Figures and data
Modulation and measurement of membrane tension in vivo. All error bars are SD.
A. B. Schematic of the strategy used to generate B. subtilis cells with membrane excess (reduced membrane tension) by titrating the overexpression of accDA to increase phospholipid synthesis.
B and C. B) De novo total lipid synthesis was quantitated by measuring the incorporation of [14C] acetate into bSW305 (amyE::tet-pXyl-accDA) cells. At a low induction (1 mM xylose), cells produced approximately 4-fold more lipids with no change in C) cell morphology as observed by SIM microscopy of cell membranes stained with Nile Red. P-values are in Table S1. Scale bar in C is 1 µm.
D. Fluorescence lifetime microscopy images of cells stained with Flipper-TR dye. A color code per pixel reflects the results of the double exponential fit to the fluorescent lifetimes: red pixels indicate high membrane tension and blue low membrane tension. Scale bar is 1 µm.
E. Plot of FlipperTR lifetimes at different accDA inductions in bSW305. Photons from different single cells were analyzed. Each data point corresponds to the fitted value gained from all the cells within one field of view. P-values are in Table S1.
F. Change in the generalized polarization of Laurdan dye of cells with different inductions of accDA. Inset - Change in lipid packing over time when cells were subjected to hyperosmotic shocks.
FtsZ condensation by FtsZ bundling proteins serves to overcome membrane tension to initiate division.
All error bars are SD.
A. Schematic of experimental rationale. Top - FtsZ inside GUVs are known to deform the membrane when the GUV is beneath a given membrane tension. Bottom - Similarly, if the condensation of FtsZ filaments by FtsZ bundling proteins helps to overcome the cellular membrane tension to bend the membrane inward, cells lacking FtsZ bundling proteins should only be able to initiate division at lower membrane tensions. We made 2 B. subtilis strains (bDR110 and bDR112) where we could titrate expression of 1) The FtsZ bundling protein EzrA with xylose (regulating filament condensation) and 2) accDA with IPTG (regulating the amount of membrane excess) in ΔsepF backgrounds. bDR112 also contains mNeonGreen-FtsA(sw) expressed at the native locus.
B. Long filamented cells (> 10 μm) lacking invaginations formed when bDR112 cells (lacking sepF) were depleted of EzrA at low accDA induction (5 μM IPTG), and FtsZ only formed decondensed Z rings. Nile Red membrane stain is orange, mNeonGreen-FtsZ is green. EzrA was depleted by growth in 0 mM xylose for 4 hours.
C. Smaller cells (∼4 μm) capable of dividing formed when accDA was induced at higher levels (25 μM IPTG) in bDR112 cells depleted of EzrA. Scale bar is 1 µm.
D. Cell length distributions as a function of accDA induction in EzrA depleted bDR112 cells. At low accDA inductions, the distributions are dominated by long cells (> 10 μm), as in Fig. 2B. At inductions of 15 μM IPTG and higher, cell division is evidenced by the increasing number of smaller cells (emerging peaks around 3-6 μm). Inset - The corresponding decreasing fraction of long cells as accDA induction is increased.
E. A phase diagram of cell division plotted as a function of FtsZ filament condensing activity (ezrA induction) vs. membrane excess (accDA induction). Average cell length for each induction pair is represented by spot color. For these ΔsepF cells to divide, they require i) higher accDA inductions when EzrA is depleted or ii) higher ezrA expression (0.25 mM xylose) at low accDA inductions (5 µM IPTG). At the threshold points of cell division along each axis (0.25 mM xylose with 5 μM IPTG, and 15 μM IPTG with 0 mM xylose), cells appear as a mix of small cells and long filaments. Cells were unable to divide when ezrA was induced with 0.25 mM xylose and accDA with 5 µM IPTG, but they regained the ability to divide with near wild-type lengths when accDA induction was increased to 10 µM IPTG, demonstrating filament condensation caused by FtsZ bundling proteins serves to overcome membrane tension to deform the membrane inward.
F. Membrane tension in EzrA depleted bDR110 cells as a function of accDA induction. Cells induced with 5 μM IPTG did not divide compared to cells induced at 15µM IPTG and above. To correct for the difference in surface area between filaments and dividing cells (“adjusted membrane tension”), we used cells induced with 5 μM IPTG (accDA) and 1mM xylose (ezrA) as a proxy for the 5 μM IPTG point, as these cells can divide (see main text). Intermediate values (10, 20, 30 μM IPTG) were linearly interpolated. Inset - Diagram of the difference in membrane surface area in dividing and filamented cells. P-values are in Table S1.
G. Fluorescence lifetime microscopy images of EzrA depleted bDR110 cells stained with Flipper-TR membrane tension dye at different accDA inductions. Scale bar is 1 µm.
Increasing membrane excess leads to faster Z ring constriction.
All error bars are SD.
A. Time-averaged images of FtsZ rings in ezrA depleted bDR112 cells at various accDA inductions (IPTG) relative to cells with high ezrA induction (1mM xylose). Dotted red line shows an example measurement of ring width. Scale bar is 1 µM.
B. Plot of Z ring constriction rates vs. ring thickness at different accDA inductions in EzrA depleted cells. Inset - Thickness distributions of Z rings in bDR112 cells at different accDA inductions. Z rings transition from a wide state (∼0.9 μm) at 10 μM IPTG to a compressed state of ∼0.4 μm at 15 μM IPTG.
C. Kymographs of Z rings showing faster constriction rates at increasing amounts of IPTG controlled accDA induction in EzrA depleted bDR112 cells. As before, cell division was only observed at 15 μM IPTG and higher.
D. FtsZ treadmilling speed vs. Z ring constriction rate at different accDA inductions in ZBP deficient cells. While the constriction rate (blue squares) increases as membrane tension is reduced in ZBP deficient bDR112 cells, the speed of FtsZ filament treadmilling remains constant. This contrasts with the previously observed relationship (green line) in B. subtilis cells with wild-type membrane tension, where the treadmilling speed limits the constriction rate (Bisson-Filho et al., 2017). Error bars are SD.
E. FtsZ treadmilling speed vs. Z ring constriction rates at different accDA inductions in cells with native ZBP expression. As in Fig. 3D, constriction rates also increase with accDA overexpression in these otherwise wild-type cells. As above, the speed of FtsZ filament treadmilling remains constant across accDA inductions. P-values are in Table S1.
F. Ring constriction rates plotted against scaled membrane tension. Constriction rates scale linearly with membrane tension in ZBP-deficient cells at accDA inductions above 15 µM.
Estimates and Models
A. The excess of membrane 
B. By using the constriction rates as proxy for lipid flux velocity v, the scaled membrane tension from Flipper TR lifetimes, and the equation 
C. Tomograms of FtsZ filaments at division septa. Used with permission from (Khanna et al., 2021) Left - Slice through a tomogram of a dividing B. subtilis cell. Membrane distal filaments (FtsZ with a Q-rich linker) are blue, membrane-proximal filaments (FtsA) are pink. Blue arrows indicate doublets, and an orange arrow indicates a possible triplet. Right - Reconstructed cell membrane, with membrane-proximal and membrane-distal filaments corresponding to the figure on the left. Blue arrows point to doublets of membrane-distal filaments. Similar arrangements were observed in cells containing native FtsZ
D. Model for how FtsZ filaments induce opposing curvatures at the division septa. After FtsA/FtsZ filament crowding initiates division, the membrane (grey) around the septal invagination forms a saddle shape composed of two opposing principle curvatures (blue and red lines). FtsA/FtsZ filaments and their condensation may play a role in establishing this geometry: FtsA/FtsZ filaments (green) have an intrinsic inward curvature, which may induce (and stabilize) the deformation parallel to the rod width (red axis). As expected for a crowding induced deformation, FtsA/FtsZ filaments are spaced around the septal invagination (blue axis), an arrangement that induces and stabilizes the membrane deformation along the cell length.
E. Model for how the membrane is deformed to initiate cell division. The membrane (dotted line) has a tension that resists deformation. FtsA/FtsZ filaments (green circles) are bound to (and treadmill along) the membrane, but individually cannot deform the membrane. FtsZ bundling proteins (blue circles) cause FtsA/FtsZ filaments to laterally associate. This local filament crowding works to overcome the membrane’s tension to deform it inward, giving the space needed for cell wall synthetic enzymes to begin building the division septa.
F. Model for inward progression of the division septa. Membrane fluctuations occur all over the membrane surface as well as at division septa. These fluctuations give the cell wall synthesis enzymes (red hexagons) that are associated with FtsA/FtsZ filaments room to insert new cell wall material at the base of the septa. Each insertion event rectifies the inward membrane fluctuation, thereby causing the division septa to ratchet inward. Membrane flux of into the invagination may also increase membrane fluctuations at the tip of the invagination.
A. bSW305 growth in CH media as a function of different accDA expression levels (xylose) assayed by OD600 readings of liquid culture.
B. 2D-Structured Illumination Microscopy (2D-SIM) images of bSW305 cells stained with Nile red with accDA induced with 2 mM xylose. Scale bar is 1 µm.
C. Cell length distribution for different xylose inductions of accDA in bSW305. Scale bar is 1 µm.
D. 2D-SIM images of bSW305 cells stained with Nile red with accDA induced with 8 mM xylose. Scale bar is 1 µm.
E. Fluorescence lifetime microscopy images of GUVs made with different lipid compositions stained with Flipper-TR dye.
F. Plot of FlipperTR lifetimes for different GUVs with different lipid compositions shown in E. Each data point corresponds to the fitted value gained from all GUVs within one field of view. P-values are in Table S1.
G. Phospholipid polar head group distribution for bSW305 with 8 mM xylose accDA levels compared to PY79 wildtype wherein PG represent phosphatidylglycerol, PE:Phosphatidylethanolamine, CL: cardiolipin and L1&L2 unidentified polar elements in the TLC analysis.
H. Fatty acid distribution of PY79 (wild type cells) and bSW305 with 0 mM and 8 mM xylose. AccDA levels were measured by mass spectrometry.
A. Transmission Electron microscopy images of bDR026 with accDA induced with different amounts of IPTG. Scale bar = 100 nm.
B. Widefield fluorescence images of bDR112 cells stained with Nile red for different IPTG inductions of accDA in EzrA depleted ΔsepF cells. Scale bar = 1 µm.
C. Plot of FlipperTR lifetimes at different for bDR110 at different IPTG inductions of accDA at high EzrA (1 mM xylose) and after EzrA was depleted for 4 hours. Photons from different single cells were analyzed. Each data point corresponds to the fitted value gained from all the cells within one field of view.
A. Top - EM images of septa from EzrA depleted bDR110 cells induced with 30 μM IPTG. Bottom - EM images of septa from wild-type cells.
B. Septum thickness for bDR110 for different accDA inductions with 0mM xylose compared to wild type (PY79). Error bars are SD. P-values are in Table S1.
C. Immunoblots of FtsZ in strain bDR110 with accDA induced at 0 µM, 15 µM, 25 µM, and 30 µM IPTG. Quantification of relative FtsZ level to strain bDR110 induced with 0 µM IPTG. Bar represents mean ± SD, P> 0.997, ns, ordinary one-way ANOVA.
A. FtsZ-YFP-mts form rings with high curvature inside GUVs with low membrane tension. When membrane tension is slowly increased by ∼0.03 ns, rings turn into longer bundles with a smaller curvature. Scale bar is 1 µm.
B. De novo total lipid synthesis was quantitated by measuring the incorporation of [14C] acetate into bDR026 with accDA induced at 0 μM and 15 μM IPTG.
Statistics
Strains Used in this Study
Primers Used in this Study
