Recombinant protein expression and purification:

Recombinant v-SNARE (VAMP2) and t-SNARE (SNAP25+syntaxin1A) were expressed and purified as previously described^36,49,50 . Briefly, proteins were expressed in the E. coll BL21 strain using 0.5mM IPTG for 4 hours at 37°C. Cells were pelleted and lysed using a cell disruptor (Avestin, Ottawa, Canada) in HEPES buffer (25mM HEPES, 400mM KCl, 4% Triton X-100, 10% glycerol, pH7.4) containing 1mM DTT. Samples were clarified using a 45Ti rotor (Beckman Coulter, Atlanta, GA) at 142159xg for 30 minutes and subsequently incubated with Ni-NTA resin (Thermofisher Scientific, Waltham, MA) overnight at 4°C. For t-SNARE, the resin was washed with HEPES buffer supplemented with 50mM Imidazole, 1% octylglucoside (OG), 1mM DTT pH 7.4. Proteins were eluted using HEPES buffer supplemented with 500mM Imidazole, 1% octylglucoside (OG), 1mM DTT, pH 7.4. For v-SNARE resin was washed with HEPES buffer supplemented with 15mM Imidazole, 1% octylglucoside (OG), 1mM DTT, and then with HEPES buffer supplemented with 25mM Imidazole, 1% octylglucoside (OG), 1mM DTT pH 7.4. The resin was incubated with SUMO protease in HEPES buffer supplemented with 1% octylglucoside (OG), 1mM DTT overnight at 4°C. The next day, the protein was collected as flow-through from the gravity flow column.

semiSWEET, MscL, and AqpZ were purified as described previously^4,13. Briefly, 20 ml of overnight Rosetta (DE3)pLysS cells with plasmids coding semisweet, MscL, and AqpZ were grown overnight. Then after refreshing the culture, bacteria were grown to OD – 0.8. At this point, the protein expression was induced with IPTG at a final concentration of 0.2 mM, and the bacteria were grown at 37°C for 4 hours except for semiSWEET (22°C for 15 hours). Cells were harvested by centrifugation (4000xg, 10 minutes, 4°C) and then resuspended in 100 ml of lysis buffer (20 mM Tris, 300 mM NaCl, pH 7.4 supplemented with protease inhibitor cocktail tablets (Pierce protease inhibitor mini tablets). Cells were lysed using a cell disruptor (Avestin, Ottawa, Canada) and then cell debris was removed by centrifugation (20,000xg, 20 minutes, 4°C). The supernatant was taken and the membranes were pelleted by ultracentrifugation (100,000xg, 2 hours 15 minutes, 4°C). The membranes were homogenized in 30 ml ice-cold membrane resuspension buffer (20 mM Tris, 100 mM NaCl, 20% Glycerol, pH 7.4) using a Potter-Elvehjem Teflon pestle and 2% (w/v) powder DDM was added except for MscL (1%OGNG was used) and left to tumble at 4°C for 2 hours. The samples were clarified by centrifugation (20,000xg, 40 minutes, 4°C) and then the supernatant was filtered through 0.22 μm filters. Then proteins were first purified by His-tag affinity chromatography in 20 mM Tris, 150 mM NaCl, 10% Glycerol, 0.02% DDM, pH 7.4. Then protein was cleaved from respective tags using appropriate protease (semiSWEET - Precission Protease, AqpZ, and MscL–TEV protease) and further purified using reverse Ni chromatography and size exclusion chromatography. During size exclusion chromatography the detergent was exchanged to OG for protein reconstitution into lipid vesicles.

LacY and MelB was purified as described previously^51,52.

All lipids were purchased from Avanti Polar Lipids. For each of the vesicle preparation, the composition of the lipids is described in Supplementary Table 2. The lipid mixtures were dried down in 12 x 75 mm Borosilicate glass tubes by a gentle stream of nitrogen, and any remaining traces of organic solvents were then removed under vacuum for 1 hour. A sample to produce protein-containing vesicles was prepared by dissolving the lipid film in 100μl of a reconstitution buffer containing the desired amount of the purified protein (lipid: protein ratio in the range of 100 to 1000 for different proteins). The lipids were dissolved in this solution by gentle agitation for 30 minutes at room temperature. Vesicles were then formed by rapid dilution by adding 200μl respective reconstitution buffer supplemented with 2mM DTT without detergent. Then detergent was removed by dialysis in constant flow dialyzing cassette using Spectrapore 6-8 kDa cutoff dialysis membrane against 4 liters of respective reconstitution buffer supplemented with 2mM DTT at 4°C. Vesicles were recovered and concentrated by floatation in a Nycodenz (Sigma) step gradient. Each 300μl dialysate was mixed with 300μl of 80% (w/v) Nycodenz dissolved in reconstitution buffer and then divided equally into two 5 x 41 mm ultracentrifuge tubes (Beckman). Then, each sample was overlayed with 300μl 30% (w/v) Nycodenz in reconstitution buffer followed by 100μl reconstitution buffer without Nycodenz. The samples were then centrifuged in an SW55 rotor (Beckman) at 279482xg for 4 hours at 4°C. The vesicles were carefully collected from the 0 −30% Nycodenz interface from each tube and combined. These samples were then analyzed for size in electron microscopy and dynamic light scattering (DynaPro Nanostar, Wyatt Technology). 30μl of these samples were dialyzed in constant flow dialyzing cassette using Spectrapore 6-8 kDa cutoff dialysis membrane against 4 liters of mass spec buffer (200 mM ammonium acetate, 2mM DTT) at 4°C. These dialyzed samples were used for nMS.

This method was adapted from the previous work that enabled reconstitution of membrane proteins into liposomes for CryoEM structure determination of membrane protein directly from proteoliposomes⁵³. The day before the MS experiments, Sephadex G-50 powder was dissolved in ammonium acetate buffer and sonicated in a water bath for 5 minutes. This suspension was then swelled overnight while degassing under a vacuum. On the day of the experiment, the Sephadex column was prepared by filling an empty column packed with the pre-swollen Sephadex gel. Separately, dried lipid film was resuspended in ammonium acetate buffer (200 mM ammonium acetate, 2 mM DTT). The lipid: protein ratio in the range of 100 to 1000 for different proteins. Then the solution was sonicated for 15 minutes in a bath sonicator and ten freeze-thaw cycles were performed (liquid nitrogen was used for freezing and a water bath set at 50°C was used for thawing). Then the appropriate detergent was added to the final concentration 2X CMC. This solution was then kept on ice for 30 minutes. After a 30-minute incubation, the desired amount of protein in 2X CMC detergent was added and the mixture was incubated on ice for 2 hours. This sample was placed on top of the prepared column and separated through Gel filtration to collect the fraction containing the proteoliposomes. All liposomes were prepared using 1% fluorescent lipid to conveniently track the elution of the liposomes. As a control, the same amount of protein in 2xCMC detergent was added to a blank buffer solution containing no liposomes and passed through sephadex G-50 column. Here, no proteins were observed in the elution volume corresponding to the elution volume of the proteoliposomes in the sample. Part of the proteoliposome samples were then analyzed using electron microscopy and dynamic light scattering (DynaPro Nanostar, Wyatt Technology) to confirm the size. If needed, an optional step of Nycodenz floatation, as described above, can be added to further separate the proteoliposomes from the empty liposomes.

Curvature of a curved surface is related to the radius of curvature of that point by the formula: k = 2/(R); where k is the mean curvature and R is the radius of curvature.

For a spherical liposome this R translates to the radius of the liposome. Hence, by precisely varying the radius of liposomes the curvature of the membrane can be modulated. To create liposomes of specific radius, we extruded the liposomes (prepared either by Method A or B) through specific size membrane filters 21 times. The size of the filter defines the diameter of the liposomes. In the current work, for Fig. 1d, to generate liposomes of 0.08, 0.04, and 0.02 nm⁻¹ curvatures, we extruded the liposomes through a 50, 100, and 200nm membrane filters, respectively. The size of the extruded liposomes was further confirmed through negative stain imaging (Inset to Fig. 1d).

Fluidity of a lipid membrane can be modulated by several independent factors which include chain lengths of the lipids used and the percentage of cholesterol in the lipid membrane. In this current work, for Extended Data Fig. 5, to make liposomes with varying fluidity, we systematically varied the CH% (as reported in the Extended Data Fig. 5) in the bilayer. The change in membrane fluidity was measured by FRAP experiments on the lipid mixtures used to form the liposomes.

All proteins were first checked for exact mass measurement without incorporations into lipid vesicles. For this purpose, all proteins were buffer exchanged to 200 mM ammonium acetate, 2 mM DTT with Zeba^™ spin desalting columns (Thermo Fisher Scientific). The protein concentration was kept in the range between 2μM and 10μM. Stable electrospray ionization was achieved using in-house nano-emitter capillaries. The nano-emitter capillaries were formed by pulling borosilicate glass capillaries (O.D – 1.2mm, I.D – 0.69mm, length – 10cm, Sutter Instruments) using a Flaming/Brown micropipette puller (Model P-1000, Sutter Instruments). After the tips were formed using this puller, the nano-emitters were coated with gold using rotary pumped coater Q150R Plus (Quorum Technologies). These nano-emitters were used for all nativeMS analyses. All nativeMS analysis were performed in Q Exactive UHMR (Thermo Fisher Scientific). To perform nativeMS of proteins from lipid vesicles, the nano-emitter capillary was filled with the prepared proteo-lipid vesicles and installed into the Nanospray Felx^™ ion source (Thermo Fisher Scientific). The MS parameters were optimized for each sample. The parameters are as follows: spray voltage was in the range between 1.2 – 1.5 kV, the capillary temperature was 270°C, the resolving power of the MS was in the range between 3,125 – 50,000 (for MS analysis) and 100,000 (for MS/MS analysis), the ultrahigh vacuum pressure was in the range of 5.51e-10 to 6.68e-10 mbar, the in-source trapping range was between 50V and 300V. The HCD voltage was optimized for each sample ranging between 0 to 200V. We observed that for certain proteoliposomes the use of a small amount of supercharger reduced the energetic requirements to ablate the IMPs form the bilayer and enhanced the detection of membrane proteins out of lipid bilayer. Accordingly, Glycerol 1,2-Carbonate (GC) was used while studying semiSWEET, MelB, LacY, MscL without GFP tag, and AqpZ from the bilayer. For the MS/MS analysis of the lipid-bound VAMP2, the 4+ charge state was isolated in quadrupole and fragmented in the HCD cell. The isolation of the lipid-bound peak was achieved just by applying in-source trapping voltage without any HCD energy. During MS/MS, the HCD energy was set to 100V. Nitrogen was used as collisional gas. All the mass spectra were visualized and analyzed with the Xcalibur software and assembled into figures using Adobe illustrator.

For semiSWEET oligomer percentage calculation, we used two alternative approaches. In the first approach, we used UniDec⁵⁴ (Fig. 2f and Extended Data Fig. 10). The outcome of spectra deconvolution in UniDec is shown in the inset of respective spectra in Extended Data Fig. 10 and in Fig. 2f. After spectral deconvolution, the intensity of monomer mass, dimer mass along with lipid-bound monomer and dimer mass were taken for calculations of the percentage of oligomers in detergent, 0%CL, 2.5% CL, and 10% CL containing bilayer and the final bar graph of this analysis is shown as an inset in Fig. 2e and the final mass plot is shown in Fig. 2f and Extended Data Fig. 10. In the second approach, the individual peaks were plotted in the Origin software and then the area under the curve was obtained by fitting the peak with constant baseline mode. The 6+ monomer charge state and 9+ and 11+ dimer charge states were considered for oligomer % calculation. The peak areas of protein and lipid-bound protein from Origin software were then taken for percentage calculations and the final bar graph of this calculation is shown in inset in Extended Data Fig. 10a.

Extracted E. Coli lipids were diluted 300 times in 4:1 MeOH: CHCl3 and spiked with Avanti Splash mix standards. 3μl of this mixture was loaded onto an Agilent Poroshell 120 (EC-C18 2.7 um, 1000bar, 2.1 x 100 mM) column using an Agilent 1290 Infiniti II LC. Mobile Phase A (60% Acetonitrile, 40% H20, 7.5 mM Ammonium Acetate) and mobile phase B (90% IPA, 10% Acetonitrile, 7.5 mM Ammonium Acetate) were used to separate peaks for detection on an Agilent Quadrupole Time-Of-Flight 6546 mass spectrometer. The gradient started at 85% (15%B) and decreased to 70% A over 2 minutes and then 52% A over 30 seconds. The gradient then slowly decreased to 18% A over 12.5 minutes and then 1% A in one minute and these concentrations held for four minutes. The gradient is then restored to 85% A and the column is washed for 5 minutes. Lipids were identified using MSDIAL 4.7⁵⁵. Further, these identified were manually validated through observation of class specific fragment ions and class specific retention time window. The molar concentrations were normalized to the amount of Avanti Splash mix standards spiked onto each of the samples.

Bulk fusion assay was performed as previously described ³⁶. Briefly, for the fusion assay 1 mol% NBD-PE and 1 mol% Rhodamine-PE were used along with other lipids in the v-SNARE (VAMP2) containing vesicles. The t-SNARE vesicles were unlabeled. During the fusion assay, NBD is excited at 460 nm and its emission is read at 538 nm. The Rhodamine-PE excites at 560 nm and emits at 583 nm. Since in the v-SNARE vesicles the fluorescent lipids NBD-PE and Rhodamine-PE are in close proximity, they quench each other. Consequently, upon excitation of 460nm, the NBD emission signal at 538 nm was not detected from the V-SNARE vesicles. During bulk fusion assay, prewarmed at 37°C non-labeled t-SNARE vesicles (45 μl) was mixed with labeled v-SNARE vesicles (5 μl) and the fluorescence reading was recorded immediately. Because of the fusion of v-SNARE with t-SNARE, the NBD-PE and Rhodamine-PE become distant and emission signal at 538 nm starts emerging. This signal is continuously recorded at a 30-second interval. After 90 minutes of recording, 2.5 %(w/v) DDM was added to each well and again the plate was read for 5 minutes to get the maximum emission signal of NBD-PE and the data were normalized as previously described³⁶. The fusion was plotted as the percentage of the NBD-PE fluorescence. For the final comparison between different conditions, the fusion at the 21^st minute was taken. In all experiments, as a control, we incubated the t-SNARE vesicles (45μl) with the excess soluble cytoplasmic domain of VAMP2 (CDV) (5μl) that competes with the full-length VAMP2 in v-SNARE vesicles for binding the t-SNARE and consequently quenches the vesicle fusion.

Lipid vesicles were diluted 1:150 in 50mM Tris-HCl pH7.4. Five microliters of sample were applied to a carbon, Type-B 400 mesh, Cu grid that was previously glow discharged for 30 seconds. The sample was kept on the grid for 1 minute then removed with blotting paper. Five microliters of 2% uranyl formate (UFo) were applied to the grid for 5 seconds then removed with blotting paper. Immediately, 5μl UFo was applied to the grid and kept for 1 minute. The stain was removed with blotting paper, and the grid was left to dry for 30 minutes. All images were acquired using a JEOL JEM 1400-plus 120kV.

For the molecular dynamics (MD) simulations of VAMP2, we used the solution NMR structure, PDB ID: 2KOG⁵⁶, taking only the ordered region (residues 26-116), and using structure 1 of the NMR ensemble. Martinize (v2.6) was used to convert the atomistic NMR structure to a coarse-grain (CG) MARTINI model, using the Martini 2.2 forcefield^57,58 and ElNeDyn elastic network⁵⁹.

The INSANE python script⁶⁰ was used to generate a bilayer and embed the CG protein structure therein. The lipid composition of this membrane was selected based on the Takamori et al.³⁸, to ensure that the simulations were performed in a similar membrane environment to that of synaptic vesicles in vivo. Thus, the membrane was comprised of di-C16:3-C18:3 phosphatidylethanolamine (PE), di-C16:3-C18:3 phosphatidylserine (PS), cholesterol (CH), and C16:0-18:1 phosphatidylcholine (PC), in the ratio 25:10:45:20. The box size was 17nm by x 17nm x 15nm and a physiological salt concentration of 0.15M was used. Three repeats were carried out, each of 10ms.

All simulations were performed using GROMACS 2019.1^61,62. A time step of 20 fs was used and periodic boundary conditions were applied. The temperature was maintained at 323 K using a V-rescale thermostat, and the pressure was maintained at 1 bar using a Parinello-Rahman barostat⁶³, with a time coupling constant of 12 ps and compressibility of 5 x 10⁻⁶ bar⁻¹. Lennard Jones and Coulomb interactions were cutoff at 1.1 nm and potential modifiers were used⁶⁴. Simulation analysis was performed using the GROMACS 2019.1 radial distribution function and plots prepared using Xmgrace⁶². The RDF was calculated for residues 26-87 of VAMP2.

FRAP experiment on supported bilayer was performed as previously reported⁶⁵. Briefly, supported lipid bilayers were formed onto the glass-bottom part of a petri dish using the Langmuir-Blodgett deposition technique⁶⁶. Glass-bottom petri dishes (35 mm dishes from MatTek with uncoated glass cover slip of 14 mm diameter, thickness number 1.5) were soaked for 1 hour at ^~ 60°C in 2% v/v cleaning detergent (MICRO-90, VWR), thoroughly rinsed with 18.2 MΩ ultra-pure water, and then dried under a stream of nitrogen. A chloroform solution of DOPC constituting the inner monolayer was first spread on degassed water in a NIMA Langmuir trough (model 611 equipped with the PS4 surface pressure sensor) and allowed to dry for 15 min at room temperature. After solvent evaporation, the film was compressed up to 33 mN/m and the monolayer was transferred onto the glass cover slip (with its lipid headgroups facing the glass surface) as the petri dish was slowly (0.5 cm/min) raised out of water. This first monolayer was allowed to dry for at least 30 min at room temperature. Meanwhile, a chloroform solution of the lipid mixture constituting the outer monolayer was spread at the air/water interface. This lipid mixture is a series of increasing cholesterol mol% (0, 15, 25, and 35) with other synaptic vesicle lipids and 1mol% NBD-DOPE. After 15 minutes, this second monolayer was transferred at a constant pressure of 33 mN/m and with a dipping speed of 0.5 cm/min onto the first monolayer (with its lipid headgroups facing the aqueous medium). After this second deposition, the bilayers remained immersed in aqueous medium throughout experiment.

FRAP experiments were performed on a confocal microscope TCS SP8 from Leica equipped with the LCS software, using an HC PL FLUOTAR 10X Dry objective (numerical aperture: 0.3; zoom: 10X) with a pinhole opened at 70.8 μm and an acquisition rate of 1 picture every 635 ms. Fluorescence bleaching and recovery were conducted as follows. For NBD: λ_exc = 488 nm; λ_em = 500–600 nm with 1 scan at 100% laser power for bleaching, and monitoring recovery at 2% of the maximum laser power. For each sample, recovery curves (average over 3 independent experiments, i.e. performed on a different region of the sample using the same bleaching conditions) were fitted with the software Mathematica (FRAP experiments described by modified Bessel functions; code provided upon request).

In order to take into account the contribution of fluorescence recovery that occurs during the photobleaching phase and can thus affect data analysis⁶⁷, the time t = 0 of the recovery phase in all FRAP experiments was set as the time of the last bleaching frame.

When working with photosensitive probes, such as NBD, one has to correct for the intrinsic photobleaching that occurs during the recovery phase and can also affect data analysis^68,69. In FRAP experiments, we measured the intrinsic photobleaching in a region far away from the bleaching zone. The corrected fluorescence recovery signal was then calculated by dividing the raw fluorescence recovery signal by the intrinsic photobleaching signal.

Diffusion coefficient (D) of 18:1 NBD-PE lipids in the outer leaflet of different supported lipid bilayer containing different amount of cholesterol were deduced by varying the area of the bleached region (disks of diameters d= 5, 10, 15 μm). The linear relationship between the bleaching area d² and the recovery time (τ) proves that lipid diffusion is controlled by Brownian motion and then diffusion coefficient (D) is calculated from the slope of these straight lines: D = d²/16τ where D, the diffusion coefficient is in μm²/s, d is bleached regions diameter, and τ is the recovery time in [s]. The diffusion coefficient (D) is then plotted against the mol% of cholesterol in the bilayer. As can be seen from the plot (Extended Data Fig. 5b), with increase in mol% of cholesterol in the bilayer the diffusion coefficient decreases. That means the fluidity of the bilayer decreases with increase in cholesterol in the bilayer.