Stairway to Asymmetry: Five Steps to Lipid-Asymmetric Proteoliposomes

Abstract

Membrane proteins are embedded in a complex lipid environment that influences their structure and function. One key feature of nearly all biological membranes is a distinct lipid asymmetry. However, the influence of membrane asymmetry on proteins is poorly understood, and novel asymmetric proteoliposome systems are beneficial. To our knowledge, we present the first study on a multispanning protein incorporated in large unilamellar liposomes showing a stable lipid asymmetry. These asymmetric proteoliposomes contain the Na⁺/H⁺ antiporter NhaA from Salmonella Typhimurium. Asymmetry was introduced by partial, outside-only exchange of anionic phosphatidylglycerol (PG), mimicking this key asymmetry of bacterial membranes. Outer-leaflet and total fractions of PG were determined via ζ-potential (ζ) measurements after lipid exchange and after scrambling of asymmetry. ζ-Values were in good agreement with exclusive outside localization of PG. The electrogenic Na⁺/H⁺ antiporter was active in asymmetric liposomes, and it can be concluded that reconstitution and generation of asymmetry were successful. Lipid asymmetry was stable for more than 7 days at 23°C and thus enabled characterization of the Na⁺/H⁺ antiporter in an asymmetric lipid environment. We present and validate a simple five-step protocol that addresses key steps to be taken and pitfalls to be avoided for the preparation of asymmetric proteoliposomes: 1) optimization of desired lipid composition, 2) detergent-mediated protein reconstitution with subsequent detergent removal, 3) generation of lipid asymmetry by partial exchange of outer-leaflet lipid, 4) verification of lipid asymmetry and stability, and 5) determination of protein activity in the asymmetric lipid environment. This work offers guidance in designing asymmetric proteoliposomes that will enable researchers to compare functional and structural properties of membrane proteins in symmetric and asymmetric lipid environments.

Significance

Distribution of lipids in natural membranes is highly asymmetric. For example, bacterial plasma membranes accumulate negatively charged lipids in the outer leaflet. Until now, only lipid-symmetric proteoliposomes were reported. This is due to the lack of proven applicability of novel methods for the preparation of lipid-only asymmetric liposomes to asymmetric proteoliposomes. To our knowledge, we present the first proteoliposomes with a multispanning membrane protein comprising a week-long stable lipid asymmetry. Starting from this, we discuss the crucial steps to take and pitfalls to avoid on the way to asymmetric proteoliposomes. This will help to close the interdisciplinary gap between fundamental membrane biophysics and the life-science-oriented community and may help render lipid-asymmetric proteoliposomes a standard, default model of membrane biophysics.

Introduction

Virtually all biological membranes show a distinct asymmetry in lipid composition across the bilayer (1). Cells spend a substantial amount of free energy on maintaining this entropically unfavorable state. The best-studied case of asymmetry is the eukaryotic cellular membrane, where the outer leaflets are enriched in sphingomyelin and phosphatidylcholine (PC) and the inner leaflets are rich in phosphatidylethanolamine (PE), phosphatidylserine, and phosphatidylinositol (2). Bacterial membranes also show asymmetry, such as the outer membrane of Gram-negative bacteria, where phospholipids are on the inner leaflet and lipopolysaccharides are on the outer (3). The bacterial plasma membranes accumulate negatively charged phosphatidylglycerol (PG) in their extracellular leaflets and keep PE and phosphatidylinositol mostly on the intracellular side of the membrane; cardiolipin (CL) is symmetrically distributed (4,5).

In the last decade, asymmetric model membranes became possible through several newly established approaches (6, 7, 8, 9, 10). However, these methods have been used primarily for lipid bilayer experiments until now (11, 12, 13, 14, 15), and their use for membrane protein studies has been limited. In the handful of studies on lipid asymmetry in the context of peptides and proteins, asymmetry was rapidly lost. Kol et al. (16) reported in 2003 that peptides result in structural disturbances in asymmetric large unilamellar vesicles (LUVs), thus creating possible flop sites that cause a decay of asymmetry over minutes or a few hours. Larger proteins with multiple transmembrane helices showed slower flip-flop in the range of several hours (16). In 2019, Doktorova and co-workers (15) also showed a loss of asymmetry via the peptide gramicidin within hours to days. Nguyen et al. reported similar behavior for other peptides like pHLIP in the range of minutes to hours (17). Lin et al. (18) studied the influence of a pore-forming protein without checking stability of asymmetry. The only study on a multi-transmembrane helix protein in asymmetric small unilamellar vesicles was published in 2015 by Vitrac et al. (19). Here, protein behavior and structure were checked before and after applying the protocol to render membranes asymmetric. Presence of the protein induced rapid flip-flop within minutes to hours as well. To our knowledge, no further findings are reported. Given the functional role of lipid asymmetry and the requirement of energy to maintain it, one should expect significant scramblase activity in asymmetric membranes to be limited to proteins with specific, asymmetry-related function (20). It is important to state that studies of scramblase activity of proteins in proteoliposomes require a rigorous elimination of other causes of lipid flip-flop such as residual detergent or inherent bilayer instability of the lipid composition chosen.

To close the interdisciplinary gap between lipid bilayer studies on asymmetry and membrane protein studies, we expanded our asymmetry assay (8) of outer-leaflet adjustment with cyclodextrin-lipid complexes to protein-containing liposomes. We focused on asymmetric proteoliposomes containing lipids that closely mimic those found naturally around bacterial membrane proteins.

Of medical interest is Salmonella Typhimurium, which has an asymmetric inner membrane with PG enriched in the outer layer (21,22) in which the integral membrane protein NhaA (ST-NhaA) is located. ST-NhaA belongs to the class of Na⁺/H⁺ antiporters (Nha) that regulate intracellular Na⁺ and pH homeostasis by exchanging small alkali cations for protons. The specificity of ST-NhaA for Li⁺ and Na⁺ and its pH-dependent activity was previously demonstrated in equilibrated, symmetric proteoliposomes (23). ST-NhaA has a mass of 41.4 kDa and is expected to have 12 transmembrane helices connected with short loops, as shown for the structure of the highly homologous NhaA from Escherichia coli (24,25).

Here, we present the first protocol, to our knowledge, that enables functional studies of a multispanning membrane protein in an asymmetric lipid membrane stable for several days.

Before presenting the exemplary, specific process with ST-NhaA, we start with a general discussion about the crucial considerations and procedures to produce such proteoliposomes with lipid asymmetry, based on the five steps illustrated by Fig. 1:

• 1)Selecting lipids: the lipid compositions of the outer and inner leaflets of the asymmetric proteoliposome (aPLUV) have to be chosen and optimized carefully to meet scientific and technical criteria. Scientifically, the aPLUVs should mimic crucial properties of the biomembrane. This may include electrostatic and mechanical properties, order, dynamics, permeability, thermodynamics of (de)mixing, and the presence of structural lipids (26) that interact specifically with the protein under study. Technically, the aPLUVs and their symmetric precursors made of inner-leaflet lipids must form stable, unilamellar liposomes with sufficiently slow lipid flip-flop. Furthermore, they should not freeze into a gel phase within the temperature range needed for study and even storage because freezing or melting is typically accompanied by membrane leakage and lipid scrambling (27,28).
• 2)Protein reconstitution: detergents are commonly used to stabilize the isolated membrane protein in solution and to destabilize the target membrane to facilitate protein insertion. The detergent is subsequently extracted by methods such as hydrophobic adsorption to BioBeads (29), rapid dilution (30), dialysis (31), or gel filtration (32). Tiny amounts of residual detergent may enhance lipid flip-flop and hence jeopardize the preparation of aPLUVs. This means a detergent-removal procedure that has proven useful for producing symmetric proteoliposomes (PLUVs) might still not fulfill the much higher requirements for aPLUVs. In addition, protein orientation in the membrane can be important. Only right-side-out proteins will experience the correct lipid environment in right-side-out aPLUVs. This requires a preferential orientation of the proteins or a selective activation of one orientation in the experiment.
• 3)Generation of asymmetry: various methods for the preparation of asymmetric model membranes are available. Cyclodextrins can exchange a wide variety of lipids. When the whole outer layer should be exchanged, the protocol using donor vesicles (6) is most suitable. The precise adjustment of the outer leaflet with a minor component is achieved using the method with donor complexes (8). For eukaryotic membrane models, the phosphatidylserine in the outer leaflet can be depleted via enzyme treatment (7,33).
• 4)Validation of stable asymmetry: the asymmetry generated in step 3 has to be verified through methods such as ζ potential (ζ) (8), fluorescence (34), NMR (14), SAXS and SANS (35), and others, depending on the lipid species to be monitored.
• 5)Characterization of protein activity: to ensure functional reconstitution, protein activity should finally be checked with an assay depending on the individual protein.

Figure 1.

Cartoon showing the five steps to lipid-asymmetric proteoliposomes. Captions describe the general steps; pictures show specific procedure for the bacterial membrane protein ST-NhaA as an example. The lipids PE, PC, and CL are shown as gray spheres with gray fatty acid chains. The lipid PG is depicted as blue spheres with orange fatty acid chains together with bound and free cyclodextrin (green spheres). The Na⁺/H⁺ antiporter ST-NhaA is shown as a green rectangle either solubilized in detergent (magenta rectangles with one hydrophobic chain) or reconstituted in liposomes. Asymmetry as measured by ζ potential (ζ) is highlighted by cyan color. ST-NhaA activity was measured using a fluorescent-probe-based assay for electrogenic transport (white star). To see this figure in color, go online.

Lipid asymmetry and protein activity should be validated again after finishing all experiments.

Materials and Methods

Materials

1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) and 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-rac-glycerol ammonium salt (POPG) were kindly provided by Lipoid (Ludwigshafen, Germany). 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE) and 1′,3′-bis[1,2-dioleoyl-sn-glycero-3-phospho]-glycerol sodium salt (TOCL) were purchased from Avanti Polar Lipids (Alabaster, AL), and randomly methylated β-cyclodextrin (MβCD) was purchased from Sigma-Aldrich (Munich, Germany).

The following detergents were purchased: n-dodecyl-β-D-maltoside (β-LM) and n-octyl-β-D-glucoside (OG) from Glycon (Luckenwalde, Germany) and n-dodecyl-α-D-maltoside (α-LM) from Anatrace (Maumee, OH). 3,3′-dipropylthiadicarbocyanine iodide (DiSC₃5), 4–12% Bis-Tris precast gel, and 4× LDS loading buffer were purchased from Thermo Fisher Scientific (Waltham, MA). Colloidal Coomassie was purchased from Expedeon (Heidelberg, Germany).

All other chemicals were purchased from Carl Roth (Karlsruhe, Germany) or Sigma-Aldrich and were of analytical grade. Vivaspin Turbo 4 centrifugal filters (MWCO 100 kDa) were purchased from Sartorius (Göttingen, Germany) and BioBeads SM-2 from Bio-Rad (Hercules, CA).

Vesicle preparation

In the first step, so-called acceptor LUVs are prepared as described previously (36) from the lipids to finally form the inner leaflet; they will be subjected to the exchange procedure to accept outside PG later. Briefly, chloroform stock solutions of lipids were mixed in round-bottom flasks and dried to a smooth film using a rotary evaporator. After 2 h of further drying under high vacuum, the resulting film was hydrated with reconstitution buffer (10 mM Tris (pH 7.4) (at 28°C), 100 mM KCl, 0.5 mM EDTA). Five freeze-thaw cycles followed, and then the lipid dispersion was extruded 51 times through an 80 nm polycarbonate membrane (Nuclepore Whatman; Little Chalfont, UK) at 35°C.

Multilamellar vesicles composed of POPG were prepared equally, except for omission of the extrusion step.

Adjustment of lipid composition to S. Typhimurium membrane

To mimic the plasma membrane of S. Typhimurium, we tried to have the lipid composition closest to S. Typhimurium as derived from Olsen and Ballou, which was reported to contain roughly 87.5 mol% PE, 10 mol% PG, and 2.5 mol% CL averaged over different growth phases (21). PG is postulated to be exclusively on the outer leaflet (22), so we aimed for acceptor vesicles with a composition of PE/CL to mimic the inner leaflet and then exchanged 20 mol% of outer-leaflet lipid with PG (20 mol% PG exclusively in the outer leaflet is equivalent to 10 mol% PG of the total phospholipid content). Because pure POPE with a small amount (2.5 mol%) of TOCL does not form stable vesicles at room temperature, we needed to add a certain amount of POPC. To test the needed amount of POPC, we prepared vesicles with 10, 20, 40, and 50 mol% of POPC and performed differential scanning calorimetry (DSC) and cryo-transmission electron microscopy (Cryo-TEM) (Figs. S1 and S2). Reconstitution of ST-NhaA into acceptors without TOCL was not successful, indicating that ST-NhaA requires CL for structural and functional integrity.

ζ potential

ζ-potential (ζ) measurements were performed as described previously (8). Briefly, liposome dispersions were filled into a flow-through, high-concentration ζ-potential cell (HCC) (Malvern) and pushed forward a few millimeters in the tubing for each measurement with Zetasizer Nano ZS at 28°C. Buffer effects like viscosity and refractive index were taken into account for data analysis (for details, see Supporting Material for Markones et al. (8)). This is especially important for MβCD-containing solutions. HCC was cleaned (with 1% Hellmanex II and ddH₂O). To ensure the quality of data, the transfer standard DTS1235 was measured before, after, and between measurements. If the standard did not meet its specification (−42 ± 4.2 mV), the cell was cleaned, and samples measured previously were remeasured.

For ζ calibration, ζ potentials of symmetric liposomes with known composition were measured. Those calibration liposomes, which should resemble the outer-leaflet composition of asymmetric LUVs (aLUVs), contained a constant fraction of 2.5 mol% TOCL because we assumed that TOCL would not be solubilized by MβCD to any significant amount during lipid exchange. It appears safe to assume that the CL content is not affected by the cyclodextrin treatment because the cyclodextrin concentration needed to complex a lipid depends strongly on stoichiometry. It needs on the order of 3 mM MβCD to form significant concentrations of 2:1 MβCD-cholesterol complexes (37) and ∼30 mM for 4:1 MβCD-POPC complexes (38). For CL, one should expect a stoichiometry of 8:1 and a much higher MβCD concentration to initiate significant complexation. This expectation has, indeed, been proven experimentally by Epand and co-workers (39). The fraction of POPG in the calibration LUVs was varied from 0 to 97.5 mol%, and the remaining lipid fraction was split to 60 mol% POPE and 40 mol% POPC.

Reconstitution and functional assay

Reconstitution

His-tagged ST-NhaA (NHAA_SALTY-(TNSENLYFQGGRGS)-His6) was produced and purified as previously described (23). The protein concentration of purified ST-NhaA was measured spectroscopically at 280 nm using the respective extinction coefficient (23). The protein was loaded onto a spin desalting column (7k, Zeba; Thermo Fisher Scientific) equilibrated in reconstitution buffer that was supplemented with 0.03% (w/v) α-LM. Vesicles were destabilized by addition of OG at 1% (w/v) final concentration (f.c.). ST-NhaA was added to the lipid-detergent suspension at a lipid/protein mass ratio of 50:1, and the solution was incubated for 1 h. Detergent was removed by addition of 30% (v/v) BioBeads shaking overnight. The next morning, BioBeads were settled and exchanged for a fresh aliquot, and the dispersion was shaken for 1 h. After removal of the BioBeads, two freeze-thaw cycles followed to obtain more unilamellar vesicles (40). The proteoliposomes were consequently extruded 51 times through a 100 nm polycarbonate membrane (LFM-100; Avestin, Ottawa, Canada). All reconstitution steps were performed at room temperature.

Na⁺/H⁺ antiport activity measurement

Electrogenic Na⁺/H⁺ antiport activity of ST-NhaA was measured in proteoliposomes using the membrane-potential-sensitive dye DiSC₃(5) and driving transport with a Na⁺ gradient.

The reaction was started by addition of 30 μL of a proteoliposome suspension to 500 μL reaction buffer (10 mM Tris (pH 8.5), 100 mM KCl, 1.8 μM DiSC₃(5). Emission of DiSC₃(5) was monitored at 668 nm using an excitation wavelength of 651 nm in a fluorescence spectrophotometer (Cary Eclipse; Agilent Technologies, Santa Clara, CA). After 30 s, transport was started, and ion selectivity was probed by addition of 100 mM (f.c.) LiCl, NaCl, KCl, or choline chloride from a 2 M stock solution. After reaching equilibrium, the proton gradient was collapsed by addition of 10 μM (f.c.) carbonyl cyanide m-chlorophenyl hydrazone (CCCP) from a 10 mM stock solution. To test the stability of ST-NhaA in asymmetric proteoliposomes, activity measurements were repeated after 14 days of proteoliposome storage at 23°C.

aLUVs

Preparation of aLUVs

Exchange was performed as described previously (8). In brief, acceptor vesicles of ∼10–15 mM lipid were mixed with donor cyclodextin-lipid complexes to enable lipid exchange of the outer leaflet in Eppendorf reaction tubes. In this step, acceptor vesicles are diluted to a final concentration of 5 mM. After 20 min of incubation at 28°C and low-speed mixing at 300 rpm (to avoid foam formation and lower the shear stress for proteoliposomes) in an Eppendorf Thermomixer, the exchange process was completed.

Separation of postexchange aLUVs from free MβCD and MβCD-POPG complexes

The separation of postexchange aLUVs from free MβCD and MβCD-lipid complexes was performed as described previously (8). In brief, vesicles and MβCD was separated by ultrafiltration using Vivaspin Turbo 4 centrifugal filters (MWCO 100 kDa). Dilution was repeated until the calculated MβCD concentration was below 1 mM (typically three cycles). The asymmetry and integrity of LUVs after the separation process was checked by ζ and dynamic light scattering (DLS).

Scrambling of postexchange LUVs

To scramble the lipid composition of postexchange LUVs, we permeabilized the purified aLUVs with OG, the same detergent as used for protein reconstitution. To destabilize the membrane, we added OG to 1% (w/v) (f.c.) and the solution was incubated for 1 h. Detergent was removed by addition of 30% (v/v) BioBeads shaking overnight. The next morning, ΒioBeads were settled and exchanged for a fresh aliquot, and the solution was shaken for 1 h. After removal of the BioBeads, two freeze-thaw cycles followed, and the proteoliposomes were consequently extruded 51 times through a 100 nm polycarbonate membrane. Through this procedure, asymmetry was abolished. The fraction of PG in the outer leaflet of these scrambled liposomes was probed by ζ measurements giving the fraction of PG in the total bilayer.

Stability test

For long-term stability studies, the MβCD was removed, the asymmetric liposomes were stored at 23°C, and ζ was measured to check the stability of the asymmetric lipid distribution.

Results and Discussion

In the following, we will describe the specific procedure of forming a stable, bacterial-membrane-mimicking aPLUVs with the Na⁺/H⁺ antiporter ST-NhaA.

• 1)Finding suitable lipid compositions turned out to be not trivial in our example. The inner membrane of S. Typhimurium consists mainly of PE distributed over both leaflets but enriched in the inner one. A small amount of CL is equally distributed, and PG is enriched in the outer leaflet (21,22). The first step of our protocol required the preparation of symmetrical liposomes of the desired inner-leaflet composition. Liposomes from POPE with and without 2.5 mol% TOCL failed to form quasispherical, mostly unilamellar vesicles. Their gel-to-fluid melting transition is close to room temperature, which limits and complicates handling, especially concerning lipid asymmetry. A PE lipid with saturated chains could stabilize the liposomes but would render T_m even higher, i.e., worse. We therefore decided to stabilize the bilayer by adding some POPC, deviating from in vivo composition but reestablishing, to some extent, biological membrane stability properties and resolving issues with handling. We tested liposomes with varying amounts of POPC by DSC, DLS, and cryo-TEM; for details, see the Supporting Materials and Methods. In our case, the optimal composition of our precursor liposomes was determined to be 58.5:39.0:2.5 POPE/POPC/TOCL (molar fractions). To mimic the natural lipid environment of ST-NhaA, CL turned out to be essential for successful protein reconstitution. After partial exchange of the outside lipid with POPG, the composition of the precursor liposomes represents the composition of the inner leaflet and the ratio between POPE/POPC in the outer leaflet remains ∼60:40.

  We emphasize that the desired lipid composition is not primarily guided by matching the biological lipid composition but by obtaining the desired properties of the liposomal membranes. If the lipids contained in a stable, fluid, in vivo membrane do not form a stable, fluid liposome on their own, it needs a tradeoff of matching lipid composition for desired liposome behavior. We have used non-native PC to stabilize the liposomes and to extend the available temperature range for storage and study of the aPLUVs down to room temperature because PC is a rather generic, zwitterionic lipid that should serve the purpose while being rather inert and inactive in other respects.

  As a control of the chosen lipid selection, we prepared aLUVs (step 3) and confirmed their asymmetry and its long-term stability (step 4; see first row of Table 1) without adding protein or detergent.

Table 1.

Mole Fractions of Outer-Layer POPG

	Lipid-Selection Control	Detergent-Removal Control	PLUVs
${{\rm X}}_{POPG}^{asym,out}$	19%	23%	24%
${{\rm X}}_{POPG}^{scramb,out}$	10%	11%	12%
α	0.91	1.1	0.98

For asymmetric ${{\rm X}}_{POPG}^{asym,out}$ and scrambled ${{\rm X}}_{POPG}^{scramb,out}$ versions of lipid-selection control, detergent-removal control, and proteoliposomes calculated from ζ calibration.

Asymmetry parameter α of 1 indicating perfect, outside-only asymmetric lipid distribution; 0 would stand for symmetric distribution.

• 2)In the second step, the acceptor vesicles are destabilized to facilitate the reconstitution of ST-NhaA. We used the detergent OG because it has a relatively high critical micelle concentration and is a standard detergent in proteoliposome preparation (41). ST-NhaA was produced by heterologous expression (23), purified (23), and reconstituted as described in the Materials and Methods.

  After reconstitution, the detergent was removed by BioBeads as established for symmetric proteoliposomes before (42). To check whether the extent of detergent removal also meets the higher requirements for aPLUVs, we subjected precursor liposomes (POPE-POPC-TOCL) to the same addition and removal of detergent procedures in the absence of protein. These “detergent-removal control” liposomes were turned into stable aLUVs (see Table 1, second row), thus qualifying the detergent-removal protocol as appropriate for aPLUV production.

  ST-NhaA concentration in symmetric and asymmetric liposomes was probed by densitometric quantification of SDS-PAGE-separated monomer and dimer bands using a dilution series of purified ST-NhaA as quantification standard (Fig. S4). The amount of ST-NhaA in PLUVs and aPLUVs remained at ∼80% for both compared to the amount added during reconstitution. The protein loss of ∼20% after reconstitution corresponded to a lipid loss of ∼20% as determined by the Bartlett assay. Thus, the lipid/protein ratio remained constant. In addition, there was no substantial protein loss during generation of asymmetry when comparing PLUVs and aPLUVs directly. ST-NhaA orientation in proteoliposomes was probed by trypsin digestion, which targets the most prominent trypsin cleavage site of Nha in loop VIII-IX, which is located on the periplasmic side of the protein (Fig. S5; (43)). ST-NhaA was reconstituted nearly equally in both orientations. In addition, its transport function is fully reversible (44).

• 3)For creating lipid asymmetry, we adjusted the outer leaflet by following the lipid-exchange protocol as explained in detail in Markones et al. (8) with donor cyclodextrin complexes. Slight adjustment to the optimal treatment of proteins, including less agitation to avoid shear stress and foaming, were made. The exchange aimed for 20 mol% negatively charged PG in the outer leaflet. 20 mol% PG in the outer leaflet is equivalent to 10 mol% PG of the total phospholipid content and approximates the composition reported for the natural S. Typhimurium membrane (21).

  This method is founded on the assumption that the lipid in the MβCD complexes equilibrates with the lipid in the outer leaflet of the acceptors and that PG, PE, and PC share roughly the same affinity to cyclodextrin. This assumption is discussed in Markones et al. (8) and is a basic assumption of all cyclodextrin-based lipid-exchange methods that are widely used (6).

• 4)For validation of asymmetry, we measured the ζ potential of the proteoliposomes as well as the lipid-selection and detergent-removal controls. The technique is sensitive to negatively charged PG localized in the outer leaflet only. A postexchange ζ value that agrees with the “asymmetry curve” (Figs. 2 A and S3, red dashed line) indicates a successful preparation of asymmetric aPLUVs. To be able to determine the fraction of successfully exchanged PG, we established a ζ-calibration curve with symmetrical LUVs with known composition of various fractions of PG (black curve, Figs. 2 A and S3), empirical fit yielding ζ = −52.6 mV ($X_{POPG}^{out}$x)³ + 139 mV ($X_{POPG}^{out}$)² − 120 mV ($X_{POPG}^{out}$) − 15.0 mV. With this, we calculated a second calibration curve of ideally asymmetric LUVs as described previously (8) by plotting the theoretical ζ values of $X_{POPG}^{outer}$, the fraction of PG in the outer leaflet, as a function of $X_{POPG}^{bulk}$, the fraction of PG in the exchange mixture (acceptor liposomes and MβCD-lipid complexes). The composition of the outer leaflet of postexchange LUVs can be extracted from the symmetric calibration curve (Fig. 2 A, blue arrows, Table 1, first row).

  Successful preparation of asymmetric liposomes and proteoliposomes can be confirmed with the asymmetry curve. Fig. 2 A shows data from a representative experiment of protein-containing and control liposomes: pre-exchange LUVs and PLUVs are represented by open circles and postexchange aLUVs and aPLUVs by solid circles. The exchange was successful, and the postexchange aLUVs must have an asymmetric, outside-only POPG distribution because the solid circles all lie on the “asymmetry curve.” That is true for aPLUVs, as well as for both controls. To further validate the asymmetry of postexchange aLUVs and aPLUVs, we scrambled purified aLUVs and aPLUVs by adding OG to destabilize the membrane and promote rapid flip-flop. The vesicles subsequently underwent the detergent removal, freeze-thaw, and extrusion treatment as described in the Materials and Methods to ensure a complete mixing of all lipids across both leaflets.

  The ζ potential of these re-extruded, scrambled liposomes corresponded to about half of the local outside density of PG before scrambling (see Table 1, second row), as expected if all PG had been outside before and equilibrated to be half-in, half-out after scrambling. Therefore, the ζ-values after scrambling lay exactly on the calibration curve for symmetric liposomes (Fig. 2 A, half-solid circles). To account for minor lipid losses, e.g., upon detergent removal during protein reconstitution, a Bartlett phosphorus assay was conducted before lipid exchange (for details, see Supporting Materials and Methods).

  Table 1 also lists the asymmetry parameter α:

  $$\alpha =\frac{{{\rm X}}_{POPG}^{\mathit{asym},out}-{{\rm X}}_{POPG}^{scramb,out}}{{{\rm X}}_{POPG}^{scramb,out}},$$

  where ${{\rm X}}_{POPG}^{asym,out}$ stands for the average mole fraction of POPG in the outer leaflet of the asymmetric liposomes and ${{\rm X}}_{POPG}^{scramb,out}$ in the outer leaflet of the scrambled liposomes corresponding to the fraction of POPG in the total bilayer of the asymmetric liposomes. In all cases, α amounts to +1 ± 0.1, which indicates outside-only localization of PG. According to this definition, α = 0 would denote a symmetrical membrane and −1 stand for inside-only localization.

  With the preparation of aPLUVs, the first objective was achieved. However, for such aPLUVs to be a suitable membrane model, the lipid asymmetry must be stable for several days to have sufficient time for the desired experiments on protein structure and function. To that end, we measured the ζ potential of aLUVs and aPLUVs over several days. The storage temperature was 23°C to ensure that the liposomes stayed above the transition temperature of the lipid mixture (Fig. S1) to minimize partial demixing and flip-flop. The results of three independent experiments are shown in Fig. 2 B.

  The ζ-potential readings remained constant within the error of the measurement for at least 7 days after exchange, showing that the asymmetric PG distribution was maintained for a sufficient amount of time.

  The ζ-potential readings remained constant within the error of the measurement for at least 7 days after exchange, showing that the asymmetric PG distribution was maintained for a sufficient amount of time.

  If generation or stability of asymmetry had not been successful, the controls (lipid-selection control blue, detergent-removal control yellow; Fig. 2 A) could provide valuable information about the reasons why it did not work. If the problem is the lipid composition itself, none of the three samples would have a stable lipid asymmetry, and their ζ would approach that of the symmetric curve rather rapidly. If the detergent removal was incomplete, the lipid-selection control would be asymmetric, but the detergent-removal control and aPLUVs would not have a stable asymmetry. The last case would be that the protein itself increases lipid flip-flop rates; then lipid-selection and detergent-removal controls would be asymmetric, but aPLUVs would not retain asymmetry.

  Additionally, the detergent-removal control can be used to check the influence of the protein on the ζ potential. Because aPLUVs and the detergent-removal control aLUVs show no difference in ζ, one can conclude that ST-NhaA has no detectable influence on the ζ potential.

Figure 2.

(A) Measurements of ζ potential during the exchange to quantify exchanged POPG. Averaged ζ potentials of precursor liposomes of POPE-POPC-TOCL (open circles), postexchange aLUVs (solid circles), and scrambled LUVs (half-solid circles) are shown. Error bars indicate the standard deviation of three measurements of the same sample. Symbol color denotes pre-exchange modifications. LUVs without further modification (control of lipid selection) are shown in blue, PLUVs in green, and LUVs for control of detergent removal in yellow. The black line shows ζ calibration established with symmetric LUVs of known composition. The red dashed line is a second, calculated calibration, in which perfectly asymmetrically exchanged aLUVs are expected. For details, see Markones et al. (8). aLUVs are on asymmetric calibration showing a successful asymmetric exchange, supported by the fact that scrambled LUVs go back to the symmetric curve. (B) Measurements of ζ potential during the storage at 23°C (above phase transition to avoid partial demixing and thus enhanced scrambling) to check for stability of asymmetry are shown. Averaged ζ potential of three independent experiments is shown: first (squares), second (diamonds), and third (circles; t₀ also depicted in A). Error bars indicate the standard deviation of three measurements of the same sample. Color specifies pre-exchange modifications. aLUVs without further modifications are depicted in blue, aPLUVs in green, and detergent removal control in yellow. Green lines were drawn to guide the eye, showing that ζ potential of aPLUVs does not change more than in the error of the measurement over a time of at least 8 days, showing that asymmetry remains stable long enough to perform experiments with asymmetric proteoliposomes. To see this figure in color, go online.

• 5)One last issue was that the protein might suffer from the treatment of PLUVs during the exchange. To rule out this concern and to ensure normal protein function after exchange, we performed a fluorescence-based protein activity assay (42). Probed was the electrogenic transport (exchange of one Na⁺ against 2 H⁺). Briefly, the membrane-potential-sensitive fluorophore DiSC₃(5) (45) was added to monitor electrogenic import of Na⁺ by ST-NhaA. In turn, the protons were exported, producing a membrane-potential change that was detectable by a decrease in fluorescence intensity (FI) (Fig. 3 A). The proton ionophore CCCP was added to collapse the generated membrane potential, resulting in a FI increase. In comparison, FI was unaffected by NaCl addition in the protein-free symmetric and asymmetric control vesicles. If a nonsubstrate ion was added (e.g., K⁺), the FI does not change, demonstrating that ST-NhaAs selectivity for small alkali cations is preserved in asymmetric liposomes (Fig. 3 B). ST-NhaA remained active for at least 14 days in an asymmetric lipid environment and when stored at room temperature. (Fig. 3 C).

Figure 3.

ST-NhaA is active, selective, and stable in aPLUVs as probed by DiSC₃(5) fluorescence intensity (FI) with excitation at 651 nm and emission at 668 nm. (A) ST-NhaA is active in asymmetric liposomes (dark green, aPLUVs), as well as in symmetric liposomes (light green, PLUVs), because FI decreases after addition of Na⁺. There is no change in FI upon Na⁺ addition for the symmetric (cyan, LUVs) and asymmetric (blue, aLUVs) extruded lipid-selection control and symmetric (yellow, LUVs det. removal) and asymmetric (orange, aLUVs det. removal) detergent-removal control. (B) Ion selectivity of ST-NhaA in aPLUVs is shown: Li⁺ (dark gray) and Na⁺ (light gray) are transported, K⁺ (black) does not reduce FI after addition and is therefore not transported. (C) ST-NhaA in aPLUVs remains stable after 14 days of storage above 23°C and remains active as compared to the first day. For all experiments, the proton ionophore CCCP was added after 100 s to collapse the generated membrane potential. To see this figure in color, go online.

Conclusion

We present the first, to our knowledge, aPLUVs containing a multispanning protein with an asymmetry being stable for several days. We believe that the five-step workflow proposed here can be adjusted to a variety of membrane proteins and their natural membrane environment. As described, the specific procedure has to be optimized for every protein and purpose to obtain a model that is relevant to address the question at hand and, at the same time, as simple and practicable as possible. The protocol introduced here shall open new avenues to finally study lipid asymmetry effects on detailed, functional parameters of membrane proteins.

Author Contributions

M.M., H.H., A.F., and C.H. conceived and designed the study. M.M. coordinated the project, prepared liposomes, and did asymmetrification, DSC, DLS, Bartlett, and ζ-analysis. A.F. purified and reconstituted protein, did NhaA activity assay, scrambled liposomes, probed orientation and concentration of reconstituted protein, and wrote respective parts of the manuscript. M.K. worked on stable, asymmetric, PE-rich liposomes for his MSc thesis, which was supervised by M.M. and H.H. C.H. and H.H. contributed to data analysis and writing of the manuscript. All authors discussed the results and commented on the manuscript. All authors have given approval to the final version of the manuscript.

Editor: Georg Pabst.

Supporting Citations

References (46, 47, 48, 49) appear in the Supporting Material.