FtsZ dynamics in bacterial division: what, how, and why?

Abstract

Bacterial cell division is orchestrated by the divisome, a protein complex centered on the tubulin homolog FtsZ. FtsZ polymerizes into a dynamic ring that defines the division site, recruits downstream proteins, and directs peptidoglycan synthesis to drive constriction. Recent studies have documented treadmilling of FtsZ polymer clusters both in cells and in vitro. Emerging evidence suggests that FtsZ dynamics are regulated largely by intrinsic properties of FtsZ itself and by the membrane anchoring protein FtsA. While FtsZ dynamics are broadly required for Z-ring assembly, their role(s) during constriction may vary among bacterial species. These recent advances set the stage for future studies to investigate how FtsZ dynamics are physically and/or functionally coupled to peptidoglycan metabolic enzymes to direct efficient division.

Introduction

Cell division in bacteria is a carefully controlled process that requires the coordination of multiple complex systems. In coordination with chromosome replication and segregation, cells must localize peptidoglycan (PG) cell wall enzymes to the site of division and coordinate their activities to synthesize PG, which is thought to be the driving force underlying constriction. The process of cytokinesis is carried out by the divisome, a protein complex responsible for directing division through spatiotemporal control of PG enzymes [1].

At the heart of the divisome lies FtsZ, an essential, polymerization-competent tubulin homolog that localizes to the division site to form a structure known as the “Z-ring”. The site of Z-ring assembly defines the site of division, and FtsZ acts to both recruit and regulate roughly a dozen essential cell division proteins that mediate constriction. Through signals that are not yet clear, the assembled divisome is activated to synthesize peptidoglycan, and the Z-ring constricts in diameter throughout the division process [1].

In recent years, significant advances have been made in understanding the nature, regulation, and function of the dynamic properties of FtsZ and the Z-ring at the polymer and cellular levels. This review aims to highlight recent work that has advanced our understanding of how the dynamics of this critical protein are regulated, and how FtsZ dynamics affect and contribute to the process of division.

FtsZ is dynamic on a cellular scale and within the assembled Z-ring

The Z-ring marks the site of cell division, and therefore Z-ring assembly and constriction must be coordinated with cell cycle events like chromosome replication and segregation. Z-ring assembly occurs during S phase and requires a shift in FtsZ polymer concentration toward mid-cell and away from the poles. In most bacteria, mid-cell assembly of FtsZ is driven by extrinsic positioning systems that exploit and modify intrinsic FtsZ polymer dynamics (Figure 1). The Min system in Escherichia coli and Bacillus subtilis and MipZ in Caulobacter crescentus function by localizing factors that prevent FtsZ polymerization near cell poles, effectively constraining Z-ring formation to mid-cell. Conversely, some bacteria instead rely on positive regulators of Z-ring assembly, such as PomZ in Myxococcus xanthus and MapZ in Streptococcus pneumoniae [2]. Walker et al. recently examined fluorescently labeled FtsZ through the cell cycle and observed that FtsZ forms short, transient, membrane-anchored assemblies moving mostly circumferentially along the length of the cell preceding Z-ring assembly [3]. The presence of transient precursors implies that FtsZ polymers stochastically sample the membrane via interactions with membrane anchors in the early stages of Z-ring assembly. The combined actions of Z-ring positioning systems and recruitment of additional factors promote maturation of these transient structures into a stably localized Z-ring at mid-cell (Figure 1).

Figure 1: A host of localization systems contribute to Z-ring assembly throughout the cell cycle.

Early in the cell cycle, FtsZ (blue haze) is rapidly redistributed from the poles throughout the cell. Prior to the subsequent division event as the chromosome (bronze line) is being replicated, the division site is selected by spatially regulated polymerization of FtsZ into a mid-cell ring. Depending on the organism, negative (green arrows) Z-ring regulators (Min system or MipZ) promote disassembly of FtsZ polymers near the poles, or positive (purple arrows) Z-ring regulators (MapZ or PomZ) promote local mid-cell assembly of FtsZ polymers (blue gradient lines). Z-ring assembly and condensation are further facilitated by association with membrane anchors (red spheres) and Z-binding proteins (orange ellipses), e.g. ZapA, at mid-cell.

FtsZ is also highly dynamic within the assembled Z-ring, with molecules in the ring turning over with a half-time of ~10–30 s as measured by fluorescence recovery after photobleaching (FRAP) [4,5]. Notably, the recovery time is significantly extended upon expression of a GTPase-deficient FtsZ variant, indicating a dependence of polymer dynamics on FtsZ’s GTPase activity. This observation is supported by previous studies that established FtsZ’s ability to form dynamic polymers upon binding and hydrolyzing GTP in vitro [6].

Recently, studies in several bacterial species have characterized a specific dynamic mode of FtsZ known as treadmilling, previously observed in vitro [7], wherein subunits are added to one end of a filament cluster and dissociated from the other resulting in the apparent directional motion of FtsZ clusters while single molecules of FtsZ are stationary. Using total internal reflection fluorescence microscopy (TIRFM) or imaging Z-rings in vertically-immobilized bacteria, FtsZ treadmilling has been observed in vivo with rates of ~30–40 nm/s - corresponding to roughly 7–9 FtsZ monomers/s - depending on the organism [8–12]. Cell geometry does not appear to affect FtsZ dynamics, as recovery times following photobleaching and treadmilling rates of FtsZ in E. coli cells induced to acquire different diameters or shapes are similar to those in wild-type cells [13]. However, perturbations that decrease or increase GTPase activity lead to a decrease or increase in treadmilling rates, respectively [8,9,12]. Interestingly, while treatment of cells with the GTPase inhibitor PC190723 eliminates treadmilling motion in both Staphylococcus aureus and B. subtilis [8,12], FtsZ structures remain dynamic by FRAP [14], suggesting the existence of yet uncharacterized dynamic modes beyond treadmilling.

Intrinsic regulation of FtsZ polymer dynamics

How are FtsZ dynamics regulated to potentiate the observed in vivo behaviors? In vitro, FtsZ assembles into straight or gently curved single protofilaments under physiological salt and buffer conditions [6]. Early studies employing transmission electron microscopy, GTPase assays, and light scattering to examine FtsZ assembly in solution identified GTP binding and hydrolysis as key factors driving polymer dynamics. Increasing Mg²⁺, Ca²⁺, or DEAE-dextran (a polycationic crowding agent) concentrations results in more stable polymers that form thick bundles or sheets, apparently by favoring abundant lateral filament interactions and reducing polymer disassembly [6].

The implementation of supported lipid bilayers (SLBs) coupled with TIRFM has allowed the in vitro study of FtsZ dynamics and superstructure in the physiologically relevant context of membrane association. Loose and Mitchison first observed treadmilling of purified E. coli FtsZ on SLBs with rates similar to those subsequently observed in vivo (~40 nm/s). They found that treadmilling requires sufficient local FtsZ concentration, GTP hydrolysis, and association with the SLB through a membrane anchoring protein, in their case E. coli FtsA [7] (Figure 2A). Wagstaff et al. recently provided a structural explanation for both cooperative assembly and treadmilling of single FtsZ protofilaments, proposing that both are enabled by a switch between open and closed FtsZ conformations that depends on its polymerization state [15]. These concepts have also been explored by Corbin and Erickson in the design of a Monte Carlo model for FtsZ dynamics that accurately recapitulates properties of FtsZ nucleation and treadmilling in vitro and in vivo and corroborates the hypothesis that these processes are driven by a conformational switch [16].

Figure 2: FtsZ treadmilling on SLBs depends on membrane association, GTPase activity, and proper lateral interactions.

Model for polymer behavior (left) and representative experimental view (right) of each scenario are represented. A. Treadmilling of E. coli FtsZ (blue gradient arrow and lines) occurs on SLBs when FtsZ can bind (dark blue) and hydrolyze (light blue) GTP and the local concentration of FtsZ at the membrane, facilitated by a membrane anchor (red spheres) or a membrane targeting sequence (not shown), is sufficient to allow for both polymerization and depolymerization. Treadmilling FtsZ is typically observed in either single clusters or chiral swirls, from each of which treadmilling rates can be determined. B. E. coli FtsZ forms asters or bundles/mesh-like structures under conditions of high E. coli ZipA or high free Mg²⁺ concentrations, respectively. Treadmilling has not been observed under these conditions, although polymers are still dynamic. C. C. crescentus FtsZ fused to a membrane targeting sequence (FtsZ-MTS) forms small dynamic clusters that do not appear to move by treadmilling under conditions tested thus far. D. C. crescentus FtsZ lacking its C-terminal linker (ΔCTL-MTS) forms stable, mesh-like structures that are more stable compared to WT.

Building on the above observations, Ramirez-Diaz et al. observed that FtsZ fused to a membrane targeting sequence (FtsZ-MTS) exhibits treadmilling on SLBs within a narrow range of Mg²⁺ concentrations, below which FtsZ cannot bind GTP and above which FtsZ instead forms dense mesh-like structures with increased single molecule residence times [17] (Figure 2B). These structures are likely facilitated by increased lateral interactions favored by high concentrations of free Mg²⁺ and are similar to the structures formed by a hyper-bundling variant of C. crescentus FtsZ-MTS on SLBs [18] (Figure 2D). These observations suggest a model in which treadmilling activity is highly dependent on local concentrations of FtsZ, GTP, and Mg²⁺ at the membrane, which brings into question what happens in vivo when these requirements are not met. In contrast to E. coli FtsZ, wild-type C. crescentus FtsZ-MTS does not appear to move directionally on SLBs. Instead, it forms discrete clusters that form and disassemble rapidly, suggesting potential species-specific differences in the intrinsic propensity of FtsZ clusters to undergo treadmilling motion [18] (Figure 2C).

In addition to the GTPase domain, FtsZ includes a conserved C-terminal peptide (CTC), a disordered C-terminal linker (CTL) between the GTPase domain and CTC and, sometimes, a short variable sequence at the extreme C-terminus (CTV), each of which may regulate polymer dynamics. In vitro assembly of FtsZ lacking its C-terminal linker (ΔCTL) results in bundled polymers that exhibit greater stability as compared to wild-type FtsZ from C. crescentus [18–20] (Figure 2C–D) and B. subtilis [21]. Production of ΔCTL in vivo results in disruptions in Z-ring morphology and cell division in B. subtilis [22], E. coli [23], and C. crescentus [19,24]. Taken together, these observations indicate a role for the CTL in regulating interfilament interactions, and consequently, FtsZ dynamics. Conversely, replacement or removal of the CTC, CTV, or both results in either a decrease (B. subtilis [25]) or no change (E. coli and C. crescentus) in bundling propensity and a modest decrease in GTPase rate [20,25]. Thus, the FtsZ C-terminus influences polymer dynamics and lateral interactions, serving as an intrinsic regulator of Z-ring dynamics in addition to the GTPase domain.

The role of binding partners in FtsZ dynamics: is bundling relevant in cells?

FtsZ binding partners have long been known to affect FtsZ’s localization and structure in cells and polymer dynamics and/or superstructure in purified in vitro systems. Specifically, a number of proteins can induce bundling – here defined as association of protofilaments into stable multi-filament structures with tight lateral interactions - of FtsZ filaments in vitro and generally reduce the apparent GTPase activity of FtsZ in the process [26]. The relevance of FtsZ bundling in cells is unclear, however, and mounting evidence suggests that at least some of the proteins previously described as “bundling” proteins likely impact Z-ring organization or function through bundling-independent mechanisms – and/or without impacting FtsZ dynamics - in cells.

One of the best-studied of these is the conserved coiled coil protein ZapA. Purified ZapA causes FtsZ polymers to form stable, multi-filament bundles under some conditions in vitro [27,28], and ZapA promotes formation of coherent Z-rings in vivo [27,29]. These observations led to a model wherein ZapA serves as a bundling protein that stabilizes the Z-ring by promoting lateral interactions and slowing polymer turnover. However, Caldas et al. recently demonstrated that ZapA does not affect FtsZ dynamics on SLBs in vitro, though it promotes organization of polymers into coherent superstructures [30]. These observations are consistent with findings in E. coli, C. crescentus, and B. subtilis that indicate that ZapA does not stabilize bundled FtsZ polymers in cells, but contributes to alignment of FtsZ clusters into a coherent, condensed ring [31–33].

The evolving model for the mechanism by which ZapA influences Z-ring assembly in cells calls into question the physiological relevance of bundling and polymer stability that can be induced by other FtsZ-binding proteins in vitro. Indeed, in C. crescentus the FtsZ binding protein FzlA can induce formation of hyperstable, helical bundles of FtsZ in vitro, but FzlA mutants that lack this activity are still able to support cell division, suggesting it may not be relevant in cells [34]. Similarly, EzrA and SepF, which can inhibit or promote FtsZ bundling in vitro, respectively, are dispensable in B. subtilis and do not affect FtsZ filament dynamics in cells [33].

Membrane anchors: mixed effects on FtsZ dynamics

Proteins that facilitate association of FtsZ with the membrane are important for Z-ring assembly and function, and they impact FtsZ dynamics and superstructure. The two best characterized of these are the broadly conserved actin homolog FtsA and the γ-proteobacterial protein ZipA. In E. coli, both FtsA and ZipA bind to the CTC of FtsZ and, although both are individually essential for cell division, either of the two is sufficient for Z-ring assembly [35]. ZipA comprises a short N-terminal transmembrane sequence, a proline- and glutamine-rich central domain, and a large cytosolic C-terminal region that binds the FtsZ CTC. Early experiments demonstrated that ZipA can induce bundling of FtsZ filaments in vitro [36]. In vitro experiments using SLBs have since observed that ZipA is sufficient to recruit FtsZ to a membrane where it ultimately forms straight bundles and asters [7,37] (Figure 2B). While these bundles are dynamic, the lack of individual clusters under these conditions makes it unclear whether treadmilling occurs in FtsZ-ZipA structures. Lowering the ZipA concentration yielded rings that appeared to be dynamic in one study [38] and more static in another [39], but treadmilling was not documented in either case. As ZipA can bind to both monomers and polymers of FtsZ with comparable affinity [7,40], ZipA likely functions to increase the local concentration of FtsZ at the membrane to facilitate polymerization.

FtsA is a widely conserved actin homolog that interacts with downstream divisome proteins, is implicated in regulating constriction activation, and was shown to treadmill with FtsZ in B. subtilis [1,8]. Purified E. coli FtsZ forms co-assemblies with FtsA and exhibits treadmilling activity on SLBs (Figure 2A), with higher FtsA concentrations favoring short, highly dynamic FtsZ polymers [7]. Interestingly, FtsA stimulates dynamics of membrane-associated FtsZ assemblies without affecting the apparent GTPase rate [13]. FtsA overproduction results in a division defect in all species tested and dispersed Z-rings in E. coli [41] and C. crescentus [24], suggesting that overabundant FtsA destabilizes the Z-ring, an effect similar to the destabilization of SLB-associated filaments observed in vitro. Collectively, the above observations lead to the hypothesis that FtsA not only serves as a membrane anchor, but also destabilizes FtsZ polymers and facilitates treadmilling.

While FtsA has been reported in several studies to form polymers in vitro [42,43], the role of FtsA self-assembly remains poorly understood. FtsA binds and hydrolyzes ATP, which is required for its activation of FtsZ dynamics on SLBs [7]. An FtsA variant deficient in ATP hydrolysis fails to self-assemble in vitro and results in shorter cell length when produced with a GFP tag in E. coli [44], suggesting that monomers of FtsA promote cell division. Indeed, FtsA variants predicted to be deficient in self-interaction are able to suppress deletion of multiple divisome components in E. coli, including ZipA [45,46], and can partially suppress the toxic effects of deleting the CTL in C. crescentus [24]. These mutants have also been observed to facilitate bundling of FtsZ in vitro and suppress a bundling-defective FtsZ variant in vivo [47]. Overall, FtsA monomers appear to allow FtsZ lateral interactions in vitro, while FtsA polymers may destabilize FtsZ filaments and stimulate FtsZ dynamics on membranes.

While genetic evidence suggests distinct roles for monomeric versus polymeric FtsA, the presence of FtsA polymers in cells has been enigmatic. Structures proposed to be FtsA filaments parallel to FtsZ filaments were recently observed using cryo-electron tomography (cryoET) in both vegetatively growing and sporulating B. subtilis [48], suggesting that polymers may indeed be present in vivo in this organism. However, FtsA polymers have not been detected in wild-type cells of C. crescentus or E. coli by cryoET, although FtsZ polymers are readily observed [43,49,50]. The in vivo polymerization state of FtsA, and how it relates to the regulation of FtsZ dynamics and function, therefore remains an important outstanding question.

So what? The function of FtsZ dynamics in division

The study of FtsZ dynamics has yielded a plethora of insights into its function, as well as the function of the divisome as a whole. Dynamic FtsZ filaments are clearly required for division site selection, as spatial regulators of Z-ring assembly exploit the dynamic nature of FtsZ polymers. However, what is the relationship between FtsZ dynamics in the Z-ring and the PG enzymes that drive constriction (Figure 3 and Table 1)? The rate of FtsZ treadmilling is correlated with movement of PG remodeling complexes in E. coli and B. subtilis, but not in S. pneumoniae [8–10] (Figure 3A–C). The rate of treadmilling impacts the rate of PG metabolism and constriction in B. subtilis, suggesting FtsZ-mediated regulation of PG synthase activity [7]. Conversely, in E. coli, FtsZ dynamics primarily serve to distribute PG enzymes evenly about the circumference of the cell to maintain septum geometry [8]. Yang and colleagues recently discovered that in E. coli the PG synthase FtsW exists in two motile populations: an actively synthesizing, slow-moving population independent of FtsZ and an inactive, fast-moving population that is likely associated with FtsZ [51] (Figure 3A). This distribution model is supported by evidence of a diffusion and capture mechanism that uses FtsZ treadmilling to direct movement of the E. coli divisome regulators FtsN and FtsQ on SLBs [52] and PG synthesis enzymes in vivo [51,53].

Figure 3: The relationships between FtsZ treadmilling and PG enzyme complex movement and activity vary across organisms.

A. E. coli bears both fast, inactive (dark green) and slow, active (light green) moving PG enzyme populations, and FtsZ treadmilling rates correlate with the former. B. B. subtilis inactive and active PG enzyme complexes comprise a single population of moving molecules that correlates with and depends on FtsZ treadmilling. C. S. pneumoniae inactive and active PG enzyme complexes comprise a single population of moving molecules that does not correlate with or depend on FtsZ treadmilling. D. S. aureus has a moving population of PG enzyme complexes and treadmilling FtsZ, but the rates of each have yet to be determined.

Table 1:

Impacts of perturbing FtsZ GTPase activity, either due to mutation or treatment with PC190723 following Z-ring assembly

Species	PG Enzyme Recruitment	Constriction	PG enzyme movement rate	Septum morphology
E. coli	Still occurs	Still occurs	Fast, inactive population correlates with FtsZ treadmilling rate; slow, active population does not	Aberrant morphology
B. subtilis	Still occurs	Still occurs if already initiated; rate correlates with FtsZ treadmilling rate	Correlates with FtsZ treadmilling rate	No reported change
S. pneumoniae	Still occurs	Fails for GTPase-deficient FtsZ; ??? for PC addition after start of constriction	Does not correlate with FtsZ treadmilling rate	No reported change
S. aureus	Still occurs	Still occurs if already initiated	???	No reported change

Interestingly, FtsZ treadmilling is only required until constriction has initiated in B. subtilis and S. aureus, as treatment with PC190723 after that point does not prevent continued constriction [12,54]. A requirement for dynamic FtsZ only early in the constriction process is similar to the observation in fission yeast that actin is dispensable for completion of cytokinesis if the septum has progressed sufficiently, with septum assembly being sufficient at later stages [55]. Collectively, these findings suggest that FtsZ dynamics are broadly required for division site selection and Z-ring assembly, and that they contribute to efficiency and directionality of constriction through distribution and, in some organisms, regulation of PG synthesis complexes.

Conclusions and Outlook

Recent investigations into the mechanisms and functions of FtsZ dynamics have honed our understanding of the role of this crucial cell division mediator. Notably, treadmilling motion has been characterized in a number of bacterial species. FtsZ dynamics appear to be predominantly regulated by its intrinsic polymerization properties including GTPase activity and lateral interactions, but the membrane anchor FtsA also likely contributes to treadmilling and possibly other dynamic behaviors in vivo. In contrast, other binding partners may not function to regulate dynamics, but to promote assembly of a localized, condensed ring that can efficiently direct division. While FtsZ dynamics are broadly important for division site selection in diverse bacterial species, their function within an assembled and constricting Z-ring may be more species-specific, regulating protein distribution in some organisms and activity in others.

Of course, there are important questions that have yet to be addressed. Which, if any, dynamic behaviors beyond treadmilling are important for FtsZ’s function? Studies conducted until now have focused on treadmilling out of necessity, as it is relatively easy to observe on a macro scale both in vitro and in vivo. There may be other dynamic modes that complement treadmilling, as evidenced by the observation that Z-rings remain dynamic even after inhibition of treadmilling. Characterizing additional dynamic behaviors and connecting each – including treadmilling – to specific aspects of FtsZ function in cells will require the development of methods to disrupt them individually without affecting GTPase rate, as this drives global FtsZ dynamics. Also perplexing is whether treadmilling occurs through association and disassociation of FtsZ monomers, oligomers, polymers, or all of the above with clusters of FtsZ. Although addition and loss of monomers is intuitive, in vivo treadmilling rates are faster than would be predicted from in vitro GTPase rates [56] if monomers are solely involved. Finally, significant differences have been reported for the impact of FtsZ dynamics on cell wall synthesis in different species. Are these the result of divergent evolution, or is there a conserved mechanism of division that can be modulated depending on growth conditions and other factors? We have a long way to go in terms of understanding how Z-ring structure and dynamics are regulated, and by extension, the effects that regulation has on cell division.

Highlights.

• Z-ring spatial regulators exploit intrinsic properties of FtsZ polymer dynamics

• Treadmilling requires GTP hydrolysis, low levels of Mg²⁺, and membrane association

• FtsA concentration affects FtsZ polymer dynamics and Z-ring architecture

• FtsZ dynamics are crucial for cell wall synthase recruitment and distribution

• In some species, FtsZ treadmilling regulates cell wall synthesis rate