Cell-free reconstitution has been a key tool to understand the cytoskeleton assembly, and work in the last decade has established approaches to study septin dynamics in minimal systems. Presented here are three complementary methods to observe septin assembly in different membrane contexts: planar bilayers, spherical supports, and rod supports.
Most cells can sense and change their shape to carry out fundamental cell processes. In many eukaryotes, the septin cytoskeleton is an integral component in coordinating shape changes like cytokinesis, polarized growth, and migration. Septins are filament-forming proteins that assemble to form diverse higher-order structures and, in many cases, are found in different areas of the plasma membrane, most notably in regions of micron-scale positive curvature. Monitoring the process of septin assembly in vivo is hindered by the limitations of light microscopy in cells, as well as the complexity of interactions with both membranes and cytoskeletal elements, making it difficult to quantify septin dynamics in living systems. Fortunately, there has been substantial progress in the past decade in reconstituting the septin cytoskeleton in a cell-free system to dissect the mechanisms controlling septin assembly at high spatial and temporal resolutions. The core steps of septin assembly include septin heterooligomer association and dissociation with the membrane, polymerization into filaments, and the formation of higher-order structures through interactions between filaments. Here, we present three methods to observe septin assembly in different contexts: planar bilayers, spherical supports, and rod supports. These methods can be used to determine the biophysical parameters of septins at different stages of assembly: as single octamers binding the membrane, as filaments, and as assemblies of filaments. We use these parameters paired with measurements of curvature sampling and preferential adsorption to understand how curvature sensing operates at a variety of length and time scales.
The shapes of cells and many of their internal compartments are dependent on the lipid membranes that surround them. Membranes are viscoelastic structures that can be deformed through interactions with proteins, lipid sorting, and acting internal and external forces to generate a variety of shapes1,2,3,4. These shapes are often described in terms of membrane curvature. Cells use a diverse suite of proteins capable of preferentially assembling onto, or "sensing", particular membrane curvatures to ensure defined spatio-temporal control over processes including cell trafficking, cytokinesis, and migration5,6. The dynamics of cell machinery at the membrane are notably difficult to observe due to the difficulty of balancing time and spatial resolution with cell health. While super-resolution techniques can offer a detailed view of such structures, they require lengthy acquisitions that are not amenable to the timescales of assembly/disassembly for most machinery. Additionally, the molecular complexity of these assemblies in their native environment and the multitude of roles a single component can play make minimal reconstitution systems a valuable tool for studying the functional capacity of molecules.
Minimal membrane mimetics have been developed to study membrane properties and protein-membrane interactions outside of the cell. Membrane mimetics vary from free-standing lipid bilayers, such as liposomes or giant unilamellar vesicles, to supported lipid bilayers (SLBs)7,8,9,10. SLBs are biomimetic membranes anchored to underlying support, typically composed of glass, mica, or silica11,12. A variety of geometries can be used, including planar surfaces, spheres, rods, and even undulating or micropatterned substrates to probe protein-membrane interactions on both concave and convex curvatures simultaneously13,14,15,16,17,18. Bilayer formation begins with vesicle adsorption onto a hydrophilic surface, followed by fusion and rupture to form a continuous bilayer (Figure 1)19. Supported bilayers are particularly amenable to light and electron microscopy, providing both better time and spatial resolution than is often achievable in cells. Curved SLBs especially provide an attractive means to probe protein curvature sensitivity in the absence of significant membrane deformation, allowing one to distinguish between curvature sensing and curvature induction, which are often impossible to separate in free-standing systems.
Septins are a class of filament-forming cytoskeletal proteins well known for their ability to assemble on positively curved membranes6,18,20. Over the course of the cell cycle in yeast, septins assemble into a ring and must rearrange to form the hourglass and double ring structures associated with bud emergence and cytokinesis, respectively21. While beautiful work has been done using platinum replica electron microscopy to observe septin architecture at varying cell cycle stages22, watching septin assembly over time using light microscopy in yeast has met with limited spatial resolution. Previous work on septins using lipid monolayers visualized by transmission electron microscopy (TEM) was able to reconstitute several interesting septin structures such as rings, bundles, and gauzes23. However, EM techniques are likewise limited in their temporal resolution, unlike fluorescence microscopy. In order to better resolve the kinetic parameters of the multi-scale process of septin assembly, we turned to supported membrane mimetics, where one can carefully control membrane geometry, sample conditions, and imaging modality.
The protocols described here use planar or curved SLBs, purified protein, and a combination of microscopy techniques. Quantitative fluorescence confocal microscopy and total internal reflection fluorescence microscopy (TIRFM) were used to measure both bulk protein binding onto various membrane curvatures, as well as to measure the binding kinetics of single molecules. Furthermore, this protocol has been adapted to be used with scanning electron microscopy (SEM) to examine protein ultrastructure on different membrane curvatures. While the focus of these protocols is on the septin cytoskeleton, the protocols can be easily modified to investigate the curvature sensitivity of any protein the reader finds interesting. Additionally, those working in fields such as endocytosis or vesicular trafficking may find these techniques useful for probing the curvature-dependent assemblies of multi-protein complexes.
NOTE: Forming supported lipid bilayers requires the preparation of monodispersed small unilamellar vesicles (SUVs). Please refer to a previously published protocol24 on SUV formation. Briefly, all SUVs are formed by probe sonication for 12 min in total at 70% amplitude via 4 min sonication periods followed by 2 min rest periods in ice-water. SUV solutions must be well clarified and monodispersed in size. Size distributions of SUVs can be measured, for example, by dynamic light scattering25.
1. Planar lipid bilayers
Supported Lipid Bilayer Buffer (SLBB) | ||
Stock | Volume | Final concentration |
2 M KCl | 1.5 mL | 300 mM |
1 M HEPES | 200 µL | 20 mM |
500 mM MgCl2 | 20 µL | 1 mM |
Water | 8 mL | |
Pre-Reaction Buffer (PRB) | ||
Stock | Volume | Final concentration |
2 M KCl | 166 µL | 33.3 mM |
1 M HEPES | 500 µL | 50 mM |
Water | 9.33 mL | |
Reaction Buffer | ||
Stock | Volume | Final concentration |
2 M KCl | 166 µL | 33.3 mM |
1 M HEPES | 300 µL | 50 mM |
10 mg/mL BSA | 1.39 mL | 1.39 mg/mL |
1% Methylcellulose | 1.39 mL | 0.0014 |
Water | Up to 10 mL | |
BME | 0.7 µL | 1 mM |
Septin Storage Buffer (SSB) | ||
Stock | Volume | Final concentration |
2 M KCl | 1.5 mL | 300 mM |
1 M HEPES | 500 µL | 50 mM |
Water | Up to 10 mL | |
BME | 0.7 µL | 1 mM |
Table 1: Buffer components for preparation of supported lipid bilayer and reactions. Volumes of stock solutions that are incorporated into buffers and the final concentrations of each component are shown. SLB and PRB can be stored at room temperature and reused between experiments. Reaction buffer and SSB are made fresh for each experiment.
2. Spherical supported lipid bilayers
NOTE: This protocol uses silica microspheres suspended in ultrapure water at 10% density. For any work on the kinetic parameters of protein assembly, it is important to strictly control the total membrane surface area between experiments and curvatures. Table 2 shows the corrected volumes of beads and buffer to maintain 5 mm2 of total membrane surface area. This protocol expands on a previously published method8,18.
Bead Diameter (µm) | Volume of well-mixed beads (µL) | Volume of SLB buffer (µL) | Volume of SUVs (µL) |
6.46 | 8.94 | 61.1 | 10 |
5.06 | 7 | 63 | 10 |
3.17 | 4.39 | 65.6 | 10 |
0.96 | 1.33 | 68.7 | 10 |
0.54 | 0.75 | 69.3 | 10 |
0.31 | 0.43 | 69.6 | 10 |
Table 2: Normalized volumes of microspheres. In order to maintain an equal surface area of each bead size and to keep the total membrane surface area consistent between experiments, volumes for each bead size and buffer that normalized the total surface area were calculated.
Bead diameter (µm) | Sedimentation velocity (RCF) |
0.31 | 4.5 |
0.54 | 4.5 |
0.96 | 2.3 |
3.17 | 0.8 |
5.06 | 0.3 |
Table 3: Sedimentation velocities for microspheres of varying diameters. For each bead diameter, the shown minimum sedimentation velocities were used to pellet the beads for washing away unbound liposomes.
3. Rod supported lipid bilayers
NOTE: In contrast to the other assays presented here, the rod assay does not allow for careful control of the total membrane surface area. One can be consistent in amounts and volumes between experiments, but because this results in rods of different lengths and diameters, it is difficult to extrapolate the total membrane surface area in the reaction. Thus, while this is an excellent assay for exploring curvature sensing with multiple curvatures on a single surface and has been useful for exploring septin ultrastructure, it is not recommended for kinetic measurements. This method was previously reported18 and is being expanded upon here.
Following the preparation of each SLB, septins or the protein of interest may be incubated with the desired support and imaged via TIRFM, confocal microscopy, or SEM. The results shown here use septins recombinantly expressed and purified from E. coli17. Using TIRFM on planar SLBs, it is possible to determine the length of filaments and their flexibility, measure the diffusion coefficients and observe assembly over time28,29. In order to collect the highest quality measurements, it is first necessary to ascertain the quality of bilayers, especially when preparing them for the first time or when changing lipid compositions. A visual inspection of the protein distribution on the bilayers by TIRFM can help identify regions of the bilayer that have been scratched or are malformed. Protein distribution should be homogeneous (Figure 3A), and there should not be holes or gaps in the bilayer (Figure 3B). It is best to avoid membranes with holes, which can form from dust contamination or smudges from slide handling, as this can change the protein distribution in other areas of the bilayer. To visualize the membrane itself, trace rhodamine-PE can be incorporated into SUVs for bilayer formation. In high-quality bilayers, the field will appear even (images not shown), but if liposomes are old or if washes are not stringent enough, unburst and tubulated liposomes may accumulate on the surface (Figure 3C). Additionally, while solid supports will hamper the free diffusion of lipids8, the lipids should not be immobile. FRAP experiments can be used to assess the mobility of lipids on planar bilayers, which may vary by composition and should show recovery rates in the order of seconds30.
Spherical supports can be used to examine protein binding on membranes of defined curvatures either in isolation (one curvature) or with several membrane curvatures in the same well to observe competition between curvatures6,18. Near-TIRFM on beads >1 µm can also be used to measure the number of association events for a given area27. As with planar bilayers, we use rhodamine-PE to look for smooth bilayer deposition (Figure 4A), i.e., no lipid clumps, which indicate multi-lamellarity or unburst liposomes (Figure 4B), and no gaps in the bilayer (Figure 4C). For measuring the total protein adsorption or the protein adsorption over time on curved surfaces, it is necessary to isolate individual beads from each other in order to create a discrete volume for which sum lipid intensity and sum septin intensity can be measured. Thus, the beads should be well-separated rather than clumped together as in Figure 4D. We address troubleshooting options for all of these potential issues in the discussion section.
This SLB assay can also be applied to rod supports, which offer an environment for proteins to sample multiple curvatures on a single surface. Pairing this assay with scanning electron microscopy allows the user to examine curvature preference, alignment, and length distribution of septins18 or other proteins of interest. While similar to the other assays presented here, the rod assay does not allow for careful control of total membrane surface area because of the heterogeneous nature of the substrate that is derived from filter paper. The material properties of the filter paper used here result in rods of different lengths and diameters (Figure 5A); this is useful for exploring protein curvature sensing and organization on curved surfaces (Figure 5B), but because the ratio of protein to the membrane cannot be controlled, the rods are of limited utility for generating saturation-binding curves or measuring parameters such as binding constants.
Figure 1: Overview of supported lipid bilayer formation on supports with various curvatures. SUVs are incubated with solid supports of different geometries in order to change the curvatures available for sampling on a given membrane. SUVs adsorb onto the solid support surface and rupture to create lipid bilayers. Please click here to view a larger version of this figure.
Figure 2: Schematic of reaction chamber set up. To prepare custom chambers, a 0.2 mL PCR tube is cut where the tube begins to taper and at the cap (red dashed lines). The uncut rim of the cut tube is then coated with a thin layer of UV-activated glue (blue) and placed glue-down on a coverslip. Please click here to view a larger version of this figure.
Figure 3: Representative TIRF micrographs of planar bilayers. (A) A representative image of a high-quality lipid bilayer (75% DOPC, 25% Soy PI)with non-polymerizable septins bound (green). Protein is not clustering in any specific regions, and there are no liposomes attached to the membrane and no holes. (B) This bilayer was made with poorly cleaned glass coverslips and exhibits what appear to be "holes" (white arrows) in the membrane where there are fewer septins. Septins can be seen crowded at the edges of these defects in the bilayer (denoted by the white arrows in the zoomed region on the right). (C) A representative image of a low-quality bilayer using trace rhodamine-PE. Unburst liposomes and tubulated lipids are visible on the surface. Please click here to view a larger version of this figure.
Figure 4: Representative micrographs of lipid-coated microspheres. Representative images from competition assays where lipid-coated beads (75% DOPC, 25% Soy PI, 0.1% Rhodamine PE)with diameters of 0.3 µm, 0.5 µm, 1 µm, 3 µm, and 5 µm were mixed. All images are maximum Z projections. (A) Representative image of a high-quality mixture of lipid-coated microspheres. The spherical supports are evenly coated by the membrane, and there are few bead clusters. (B) Representative image of uneven membrane coating, likely caused by insufficient washing of excess lipids. (C) Representative image of beads with gaps in membrane coverage (white arrows) due to improper handling. (D) Representative image of densely clustered beads, likely caused by insufficient mixing throughout the procedure, especially when combining the different beads together. Please click here to view a larger version of this figure.
Figure 5: Representative electron micrographs of lipid and septin-coated rods. (A) SEM image showing the distribution of lipid-coated (75% DOPC, 25% Soy PI, 0.1% Rhodamine PE) rod lengths and diameters. (B) The edge of an isolated membrane-coated rod with septin filaments aligned along the axis of positive curvature. Please click here to view a larger version of this figure.
Cell membranes take on many different shapes, curvatures, and physicochemical properties. In order to study the nanometer-scale machinery through which cells build micrometer-scale assemblies, it is necessary to design minimal reconstitution systems of membrane mimetics. This protocol presents techniques that precisely control both membrane curvature and composition while allowing the user to easily take quantitative fluorescence measurements using widely available microscopy techniques.
The most critical components of this protocol are assembling wells on the appropriate surface and handling the lipids. The wells described here are handmade using PCR tubes and UV-activated adhesives; it is important not to get glue inside of the reaction area during assembly. This can be checked for during the experiment by imaging along the edges of the well and looking for high protein adsorption onto the cover glass. The user may choose to use alternative materials for wells, such as silicone, which can be ordered commercially, but the presented method provides sturdy wells that will not shift or leak and can support larger reaction volumes. The surface of the cover glass, in our hands, has been very important for forming planar bilayers. Optimization of plasma cleaning times or even the brand of cover glass may be needed to see even formation of the bilayers, as appropriate surface treatment and charge are critical to vesicle adsorption and fusion31,32,33. When setting up these experiments for the first time or working with new lipid compositions, fluorescence recovery after photobleaching (FRAP) experiments should be used to assess the fluidity of the membrane compared to free-standing systems30. It is expected that support systems will be less fluid than their free-standing counterparts but not immobile12.
When assessing the bilayer quality by fluorescence, it is important to look for unburst liposomes or defects in the bilayer (Figure 3C), as these can alter local protein adsorption. The following troubleshooting options may be considered. a) One can adjust the SUV preparation to improve bilayer quality. Sonication, freeze-thaw, and extrusion are all commonly used methods that may vary in ease and success based on lipid compositions. In our hands, a fully clarified SUV solution with a monodispersed size distribution results in the most homogeneous bilayers. Dynamic light scattering (DLS) can be used to assess size distribution25. b) One can increase the difference in salt concentration between the inside of the SUVs and the external buffer. By reducing the internal ion concentration (or increasing the external one), osmotic stress on the SUVs increases the likelihood of rupture31. Alternatively, altering the monovalent cation present may change the kinetics of SLB formation and aid in optimizing SLBs of new lipid compositions34. c) One can increase the number or force of washes. In the authors' experience, vigorous mixing during the washes leads to cleaner bilayers and does not appear to be disruptive; instead, the biggest issue comes from accidentally scraping the bilayer with pipette tips, which will appear as a gash in the membrane. For curved bilayers, it is best to minimize handling with narrow pipette tips and mix gently when performing the wash steps, especially for larger (>1 µm) beads, as the membrane may be sheared from the glass beads. If persistent gaps in the spherical membrane bilayers are observed (Figure 4C), increasing the width of the pipette tips (by cutting off the end of the tip or buying wide tips) and reducing handling as much as possible, while still completely mixing the bead solutions, can help. One common technical issue with this technique is the tendency of the beads to clump together (Figure 4D). This can be partially resolved by increasing the sonication time before bilayer formation; however, some clustering is unavoidable as membrane-coated beads may also tend to clump together over time in the well. Clumped beads should be avoided for quantitative analyses.
Limitations to quantitative measurements of single-molecule dynamics on planar bilayers include the need to stay at low concentrations in order to accurately track and count the particles. However, this can be rectified by under-labeling the protein of interest during acquisition to reduce the number of particles visible during tracking35. Additionally, spherical and rod SLBs were developed specifically to quantify the curvature sensitivity of proteins on micrometer-scale membranes, and the silica microspheres used here are only available down to 100 nm in diameter, at which point they are diffraction-limited puncta when viewed by light microscopy. Thus, those looking to assess precise curvature specificity on smaller diameters will be unable to do so using this method. However, this technique will still provide valuable information about the trend of curvature preference and allow for the dissection of curvature-dependent assemblies.
Compared to free-standing lipid bilayer systems, this technique is amenable to a wide range of buffer conditions and does not require careful osmotic balancing during preparation. Additionally, because it uses glass supports, the bilayers readily sink to the bottom of the well, removing the need for tethering or the use of sucrose or other thickening agents36. While free-standing lipid bilayer systems provide a useful means of investigating membrane deformation by membrane-binding proteins, it is difficult to study the relationship between membrane curvature and complex assembly. Unlike easily deformable membrane systems, this approach allows the user to use a variety of lipid compositions without the intrinsic shape of individual lipid species dictating the global geometry of the membrane. Lastly, rod-shaped SLBs allow sampling of multiple curvatures on the same continuous membrane for proteins of interest (Figure 5B), which is uniquely informative for evaluating curvature-dependent assemblies or the colocalization of multiple components.
The authors have nothing to disclose.
This work was supported by the National Institutes of Health (NIH) Grant no. R01 GM-130934 and National Science Foundation (NSF) Grant MCB- 2016022. B.N.C, E.J.D.V., and K.S.C. were supported in part by a grant from the National Institute of General Medical Sciences under award T32 GM119999.
0.2 mL PCR Tubes with flat cap, Natural | Watson | 137-211C(EX) | |
0.5 mL low adhesion tubes | USA Scientific | 1405-2600 | |
Beta mercaptoethanol (BME) | Sigma-Aldrich | M6250-100ML | |
Bovine Serum Albumin (BSA) | Sigma-Aldrich | A4612-25G | |
Coverglass for making PEGylated coverslips | Thermo Scientific | 152450 | Richard-Allan Scientific SLIP-RITE Cover Glass 24×50 #1.5 |
DOPC | Avanti Polar Lipids | 850375 | |
Egg Liss Rhodamine PE | Avanti Polar Lipids | 810146 | |
EMS Glutaraldehyde Aqueous 25%, EM Grade | VWR | 16220 | |
EMS Sodium Cacodylate Buffer | VWR | 11652 | |
Ethanol, 200 proof | Fisher Scientific | 04-355-223EA | |
HEPES | Sigma Aldrich | H3375-1KG | |
Hexamethyldisilazane | Sigma-Aldrich | 440191 | |
Magnesium chloride | VWR | 7791-18-6 | |
Methyl cellulose 4000cp | Sigma-Aldrich | M052-100G | |
Microglass coverslips for planar bilayers | Matsunami | Discontinued | 22×22 |
Mini centrifuge | |||
Non-Functionalized Silica Microspheres | Bangs Laboratories, Inc. | Depends on size: SS0200*-SS0500* | Silica in aqueous suspension |
Optical Adhesive | Norland Thorlabs | NOA 68 | Flexible adhesive for glass or plastics |
Osmium tetroxide | Millipore Sigma | 20816-12-0 | |
Parafilm | VWR | 52858-000 | |
Plasma Cleaner | Plasma Etch | PE-25 | Voltage: 120V, 60Hz. Current: 15 AMPS |
Potassium chloride | VWR | 0395-1kg | |
Round coverglass, #1.5 12mm | VWR | 64-0712 | |
Sonicator bath | Branson | 1510R-MT | Bransonic Ultrasonic cleaner. 50-60 Hz. Output: 70W |
Soy PI | Avanti Polar Lipids | 840044 | |
Tabletop centrifuge | Eppendorf | 22331 | |
UV Lamp | Spectroline | ENF-260C | 115 Volts, 60 Hz, 0.20 AMPS |
WhatmanGlass Microfiber Filter Paper | VWR | 28455-030 | 42.5 mm diameter, Grade GF/C |