Z-Ring-Associated Proteins Regulate Clustering of the Replication Terminus-Binding Protein ZapT in Caulobacter crescentus

Rapidly growing bacteria experience dynamic changes in chromosome architecture during chromosome replication and segregation, reflecting the importance of mechanisms that organize the chromosome globally and locally within a cell to maintain faithful transmission of genetic material across generations. During cell division in the model bacterium Caulobacter crescentus, the replication terminus of the chromosome is physically linked to the cytokinetic Z-ring at midcell.

morphogenesis. During the G 1 -to-S transition, the cell undergoes flagellar ejection and stalk morphogenesis and concomitantly regains the ability to initiate replication at the origin. Subsequently, one of the replicated origins translocates dynamically from the incipient stalked cell pole to the opposite cell pole. This active translocation of the origin is mediated by the essential ParA-ParB-parS system (13)(14)(15)(16). The centromere-like parS sequence proximal to the origin is specifically recognized by the ParB DNA-binding protein (17)(18)(19). The resultant ParB-parS complexes associate with the ParA ATPase, which generates a driving force for translocation of the parS-proximal DNA region including the origin (16,20). Finally, the pole-localizing protein PopZ captures the segregated ParB complexes at cell poles, ensuring bipolar sequestration of the sister origins (21)(22)(23).
The replication terminus of C. crescentus, which is located on the opposite side of the chromosome from the origin of replication, also undergoes dynamic changes in subcellular position during S phase (7,24) (Fig. 1A). The terminus region is initially sequestered at the cell pole opposite the origin-tethering cell pole in G 1 phase. Upon S phase entry, the terminus region relocates to the midcell region, where cell division takes place. This central positioning of the terminus DNA persists as the cell cycle progresses, and each of the sister-replicated terminus DNAs is deposited to the new cell poles of the individual daughter cells following division. Notably, this dynamic behavior of the terminus DNA coincides with the subcellular positioning of the cytokinetic Zring, a bundle of filamentous polymers formed by the essential GTPase FtsZ (24). The midcellular localization of the Z-ring at S phase initiates cell division and provides a scaffold for the divisome, an assembly of protein complexes involved in this process (15,(25)(26)(27)(28).
To date, we know of two different systems that account for the physical linkage between the Z-ring and the terminus DNA: the MatP-ZapB-ZapA EC system in the gammaproteobacterium Escherichia coli and the ZapT-ZauP-ZapA system in C. crescentus. MatP, a terminus-specific binding protein, forms a dimer that specifically recognizes the 13-mer matS sites that are distributed within an 800-kb stretch around the terminus region (29)(30)(31)(32). Structural and biochemical studies revealed that the C-terminal coiled-coils of DNA-bound MatP dimers interact to form a tetramer, allowing DNA bridging between distal matS sites (33). The resultant MatP-mediated DNA clustering organizes the replication terminus DNA region into a compacted DNA called the Ter macrodomain (30,31,34). Meanwhile, MatP binds directly to the coiled-coil ZapB protein (35,36). Recombinant ZapB proteins exist as dimers in the crystal form and can form polymers in solution (35). Although ZapB exhibits no direct interaction with E. coli FtsZ EC , it physically associates with the E. coli ZapA tetramer (ZapA EC ) (37). ZapA EC consists of a long C-terminal helix and an N-terminal globular domain containing several charged residues that are important for the direct interaction with FtsZ EC (38,39). Collectively, the sequential interaction of matS, MatP, ZapB, ZapA EC , and FtsZ EC is thought to physically link the terminus DNA to the Z-ring. Fluorescence microscopy revealed that the MatP-mCherry fusions form a discrete focus independently of ZapA EC or ZapB (29), implying that the preassembled MatP-matS cluster, in association with the divisome, plays a predominant role in establishing the physical linkage.
C. crescentus uses the ZapT-ZauP-ZapA system instead of the MatP-ZapB-ZapA EC system. Despite the low degree of conservation of MatP and ZapB, ZapT and ZauP homologs are widely conserved among Gram-negative bacteria (24,40). In this system, the MerR DNA-binding protein ZapT plays a key role in recognition of the terminus DNA (24). Members of the MerR family contain similar N-terminal DNA-binding domains, central dimerization domains, and dissimilar C-terminal domains that interact with specific effectors, such as metal ions, chemicals, and oligopeptides (41, 42) (see Fig. S1 in the supplemental material). Although ZapT and MatP are not related at the sequence level, ZapT preferentially binds to Caulobacter genome positions ranging from 1.3 to 2.2 MB, corresponding to the chromosome terminus and its flanking regions. Moreover, ZapT directly or indirectly interacts with C. crescentus ZapA and ZauP in vivo. ZapA binds directly to FtsZ, whereas ZauP is a functional ZapB homolog with affinity for ZapA but not FtsZ. Therefore, a functional complex containing ZapT, ZapA, and ZauP is thought to mediate the physical interaction between the terminus and Z-ring in C. crescentus. However, the molecular functions underlying the ZapTmediated interaction between the terminus DNA and divisome remain to be fully elucidated.
Here, we provide evidence that clustering of DNA-bound ZapT is stimulated by its association with the divisome. We demonstrate that recombinant ZapT, ZauP, and ZapA proteins interact sequentially, presumably in that order, to form a ternary complex. Moreover, we show that multiple ZapT molecules are recruited by each ZauP oligomer in vitro. Consistent with these observations, the subcellular localization of ZapT strictly depends on ZauP and ZapA. Finally, mutant analyses revealed that the Cterminal domain of ZapT is functionally specialized to directly interact with ZauP. Because the hydrophobic properties of the C terminus are shared among ZapT homologs, our findings suggest that the mechanism of terminus sequestration to the Z-ring is conserved. Moreover, the ZauP-dependent clustering of ZapT-DNA complexes provides insight into the role of the divisome in chromosome organization.

RESULTS
ZapA and ZauP are required for ZapT localization but dispensable for the interaction between ZapT and terminus DNA. ZapT-binding sites are distributed preferentially at genome positions ranging from 1.3 to 2.2 MB, which correspond to the chromosome terminus and its flanking regions (24). Because the subcellular localization of ZapT coincides with that of the terminus and the divisome, we assumed that DNA-bound ZapT molecules interact with each other to form a nucleoprotein complex to which the divisome is recruited (Fig. 1A). Alternatively, multiple ZapT molecules could form a proteinaceous complex with the divisome to which the chromosomal ZapT-binding sites are recruited (Fig. 1A). To distinguish between these models, we assessed the dependence of the ZapT-DNA interaction on the divisome components ZapA and ZauP, both of which form a complex with ZapT in vivo. To investigate the DNA-binding activity of ZapT in vivo, we performed chromatin immunoprecipitation (ChIP) assays using strains expressing a C-terminally 3ÂFLAG-tagged ZapT (ZapT-3F) from a cumate-dependent promoter on a low-copy-number plasmid (pQF::zapT). Deep sequencing revealed that the distribution pattern of plasmid-borne ZapT-3F was indistinguishable from that of chromosomally expressed ZapT-3F (Fig. 1B). Moreover, binding of ZapT-3F to terminus-proximal regions such as the CCNA_01498, CCNA_01600, CCNA_01763, and CCNA_01936 loci ( Fig. 1B, a, b, c, and d, respectively) was confirmed by quantitative PCR (ChIP-qPCR) (Fig. 1C). The origin region (Cori) was used as a negative control. Next, we performed a similar ChIP-qPCR analysis in the DzapA and DzauP backgrounds. The profiles of DNA recovery in these mutants were comparable with those in the wild-type background (Fig. 1C). Therefore, ZapA and ZauP are dispensable for the DNA-binding activity of ZapT, arguing against the idea that formation of ZapT complexes with ZapA and ZauP is a prerequisite for the ZapT-terminus DNA interaction.
Next, we assessed the localization dependence of ZapT on zapA and zauP. To visualize the localization of ZapT, we used a strain expressing a C-terminal mNeonGreen fusion of ZapT (ZapT-mNeonGreen) from the native locus (24). We reported previously that in the wild-type strain, the localization pattern of ZapT-mNeonGreen parallels that of the Z-ring, i.e., the cells form a single discrete focus of ZapT-mNeonGreen at one cell polar region in shorter cells and at the midcell region in larger cells (24). Moreover, colocalization analysis revealed that a ZapT-mNeonGreen focus coincides with ZapA and the terminus in the same wild-type cell. Consistent with these observations, nearly all wild-type cells formed a discrete focus of ZapT-mNeonGreen at the midcell region or in one cell polar region (Fig. 2). In contrast, in cells lacking ZapA or ZauP, ZapT-mNeonGreen was completely dispersed (Fig. 2). These observations were not due to the instability of ZapT-mNeonGreen, as we obtained similar results with wild-type, DzapA, and DzauP strains transformed with a pQF::zapT derivative expressing ZapT-mNeonGreen instead of ZapT-3F (Fig. S2A). Western blotting demonstrated that the levels of ZapT-mNeonGreen in these mutants were indistinguishable from those in the wild type (Fig. S2B). Therefore, ZapA and ZauP are required for proper localization of ZapT. Together, these results favor a model in which ZapT-DNA complexes are recruited to the divisome through interactions with ZapA and ZauP.
ZapT directly binds to ZauP. To map interactions between ZapT, ZapA, and ZauP, we performed size exclusion chromatography on the purified recombinant proteins: ZapT with a C-terminal hexahistidine tag (ZapT-His; 21 kDa), ZapA with an N-terminal hexahistidine-MBP tag (His-ZapA; 56 kDa), and ZauP with an N-terminal hexahistidine-SUMO tag (His-ZauP; 26 kDa). First, we determined the oligomeric state of His-ZapA and His-ZauP using a Superdex 200 column (Fig. 3). Previously, ZapA homologs from P. aeruginosa and E. coli were reported to form homotetramers (38,43). Consistent with those studies, our His-ZapA eluted at a position corresponding to the average molecular weight of a His-ZapA tetramer (220 kDa) (Fig. 3A). Likewise, we found that His-ZauP formed multimers with the average molecular weight of a His-ZauP hexamer (150 kDa) (Fig. 3B). Moreover, when mixed together, His-ZapA and His-ZauP coeluted earlier than either protein alone ( Fig. 3A to C). Thus, His-ZapA and His-ZauP retain the ability to form heteromultimers in vitro.
Next, we used a Superdex 200 column to analyze an oligomeric state of ZapT-His. ZapT shares structural homology with transcriptional regulators of the MerR family (E value of 1.0 Â 10 216 ), whose members typically form stable dimers in vitro. In our setup, ZapT-His eluted broadly near a position corresponding to dimeric ZapT-His (43 kDa) (Fig. 3D). To better resolve the oligomeric state, we analyzed ZapT-His using a Superdex 75 column with a separation range between 3 and 70 kDa. We detected a single elution peak of ZapT-His at a position corresponding to dimeric ZapT-His (Fig. S3). These observations argue that, as with other MerR family proteins, ZapT preferentially forms a dimer in solution.
To probe for a direct interaction between ZapT and ZapA or ZauP, we analyzed a mixture of the two proteins using a Superdex 200 column. The chromatograms of ZapT-His and His-ZapA were unaffected by mixing the proteins together (Fig. 3A, D, (DzauP) strains expressing ZapT-mNeonGreen from the native locus were grown exponentially in PYE medium, followed by analysis by fluorescence microscopy. Representative phase-contrast and fluorescence microscopy images are shown. Demographs were generated using Oufti software (63).
Divisome Organizes the Replication Terminus ® and E). In contrast, in a mixture of ZapT-His and His-ZauP, the proteins were coeluted earlier than either protein alone (Fig. 3A, B, and F). This strongly argues that ZapT directly interacts with ZauP but not ZapA. The average molecular weight of the heterooligomer was 310 kDa, consistent with the idea that three or four ZapT-His dimers (130 to 170 kDa) bind to each His-ZauP hexamer (150 kDa).
Next, we analyzed a mixture of ZapT-His, His-ZapA, and His-ZauP to determine whether ZapT forms a ternary complex with ZauP and ZapA. Previously, we showed that ZapA, ZauP, and ZapT reside in the same protein complex in vivo (24). Consistent with this, when mixed together, ZapT-His coeluted with His-ZapA and His-ZauP at a position corresponding to an average molecular weight of .600 kDa ( Fig. 3G and H), which is distinguishable from ZapT-ZauP complexes ( Fig. 3F and H). Thus, ZapT sustains the ability to form a ternary structure with ZapA and ZauP.
The C-terminal sensor domain of ZapT is required for binding to ZauP. The direct interaction between ZapT and ZauP motivated us to identify the functional if not all, members of this family interact with a specific effector, such as metal ions, chemicals, or oligopeptides, through their C-terminal sensor domains, we hypothesized that the C terminus of ZapT is functionally specialized to interact with ZauP.
To determine whether the C terminus of ZapT is required for ZauP binding, we analyzed a truncated variant of ZapT-His that lacks the C-terminal sensor domain (ZapTDC-His) using gel filtration (Fig. 4A). First, we injected the ZapTDC-His protein into a Superdex 200 column to determine whether this variant retains dimerization activity. As shown in Fig. 4B, elution of ZapTDC-His peaked at a position corresponding to a ZapTDC-His dimer (29 kDa). Consistent results were obtained using Superdex 75 column chromatography (Fig. S3). Thus, the C-terminal sensor domain of ZapT is dispensable for dimerization, a finding consistent with the fact that MerR family proteins self-dimerize through the central dimerization domain. Next, we applied a mixture of ZapTDC-His and His-ZauP to a Superdex 200 column (Fig. 4C to E). The chromatograms Divisome Organizes the Replication Terminus ® of ZapTDC-His and His-ZauP were unaffected by mixing the proteins together (Fig. 4C), indicating that ZapTDC-His impairs the ZauP binding activity. To consolidate this, we performed a similar experiment in the presence of wild-type ZapT-His as an internal control. Gel filtration of the mixture revealed that wild-type ZapT-His, but not ZapTDC-His, coeluted with ZauP ( Fig. 4D and F). Wild-type ZapT-His did not interact with ZapTDC-His ( Fig. 4E and F). Taken together, these findings strongly argue that the Cterminal domain of ZapT is crucial for interaction with ZauP.
ZauP sequesters multiple ZapT dimers. To validate our model in which oligomeric ZauP sequesters multiple ZapT dimers, we carried out a pulldown assay using ZapT-His and intact ZapT and ZauP proteins (Fig. 5A). When ZapT-His was incubated with ZauP, we observed coelution of the two proteins (Fig. 5B). In contrast, when ZapT-His was mixed with ZapT instead of ZauP, only ZapT-His was recovered in an elution fraction ( Fig. 5B and C). These observations are fully consistent with the results for the gel filtration assay (Fig. 3 and 4). Notably, when ZapT-His, ZapT, and ZauP were coincubated, we observed the specific recovery of ZapT together with His-ZapT and ZauP ( Fig. 5B and C), indicating that His-ZapT, ZauP, and ZapT coexist in the same complex. Moreover, when we performed a similar analysis using His-ZapTDC instead of His-ZapT, neither ZauP nor ZapT was coeluted with His-ZapTDC. Thus, the C terminus of ZapT is required for formation of higher-ordered ZapT multimers on ZauP.
To further examine if the terminus-bound ZapT proteins are sequestered by ZauP in vivo, we carried out ChIP-qPCR experiments using a strain expressing an N-terminally 3ÂFLAG-tagged ZauP (3F-ZauP). When cells were cross-linked in 1% formaldehyde as with the ZapT-ChIP assay, neither the terminus nor Cori was recovered (data not shown), which is consistent with the idea that ZauP per se has no affinity for DNA (40). We reasoned that 1% formaldehyde cross-linking was inefficient to pull down DNA molecules that bound indirectly to ZauP. Therefore, we adopted 3.6% formaldehyde to improve the efficiency of cross-linking. As a result, we found that the terminus is markedly enriched by 3F-ZauP (Fig. 5D). This is not due to an artifact caused by a higher dose of formaldehyde, because ZauP-ChIP recovered only background levels of the origin DNA. Besides, neither the terminus nor the origin was enriched in control ChIP analyses using the wild-type NA1000 strain. These findings suggested that ZauP is able to recruit ZapT in complex with the terminus DNA.
A fluorescent protein bearing the C-terminal sensor domain of ZapT coincides with ZauP. To corroborate our idea that the C terminus of ZapT mediates its localization to the Z-ring in vivo, we investigated whether a fluorescent protein grafted onto the C-terminal domain of ZapT would be recruited to ZauP in living cells (Fig. 6). To monitor the position of ZauP in a cell, we generated a strain (SHQ230) expressing an N-terminal GFP fusion of ZauP from the native zauP-zapA operon of the chromosome. In addition, the zapA gene of SHQ230 was replaced by a gene encoding a C-terminal mCherry fusion of ZapA, which marks the position of the Z-ring (24,26). Previously, ZauP was reported to colocalize with ZapA and the Z-ring (40). When SHQ230 was transformed with a low-copy-number plasmid with a superfolder mTurquoise 2 fluorescent protein grafted onto the N terminus of full-length ZapT (sfTq2-ZapT), approximately 66% of the cells had a discrete sfTq2 focus ( Fig. 6A and B). In this assay, production of sfTq2-ZapT relied on leaky expression from the xylose promoter in the absence of the inducer xylose, which might result in moderately compromised formation of the focus. Of these cells with an sfTq2 focus, 97% of the transformed cells contained a discrete sfTq2-ZapT focus at a position in close proximity to the GFP-ZauP and ZapA-mCherry foci (Fig. 6B). Moreover, demographic representation of sfTq2-ZapT largely coincided with that of GFP-ZauP and ZapA-mCherry (Fig. S4). These observations are consistent with the idea that ZapT, ZauP, and ZapA form a ternary complex in vivo. Notably, when a plasmid expressing sfTq2 grafted onto the N terminus of the truncated ZapT bearing its C-terminal region between Gly113 and Gly141 (sfTq2-C) alone was transferred to SHQ230, the resultant strain produced cells with a discrete sfTq2-C focus that colocalized with GFP-ZauP and ZapA-mCherry foci with a frequency comparable to that of the sfTq2-ZapT strain ( Fig. 6A and C). Consistently, demographic representation of sfTq2-C was similar to that of sfTq2-ZapT (Fig. S4). As a control, expression of sfTq2 alone yielded cells with dispersed sfTq2 signals, affecting the formation of neither GFP-ZauP nor ZapA-mCherry foci (Fig. 6A and D and Fig. S4). Together, these results argue that the ZapT C terminus includes a functional motif that interacts with ZauP to form a cluster in vivo.
Finally, to further gain physiological insight into the role of the ZapT-ZauP interaction, we analyzed the localization of the Z-ring using a C. crescentus strain expressing an FtsZ-yellow fluorescent protein (YFP) fusion ectopically from the vanillate-dependent promoter (15). In wild-type cells, a discrete YFP focus is formed at the new cell pole in shorter G 1 cells and relocates to the midcell region as the cell grows. We previously reported that the timing of FtsZ relocation from the cell pole to midcell is slightly delayed in the DzapT mutant strain (24). Consequently, DzapT cells with unipolar FtsZ focus are more elongated than wild-type cells with unipolar FtsZ. Consistent with this, when the size distributions of those unipolar FtsZ cells were analyzed, DzapT mutant ZapT is depicted as a blue oval with an extension (light blue) that corresponds to its C terminus. His tag (black) and ZauP (green) are indicated. (B and C) ZapT-His or ZapTDC-His (200 pmol as monomer) was incubated in the presence (1) or absence (-) of ZauP (600 pmol as monomer) and native ZapT (300 pmol as monomer), followed by pulldown using Ni-conjugated Sepharose beads. After washing, materials retained on the beads were analyzed using SDS-15% PAGE and Coomassie brilliant blue staining. For the protein standard, the indicated amounts of purified proteins were loaded on the last two lanes of the same gel, which is used to draw a standard curve to deduce recovery of native ZapT. (C) Mean values and standard deviations obtained from two independent experiments were plotted. (D) ChIP-qPCR assay for 3F-ZauP. SHQ247 (3F-ZauP) and NA1000 were grown exponentially and cross-linked in 3.6% formaldehyde, followed by chromatin immunoprecipitation with an anti-FLAG antibody. Recovery of Cori (primers 11/12) and the terminus (position c; primers 9/10) was shown as in Fig. 1C. The P value was calculated using Student's t test.
Divisome Organizes the Replication Terminus ® cells bearing the plasmid with wild-type zapT were shorter than those with the empty vector ( Fig. 7A and B). Strikingly, introduction of the plasmid with the zapTDC allele failed to suppress the observed phenotype of the DzapT mutant, although expression of ZapT and ZapTDC was comparable ( Fig. 7B and C). Signal intensities of FtsZ-YFP foci appeared unaffected in these strains (Fig. 7A). Thus, these observations argued that ZauP interaction with the C terminus of ZapT is important for the timely formation of the Z-ring at midcell in vivo.

DISCUSSION
In C. crescentus, the replication terminus and divisome are spatially coordinated, localizing in proximity to each other throughout most of the cell cycle. ZapT acts as a terminus-binding protein, physically linking the terminus and divisome. However, the functions and molecular mechanisms of this linkage remain obscure. In this study, we found that ZapT and divisome components ZauP and ZapA interact directly,  presumably in that order, to form ternary complexes. Focusing on the ZapT-ZauP interaction, we found that ZapT forms a stable dimer and that multiple ZapT dimers assemble on ZauP oligomer(s). At the early stage of the cell cycle, ZauP is recruited to the Zring in a ZapA-dependent manner (40). Therefore, it is reasonable to assume that multiple ZapT molecules are recruited to the early divisome through the interaction with ZauP. Consistent with this idea, we observed that subcellular localization of ZapT depends on ZauP and ZapA and parallels the localization of those two factors. Notably, in mutant cells lacking ZauP or ZapA, ZapT retained a specific affinity for the terminus DNA. Thus, these observations suggested that staged assembly of the protein complexes underlies the physical linkage between divisome and the terminus, i.e., ZapT initially binds to the terminus DNA independently of ZauP and ZapA and subsequently forms a nucleoprotein cluster on the early divisome in a manner dependent on ZauP and ZapA (Fig. 8). We infer that this cluster helps to organize ZapT-bound DNA loci Divisome Organizes the Replication Terminus ® into a compact structure. To the best of our knowledge, this is the first evidence that chromosome organization is driven by the early divisome. Notably, ZapT and ZauP homologs are widespread in diverse Gram-negative pathogens, including the alphaproteobacterium Brucella abortus, the betaproteobacterium Paraburkholderia rhizoxinica, and the gammaproteobacterium Pseudomonas aeruginosa. Sequence alignment revealed that the ZapT C-terminal domain contains several leucine residues that are highly conserved among ZapT homologs (see Fig. S1 in the supplemental material), suggesting that a hydrophobic interaction underlies the interaction between ZapT and ZauP homologs. Hence, our findings may reveal a general mechanism by which the chromosome terminus is coordinated with the divisome in diverse bacterial species.
The interaction between the C terminus of ZapT and ZauP provides mechanistic insight into the organization of the terminus DNA in C. crescentus. ZapT belongs to the MerR family proteins, which contain an N-terminal DNA-binding domain and a C-terminal sensor domain to which cognate effector molecules bind to remodel the DNA structure (41,42,44) (Fig. S1). The effectors include heavy metals, chemicals, and oligopeptides. Recent structural and biochemical studies suggest that a MerR family transcriptional regulator forms a homodimer to recognize its cognate DNA within the target promoter regardless of whether the specific effector is present and that binding of the effector induces a scissor-like movement of the dimerization helices. Consequently, the altered domain constellation enables the dimer to introduce a distortion in the DNA structure; this, in turn, promotes transcriptional initiation by RNA polymerase (45,46). By analogy, a similar conformational change could operate through binding of the ZapT C terminus to ZauP, leading to remodeling of ZapT-bound DNA in the terminus region. Chromosome contact mapping of wild-type C. crescentus revealed that the chromosome comprises multiple, largely independent spatial domains (47,48). Of these, the largest one is located at 1.6 to 1.9 kb, which overlaps the region to which ZapT binds preferentially. Therefore, it is plausible that ZapT, together with the divisome, plays a part in construction of the chromosome interaction domains at the terminus. The observation that DNA recovery in the ZapT-ChIP assays is moderately The function of ZapT contrasts with that of the E. coli terminus-binding protein MatP, which organizes the chromosome structure of the replication terminus independently of the divisome (Fig. 8). Biochemical and structural studies suggest that DNA-bound MatP dimers interact with each other to self-organize a nucleoprotein cluster, shaping the chromosome terminus region into a condensed structure (33). Although MatP interacts with the Z-ring through interaction with ZapB, mutant cells lacking ZapB are still able to form a discrete focus of MatP-fused fluorescent proteins and largely maintain the same chromosome structure at the replication terminus as wild-type cells (29,31). These observations contrast with the ZauP-dependent clustering of the ZapT-DNA complexes in C. crescentus. Given that MatP and ZapB homologs  are conserved only in enteric-related species of gammaproteobacteria, the ability of a terminus recognition protein to self-organize the nucleoprotein cluster might have been acquired late in gammaproteobacterial evolution. In this context, we infer that self-organized clustering of terminus recognition proteins generates multiple contact points by which the overall affinity to the divisome is markedly reinforced, likely through a linkage effect (49). Faithful segregation of replicated sister chromosomes relies on accurate spatial control of specific chromosome loci with which dedicated proteins form clusters of nucleoprotein complexes. In C. crescentus, ParB binds the origin-proximal parS site to form a partition complex that is essential for polar sequestration of the origin (13,14,16,17). The subcellular localization of the origin is also ensured by structural maintenance of chromosome (SMC) complexes (50,51). Together with accessory proteins ScpA and ScpB, SMC forms a ring-shaped complex within which DNA strands are topologically entrapped (48). Loading of SMC onto the chromosome is stimulated at parS, thereby assisting in chromosome organization near the origin. In E. coli, the SMC family DNAbinding protein MukB also plays a crucial role in chromosome segregation (34,(52)(53)(54). As with SMC of C. crescentus, MukB localizes primarily in the vicinity of the origin and organizes the chromosome into a compact structure (55,56). In addition, E. coli has coopted multiple DNA-associating proteins to spatially control the nascently replicated  (55,(57)(58)(59). Thus, our finding that the divisome can control organization of the ZapT-terminus DNA complexes reveals another layer of regulation of chromosome dynamics in coordination with cell division.

MATERIALS AND METHODS
Bacterial strains and DNA. The strains, plasmids, and primers used in this study are listed in Tables 1,  2, and 3, respectively. Caulobacter strains were grown at 30°C in peptone-yeast extract (PYE) supplemented with appropriate antibiotics, as described previously (60,61). When necessary, cumate (1mM) or vanillate (1 mM) was added to the culture medium as indicated. Detailed procedures for construction of the strains and plasmids are described in the supplemental material (Text S1).
Recombinant proteins. Recombinant proteins were expressed in E. coli and purified as described in the supplemental material (Text S1).
Size exclusion chromatography. The size exclusion chromatography assay was performed essentially as described previously (57,62). Briefly, proteins were loaded onto a Superdex 200 PC3.2/30 column (2.4-ml column volume) equilibrated with SEC buffer (25 mM Tris-HCl [pH 7.5], 300 mM sodium chloride, and 20% sucrose) and fractionated at a flow rate of 20 ml/min, followed by SDS-15% PAGE and Coomassie brilliant blue staining. When a Superdex 75 HR 10/30 column was used, proteins were separated at a flow rate of 0.2 ml/min.
ChIP sequencing and ChIP-qPCR. ChIP was performed as described previously (24). Briefly, exponentially growing cells (200 ml in PYE) were fixed for 10 min using 1% formaldehyde solution, washed thoroughly, resuspended in buffer, and lysed through two passages in a French press. After DNA shearing with sonication, the cell debris was removed by ultracentrifugation, and the cleared cell lysate was incubated with anti-FLAG M2 magnetic beads (Wako or Sigma). After the beads were washed, the bound materials were incubated at 65°C overnight to reverse cross-linking. The resultant DNA samples were purified using a DNA cleanup kit (Zymo Research). When SHQ247 (3F-ZauP) cells (100 ml in PYE) were analyzed, cross-linking was carried out for 5 min in 3.6% formaldehyde instead of 1% formaldehyde.
For deep sequencing, samples were indexed using a NEBNext Ultrall DNA library prep kit and analyzed on an Illumina HiSeq 2500 instrument (single-end).
For qPCR, samples were analyzed by a standard percent input method using TB green premix ExTaqII and Thermal Cycler Dice TP800 (TaKaRa). Locus-specific primers are listed in Table 3.
Microscopy. Differential interference contrast (DIC), phase-contrast, and fluorescence microscopy analyses were performed using a Nikon Eclipse 80i microscope equipped with an X-Cite TURBO multiwavelength LED illumination system and an Andor Zyla 4.2 sCMOS camera, as described previously (24). Quantitative image analyses were performed using the Oufti and MicrobeJ software packages (63,64).
Western blotting. The Western blot assay was performed as described previously (24). The anti-mNeonGreen and anti-Flag antibodies were purchased from Chromotek and Thermo, respectively.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.02 MB.