A convenient protocol for generating giant unilamellar vesicles containing SNARE proteins using electroformation

Reconstitution of membrane proteins in artificial membranes is an essential prerequisite for functional studies that depend on the context of an intact membrane. While straight-forward protocols for reconstituting proteins in small unilamellar vesicles were developed many years ago, it is much more difficult to prepare large membranes containing membrane proteins at biologically relevant concentrations. Giant unilamellar vesicles (GUVs) represent a model system that is characterised by low curvature, controllable tension, and large surface that can be easily visualised with microscopy, but protein insertion is notoriously difficult. Here we describe a convenient method for efficient generation of GUVs containing functionally active SNARE proteins that govern exocytosis of synaptic vesicles. Preparation of proteo-GUVs requires a simple, in-house-built device, standard and inexpensive electronic equipment, and employs a straight-forward protocol that largely avoids damage of the proteins. The procedure allows upscaling and multiplexing, thus providing a platform for establishing and optimizing preparation of GUVs containing membrane proteins for a diverse array of applications.

Nevertheless, the preparation of proteo-GUVs still requires time-consuming optimization as protocols are frequently difficult to reproduce between laboratories (own experience and personal communication with other researchers in the field), largely because not all variables affecting the outcome are controlled and optimised. This includes, for instance, design of the electroformation chamber including slide resistance when using ITO slides, spacer thickness (that is necessary for calculating the electric field), and for Pt electrodes information about the thickness and the axial distance between the wires. Moreover, the parameters of the applied electric field are critical for the outcome, with the results depending on the chamber geometry and the precise voltage-time profile of the applied electric field.
Here we report a convenient protocol for the preparation of proteo-GUVs containing functionally active neuronal SNARE (soluble N-ethylmaleimide-sensitive factor activating protein receptor) proteins for the study of membrane fusion in vitro. SNARE proteins represent a superfamily of small, mostly membrane-anchored proteins that catalyse the fusion of membranes in all eukaryotic cells. In neurons, exocytosis of synaptic vesicles is mediated by the SNARE proteins syntaxin-1A and SNAP-25 present at the plasma membrane, and synaptobrevin-2 present on the vesicles. Our protocol is straightforward and requires only a simple and affordable, in-house-built setup, therefore it can be easily adapted for other proteins and lipid compositions.

Results
Setup Design. Electroformation of GUVs can be performed by applying an alternating electric field in the formation chamber that either consists of two glasses coated with conductive material (such as ITO) that are separated by a spacer (see e.g. refs 18,[20][21][22] ; Fig. 1a), or that contains two Pt electrodes (presented in this work; Fig. 1b). In the first approach, the parameters that critically influence the electroformation are the electrical resistance of the ITO coat and the distance between the two conductive surfaces. When Pt wires are used for preparation of GUVs, the two important parameters are the axial distance and the thickness of the two parallel electrodes. In both cases, additional parameters have a strong influence on GUV formation such as chamber volume, chamber cleanliness and deterioration due to repeated use, concentration of lipids, the method of lipid drying, and the composition of the buffer used for electroformation.
Previously, we prepared SNARE-containing GUVs using ITO-coated glasses 18,20 (see also Fig. 1a). However, GUV formation was not easily reproducible and yielded relatively small GUVs (often below 5 µm in diameter), prompting us to work out a more reliable protocol that would yield larger GUVs. We therefore designed a Pt wire-based electroformation chamber (in short Pt chamber; Figs 1b and 2a; other designs are presented for example in refs 13,23 ). The main idea was to design a chamber that could be effectively cleaned with organic solvents (to remove residual lipids), and that allows for easy monitoring of GUV formation (compatibility with a standard Zeiss microscopy stage). For this purpose, a largely chemically inert PTFE (polytetrafluoroethylene) was chosen as the main material for the chamber with two Pt wires (0.5 mm in diameter) embedded close to the chamber bottom. The dimensions of the chamber were chosen to allow for sealing with a standard size (25 mm in diameter) microscopic coverslip and to fit, together with the wiring, on a microscopy stage (Figs 1b and 2a,e). For electroformation we used a digital function generator (Velleman PCGU1000), connected via USB to a Windows PC (Fig. 1c). This function generator is inexpensive in comparison to other (usually stand-alone) laboratory function generators, and the output voltage waveform can be easily programmed within the accompanying software (PcLab2000SE, Velleman). The Pt chamber was connected to the function generator using cable with BNC connector and pin socket. In this setup socket pitch of 2.54 mm fits the 2.5 mm axial distance of Pt electrodes, while for the ITO chamber the latter were replaced with crocodile clips (Fig. 1c). Additionally, by using BNC Y-splitters, multiple chambers can be connected to one function generator. In conclusion, the whole electroformation setup consists of a PC, a function generator, connecting cables, and electroformation chambers (Fig. 1c). One electroformation chamber can be then placed on a microscope (as shown in Fig. 2e, to allow live monitoring, see Supplementary Video 1), while others can be placed for stability in a suitable stand (like the one made from the polyethylene foam shown in Fig. 2d).

Formation of GUVs in a Pt chamber. The most common approach for formation of GUVs containing
transmembrane proteins is to start with small liposomes reconstituted with membrane proteins using standard protocols, e.g. using detergent removal by size exclusion chromatography 3,20,24 (Fig. 3a). These proteoliposomes are deposited on the conductive surface in the electroformation chamber and dried in order to remove the aqueous buffer. For instance, for preparing SNARE-GUVs with a Pt chamber 5-7 × 1 µl drops of SUVs are deposited on each Pt wire (10-14 drops/Pt chamber) and dried under vacuum for around 30 minutes (Fig. 3a). Next, the Pt chamber is sealed with a coverslip (25 mm diameter, coated with β-Casein to prevent bursting of GUVs making contact with the glass surface) and a silicone glue (see photo in Fig. 2c). The sealed chamber is then connected to the function generator and filled with an electroformation solution -typically water with sucrose (we used 800 µl of 200 mM sucrose solution in each chamber; see photos in Fig. 2d and e). Immediately afterwards electroformation is started by switching on the AC field. In our hands, the best GUV quality and highest protein activity was obtained when electroformation was performed for 1 h at 10 Hz, 2.2 V pp (peak-to-peak voltage, sine wave shape), followed by a detachment phase (detaching GUVs from Pt wires into the solution) of 30 min at 2-4 Hz, 2.2 V pp (sine wave shape, Fig. 3b and Supplementary Video 1). After detachment, GUVs are collected by pipetting with a cut 1 ml micropipette tip and transferred directly to the imaging chamber, or stored refrigerated for up to a week in a microcentrifuge tube (Fig. 3a).
GUV quality analysis -vesicle diameter and efficiency of protein reconstitution. Depending on the biological problem to be studied, the average diameter of GUVs may be critical. For instance, in some experiments handling and visualization of larger GUVs may be beneficial. The SNARE-GUVs prepared with a Pt chamber have diameters ranging from around 5 to 30 µm (Fig. 4a and c). Thus, the average diameter (13.5 µm, Fig. 4a) is substantially larger than that of the same GUVs prepared with ITO slides (5.8 µm) 18 . Another parameter critical for the assessment of GUV quality is the amount of protein incorporated in the membrane. In GUVs prepared with a Pt chamber we observe efficient protein incorporation by monitoring fluorescence intensity of Texas Red labelled proteins in the GUV membrane (Fig. 4b). By comparing these intensities with those of labelled lipid (for details see Materials and Methods and ref. 18 ), the protein concentration in the membrane can be estimated (see histogram in Fig. 4b).
Although the protein to lipid ratio showed some variability, there was no correlation with the size of the GUVs.
GUV quality analysis -protein activity. SNARE proteins catalyse most membrane fusion reactions in eukaryotic cells 25 . Therefore, the best test for their activity upon membrane reconstitution and vesicle formation is to perform fusion assays 26 . Here we measured fusion using a lipid mixing assay 18,27 . In this experiment, immobilised GUVs containing labelled lipid NBD-PE as fluorescence donor and a stabilised complex of plasma membrane SNARE proteins 28 , were incubated with SUVs containing Lissamine Rhodamine-PE (Rho-PE) as fluorescence acceptor and the vesicular SNARE (schematic illustration in Fig. 4d). Upon SNARE-mediated membrane fusion, these two labels are in the same membrane and undergo Förster resonance energy transfer (FRET), causing quenching of NBD. If Rho is then bleached, a corresponding recovery of the NBD fluorescence intensity is observed (Fig. 4e, red). As a control for the specificity of this reaction, we used a synaptobrevin mutant (Δ84) 29 that stops the fusion reaction at the docked state, preventing mixing of lipids and thus reducing FRET (Fig. 4e, grey).

Discussion
Here we describe a convenient procedure for preparing proteo-GUVs containing SNARE-proteins of the presynaptic plasma membrane using in-house-built devices. The electroformation chamber described here is made from PTFE and thus can be cleaned with organic solvents. Moreover, the chamber allows for directly monitoring the formation of GUVs under a microscope. Additionally, the function generator used in this study can be easily programmed, allowing for testing of multiple electroformation protocols (a crucial step when establishing a protocol for proteo-GUV formation). The protocol described here is convenient and avoids some of the problems associated with other methods. For instance, the procedures involving osmotic shock 6 require repetitive drying-rehydration cycles, which are likely to be detrimental for maintaining membrane proteins in a functional state. Furthermore, gel-assisted swelling 30 was reported to yield GUVs with altered mechanical properties 31 . Another possibility is to reconstitute proteins into the preformed GUVs with the aid of low concentrations of a detergent 10 , yet it requires extensive optimization of detergent type and concentration 3 , and it is very difficult to achieve efficient protein insertion while maintaining  the GUVs intact. For special purposes, i.e. when membrane asymmetry is required, GUVs may be prepared with inkjet method 12 , however these technique requires a more specialised, and expensive equipment.
We conclude that our protocol offers a convenient method for the preparation of large GUVs containing moderate-high concentrations of membrane proteins. The yield of high-quality GUVs is comparably high, and only a single drying step is required, helping to preserve protein activity. For sensitive proteins, additional protection during the drying process may be necessary, for instance by adding disaccharides or ethylene glycol 3,32-34 .
Labelled ΔN complex was formed by replacing SNAP-25 with a S130C mutant labelled with Texas Red.
Preparation of small unilamellar vesicles and fluorescent labelling of vesicles. Small unilamellar vesicles (SUVs) containing SNARE proteins (the plasma membrane SNARE complex or synaptobrevin) were prepared by co-micellization followed by size exclusion chromatography as described before 18

Preparation of giant unilamellar vesicles.
GUVs containing SNARE proteins were prepared from vacuum-dried proteo-SUVs with the electroformation procedure using an in-house-built Pt electrode electroformation chamber (referred to as Pt chamber, see Fig. 2). The detailed GUV preparation protocol is described in the Results section. Prior to use, the Pt chamber was cleaned by bath sonication (around 5-10 min) in ethanol and subsequently in chloroform. For sealing of the chamber, microscopy coverslips (25 mm in diameter) were used, that were first cleaned with ethanol and isopropanol, then coated with β-Casein (3 mg/ml, 5 min), and finally rinsed with water and dried.
Microscopy imaging and data analysis. The formation of GUVs was directly monitored at low magnification in the electroformation chamber with an epifluorescence microscope. For visualization in higher magnification, GUVs were collected after the electroformation procedure and transferred to the imaging chamber containing a coverslip functionalised with biotinylated BSA and neutravidin 18 , and imaging buffer (20 mM HEPES/KOH pH 7.4, 150 mM KCl, 1 mM MgCl 2 , at least 1.5 × volume of the GUV solution to be added). GUVs were allowed to settle for around 30 min prior to imaging, resulting in surface attachment. Microscopy imaging was done with a Zeiss Axiovert 200 epifluorescence microscope or with a Zeiss LSM 780 confocal microscope.
The efficiency of protein reconstitution was determined as described in ref. 13 , following the detailed protocol described in ref. 18 , by comparing membrane fluorescence intensity of Texas Red labelled ΔN complex with those of known concentration of Texas Red labelled DHPE. Bulk lipid mixing experiments were performed essentially as described in ref. 18 . Image analysis was performed in Fiji 41 with self-written scripts 18,42,43 . Data availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.