Condensation of FtsZ filaments can drive bacterial cell division

Abstract

Forces are important in biological systems for accomplishing key cell functions, such as motility, organelle transport, and cell division. Currently, known force generation mechanisms typically involve motor proteins. In bacterial cells, no known motor proteins are involved in cell division. Instead, a division ring (Z-ring) consists of mostly FtsZ, FtsA, and ZipA is used to exerting a contractile force. The mechanism of force generation in bacterial cell division is unknown. Using computational modeling, we show that Z-ring formation results from the colocalization of FtsZ and FtsA mediated by the favorable alignment of FtsZ polymers. The model predicts that the Z-ring undergoes a condensation transition from a low-density state to a high-density state and generates a sufficient contractile force to achieve division. FtsZ GTP hydrolysis facilitates monomer turnover during the condensation transition, but does not directly generate forces. In vivo fluorescence measurements show that FtsZ density increases during division, in accord with model results. The mechanism is akin to van der Waals picture of gas-liquid condensation, and shows that organisms can exploit microphase transitions to generate mechanical forces.

Fig. 1.

Lattice model of the Z-ring. (A) A schematic of the Z-ring, which contains FtsA/ZipA interacting with FtsZ in a nonuniform MinC background. A gradient in MinC (orange) allows FtsA/ZipA to colocalized with FtsZ. (B) A lattice representation of FtsZ interaction. Each lattice can be either empty or filled with FtsZ, FtsA, or FtsZ+FtsA. FtsZ can either orient in the cell-axis direction or the hoop direction. Two FtsZ molecules end-to-end have a longitudinal interaction (polymer bond) energy e₁. Two parallel FtsZ side-to-side have a lateral interaction energy e₂. All other adjacent orientations do not interact. (C) The basic force generation mechanism is by increasing the number of lateral contacts between filaments. Increasing the number of lateral contacts, which can occur by adding a monomer or sliding the filament end, lowers energy. The probability of adding a monomer is controlled by the cytoplasmic concentration. The sliding of the filament end performs mechanical work. Our model reveals that the second scenario is most probable and is the force generation mechanism.

Fig. 2.

Model results. (A) Snapshot of a nascent Z-ring with a weak lateral interaction e₂. The filaments are short and disordered, and can orient in any direction. Red and yellow correspond to parallel and anti-parallel to the cell-axis, respectively. Blue and black correspond to clockwise and counterclockwise filament orientation in the hoop direction. (B) Series of configurations during Z-ring contraction. Filaments during this stage become longer and more aligned in the hoop direction. (C) During contraction, the number density of FtsZ in the ring region increases. (D) The free energy of the ring decreases with decreasing ring radius, suggesting that there is a thermodynamic contractile force given by the slope of the free energy curve. (E) The number of FtsZ molecules, however, remains relatively constant during contraction. (F) The number of lateral contacts between FtsZ increases during contraction while the number of polymer bonds remains relatively constant. (G) We compute the contraction force as a function of the lateral interaction energy (symbols), which agrees with the prediction of van der Waals model of Eq. 1 (solid line).

Fig. 3.

We use fluorescence images of dividing bacterial cells to verify model predictions. (A) Images of dividing E. coli with FtsZ::GFP. (B) The ring radius decreases during division. (C) The measured total fluorescence intensity (symbols) of the ring remains constant, suggesting that the total number of FtsZ in the ring is constant during division. (D) The measured fluorescent intensity density (symbols), or the FtsZ density is increasing, reminiscent of Fig. 2C. The solid lines in C and D are computed results from our model. The data suggest that e₂ is between −1.0 and −1.5 k_BT.

Fig. 4.

Overexpression of either FtsZ or FtsA stops division. The model shows that the density of FtsZ in the ring is elevated when there is excess FtsZ or FtsA, further density increase is not possible, thus preventing condensation. (A) The density of FtsZ in the ring is significantly higher. (B) Excess FtsA+ZipA is recruited to the ring when there is overexpression of FtsA/ZipA. (C) The degree of alignment of FtsZ filaments higher when there is overexpression of FtsZ. (D–F) Model snap-shots of the Z-ring, showing normal levels of FtsZ and FtsA (D), overexpression of FtsA (E), and overexpressions of FtsZ (F).