In Situ Monitoring of Phospholipid Flip-Flop via Fluorescence Self-quenching

Abstract

Many antimicrobial peptides (AMPs) exert their activity by disrupting the integrity of the bacterial plasma membrane. One of the membrane-disrupting mechanisms of AMPs involves the formation of toroidal pores, composed of α-helices and lipid headgroups. These pores enable the diffusion of lipid molecules to the opposite leaflet without exposing their headgroups to the hydrocarbon core. Consequently, an increase in lipid transbilayer diffusion (flip-flop) in the presence of AMPs is an important characteristic for AMP structure and function. However, real-time monitoring of the rapid flip-flop in the presence of transmembrane pores has not been achieved because of the permeation of membrane-impermeable reagents and/or the low time resolution of the conventional assays. Herein, we have developed a fluorescence quenching-based flip-flop assay. The flip-flop rates obtained by our method were consistent with those measured by the conventional dithionite reduction assay, confirming the reliability of our approach. The real-time monitoring of the flip-flop process in the presence of the AMP, magainin 2, using the self-quenching assay suggested that the disordered toroidal pores composed of 2 magainin molecules facilitate flip-flop. The newly developed assay will provide a better understanding of the interactions between AMPs and lipid bilayers.

Introduction

Antimicrobial peptides (AMPs), produced in various organisms such as plants, insects, amphibians, and mammals, are naturally occurring peptides that function in the innate immune system.^1,2) AMPs are promising candidates as new antimicrobial agents because of their broad spectrum of antibiotic activity, including antibiotic-resistant bacteria that threaten global public health.³⁾ The lytic activity of AMPs is correlated with the interactions between the peptides and the lipid bilayers of the bacterial plasma membrane. The formation of a transmembrane pore is one of the most studied mechanisms of AMPs, which disrupt the bacterial membrane.^1,4) Many AMPs, such as magainin and melittin, form a toroidal pore in the membrane, which is composed of peptide helices and lipid headgroups.^1,5–7) These pores enable the rapid diffusion of lipid molecules to the opposite leaflet. Conversely, recent studies have proposed the induction of lipid transbilayer movement (flip-flop) by the surface-bound AMPs.^8,9) Therefore, the precise analysis of the AMP-mediated flip-flop process could provide a better understanding of the structure and properties of AMPs in lipid bilayers. However, phospholipid flip-flop in the presence of pores has been scarcely explored, even in model membrane systems, as pore formation presents a major obstacle to the accurate flip-flop assays.

Conventional flip-flop assays have been performed using fluorescence,^4,10–17) ESR,^12,18) NMR,¹⁹⁾ and neutron scattering.^8,20–25) The most important step in the flip-flop assay is distinguishing lipids in the outer and inner leaflets. To achieve this, most assays use the membrane-impermeable reagents that react exclusively with the reporter lipids on the reagent side. However, these methods do not apply to the pore-forming peptides, because they enable reagents to penetrate to the opposite side. Alternative techniques have been developed using intervesicular lipid transfer^10,20,21,25) and lipid extraction with bovine serum albumin.^12,18) However, these lipid transfer processes between vesicles or between bilayers and proteins occur over seconds to minutes, which is similar to or slower than the AMP-mediated flip-flop. Recently, time-resolved neutron scattering experiments with asymmetric vesicles were applied to directly detect the lipid mixing between leaflets by AMPs.⁸⁾ Owing to the low time resolution of the instruments, this study could not trace the rapid disruption process of asymmetry after the addition of AMPs.

The fluorescence intensity of 1-palmitoyl-2-{6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]hexanoyl}-sn-glycero-3-phosphocholine (C₆NBD-PC), which is present at high concentrations in lipid bilayers, is reduced by the interactions between the excited and ground state molecules.²⁶⁾ Collisions between NBD moieties occur only between lipids in the same leaflet owing to the surface localization of the probe. Therefore, the flip-flop of fluorescent lipids in asymmetrically labeled vesicles reduces the frequency of self-quenching, enabling the flip-flop to be detected through the increase in the fluorescence intensity. No additional reagent, except for the fluorescent lipids, is required for this methodology, which is an appropriate property for evaluating the flip-flop induced by AMPs.

In this study, we have developed a flip-flop assay using self-quenching. The obtained flip-flop rates were almost identical to those by the conventional method. We applied the self-quenching assay to evaluate AMP-mediated flip-flop and successfully monitored the flip-flop process immediately after AMP addition. The kinetic analysis indicated that the disordered toroidal pore facilitates the phospholipid flip-flop and concurrent leakage of calcein. The methodology developed here could be useful in unveiling the dynamic mechanisms by which AMPs interact with lipid bilayers.

Experimental

Materials

1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) and 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1′-rac-glycerol) (POPG) were purchased from NOF (Tokyo, Japan). C₆NBD-PC was obtained from Avanti Polar Lipids (Alabaster, AL, U.S.A.). Dithionite was obtained from Sigma-Aldrich (St. Louis, MO, U.S.A.). Calcein was obtained from Dojindo (Kumamoto, Japan). All other chemicals used were of the highest reagent grade.

Peptides

TMP23Q and magainin 2 F12W (Supplementary Table S1) were synthesized using Fmoc-based chemistry.²⁷⁾ The Phe residue at the 12th position from the N-terminus of magainin 2 was substituted with Trp to evaluate membrane binding using Trp fluorescence. Magainin 2 F12W has characteristics similar to those of wild-type peptides in terms of bacterial growth inhibition, membrane leakage promotion, and conformation.²⁸⁾ Synthesized peptides were purified by HPLC (>90% purity) and characterized by MS. TMP23Q and magainin 2 F12W were dissolved in methanol and 20 mM HCl, respectively. The concentrations of peptides were determined using the molar extinction coefficient of tryptophan.

Large Unilamellar Vesicles (LUVs)

Phospholipids with or without TMP23Q in methanol–chloroform solution were placed in a round-bottom flask. After removing the solvent using a rotary evaporator, the lipid-peptide film was dried under vacuum. The dried film was hydrated with Tris buffer (10 mM Tris, 150 mM NaCl, pH 7.4), Tris ethylenediaminetetraacetic acid (EDTA) buffer (10 mM Tris, 150 mM NaCl, 1 mM EDTA, pH 7.4), or 50 mM calcein solution. The vesicle suspension was freeze-thawed several times and extruded using a 100-nm pore filter. Dynamic light scattering measurements (FPAR-1000 particle analyzer; Otsuka Electronics, Osaka, Japan) showed a vesicle diameter of 120–140 nm. Phosphatidylcholine concentration was determined using the phospholipid C-test (FUJIFILM Wako Pure Chemical, Osaka, Japan).

Concentration Dependence of C₆NBD-PC Self-quenching

One hundred micromolar LUVs or 1% Triton X-100 micelles containing various concentrations of C₆NBD-PC were prepared in the Tris buffer and loaded into quartz cuvettes. Fluorescence measurements of these samples were performed using an F-4500 spectrofluorometer (Hitachi, Tokyo, Japan) at excitation and emission wavelengths of 460 and 530 nm, respectively, at 25°C. To obtain the unquenched fluorescence of C₆NBD-PC, LUVs were solubilized using 1% Triton X-100.

Flip-Flop Assay Using Self-quenching

To prepare asymmetrically labeled LUVs, C₆NBD-PC dissolved in ethanol was added to LUVs in the Tris buffer (final ethanol concentration <0.2%). This procedure resulted in the incorporation of C₆NBD-PC into the outer leaflet at 2–4 mol% (the concentration at which self-quenching occurs). The fluorescence intensity from the asymmetrically labeled LUVs was recorded immediately after preparation using an F-4500 or F-2500 spectrofluorometer (Hitachi) at excitation and emission wavelengths of 460 and 530 nm, respectively, at 25°C. After the time-resolved measurements, the LUVs were solubilized using 1% Triton X-100 to quantify the self-quenching efficiency.

Flip-Flop Assay Using Dithionite Reduction

The asymmetrically labeled LUVs were prepared using the above-described procedure, but with a lower concentration of C₆NBD-PC in the outer leaflet (0.4 mol%) to prevent the self-quenching. Conventional flip-flop assays using dithionite were conducted as described previously.^15,16) In brief, the asymmetric LUVs were incubated for 15–360 min at 25°C, and dithionite was added to the vesicles to bleach the NBD fluorescence from the lipids in the outer leaflet. The ratio of C₆NBD-PC in the inner leaflet was determined by fitting the decay profiles of fluorescence intensity using a double-exponential function.

Right-Angle Light Scattering

Vesicle rupture induced by AMPs was evaluated by light scattering from vesicles. A 100 μM suspension of LUVs was mixed with AMPs, and the right-angle light scattering at a wavelength of 650 nm was recorded using an F-4500 spectrofluorometer at 25°C.

Calcein Leakage

Calcein-encapsulated LUVs were prepared in a 50-mM calcein solution, and removal of calcein outside of LUVs was achieved using a spin column (Superdex 200, GE Health Care, Chicago, IL, U.S.A.) with Tris EDTA buffer. LUVs with and without calcein were mixed at a molar ratio of 1 : 9. The fluorescence from the mixed LUVs (100 μM) was recorded after the addition of magainin using an F-4500 spectrofluorometer at excitation and emission wavelengths of 490 and 520 nm, respectively, at 25°C. To obtain the fluorescence from the completely released calcein, the LUVs were solubilized using 0.25% Triton X-100.

Membrane Binding of Magainin 2

Binding of magainin 2 to LUVs was evaluated using tryptophan fluorescence as previously described.²⁸⁾ Briefly, 1–5 μM magainin solution was titrated with LUVs, and tryptophan fluorescence spectra were measured using an F-4500 spectrofluorometer with an excitation wavelength of 280 nm at 25°C. The binding isotherms were obtained from the change in the fluorescence intensity upon LUV addition. The details of the analysis are described in the Supplementary Materials.

Results and Discussion

Quantification of the Fluorescence Self-quenching of C₆NBD-PC in Lipid Bilayers

To quantitatively analyze the phospholipid flip-flop process using fluorescence self-quenching, we evaluated the relationship between the self-quenching efficiency and the ratio of fluorescent lipids to total lipids. The fluorescence intensity of 1–5 μM C₆NBD-PC in 100 μM POPC LUVs, F_LUV, did not increase linearly with C₆NBD-PC concentration (Fig. 1B). In contrast, the fluorescence intensity of C₆NBD-PC in 1% Triton micelles, F_mic, exhibited linearity (Fig. 1B), indicating that incident light attenuation due to the absorption by C₆NBD-PC (inner filter effect) can be ignored in this concentration range. Considering the difference in the fluorescent lipid concentrations in the assemblies (1–5 mol% in LUVs and 0.005–0.03 mol% in micelles), the nonlinearity of the intensity in LUVs is due to the fluorescence self-quenching (Fig. 1A). Because the interior of micelles is more hydrophilic than that of LUVs, the fluorescence intensity in micelles is lower than that in the LUVs even in the absence of self-quenching. Accordingly, F_LUV was normalized by F_mic to estimate the intensity in LUVs without self-quenching. Figure 1C shows the linear relationship between F_LUV/F_mic and the fluorescence lipid ratio in LUVs. The intercept at 0 mol% C₆NBD-PC, A, represents the estimated difference in the intensity between LUVs and micelles. Therefore, AF_mic corresponds to the fluorescence intensity of C₆NBD-PC in LUVs without self-quenching. We defined the self-quenching efficiency as 1 – F_LUV / AF_mic. The self-quenching constant, Q, was determined from the dependence of self-quenching efficiency on the molar ratio of C₆NBD-PC to total lipids, X_all, using the following equation (Fig. 1D):

  

$$1-F_{\text{LUV}}/A{F}_{\text{mic}}=Q{X}_{\text{all}}$$(1)

Fig. 1. Quantification of the Fluorescence Self-quenching of C₆NBD-PC in Lipid Bilayers

(A) Illustration of the experiment quantifying the self-quenching efficiency of C₆NBD-PC. At higher C₆NBD-PC/lipid ratios, the self-quenching of C₆NBD-PC occurs owing to the interactions between C₆NBD-PC molecules. The addition of an excess amount of detergent decreases the local concentration of C₆NBD-PC, reducing the self-quenching (the increase in the fluorescence intensity). Therefore, the self-quenching efficiency can be estimated from the ratio of fluorescence intensity before and after adding the detergent. (B) Fluorescence intensity of C₆NBD-PC in 100 μM LUVs (F_LUV) or 1% Triton X-100 micelles (F_mic). Factor “A” is used to compensate for the difference in the intensity of C₆NBD-PC in LUVs and micelles, which was determined in (C). (C) Ratio of F_LUV and F_mic. Factor “A” is the intercept at C₆NBD-PC/lipids = 0. (D) Efficiency of fluorescence self-quenching of C₆NBD-PC in LUVs. Self-quenching efficiency is defined as 1 – F_LUV/AF_mic. The obtained Q value was 7.56.

Although a previous study reported an exponential relationship between the self-quenching efficiency and X_all,²⁶⁾ the self-quenching efficiency obtained in this study fitted well with the linear function in Eq. (1) owing to the relatively small values of X_all.

Establishment of a Flip-Flop Assay Using C₆NBD-PC Self-quenching

The phospholipid flip-flop event was evaluated using the reduction in self-quenching. Asymmetrically labeled LUVs were prepared by adding C₆NBD-PC to LUVs. C₆NBD-PCs inserted into the outer leaflet translocate to the inner leaflet by flip-flop motion, decreasing the C₆NBD-PC/total lipid ratio in the outer leaflet. For LUVs with high C₆NBD-PC ratios, this causes a reduction in the self-quenching (Fig. 2A). The flip-flop in POPC LUVs was very slow^15–17); hence, a flip-flop-promoting peptide, TMP23Q,¹⁵⁾ was included in LUVs in this study. The fluorescence from LUV dispersion was monitored after adding 1 mol% C₆NBD-PC. Figure 2A shows the increase in the fluorescence intensity when C₆NBD-PC was inserted into LUVs containing TMP23Q, whereas no change in the intensity occurred when C₆NBD-PC was inserted into LUVs without TMP23Q. Notably, when the C₆NBD-PC/total lipid ratio was low (0.1 mol%), the intensity was unchanged even in the presence of the peptide (Fig. 2A). These results indicate that the increase in the fluorescence intensity of LUVs containing TMP23Q and high concentrations of C₆NBD-PC can be attributed to reduced self-quenching by rapid flip-flop.

Fig. 2. Evaluation of Phospholipid Flip-Flop Using Self-quenching

(A) Illustration of the flip-flop assay via self-quenching. C₆NBD-PC added to LUVs is inserted into the outer leaflet. Flip-flop motion redistributes C₆NBD-PC into the inner leaflet, decreasing the C₆NBD-PC/total lipid ratio in the outer leaflet. This reduces the self-quenching efficiency and increases the fluorescence intensity of C₆NBD-PC. (Right) Fluorescence intensity of asymmetrically labeled LUVs without peptides (blue) or containing 0.05 mol% TMP23Q (red and open circles). Time-resolved measurements began after adding 0.1 (open circles) or 1 mol% C₆NBD-PC (red and blue). (B) Change in the self-quenching efficiency obtained from (A). Solid lines represent fitting curves using Eq. (13). (C, D) Flip rate constants of C₆NBD-PC in LUVs containing 0.05 mol% TMP23Q (C) or without peptides (D) determined using the self-quenching or dithionite methods. Error bars represent ± standard deviation (S.D.) throughout the paper.

The changes in the fluorescence intensity in the flip-flop assay were converted to the changes in the self-quenching efficiency using the fluorescence from solubilized LUVs after time-resolved measurements (Fig. 2B). The flip-flop rate constants of C₆NBD-PC were determined from the decay profile of the self-quenching efficiency by the following procedure. F_LUV is composed of the intensity of C₆NBD-PC in the outer (F_out) and inner (F_in) leaflets:

  

$$F_{\mathrm{L}\mathrm{U}\mathrm{V}}=F_{\mathrm{o}\mathrm{u}\mathrm{t}}+F_{\mathrm{i}\mathrm{n}}$$(2)

F_out and F_in were calculated using Q and the molar ratios of C₆NBD-PC to total lipids in the outer (X_out) and inner (X_in) leaflets, respectively,

  

$$F_{\mathrm{o}\mathrm{u}\mathrm{t}}=(1-Q{X}_{\mathrm{o}\mathrm{u}\mathrm{t}})\frac{A{F}_{\mathrm{m}\mathrm{i}\mathrm{c}}{X}_{\mathrm{o}\mathrm{u}\mathrm{t}}}{X_{\mathrm{o}\mathrm{u}\mathrm{t}}+X_{\mathrm{i}\mathrm{n}}}$$(3)

  

$$F_{\mathrm{i}\mathrm{n}}=(1-Q{X}_{\mathrm{i}\mathrm{n}})\frac{A{F}_{\mathrm{m}\mathrm{i}\mathrm{c}}{X}_{\mathrm{i}\mathrm{n}}}{X_{\mathrm{o}\mathrm{u}\mathrm{t}}+X_{\mathrm{i}\mathrm{n}}}$$(4)

From Eqs. (2) to (4), the following relationship was obtained:

  

$$1-F_{\mathrm{L}\mathrm{U}\mathrm{V}}/A{F}_{\mathrm{m}\mathrm{i}\mathrm{c}}=\frac{Q({X_{\mathrm{o}\mathrm{u}\mathrm{t}}}^2+{X_{\mathrm{i}\mathrm{n}}}^2)}{X_{\mathrm{o}\mathrm{u}\mathrm{t}}+X_{\mathrm{i}\mathrm{n}}}$$(5)

The kinetics of the C₆NBD-PC ratio in each leaflet can be described as follows:

  

$$-\frac{d{\left[NBD\right]}_{\mathrm{o}\mathrm{u}\mathrm{t}}}{dt}=k_{\text{flip}}{\left[NBD\right]}_{\mathrm{o}\mathrm{u}\mathrm{t}}-k_{\text{flop}}{\left[NBD\right]}_{\mathrm{i}\mathrm{n}}$$(6)

  

$$-\frac{d{\left[NBD\right]}_{\mathrm{i}\mathrm{n}}}{dt}=k_{\text{flop}}{\left[NBD\right]}_{\mathrm{i}\mathrm{n}}-k_{\text{flip}}{\left[NBD\right]}_{\mathrm{o}\mathrm{u}\mathrm{t}}$$(7)

where [NBD]_out, [NBD]_in, k_flip, and k_flop represent the concentrations of C₆NBD-PC in the outer and inner leaflets and the rate constants of flip and flop, respectively. For the energy-independent flip-flop process, k_flip and k_flop are almost the same.^14,15,17) Thus, the above differential equations can be solved under the assumption k_flip = k_flop as follows:

  

$$[NBD{]}_{\mathrm{o}\mathrm{u}\mathrm{t}}=(\frac{1}{2}-{\mathrm{\Phi }}_{\mathrm{i}\mathrm{n},0})[NBD{]}_{\mathrm{o}\mathrm{u}\mathrm{t}}\exp (-2k_{\text{flip}}t)+\frac{[NBD{]}_{\mathrm{a}\mathrm{l}\mathrm{l}}}{2}$$(8)

  

$$[NBD{]}_{\mathrm{i}\mathrm{n}}=-(\frac{1}{2}-{\mathrm{\Phi }}_{\mathrm{i}\mathrm{n},0})[NBD{]}_{\mathrm{o}\mathrm{u}\mathrm{t}}\exp (-2k_{\text{flip}}t)+\frac{[NBD{]}_{\mathrm{a}\mathrm{l}\mathrm{l}}}{2}$$(9)

where [NBD]_all and Φ_in,0 are the concentration of C₆NBD-PC in the bilayer and the proportion of C₆NBD-PC localized in the inner leaflet at t = 0. The mole fractions of C₆NBD-PC are represented using the total lipid concentration as

  

$$X_{\mathrm{a}\mathrm{l}\mathrm{l}}=\frac{[NBD{]}_{\mathrm{a}\mathrm{l}\mathrm{l}}}{[\mathit{l}\mathit{i}\mathit{p}\mathit{i}\mathit{d}]}, X_{\mathrm{o}\mathrm{u}\mathrm{t}}=\frac{2[NBD{]}_{\mathrm{o}\mathrm{u}\mathrm{t}}}{[\mathit{l}\mathit{i}\mathit{p}\mathit{i}\mathit{d}]}, X_{\mathrm{i}\mathrm{n}}=\frac{2[NBD{]}_{\mathrm{i}\mathrm{n}}}{[\mathit{l}\mathit{i}\mathit{p}\mathit{i}\mathit{d}]}$$(10)

Summing up Eqs. (8)–(10), X_out and X_in can be described using the following equations:

  

$$X_{\mathrm{o}\mathrm{u}\mathrm{t}}=(1-2{\mathrm{\Phi }}_{\mathrm{i}\mathrm{n},0})\exp (-2k_{\text{flip}}t)+X_{\mathrm{a}\mathrm{l}\mathrm{l}}$$(11)

  

$$X_{\mathrm{i}\mathrm{n}}=-(1-2{\mathrm{\Phi }}_{\mathrm{i}\mathrm{n},0})\exp (-2k_{\text{flip}}t)+X_{\mathrm{a}\mathrm{l}\mathrm{l}}$$(12)

Finally, the relationship between the self-quenching efficiency and the flip rate constant is obtained as

  

$$1-F_{\mathrm{L}\mathrm{U}\mathrm{V}}/A{F}_{\mathrm{m}\mathrm{i}\mathrm{c}}=Q{X}_{\mathrm{a}\mathrm{l}\mathrm{l}}\{{(1-2{\mathrm{\Phi }}_{\mathrm{i}\mathrm{n},0})}^2\exp (-4k_{\text{flip}}t)+1\}$$(13)

Using Eq. (13), the flip rate constants were determined as 91.8 ± 11.7 × 10⁻⁴ and 0.701 ± 0.317 × 10⁻⁴ min⁻¹ for LUVs with and without TMP23Q, respectively (Fig. 2B).

TMP23Q does not affect the membrane permeability of dithionite, and therefore, flip-flop of C₆NBD-PC in LUVs containing TMP23Q was also evaluated using the conventional method with dithionite (Supplementary Fig. S1). The time course of the proportion of C₆NBD-PCs localized in the inner leaflet, Φ_in, was determined from the fluorescence intensity before and after adding dithionite. The obtained flip rate constants corresponded to those by the self-quenching method (Figs. 2C and 2D), demonstrating the validity of the newly developed method. In addition, the difference in the amount of C₆NBD-PC inserted into the outer leaflet (2 and 0.4 mol% for self-quenching and dithionite methods, respectively) did not affect the spontaneous and peptide-mediated flip-flop.

Flip-Flop Assay in the Presence of Transmembrane Pores Composed of AMPs

We applied the developed self-quenching assay to evaluate phospholipid flip-flop in the presence of the transmembrane pores composed of the AMP, magainin 2. Because magainin 2 binds to the membranes containing negatively charged lipids,²⁹⁾ POPC/POPG = 1/1 LUVs were used for the magainin experiments. The calcein leakage experiments were performed to examine the pore formation of magainin. The fluorescence of calcein encapsulated in LUVs at a high concentration (50 mM) is self-quenched. The presence of magainin pores facilitates the leakage of calcein from LUVs, reducing the self-quenching. A significant rise in calcein fluorescence was observed when the concentration of magainin added to LUVs was ≥0.75 μM, indicating the formation of magainin pores (Fig. 3A). The concentration required for pore formation was lower than the minimal inhibitory concentration of magainin for Escherichia coli (31.4 μM), possibly due to the higher PG content in LUVs than that in bacterial membranes.²⁸⁾

Fig. 3. In Situ Monitoring of Phospholipid Flip-Flop in the Presence of Pores

(A) Calcein leakage from LUVs after the addition of magainin 2 F12W. (B) Initial leakage rate of calcein from LUVs with magainin (n = 3). (C) Self-quenching flip-flop assay of LUVs after the addition of magainin 2 F12W. (D) Initial flip rate of C₆NBD-PC in LUVs with magainin (n = 3–6).

Adding magainin to the LUVs asymmetrically labeled with C₆NBD-PC increased the fluorescence intensity of C₆NBD-PC (Supplementary Fig. S2), demonstrating the real-time monitoring of flip-flop in the presence of the pores. Monitoring the intensity increase immediately after adding magainin enabled a precise analysis of the flip-flop process in the presence of transmembrane pores. Because the self-quenching efficiency of C₆NBD-PC in POPC/POPG = 1/1 LUVs was almost identical to that in POPC LUVs (Supplementary Fig. S3), the same equation, Eq. (1), was used to convert fluorescence intensity to self-quenching efficiency (Fig. 3C). Notably, magainin bound to lipid vesicles moves across the bilayer via the formation of toroidal pores.³⁰⁾ The pores of magainin are more likely to form in the early stages of binding because of the imbalance in the peptide density in each leaflet.³¹⁾ Indeed, at 0.75 μM magainin, the decay of self-quenching efficiency leveled off after 2 min (Fig. 3C), suggesting that after lipid flip-flop through the pores occurred, the pores disappeared owing to peptide migration to the inner leaflet. Thus, the initial rates of flip-flop and leakage were used for the analysis. Figures 3B and 3D show the dependence of leakage and flip-flop rates on the magainin concentration, respectively. The leakage and flip-flop rates increased upon adding ≥0.75 μM magainin, suggesting that similar mechanisms mediate both processes.

To further characterize the magainin-mediated flip-flop, the binding experiments of magainin to LUVs were conducted. Utilizing the shift in tryptophan fluorescence spectra of magainin upon membrane binding,²⁸⁾ a sigmoidal binding isotherm of magainin to POPC/POPG = 1/1 LUVs was obtained (Supplementary Fig. S4), which is consistent with a previous report.²⁸⁾ Considering this isotherm, the constant flip rate at high concentrations of magainin was attributed to the saturation of magainin binding (Fig. 3D). We confirmed by light scattering that no vesicles ruptured upon binding (Supplementary Fig. S5). Figure 4 shows that the flip-flop and leakage rates are proportional to the approximate square of the amount of membrane-bound magainin per lipid, r. This suggests that magainin dimers induce both phospholipid flip-flop and calcein leakage in lipid bilayers. Although toroidal pores of magainin were estimated to be tetra- or pentameric in pure POPG membranes,^28,31) the dynamic pore size variation was reported for DOPC/DOPG = 1/1 membrane.³²⁾ Several simulation studies have also shown the formation of disordered toroidal pores composed of 1 or 2 AMP molecules.^33,34) Thus, the observed phenomena can be explained by assuming the presence of disordered pores consisting of 2 molecules of magainin.

Fig. 4. Dependence of Initial Flip Rate (A) and Leakage Rate (B) on the Amount of Membrane-Bound Magainin per Lipid, r

Solid lines represent linear regression curves with the slope values of 1.9 and 2.1 for (A) and (B), respectively.

Conclusion

In this study, we developed a phospholipid flip-flop assay based on fluorescence self-quenching, enabling accurate evaluation of flip-flop rates in lipid vesicles. Unlike conventional assay methods, our assay is superior and does not rely on membrane permeability, making it suitable for systems involving transmembrane pores. Using this approach, we have demonstrated in situ monitoring of phospholipid flip-flop in the presence of transmembrane pores of the AMPs. Moreover, the self-quenching assay has high temporal resolution. Changes in the fluorescence intensity were detected immediately after peptide addition, enabling detailed kinetic analysis. The permeability-independent flip-flop assay with high time resolution would be a valuable tool for understanding the interactions of AMPs with lipid membranes as well as transbilayer lipid mixing during morphological changes in bilayers, such as membrane fusion and fission.

Acknowledgments

This study was supported by JSPS KAKENHI under Grant Numbers: JP19K16086 and JP23K14144 (to H.N.), JP20K06998 and JP23H02620 (to K.I.), and JP17H02941 and JP22H02194 (to M.N.).

Conflict of Interest

The authors declare no conflict of interest.

Supplementary Materials

This article contains supplementary materials.

References

• 1)  Zasloff M., Nature, 415, 389–395 (2002).
• 2)  Melo M. N., Ferre R., Castanho M. A., Nat. Rev. Microbiol., 7, 245–250 (2009).
• 3)  Hancock R. E., Sahl H. G., Nat. Biotechnol., 24, 1551–1557 (2006).
• 4)  Matsuzaki K., Murase O., Fujii N., Miyajima K., Biochemistry, 35, 11361–11368 (1996).
• 5)  Matsuzaki K., Biochim. Biophys. Acta Biomembr., 1462, 1–10 (1999).
• 6)  Shai Y., Biochim. Biophys. Acta Biomembr., 1462, 55–70 (1999).
• 7)  Yang L., Weiss T. M., Lehrer R. I., Huang H. W., Biophys. J., 79, 2002–2009 (2000).
• 8)  Nguyen M.H. L., DiPasquale M., Rickeard B. W., Doktorova M., Heberle F. A., Scott H. L., Barrera F. N., Taylor G., Collier C. P., Stanley C. B., Katsaras J., Marquardt D., Langmuir, 35, 11735–11744 (2019).
• 9)  Taylor G., Nguyen M. A., Koner S., Freeman E., Collier C. P., Sarles S. A., Biochim. Biophys. Acta Biomembr., 1861, 335–343 (2019).
• 10)  Fattal E., Nir S., Parente R. A., Szoka F. C. Jr., Biochemistry, 33, 6721–6731 (1994).
• 11)  Homan R., Pownall H. J., Biochim. Biophys. Acta Biomembr., 938, 155–166 (1988).
• 12)  Marx U., Lassmann G., Holzhutter H. G., Wustner D., Muller P., Hohlig A., Kubelt J., Herrmann A., Biophys. J., 78, 2628–2640 (2000).
• 13)  Kol M. A., de Kroon A. I., Rijkers D. T., Killian J. A., de Kruijff B., Biochemistry, 40, 10500–10506 (2001).
• 14)  Nakao H., Ikeda K., Ishihama Y., Nakano M., Biophys. J., 110, 2689–2697 (2016).
• 15)  Nakao H., Hayashi C., Ikeda K., Saito H., Nagao H., Nakano M., J. Phys. Chem. B, 122, 4318–4324 (2018).
• 16)  Nakao H., Sugimoto Y., Ikeda K., Saito H., Nakano M., J. Phys. Chem. Lett., 11, 1662–1667 (2020).
• 17)  Nakao H., Tsujii T., Saito H., Ikeda K., Nakano M., Colloids Surf. B Biointerfaces, 251, 114612 (2025).
• 18)  Buton X., Morrot G., Fellmann P., Seigneuret M., J. Biol. Chem., 271, 6651–6657 (1996).
• 19)  Marquardt D., Heberle F. A., Miti T., Eicher B., London E., Katsaras J., Pabst G., Langmuir, 33, 3731–3741 (2017).
• 20)  Nakano M., Fukuda M., Kudo T., Endo H., Handa T., Phys. Rev. Lett., 98, 238101 (2007).
• 21)  Nakano M., Fukuda M., Kudo T., Matsuzaki N., Azuma T., Sekine K., Endo H., Handa T., J. Phys. Chem. B, 113, 6745–6748 (2009).
• 22)  Kaihara M., Nakao H., Yokoyama H., Endo H., Ishihama Y., Handa T., Nakano M., Chem. Phys., 419, 78–83 (2013).
• 23)  Nakao H., Kimura Y., Sakai A., Ikeda K., Nakano M., Struct. Dyn., 8, 024301 (2021).
• 24)  Nakao H., Nagao M., Yamada T., Imamura K., Nozaki K., Ikeda K., Nakano M., Colloids Surf. B Biointerfaces, 228, 113396 (2023).
• 25)  Nielsen J. E., Bjornestad V. A., Pipich V., Jenssen H., Lund R., J. Colloid Interface Sci., 582(Pt. B), 793–802 (2021).
• 26)  Nichols J. W., Pagano R. E., Biochemistry, 20, 2783–2789 (1981).
• 27)  Chan W., White P., “Fmoc solid phase peptide synthesis: a practical approach,” Oxford University Press, Oxford, 1999.
• 28)  Matsuzaki K., Murase O., Tokuda H., Funakoshi S., Fujii N., Miyajima K., Biochemistry, 33, 3342–3349 (1994).
• 29)  Matsuzaki K., Harada M., Funakoshi S., Fujii N., Miyajima K., Biochim. Biophys. Acta Biomembr., 1063, 162–170 (1991).
• 30)  Matsuzaki K., Murase O., Fujii N., Miyajima K., Biochemistry, 34, 6521–6526 (1995).
• 31)  Matsuzaki K., Murase O., Miyajima K., Biochemistry, 34, 12553–12559 (1995).
• 32)  Tamba Y., Ariyama H., Levadny V., Yamazaki M., J. Phys. Chem. B, 114, 12018–12026 (2010).
• 33)  Leontiadou H., Mark A. E., Marrink S. J., J. Am. Chem. Soc., 128, 12156–12161 (2006).
• 34)  Sengupta D., Leontiadou H., Mark A. E., Marrink S. J., Biochim. Biophys. Acta Biomembr., 1778, 2308–2317 (2008).