A single-particle analysis method for detecting membrane remodelling and curvature sensing

Abstract

In biology, shape and function are related. Therefore, it is important to understand how membrane shape is generated, stabilised and sensed by proteins and how this relates to organelle function. Here, we present an assay that can detect curvature preference and membrane remodelling with free-floating liposomes using protein concentrations in physiologically relevant ranges. The assay reproduced known curvature preferences of BAR domains and allowed the discovery of high-curvature preference for the PH domain of AKT and the FYVE domain of HRS (also known as HGS). In addition, our method reproduced the membrane vesiculation activity of the ENTH domain of epsin-1 (EPN1) and showed similar activity for the ANTH domains of PiCALM and Hip1R. Finally, we found that the curvature sensitivity of the N-BAR domain of endophilin inversely correlates to membrane charge and that deletion of its N-terminal amphipathic helix increased its curvature specificity. Thus, our method is a generally applicable qualitative method for assessing membrane curvature sensing and remodelling by proteins.

Keywords: Endophilin; Membrane curvature preference; Membrane remodelling; Single-particle analysis.

Fig. 1.

Working principle and validation. (A) Scheme of how liposomes are detected using diffraction or fluorescence modes. Upon exiting the glass optical flat, the laser beam is refracted through the liquid sample and diffracted by particles in the sample. Introduction of an emission filter that blocks diffracted light allows detection of fluorescent particles only. (B) Scheme of setup. Liposome samples (blue) are introduced in the sample chamber, where they freely move due to Brownian motion. Each particle is tracked and its size is calculated according to Stokes–Einstein equation. Size distribution is given as the concentration of particles in each 5 nm bin. (C,D) Sizing of a fluorescent subpopulation of liposomes is unaffected by the presence of non-fluorescent liposomes. Small (C) or large (D) fluorescent liposomes were sized in the absence (red) or presence (blue) of a population of non-fluorescent liposomes of a different size (black). The black trace shows the size distribution of the non-fluorescent liposomes population detected by diffraction (no emission filter). Graphs are displayed as mean±s.d. n=3.

Fig. 2.

Method principle and validation using BAR domains. (A) Scheme for the curvature sensitivity assay. The fluorescent protein (purple) is incubated with non-fluorescent liposomes (black circles) and measured in our setup. Comparison between the size distributions of all liposomes (black curve) with the fluorescent signal of the protein-bound subpopulation (purple) indicates whether there is a curvature preference for small liposomes or large liposomes, or an absence of curvature preference. The data can then be displayed using boxplots as shown in B – the pattern-filled box represents the fluorescent, protein-bound subpopulation. (B) Box plot display of results. Box plots show the mode of the size distribution as the middle line, whereas boundaries represent 50% of the data on each side of the mode. (C–E) Measurements for (C) amphiphysin N-BAR (PDB: 1URU), (D) FCHo2 F-BAR (PDB: 2V0O) and (E) IRSP53 I-BAR (PDB: 1Y2O). (C) Amphiphysin N-BAR shows specificity for high curvatures. (D) FCHo2 F-BAR is insensitive to curvature and binds all sizes of liposomes equally. (E) IRSp53 I-BAR preferentially binds to the largest liposomes and does not bind to smaller liposomes extruded at 200 or 50 nm. Box plots depict mode values (middle line)±50% data on each side of the mode (indicated by bottom and top lines). n=3. ns, not significant; **P<0.01; ****P<0.0001. P-values were tested using one-way ANOVA followed by Bonferroni correction for multiple comparisons.

Fig. 3.

Curvature preference for diverse lipid-binding domains. (A) The FYVE domain of HRS (PDB: 1DVP) displays preference for higher curvatures (smaller liposomes). The HRS FYVE domain preferentially binds to the subpopulation of small liposomes on the 800 or 200 nm extruded samples and all sizes on the smaller, 50 nm extruded sample. (B) The PH domain of AKT (PDB: 1UNQ) displays preference for higher curvatures (smaller liposomes). The AKT PH domain preferentially binds to the subpopulation of small liposomes on the 800 or 200 nm extruded samples and all sizes on the smaller, 50 nm extruded sample. (C) The C1B domain of PKCβ2 (PDB: 3PFQ) displays no curvature specificity and binds all sizes of liposomes in 800 and 200 nm extruded samples. (D) The FERM domain of talin-1 (PDB: 3IVF) displays no curvature specificity and binds all sizes of liposomes indifferently in all samples. Box plots depict mode values (middle line) ±50% data on each side of the mode (indicated by bottom and top lines). n=3. ns, not significant; *P<0.05; ***P<0.001; ****P<0.0001. P-values were tested using one-way ANOVA followed by Bonferroni correction for multiple comparisons.

Fig. 4.

Monitoring protein-induced vesiculation. (A) Scheme for the assay to detect membrane vesiculation. Incubation of liposomes with ENTH (epsin N-terminal homology domain; PDB: 1H0A) induces formation of smaller liposomes by vesiculation. This is observed by a reduction in mean liposome size and an increase in the concentration of particles. (B,C) Vesiculation by ENTH wild-type (wt, B) or the hyperactive L6W mutant (C) after incubation of liposomes with different ENTH concentrations. (D) Corresponding dose-response curves for vesiculation by ENTH wt or L6W. The concentration of particles in the bin centred around 82.5 nm (marked as a dotted line in B,C) was used as a marker of vesiculation efficiency. (E) Comparison of vesiculation efficiency between ENTH, PiCALM ANTH and Hip1R ANTH domains. PiCALM ANTH and Hip1R ANTH both vesiculate with PiCALM producing larger vesicles than ENTH and Hip1R producing fewer vesicles. Graphs in B–E have n=3 and are displayed as mean±s.d. (F) Negative-stain electron micrograph of liposomes before (black frame) and after incubation with 4 μM ENTH (blue frame), PiCALM ANTH (red frame) or Hip1R ANTH (green frame). Images representive of three independent experiments. Scale bars: 200 nm.

Fig. 5.

Mechanism of curvature sensing by endophilin. (A) Influence of charge on membrane curvature sensing. Increasing the concentration of Folch lipids containing charged phospholipid headgroups in a phosphatidylcholine background reduces the stringency of curvature sensing by Endophilin N-BAR. (B) Structure of N-BAR and BAR ΔH0 constructs of endophilin used (PDB: 2C08). The H0 was added manually for illustration purposes as the H0 is not visible in the available endophilin crystal structures. (C) Curvature sensitivity of endophilin N-BAR and BAR ΔH0 to highly charged liposomes (1:1 FolchS/FolchA). Endophilin N-BAR (blue) binds all sizes of 200 nm extruded liposomes (black), whereas endophilin BAR ΔH0 (red) specifically binds a subpopulation of smaller liposomes. The graph is displayed as mean±s.d. n=3. (D) Model of endophilin N-BAR and BAR ΔH0 binding to small and large liposomes. Endophilin BAR ΔH0 is specific for small liposomes and does not bind to large liposomes. Endophilin N-BAR binds both small and large liposomes, in this case, by inducing high local curvature to which it can bind through the BAR domain. (E) Negative-stain electron micrograph of liposomes before (black frame) and after addition of endophilin BAR ΔH0 (red frame) or endophilin N-BAR (blue frame). Endophilin N-BAR induces formation of areas with high local curvature on large liposomes indicated by arrows. Scale bars: 100 nm. (F) Quantification of the fraction of liposomes showing protrusions in each micrograph for liposomes alone (n=9) or in the presence of endophilin BAR ΔH0 (n=8) or endophilin N-BAR (n=19). Bars show the mean±s.d. Box plots in A and C depict mode values (middle line) ±50% data on each side of the mode (indicated by bottom and top lines). P-values were tested using one-way ANOVA followed by Bonferroni correction for multiple comparisons. ns, not significant; **P<0.01; ****P<0.0001.