Effect of a scramblase activator upon scrambling and membrane domain formation in HEK293T cells

Abstract

Lipid asymmetry, a difference between the lipid composition of the inner and outer leaflets (monolayers) of a membrane, is a characteristic of mammalian plasma membranes. In artificial lipid vesicles, asymmetry can either suppress or induce the formation of co-existing ordered and disordered lipid domains depending on lipid composition. In mammalian plasma membrane preparations, loss of asymmetry induces the formation of ordered domains. In this report, we studied the effect of a scramblase activator, the ionophore BrA23187 (BrA) plus Ca²⁺, upon ordered domain (lipid raft) formation in human embryonic kidney 293T cells. Addition of BrA induced a decrease in FRET between the plasma membrane outer leaflet probe TMADPH, which partly associates with ordered domains and ODRB, which localizes largely in liquid disordered domains. This is consistent with the formation of coexisting ordered and disordered domains in the plasma membrane. In addition, upon BrA addition, the plasma membrane outer leaflet probe Pro12A exhibited a decrease in generalized polarization (GP) suggesting a decrease in outer leaflet membrane order, perhaps due to a decrease in outer leaflet cholesterol. To test if BrA induced scrambling of fluorescent membrane probes, which would complicate interpretation of the experiments described above, we measured the effect of BrA upon extractability of outer leaflet probes with MβCD. (In most cases, MβCD was more effective for extraction than BSA.) These experiments showed that at most a small fraction of probes migrate to the inner leaflet upon addition of BrA. Other experiments raise the possibility that BrA binding to membranes may directly influence ordered domain formation and properties.

INTRODUCTION

Membrane lipid asymmetry refers to the difference in lipid composition between the inner and outer leaflets (lipid monolayers) of artificial or natural membranes containing a lipid bilayer. A distinct lipid composition is found in mammalian plasma membrane, with sphingolipids, including sphingomyelin (SM), and phosphatidylcholine (PC) enriched in the outer (exofacial) leaflet, and the anionic phospholipids phosphatidylserine (PS) and phosphatidylinositol (PI) plus the zwitterionic lipid phosphatidylethanolamine (PE) enriched in the inner (cytofacial, cytoplasmic facing) leaflet¹.

The lipids found in the inner and outer leaflet have different propensities to form an ordered state. In particular, outer leaflet enriched sphingolipids have a high propensity to form the liquid ordered (Lo) state when the membrane has a substantial concentration of cholesterol. In contrast, the inner leaflet is enriched in lipids with a higher degree of acyl chain unsaturation, which tend to form the liquid disordered (Ld) state². It has been found that, at least in the resting state, the lipids in the outer leaflet of the plasma membranes have a higher degree of order than those in the inner leaflet¹, and although cholesterol distribution between leaflets remains controversial, recent evidence suggests a higher cholesterol concentration in the outer leaflet under usual conditions controversial^3–8. For example, in a high-resolution structure of an aquaporin, containing bound lipids⁸, it has been observed that the outer leaflet lipids in contact with protein are both in a more ordered configuration (with more extended acyl chains) and are more cholesterol enriched than the inner leaflet lipids in contact with protein.

Studies in artificial and natural membranes have shown that asymmetry can greatly impact the formation of ordered lipid domains (lipid rafts)^9–11. This may be biologically important because a transient loss of lipid asymmetry can occur during signal transduction¹². In addition, changes in asymmetry can influence membrane protein conformation¹³. In artificial lipid membranes in which only one leaflet is composed of lipids that spontaneously form ordered domains, it has been observed that, depending upon lipid composition, lipid asymmetry can either induce the formation of ordered domains in a leaflet formed of lipids that do not spontaneously form ordered domains^14–18, or suppress ordered domain formation in a leaflet formed of lipids that would spontaneously form ordered domains in a symmetric vesicle^{9, 17–19}. The latter behavior has been observed in both lipid vesicles with a composition mimicking that of plasma membranes⁹, and plasma membrane vesicles from rat basophilic leukemia 2H3 (RBL-2H3) cells¹⁰.

In this report, we studied the effect of altering the living cell plasma membrane lipid asymmetry upon membrane organization by adding a lipid scramblase activator to cells. Using plasma membrane outer leaflet probes, changes indicative of domain formation and altered membrane order were observed. Explanations for the origin of these changes are discussed.

EXPERIMENTAL

Materials

Lipids:

1-oleoyl-2-(6-[(7-nitro-2-1,3-benzoxadiazol-4-yl) amino] hexanoyl)-sn-glycero-3- phosphocholine (6NBD-PC) from Avanti Polar Lipids (Alabaster, AL) was stored at 1mM in DMSO. Brain sphingomyelin (SM), 1-palmitoyl-2-oleoyl phosphatidylcholine (POPC), dioleoylphosphatidylcholine (DOPC), and cholesterol were from Avanti Polar Lipids. All lipids were stored at −20°C

Probes:

1-(4-Trimethylammoniumphenyl)-6-phenyl-1,3,5-hexatriene p-toluenesulfonate (TMADPH) was purchased from the Molecular Probes division of Invitrogen (Carlsbad, CA), and stored at 24 μM in ethanol. 3-(N-dodecyl-2-(methyl(6-propionylnaphthalen-2-yl) amino) acetamido) propane-1-sulfonate (Pro12A) was from Cytoskeleton, Inc (Denver, CO), and stored at 20 μM in DMSO). 6-dodecanoyl-2-dimethylaminonaphthalene (Laurdan) was purchased from Invitrogen (Carlsbad, CA), and stored at 32 μM in ethanol. Octadecylrhodamine B (R18, ODRB) was from Invitrogen (Carlsbad, CA), and stored at 100 μM in ethanol or at 8.6 mM in ethanol for cell experiments. All probes were stored at −20°C.

Reagents:

Alexa Fluor 488 labeled Annexin V (Annexin V-F488) was from Invitrogen and stored at 4°C. BrA (4-bromo-A23187, 4-bromo-calcimycin) was from Millipore Sigma (Burlington MA) and MedChemExpress (Monmouth Junction, NJ). BrA was dissolved at 2 mM in ethanol or DMSO. No difference in behavior between experiments was noted depending on solvent or manufacturer. Stored at −20°C. Fluo-4 AM in DMSO was from Invitrogen. Methyl-β-cyclodextrin (MβCD) was from Millipore Sigma (Burlington MA). Bovine Serum Albumin (BSA), Fatty Acid free was from Chem Cruz (Dallas TX). Dulbecco’s PBS (DPBS) (200 mg/l KCl, 200 mg/l KH₂PO₄, 8 g/l NaCl, and 2.16 g/l Na₂HPO₄), was from ThermoFisher Scientific (Waltham, MA). All other chemicals were reagent grade.

Cell lines and cell culture

Human embryonic kidney (HEK) 293T cells (ATCC CRL-11268) were maintained in DMEM (Gibco), 10% FBS, 1% Pen Strep (Gibco) at 37 °C, and 5% CO₂, sub-cultured every two days, and used between passages 4–25.

Measurement of time dependence of FRET in living cells before and after BrA treatment

Cells were detached from plates with DPBS. Two samples were prepared each with 1×10⁶ cells in 1 ml DPBS. We estimate that each of these samples would roughly have 50–100 μM phospholipid (data not shown). For the “F sample” the FRET acceptor ODRB was added to a concentration of 12 μM, while the “Fo sample” lacked ODRB. Both samples were incubated at 37 °C for 10 min and cells pelleted (200 × g /3 min) to remove unbound ODRB. The cell pellets were resuspended in 1 ml of ABB (Annexin-binding buffer: 10 mM HEPES, 140 mM NaCl, and 2.5 mM CaCl₂), and then FRET donor TMADPH was added to both F and Fo samples at a final concentration was 0.07 μM, incubated at RT for 10 min in the dark before fluorescence was measured. TMADPH intensity was read every 2 min for up to 10 min at RT (~23°C) and then BrA was added to both F and Fo samples to the desired final concentration (maximum 8 μM). Readings were made every 2 min (up to 10 min). TMADPH fluorescence intensity was measured using a Horiba QuantaMaster Spectrofluorometer (Horiba Scientific, Edison, NJ) at Ex 358 nm, Em at 427 and bandpass set 3 nm for both excitation and emission. The fluorescence intensity of background samples (lacking TMADPH but containing equivalent organic solvent) was negligible (less than 1%). Then the fraction of fluorescence unquenched by FRET, F/Fo, was calculated. (FRET efficiency is 1 − F/Fo).

Preparation of large unilamellar vesicles

Large unilamellar vesicles (LUV) composed of various combinations of bSM, POPC, DOPC and cholesterol were prepared. Lipids initially dissolved in chloroform were mixed, dried by a stream of nitrogen and then subject to high vacuum for 1h. Multilamellar vesicles (MLV) were prepared by adding 1 ml of DPBS and incubated at 70°C for 5 min. LUVs were prepared from MLVs with 7 freeze-thaw cycles. To avoid losses of lipid during filtration, the samples were not extruded to achieve uniform size. Final lipid concentration was 1mM.

Measurement of the temperature dependence of FRET in lipid vesicles

FRET measurements on vesicles were carried out as previously described¹⁰. Briefly, vesicles contained 50 μM lipid dispersed in 1 ml of ABB. TMADPH was the FRET donor and ODRB as the FRET acceptor. First, the F sample containing ODRB (0.4 μM) and Fo sample lacking ODRB were incubated at 37°C for 1 in dark. Next TMADPH was added to both F and Fo samples (final concentration 0.07 μM) and then incubated at room temperature for 10 min in dark. TMADPH fluorescence intensity was measured as above. The data was collected at every 4°C while increasing temperature from 16 to 64°C. Between readings, temperature was increased at a rate of 4°C /6 min. The fraction of fluorescence not quenched by FRET is equal to F/Fo. (FRET efficiency is 1 − F/Fo).

Determining exposure PS on the cell surface by spectrofluorimetry and fluorescence microscopy

To evaluate exposure of PS on the outer leaflet of the plasma membrane the fluorescence of conjugated annexin V (Annexin V-F488) bound to cell surface PS was measured. 4 × 10⁶ cells were suspended in 4 ml of DPBS at 37°C for 10 min and centrifuged. The cell pellets were then suspended with 4 ml of ABB and 1 ml aliquots added to four 1.5-ml tubes. BrA was added to each cell suspension at a concentration of 0, 2 ,4 and 8 μM and then incubated at RT for 6 min. Next, 0.2 ml from each sample was transferred to 1 ml cold ABB and the cells were pelleted. Annexin V-F488 was diluted twenty-fold (according to manufacturer’s directions) with ABB. Then 50 μl was added and incubated at RT for 5 min in dark. After washing two times with ABB, cells were suspended 1 ml ABB. Cell surface bound Annexin V-F488 fluorescence intensity was measured with Ex 488 and Em 510; and bandpass set 5 nm for both excitation and emission.

For imaging living cell PS exposure, cells were seeded on 5 μg/ml poly-D-lysine coated Glass Bottom Culture Dishes (MatTek corporation). The next day, cells were washed with ABB and stimulated with 8 μM of BrA at RT for 10 min. After washing, 200 μl of Annexin V-F488 containing in ABB solution was added and incubated at RT for 10min in the dark. After removing the Annexin V solution, 1 ml ABB was added. Samples were imaged with a Zeiss LSM 980 confocal microscope. Annexin V-F488 was excited with 488 nm and emission collected between 503 and 700 nm.

Estimated internalization of 6NBD-PC by extraction of outer leaflet 6NBD-PC and dithionite treatment

Flip of 6NBD-PC added to cells across the plasma membrane was measured by accessibility to extraction by externally added BSA. To do this 4×10⁶ cells were suspended in 4 ml of DPBS at 37°C for 10 min and then pelleted. The cell pellets were resuspended with 4 ml of ABB containing 0.5 μM of 6NBD-PC and incubated at RT for 10 min. 1 ml aliquots were transferred to four 1.5 ml tubes. BrA was added to each cell sample at 0, 2 ,4 and 8 μM and then incubated at RT for 6 min. Then 0.2 ml from each sample was added to 0.2 ml of prechilled 5 mg of fatty acid-free-BSA in ABB and incubated on ice for 10 min to extract outer leaflet 6NBD-PC. After washing two times with 1ml ice-cold ABB and then adding 1 ml of RT ABB for 6NBD-PC fluorescence was measured at Ex 464 and Em 531, with bandpass set 5 nm for both excitation and emission. Background values for samples without 6NBD-PC were low and not subtracted from the values with 6NBD-PC.

Flip of 6NBD-PC was confirmed by the loss of exposure of 6NBD-PC to reduction by sodium dithionite. To do this, 1 ml samples of BrA treated cells with 6NBD-PC, with or without BSA extraction were used, as described above. Samples were then mixed with 25 μl of 1 M sodium dithionite in 1 M Tris-HCl buffer pH10. 6NBD-PC fluorescence intensity was recorded before dithionite addition and at 2 min intervals after dithionite addition. Dithionite abolishes NBD fluorescence by reduction of NBD groups.

Laurdan and Pro12A GP measurements in cells and lipid vesicles

2×10⁶ cells in 2ml of DPBS were incubated at 37°C for 10 min and then pelleted. The cell pellets were resuspended to 2 ml of ABB for two samples. LUV with various lipid compositions were prepared as described above and then diluted to 100 μM lipid with ABB. In one cell or vesicle sample BrA was added at 8 μM and in the other (control) an equivalent amount of solvent was added. After incubation in cuvettes at RT for 6 min fluorescence backgrounds were measured. Then the fluorescence probe was added (0.08 μM for Laurdan or 0.05 μM for Pro12A), incubated at 23°C for 10 min, and fluorescence measured using Ex 395 nm and Em 440 and 490 nm for both probes. Bandpass was set 5 nm for both excitation and emission). GP was calculated using the formula. GP = (F₄₄₀ - F₄₉₀)/ (F₄₄₀ + F₄₉₀). Background values from samples lacking Laurdan or Pro12A were subtracted.

BSA extraction of fluorescent probes in living cells

From 1.5 × 10⁶ cells in 1.5 ml DPBS three 0.5 ml samples were prepared. One was background (cells only) and to the other two were added 0.07 μM TMADPH. After incubation at RT for 10 min in dark, the cells were then pelleted, and 1 ml of ABB was added to each cell pellet. To the cells with fluorescent probe either 0.5 ml of 10 mg/ml BSA in ABB or 0.5 ml ABB was added, and then all samples incubated at RT for 20 min. The cells were then pelleted, and 1 ml ABB was added. TMADPH fluorescence intensity measurements used Ex 358 nm, Em at 427, and bandpass set 3 nm for both excitation and emission. Background fluorescence from sample lacking fluorescence probe was subtracted from reported fluorescence.

For other probes the same experimental procedure was followed. The concentration of probes and fluorescence measurements were: Pro12A 0.07 μM, Ex 395 nm, Em at 440 and 490 nm, bandpass 5 nm; 6NBD-PC 0.4 μM, Ex 464 nm, Em 531 nm, bandpass 5 nm; ODRB 12 μM, Ex 556 nm, Em at 578 nm, bandpass 3 nm.

For ODRB samples, cells were incubated at 37°C for 10 min and then pelleted to remove unbound ODRB. Then the cells were resuspended to 1 ml of ABB and to abolish any ODRB self-quenching 0.1 ml of Triton X-100 (w/v) was added before measuring fluorescence.

MβCD extraction of fluorescent probes in living cells with or without BrA treatment

From 2.5 × 10⁶ cells in 2.5 ml DPBS five 0.5 ml samples were prepared. Two were backgrounds (cells only and with BrA) and to the other three were added 0.07 μM TMADPH. After incubation at RT for 10 min in the dark BrA was added to 8 μM to one sample containing fluorescent probe and one background sample, and incubated at RT for 6 min in the dark. Next, 55 μl of 100 mM MβCD in PBS was added to one sample with fluorescent probe only and one sample with fluorescent probe and BrA. All samples were then incubated at RT for 20 min. Finally, cells were pelleted, washed once with 1 ml ABB, and suspended in 1 ml ABB for fluorescence measurements. Background fluorescence from samples lacking fluorescence probe, with or without BrA, was subtracted from reported values. The same protocol was used for the other probes, with concentrations and fluorescence measurements as above.

Fluorescence microscopy images of probe localization in living cells

All imaging was carried out at room temperature (~23 °C). Cells were seeded on Glass Bottom Culture Dishes (MatTek corporation) a day before imaging. Next day the cells were washed once with DPBS and were incubated with TMADPH (3 μM), ODRB (3 μM), Pro12A (1μM), or Laurdan (20 μM) for 5 min at RT in the dark. After labeling, the cells were washed once with ABB, and imaging was performed after 20–30 min incubation at RT. Cells were imaged using a Zeiss LSM 980 confocal microscope. For TMADPH, Pro12A and Laurdan, excitation was 405 nm and emission collected between 403 and 621 nm. For ODRB excitation was at 568 nm and emission collected between 560 and 747 nm.

Staining cytosolic Ca²⁺ with Fluo-4 AM

Cells were plated on poly-D-lysine coated 48-well plates. The next day, after washing cell with DPBS three times, Fluo-4 AM in DMSO was diluted 1000-fold into 0.2 ml ABB or ABB without Ca²⁺. Samples were incubated at RT for 10 min and then 8 μM BrA, or an equivalent amount of solvent, was added and incubated at RT for another 10 min. The images were taken after one wash with ABB with or without Ca²⁺ using an Olympus IX 83 with a GFP filter cube.

Measurement of ODRB binding to cells

To 2×10⁶ cells in 2ml DPBS, ODRB was added to 12 μM then incubated at 37°C for 10 min, divided into two equal samples, and then cells pelleted. To one sample 1ml ABB was added. To the other 1ml ABB without Ca²⁺ but with 0.4mM EGTA was added. Next BrA was added to a concentration of 8μM. Samples were incubated 6 min at RT. Then the cells were pelleted, and the supernatant removed carefully. The cell pellets were suspended in 1 ml ABB with Ca²⁺ or without Ca²⁺ but with 0.4mM EGTA. ODRB fluorescence was measured after taking 400 μl of supernatant or 400 μl of the cell suspension, adding 500 μl of ABB or ABB without Ca²⁺, respectively, and 100 μl of 10% (w/v) Triton X-100. ODRB fluorescence was measured as described above.

RESULTS AND DISCUSSION

BrA-induced loss of membrane asymmetry in HEK 293T cells.

To induce loss of plasma membrane lipid asymmetry, HEK 293T cells were incubated with Ca²⁺ and an ionophore that allows Ca²⁺ entry into the cytoplasm. High cytosolic Ca²⁺ concentration activates TMEM16F²⁰ and other lipid scramblases. We chose HEK 293T cells because they exhibit strong scramblase activity following the treatment with an ionophore plus Ca^{2+ 21}. To test for the loss of membrane asymmetry, we adopted widely used fluorescence-based lipid scrambling assays. The first was detecting scrambling-induced PS exposure on the outer leaflet of the plasma membrane, because in the absence of scrambling PS is a cytosolic leaflet lipid. To do this the amount of cell surface bound fluorescently conjugated Annexin V, a PS binding protein, was measured. The second assay was monitoring the decrease of 6NBD-PC localization in the plasma membrane outer leaflet. In all of these studies, we used the non-fluorescent ionophore BrA (4 -bromo-A23187) to avoid interference from the fluorescence if ordinary A23187 was used, and ABB, a Ca²⁺ containing Annexin-binding buffer (see Methods).

We confirmed that addition of Ca²⁺ and BrA increased cytosolic Ca²⁺ levels using a Ca²⁺ indicator Fluo-4 AM. Supplemental Figure 1 shows a dramatic increase in cytosolic Ca²⁺ levels when cells were incubated with both the Ca²⁺ and BrA, which should activate scramblase.

To assay scrambling of PS, cells were incubated with different concentrations of BrA and then the binding of PS to Annexin V-F488 was measured. Figure 1A shows that there was a BrA dose-dependent increase in outer leaflet PS. As shown by confocal microscopy images (Figure 1B) a significant population of cells showed PS exposure on the cell surface. This observation is consistent with the A23187 activation of NK cells and mouse Ba/F3 cells after which a significant subpopulation of cells exhibited PS exposure^{22, 23}.

Figure 1. BrA-induced PS exposure in cells.

A. HEK 293T cells were incubated with BrA at the indicated concentrations and externalized PS assayed by the binding of Annexin V-F488. Graph shows mean values of Annexin fluorescence and standard deviations for n = 3. Value in samples without BrA was normalized to 1. B. Microscope image showing of cells exhibiting Annexin V-F488 binding after treatment with 8 μM BrA.

To monitor scrambling of outer leaflet PC, we measured the decrease in extractability of the fluorescent lipid 6NBD-PC by BSA and MβCD added to the cells²⁴. BSA has a high (but not 100%) extraction efficiency. PC is normally enriched in the plasma membrane outer leaflet, and 6NBD-PC added to cells inserts in the outer leaflet. Movement to the inner leaflet prevents extraction. We found that the amount of cell-associated 6NBD-PC protected from BSA or MβCD extraction increased when BrA was added to the cells (Figure 2A and Table 1). This indicates that a substantial fraction of 6NBD-PC translocated across the membrane in a BrA does-dependent fashion.

Figure 2. Assays of BrA induced scrambling of 6NBD-PC added to cells.

A. Effect of BrA upon protection of 6NBD-PC from extraction by BSA. Y-axis shows the amount of 6NBD-PC fluorescence unextracted by BSA in arbitrary units. B. and C. Protection from dithionite after the addition of BrA without pre-extraction of 6NBD-PC (B), or after pre-extraction by BSA (C). Fraction of the initial 6NBD-PC fluorescence before dithionite that is unreduced 2 min after addition of 25 mM sodium dithionite is shown. Samples contained 0.5 μM 6NBD-PC added to HEK 293T cells suspended in ABB. Mean values (n=3) and standard deviations are shown.

Table 1:

Extraction of membrane probes by BSA and MβCD at RT¹. Mean values for the fraction probe molecules that were unextracted and standard deviations for n = 3 are shown.

	BSA−BrA	MβCD−BrA	MβCD + BrA2
6NBD-PC3	0.182 ± 0.028	0.163 ± 0.022	0.327 ± 0.026
TMADPH	0.576 ± 0.125	0.300 ± 0.047	0.333 ± 0.026
ODRB	0.946 ± 0.158	0.482 ± 0.082	0.499 ± 0.037
Pro12A4	0.235 ± 0.081	0.092 ± 0.012	0.192 ± 0.029

¹There was less extraction of probes when MβCD extraction was carried out on samples on ice (data not shown).

²BrA 8 μM

³Extraction of 6NBD-PC and the other fluorescent probes were carried out at RT for these experiments.

⁴Emission measured at 440nm. Values were similar when emission was measured at 490nm.

To confirm the decrease in outer leaflet 6NBD-PC, accessibility of 6NBD-PC to reduction by externally added sodium dithionite was measured²⁵. Dithionite reduces the NBD nitro group and thereby abolishes NBD fluorescence. Outer leaflet NBD lipids are accessible to rapid reduction by dithionite. Figures 2B and 2C and Supplemental Figure 2 show that minutes after dithionite addition, there was almost complete reduction of the NBD group in the absence of BrA, even if samples were pretreated with BSA to remove some of the outer leaflet 6NBD-PC.

However, in the presence of BrA there was substantial protection of 6NBD-PC from dithionite. Protection was further increased after BSA extraction of some of the outer leaflet 6NBD-PC, which would increase the fraction of NBD in the inner leaflet (compare Figure 2B and 2C). This confirms that some of the 6NBD-PC migrated out of the outer leaflet upon BrA addition, and that the extraction step indeed removes outer leaflet 6NBD-PC. It should be noted that dithionite can cross membranes to leak into cells. Extrapolation to time zero after the addition of BrA indicated that the initial protection from dithionite is even greater than shown in Figure 2B and 2C (see Supplemental Figure 2).

BrA induced decrease in FRET between plasma membrane outer leaflet probes TMADPH and ODRB

The effect of BrA treatment upon membrane domain formation was investigated using Förster resonance energy transfer (FRET). TMADPH was used as the FRET donor and ODRB as FRET acceptor. Both are cationic molecules that associate with the outer leaflet of plasma membranes when added to cells²⁶, although ODRB, in which the positive charge density is low, can cross membranes²⁷. Microscopy confirmed their localization in plasma membranes (Figure 3). Prior studies have shown that upon formation of co-existing ordered and disordered domains in membranes there is a segregation of some TMADPH molecules into the ordered domains, while the ODRB very largely remains in disordered domains. This results in a decrease in FRET which can be detected by the increase in the ratio of donor fluorescence (F) relative to donor fluorescence in the absence of acceptor (Fo), i.e. an increase in F/Fo¹⁰. Figure 4 shows that when BrA was added to the HEK293T cells there was a rapid decrease in FRET (increase in F/Fo). The increase was BrA dose dependent (Figure 4A). A similar, but more rapid increase in F/Fo (and larger at 4 μM) was observed when BrA was added to cells at 37°C (Figure 4B). The increase in F/Fo was blocked when BrA was added to cells incubated in ABB without Ca²⁺ but with EGTA (Figure 4C).

Figure 3.

Confocal microscopy images of fluorescent probe localization in HEK 293T cells at RT. Images taken 20–30 min after washing out excess probes. Laurdan image in psuedocolor.

Figure 4.

Effect of BrA on FRET between TMADPH (donor) and ODRB (acceptor) incorporated into HEK 293T cells in ABB. F/Fo is the ratio of (F) TMADPH fluorescence in a sample with ODRB to (Fo) fluorescence in a sample lacking ODRB. (A) Increase in F/Fo at room temperature after addition of different BrA concentrations. Time of addition of BrA defined as t=0. (B) Increase in F/Fo at 37°C after addition of BrA. (C) Increase in F/Fo after addition of BrA with Ca²⁺ or without Ca²⁺ but plus 0.4 mM EGTA present. BrA was added immediately after t = 0. Mean values (n=3) and standard deviations are shown. See Supplemental Table 1.

The change in FRET is consistent with the formation of ordered domains upon addition of BrA. However, we were concerned that there might be other variables that could affect F/Fo. For example, if BrA in the presence of Ca²⁺ reduced the binding of ODRB to membranes this would increase F/Fo specifically in such samples. However, ODRB binding experiments did not detect an increase in the amount or fraction of ODRB in the supernatant when cells incubated with ODRB were pelleted in the presence of BrA and Ca²⁺ relative to the amount or fraction of ODRB in the supernatant for samples incubated in the presence of BrA without Ca²⁺ but with 0.4 mM EGTA (Supplemental Figure 3). Other parameters that could also affect FRET (see below).

Effect of BrA addition upon membrane order detected by Pro12A

The effect of BrA addition upon membrane order was also measured. The probes used were Laurdan, which resides largely, but not fully, in the plasma membrane²⁸ (and see Figure 3), and the anionic Laurdan derivative Pro12A, which does not cross membranes²⁸ and so localizes in the plasma membrane outer leaflet when added to cells. Generalized polarization (GP), a measure of the relative florescence emission intensity at short and long wavelengths for these probes²⁹, was used to assay membrane order. GP values are high in an ordered membrane environment.

Experiments with artificial vesicles confirmed that both Laurdan and Pro12A GP are very sensitive to both ordered state formation and cholesterol content. Vesicles were composed of POPC, which should be in the disordered Ld state at RT, brain sphingomyelin (SM), which should be in an ordered state (gel without cholesterol, liquid ordered (Lo) with cholesterol), and SM/POPC, which should have a combination of disordered (Ld) and ordered domains, were prepared, all with and without cholesterol. Table 2 shows Pro12A and Laurdan GP show similar, but not identical, sensitivities to membrane composition. For both probes, GP, and thus membrane order, increased with increasing formation of ordered state. GP measured by Pro12A and Laurdan in SM/POPC/cholesterol vesicles, which form co-existing Ld and Lo domains³⁰, was intermediate between GP in POPC/cholesterol vesicles, and SM/cholesterol vesicles. The intermediate GP values suggest that both Laurdan and Pro12A localize within both Lo and Ld domains and do so to a similar extent. In addition, GP was higher in samples with cholesterol relative to those without cholesterol, even for POPC in which the membrane should be in the Ld state whether cholesterol was present or not. This is not unexpected, because even in the Ld state, addition of cholesterol increases membrane order³¹. The addition of BrA to these vesicles using the same BrA concentration as used in cells induced a small increase in GP in most cases. This shows that BrA has an effect on membrane order that is independent of activating a lipid scramblase.

Table 2:

Effect of vesicle lipid composition on GP values with or without 8 μM BrA. Mean values for the fraction probe molecules that were unextracted and standard deviations for n = 3 are shown.

	−BrA	+BrA		−BrA	+BrA
Laurdan			Pro12A
POPC	−0.049 ± 0.091	0.012 ± 0.046	POPC	−0.064 ± 0.010	−0.062 ± 0.017
1:1 SM/POPC	0.179 ± 0.033	0.269 ± 0.026	1:1 SM/POPC	0.094 ± 0.007	0.063 ± 0.019
SM	0.302 ± 0.016	0.487 ± 0.043	SM	0.136 ± 0.032	0.093 ± 0.014
2:1 POPC/chol	0.391 ± 0.053	0.435 ± 0.054	2:1 POPC/chol	0.358 ± 0.027	0.418 ± 0.047
1:1:1 SM/POPC/chol	0.511 ± 0.019	0.551 ± 0.001	1:1:1 SM/POPC/chol	0.530 ± 0.019	0.570 ± 0.040
2:1 SM/chol	0.650 ± 0.034	0.650 ± 0.022	2:1 SM/chol	0.682 ± 0.032	0.687 ± 0.037

Table 3 shows the level of membrane order in HEK293T cell membranes with and without BrA treatment. In the absence of BrA. The GP of both Laurdan and Pro12A was similar to that in cholesterol-containing artificial vesicles with co-existing ordered and disordered domains (i.e., SM/POPC/cholesterol). The presence of BrA slightly decreased Pro12 A GP, but slightly increased Laurdan GP. The difference in the behavior of these probes is probably a reflection of the difference in their membrane locations. Pro12A GP should reflect membrane order only in the plasma membrane outer leaflet, while Laurdan would probe the average order in both leaflets. The behavior of Pro12A is thus consistent with a loss of cholesterol and/or a decrease in average order in the outer leaflet under conditions in which overall membrane order is maintained (see below).

Table 3:

Effect of BrA on GP values in cells. Mean values for the fraction of probe molecules that were unextracted and standard deviations for n = 3 are shown.

	−BrA	+BrA1
Laurdan	0.489 ± 0.004	0.542 ± 0.005
Pro12A	0.482 ± 0.028	0.441 ± 0.027

¹BrA 8 μM

Effects of BrA upon extractability/localization of membrane probes

Interpretation of FRET and GP data for the experiments described above could be complicated by changes in probe localization when BrA is added to cells. Therefore, we considered the possibility that BrA activation of a scramblase would alter the fraction of fluorescent probes in the inner and outer leaflet. To investigate the leaflet localization of probes, their accessibility to extraction was assayed. For these experiments a suitable molecule for extracting probes from the outer leaflet had to be identified. To do this, we compared extraction by BSA and by MβCD. Except for 6NBD-PC, BSA did not extract membrane probes from cells as well as did MβCD (Table 1). Therefore, MβCD was used in subsequent experiments.

Table 1 shows 10 mM MβCD extracted roughly 70% of membrane bound TMADPH and 90% of Pro 12A as judged from the decrease in cell bound fluorescence. Only about 50% of ODRB added to the outer leaflet was extracted by MβCD. This may mean ODRB locates equally in both the inner and outer leaflets. Indeed, studies have shown that ODRB is able to cross membranes under some conditions^{27, 32}. Concluding that ODRB equally localizes in the outer and inner plasma membrane leaflet is valid if ODRB does not flip across the membrane quick enough for inner leaflet ODRB to be extracted by MβCD. Alternately, the low level of ODRB extraction could result from strong ODRB binding to membranes relative to its binding to MβCD. Residual TMADPH and Pro12A in membranes after MβCD extraction could also reflect incomplete extraction of outer leaflet probes.

Unless BrA induced flip-flop of molecules is extremely rapid, if there was BrA-induced scrambling that moved probes into the inner leaflet, their extraction from the outer leaflet would decrease. This was observed for 6NBD-PC. However, there was only a very small, if any, decrease in the accessibility of TMADPH or ODRB to extraction after incubation with BrA (Table 1). This suggests that these probes are not significantly scrambled after addition of BrA, i.e., not subject to flip-flop due to BrA activation of scramblase. Pro12A did show a somewhat larger decrease in accessibility to extraction, which might reflect scramblase-induced Pro12A flip to inner leaflet, or a direct effect of membrane bound BrA such that it decreases the ability of MβCD to extract Pro12A from membranes.

Direct effects of BrA upon membrane domain formation

The decrease in FRET (increase in F/Fo) upon BrA addition to cells could reflect direct induction of ordered domains formation resulting from the binding of BrA to the plasma membrane. To test if this was possible, we compared membrane domain formation in 1:1:1 SM/DOPC/cholesterol vesicles and 2:1 DOPC/cholesterol vesicles with and without BrA. Figure 5 shows the temperature dependence of FRET for these samples. The F/Fo curves for DOPC/cholesterol, which do not form ordered domains in the temperature range studied, show a baseline for samples without ordered domain formation (Figure 5A and 5B). The F/Fo curves for the SM/DOPC/cholesterol vesicles show higher F/Fo at low temperature than those for DOPC/cholesterol due to ordered domain formation³³. At higher temperature these ordered domains melt, abolishing domain formation such that homogeneous bilayers form, and F/Fo values become similar to those in the samples DOPC/cholesterol vesicles which lacking ordered domains³⁴. It appears that the ordered domains are more thermally stable with BrA (Figure 5A and 5B). However, interpreting the effect of BrA on domain formation depends what baseline is chosen for FRET. If we assume that the baseline without domains is represented by the slope of the F/Fo curves above the temperature at which domains normally form (about 45°C³⁴) (Figure 5C and 5D), then there appears to be little if any effect of BrA on domain stability. There is also very little effect Ca²⁺ on FRET with or without BrA, indicating it has little effect on domain formation (Supplemental Figure 4).

Figure 5.

Effect of BrA upon FRET between TMADPH and ODRB in lipid vesicles. F/Fo vs temperature for: 50 μM lipid vesicles composed of 1:1:1 bSM/DOPC/cholesterol or 2:1 DOPC/cholesterol for samples: A and C. in the absence of BrA, or B and D. in the presence of 8 μM BrA. A. and B., show superimposed curves for the two lipid compositions. C. and D, F/Fo for 1:1:1 bSM/DOPC/cholesterol vesicles with and without BrA superimposed over baselines estimated from F/Fo vs temperature trends at high temperature. Vesicles were suspended in ABB. Mean values (n=3) and standard deviation are shown.

Effect of loss of lipid asymmetry upon domain formation in HEK 293T cells.

Prior studies support the conclusion that under basal conditions in the mammalian plasma membrane outer leaflet is in a Lo-like state, while the inner leaflet is more Ld-like^{1, 35}. In addition, under basal conditions the outer leaflet of the plasma membrane is enriched in sphingolipids, and although still controversial^3–7may be enriched in cholesterol. The extent to which laterally segregated ordered and disordered domains co-exist in individual leaflets of the plasma membrane under basal conditions is unclear, and may depend on cell type and experimental conditions. As in the case of asymmetric lipid vesicles with a SM/POPC/cholesterol outer leaflet and POPC/cholesterol inner leaflet, there might be no or little lateral domain formation in cells under many conditions. As noted in the Introduction, in vitro the loss of asymmetry in SM/POPC/cholesterol vesicles, which somewhat mimic the plasma membrane, induces lateral separation into Lo and Ld domains⁹. In addition, studies in plasma membrane vesicles from mammalian cells have also shown that loss of lipid asymmetry strongly promotes the formation of ordered lipid domains¹⁰.

The aim of this study was to explore whether loss of asymmetry in intact cells can induce lateral lipid domain formation in the plasma membrane of intact cells. In particular, we wished to determine if scramblase-activating conditions, i.e. incubation with an ionophore in the presence of extracellular Ca²⁺, would induce lateral segregation of lipids into co-existing Lo and Ld domains. Under these conditions, a decrease in FRET consistent with a segregation of a population of TMADPH molecules from ODRB molecules, and thus lateral domain formation, was observed. This could reflect an increase in the formation and/or size of co-existing Lo and Ld domains in the plasma membrane upon loss of membrane lipid asymmetry^{34, 36}.

GP values for Pro12A suggest the loss of asymmetry is also accompanied by a decrease in average outer leaflet membrane order which could reflect a decrease in outer leaflet cholesterol, without a significant change in overall membrane order, as shown by Laurdan GP. It should be noted that in addition to random loss of asymmetry, domain formation could be impacted changes in the outer leaflet PC/SM ratio upon activation of scrambling, as has been observed in the presence of active TMEM63B³⁷. A model summarizing the hypothesis for the changes that occur under scramblase activating conditions is schematically illustrated in Figure 6.

Figure 6.

Schematic illustration of possible changes in membrane organization upon addition of BrA. Only a few lipids are shown. The space between the solid lines represents a lipid bilayer. The dashed lines indicate boundaries between different physical states. D and A are FRET donor and acceptor, respectively. Lo and Lo’ and Ld and Ld’ are used to indicate the Lo and Ld states do not have to be identical in properties before and after scrambling. Note: FRET acceptor level in inner leaflet is not shown. An alternate model is that the FRET acceptor is distributed in both leaflets before and after BrA addition.

Potential biological implications

A reversible loss of lipid asymmetry may be associated with those types of signal transduction that induce Ca²⁺ influx into the cytoplasm and thus might activate scramblases^{12, 38, 39}. In processes in which ordered domain (lipid raft) formation is triggered by and/or promotes clustering and activation of raft-associating receptors and associated proteins, such as in immune cells^40–42, activation-triggered Ca²⁺ influx could induce a loss of asymmetry and thus initiate a positive feedback loop that increases the level of raft formation and activation.

Factors that complicate interpretation of FRET and membrane order experiments

Two factors complicate interpretation of the experiments in these studies. One is the possibility that activation of scrambling results in scrambling of the fluorescent probes used. The rationale for such scrambling is that scramblase action may arise partially from membrane thinning⁴³. Thinning reduces the size of the energy barrier for flipping of membrane molecules between leaflets and could potentially induce flipping of fluorescent probes as well as lipids. If flipping of probes occurred, the changes in FRET and membrane order that fluorescent probes detect upon BrA addition could arise from BrA induced probe scrambling. For example, scrambling of TMADPH and/or ODRB between the inner and outer leaflets of the plasma membrane would decrease FRET by increasing the average distance between donor and acceptor. In addition, probe scrambling could decrease Pro12A GP by moving Pro12A from a high order outer leaflet to the lower order inner leaflet. In contrast, the environment sensed by a probe expected to localize in both leaflets, such as Laurdan, might not be significantly affected by scrambling. As judged by accessibility to extraction there is very little change in probe exposure on the outer leaflet upon BrA addition. Nevertheless, we cannot rule out a small extent of probe scrambling upon addition of BrA, especially in the case of Pro12A.

A second factor complicating interpretation is that BrA binding to membranes might directly impact ordered domain formation and/or physical properties in addition to altering lipid asymmetry. There was an increase in F/Fo seen in when BrA was added to artificial vesicles composed of SM/DOPC/cholesterol. BrA has a chemical structure that should not pack well in Lo domains, and so is very likely to localize in Ld domains. At the high BrA concentrations used in these studies, this could change the properties of Ld domains to decrease the miscibility of the Ld and Lo lipids and/or increase domain size to make them large enough to detect by FRET³⁴. The experiments in artificial lipid vesicles described above are ambiguous as to whether BrA actually induces domain formation. In addition, F/Fo could increase due to changes in the partition of ODRB or TMADPH between Lo and Ld domains such that TMADPH molecules are more segregated from ODRB molecules when BrA is added.

However, the change in F/Fo in artificial vesicles upon BrA addition was observed both in the presence and absence of Ca²⁺ (Supplemental Figure 5), while the increase observed in cells was Ca²⁺-dependent (Figure 4C), indicating that the FRET change in cells is likely to reflect the change in lipid asymmetry.

CONCLUSIONS

The studies in this report show that adding the scramblase activator BrA to HEK293T cells induces lipid scrambling and an apparent increase in FRET-detected membrane domain formation. BrA has at most modest affects upon the leaflet localization of the non-lipid fluorescent membrane probes used. The bulk FRET approach may also be useful for detecting plasma membrane ordered domain formation induced by biological phenomena that activate lipid scrambling.

DATA AVAILABILITY

All data supporting the findings of this study are available within the paper and its Supplementary Information.