Oscillation helps to get division right

A conserved feature of proliferating cells is their need to divide at the appropriate time and place using precise positioning mechanisms to ensure that each daughter cell has the same genetic complement as the parent, as well as the correct shape and size. In rod-shaped bacteria, division typically occurs precisely at midcell after the bacterial chromosome has replicated and segregated away from the cell center (1–3) (Fig. 1, Top), thereby generating daughter cells with the same chromosome content and shape as their parent. Early genetic studies in Escherichia coli identified three Min (Minicell) genes whose products are important for defining the normal pattern of cell division. Mutations in these genes lead to inappropriate divisions close to the cell poles, generating chromosome-less minicells. Because Min⁻ mutants still undergo cell divisions at midcell in addition to polar divisions (1, 3), precise chromosome positioning, and perhaps other processes, must act additionally to define midcell (3, 4). Subsequent genetic, biochemical, and cell biological studies have revealed features of how the three Min proteins function (3, 5–9). ATP binding to MinD promotes its dimerization and inner membrane binding, whereas its partner protein, dimeric MinE, interacts with MinD to stimulate its ATPase and consequent release from the membrane. The ensuing dynamic pattern of MinD ATPase imparts positional information to the cell (discussed below). A third protein, MinC, which binds and travels as a passenger with MinD, is a division inhibitor but is not required for dynamic patterning. In PNAS, Vecchiarelli et al. (10) extend their analysis of MinCD behavior on supported lipid bilayers in vitro to provide much needed new mechanistic insight into the dynamic patterning mechanism, which helps direct the spatial positioning of division. The work reveals the nonlinear protein interactions that drive the observed Min oscillatory behavior and demonstrates roles of MinD and MinE in patterning additional to those identified in earlier studies.

Fig. 1.

(Top) Schematic of the oscillatory behavior of MinD (blue, and its passenger, MinC, which exhibits identical oscillatory behavior; not shown) as E. coli cells grow and divide. MinE also oscillates, forming a ring at the leading edge of MinD (for simplicity, not shown). Cells can divide when newly replicated nucleoids have segregated away from the cell center and the divisome has assembled at midcell, where the time-averaged MinC concentration is at a minimum. The bacterial nucleoids are shown in gray. (Bottom) Schematic of the experimental flowcell setup in ref. 10, with the gradient of MinD cartooned. Examples of spirals (solution MinD concentration relatively high; MinD in blue and MinE in red) and bursts (solution MinD concentration low) are shown. For further details, including movies of spiral and burst dynamics, see ref. 10.

Early live-cell imaging studies showed that the Min proteins oscillate from one cell pole to the other, with a period of <1 min, in reactions requiring MinD and MinE (Fig. 1, Top) (3, 5–7). The oscillations result from the perpetual chase and release of MinD (and its associated MinC) from the inner membrane by MinE, thereby forming a standing wave at the cell center, leading to a time-averaged minimum concentration of MinC at midcell, thereby allowing midcell divisome assembly (3). Even though the basic biochemical features of MinDE have been known for some time, it would be a mistake to believe that these past studies of Min oscillatory behavior provide a comprehensive molecular understanding of the process. One of the many strong points of the paper by Vecchiarelli et al. (10) is that the new biochemical insight that emerges could not have been gleaned from classic ensemble biochemistry, or from models based on simulations of reaction-diffusion patterning mechanisms. Importantly, the paper discusses the fact that mechanistically diverse models, using different biochemical assumptions, can capture the same self-organizing oscillatory behavior using reaction-diffusion patterning mechanisms in which there is a single nonlinear protein interaction term (11–14). This underlines the fact that modeling alone cannot inform biochemical mechanism and emphasizes the desirability of careful experimental studies and robust data if modeling is to capture biochemical detail. Whenever feasible, incisive experiments should be used to interrogate models rigorously, and decisive “killer” experiments based on predictions from the models should be undertaken.

Previous experimental work in vitro established that MinDE can form ATP-dependent dynamic patterns on lipid bilayers, but in the earlier experiments the formation of standing waves was not recapitulated, most likely because the ratio of membrane surface area to free volume in these in vitro systems was much lower than in vivo, and therefore the concentrations of the key molecules in vitro needed to be much higher than those in vivo for dynamic patterning to be observed (14–17). Nevertheless, recent work using cell-shaped membrane-bound compartments of picoliter reaction volumes and large lipid surface areas were able to reconstitute the standing wave dynamics (18), thereby underlining the importance of capturing as many in vivo characteristics as possible when developing in vitro reaction assays.

Vecchiarelli et al. (10) preincubated ATP with fluorescently labeled derivatives of MinD and MinE at close to physiological concentrations. Introduction of this mixture into a flowcell with a supported lipid bilayer and a ∼25-μm-deep aqueous channel along its length generated a gradient of MinD that decreased from inlet to outlet as limiting MinD bound the lipid bilayer (Fig. 1, Bottom). Along the MinD gradient in the flowcell spatially reproducible patterns of dynamic behavior were observed, with “amoebae” close to the inlet, “spirals” in the middle of the flowcell, and, importantly, “bursts” occurring toward the outlet where MinD is most depleted from solution. These bursts of radially expanding and contracting MinD zones, with a peripheral ring of MinE, formed standing waves. The authors argue convincingly that it is these bursts that best recapitulate in vivo Min standing wave patterning. Furthermore, MinD depletion, which is responsible for the bursts in the in vitro assays here, and in vivo because the inner membrane (∼10 μm²) can bind many more MinD dimers than exist in the cell (∼2,000; ∼3 μM), has a complementary role in shaping in vivo oscillatory behavior.

The authors then go on to explore the role of the MinE membrane-binding domain, which is required for oscillations in vivo but whose functional importance has been debated. For example, one family of models has the regulator MinE role being simply a binding partner for MinD ATP, with binding stimulating ATP hydrolysis and MinD release from the membrane (12). MinE alone interacts with membrane weakly to uncover its MinD interacting surface (7). Consequent stimulation of MinD ATP hydrolysis by MinE leads to MinD ADP release from the membrane, whereas “lingering” membrane-bound MinE forms the observed MinE ring at the periphery of the MinD polar membrane zone. The work proposes that this lingering MinE directs the periodicity of the standing waves and may be important in ensuring that subsequent membrane binding by

Vecchiarelli et al. extend their analysis of MinCD behavior on supported lipid bilayers in vitro to provide much needed new mechanistic insight into the dynamic patterning mechanism.

MinD is at the distal pole. Taken together, the results lead to the hypothesis that the patterning mechanism is driven by switches from a state in which MinE recruits MinD to the membrane when cytoplasmic MinD is high to a state in which MinE stimulates MinD ATP hydrolysis and release from the membrane when MinE is in excess. Consequently, MinE is both a “promoter” of MinD binding as well as a “dissociator” of MinD from the membrane, with the proposed “toggle switch” controlled by local relative concentrations of MinD and MinE on the membrane; coupling of this switch with MinD depletion from the cytoplasm is proposed to generate the self-organized standing wave oscillator. This new mechanistic insight should provide the platform for future robust modeling.

Whereas Min directs the spatial positioning of division, chromosome segregation plays an important role in determining timing of division. Delays in completion of DNA replication or segregation of newly replicated chromosomes prevent normally timed division at midcell, because the presence of unsegregated nucleoids at midcell prevents placement of the divisome there, a process that is enhanced by nucleoid occlusion proteins (3). In my opinion, precise chromosome positioning after segregation of newly replicated daughter chromosomes has an important part in defining midcell in bacteria; indeed, this may be a general feature of many, if not all, cell types. Intriguingly, in eukaryotes, not only may positioning of chromosomes and their associated segregation apparatus play an important role in directing cytokinesis position in both symmetric and asymmetric cell divisions (19), but pole-to-pole oscillations of dynein direct this positioning (20).

Finally, MinDE are structurally related to the ParAB proteins that act to position genetic loci and proteinaceous machines using the bacterial nucleoid as a matrix rather than the inner membrane, as is the case of MinDE (21). This leads to efficient segregation of these loci/proteinaceous machines with daughter nucleoids at cell division. Although, in outline, these systems seem to function like the Min system, with ParA being the matrix-binding ATPase and ParB stimulating ATP hydrolysis by ParA, the details of how they precisely work will only emerge from the type of painstaking quantitative in vitro assays described here, alongside more quantitative high-resolution in vivo imaging studies.