An amphipathic helix enables septins to sense micrometer-scale membrane curvature

Abstract

Cell shape is well described by membrane curvature. Septins are filament-forming, GTP-binding proteins that assemble on positive, micrometer-scale curvatures. Here, we examine the molecular basis of curvature sensing by septins. We show that differences in affinity and the number of binding sites drive curvature-specific adsorption of septins. Moreover, we find septin assembly onto curved membranes is cooperative and show that geometry influences higher-order arrangement of septin filaments. Although septins must form polymers to stay associated with membranes, septin filaments do not have to span micrometers in length to sense curvature, as we find that single-septin complexes have curvature-dependent association rates. We trace this ability to an amphipathic helix (AH) located on the C-terminus of Cdc12. The AH domain is necessary and sufficient for curvature sensing both in vitro and in vivo. These data show that curvature sensing by septins operates at much smaller length scales than the micrometer curvatures being detected.

Figure 1.

Septins bind cooperatively to curved membranes, with differences in affinity and maximal binding. SLBs (75% DOPC, 25% PI, and trace amounts of Rh-PE) were reconstituted on silica beads. Purified septins were added at several concentrations through saturation. (A) Representative images are maximum intensity projections. Bar, 10 µm. (B) Quantification of septin adsorption at equilibrium onto SLBs of varying curvature. Each data point represents the mean intensity for 98–600 beads. Error bars are SEM, n = 3.

Figure 2.

Septin filament alignment toward the axis of principal curvature is dependent on filament length. SLBs (75% DOPC, 25% PI, and trace amounts of Rh-PE) were reconstituted on borosilicate rods of different diameters ranging from ∼100 nm to 2,300 nm. (A–C) Representative images of septin filament alignment on rods from the three different categories. Bars, 200 nm. A subset of filaments was false colored to depict alignments; pink are parallel to the long axis of curvature; yellow are oriented at ∼45°; blue are aligned to the axis of principal curvature. (D) Schematic of septin filament length in terms of the number of octamers and a cartoon depicting how filament orientation relative rod was measured. (E–G) Box and whisker plot quantifying septin filament alignment on various rod diameters as a function of filament length binned to three diameter ranges. Black bars represent the median. Error bars are standard deviation. (E) 100–400 nm rods. n = 23 rods and 193 filaments. (F) 401–600 nm rods. n = 15 rods and 189 filaments. (G) 601–2,000 nm rods. n = 24 rods and 491 filaments.

Figure 3.

Septin filament orientation is dependent on membrane curvature. SLBs (75% DOPC, 25% PI, and trace Rh-PE) were generated on borosilicate rods of different diameters (100–2,300 nm). Purified septins were added to SLBs at saturating concentrations (from 50 nM to 500 nM) and imaged using scanning EM. (A–C) Representative images of septin filament alignment on rods from the three different categories at 50 nM septin concentration. Bar, 200 nm. (D) Example septin bundling (white arrowheads). Bar, 50 nm. (E) Box and whisker plots of septin filament orientation on measured rod diameters at several septin concentrations (50 nM, 100 nM, 250 nM, and 500 nM). The number of filament orientations measured per rod ranged >10 filaments. Quantification is from 49 rods. Measured filaments were >100 nm (∼3 octamers) in length. Black bars represent the median. Error bars are standard deviations.

Figure 4.

Single septin complexes have a higher association rate for optimal membrane curvatures. (A) SLBs (75% DOPC and 25% PI) were reconstituted on curvatures of 2 µm⁻¹ or 0.67 µm⁻¹ and flowed between two PEG (black)-coated coverslips (top coverslip not shown). Nonpolymerizable septin complexes were then flowed into the chamber, and septin binding and unbinding events were observed using near-total internal reflection fluorescence microscopy. (B) Representative images of a binding event on κ= 2 µm⁻¹ (1 µm bead). Line scan through the particle shows signal intensity fit to a Gaussian function. (C and D) Dwell time histograms for membrane curvature of 2 µm⁻¹ and 0.67 µm⁻¹, respectively. (E) Box and whisker plot quantifying association rate of nonpolymerizable septin complexes onto both curvatures (κ =2 µm⁻¹ and 0.67 µm⁻¹). Black bars represent the median. Error bars are standard deviations. N_beads = 10 (for κ =2 µm⁻¹ and 0.67 µm⁻¹); N_{binding events} = 844 and 315 for for κ = 2 µm⁻¹ and 0.67 µm⁻¹, respectively.

Figure 5.

Septins have conserved AHs at their C-termini. (A) Domains within the yeast septin, Cdc12. PBD (blue): polybasic domain; GTP-BD (green): GTP-binding domain; SUE (purple): Septin unique element; CTE (pink): C-terminal extension; CC (magenta): coiled-coil domain (red); AH domain (cyan). The AH domain is shown as a helical wheel, corresponding primary sequences for WT Cdc12 and Cdc12-6 (below). (B) AHs found in the C-terminal extensions of septins in multiple species represented through helical wheel diagrams. (C) 1 nM purified WT septin complexes and cdc12-6 septin complexes adsorption (green) onto SLBs (magenta) with different curvatures. Bar, 5 µm. (D) Box and whisker plot quantifying adsorption of WT or cdc12-6 septin complexes onto SLBs of various curvature. Black bars represent the median. Error bars represent the standard deviation. n > 50 beads for each curvature. (E) Representative images of 1 µM 2×-Cdc12 AH-GFP binding onto SLBs on beads with curvatures. Arrows highlight filament-like structures. (F) Box and whisker plot of 2×-Cdc12 AH-GFP adsorption onto different membrane compositions and curvatures. Black bars represent the median. Error bars at each curvature are standard deviations for n > 100 beads. (G) Septins assemble at the base hyphal branches in WT cells (Cdc11a-GFP), but rarely assemble in cdc12-6 mutants (cdc12-6-GFP) at permissive temperature. Bar, 5 µm.