A Conserved Coiled-Coil Protein Pair Focuses the Cytokinetic Z-Ring in Caulobacter crescentus

Summary

The cytoskeletal GTPase FtsZ assembles at midcell, recruits the division machinery, and directs envelope invagination for bacterial cytokinesis. ZapA, a conserved FtsZ-binding protein, promotes Z-ring stability and efficient division through a mechanism that is not fully understood. Here we investigated the function of ZapA in Caulobacter crescentus. We found that ZapA is encoded in an operon with a small coiled-coil protein we named ZauP. ZapA and ZauP co-localized at the division site and were each required for efficient division. ZapA interacted directly with both FtsZ and ZauP. Neither ZapA nor ZauP influenced FtsZ dynamics or bundling, in vitro, however. Z-rings were diffuse in cells lacking zapA or zauP and, conversely, FtsZ was enriched at midcell in cells overproducing ZapA and ZauP. Additionally, FtsZ persisted at the poles longer when ZapA and ZauP were overproduced, and frequently colocalized with MipZ, a negative regulator of FtsZ polymerization. We propose that ZapA and ZauP promote efficient cytokinesis by stabilizing the midcell Z-ring through a bundling-independent mechanism. The zauPzapA operon is present in diverse Gram-negative bacteria, indicating a common mechanism for Z-ring assembly.

INTRODUCTION

Accurate cell division is essential to produce viable progeny across all domains of life. In almost all bacteria, the execution of cytokinesis requires an essential cytoskeletal GTPase, FtsZ, which polymerizes at the incipient division site to form the cytokinetic Z-ring (Erickson et al., 2010). In the α-proteobacterium Caulobacter crescentus, the Z-ring is positioned at midcell primarily by the Walker-type ATPase MipZ, which forms a bipolar gradient with high concentration at the cell poles and lower concentration near midcell (Thanbichler and Shapiro, 2006; Kiekebusch et al., 2012). MipZ directly binds FtsZ and stimulates its GTPase activity, thus the Z-ring assembles at the site of lowest MipZ concentration, that is, roughly midcell. Localized assembly of the Z-ring nucleates recruitment of other division proteins (the divisome) in serial waves (Goley et al., 2011). Collectively, FtsZ and the divisome promote constriction and fission of the cell membrane(s) and remodeling of the peptidoglycan cell wall (Lutkenhaus et al., 2012). FtsZ is proposed to fulfill three possible functions in cytokinesis (Meier and Goley, 2014; Xiao and Goley, 2016): it acts as a scaffold to promote assembly of the division machinery, it can generate constrictive force that may guide cell envelope invagination, and it regulates peptidoglycan metabolism.

As a central player in cytokinesis, the assembly properties, structure, and regulation of FtsZ have been the subject of intense study for two decades. Like its eukaryotic homolog, tubulin, FtsZ is a polymerizing GTPase. Thus, the nucleotide binding and hydrolysis cycle of FtsZ regulates both its polymerization state and protofilament conformation (Mukherjee and Lutkenhaus, 1994; Erickson et al., 1996; Yu and Margolin, 1997; Lu et al., 2000). In vitro, purified FtsZ assembles into linear, head-to-tail protofilaments with gentle curvature. Nucleotide hydrolysis and phosphate release lead to a conformational change in the FtsZ protofilament, taking it to a more highly curved state with weakened intersubunit contacts, leading to depolymerization (Lu et al., 2000). Imposition of the gentle curvature of FtsZ protofilaments on the less curved membrane has been proposed as a mechanism of force generation for constriction (Osawa et al., 2009; Erickson et al., 2010).

In addition to its intrinsic self-assembly properties, numerous binding partners directly interact with FtsZ and can affect its polymerization state, GTPase activity, superstructure, and/or association with the inner membrane (Huang et al., 2013). One regulator that has been studied in numerous bacteria is a small coiled-coil protein called ZapA. Although not essential for viability, ZapA is broadly conserved and contributes to the efficiency of cytokinesis (Gueiros-Filho and Losick, 2002; Small et al., 2007; Kamran et al., 2016). In order for FtsZ to direct efficient cytokinesis, it must assemble into a paradoxically dynamic but stably localized cytokinetic ring. ZapA promotes the stability of the Z-ring through at least two possible mechanisms. Early studies demonstrated that ZapA from Bacillus subtilis, Escherichia coli, and Pseudomonas aeruginosa can promote bundling of parallel FtsZ protofilaments and reduce the apparent GTPase activity of FtsZ in vitro (Gueiros-Filho and Losick, 2002; Low et al., 2004; Small et al., 2007; Mohammadi et al., 2009; Pacheco-Gómez et al., 2013). For at least E. coli ZapA, however, these activities are dependent on pH, divalent cation concentration, and presence or absence of a hexahistidine tag on ZapA, with reduced bundling and effects on the GTPase activity at physiological pH and divalent cation concentration (Mohammadi et al., 2009). Using rheometry, a study of E. coli ZapA indicated that, in addition to bundling under some in vitro conditions, ZapA may cross-link FtsZ protofilaments into a network of non-parallel polymers (Dajkovic et al., 2010). It is unclear if ZapA-mediated parallel bundling or non-parallel cross-linking are functionally important in the cell. However, it is generally proposed that ZapA stabilizes the Z-ring by promoting association of FtsZ protofilaments into higher order assemblies and reducing turnover of polymers.

A second mechanism whereby ZapA may contribute to stable assembly of the Z-ring, at least in E. coli, is through its interaction with another small coiled-coil protein, ZapB. In E. coli cells, ZapB promotes the efficiency of cytokinesis, similar to ZapA, and localizes in an FtsZ- and ZapA-dependent manner to a midcell ring that is concentric to the Z-ring (Ebersbach, Galli, et al., 2008; Galli and Gerdes, 2010). ZapB does not interact directly with FtsZ, but does bind to ZapA (Galli and Gerdes, 2010; Galli and Gerdes, 2011). ZapB also forms polymers, itself, in a divalent cation-dependent manner in vitro (Ebersbach, Briegel, et al., 2008). In cells lacking ZapA or ZapB, the E. coli Z-ring is dispersed, forming clusters of FtsZ filaments that are not properly aligned at midcell (Buss et al., 2013). The FtsZ clusters formed in the absence of ZapA or ZapB still contain numerous filaments, indicating that filament bundling or cross-linking is intact in their absence (Buss et al., 2013). From these data, it was proposed that the critical function of ZapA in E. coli is to promote Z-ring localization at midcell by bridging dynamic FtsZ polymers and the stable ZapB ring, rather than by promoting filament bundling or cross-linking (Buss et al., 2013; Buss et al., 2015). As ZapB is restricted phylogenetically to the Enterobacteriales and Vibrionales orders of ɣ–proteobacteria, this role of ZapA-ZapB is presumed to be similarly confined.

In the present work, we investigated ZapA function in the α-proteobacterium Caulobacter crescentus, which undergoes an obligate asymmetric cell division and bears FtsZ regulators distinct from other well-studied bacteria (Curtis and Brun, 2010; Kirkpatrick and Viollier, 2011). We identified a small coiled-coil protein encoded immediately upstream of and in an operon with zapA that we named ZapA upstream-encoded protein (ZauP). Using biochemistry, genetics, and cell biology, we demonstrated that ZapA and ZauP comprise a coiled-coil protein pair that serves to stabilize the Z-ring at midcell through a bundling-independent mechanism in C. crescentus. Finally, we identified putative zauP homologs in diverse Gram-negative organisms, suggesting that the ZapA-ZauP mechanism of Z-ring stabilization is broadly conserved.

RESULTS

An upstream gene co-expressed with zapA is important for efficient cytokinesis

Upon examining the genomic region around the zapA gene in the laboratory-adapted C. crescentus CB15 derivative NA1000, we noted the presence of a gene upstream of zapA, CCNA_03357 (Fig. 1A), which was not annotated in the original CB15 genome. Two transcriptional start sites are found for these two genes, both of which are upstream of the start codon for CCNA_03357 (Zhou et al., 2015). No transcriptional start site for zapA, alone, was detected (Zhou et al., 2015), indicating that these two genes form an operon. Both genes were confirmed to be translated by ribosome profiling and liquid chromatography-mass spectrometry (Schrader et al., 2014). COILS prediction (Lupas et al., 1991) suggests that CCNA_03357, like ZapA and E. coli ZapB, assumes a predominantly coiled-coil structure (Fig. 1B). We named this 13.4 kD protein encoded by CCNA_03357 ZauP, for ZapA upstream-encoded protein, and set out to determine if, like ZapA, it has a role in cell division.

Figure 1. ZapA and ZauP contribute to efficient cytokinesis.

A. Genome organization around the zapA gene as annotated in the C. crescentus CB15 and NA1000 genomes. zauP and zapA are organized in an operon, with transcriptional start sites upstream of zauP. B. Probability of coiled-coil formation for ZauP (encoded by CCNA_03357) and E. coli ZapB using COILS prediction. C. Representative phase contrast images of the indicated strains. Bar = 2 µm. D. Cell length distribution of cells of the indicated genetic backgrounds. For WT, 3 biological replicates of a single strain (EG864) were combined. For ΔzapA (EG1080, EG1081, EG1082), ΔzauP (EG971, EG972, EG973), and ΔzauPzapA (EG974, EG975, EG976), data from 3 independently created deletion clones of each are plotted. Each data point represents the length of an individual cell. Grey lines = mean, bars = standard deviation (SD). n = 1531 to 2395 cells. Significant differences between the mean cell length of each strain and WT control determined using ANOVA with Dunnett’s Multiple Comparison post-test. ***: p < 0.001. E. Mean length with SD for each genetic background calculated from the data in D. F. Immunoblot against lysates for the indicated strains using antisera for the indicated proteins. Arrow indicates ZauP band and asterisk indicates cross-reacting band. SpmX was used as a loading control. G. Representative phase contrast images of strains overproducing indicated proteins. EV = empty vector. Bar = 2 µm. H. Cell length distributions (mean ± SD) of cells shown in G, n = 244 cells for each. Significant differences between the mean cell length of each strain and EV control determined using ANOVA with Dunnett’s Multiple Comparison post-test. ***: p < 0.001. I. Mean length of each strain calculated from data shown in H. J. Immunoblot against lysates for the indicated strains at the indicated time points using antisera for the indicated proteins.

To do this, we made in-frame deletions of zauP and of zapA, as well as a double deletion of both genes, and examined the morphologies of the resulting strains. zapA and zauP were dispensable alone or in combination, however deleting each had a detrimental effect on the efficiency of cytokinesis. Mildly elongated cells, as well as occasional long filaments were observed for each single deletion, with deletion of zapA being more severe than deletion of zauP (Fig. 1C–E). Together, deleting zapA and zauP had a synergistic effect on cell length, with significant cell length heterogeneity. We conclude that while neither zapA nor zauP is essential for viability, each contributes to the efficiency of division. Moreover, we suggest that these proteins may work together, but that they must have independent functions as well, as the double zauPzapA deletion is more severe than either single deletion. We did not detect changes in the levels of FtsZ or MipZ when zapA and zauP were deleted (Fig. 1F).

Using quantitative immunoblotting, we determined that in a mixed population of cells FtsZ is present at 881 ± 286 molecules per cell (mean ± standard deviation) and that ZapA is present at 109 ± 41 molecules per cell (Supporting information Fig. S1). This is comparable to the 12:1 (FtsZ:ZapA) ratio suggested from ribosome profiling experiments (Schrader et al., 2014). Though we were unable to quantify ZauP levels due to poor antibody performance, ZapA and ZauP are expected from ribosome profiling data to be present at roughly 1:1 stoichiometry. To explore the functional consequences of altered ZapA and/or ZauP levels, we overproduced them individually or together and probed for levels of each protein at 4, 8 and 24 hours (Fig. 1G–J). Since we did not observe changes in protein levels between the 4 and 8 h time points (Fig. 1J), we examined the resulting cell morphologies after 4 h of overproduction and found that individual overexpression strains were morphologically indistinguishable from WT (Fig. 1G–I). However, overproduction of both ZapA and ZauP resulted in mild cell elongation, indicating a minor defect in the efficiency of cell division when FtsZ:ZapA:ZauP stoichiometry is disrupted (Fig. 1H,I).

ZauP co-localizes with ZapA at the division site

ZapA has previously been demonstrated to be an early recruit to the division site in C. crescentus (Goley et al., 2011). To assess the localization of ZauP, we replaced zauP with zauP-venus at the zauP locus. ZauP-Venus showed midcell localization in most cells, and localized to cell poles in swarmer cells (Fig. 2A). The ZauP-Venus fusion was fully functional in promoting cell division, with a mean cell length indistinguishable from WT (2.62 ± 0.02 µm, mean ± SEM). We performed demograph analysis on a population of ZauP-Venus-producing cells to understand the cell cycle profile of its localization. The smallest cells, representing newborn swarmer cells, showed a polar focus of ZauP, followed by recruitment to a broad midcell band that focused into a ring by mid-cell cycle (Fig. 2B). This localization pattern is reminiscent of that exhibited by ZapA and other early cell division proteins (Goley et al., 2011).

Figure 2. ZauP co-localizes with ZapA and FtsZ at the new cell pole and the division site.

A. Fluorescence, phase contrast, and merged images of strain EG992, bearing zauP-venus at its native locus. Bar = 2 µm. B. Demograph showing normalized ZauP-Venus fluorescence intensity as a function of cell length (blue = least intense pixel in each cell, red = most intense pixel in each cell) of a population of cells in A. C. Merged fluorescence and phase contrast images of strain EG894, bearing zapA-mCherry at its native locus and xylose-inducible zauP-venus at the xylX locus. Cells were grown with 0.3% xylose for 1.5 h prior to imaging. Hatched arrows indicate early stalked cells with midcell assemblies of both ZauP-Venus and ZapA-mCherry. Zoomed in image shows ZapA and ZauP co-localize. Bar = 2 µm. D. Merged fluorescence and phase contrast images of strain EG1361, bearing zauP-venus at its native locus and vanillate-inducible FtsZ-CFP at the vanA locus. Cells were grown with 0.5 mM vanillate for 1 h prior to imaging. Images inside box are merged and zoomed-in to show co-localization of the two signals. E. Live-cell PALM imaging of Dendra2 fusions to FtsZ (EG2038), ZapA (EG1956), and ZauP (EG1957). Images are shown in the order of bright-field image (i), ensemble fluorescence image (ii) and PALM image displayed in pseudo color (iii). Approximate cell outlines are indicated by yellow dashed lines. Scale Bars = 1 µm. F. Summary of PALM measurement values.

To observe directly the spatial relationship between ZauP and ZapA, we imaged fluorescent fusions to both proteins in the same cells. ZauP-Venus and ZapA-mCherry showed near-perfect co-localization, localizing to overlapping structures in almost every cell (Fig. 2C). Notably, in early stalked cells, both localized to weak midcell structures, indicating that they arrive at the same time to the incipient division site (Fig. 2C, double headed arrow). We also visualized ZauP-Venus in cells also producing FtsZ-CFP and observed a similar pattern of co-localization as for ZauP and ZapA (Fig. 2D). In sum, our observations indicate that like ZapA, ZauP is an early recruit to the divisome in C. crescentus.

In E.coli, ZapA interacts with and recruits another small coiled-coil protein, ZapB, to midcell (Galli and Gerdes, 2010). ZapB stabilizes the divisome by interacting with ZapA and the nucleoid-associated protein MatP, and localizes to a ring that is concentric to ZapA and FtsZ (Galli and Gerdes, 2010; Espéli et al., 2012). To determine if ZauP is concentric to the Z-ring similar to E.coli ZapB, we turned to super-resolution photoactivated localization microscopy (PALM). We initially set out to image pairs of these proteins in the same cells by PALM, but found that the fluorophores available for two-color PALM did not perform well in C. crescentus. PAmCherry fusions to FtsZ, ZapA, or ZauP, in particular, localized to patchy, non-ring structures in most cells. Live cell PALM imaging was therefore performed on each strain individually expressing Dendra2 fusions of FtsZ, ZapA or ZauP with a spatial resolution of ~42 nm (Supporting Information Fig. S2–5).

Qualitatively, the three proteins exhibited similar, ring-like morphologies with apparently heterogeneous intensity distributions (Fig. 2E and Supporting Information Fig. S3). We measured the diameters of the rings formed by FtsZ-Dendra2 (n = 146), ZapA-Dendra2 (n = 96) and ZauP-Dendra2 (n = 106) in non-constricting cells. We found the mean diameters of the rings to be 288 ± 68 nm, 288 ± 70 nm and 296 ± 62 nm, respectively (Fig. 2F and Supporting Information Fig. S4). Neither the means nor the distributions of ring diameters among the three strains are significantly different from each other (ks test: FtsZ/ZapA p = 0.9968, FtsZ/ZauP p = 0.7923, ZapA/ZauP p = 0.7991). We conclude that FtsZ, ZapA and ZauP form similar heterogeneous ring-like structures at midcell, with no evidence that ZauP forms a ring that is concentric to ZapA and FtsZ.

ZauP requires ZapA and FtsZ for localization at midcell

Having demonstrated that ZauP contributes to the efficiency of cytokinesis and localizes to the division plane, we next sought to determine how it is recruited to the divisome. We first asked if ZauP requires FtsZ for its midcell localization. We depleted FtsZ in cells producing ZauP-Venus or, as a control, ZapA-mCherry. ZapA-mCherry was previously demonstrated to be diffuse in cells lacking FtsZ (Goley et al., 2011). As expected, both ZauP-Venus and ZapA-mCherry localized to midcell bands in the presence of xylose inducer for ftsZ, and ZapA-mCherry was completely diffuse by 4 h post-depletion of FtsZ (Fig. 3A). Interestingly, however, ZauP-Venus persisted at midcell (single headed arrow), polar foci (double headed arrow) or ectopic polls in cells depleted of FtsZ for up to 7 h (Fig. 3A, bottom). It should be noted that localized ZauP-Venus did not form rings or bands in the absence of FtsZ, but rather small foci. FtsZ protein is no longer detectable at this time point (Fig. 3B), so we conclude that FtsZ is not required to maintain ZauP foci over this time period.

Figure 3. ZauP foci persist upon FtsZ depletion, but require ZapA for formation.

A. Merged fluorescence and phase contrast images of ZapA-mCherry (EG700) and ZauP-Venus (EG1006) upon depletion of FtsZ for the indicated amount of time. Arrows indicate midcell rings or foci. Hatched arrows indicate polar foci. Bar = 2 µm. B. Immunoblot against lysates from the cells in A using FtsZ antisera (top) or HU antisera (bottom). C. Merged phase contrast and fluorescence images of ZapA-Venus in WT background (EG1094) or in cells lacking zauP (EG1224). D. Merged phase contrast and fluorescence images of ZauP-Venus in WT background (EG798) or in cells lacking zapA (EG954). E. Merged phase contrast and fluorescence images of ZauP-mCherry upon depletion of ZapA in strain EG1414 for the indicated amount of time. F. Immunoblot against lysates from the cells in E using ZapA, SpmX and FtsZ antisera. EG1080 = ΔzapA control.

We next assessed the localization dependence between ZauP and ZapA. We first examined the localization of ZapA-Venus in the zauP deletion background. ZapA-Venus was observed in midcell bands and other structures in ΔzauP cells indicating that it can be recruited to sites of FtsZ assembly independent of ZauP (Fig. 3C). Conversely, ZauP-Venus was completely diffuse in cells lacking ZapA (Fig. 3D). This result may seem in conflict with the observation that ZauP continued to localize to foci in FtsZ-depleted cells when ZapA was presumably de-localized. However, in this case, zapA is deleted and, therefore, has been absent for many generations. To observe ZauP localization in cells depleted of ZapA over a shorter time course, we made a strain expressing zauP-mCherry from its native locus in a xylose-dependent ZapA depletion background. In contrast to what we observed for FtsZ depletion, ZauP becomes completely diffuse after 8 hours of ZapA depletion (Fig. 3E) even in the presence of a small amount of ZapA (Fig. 3F). This suggests that ZauP requires ZapA to localize at midcell in the presence of FtsZ as well as to persist as foci in the absence of FtsZ since deleting or depleting ZapA causes ZauP to become diffuse. However, FtsZ is required to maintain ZauP in rings, as opposed to foci.

ZapA and ZauP physically interact

In E. coli, ZapA interacts directly with ZapB. Although zapB is not operonic with zapA, ZapB shares some similarities with ZauP. Each is a small coiled-coil protein that localizes to the division site, requires ZapA for midcell localization, and has a mild cytokinesis defect when deleted. We therefore postulated at this point that, while they share no detectable sequence similarity, ZauP might be a functional ortholog of ZapB. In E. coli, ZapB and ZapA physically interact. We therefore asked if ZauP binds directly to ZapA. We took two approaches to answer this. First, we assessed the abilities of ZapA and ZauP to self-interact and to interact with each other using the bacterial adenylate cyclase two-hybrid (BACTH) system (Fig. 4A). We tested all combinations of T18 and T25 fusions for ZapA and ZauP. We observed self-interactions for both proteins, consistent with their predicted coiled-coil structures. Importantly, we also observed ZapA-ZauP interactions. The only combinations in which we did not observe an interaction were those in which ZapA was tagged at its C-terminus, suggesting that this region of ZapA may be important both for self-interaction and interaction with ZauP.

Figure 4. ZapA interacts with ZauP and FtsZ, but does not affect FtsZ assembly in vitro.

A. E. coli BTH101 cells bearing plasmids for expression of T18 and T25 fusions to the indicated proteins plated on media with IPTG and X-Gal. C: C-terminal fusion, N: N-terminal fusion. B. Colloidal Coomassie-stained SDS-PAGE of equal fractions of supernatant (S) and pellet (P) samples after co-sedimentation of reactions containing the indicated purified proteins ± 10 mM MgCl₂. Image is representative of 3 independent experiments. C. Transmission electron micrographs of 5 µM of each of the indicated proteins in HEK50 buffer with 10 mM MgCl₂ stained with uranyl formate. Bottom: zoomed images of boxed regions in the top row of images. D. Colloidal Coomassie-stained SDS-PAGE of equal fractions of supernatant (S) and pellet (P) samples after co-sedimentation of reactions containing the indicated purified proteins ± GMP-CPP. E. GTP hydrolysis over time by 4 µM FtsZ in the absence or presence of 16 µM ZapA and/or ZauP. F. Electron micrographs of indicated 4 µM FtsZ and 8 µM ZapA and ZauP in HEK50 buffer with 2 mM GTP and 2.5 mM MgCl₂ stained with uranyl formate.

To add confidence to our BACTH findings, we set out to demonstrate a direct interaction between ZapA and ZauP using biochemical methods. We purified ZauP and ZapA as His₆-SUMO fusions and cleaved the His₆-SUMO tag to yield untagged proteins. To test for an interaction, we asked if, like E. coli ZapA and ZapB, these proteins formed large structures that co-pelleted in the presence of divalent cations. We incubated ZapA alone, ZauP alone, or both proteins together in the absence or presence of 10 mM MgCl₂, subjected these reactions to high speed centrifugation and determined the proteins present in the supernatant and pellet fractions by SDS-PAGE and colloidal Coomassie staining (Fig. 4B). Interestingly, we found that ZapA had a tendency to move to the pellet in the presence of MgCl₂, whereas ZauP remained soluble. However, when combined in the presence of MgCl₂, both proteins were found almost exclusively in the pellet. The tendency of ZapA to go to the pellet during sedimentation is MgCl₂-dependent (Fig. 4B, Supporting Information Fig. S6). We observed very little ZapA in the pellet in the absence of MgCl₂ and roughly 80% in the pellet at 10 mM MgCl₂. We observed comparable, intermediate amounts of ZapA in the pellet (~60%) at 1 mM and 2.5 mM MgCl₂.

E. coli ZapB assembles into filament bundles in the presence of divalent cations, and these filaments are further bundled by ZapA (Galli and Gerdes, 2010; Galli and Gerdes, 2011). To determine if C. crescentus ZauP and/or ZapA form polymers, we performed negative stain transmission electron microscopy (TEM) on reactions containing ZapA, ZauP, or both proteins using similar conditions that have been reported for E.coli ZapA and ZapB. By itself, ZapA assembled into heterogeneous oligomeric or aggregated structures (Fig. 4C and Supporting Information Fig. S7). However, unlike E. coli ZapB, which polymerizes on its own, ZauP did not form large polymeric structures, as we only observed a few aggregates. E.coli ZapA and ZapB together are more prone to formation of large filament bundles; therefore we imaged C. crescentus ZapA and ZauP under similar conditions. We observed structures that were larger than those formed by ZapA, alone, but did not observe any ordered polymers of ZapA and/or ZauP under any of the conditions reported to stimulate E. coli ZapA-ZapB assembly (Fig. 4C and Supporting Information Fig. S7). We conclude that, while ZapA and ZauP appear to co-assemble specifically in the presence of divalent cations, they do not form ordered polymers and are more heterogeneous in structure than what has been reported for E.coli ZapA and ZapB.

ZapA binds to FtsZ, but does not affect protofilament assembly, GTPase activity, or bundling

ZapA from B. subtilis, E. coli, and P. aeruginosa have each been demonstrated to bind to FtsZ and to promote formation of stable polymer bundles with reduced GTPase activity under some in vitro conditions. We therefore set out to ask if C. crescentus ZapA and/or ZauP interact directly with FtsZ and/or affect its polymerization properties. We attempted to test for interactions with FtsZ using BACTH, however adenylate cyclase fusions to C. crescentus FtsZ appear to be highly toxic in E. coli. The fusions we made were either not expressed in the BTH101 strain or they acquired inactivating mutations in ftsZ, disallowing use of this assay for interactions with FtsZ.

To test directly for binding, we incubated purified ZapA, ZauP, or both proteins with FtsZ under polymerizing or non-polymerizing conditions, subjected each reaction to high-speed centrifugation and assessed the amount of protein in the supernatant and pellet. For this assay, we polymerized FtsZ in the presence of 2.5 mM MgCl₂ to minimize ZapA/ZauP pelleting in the absence of FtsZ. Moreover, we used GMPCPP rather than GTP to induce FtsZ polymerization to enrich for stable polymers in the pellet, as at this MgCl₂ concentration, the GTPase activity of FtsZ is high and polymers turn over rapidly and pellet inefficiently. Using this assay, ZapA was recruited to the pellet in the presence of FtsZ under polymerizing conditions, indicating direct binding to FtsZ polymers (Fig. 4D and Supporting Information Fig. S8A,B). ZauP was not recruited to the pellet, however, unless ZapA was also included in the reaction. As demonstrated for 10 mM MgCl₂, ZapA and ZauP co-pelleted even in the absence of FtsZ (Fig. 4D and Supporting Information Fig. S8A,B). However, the fraction of each in the pellet increased further in the presence of FtsZ polymers, suggesting that ZapA can interact with both ZauP and FtsZ polymers at the same time (Fig. 4D and Supporting Information Fig. S8A,B).

We next asked if ZapA and/or ZauP affect FtsZ polymer assembly by monitoring tryptophan fluorescence of FtsZ L72W, which contains a tryptophan substitution near where two monomers interact (Milam and Erickson, 2013). Protofilament formation buries the tryptophan residue so that a higher fluorescence signal is observed in the polymer than in monomeric form. We set up reactions containing FtsZ L72W alone or with ZapA, ZauP, or both and monitored tryptophan fluorescence emission before and after addition of GTP (indicated by arrow Supporting Information Fig. 8C). We did not observe any changes in protofilament assembly kinetics or quantity in the presence of ZapA, ZauP or both (Supporting Information Fig. S8C). We further asked if ZapA and ZauP affect the GTPase activity of FtsZ using a malachite green assay to monitor inorganic phosphate (P_i) release over time. FtsZ, alone, had a turnover number of ~4 P_i per minute per FtsZ molecule (Fig. 4E). Addition of ZapA, ZauP or both proteins did not significantly affect the GTPase activity of FtsZ (Fig. 4E). We also saw no effect of ZapA or ZauP on FtsZ GTPase activity using stoichiometries close to the 8:1:1 FtsZ:ZapA:ZauP stoichiometry we calculated in cell lysates (Supporting Information Fig. S8D). The only condition under which we observed a mild ZapA-dependent reduction in the apparent GTPase activity of FtsZ was when we polymerized FtsZ at high FtsZ:ZapA ratio (1:4) in buffer containing 1 mM MgCl₂ (1.77, 1.55, 1.45, 0.99 P_i per minute per FtsZ molecule for 1:0, 1:1, 1:2, 1:4 FtsZ:ZapA ratios, respectively) (Supporting Information Fig. S9A). However, we did not observe a difference in protofilament assembly kinetics using the tryptophan assay under the same conditions (Supporting Information Fig. S9B).

In order to visualize potential effects of ZapA and ZauP on FtsZ protofilament organization, we polymerized FtsZ in the presence or absence of ZapA or ZapA and ZauP at 2.5 mM MgCl₂ and visualized the resulting polymers by TEM. FtsZ by itself formed mostly double and some single protofilaments under these conditions (Fig. 4F). In the presence of ZapA or ZapA and ZauP we saw mostly single protofilaments indicating that binding of ZapA to FtsZ protofilaments may limit the electrostatic interactions between FtsZ protofilaments that lead to filament pairing. We also imaged FtsZ protofilaments assembled in 1 mM MgCl₂ in the presence or absence of ZapA, since we observed a slight decrease in the GTPase activity at a 1:4 FtsZ:ZapA ratio and 1 mM MgCl₂ (Supporting Information Fig. S9A). We found that FtsZ protofilaments looked very similar with or without ZapA at this MgCl₂ concentration (Supporting Information Fig. S9C). Previous work has shown that FtsZ protofilaments become highly curved when polymerized in the presence of MipZ. To test if ZapA protects FtsZ from the effects of MipZ under these conditions, we performed TEM on reactions containing FtsZ and MipZ in the presence or absence of ZapA. We found that FtsZ protofilaments looked similar in the presence of MipZ regardless of the presence of ZapA (Supporting Information Figure 9C).

Together, the above results indicate that although C. crescentus ZapA binds directly to FtsZ polymers, ZapA and ZauP do not affect polymer assembly or stability and do not induce bundling of FtsZ protofilaments in vitro under the conditions tested.

ZapA and ZauP contribute to Z-ring focusing

Although ZapA and ZauP do not affect FtsZ polymer dynamics or structure in vitro using the assays described above, they interact with the Z-ring through ZapA and affect the efficiency of cell division. This prompted us to ask if loss of ZapA or ZauP affects FtsZ organization in the cell. To address this, we imaged FtsZ-CFP in WT cells or in cells lacking zapA, zauP, or both. As expected, in WT cells, FtsZ-CFP was observed as a polar focus in swarmer cells and in a tight band at midcell in stalked and pre-divisional cells (Fig. 5A). In contrast, in cells lacking either zapA or zauP, FtsZ-CFP frequently accumulated into more dispersed assemblies near midcell (Fig. 5B–C). Consistent with its more severe filamentation phenotype, cells lacking ZapA exhibited more frequent abnormal Z-ring organization than those lacking ZauP. The double mutant, ΔzauPzapA, was similar in FtsZ-CFP localization to the ΔzapA mutant (Fig. 5D). These data indicate that, although ZapA and ZauP do not affect FtsZ polymer stability or organization in vitro, they are important for assembly of a focused Z-ring at midcell in vivo.

Figure 5. The Z-ring is dispersed in cells lacking ZapA, ZauP or both.

A – D. Merged phase contrast (blue) and fluorescence (yellow) images and demographs (n = 400, maximum cell length = 4.5 µm) of WT (EG1215) ΔzapA (EG1176), ΔzauP (EG1171), and ΔzapAzauP (EG1172) cells expressing vanillate-inducible ftsZ-cfp from the vanA locus for 1 h. Arrows indicate examples of dispersed rings in cells of normal length.

Overproducing ZapA and ZauP enriches FtsZ at midcell

Since deleting zapA or zauP led to more dispersed Z-ring assembly, we next sought to test the effects of overexpressing zapA and/or zauP on Z-ring formation. To do this, we overexpressed zapA and/or zauP for 4 h in a strain bearing vanillate-induced ftsZ-cfp, synchronized cells, imaged every 12 minutes, and created demographs to visualize Z-ring distribution in each population. In cells overproducing ZapA or both ZapA and ZauP, FtsZ took longer to relocate from the new cell pole (observed as red peaks of intensity on the right side of each demograph) to midcell (Fig. 6). Specifically, cells overproducing ZapA or ZapA and ZauP had significant FtsZ localized at the pole at the 12 and 24 min time points. In contrast, in the empty vector control or cells overproducing only ZauP, FtsZ localized near midcell in almost all cells by 24 min post-synchrony (Fig. 6, Supporting Information Fig. S10). Additionally, we observed a difference in the rate of focusing of the Z-ring at midcell in cells overproducing ZapA or ZapA and ZauP as compared to the empty vector control. The control or ZauP-overproducing cells gradually assembled a focused Z-ring at midcell over the course of 30 to 50 minutes, with more diffuse FtsZ-CFP signal (exhibited as cyan flanking the central red peak in demographs) observed along the length of the cell between 12 and 48 min post-synchrony (Fig. 6, Supporting Information Fig. S10 A, B). In cells overproducing ZapA or both ZapA and ZauP, however, Z-rings became tightly focused at midcell soon after leaving the poles with little diffuse FtsZ-CFP signal observed along the length of the cells (Fig. 6, Supporting Information Fig. S10).

Figure 6. FtsZ localizes at the poles longer, but focuses more rapidly at midcell when ZapA and ZauP are overproduced.

Demographs of FtsZ-CFP in the indicated strains at the indicated times post-synchrony. Cells overproducing ZauP (EG1174), ZapA (EG1173), both (EG913), or neither (EG1175) for 4 h and expressing ftsZ-cfp for 1 h were synchronized and grown without inducer and imaged at the indicated time points post-synchrony. Maximum cell lengths used for demographs at t=12, 24, 36, 48, 60, and 72 min are 3, 3, 4, 5.1, 5.1, and 5.3 µm, respectively. n=300 for t=12 and 36, n=200 for t=24 and 72, n=275 for t=48, n=350 for t=60.

In addition to the dynamics of Z-ring assembly being perturbed by ZapA overproduction, we observed what appeared to be more intense localizations of FtsZ at the poles and in Z-rings, with less diffuse signal in the cytoplasm, for ZapA or ZapA and ZauP overproducing cells (Fig. 6, Supporting Information Fig. S10). To quantify this effect, we measured the fraction of FtsZ at midcell at 72 min post-synchrony in each strain. We found that cells overproducing ZapA, ZauP, or both had significantly higher fraction of FtsZ at midcell compared to the empty vector control (Fig. 7A). However, we did not find significant differences in the longitudinal width of the Z-ring (Full Width at Half Maximum (FWHM)) in cells overproducing ZapA or ZauP (Supporting Information Fig. S11). We did observe a small but significant decrease in the population distribution of Z-ring FWHM between the empty vector control and the double zapA zauP overexpression strain. This suggests that excess ZapA and ZauP may cause the Z-ring to become slightly narrower along the long axis of the cell.

Figure 7. FtsZ is enriched at midcell when ZapA and ZauP are overproduced.

A. Fraction of FtsZ-CFP at midcell at t = 72 min post-synchrony using cells from the experiment in Figure 6. EG1174, EG1173, EG913, and EG1175 are overproducing ZauP, ZapA, both, or neither protein, respectively. Significant differences between the mean signal of each strain and empty vector control determined using ANOVA with Dunnett’s Multiple Comparison post-test. ***: p < 0.001, **: p < 0.01. B. Percent of cells that have MipZ colocalized with FtsZ at the cell pole for the indicated strains imaged at the indicated time points after synchrony. Representative images are shown in Supporting Information Figure S12. EG1458 is overproducing ZapA and ZauP from a high copy plasmid and EG1460 is an empty vector control. Both strains express mipZ-yfp from the native locus and ftsZ-cfp from the vanA locus for 1 h. 0 min: n=142 for EG1458 and n=144 for EG1460; 10 min: n=122 for EG1458 and n=117 for EG1460; 20 min: n=178 for EG1458 and n=162 for EG1460; 30min: n= 148 for EG1458 and EG1460.

Persistence of FtsZ at the new cell pole for a longer period of time in cells overproducing ZapA or ZapA and ZauP might be explained by resistance of FtsZ to the destabilizing effects of MipZ in those cells. If this were true, we would expect more cells overexpressing zapA and/or zauP to have FtsZ and MipZ colocalized at the new cell pole prior to midcell Z-ring assembly as compared to empty vector. To test this, we constructed strains that overproduce ZapA and ZauP or bear an empty vector in a background that allows expression of mipZ-yfp from the native locus and ftsZ-cfp from the vanA locus. After 4 h of overexpression of zapA and zauP and one hour of expression of ftsZ-cfp, we synchronized cells, imaged FtsZ-CFP and MipZ-YFP at 10 minute intervals, and counted the number of cells that contained bipolar MipZ and polar FtsZ using ImageJ (Fig. 7B, Supporting Information Fig. S12). At 20 min post-synchrony, the number of cells that contained colocalized MipZ and FtsZ signal at the poles, i.e. bipolar MipZ and a polar FtsZ signal, was 54% for ZapA-ZauP overproducing cells compared to 27% in the empty vector control (Fig. 7B). Additionally, at 30 min post-synchrony, 42% of cells overproducing ZapA and ZauP had colocalized MipZ and FtsZ compared to only 5% in the empty vector control (Fig. 7B). Thus, we find that FtsZ persists at the poles in zapA and zauP overexpressing cells even though MipZ bipolarization, which requires centromere segregation, occurs normally (Supporting Information Fig. S12).

The ZapA-ZauP module is conserved in Gram-negative bacteria

ZapA homologs are found broadly across bacterial species, yet ZapB, its important partner protein in E. coli, is apparently restricted to a subset of ɣ-proteobacteria. As ZauP appears to fulfill a ZapB-like function in C. crescentus, we asked if ZauP is conserved. BLAST searching using C. crescentus ZauP as a query identified only a small number of homologs in closely related Caulobacter species. However, we reasoned that the primary sequence of ZauP might not be its most important conserved feature. Indeed, ZapB and ZauP share little similarity in primary sequence, yet exhibit some similar biochemical and physiological characteristics. We therefore searched for putative zauP homologs using three criteria: a likely homolog should immediately precede zapA in a predicted or possible operon, it should encode a small protein (<150 amino acids), and that protein should be predicted to be predominantly coiled-coil.

To come up with a list of candidate zauP homologs, we first used BioCyc (Caspi et al., 2014) to align C. crescentus CB15 zapA/CC3247 with a subset its homologs manually selected from diverse proteobacterial classes in their genomic contexts. Next, we visually searched these aligned genomic regions for small genes operonic with zapA for each organism. Finally, we used COILS (Lupas et al., 1991) to determine if the gene upstream of zapA encodes a predicted coiled-coil protein (Fig. 1B). This analysis indicates that ZauP is, in fact, broadly distributed among the proteobacteria (Supporting Information Table S1). We find representative homologs in most α-proteobacteria we searched, excluding the Rickettsia and Wolbachia species, as well as in numerous orders of ɣ- and β-proteobacteria.

Using this approach, possible identification of zauP homologs was technically limited to organisms with a gene recognizable as a homolog of C. crescentus zapA. That is, the BioCyc genome alignment tool only functions if it identifies a homolog to the gene of interest (in this case, zapA) in the target organism. The sequence conservation of zapA is poor when compared across bacteria, so to maximize identification of putative zauP homologs, we repeated this analysis starting with Pseudomonas aeruginosa zapA, Chromobacterium violaceum zapA, and Bacillus subtilis zapA. This bootstrapping approach allowed us to identify additional putative zauP orthologs in ɣ-proteobacteria, β-proteobacteria, and organisms outside the proteobacteria, respectively (Supporting Information Table S1). Table S1 provides a select subset of homologs illustrating groups in which zauP may be found and is not a comprehensive list; one representative from each order in which a zauP homolog was found is included. Although we discovered numerous putative homologs outside of the proteobacteria, no zauP homologs were identified in Gram-positive organisms using this method. In addition, zauP and zapB appeared to be mutually exclusive, as no putative zauP homologs were detected in organisms bearing a zapB homolog. We conclude that the ZapA-ZauP module is widely distributed in Gram-negative bacteria.

To test if ZauP or ZapA homologs from a closely related species might be functional in C. crescentus, we generated mNeonGreen fusions of Hyphomonas neptunium (Hn) ZapA and ZauP and observed their localization in C. crescentus cells. We found that HnZapA was diffuse in the cytoplasm in WT and ΔzapA C. crescentus cells (data not shown). However, we observed that HnZauP localizes to midcell in a ΔzauP C. crescentus strain (Supporting Information Fig. S13). Additionally, HnZauP does not localize to midcell in the ΔzauPzapA C. crescentus strain suggesting that it requires C. crescentus ZapA for midcell localization. Moreover, since we do not observe localization of H. neptunium ZauP to midcell in WT C. crescentus cells, we hypothesize that C. crescentus ZauP competes for ZapA binding (Supporting Information Fig. S13). Length analysis, to test if H. neptunium ZauP rescues elongation of ΔzauP C. crescentus cells, was inconclusive since the loss of ZauP in C. crescentus has only a mild elongation phenotype.

DISCUSSION

Here, we describe two proteins, ZapA and ZauP, that are encoded in an operon and function to focus the Z-ring at midcell for efficient cell division in Caulobacter. In the absence of ZapA and/or ZauP, cells fail to divide as efficiently as wild-type, leading to mild elongation. Both ZapA and ZauP are recruited to the division machinery early in the cell cycle, and ZauP requires both ZapA and FtsZ to localize to a midcell ring. ZapA interacts directly with both ZauP and FtsZ, bridging these two proteins. However, ZapA and ZauP do not affect the GTPase activity, assembly kinetics, or bundling properties of FtsZ in vitro. Nevertheless, ZapA and ZauP contribute to assembly of a focused midcell Z-ring in vivo, suggesting a bundling-independent contribution to Z-ring stability.

Although ZauP and ZapB lack sequence similarity, they share biochemical, genetic, and cell biological characteristics. Each is predicted to be a small coiled-coil protein and is recruited to the division site by direct interaction with ZapA (Ebersbach, Galli, et al., 2008; Galli and Gerdes, 2010) (Fig. 2–Fig. 4). ZapB in E. coli was recently reported to form FtsZ-independent structures (Buss et al., 2017), similar to our observation that ZauP can persist for hours upon depletion of FtsZ in C. crescentus. Moreover, each facilitates efficient division in its cognate organism (Ebersbach, Galli, et al., 2008) (Fig. 1). However, important distinctions exist between ZapB and ZauP. Unlike ZapB, ZauP does not form higher ordered structures on its own in vitro (Fig. 4). Also unlike ZapB, ZauP is encoded in an operon with ZapA. We can envision two possibilities for the evolutionary relationship between these two factors. In the first, ZapB and ZauP are true homologs and evolved from a common ancestor. In that case, the ancient ZauP/ZapB was likely encoded in an operon with ZapA, as the zauPzapA operon is broadly distributed in Gram-negative bacteria, including ɣ-proteobacteria (Supporting Information Table S1). In this scenario, E. coli zapB moved elsewhere in the genome some time after divergence of the ɣ-proteobacteria. Alternatively, ZapB arose by convergent evolution in a subset of ɣ-proteobacterial lineages. Because we cannot conclude that ZauP and ZapB are true homologs, we favor retaining distinct names for these proteins.

In E.coli, ZapB additionally interacts with MatP, a protein that associates with DNA loci near the chromosomal terminus (Espéli et al., 2012; Buss et al., 2015). ZapB is therefore proposed to provide a physical link between the nucleoid and divisome, additionally stabilizing the Z-ring and coordinating division with chromosome segregation. Caulobacter lacks a MatP homolog, however, raising the question of whether ZauP fulfills such a bridging function. We hypothesized that ZauP might bind to DNA directly to physically link the divisome to the nucleoid. However, we did not observe general DNA-binding activity for ZauP using electrophoretic mobility shift assay (Supporting Information Fig. S14) nor did we observe an effect on nucleoid organization in cells overexpressing zauP (data not shown). If it exists, a link between ZauP and the nucleoid in Caulobacter, either through sequence-specific DNA binding by ZauP or through an as yet unidentified protein partner, awaits identification. Interestingly, MatP appears to only be found in organisms with ZapB homologs, not with ZauP homologs, that is, a small coiled-coil protein that is encoded in an operon with zapA.

Unlike ZapA from other organisms, C. crescentus ZapA does not affect the GTPase activity, protofilament assembly, or bundling of FtsZ under any conditions tested in vitro, indicating that ZapA and ZauP promote midcell positioning of the Z-ring through a bundling-independent mechanism (Fig. 8, Fig. 4E, Supporting Information Fig. S8–9). This is consistent with previous observations that the bundling activity of E. coli ZapA is limited under physiological conditions (Mohammadi et al., 2009) and that E. coli ZapA can cross-link FtsZ protofilaments into non-parallel networks (Dajkovic et al., 2010). A bundling-independent function for ZapA is also consistent with results from super-resolution imaging of the E. coli Z-ring in cells lacking ZapA or ZapB (Buss et al., 2013). Without ZapA or ZapB, E. coli cells still assemble clusters containing multiple FtsZ filaments, but these clusters fail to align properly at midcell, similar to our observations of FtsZ localization in Caulobacter cells lacking ZapA or ZauP (Fig. 5). In fact, high resolution imaging studies of the Z-ring in vivo in a variety of organisms have failed to identify large, densely-packed bundles of FtsZ filaments resembling those observed in vitro with proteins like ZapA, calling into question their physiological relevance (Li et al., 2007; Fu et al., 2010; Strauss et al., 2012; Holden et al., 2014; Jacq et al., 2015). Specifically, electron cryotomography (ECT) imaging of Caulobacter FtsZ in vivo revealed a relatively sparse assembly of single protofilaments at midcell and no apparent bundles (Li et al., 2007). These data are supported by super-resolution light microscopy studies that suggest a loosely packed collection of protofilaments within the Z-ring (Fu et al., 2010; Holden et al., 2014; Jacq et al., 2015). As Caulobacter lacks proteins that promote large-scale bundling of FtsZ filaments in vitro and has an abundance of single filaments in vivo, we favor a model wherein single protofilaments or, perhaps, protofilament pairs are the physiologically relevant FtsZ species.

Figure 8. Bundling-independent Z-ring focusing by ZapA-ZauP.

ZapA interacts with itself, ZauP, and FtsZ at sites of FtsZ assembly, providing a stable platform for Z-ring assembly. When ZapA and/or ZauP are absent, this stable platform is absent, leading to dispersion of the Z-ring. When zapA is overexpressed alone or with zauP, FtsZ is more stable and is enriched at midcell.

How, then, do ZapA and ZauP contribute to efficient cytokinesis if not through filament bundling? Their most obvious effect is in promoting assembly of a stable, focused FtsZ ring at midcell. The Z-ring has a remarkably consistent width – roughly 100 nm – relative to the long axis of the cell in diverse Gram-negative and Gram-positive organisms (Fig. 2) (Fu et al., 2010; Strauss et al., 2012; Holden et al., 2014; Jacq et al., 2015). These observations imply that conservation of Z-ring width is under evolutionary pressure and, therefore, important for the efficiency of cytokinesis. Ribosome profiling indicates that most late divisome proteins are present in low numbers (Schrader et al., 2014), thus a more dispersed Z-ring could lead to dilution of these limiting factors and, therefore, inefficient envelope remodeling. Alternatively, or in addition, a focused Z-ring provides a more robust directional cue for radial invagination of the cell envelope than dispersed FtsZ clusters. Indeed, in E. coli cells lacking ZapA or ZapB, dispersed FtsZ clusters promote formation of multiple discontinuous invaginations, rather than a complete, ring-like septum (Buss et al., 2013).

The next question is how ZapA and ZauP promote formation of a focused Z-ring without affecting the bundling or stability of FtsZ polymers. A biophysical study of E. coli ZapA suggested that in addition to promoting formation of parallel bundles of FtsZ, E. coli ZapA is able to cross-link FtsZ filaments into an interconnected network (Dajkovic et al., 2010). Cross-linking activity is not straightforward to detect by TEM, as filaments tend to stick to the grid in a tightly packed, overlapping manner even for FtsZ, alone. Thus, it remains possible that Caulobacter ZapA has FtsZ cross-linking activity that helps condense FtsZ filaments into a narrow midcell zone in vivo. However, as proposed for ZapA-ZapB (Galli and Gerdes, 2010; Buss et al., 2013), an alternative for how ZapA and ZauP promote Z-ring focusing lies in the ability of these factors to form higher order oligomers, themselves. Although neither Caulobacter ZapA nor ZauP forms filaments in vitro under the conditions tested, ZapA formed oligomers or aggregates in the presence of divalent cations that were larger in the presence of ZauP (Fig. 4C). While we cannot be certain that these structures are relevant in the cell, ZauP persisted in stable foci upon depletion of FtsZ, indicating formation of fairly stable higher order assemblies in vivo that depend on the presence of ZapA (Fig. 3). We hypothesize, therefore, that ZapA-ZauP hetero-oligomers assemble in association with FtsZ polymers near midcell, forming a stable platform that reinforces FtsZ assembly in a narrow midcell zone (Fig. 8). This mechanism for reinforcing Z-ring dimensions increases the robustness and efficiency of divisome-mediated envelope constriction and appears to be conserved in diverse Gram-negative organisms.

EXPERIMENTAL PROCEDURES

Bacterial strains, plasmids, and growth conditions

All C. crescentus strains used are derivatives of the lab-adapted NA1000 strain and were grown in peptone yeast extract (PYE) medium at 30°C. Antibiotics were used at the following concentrations in liquid (solid) media for C. crescentus: kanamycin 5 (25) µg/mL, gentamycin 1 (5) µg/mL, spectinomycin 25 (100) µg/mL, and streptomycin (5 µg/mL). Inducers and additives were used at the following concentrations in liquid (solid) media: 0.3 (0.2)%, glucose 0.2%, and vanillate 0.5 (0.5) mM. For depletion experiments, cells were washed 3 times in plain media prior to resuspension in media with or without inducer. Cloning was carried out using standard molecular biology procedures and all plasmids were sequence-verified. Escherichia coli TOP10 and NEB Turbo strains were used for cloning purposes and grown at 37°C in Luria-Bertani broth with shaking. For E. coli, antibiotics were used at the following concentrations in liquid (solid) media: ampicillin 50 (100) µg/mL, kanamycin 30 (50) µg/mL, gentamycin 15 (20) µg/mL, and spectinomycin 50 (50) µg/mL). Strains and plasmids used in this study are listed in Supporting Information Table S2.

Phase contrast and ensemble fluorescence microscopy and image analysis

Cells were imaged during exponential growth after immobilization on 1% agarose pads. Light microscopy was performed on a Nikon Eclipse Ti inverted microscope equipped with a Nikon Plan Fluor 100X (NA 1.30) oil Ph3 objective and Photometrics CoolSNAP HQ² cooled CCD camera. Chroma filter cubes were used as follows: ET-EYFP for Venus, ET-ECFP for CFP, and ET-dsRED for mCherry. Images were processed in Adobe Photoshop. Automated cell length and demograph analyses (Hocking et al., 2012) were performed using MicrobeTracker (Sliusarenko et al., 2011) and Oufti (Paintdakhi et al., 2016).

Live cell PALM Imaging

Liquid cultures of EG2038 (xylX::ftsZ-dendra2), EG1956 (xylX::zapA-dendra2), or EG1957 (xylX::zauP-dendra2) in PYE were grown overnight at 30°C until they reached stationary phase (~OD 1.0). Cells were re-inoculated 1:1000 into M2G medium and grown again overnight at 30°C to early exponential phase (~OD 0.1). 0.003% xylose was added and the cells were grown for 2 hrs at RT at which point they had begun to reach mid-exponential phase (~OD 0.2). The culture was then centrifuged for 3 min at 4100 rpm and resuspended in fresh M2G, and was shaken for 1.5 hr at room temperature (RT) in the absence of xylose. After the RT out-growth, 1 mL of the culture was centrifuged for 1 min at 13k rpm at RT, resuspended, and centrifuged again for 1 min at 13k rpm at RT. The pellet was resuspended in 50 µL of fresh M2G and vortexed for ~40 sec to avoid cell aggregation.

Cells harboring different plasmids as described above and were applied to an agarose gel pad (3% in M2G medium) laid in a microscope chamber (FCS2, Bioptechs). The chamber was locked on the microscope stage (ASI, Eugene, OR) to minimize mechanical drifts. PALM imaging was performed on an Olympus IX71 inverted microscope with a 60×, 1.45 NA oil-immersion objective. Dendra2 was switched to the red-emitting state using a 405 nm laser with a light intensity increasing from 2 to 37 W cm⁻² for FtsZ and ZauP and 2 to 162 W cm⁻² for ZapA stepwise to compensate for the gradual depletion of inactivated Dendra2. The activated red-emitting Dendra2 was excited at 561 nm continuously, with laser intensity of 10 kW cm⁻² and exposure time of 15 ms. All fluorescence emission signal was collected using a dual-band emission filter (ZET 488/561nm, Chroma Technology) before acquisition by an EMCCD camera (iXon Ultra 897, Andor Technology). All the lasers and the EMCCD camera were controlled by the MetaMorph software (Molecular Devices). Each imaging aquisition was completed in less than 45 seconds to minimize sample drift.

All resulting PALM images were constructed from 3,000 frames of the imaging sequence. Single molecule detection and localization were carried out using the ThunderSTORM plugin in ImageJ (Ovesný et al., 2014). Resulting PALM images and corresponding dimension measurements (ring diameters and band widths) were performed using custom MATLAB (MathWorks, Inc., Natick, MA) software as described previously (Buss et al., 2013). Individual cells were selected for analysis on the basis that they met the following four criteria: (1) the cells appeared as non-constricting by their brightfield images, (2) the cells demonstrated a midcell band, (3) two-maxima were present at either end of the band along the short-axis of the cell when the structure was projected to 2-dimensional (2D) imaging planes and (4) the midcell band extended the length of the short-axis when overlaid with the brightfield image.

Measuring the fraction of the FtsZ signal at midcell

Images of the cells harboring FtsZ-CFP fusions were blinded before analysis and analyzed using ImageJ. First, the background intensity was subtracted from the CFP channel and the CFP channel was overlaid with the phase-contrast images of the cells. The total cell region-of-interest (ROI) was constructed using the phase-contrast channel by manually outlining the cell body using the polygon selections tool. Next, the Z-ring ROIs were determined using the CFP channel and were similarly outlined using the polygon selections tool. The integrated intensity for each of the ROIs were measured in ImageJ and recorded. The fraction of signal at midcell was calculated as the total integrated signal in the Z-ring ROI divided by the total integrated signal of the whole cell ROI.

Calculating the Full-Width at Half-Maximum (FWHM) of Z-rings from Oufti output

A custom MATLAB script was created to measure, in a high-throughput manner, the FWHM of Z-rings in a sample population directly from data processed using Oufti (Paintdakhi et al., 2016). Briefly, for each cell, the FtsZ-CFP intensity line scan was extracted from the Oufti output. To isolate the Z-ring signal from each intensity array, the maximum of the signal array was identified and denoted the midcell position on the basis that Z-rings form at roughly the center of the cell. Next, a threshold zone containing a quarter of the total signal elements to the left and the right the maximum position was extracted and normalized such that the maximum signal was equal to 1 and minimum signal was equal to 0. This new normalized signal array was then fit to a 2D Gaussian distribution. If the goodness-of-fit (GOF, R²) value was greater than 0.9 the fit and resulting FWHM value were recorded. Otherwise, if the GOF value was below the 0.9 threshold, the measurement was excluded from the analysis.

Immunoblotting

Cells were lysed in SDS-PAGE loading buffer by boiling for 5 min. SDS-PAGE and transfer of protein to nitrocellulose membrane were performed using standard procedures. Antisera were used at the following dilutions: FtsZ, 1:20000 (Sundararajan et al., 2015); HU, 1:2000 dilution (Bowman et al., 2010); MipZ, 1:5000 (Thanbichler and Shapiro, 2006); SpmX, 1:50000 (Radhakrishnan et al., 2008); ZapA, 1:2500; ZauP, 1:2000. Antibodies against ZapA and ZauP were generated by immunizing a rabbit and a chicken with purified ZapA or ZauP, respectively. Specificity of the antibodies was determined using cell lysates that had either protein deleted or overproduced and comparing the signal to a WT lysate.

Immunoblotting to quantify FtsZ and ZapA levels in the cell

Purified FtsZ and ZapA proteins were diluted in SDS-PAGE loading buffer, boiled for 5 minutes and loaded on a gel to serve as standards. Cells were harvested in log phase, resuspended in SDS-PAGE loading buffer at concentrations of 0.006 ODU/µL for ZapA detection and 0.003 ODU/µL for FtsZ detection, boiled, and loaded next to standards. Alexa-Flour 647 conjugated secondary antibody was used to probe for primary antibodies of FtsZ and ZapA and the signal was detected using Amersham Imager 600 RGB. For quantification, band intensity was determined using ImageJ and concentrations in lysates were calculated using standard curves generated from purified proteins. 2.275*10⁹ cells were approximated to be in 1 mL of culture at OD₆₀₀ of 1, as described by Thanbichler and Shapiro (Thanbichler and Shapiro, 2006).

Bacterial two-hybrid analysis

C- and N-terminal T18 and T25 fusions to zapA and zauP were generated using plasmids pKNT25, pKT25, pUT18, and pUT18C, sequence-verified, and co-transformed in all possible T18/T25 combinations into E. coli BTH101 (Karimova et al., 1998). At least three independent clones of each T18/T25 combination were isolated and grown in LB medium with antibiotics and 0.5 mM IPTG overnight at 30°C with shaking. Overnight cultures were used to patch cultures onto indicator LB agar plates with 100 mg/mL ampicillin, 50 µg/mL kanamycin, 40 µg/mL X-gal, and 0.5 mM IPTG. Plates were incubated at 30°C overnight prior to imaging.

Protein production and purification

FtsZ was overproduced from plasmid pMT219 and purified essentially as described previously (Sundararajan et al., 2015) with modifications. Rosetta (DE3) pLysS E. coli cells bearing plasmid pMT219 were induced with 0.5 mM IPTG for 3–4 hours at 37 °C after achieving an OD600 of 0.8–1.0. Cells were pelleted by centrifugation, resuspended in Buffer QA (50 mM Tris-HCl pH 8.0, 50 mM KCl, 1 mM EDTA, 1 mM β-mercaptoethanol, 10% glycerol), frozen in liquid nitrogen and stored at −80 °C until purification. Cell suspensions were thawed at 37 °C, and a Mini Complete Protease Inhibitor tablet (Roche), lysozyme (1 mg/mL), PMSF (2 mM), and DNAse I (New England Biolabs) (2 units/mL) were added. Cell suspensions were incubated with rocking at room temperature for 30–60 min to allow lysis, sonicated, and centrifuged for 30 min at 15,000 × g, 4°C to pellet debris and unlysed cells. The supernatant was filtered and loaded onto two HiTrap Q HP 5 mL columns (GE Life Sciences) equilibrated in Buffer QA. Protein was eluted using a linear KCl gradient (50–500 mM over 20 column volumes). Peak fractions were combined, ammonium sulfate was added to 20% saturation, and after 1 h incubation at 4°C with stirring, the suspension was centrifuged at 10,000 × g for 10 min at 4°C. Precipitates were resuspended in FtsZ Storage Buffer (50 mM HEPES-KOH pH 7.2, 50 mM KCl, 0.1 mM EDTA, 1 mM β-mercaptoethanol, 10% glycerol) and consecutively applied to a Superdex 200 10/300 GL column (GE Life Sciences) in FtsZ Storage Buffer. Peak fractions were combined, concentrated, frozen in liquid nitrogen, and stored at −80 °C.

His-SUMO fusions to ZauP and ZapA were overproduced in Rosetta (DE3) pLysS E. coli cells from plasmids pEG619 and pEG620, respectively, by 4 h induction with 0.5 mM IPTG at 37°C. Cells were harvested by centrifugation, resuspended in lysis buffer (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 20 mM imidazole, and 10% glycerol), frozen in liquid nitrogen, and stored at −80°C until purification. Cell suspensions were thawed, lysozyme (1 mg/mL), MgCl₂ (2.5 mM), CaCl₂ (1 mM), and DNAse I (2 units/mL) were added, and cells were incubated with rocking at room temperature for 1 h. Cells were sonicated and centrifuged for 30 min at 15,000 × g and 4°C. Supernatants were filtered and loaded onto a HisTrap FF 1 mL column (GE Life Sciences) equilibrated in lysis buffer. Bound protein was eluted in lysis buffer with 300 mM imidazole. His-SUMO-Protease was added at a 1:100 molar ratio (protease:fusion) and solutions were cleaved overnight while dialyzing into lysis buffer at 4°C. Cleaved, untagged ZauP or ZapA was separated from His-SUMO and His-SUMO-Protease by loading dialysate onto a HisTrap FF 1 mL column equilibrated in lysis buffer and keeping the unbound fraction. Proteins were further purified by loading onto a HiTrap Q HP 1 mL column (GE Life Sciences) equilibrated in QA Buffer (50 mM Tris-HCL pH 8.0, 100 mM NaCl, 10% glycerol) and eluting with a linear NaCl gradient (100 mM – 1 M over 20 column volumes). Peak fractions were pooled, dialyzed in FtsZ Storage Buffer, concentrated, frozen in liquid nitrogen, and stored at −80°C. ZauP and ZapA were stained with ProtoBlue Safe Colloidal Coomassie G-250 Stain (National Diagnostics), as they stained poorly with Coomassie Brilliant Blue R-250.

Protein assembly assays

High speed pelleting and co-pelleting, and GTPase assays were performed in HEK50 buffer (50 mM HEPES-KOH pH 7.2, 50 mM KCl, 0.1 mM EDTA) as described previously (Goley et al., 2010; Sundararajan et al., 2015). When indicated, MgCl₂ was added to 1, 2.5 or 10 mM, GTP was added to 2 mM, or GMP-CPP (Jena Bioscience) was added to 0.2 mM. Tryptophan assay was performed by measuring tryptophan emission at 344 nm of FtsZ L72W ± FtsZ in HEK50 buffer with ZapA or ZauP at after exciting at 295 nm using Fluoromax-3 as described (Milam and Erickson, 2013).

Electron microscopy

For imaging ZapA and ZauP, reactions containing indicated amounts of each protein were set up in HEK50 buffer and divalent cations were added as indicated. Reactions containing FtsZ were polymerized in HEK50 buffer containing indicated amounts of MgCl₂ and 2 mM GTP along with ZapA, ZauP, and/or MipZ. Reactions containing MipZ also contained 2 mM ATP. Reactions were applied to glow discharged carbon coated grids and stained with 0.75% uranyl formate. Grids were dried and imaged with Philips/FEI BioTwin CM120 TEM equipped with an AMT XR80 8 megapixel CCD camera (AMT Imaging, USA).

Electromobility shift assay

Proteins were incubated with 26 nM a nonspecific 516 bp dsDNA fragment in EMSA buffer (2.5mM MgCl₂, 25 mM Tris/HCl (pH 8.0), 100 mM KCl, 10% glycerol and 50 µg/mL BSA) for 20 min. The MipZ reactions also contained 1.25 mM ATP. Reactions were loaded on to a 4–20% Criterion-TBE acrylamide gel (Bio-Rad) and run for 90min at 4°C. Gels were stained with 1% ethidium bromide and visualized using a Bio-Rad EZ Gel Doc imager. The DNA sequence used for this assay was amplified from Mycobacterium smegmatis gDNA using primers GACTGCTCTTCTGGTATGATTTTTAAGGTCGGAGAC and AGTCGCGGCCGCTCAGGACGCGGCGGCCAAA.

Bioinformatic analysis of zauP conservation

As described in the Results, BioCyc (Caspi et al., 2014) was used to align zapA homologs in their genomic context using randomly selected representatives of diverse bacterial orders that contained a homolog of C. crescentus zapA recognized as such in BioCyc. Manual inspection of the genomic region was used to identify any small ORF upstream of zapA, and COILS (Lupas et al., 1991) was used on a subset of these possible zauP homologs to assess if the protein encoded by the upstream gene was predicted to be predominantly coiled-coil in nature. A similar procedure was used with P. aeruginosa zapA, Chromobacterium violaceum zapA, and Bacillus subtilis zapA to identify more distantly related homologs.