Micron-scale plasma membrane curvature is recognized by the septin cytoskeleton

Abstract

Cells change shape in response to diverse environmental and developmental conditions, creating topologies with micron-scale features. Although individual proteins can sense nanometer-scale membrane curvature, it is unclear if a cell could also use nanometer-scale components to sense micron-scale contours, such as the cytokinetic furrow and base of neuronal branches. Septins are filament-forming proteins that serve as signaling platforms and are frequently associated with areas of the plasma membrane where there is micron-scale curvature, including the cytokinetic furrow and the base of cell protrusions. We report here that fungal and human septins are able to distinguish between different degrees of micron-scale curvature in cells. By preparing supported lipid bilayers on beads of different curvature, we reconstitute and measure the intrinsic septin curvature preference. We conclude that micron-scale curvature recognition is a fundamental property of the septin cytoskeleton that provides the cell with a mechanism to know its local shape.

Figure 1.

Septin abundance scales with positive curvature in A. gossypii. (A) Septin higher-order structures in A. gossypii, visualized by Cdc11a-GFP using structured illumination microscopy (SIM). Straight bundles (1), thin filaments (2,) and branch assemblies (3) exist in the same cell. (B) SIM images of Cdc11a-GFP signal at the base of four lateral branches emanating from hyphae at distinct angles, producing different curvatures. (C) Mean curvature heat map produced by imaging Blankophor in the A. gossypii cell wall followed by curvature analysis. For this display, curvature was mapped onto the external surface and values were inverted to represent curvature as viewed from the cell interior. (D) Cdc11a-GFP intensity at the base of branches plotted against positive and negative principal curvatures and mean curvature as viewed from the cell interior. Diagrams illustrate the curvature measured in each plot. (E) Filament orientation at the base of branches visualized by Cdc11a-GFP SIM. (F) Filament orientation in hyphae, away from sites containing a positive curvature component. (G) Filament orientation relative to the hyphal axis was measured compared with a random simulation of filament orientations (solid lines, mean; dotted line, SD; n = 263 filaments in 13 hyphae). (H) Colocalization of Cdc11-mCherry (green) and Hsl7-GFP (magenta) at the base of branches and straight bundles in A. gossypii.

Figure 2.

Septins recognize micron-scale positive membrane curvature. (A) Supported lipid bilayer (25% PI and 75% PC) and trace Rh-PE–coated silica beads ranging from 0.3 to 6.5 µm in diameter mixed with 50 nM S. cerevisiae septin complex containing Cdc11-SNAP488 for visualization. The Rh-PE is shown in magenta, and Cdc11-SNAP488 is shown in green. (B) Mean intensity images of 10 beads for each condition. (C) Septin adsorption to each bead size at 100-nM septin complex. (D) Septin adsorption to beads as a function of concentration. (E) Heat map of fold difference in septin adsorption to beads as a function of concentration. Data were normalized to the lowest detectable septin adsorption, on 1-µm beads at 10-nM septin complex concentration. 500 nM was the highest experimentally attainable septin complex concentration that could be mixed with beads. In C–E, n ≥ 32 for each size, and error bars represent standard error. Dunn test results: ***, P < 0.005; *, P < 0.05. n.s., not significant.

Figure 3.

Septin affinity for membranes varies depending on curvature. (A) Adsorption of septins to silica beads over time on 1 µm (κ = 2 µm^–1) and 5 µm (κ = 0.4 µm^–1) at 250-nM septin complex concentration. Solid lines represent means, and shaded areas represent SD (average n/time point: 2 µm^–1 = 158 beads and 0.4 µm^–1 = 18 beads). (B) Adsorption of 500-nM septin complexes on 1-µm (κ = 2 µm^–1) and 5-µm (κ = 0.4 µm^–1) beads in 100 mM KCl. (C) Adsorption of 500-nM septin complexes on 1-µm (κ = 2 µm^–1) and 5-µm (κ = 0.4 µm^–1) bilayer-coated beads in 50 mM KCl. In B and C, n ≥ 50 for each size; black bars represent medians.

Figure 4.

Curvature preference remains intact in single complexes, but filament formation is required for stable membrane association. (A) Adsorption of 25 nM 6xHIS WT Cdc11–SNAP488 complexes to 1-µm (κ = 2 µm^–1) and 3-µm (κ = 0.67 µm^–1) beads coated in 2% DGS-Ni²⁺ NTA lipids in 100 mM KCl. For visualization of fold enrichment, data were normalized to 3-µm beads. (B) WT Cdc11–SNAP complexes and Cdc11-α6–SNAP complexes, containing point mutations in the polymerization interface of Cdc11, were diluted to 250 nM in 50 mM KCl to promote filament formation and visualized on a polyethylene glycol–coated coverslip. (C) Adsorption of 100 nM 6xHIS Cdc11-α6–SNAP488 complexes to 1-µm (κ = 2 µm^–1) and 3-µm (κ = 0.67 µm^–1) beads coated in 2% DGS-Ni²⁺ NTA lipids in 100 mM KCl. (D) Adsorption of 25 nM 6xHIS WT Cdc11–SNAP488 complexes to 1-µm (κ = 2 µm^–1) and 3-µm (κ = 0.67 µm^–1) beads coated in 2% DGS-Ni²⁺ NTA lipids in 300 mM KCl. (E) WT and α6 mutant Cdc11-GFP expressed and imaged in live A. gossypii. (F) Membrane adsorption of 100 nM WT and Cdc11-α6–SNAP488 complexes on anionic-supported lipid bilayer (25% PC, 75% PI, and trace RhPE)–coated 1-µm beads. In A, C, D, and F, n ≥ 43 for each size. Dunn test results: ***, P < 0.005.

Figure 5.

Mammalian septins recognize membrane curvature. (A) Phalloidin (magenta) and SEPT7 (green) localization, visualized by α-SEPT7 immunofluorescence, in NIH 3T3 fibroblasts. (B) SEPT7 localization (green) in NIH 3T3 fibroblasts treated with 1 µM latrunculin A for 15 min to disrupt actin-dependent septin localization. The cell outline (magenta), was produced by imaging Alexa Fluor 647–conjugated wheat germ agglutinin. (C) 50 nM human septin complexes (SEPT2–SEPT6–SEPT7) labeled with NHS-Alexa Fluor 488 on supported lipid bilayer–coated silica beads. (D) Adsorption of 50 nM human septin complex on bilayer-coated beads (25% PC, 75% PI, and trace RhPE). n ≥ 67 for each size. Dunn test results: ***, P < 0.005; n.s., >0.05. (E) Model displaying the scale of septin curvature sensing compared with other established proteins that interact with curved membranes.