Flippase‐Mediated Hybrid Vesicle Division

2.1. Small Hybrid Vesicle Assembly and Flippase Activity

Transmembrane proteins require a lipid‐based bilayer for stability and appropriate function. However, the cell membrane has more building blocks than lipids, and using polymer lipid hybrid vesicles (HVs) instead of liposomes for protein reconstitution is a first step toward implementing a higher level of membrane complexity. HVs are self‐assembled vesicles that combine the properties of (phospho)lipids and amphiphilic block copolymers as recently discussed as parts of several reviews.^{[ 24 ]} Various types of proteins have been reconstituted into HVs including OmpF from E. coli,^{[ 25 ]} Cyt bo₃,^{[ 26 ]} the efflux Pumps NaAtm1 and P‐glycoprotein,^{[ 27 ]} and the enzyme pair bacteriorhodopsin and F₁F₀‐ATP synthase.^{[ 28 ]} Integration of fusogenic SNAREs into HVs illustrated that dynamic membrane phenomena could be reconstituted in hybrid membranes.^{[ 29 ]}

We chose to assemble HVs using 1,2‐dioleoyl‐sn‐glycero‐3‐phosphocholine (DOPC) as the major phospholipid and three types of amphiphilic block copolymers that all had a poly(carboxyethyl acrylate) hydrophilic extension, made of ≈70–80 units (10–11 kDa). Amphiphilic block copolymers that had a hydrophobic block made of only poly(cholesteryl methacrylate) (14 units, ≈6 kDa) are referred to as poly(cholesteryl methacrylate)‐block‐poly(2‐carboxyethyl acrylate) (BCP1), while poly(cholesteryl methacrylate‐co‐butyl methacrylate)‐block‐poly(2‐carboxyethyl acrylate) (BCP2) had a hydrophobic part made of a statistical copolymer between cholesteryl methacrylate and butyl methacrylate (8 units of each, ≈5 kDa). The hydrophobic block in poly(butyl methacrylate)‐block‐poly(2‐carboxyethyl acrylate) (BCP3) consisted of only poly(butyl methacrylate) (50 units, ≈7 kDa) (Scheme 1bi/ii). We previously characterized these three block copolymers in their ability to assemble as HVs with different lipids.^{[ 30 ]} Therefore, we chose 10 mol% of block copolymers for the assembly of small hybrid vesicles (sHVX), where X referred to the type of block copolymer. All sHVX contained DOPC lipids supplemented with 10 mol% DOPS lipids, the substrate for the Flippase used here, and 1 mol% PI4P, a regulatory phospholipid for the Flippase, except for the control ones (for more details on the composition, refer to Table S1 in the Supporting Information). The different sHVX were successfully assembled using the rehydration method as observed from negative stain transmission electron microscopy (TEM) images (Figure S1, Supporting Information) with average diameters between 150 and 210 nm (Table S2, Supporting Information). Dynamic light scattering (DLS) confirmed the relatively narrow size distribution of the as‐assembled sHVX populations. The hydrodynamic diameter for sHV3 was substantially smaller compared to sHV1 and sHV2, suggesting that the absence of the cholesteryl‐pendant groups affected the self‐assembly process (Table S2, Supporting Information).

There are several techniques for membrane protein reconstitution into vesicles as recently reviewed in detail.^{[ 31 ]} In our case, detergent‐based reconstitution using 3‐((3‐cholamidopropyl) dimethylammonio)‐1‐propanesulfonate (CHAPS) was employed to equip sHVX with Flippases resulting in sHVX_Flip. For comparison, sHVX went through the reconstitution process using a CHAPS‐containing buffer lacking Flippase with the aim to discriminate effects of the detergent (i.e., micelle formation, multilamellar structures, transmembrane lipid motion, or increased permeability) from the protein insertion on the morphology of the sHVX, resulting in sHVX_C. The hydrodynamic diameters of sHVX_C were smaller than the as‐assembled sHVX, with sHV3_C being the smallest. The overall smaller size of sHVX_C compared to sHVX was confirmed by negative‐stain TEM images (Figure 1 ). In addition, cryogenic transmission electron microscopy (cryoEM) images confirmed the vesicular structure of sHVX_C, i.e., the exposure to CHAPS was not detrimental for the assemblies although multilamellar structures were observed in most of the samples. Further, the hydrodynamic diameters of sHVX_Flip were comparable to the corresponding sHVX_C. In addition, sHV1_Flip and sHV2_Flip had a diameter of ≈50 nm, which was comparable to the corresponding sHVX_C (Table S2 and Figure S2, Supporting Information). However, the diameter of sHV3_Flip increased compared to sHV3_C, suggesting that the presence of the Flippase and the type of hydrophobic block incorporated in the hybrid membrane (i.e., the absence of the cholesterol pendant groups) affected the morphology.

We used sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS‐PAGE) to confirm that the Flippase was present in sHV1_Flip and sHV2_Flip. The bands at ≈130 and ≈55 kDa corresponded to Drs2p and Cdc50p, respectively, confirming the presence of Flippase in the sHV1_Flip and sHV2_Flip (Figure S3, Supporting Information).

In a next step, we aimed to illustrate that the reconstituted Flippase was active. To this end, we first compared the ATP consumption of sHVX_Flip to sHVX_C over 4 h since ATP hydrolysis is the initial step in the transport cycle of the Flippase. All three types of sHVX_Flip showed complete ATP consumption over this time, while the ATP remained at the initial level for pristine sHVX_Cs (Figure 2a‐i). It should be noted that sHVX_Flip without exposure to ATP did not show any signal in the ATP detection assay. Further, sHV2_Flip without PI4P phospholipids, but still containing PS, were also assembled, showing the expected slow ATP consumption. The residual ATP consumption was likely due to carry over PI4P added during the Flippase purification (Figure 2a‐ii). Finally, sHV2_Flip without DOPS lipids were made and complete ATP consumption over 4 h was observed (Figure 2a‐iii). The substrate specificity for PS of this Flippase has previously been reported.^{[ 23b ]} Therefore, we hypothesized that the ATP consumption was due to copurified DOPS from the Flippase purification process.

We also attempted to reconstitute Flippase into DOPC‐based liposomes (i.e., into vesicles without the block copolymer with 8:1:1 mol ratio of DOPC:DOPS:PI4P), but no ATP consumption was measured in this case, suggesting that the Flippase was either not present or the reconstitution conditions required optimization (Figure S4, Supporting Information). Therefore, no further experiments were conducted with liposomes, and we focused our current effort on HVs.

Finally, the aim was to illustrate the maintained catalytic function of the Flippase in sHVX_Flip for a prolonged period. To this end, ATP was added every 60 min for up to 180 min while the ATP consumption was monitored (Figure 2b). Repeated complete ATP consumption for all sHVX_Flip within 60 min was found, confirming the retained enzymatic activity.

The second aspect to validate the activity of Flippase reconstituted in sHVXs was to demonstrate that the PS lipids were flipped between the lipid leaflets.

To this end, we employed NBD‐tail labeled PS lipids (1‐palmitoyl‐2‐{12‐[(7‐nitro‐2‐1,3‐benzoxadiazol‐4‐yl)amino]dodecanoyl}‐sn‐glycero‐3‐phosphoserine, NBD‐PS) during the assembly of sHVXs resulting in sHVX^F‐PS. The tail‐labeled version was chosen because the Flippase recognizes the head group of the PS lipids. Dithionite is a reducing compound that has previously been used to quench NBD.^{[ 20} , ^{32 ]} sHVX^F‐PS exposed to dithionite were anticipated to have selective quenching of NBD in their outer leaflets, and upon addition of Deviron C16 detergent (Deviron), sHVX^F‐PS disintegrated and all NBD fluorophores were expected to be quenched (Figure S5a‐i, Supporting Information). This approach allowed for the distribution determination of NBD‐PS in the inner and outer leaflets of the sHVX^F‐PS. For concept illustartion, a hypothetical ≈60% loss in normalized fluorescence signal originating from the NBD‐PS after the addition of dithionite in $\text{sHV}{\mathrm{X}}_{\mathrm{C}}^{\mathrm{F}-\mathrm{PS}}$ (and the remaining 40% of this fluorescence signal was lost after Deviron addition) compared to a hypothetical ≈70% loss in normalized fluorescence signal in $\text{sHV}{\mathrm{X}}_{\mathrm{Flip}}^{\mathrm{F}-\mathrm{PS}}$ would indicate that NBD‐PS was flipped to the outer leaflet when Flippase was present (Figure S5a‐ii, Supporting Information).

The flipping of NBD‐PS was assessed in $\text{sHV}{\mathrm{X}}_{\mathrm{C}}^{\mathrm{F}-\mathrm{PS}}$ and in $\text{sHV}{\mathrm{X}}_{\mathrm{Flip}}^{\mathrm{F}-\mathrm{PS}}$ preincubated with ATP for 4 h. The loss of normalized fluorescence intensity after dithionite addition was higher for $\text{sHV}{\mathrm{X}}_{\mathrm{Flip}}^{\mathrm{F}-\mathrm{PS}}$ than for $\text{sHV}{\mathrm{X}}_{\mathrm{C}}^{\mathrm{F}-\mathrm{PS}}$, i.e., NBD‐PS lipids were actively flipped from the inner to the outer leaflet in the former case, where they were quenched by dithionite resulting in higher fluorescence intensity losses (Figure S5b, Supporting Information). This trend was observed in all the three different $\text{sHV}{\mathrm{X}}_{\mathrm{Flip}}^{\mathrm{F}-\mathrm{PS}}$. It should be noted that $\text{sHV}{\mathrm{X}}_{\mathrm{Flip}}^{\mathrm{F}-\mathrm{PS}}$ in the absence of ATP had lower fluorescence intensity loss than $\text{sHV}{\mathrm{X}}_{\mathrm{Flip}}^{\mathrm{F}-\mathrm{PS}}$ exposed to ATP but higher than for $\text{sHV}{\mathrm{X}}_{\mathrm{C}}^{\mathrm{F}-\mathrm{PS}}$. This aspect could be partly explained by the fact that the fluorescence read‐out by the multiplate reader included the entire population and effects like uneven block copolymer and Flippase distribution between the individual sHVs of the population together with the presence of micelles and other nonvesicular assemblies, limited the characterization and induced considerable variations between independent repeats. As a control experiment, we assembled $\text{sHV}{\mathrm{2}}_{\mathrm{Flip}}^{\mathrm{F}-\mathrm{PC}}$, which contained NBD‐tail labeled PC lipids (1‐palmitoyl‐2‐{12‐[(7‐nitro‐2‐1,3‐benzoxadiazol‐4yl)amino]dodecanoyl}‐sn‐glycero‐3‐ phosphocholine, NBD‐PC) that were not substrates for the reconstituted Flippase. The normalized fluorescence intensity loss of $\text{sHV}{\mathrm{2}}_{\mathrm{C}}^{\mathrm{F}-\mathrm{PC}}$ and $\text{sHV}{\mathrm{2}}_{\mathrm{Flip}}^{\mathrm{F}-\mathrm{PC}}$ after dithionite was comparable, suggesting no transmembrane movement of the NBD‐PC lipids (Figure S5c, Supporting Information).

Overall, these results suggested that DOPS lipids were flipped from the inner to the outer leaflet in sHVX_Flip.

As a complementary experiment, we used Annexin V Alexa Fluor 555 conjugate (Annexin^F) as a calcium‐mediated fluorescent indicator for the presence of DOPS in the outer leaflet (Figure 3a). Specifically, we preincubated sHVX_C and sHVX_Flip with ATP for 4 h before the addition of Annexin^F and visualization by confocal laser scanning microscopy (CLSM) with Airyscan. The fluorescence signals originating from Annexin^F were compared, and the sHVX_C and sHVX_Flip exposed to Annexin^F without ATP preincubation were used as controls. The lipid concentration was kept consistent across the different sHV types to allow for comparison between conditions (Figure S6a, Supporting Information). First, a significant increase in fluorescence intensity of Annexin^F was observed on samples containing ATP‐treated sHV1_Flip (sHV1_Flip + ATP), while the controls sHV1_C and sHV1_Flip without ATP had a comparable low Annexin^F signal, suggesting that the DOPS lipids were only flipped to the outer leaflet in ATP treated sHV1_Flip (Figure 3b). Second, the Annexin^F signal from ATP‐treated sHV2_Flip was significantly higher than all the other samples, confirming DOPS flipping (Figure 3c). However, sHV2_Flip without ATP (sHV2_Flip – ATP) exhibited a significantly higher fluorescence signal from Annexin^F compared to sHV2_C, likely suggesting that the Flippase reconstitution process contributed to the rearrangement of the lipids. Finally, only ATP‐treated sHV3_Flip samples showed a significantly higher fluorescence signal from Annexin^F, while all the other sHV3‐based samples had an equally low signal (Figure 3c). Complementary, DLS analysis of the sHV1‐ and sHV3‐based assemblies showed a single population independent of the presence of Flippase or ATP (Figure S6b‐i/iii, Supporting Information). By contrast, the DLS data of sHV2_Flip before and after the addition of ATP showed two populations compared to sHV2_C (Figure S6b‐ii, Supporting Information), while the DLS data of sHV2_Flip assembled without DOPS had a single population. Although no morphological changes were observed in the cryoEM images of sHV2_Flip, the DLS data and the super‐resolution images suggested that sHV2_Flip exhibited a different behavior compared to sHV1_Flip and sHV3_Flip.

Taken together, these results illustrated the insertion and activity, i.e., both, ATP consumption and translocation of DOPS lipids, of Flippase in sHVs.

2.2. Giant Hybrid Vesicles

Giant vesicles have been used for a variety of purposes from membrane dynamic studies,^{[ 33 ]} to the assembly of minimal cells to address questions related to the origin of life,^{[ 34 ]} and the design of imitates of biological cells to elucidate fundamental biological phenomena in a controlled and simplified environment.^{[ 35 ]} In addition, giant vesicles allow for direct observation of single vesicles and not only the population as is often the case for nanosized vesicles. We used electroformation to assemble giant hybrid vesicles (GHVX) where X indicates the used block copolymer (1 mol% of cholesteryl‐pendant groups unless indicated otherwise) in addition to DOPC and DOPS (94 and 4 mol%, respectively) as well as PI4P (1 mol%) in all cases (Table S1 in the Supporting Information for detailed compositions). Furthermore, we included poly(cholesteryl methacrylate‐co‐butyl methacrylate)‐block‐poly(6‐O‐methacryloyl‐d‐galactopyranose) (BCP4), which was composed of cholesteryl methacrylate and butyl methacrylate (8 units each, ≈5 kDa) as the hydrophobic block and 30 units of 6‐O‐methacryloyl‐d‐galactopyranose (≈10 kDa) as the hydrophilic extension.^{[ 36 ]} The chosen lipid composition together with the batch of BCP1 used here (14 units of cholesteryl methacrylate (6 kDa) and 80 units carboxyethyl acrylate (11.5 kDa)) did not result in the assembly of GHV1, although we had previously used a block copolymer with a similar hydrophobic block but an ≈2× longer hydrophilic block that allowed for giant hybrid vesicle assembly.^{[ 30a ]} By contrast, GHV2, GHV3, and GHV4 were successfully assembled, confirmed by CLSM images where 1,2‐dimyristoyl‐sn‐glycero‐3‐phosphoethanolamine‐N‐(lissamine rhodamine B sulfonyl) (Rho‐PE) and either Oregon Green 488‐labeled BCP2 (^OGBCP2), Oregon Green 488‐labeled BCP3 (^OGBCP3), or nitrobenzofurazan (NBD)‐labeled BCP4 were used (Figure S7, Supporting Information). The signals originating from the fluorescently labeled building blocks were homogeneously distributed in the GHVX membranes. However, not all the GHVXs had the same signal intensity, suggesting that the lipids and block copolymers were not evenly distributed between the individual GHVX in the population, an observation we previously made for similar assemblies.^{[ 30a ]} We speculated that this observation might be due to the uneven distribution of ^OGBCPX in the dried film before the rehydration step, resulting in GHVs with different amounts of incorporated ^OGBCPX. The distribution of the membrane constituents of an individual GHVX probably equilibrated via lateral movement within the membrane without exchange of the building blocks between different GHVXs. Interestingly, GHV4 and GHV4^2.5 (2.5 mol% BCP4) showed a similar distribution of the BCP4 between all the vesicles without BCP4 trapped in their lumen (Figure S7c/d, Supporting Information).

Subsequently, Flippase was fluorescently labeled with DyLight 633 NHS Ester (Flippase^F) for visualization purposes by CLSM. First, when DOPC‐based GUVs (no block copolymer, 8:1:1 mol ratio of DOPC:DOPS:PI4P) were used for Flippase^F reconstitution, no insertion was observed independent on the presence of CHAPS when using our current protocol (Figure S8a/b, Supporting Information). Consequently, no further experiments were conducted with these GUVs. Next, GHVXs were used for detergent‐mediated Flippase reconstitution, but the GHVXs aggregated and disintegrated in the presence of CHAPS (Figure S8c, Supporting Information). Inspired by previous reports where membrane proteins were directly inserted without the need for detergent‐mediated destabilization when amphiphilic block copolymers were present in the lipid bilayers,^{[ 26} , ^{37 ]} we incubated fluorescently labeled GHVXs with Flippase^F overnight without additional detergent, only the 62 µg mL⁻¹ N‐dodecyl‐β‐d‐maltoside (DDM) present in the solution originating from the purification, before visualization by CLSM (Figure 4a). The CLSM images revealed that Flippase^F inserted homogenously into GHV2, GHV3, and GHV4 as well as GHV4^2.5, and these assemblies will be referred to as GHV2_Flip, GHV3_Flip, GHV4_Flip, and $\text{GHV}{\mathrm{4}}_{\mathrm{Flip}}^{2.5}$. For comparison, GHVX went through the reconstitution process lacking Flippase, resulting in GHVX_C. Interestingly, Flippase^F was only present in hybrid membranes (Figure 4a‐ii/b‐ii/c‐ii/d‐ii) and not in giant vesicles that were only made of lipids (Figure 4a‐iii/b‐iii), although both types of vesicles were present in the same population. It should be noted that when using BCP4 during the assembly, only hybrid vesicles were observed. In addition, SDS‐PAGE with silver staining confirmed the presence of Flippase in all the different GHVX_Flip, as indicated by bands at ≈130 and 55 kDa, corresponding to Drs2p and Cdc50p, respectively (Figure S9, Supporting Information). In addition, we successfully reconstituted Flippase into sHV2 in the absence of additional detergent (75 µg mL⁻¹ final concentration of DDM) as determined by ATP consumption measurements of sHV2_Flip (Figure S10, Supporting Information).

Previous efforts suggested that the increased elasticity of hybrid membranes is a major contributing factor that facilitates improved membrane protein insertion into membranes containing amphiphilic block copolymers.^{[ 37a ]} We, therefore, assessed the membrane elasticity of the different GHVXs using micropipette aspiration where brightfield images of aspirated vesicles were acquired (Figure S11, Supporting Information), and determined the stretching modulus (or area compressibility modulus, K _a) and the bending rigidity modulus (K _b) (Table S3, Supporting Information). While K _a describes how much the membrane resists stretching or compression in the plane of the bilayer, i.e., higher K _a means the membrane is more resistant to expansion, K _b defines the membrane's resistance to bending or curvature deformations, i.e., higher K _b means the membrane is more resistant to bending. The least elastic membranes corresponded to GHV4^2.5 (K _a ≈ 160 mN m⁻¹) followed by GUVs (K _a ≈ 110 mN m⁻¹), GHV3, and GHV4 (K _a ≈ 70 mN m⁻¹ for both) with GHV2 (K _a ≈ 40 mN m⁻¹) having the most flexible hybrid membranes. In other words, the hybrid membranes of GHV2 were almost 2–4 times more elastic than the other assembled giant vesicles. It should be noted that, although the hydrophobic blocks of BCP2 and BCP4 were the same, the amphiphilic nature of block copolymers was different, resulting in different mechanical properties of the hybrid membranes when assembled together with lipids. It should be noted that there was a higher viscosity environment (300 mm sucrose) inside of the GHVXs compared to the 4‐(2‐hydroxyethyl)piperazine‐1‐ethane‐sulfonic acid (HEPES) buffer surroundings in all the experiments.

In addition, we followed the time‐dependent ATP consumption of GHV2_Flip, GHV3_Flip, GHV4_Flip, and $\text{GHV}{\mathrm{4}}_{\mathrm{Flip}}^{2.5}$ in comparison to the corresponding controls without reconstituted Flippase (Figure 4a‐iv/b‐iv/c‐iv/d‐iv). All Flippase‐reconstituted GHVs consumed ATP over time, but the variations in vesicle concentration made it impossible to compare the efficacy between them.

Next, we employed dithionite to visualize the transmembrane distribution of NBD‐PS in GHV2_Flip by CLSM (Figure 5a). To this end, we preincubated GHV2_C and GHV2_Flip for 4 h with ATP, before recording CLSM images before and after exposure to dithionite (Figure 5b). As expected, the initial fluorescence signal originating from NBD‐PS was comparable in all cases. Following incubation with dithionite, this fluorescence signal was almost nondetectable for GHV2_Flip preincubated with ATP, suggesting that most of the NBD‐PS (and likely DOPS lipids in general) were located in the outer leaflet (Figure 5b‐i). On the contrary, GHV2_Flip without preincubation with ATP retained fluorescence even after exposure to dithionite (Figure 5b‐ii). Similarly, the fluorescence signal originating from NBD‐PS in GHV2_C remained visible after dithionite exposure, regardless of whether ATP was present (Figure 5b‐iii/iv).

With the aim to get a more quantitative analysis of the NBD‐PS transport, we determined the fluorescence intensity of 400 vesicles and plotted the data as box plots (Figure 5c). Not surprisingly, the initial fluorescence intensities of the vesicles were comparable, illustrated by similar means and medians. Following the dithionite exposure, the mean and median of GHV2_Flip preincubated with ATP was lower compared to the other three cases, illustrating that the NBD‐PS (and likely DOPS lipids in general) were flipped from the inner to the outer leaflet in the former case. It should be noted that dithionite did not penetrate the membrane of the GHV2s because the small vesicles inside of the GHV2s maintained their fluorescence signal after exposure to dithionite at the same level (Figure S12, Supporting Information).

Additionally, we assembled GHV2_Flip and GHV2_C using NBD‐PC lipids without adding DOPS lipids. First, they exhibited ATP consumption over time when the Flippase was reconstituted, likely due to the presence of PS lipids from the Flippase expression and purification (Figure S13a, Supporting Information). Further, these vesicles were exposed to dithionite with the aim of ensuring the specificity of the Flippase toward DOPS lipids. Not surprisingly, CLSM images showed a similar loss of fluorescence intensity after dithionite exposure for GHV2_C and ATP‐treated GHV2_Flip (both without DOPS lipids), which was confirmed by similar means and medians in the box plots of their fluorescence intensities, illustrating that NBD‐PC lipids (and DOPC lipids in general) were not actively flipped between the leaflets (Figure S13b/c, Supporting Information).

We also assembled GHV3_Flip and conducted the above‐described dithionite experiments with the aim of assessing if there were any differences based on the type of hydrophobic polymer present in the hybrid membranes. The experiments showed no distinctive differences between assemblies made with BCP2 or BCP3 (Figure S14, Supporting Information).

Taken together, these observations strongly pointed toward the ATP‐dependent flipping of NBD‐PS lipids (and likely the DOPS lipids in general) in GHV2_Flip (and GHV3_Flip) from the inner to the outer leaflet due to the reconstituted active Flippase. Our GHVX_Flip were generated by externally adding the Flippase to preformed GHVXs. With the aim to experimentally verify the Flippase orientation in the hybrid membrane, we compared the ATP consumption of intact GHV2_Flip and permeabilized GHV2_Flip where the ATP was accessible to all the Flippases. Our data showed that ≈75% of the Flippases were oriented with the ATPase domains pointing outward, consistent with the asymmetric insertion of the Flippase (Figure S15a, Supporting Information). Furthermore, CLSM images confirmed that fluorescently labeled ATP was not able to cross the membrane of GHV2, suggesting that externally added ATP only activated Flippases that were inserted with their ATPase domains pointing outward (Figure S15b, Supporting Information).

As a complementary experiment, we used Annexin^F as a calcium‐mediated indicator for the presence of DOPS lipids in the outer leaflet (Figure 6a). Specifically, we preincubated GHV2_C and GHV2_Flip with ATP for 4 h before the addition of Annexin^F and visualization by CLSM (Figure 6b‐i). The fluorescence signals originating from Annexin^F were compared, and the GHV2_C and GHV2_Flip directly exposed to Annexin^F (without ATP preincubation) were used as controls. First, GHV2_C had low fluorescence signals, i.e., comparable Annexin^F binding, independent of ATP exposure. By contrast, the immediate exposure of GHV2_Flip to Annexin^F resulted in parts of the population exhibiting higher fluorescence signals. This observation was attributed to the fact that during the reconstitution process of the Flippase in the bilayer, lipid rearrangement might have occurred or DOPS copurified with the Flippase was added in the reconstitution step. Importantly, when GHV2_Flip was incubated with ATP before exposure to Annexin^F, all GHV2_Flip in the population had a high fluorescence signal, suggesting higher and more homogenous Annexin^F binding capacity. The CLSM images were quantitatively analyzed by determining the fluorescence intensities of at least 400 vesicles per sample. The data were plotted as box plots (Figure 6b‐ii), confirming that less Annexin^F was attached to GHV2_C than to GHV2_Flip. In addition, the median and average of the fluorescence signal from Annexin^F were significantly lower for GHV2_Flip not preincubated with ATP as opposed to those exposed to ATP. Interestingly, the data for GHV2_Flip not preincubated with ATP suggested asymmetric distribution since the average was above the median, pointing toward a larger population of low intensity GHV2_Flip. By contrast, the data were symmetrically distributed for GHV2_Flip preincubated with ATP. In other words, the observation from the CLSM images, which suggested higher amounts of DOPS lipids present in the outer leaflet for GHV2_Flip preincubated with ATP, were reflected in the quantitative analysis when high numbers of vesicles were considered.

The shape of GHV2_Flip preincubated with ATP was distinctively different from all other samples. Specifically, ≈13% of these GHV2_Flip in the population had a “snowman‐” like shape without any indication of a membrane present between the two hemispheres (Figures 6b‐iii and S16a, Supporting Information). We would like to emphasize that this observation was a snapshot of a dynamic process, i.e., 13% of the entire GHV2_Flip population showed constriction at the time of observation. By contrast, none of the GHV2_C exhibited notably changes from their spherical shape. These results suggested that GHV2_Flip underwent a constriction process that might result in division. Further, the observed morphological changes in GHV2_Flip likely also explained the DLS data of sHV2_Flip, i.e., sHV2_Flip underwent shape changes when functional Flippase was reconstituted.

Additionally, Annexin^F was also employed to assess the DOPS distribution in GHV3_Flip in comparison to GHV3_C (Figure 6c). The CLSM images showed ATP‐dependent DOPS lipid flipping from the inner to the outer leaflet for GHV3_Flip, similar to GHV2_Flip (Figure 6c‐i). The quantitative analysis of the CLSM images of GHV3_C and GHV3_Flip after Annexin^F incubation qualitatively confirmed the same outcome as for GHV2_C and GHV2_Flip, i.e., less Annexin^F was attached to GHV3_C than GHV3_Flip when preincubated with ATP (Figure 6c‐ii). However, the difference was more pronounced, suggesting either higher numbers of active Flippase reconstituted per GHV3 or higher orientation of the reconstituted Flippases, which resulted in more efficient unidirectional DOPS flipping. Additionally, no GHV3_Flip with “snowman‐” like shapes were observed, suggesting differences in the membrane properties depending on the type of hydrophobic block incorporated in the hybrid membrane (statistical copolymer between cholesteryl methacrylate and butyl methacrylate for GHV2, poly(butyl methacrylate) for GHV3).

Finally, we aimed to identify if the hydrophilic extension in the BCP used during assembly affected the DOPS lipid rearrangement. CLSM image of ATP‐preexposed GHV4_Flip and $\text{GHV}{\mathrm{4}}_{\mathrm{Flip}}^{2.5}$ showed similar results to GHV2_Flip (Figure 6d‐i/e‐i), i.e., ATP‐dependent DOPS flipping to the outer leaflet independent of the amount of BCP4 used during assembly. The quantitative analysis confirmed that GHV4_Flip − ATP and GHV4_Flip + ATP had significantly higher fluorescence intensity than GHV4_C − ATP and GHV4_C + ATP (Figure 6d‐ii/e‐ii). Furthermore, no “snowman‐” like morphologies were observed suggesting that the hydrophilic extension of the block copolymer also had a consequence on the morphological changes.

Additionally, we changed the conditions with the aim of getting a better understanding of the process using assemblies made with BCP2 as an example. First, half the amount of Flippase was reconstituted in GHV2 (GHV2_Flip/2) to assess how the DOPS translocation was affected (Figure S16b, Supporting Information). CLSM images of GHV2_C and GHV2_Flip/2 independent on the preexposure to ATP when incubated with Annexin^F showed similar fluorescence intensities, and the quantitative analysis confirmed that no significant differences between the different vesicles were observed, suggesting that the amount of reconstituted Flippase was too low to facilitate detectable DOPS translocation within the considered time frame. In addition, a lower amount of ATP was employed to assess the relation between ATP hydrolysis and DOPS translocation (Figure S16c, Supporting Information). CLSM images of GHV2_Flip and the relevant controls exposed to 0.3 mM ATP, which corresponded to half the typically used ATP concentration, and after incubation with Annexin^F had similar intensities between the samples. The quantitative analysis confirmed that no significant differences in Annexin^F signal were observed between GHV2_Flip − ATP and GHV2_Flip + ATP although the Annexin^F signals were significantly lower than the controls. This result suggested that DOPS was still translocated, but to a lower extent within the used time frame. Second, we expressed and purified a Flippase mutant (Flippase^E342Q) (Figure S17, Supporting Information). The E342Q variant in Flippase prevents phosphoenzyme dephosphorylation, resulting in an ATPase and lipid transport deficient Flippase. The successful reconstitution of Flippase^E342Q in GHV2 (GHV2_M) was shown by CLSM using fluorescently labeled Flippase^E342Q (Figure S18a, Supporting Information). The CLSM images together with the quantitative analysis showed neither significant differences in the fluorescence signal originating from Annexin^F binding nor snowman‐like shaped vesicles between all the samples (Figure S18b/c, Supporting Information), confirming that no DOPS translocation was possible without the presence of an active Flippase.

2.3. Flippase‐Induced GHV Division

The constriction of GHV2_Flip was further examined to investigate whether increased transmembrane DOPS asymmetry could drive division, given the well‐established relationship between DOPS distribution and membrane curvature.^{[ 38 ]} To this end, GHV2_Flip with a fluorescent building block, i.e., Rho‐PE, were employed to assess the constriction followed by division. The ATP‐dependent shape changes of GHV2_Flip were observed by CLSM for 60 min (Figure 7a). GHV2_Flip showed initial shape changes within 10 min of ATP addition, while GHV2_Flip without ATP remained unaffected. Interestingly, constrictions were formed in GHV2_Flip after ≈20 min incubation with ATP (Figure 7b). In some cases, two constriction sites were observed in the same GHV2_Flip (Figure 7b‐iv). Not all constricted GHV2_Flip led to successful division events in the time frame of the analysis whereas others did separate within ≈30 min of ATP incubation (Figure 7c). The division process lasted ≈5 min and both building blocks were homogeneously distributed in the membrane of the parent and daughter GHV2_Flip (Figure S19a/b and Movies S1–S3, Supporting Information). Interestingly, even subsequent division of daughter cells was observed (Figure 7c‐v). It should be noted that the observed phenomenon was not due to the interaction of ATP with the GHVs as GHV2_C exposed to ATP did not show any shape changes over the imaged time (Figure S19c, Supporting Information). Furthermore, the cross‐section area ratio of GHV2_Flip that underwent division was quantified, showing that the initial vesicles had a larger cross‐section area than the constricted ones and the daughters, i.e., the cross‐section area ratio was below 1 (Figure S20a, Supporting Information). Additionally, the 2D projected perimeter ratio was close to 1 suggesting that, although the area varied when the vesicles underwent division, the membrane length remained the same during the process. We chose this way of representing our data instead of the more typical determination of the surface‐to‐volume ratio^{[ 10} , ^{39 ]} because the estimation of our vesicle shapes from the CLSM images was challenging to reliably determine the surface area before and after division. Nonetheless, this analysis provided similar insight. Although the 2D projected perimeters remained constant during GHV2_Flip constriction and division, the 3D surface‐to‐volume ratio increased, leading to the observed reduced cross‐sectional area. This reflected enhanced membrane curvature and remodeling, even in the absence of changes to the projected contour length.

After a similar analysis, GHV3_Flip did not show any morphological changes within 1 h after ATP addition (Figure S21a, Supporting Information). We attributed this observation to the stiffer membrane of GHV3 compared to GHV2.

In addition, the morphology changes of GHV4_Flip were also explored to investigate the influence of the hydrophilic extension on the spontaneous curvature changes when DOPS was translocated to the outer leaflet. To this end, GHV4_Flip containing Rho‐PE were assembled, and the shape changes were followed over 30 min using CLSM after the addition of ATP (Figure 7d). GHV4_Flip in the absence of ATP showed no morphological changes whereas the ones incubated with ATP showed constrictions after ca. 15 min. Although this constriction was faster than observed for GHV2_Flip, GHV4_Flip returned to their initial spherical shape in less than 2 min (Figure 7e and Movie S4,Supporting Information). Additionally, the cross‐section area ratio as well as the 2D projected perimeter ratio was quantified to gain insights into the constriction process (Figure S20b, Supporting Information). The cross‐section area ratio was close to 1, i.e., similar areas were observed when the original GHV4_Flip were compared to the constricted ones, as well as when the original GHV4_Flip and final ones were analyzed. In addition, the 2D projected perimeter ratios of the GHV4_Flip were also close to 1, indicating that no permanent membrane remodeling occurred. We hypothesized that the differences in morphological changes between GHV2_Flip and GHV4_Flip were due to the variations in the chemical nature of the hydrophilic block in BCP2 and BCP4. Furthermore, GHV3 and GHV4 had similar membrane elasticity, but only GHV4_Flip exhibited a (temporary) shape change. We attributed this observation to the difference in the chemical nature of the BCPs. Specifically, BCP3 had no cholesteryl pendant groups, which seemed to affect the ability of the hybrid membrane to exhibit spontaneous positive curvature due to the DOPS rearrangement.

Finally, no shape changes in $\text{GHV}{\mathrm{4}}_{\mathrm{Flip}}^{2.5}$ were observed when they were followed over 30 min after the addition of ATP by CLSM (Figure S21b, Supporting Information). The stiffer hybrid membrane due to the higher BCP4 content was likely unable to exhibit increased positive spontaneous curvature due to the Flippase‐mediated DOPS rearrangement.

Taken together, our results illustrated Flippase‐mediated GHV2_Flip division due to changes in the spontaneous curvature of the leaflets in the membranes, as a consequence of the active movement of the DOPS lipids from the inner to the outer leaflet. Additionally, GHV4_Flip showed constriction followed by recovery of the original spherical shape, highlighting the importance of the block copolymer on the morphological changes.

Materials

Adenosine 5’‐triphosphate (ATP), ATP Colorimetric/Fluorometric Assay Kit, sucrose, 4‐(2‐hydroxyethyl)piperazine‐1‐ethane‐sulfonic acid (HEPES, ×99.5%), potassium chloride (KCl), magnesium chloride (MgCl₂), sodium chloride (NaCl), calcium chloride (CaCl₂), N‐dodecyl‐β‐D‐maltoside (DDM), DL‐dithiothreitol (DTT), 3‐(N‐morpholino)propane sulfonic acid (MOPS), tris(hydroxymethyl)aminomethane, sodium dithionite, leupeptin, pepstatin, chymostatin and Deviron C16 detergent were purchased from Sigma‐Aldrich. DyLight 633 NHS Ester, Pierce Dye Removal Columns, (3‐((3‐cholamidopropyl) dimethylammonio)‐1‐propanesulfonate) (CHAPS), hydrochloric acid (HCl), sorbitol, BODIPY FL ATP (BODIPY FL 2′‐(or‐3′)‐O‐(N‐(2‐aminoethyl)urethane), Pierce Silver Stain Kit, and Annexin V Alexa Fluor 555 conjugate were purchased from Thermo Fisher Scientific. Sepharose 2B, methanol, ethylenediaminetetraacetic acid (EDTA), glycerol, thrombin, and chloroform were purchased from VWR. Phenylmethyl sulphonyl fluoride (PMSF) was purchased from Roth. Bio‐Beads SM‐2 Resin were purchased from Bio Rad. Streptavidin resin and Superdex 200 increase 10/300 were purchased from Cytiva Life sciences. 1,2‐dioleoyl‐sn‐glycero‐3‐phospho‐L‐serine (DOPS), 1,2‐dioleoyl‐sn‐glycero‐3‐phosphocholine (DOPC), 2‐dioleoyl‐sn‐glycero‐3‐phospho‐(1'‐myo‐inositol‐4'‐phosphate) (PI4P), 1‐palmitoyl‐2‐{12‐[(7‐nitro‐2‐1,3‐benzoxadiazol‐4yl)amino]dodecanoyl}‐sn‐glycero‐3‐ phosphocholine (NBD‐PC), 1,2‐dimyristoyl‐sn‐glycero‐3‐phosphoethanolamine‐N‐(lissamine rhodamine B sulfonyl) (ammonium salt) (Rho‐PE) and 1‐palmitoyl‐2‐{12‐[(7‐nitro‐2‐1,3‐benzoxadiazol‐4‐yl)amino]dodecanoyl}‐sn‐glycero‐3‐phosphoserine (NBD‐PS) were purchased from Avanti Polar Lipids. 10× BoltTM Sample Reducing Agent and 4× BoltTM lithium dodecyl sulfate (LDS) Sample Buffer were purchased from Invitrogen.

HEPES buffer consisted of 10 mm HEPES and 150 mm NaCl at pH 7.4. Ultrapure water (ddH₂O) (18.2 MΩ cm resistance) was provided by Synergy UV by Millipore.

Reaction buffer consisted of 50 mm MOPS/Tris (pH 7.0), 100 mm KCl, 5 mm MgCl₂, 1 mm DTT, and 0.5 mg mL⁻¹ DDM.

BCP1 was composed of 14 units cholesteryl methacrylate (6300 Da) and 80 units carboxyethyl acrylate (11 520 Da). BCP2 and ^OGBCP2 were composed of cholesteryl methacrylate and butyl methacrylate (8 units each, 4800 Da) and 70 units carboxyethyl acrylate (10 080 Da). BCP3 and ^OGBCP3 composed of 50 units butyl methacrylate (7100 Da), and 70 units carboxyethyl acrylate (10 080 Da). BCP1, BCP2, BCP3, and the fluorescently labeled versions were synthesized as previously described.^{[ 30a ]} BCP4 composed of cholesteryl methacrylate and butyl methacrylate (8 units each, 4800 Da) and 30 units of 6‐O‐methacryloyl‐d‐galactopyranose (9900 Da) was synthesized as previously described.^{[ 36 ]} The block copolymers will be referred to as BCPX or ^OGBCPX where X denoted the BCP employed.

Expression and Purification of ∆NC and EQ Drs2p–Cdc50p

C‐terminal biotin acceptor domain‐tagged Drs2p and C‐terminal His₁₀‐tagged Cdc50p on a pYeDP60 plasmid were coexpressed in a ∆pep4 S. cerevisiae yeast strain, (W303‐1a/∆pep4 (MATα, leu2‐3, his3‐11, ura3‐1, ade2‐1, ∆pep4, can^r, cir⁺)) kindly provided by Rosa LopezMarques. Drs2p^{∆N104/C1247}–Cdc50p (∆NC Drs2p–Cdc50p) and the EQ Drs2p^{∆N104/C1247}–Cdc50p (EQ Drs2p–Cdc50p) were designed with a thrombin cleavage site at residue 104 and 1247. Yeast transformation and growth were carried out as previously described.^{[ 23} , ^{40 ]} Yeast cells were pelleted and resuspended in cold MQ and subsequently in buffer A (50 mM Tris‐HCl pH 7.5, 1 mm EDTA, 0.1 M KCl, and 0.6 m sorbitol). The cells were pelleted and resuspended in buffer B (50 mM Tris‐HCl pH 7.5, 1 mM EDTA, 0.6 m sorbitol, 1 mM PMSF) including 1 µg mL⁻¹ of leupeptin, pepstatin, and chymostatin. The cells were broken with 0.5 mm glass beads in a bead beater and centrifuged for 20 min at 2000 × g. The resulting supernatant was centrifuged for 20 min at 20 000 × g, and the resulting supernatant was centrifuged for 1 h at 185 000 × g. The yeast membranes were resuspended in buffer A containing 1 µg mL⁻¹ of leupeptin, pepstatin, and chymostatin. The membranes were diluted to 5 mg mL⁻¹ of total protein in buffer C (50 mM MOPS‐Tris pH 7, 100 mM KCl, and 5 mM MgCl₂, 1 mM DTT, and 20% w/v glycerol) containing 1 mM PMSF and 1 µg mL⁻¹ of leupeptin, pepstatin, and chymostatin. They were solubilized in 5 mg mL⁻¹ DDM for 15 min. The solubilized membranes containing the Drs2p–Cdc50p complex were batch‐bound to streptavidin resin for 1 h at 4 °C and subsequently washed with 10 column volumes (CVs) of buffer C containing 0.5 mg mL⁻¹ DDM. The protein was cleaved from the resin overnight with 4 U mL⁻¹ thrombin at 4 °C. After elution in 10 CV buffer C containing 0.5 mg mL⁻¹ DDM, the protein was concentrated 40‐fold in a 100 kDa MWCO spin column (VivaSpin). To remove the autoregulatory domain, 25 µg mL⁻¹ PI4P and 5 U mL⁻¹ thrombin were added and incubated for 1 h at RT, resulting in the N‐ and C‐terminally truncated ∆NC Drs2p–Cdc50p or EQ Drs2p–Cdc50p protein, as previously described.^{[ 22} , ^{23 ]} Size exclusion chromatography (SEC) was performed using a Superdex 200 increase 10/300 on an Äkta Go (Cytiva Life Sciences) using a buffer containing 50 mM MOPS‐Tris pH 7, 100 mM KCl, 5mM MgCl₂, 1 mM DTT and 0.5 mg mL⁻¹ DDM. Peak fractions were analyzed by SDS‐PAGE with appropriate fractions pooled, resulting in a protein concentration of 0.8 mg mL⁻¹, referred to as Flippase or Flippase^E342Q, respectively.

Flippase Labeling

Drs2p–Cdc50p Flippase or Drs2p–Cdc50p Flippase^E342Q was labeled using DyLight 633 NHS Ester. First, 100 µL of Flippase (≈0.7 mg mL⁻¹ in reaction buffer) were mixed together with 145 µL reaction buffer and 5 µL of dye (1 mg mL⁻¹ stock concentration in dimethyl sulfoxide) and incubated for 3 h while shaking at room temperature. The excess of dye was removed using Pierce Dye Removal Columns resulting in Flippase^F.

Small Hybrid Vesicle (sHVX) Assembly

Polymer–lipid hybrid vesicles were assembled by the film rehydration method. 10 mol% of the different BCPs, 1.24 mg DOPC (25 mg mL⁻¹ in chloroform), 0.15 mg DOPS (10 mg mL⁻¹ in chloroform), 0.015 mg PI4P (5 mg mL⁻¹ in chloroform), and if required 0.037 mg of fluorescent lipid (1 mg mL⁻¹ in chloroform) were mixed in a 25 mL round bottom flask (see Table S1 in the Supporting Information for details). Samples either without DOPS or PI4P were assembled as controls. The solvent was evaporated under a nitrogen stream for 5 min. The flask was left to dry overnight under vacuum. The film was rehydrated using 1 mL HEPES buffer (pH 7.4) followed by sonicating for 5 s and vortexing for 5 s. The solution was extruded through 400 nm (×21) and 200 nm (×21) polycarbonate filter at room temperature using an Avanti Mini Extruder followed by size exclusion chromatography (SEC, Sepharose 2B) to remove impurities resulting in sHVX where X indicated the used type of block copolymer.

Small Hybrid Vesicle Characterization

The hydrodynamic diameter (D _h) and polydispersity index were determined by using dynamic light scattering (DLS, Malvern Zeta sizer Nano‐590 at λ = 632 nm at 25 °C).

The sHVXs were visualized by transmission electron microscopy (TEM). Electron microscopy grids were made hydrophilic by glow discharging (45 s, air, 15 mA) 300 mesh copper formvar/carbon grids (Ted Pella). 3 µL of sample was deposited on the grid and left to adsorb for 45 s before blotting the excess. The grids were stained with 3 µL of 2% uranyl for 5 s. TEM images were taken using a Tecnai G2 spirit instrument (TWIN/BioTWIN, FEI co.). Additionally, the sHVXs were visualized using cryoEM. The grids were prepared by adding 3 µL sHVX solution onto a 1.2/1.3 µm 300 mesh C‐flat grid. The grids were plunge‐frozen in liquid ethane using a Mark IV vitrobot (Thermo Fisher) with a blotting time of 4.5 s and a blot force of 0. The vitrified cryoEM grids were imaged Titan Krios with an X‐FEG operated at 300 kV. Movies were acquired using a Gatan K3 camera with a Bioquantum energy filter operated at a slit width of 20 eV. A nominal magnification of 130 000× and a pixel size of 0.647 Å were used for sHV1^C and sHV2_Flip. Exposures of 7.6 s fractioned into 38 frames were collected through EPU software (Thermo Fisher) at a dose rate of 60 e− Å⁻². A nominal magnification of 33 000× with a pixel size of 2.7 Å was used for the sHV2^C and sHV3_Flip. Exposures of 4 s fractioned into 52 frames were collected through EPU software (Thermo Fisher) at a dose rate of 10.3 e− Å⁻². The size distributions of vesicles were calculated from the TEM and cryoEM images by measuring at least 100 particles employing the software Fiji. A Gaussian amplitude function was used to fit the resulting histograms, and the sizes are given as the mean ± 2σ.

The total lipid amount of the vesicles was determined using the phosphovanillin method. First, 150 µL sulfuric acid (18 m) was added to 15 µL of the sample, heated to 90 °C for 10 min followed by 5 min cooling in the fridge at 4 °C. Then, 100 µL of the mixture was transferred to a 96‐well plate and 100 µL vanillin reagent (6 mg mL⁻¹ in ultrapure water/phosphoric acid mixture 1:4) was added, and the mixture was incubated for 30 min at 37 °C before recording the absorbance (λ = 540 nm) using a multimode plate reader.

Flippase Reconstitution in sHVXs

Reconstitution of Flippase in the sHVXs was performed by detergent removal using Bio‐beads SM‐2 adsorption. First, 800 µL of as‐prepared sHVX solution was incubated together with 7.5 µL HEPES buffer and 42.5 µL CHAPS (20 mm stock, 1 mM final concentration) for 1.5 h while rotating (20 rpm) at room temperature. Then, 25 µL (0.5 mg mL⁻¹ stock of Flippase) and 125 µL reaction buffer were added, and the mixture was incubated for 1 h with end‐over‐end mixing at room temperature. Then, the detergent was removed using extensively washed SM‐2 bio‐beads. First, 80 mg biobeads was added to the mixture and incubated for 1 h followed by the addition of 160 mg biobeads and incubation for 2 h. Finally, the sample was transferred to a clean tube and 160 mg of fresh biobeads was added and the mixture was incubated overnight at 4 °C resulting in sHVX_Flip. In case of controls samples without Flippase, 150 µL reaction buffer was added to the sHVX, and they were treated as above described resulting in sHVX_C.

The reconstituted vesicles were characterized by DLS and visualized by TEM and cryoEM as outlined above.

SDS‐PAGE

13 µL sHVX_C or sHVX_Flip were mixed with Bolt sample reducing agent (Invitrogen) and Bolt LDS sample buffer (Invitrogen) and heated to 80 °C for 10 min. The proteins were loaded on 4–12% Bis–Tris polyacrylamide gels and run at 120 V for 1 h using electrophoresis. Then, the gel was stained with Coomassie for 1 h.

ATP Assay in sHVXs

The ATP assay was performed as described in the manufactured protocol. Briefly, 300 µL of sHVX_Flip and sHVX_C after CHAPS treatment were incubated together with 1 mM final concentration of ATP in HEPES buffer for 4.5 h at 37 °C. 10 µL aliquots of the mixture were transferred to a black 96‐well plate after 10, 30, 60, 90, 120, 150, and 240 min and incubated for 30 min with the reagents before measuring the fluorescence intensity (λ _ex = 535 nm/λ _em = 587 nm) using a multimode plate reader (PerkinElmer). The data were normalized to the value of fluorescence intensity of 1 mM ATP (final concentration) and presented as normalized fluorescence intensity (nFI). Three independent repeats were performed.

Annexin Assay in sHVXs

Either 400 µL sHVX_Flip or sHVX_C was incubated together with 0.6 mM ATP (final concentration in HEPES buffer) for 4 h at 37 °C. Then, 50 µL of the mixture was transferred to a 96‐well black/clear bottom plate, TC surface (Thermo Fisher) and 10 µL Annexin^F, 20 µL CaCl₂ (2 mg mL⁻¹ in ddH₂O), and 20 µL HEPES buffer were added. The mixture was incubated for 30 min before it was visualized by using a LSM800 inverted laser scanning confocal microscope with Airyscan (Carl Zeiss, Germany) and 63 × magnification (λ _ex = 555 nm). Two independent repeats were performed and at least 5 images were taken per repeat. The subsequent analysis was carried out using the ImageJ software where the Flippase^F area (in %) above the threshold was analyzed for each condition. The statistical significance was determined using one‐way analysis of variance (ANOVA) with Šídák's multiple‐comparisons test, * p < 0.1; ** p < 0.05.

Dithionite Assay in sHVXs

First, 90 µL of sHVX_Flip and sHVX_C in the presence or absence of ATP (1 mM final concentration in HEPES buffer) after 4 h incubation at 37 °C were transferred to a black 96‐well plate and the fluorescence intensity was recorded for ≈2 min (λ _ex/λ _em = 478/540 nm). Then, 10 µL of dithionite salt (10 mg mL⁻¹ stock solution in HEPES buffer, pH 8.5) was added and the fluorescence intensity was recorded for an additional ≈5 min. After, 10 µL Deviron C16 detergent (20% stock solution) was added, and the fluorescence intensity was recorded for ≈3 min. The data were normalized to the value before dithionite addition and were presented as % of loss normalized fluorescence. Three independent repeats were performed.

Giant Hybrid Vesicle (GHVX) Assembly

2 µL of BCPX or ^OGBCPX (2.9 mg mL⁻¹ stock concentration), 8 µL DOPC in chloroform (25 mg mL⁻¹ in chloroform), 1 µL DOPS (10 mg mL⁻¹ in chloroform), 1 µL PI4P (5 mg mL⁻¹ in chloroform), and if required Rho‐PE, NBD‐PS, or NBD‐PC (1 mg mL⁻¹ in chloroform) were mixed in a vial. Samples without DOPS were assembled as controls. It should be noted that BCP4 was directly dissolved into the lipid mixture due to its low solubility. Specifically, 267 µL DOPC in chloroform (25 mg mL⁻¹ in chloroform), 33.4 µL DOPS (10 mg mL⁻¹ in chloroform), 33.4 µL PI4P (5 mg mL⁻¹ in chloroform) were added to either 0.65 mg or 1.3 mg BCP4 and the mixture was sonicated until dissolution of the BCP4. Then, 10 µL was used for electroformation. The mixture was evenly spread on an indium‐tin‐oxide (ITO)‐coated glass coverslips (VesiclePrepChamber, Nanion Technologies GmbH) to form a thin film, and dried in a vacuum chamber overnight. An 18 × 1 mm O‐ring was placed on this coverslip and another ITO‐coated coverslip was placed on top. The space between the coverslips was filled with 280 µL buffer solution (300 mM sucrose in ddH₂O) to rehydrate the film. An AC electric field (5 V, 10 Hz) was applied for 2 h at 26 °C to generate GHVXs. All GHVX contained 1 mol% cholesterol as part of the BCPX compared to the total amount of lipids, apart from GHV4^2.5, which referred to 2.5 mol% cholesterol as part of BCP4 compared to the total amount of lipid.

Micropipette Aspiration

Experiments were performed using a homemade setup.^{[ 41 ]} Micropipettes were prepared by pulling borosilicate capillaries (WPI, 1/0.58 mm o.d./i.d.) using a puller (P‐31, Narishige). Afterward, the pipettes were sized to about 10–15 µm in inner diameter and bent using a microforge (MF‐900, Narishige). The pipettes were passivated using a plasma cleaner (HPT‐100, Henniker plasma, parameters: power 80% for 3 min) followed by incubation with casein (0.1 mg mL⁻¹ in 300 mm sucrose) for 1 h at room temperature to limit the adhesion of the vesicles to the pipettes. The pipettes were connected to a water tank attached to a piezoelectric pressure controller (OB1 Mk3, Elveflow). The tubing was filled with distilled water, and the pipette was completely filled with 300 mM sucrose solution. The experiments were carried out using a homemade observation chamber. Several layers of Parafilm were cut with the desired shape to provide a gap. Then, the parafilm was placed between 2 glass slides (24 × 50 mm²) and heated on a hot plate which resulted in the Parafilm adhering to the glass slides. The observation chamber was filled with 10 µL of sample, and the pipette was inserted. The pressure was carefully adjusted to eliminate any flow within the pipette before bringing it into contact with the vesicle. The step pressure was gradually increased in increments every few seconds to quantify the increase in L with ΔP. Typical applied pressures P were from 0 to 200 Pa, until complete aspiration of the vesicle. The vesicles were visualized with an inverted microscope (Nikon Eclipse Ti, 20× objective, 1.5× magnification). Bright field images were recorded using a Hamamatsu camera (either ORCA‐Flash 4.0 LT or ORCA‐quest qCMOS C15550). The camera was operated and the images were analyzed using open‐source microscopy softwares µManager (2.0) and FIJI, respectively. From the images, the radius of the micropipette R _p, the radius of the spherical portion of the vesicle outside of the pipette R, and the length of the tongue of the vesicle inside the micropipette L were determined. From L, the areal strain of the vesicle α was calculated using a previously reported equation (Equation (1))

where R ₀ corresponds to the initial radius of the vesicle.

The “initial state” used to measure R ₀ corresponded to the lowest applied pressure for which a value of L could be measured.^{[ 43 ]} This lowest applied pressure corresponded to ΔP = 0. From R, the tension of the vesicle σ for each step pressure, ΔP was deduced (Equation (2))^{[ 42 ]}

In the low surface tensions domain (in our case σ ≲ 0.05 mN m⁻¹, corresponding to ΔP ≲ 10 Pa), the bending modulus K _b was determined by plotting ln σ as a function of α following Equation (3)^{[ 42 ]}

In the high‐tension regime (typically ≥ 0.75 mN m⁻¹), the stretching modulus K _a was determined by plotting σ as a function of α, following Equation (4)^{[ 42 ]}

Two independent repeats were performed per GHVX.

Confocal Laser Scanning Microscopy (CLSM)

 ≈50 µL solution was placed onto the cover slide and let to adsorb onto the slide for 5 min before visualization using a Zeiss LSM700 CLSM (Carl Zeiss, Germany). At least 5 images were taken from different areas for each sample. The following settings were used: 63 × magnification, images recorded with 1024 × 1024 ppi resolution, and 4 line averaging was used. In all cases, vesicles were analyzed at the cross‐sectional area (defined by the contour at the midplane) to avoid misinterpretation of imaging results caused by variations in vesicle positioning within the 3D space.

Protein Reconstitution in GHVXs

50 µL of GHVXs were incubated together with 50 µL (0.5 mg mL⁻¹ stock) of Flippase or Flippase^F and 300 µL HEPES buffer in presence or absence of CHAPS (0.8 mm final concentration) in a final volume of 400 µL overnight shaking at room temperature, resulting in GHVX_Flip and GHVX_C. The Flippase^F reconstitution was visualized by CLSM using λ _ex = 488 nm for ^OGBCPX or NBD‐PC, λ _ex = 555 nm for Rho‐PE, and λ _ex = 639 nm for Flippase^F.

Silver Staining

GHVX and GHVX_Flip were mixed with Bolt sample reducing agent (Invitrogen) and Bolt LDS sample buffer (Invitrogen) and heated to 80 °C for 10 min. The proteins were loaded on 4–12% Bis–Tris polyacrylamide gels and run at 120 V for 1 h using electrophoresis. Then, the gel was stained as described in the manufactured protocol from Pierce Silver Stain Kit.

ATP Assay in GHVXs

400 µL of GHVX_C or GHVX_Flip were incubated together with 25 µL ATP (0.6 mM final concentration in HEPES buffer) and the ATP assay was performed as previously described.

Annexin Assay in GHVXs

400 µL of GHVX_Flip or GHVX_C was incubated together with 0.6 mM ATP (final concentration in HEPES buffer) for 4 h at 37 °C. Then, 50 µL of the mixture was transferred to a clean tube and 20 µL Annexin^F, 20 µL CaCl₂ (2 mg mL⁻¹ in ddH₂O), and 20 µL HEPES buffer were added. The mixture was incubated for 30 min before it was visualized by CLSM (λ _ex = 555 nm). Two independent repeats with two technical repeats were performed and at least 5 images were taken counting at least 200 GHVX_C or GHVX_Flip per repeat (n = 2; N = 400). The subsequent analysis was carried out using ImageJ software by drawing a line across the membrane and measuring the maximum intensity corresponding to it. The statistical significance was determined using one‐way ANOVA with Šídák's multiple‐comparisons test, *p < 0.05.

Dithionite Assay in GHVXs

400 µL of GHVX_C or GHVX_Flip was incubated together with 0.6 mm ATP (final concentration in HEPES buffer) for 4 h at 37 °C. 40 µL of the mixture was visualized by CLSM (λ _ex = 488 nm). Then, 7 µL of dithionite (10 mg mL⁻¹ stock solution in HEPES buffer, pH 8.5) was added and incubated for 5 min prior imaging. Two independent repeats with two technical repeats were performed. At least 5 images were taken counting at least 200 GHVX_C or GHVX_Flip per repeat (n = 2; N = 400). The subsequent analysis was carried out using ImageJ software. The statistical significance was determined using one‐way ANOVA with Šídák's multiple‐comparisons test, *p < 0.05. The distribution of constricted (“snowman‐” like shaped) GHV2_Flip was determined by counting 400 GHV2_Flip from two independent repeats.

Division Experiments in GHVXs

50 µL of GHVX_C or GHVX_Flip was placed onto a cover slide and 10 µL ATP (10 mm stock concentration, 2 mM final concentration in HEPES buffer) or HEPES buffer was carefully added on top. Then, the samples were visualized for 1 h by CLSM (λ _ex = 488 nm to visualize ^OGBCP2 and λ _ex = 555 nm to visualize the Rho‐PE). The area and perimeter of the different stages of the division were determined using ImageJ by outlining the vesicle's shape. The cross‐section area ratio was calculated according to Equation (5)

The 2D projected perimeter ratio was calculated according to Equation (6)

Statistical Analysis

Data processing was described in the different parts of the Experimental Section. The sample size (n) for each statistical analysis was 3 unless indicated otherwise. Data were presented as mean ± standard deviation. One‐way ANOVA followed by Tukey's multiple‐comparisons post hoc test was used to compare mean values.