unreacted fluorescent dye (5 times, 500 ml buffer, and 30 min each).
Fluorescence polarization binding assays were performed by mixing 100 nM labeled proteins with 0, 0.25, 0.5, 1, 2, 4, 8, 16 μM partner protein (PopZ, CckA, PleC, DivL, or BSA) for 45 min to reach binding equilibrium at the room temperature. Fluorescent proteins were excited at 470 nm, and emission polarization was measured at 530 nm in a UV-visible Evol 600 spectrophotometer (Thermo Fisher). Fluorescent polarization measurements were performed in triplicates, and three independent trials were averaged with error bars representing the standard deviation.
Quantification and statistical analysis FIJI/ImageJ (57) and MicrobeJ (58) were used for image analysis. More than 100 representative droplets were selected for partitioning ratio calculation, and each droplet's fluorescent intensity inside was divided by the background intensity outside. The mean and standard deviation for each measurement is shown. The number of replicates and the number of cells analyzed per replicate is specified in corresponding legends. All experiments were replicated three times, and statistical comparisons were carried out using GraphPad Prism with two-tailed Student's t tests. Differences were significant when p values were below 0.05. In all figures, measurements are shown as mean ± standard deviation (SD).

Fluorescence microscopy imaging of biomolecular condensates
sfGFP-PodJ(1-635), PleC PAS AB-HK-mCherry, SNAP-PodJ(1-635), sfGFP, mCherry protein aliquots were thawed on ice along with KCl, Tris-HCl (pH=8.0), 1,6-hexanediol, ATP, ADP, and sterile water. Working solutions of protein, Tris-HCl (pH=8.0), and KCl were combined, diluted with water to various concentrations, and incubated at room temperature for 15 min before imaging. Then the incubated sample mix was pipetted onto the slides with SureSeal Imaging Spacers (Electron Microscopy Sciences) and covered with coverslips (VWR). All images were taken with an Eclipse Ti-E inverted microscope (Nikon) in both phase-contrast and fluorescent channels using a Plan Apo 100x objective.
His-PodJ(1-635) was also analyzed by running a native gel. Protein was separated by gel electrophoresis (8% resolving gel) at 80 V for 4 h at 4 C, using a native protein ladder (range from 66 to 669 kDa, Thermo Fisher).

Western blotting
We analyzed protein levels and potential proteolysis for each protein construct expression in E. coli and C. crescentus through Western blot analysis. These assays indicated that each PodJ and PleC variant was expressed and exhibited little to no proteolysis (Fig. S10). For the Western blotting, logphase cells were induced with 0.002 to 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) or 0.5 to 10 mM arabinose in E. coli for 0.5 to 2 h, and 0.003% to 0.3% xylose or 0.005 to 0.5 mM vanillic acid in C. crescentus for 3 to 5 h unless otherwise stated. After induction was complete, the cells were pelleted and resuspended in 250 ml of 2x Laemmli buffer for each 1.0 A 600 unit. The samples were boiled at 95 C for 10 min, then vortexed. Next, 10 μl of samples was loaded in a 10% SDS-PAGE gel and run at 125 V for 90 min. Then the transfer was done at 20 V for 80 min at 0 C. Blocking was done for 1 h using 25 ml of blocking buffer (25 ml 1x TBST, 1.25 g nonfat dry milk) at 0 C with gentle shaking. For primary antibody blotting, the membrane was submerged in 1:5000 dilution of the anti-GFP (#2956S, Cell Signaling) or anti-mCherry (#43590S, Cell Signaling) antibody in the buffer (10 ml 1x TBST, 0.5 g BSA, 10 μl antibody) and shacken gently for 1 h at room temperature. After washing the membrane 3 times with 1x TBST for 5 min each, the membrane was incubated with secondary antibody (1:10,000) anti-goat IgG secondary antibody (A0545, Sigma Aldrich) in the buffer (10 ml 1x TBST, 0.5 g nonfat dry milk, 1 μl antibody) for 1 h with gentle shaking at room temperature. Next, the membrane was washed 3 times, 5 min each, with 1x TBST buffer with gentle shaking. After the wash, the membrane was placed in Pierce chemiluminescence substrates for 5 min and imaged on film using ChemiDoc (Bio-Rad).

PleC-CcaS chimera reporter assay
PleC-CcaS chimera reporter assays were performed based on the following steps. Starting from a −80 C dimethyl sulfoxide freezer stock, strains were inoculated into 5 ml LB Miller Broth in culture tubes containing appropriate antibiotics and grown at 37 C 220 rpm for overnight. Cultures were then diluted with fresh and sterile LB media to A 600 =1.0 using a UV-visible spectrophotometer (VWR). The cells were then inoculated into fresh LB media with a density of 25 μl per 1 ml LB media with appropriate antibiotics. The tubes were shaken at 37 C 220 rpm until A 600 reached 0.4. Then PodJ was induced with 5 mM arabinose for another 4 h. After that, cells were transferred into 96-well plates. Fluorescence was measured using a 5 nm bandpass with excitation/emission for mCherry (585/nm/610 nm)/CFP (456 nm/480 nm)/YFP (513 nm/527 nm) with a manually set gain of 50. Each construct was repeated with three independent biological replicates as indicated in the standard error in the bar graph.

Homology modeling
The PleC protein sequence was submitted into HHpred to predict protein features, fetch published crystal structures as templates and generate multiple sequences alignment (59). A template (PDB:4GCZ) was selected to model the homology structure of PleC PAS-A and PAS-B, respectively. Homology models were then downloaded and edited with PyMol to highlight secondary structures and signal transmission motif DI/VT residues at the C terminus of each PAS domain.

Coupled enzyme activity assay
Kinase activity was measured using a coupled enzyme assay (44). Purified proteins, 7.5 μM, were mixed in kinase buffer supplemented with 1.0 mM ATP, 10 mM MgCl 2 , 3 mM phosphoenolpyruvate, 0.2 mM NADH, 2 units of pyruvate kinase, and 6.6 units of lactate dehydrogenase (P0294, Sigma). Reactions were performed in three replicates in a 100 μl volume and loaded into a clear polystyrene 384-well plate. Each reaction was initiated by adding ATP, and 340 nm absorbance was recorded every 10 s for 90 min on a Tecan M1000 microplate reader (Tecan). The slope of a stable, linear absorbance decay was measured to calculate ATP hydrolysis rates (60). Background rates of ATP hydrolysis and NADH oxidation were measured and subtracted from observed ATP hydrolysis rates without adding any protein. The mean observed rate and SD were determined and analyzed using Prism (GraphPad).