Lipid Bilayers in the Gel Phase Become Saturated by Triton X-100 at Lower Surfactant Concentrations Than Those in the Fluid Phase

Abstract

It has been repeatedly observed that lipid bilayers in the gel phase are solubilized by lower concentrations of Triton X-100, at least within certain temperature ranges, or other nonionic detergents than bilayers in the fluid phase. In a previous study, we showed that detergent partition coefficients into the lipid bilayer were the same for the gel and the fluid phases. In this contribution, turbidity, calorimetry, and ³¹P-NMR concur in showing that bilayers in the gel state (at least down to 13–20°C below the gel-fluid transition temperature) become saturated with detergent at lower detergent concentrations than those in the fluid state, irrespective of temperature. The different saturation may explain the observed differences in solubilization.

Introduction

Soluble amphiphiles, also called surfactants or detergents, are tools of prime importance in the study of cell membranes (1–4). However, their use is largely empirical, probably because essential aspects of bilayer-detergent interaction have eluded us in the past, and continue doing so despite extensive research. Recent studies from our laboratory have focused on the early stages of membrane-detergent interaction, when solubilization has not started (5,6). Ahyayauch et al. (6) after studying a variety of detergent effects on membranes at subsolubilizing concentrations (transmembrane lipid motion, bilayer permeabilization, vesicle lysis and reassembly) with five different surfactants, concluded that each detergent behaves in a unique way, and that the various effects observed are independent from each other, and not different parts or stages of a single solubilization process.

Arnulphi et al. (5) took a more quantitative approach, studying Triton X-100 partitioning into palmitoylsphingomyelin (SM) or dipalmitoylphosphatidylcholine (DPPC) bilayers at concentrations well below those causing solubilization. Using a combination of isothermal titration calorimetry with other physical techniques they found that, irrespective of the chemical nature of the lipid and the gel or fluid phases, ΔG of partitioning remained virtually constant, in the range of 27–30 kJ/mol lipids. This was surprising because it had been established that, in general, lipids in the fluid state required more detergent for their solubilization than those in the gel phase (7–11). Thus, Arnulphi et al. (5) concluded that the observed different susceptibilities to solubilization could not be attributed to differential detergent binding to the liquid or gel phases, but to events downstream in the solubilization process. The differences might be due to i), a different saturation capacity of the various bilayers in different phase states, gel or fluid, so that saturation, i.e., the onset of solubilization (1–4) would occur at a different lipid/detergent ratio; or ii), differences within the far less well-understood stage II (1) or micellization step.

This contribution was intended to clarify whether or not the detergent saturation capacity of bilayers changed with the gel-fluid transition. Saturation was measured by the method described by Schurtenberger et al. (12), in which the detergent concentrations causing the onset of solubilization are plotted versus lipid concentration. Solubilization was assayed as a decrease in turbidity at different temperatures (13). At 50°C the temperature was above the cloud point of Triton X-100 (14), and this prevented turbidity measurements, thus a colorimetric method developed in our laboratory was used instead (15). Our results indicate that the amount of detergent required to saturate the bilayers is higher for lipids in the fluid phase than for those in the gel phase (about +7–10°C and −13–20°C with respect to the gel-fluid transition temperature, respectively).

Materials and Methods

Materials

Egg SM (86% N-palmitoyl), DPPC, dimyristoyl phosphatidylcholine (DMPC), and egg phosphatidylcholine (PC) (33% C16:0, 32% C18:1, 17% C18:2, 12% C18:0) were supplied by Avanti Polar Lipids (Alabaster, AL). Triton X-100 (batch no. 125H0569) and merocyanine 540 MC540 were purchased from Sigma (St. Louis, MO). All other reagents were of analytical grade.

Preparation of large unilamellar vesicles

The lipids were dissolved in chloroform/methanol (2:1,v/v) and mixed as required, and the solvent was evaporated exhaustively. Multilamellar vesicles were prepared by hydrating the dry lipids in buffer, with vortex shaking. Lipids were hydrated in 10 mM HEPES, 150 mM NaCl, pH 7.4. Large unilamellar vesicles were prepared by the extrusion method (10 passages) with filters 0.1 μm in pore diameter (16). Vesicle size was measured by quasielastic light scattering in a Malvern Zeta-Sizer 4 spectrometer. The average diameter of the vesicles was in all cases ∼100 nm.

Turbidity assay

Liposome dispersions were mixed with the same volumes of the appropriate Triton solutions in the same buffer. Final lipid concentration, measured as lipid phosphorus, was ranging from 0.3 to 2.0 mM. The mixtures were left to equilibrate for 1 h at desired temperature, and solubilization was assessed from the changes in turbidity (13), measured as absorbance at 500 nm in a double-beam Uvikon Kontron spectrophotometer (Kontron Instruments, Milan, Italy). Temperatures were kept constant with <0.5°C deviation.

Absorption spectra of merocyanine (MC540)

Due to the cloud point of Triton X-100, at ≈50°C, membrane solubilization was monitored following detergent-induced changes in the visible absorption spectrum of MC540 (15). This dye is known to show a shift in the wavelength maximum (λ_max) due to the presence of micelles (17,18). The lipid and MC540 (at a molar ratio 250:1) mixtures in organic solution were evaporated under a vacuum for 2 h. The large unilamellar vesicles (LUVs) were prepared following the protocol cited previously. The liposome dispersions containing MC540 were mixed with the same volume of the appropriate Triton solutions. Absorption spectra were recorded between 400 and 600 nm with a double-beam Uvikon Kontron spectrophotometer.

NMR measurements

A dispersion of LUVs was incubated at 20 ± 0.5°C or at 50 ± 0.5°C with Triton X-100 at a 0.42 detergent/lipid ratio. The final phospholipid concentration was 20 mM. The mixture was transferred to 5-mm glass NMR tubes. All the samples were allowed to equilibrate to the appropriate temperature for 1 h before experimentation. NMR spectra were recorded in a Bruker (Rheinstetten, Germany) AV500 spectrometer operating at 500 MHz for protons, 202.4 MHz for ³¹P and full proton decoupling. The instrument was equipped with an inverse broadband probe of 5 mm and gradients in the z axis. Scans (24,000) were averaged for each measurement. The data were recorded and processed with software TOPSPIN 1.3 (Bruker).

DSC

For DSC, both the lipid suspension and buffer were degassed before being loaded into the sample or reference cell of an MC-2 high-sensitivity scanning calorimeter (MicroCal, Northampton, MA). The final lipid concentration was 2 mM. Three heating scans, at 45°C/h were recorded for each sample. After the first one, successive heating scans on the same sample always yielded superimposable thermograms. Transition temperatures, enthalpies, and widths at half height were determined using the software ORIGIN (MicroCal) provided with the calorimeter.

Results

Solubilization of SM in the gel or fluid phase

LUV consisting of pure SM, whose gel-fluid transition occurs at ≈40°C (19), was treated with Triton X-100 either at 20°C or at 50°C as detailed in the Methods section. The onset of solubilization, i.e., the lowest detergent concentration causing solubilization of the LUV suspension was recorded at SM concentrations ranging from 0.3 to 2.0 mM. Solubilization at 20°C was assessed as a decrease in turbidity (13) (Fig. 1 A). Turbidity could not be used at 50°C because at this temperature the SM/Triton mixture is above the cloud point, and the suspension is highly turbid (14). Kaschny et al. (15) had described a colorimetric method, based on changes in the absorption spectrum of merocyanine 540 (MC540), which allows the detection of lipid bilayer solubilization. A representative experiment of SM bilayer solubilization monitored by MC540 absorption spectroscopy is shown in Fig. 1 B. Preliminary experiments had shown that the MC540 method could be used even with very cloudy dispersions. Moreover, at 20°C it gave identical results to those obtained with the turbidimetric procedure (Fig. 1 C).

Figure 1.

Solubilization of SM vesicles. (A) Solubilization at 20°C, characterized by the decrease in turbidity as a function of Triton X-100 concentration. SM concentration was 1 mM. (D_on) is the total detergent concentration required for the onset of lipid solubilization. (B) Solubilization at 50°C, monitored by MC540 absorption spectra as a function of Triton X-100 concentration. SM concentration was 1 mM. (C) Comparison of detergent concentrations required for the onset (D_on) of solubilization obtained by the turbidity assay (○) or from MC 540 absorption spectra (●). D_on is plotted as a function of SM concentration at 20°C. Data are average values (n = 3). SD is the size of the symbols, or smaller.

Consequently, turbidity was used to assess solubilization at 20°C, for convenience, whereas at 50°C the MC540 method was applied. The results are summarized in Fig. 2. The effective detergent/lipid ratios R_e^sat are obtained from the slopes of the D_on versus [lipid] plots, D_on being the total detergent concentration causing onset of solubilization at a given lipid concentration. R_e^sat for pure SM bilayers at 20 and 50°C are respectively 0.24 and 0.48, i.e., solubilization starts at 20°C (or 50°C) when the detergent/lipid molar ratio in the bilayer is 0.24 (or 0.48). Thus, SM bilayers in the fluid state become saturated with Triton X-100 at twice the effective concentration than the same bilayers in the gel state (Table 1). This fact alone may explain the observation that SM in the gel state is solubilized more easily than in the fluid state, although additional factors leading to the same effect cannot be excluded.

Figure 2.

Computation of R_e^sat for SM and Triton X-100. The detergent concentrations required for the onset (D_on) of solubilization are plotted as a function of SM concentration at 20°C (○) and 50°C (●). Measurements at 20 and 50°C were carried out with the turbidity and the MC540 methods, respectively. Data are average values (n = 3). SD is the size of the symbols, or smaller. Regression coefficients for the straight lines are 0.93 (20°C), 0.99 (50°C).

Table 1.

Triton X-100 solubilization parameters for lipid bilayers at different temperatures and varying phase structures

Lipid	Temperature (°C)	Phase	Resat	Resol	Dw (mM)
Egg SM	20	gel	0.24 ± 0.04	0.66 ± 0.06	0.03 ± 0.03
Egg SM	50	fluid	0.48 ± 0.01	3.05 ± 0.08	0.03 ± 0.01
DPPC	20	gel	1.15 ± 0.07	1.45 ± 0.10	0.06 ± 0.03
DPPC	50	fluid	2.04 ± 0.03	2.8 ± 0.20	0.09 ± 0.03
DMPC	10	gel	0.25 ± 0.02	1.48 ± 0.01	0.05 ± 0.02
DMPC	30	fluid	0.94 ± 0.04	2.75 ± 0.12	0.09 ± 0.05
Egg PC	20	fluid	0.87 ± 0.05	2.29 ± 0.12	0.13 ± 0.05
Egg PC	50	fluid	0.94 ± 0.04	2.31 ± 0.13	0.11 ± 0.04

R_e^sat is the effective detergent/lipid ratio in the bilayer at the onset of solubilization. D_w is the limit concentration of free detergent in water at zero lipid concentration, at the onset of solubilization. R_e^sol is the detergent/lipid ratio above which all the bilayer is solubilized into mixed micelles. Data obtained from experiments as shown in Figs. 1, 2, and 4. Averages of three independent measurements ±SD.

In confirmation of the previous results SM solubilization was assessed by ³¹P-NMR, a method that allows direct observation of micelle formation (20). The method is based on the fact that when lipids occur in large aggregates, e.g., multilamellar liposomes, the NMR line shape associated with the phosphoryl group is inhomogeneously broadened, however the much smaller detergent/lipid mixed micelles yield narrow NMR signals (20). In our case, a dispersion of SM liposomes was incubated at 20°C with Triton X-100 at a 0.4 detergent/lipid ratio, i.e., at the start of the solubilization stage. The NMR spectrum (Fig. 3, bottom) shows a narrow symmetric spectrum compatible with small particles tumbling rapidly in the magnetic field. The sample was then equilibrated at 50°C, and a spectrum taken under those conditions. The spectral line at 50°C (Fig. 3, top) is much broader and asymmetric, characteristic of a phospholipid in the lamellar phase (21). A series of spectra taken while the temperature was gradually increased from 25 to 35°C (Fig. S1, in the Supporting Material) shows an abrupt change from an isotropic to anisotropic signal between 30 and 32°C. In fact, the ³¹P-NMR signal in the 32–35°C range appears rather symmetric, i.e., the low-frequency shoulder characteristic of the lamellar phase spectra is not seen. The homogenous signal at ≈−10 ppm is similar to that observed by Funari et al. (22) for detergent-phospholipid mixed micelles with a crystalline phospholipid core. This deserves further investigation. In our case NMR confirms that at the given detergent/SM ratio, temperature may determine whether or not solubilization occurs, as predicted by the data in Fig. 2.

Figure 3.

³¹P-NMR spectra of SM vesicles in the presence of Triton X-100 at 20°C (lower spectrum) and 50°C (upper spectrum). SM concentration was 20 mM and effective detergent/lipid molar ratio was 0.42.

The effect of temperature

The SM experiment indicated that SM bilayers at 20°C, i.e., in the gel phase, became saturated with Triton X-100 at lower detergent concentrations than those at 50°C, in the fluid phase. However, generalization of this observation required to elucidate whether the different saturabilities were due to the difference in phase state, or to the difference in temperature. This was clarified after performing experiments as shown in Fig. 4. First, similar measurements were performed with DPPC, which has a gel-fluid transition temperature at 41°C (23). Again, it was found that R_e^sat at 50°C was about twice the value at 20°C, although in absolute terms at both temperatures, R_e^sat for DPPC was about fourfold that of SM (Table 1). It was known that SM required for solubilization less Triton X-100 than DPPC (24). A similar experiment was performed on DMPC bilayers, whose gel-fluid transition temperature is 23°C (23), at both 10 and 30°C. Furthermore, in this case R_e^sat, i.e., the effective detergent/lipid ratio causing membrane saturation with detergent was higher, by ∼fourfold in this case, for bilayers in the fluid state than for those in the gel state.

Figure 4.

Solubilization of various PC bilayers at different temperatures. The detergent concentrations required for the onset (D_on) of solubilization are plotted as a function of phospholipid concentration at 20°C (○) and 50°C (●). (A) DPPC; (B) DMPC; (C) egg PC. Data are average values (n = 3). SD is the size of the symbols, or smaller. Regression coefficients: (A) 20°C, 0.97; 50°C, 0.99; (B) 10°C, 0.97; 30°C, 0.98; (C) 20°C, 0.99; 50°C, 0.99.

A further control involved turbidity and MC540 measurements with egg PC, whose gel-fluid transition temperature is near 4°C, performed at 20 and 50°C. The bilayer remains in the lamellar fluid state at both temperatures, and R_e^sat stays constant at a value (≈0.9) similar to that of DMPC in the fluid state, confirming that the observed differences in R_e^sat with SM, DPPC, and DMPC are due to the thermotropic change in the physical state of the lipid.

Phospholipid phase structure in the presence of Triton X-100

In the previous paragraphs, the phospholipids have been considered to be in the gel phase at 20°C (SM, DPPC) or at 10°C (DMPC) and in the fluid phase at 50°C (SM, DPPC) or at 30°C (DMPC). This assumption was mainly based on the phase behavior of the pure phospholipid. However, the presence of detergent could shift the gel-fluid phase transition temperature, thus giving rise to artifactual results. To clarify this point, mixtures containing phospholipid and detergent under conditions leading to saturation in the presumed gel phase (see values in Table 1 for R_e^sat at 20°C for SM and DPPC, at 10°C for DMPC) were studied by DSC. The results in Fig. S2 show that the detergent shifts the onset of the gel-fluid transition temperature to lower temperatures, however both SM and DPPC detergent mixtures are in the gel phase at 20°C and in the fluid phase at 50°C, and correspondingly the detergent-saturated DMPC bilayers are in the gel state at 10°C and in the fluid state at 30°C. This confirms our interpretation of the data in Figs. 1–4 and Table 1. Similar data for the PC-detergent mixtures had been published by this laboratory (25,26).

A further question arises because DMPC and DPPC may exist in two different gel phases, respectively the Lβ′ and the rippled Pβ′ (27) in the temperature range of our study. However, previous studies (5,25,26) have shown that even small detergent concentration, well below R_e^sat, abolish the so-called pretransition from Lβ′ to Pβ′. Thus, we can conclude that in our experiments at low temperatures the bilayers are in the Lβ′ (DMPC, DPPC) or Lβ (SM) gel state.

Discussion

In an effort to understand why lipid bilayers in the gel phase are solubilized at lower detergent concentrations than those in the fluid phase, Arnulphi et al. (5) measured ΔG of Triton X-100 partitioning into pure SM or pure DPPC bilayers at 20 and 50°C, i.e., when the lipid was respectively in the gel and fluid phase. Surfactant concentration was kept well below those causing solubilization, so that, under their conditions, detergent partitioning to bilayers was observed in the absence of solubilization events. The main result of Arnulphi et al. was that ΔG of partition was independent of temperature, i.e., it was the same in the gel and fluid phases, for SM as well as for DPPC. Consequently, the observed preferential solubilization of gel over fluid bilayers (7–11) could not be attributed to a preferential partitioning of Triton X-100 in the gel phase.

In the well-known description of the solubilization stages by Helenius and Simons (1) stage I corresponded to the early steps of detergent partitioning into the bilayer, whereas stage II included bilayer-to-micelle transition, i.e., solubilization. The transition from stage I to stage II was thus marked by the saturation of lipid bilayers with detergent, and by the onset of lipid-detergent mixed micelle formation. The transition point could be experimentally determined as the detergent concentration at which solubilization was first observed. Measuring accurately the onset of solubilization had in our case two difficulties. One was the well-known problem of the multiple equilibria of Triton X-100 between monomers, micelles, and bilayer-bound forms, that complicate the measurement of effective, i.e., in the bilayer detergent/lipid ratios. This was circumvented by the procedure described by Schurtenberger et al. (12) and widely used by Lichtenberg and co-workers (2), consisting of determining the onset of solubilization at various lipid concentrations, and taking the slope of the total detergent concentration producing onset of solubilization versus lipid concentration straight line as the effective detergent/lipid ratio causing bilayer saturation with detergent, R_e^sat.

Another, less frequent, problem was posed by the property of Triton X-100 to self-organize into large aggregates above a certain temperature, cloud point, so that the suspension acquires a milky aspect and measurements of light scattering become impossible. The cloud point temperature decreases with the presence of lipid vesicles, so that, for SM vesicle suspensions at 50°C, the addition of even tiny amounts of Triton X-100 lead to extensive cloud formation. The problem was circumvented by the use of an alternative method of solubilization assay, based on the changes in the absorption spectrum of MC540 when this dye is transferred to a micellar environment (15). This procedure is not affected by turbidity changes, thus it can be used at 50°C in a convenient way. Because of the low sensitivity of turbidity methods to detect micelles the assignment of turbidity maxima as the onset of solubilization is not always correct (28), but may rather reflect a R_e^sat value higher than real. However, the identity of the results obtained by the turbidity and the MC540 methods (Fig. 1 C) supports the idea that turbidity measurements are reliable in our system.

The results in Fig. 2 show that, irrespective of lipid concentration (in the 0.3–2 mM range), more detergent is required to start SM bilayer solubilization at 50°C than at 20°C. The ratio R_e^sat (50°C)/R_e^sat (20°C) for pure SM is ≈2. These results are independently confirmed by the ³¹P-NMR method (Fig. 3). Moreover, the data in Fig. 4 show the effect of temperature and phase structure on the different bilayer saturability in detergents. Indeed, the temperature-induced micelle-to-vesicle transition is a well-known phenomenon (29–32) whose significance is broadened here in the context of phase-dependent membrane saturation by detergents.

We have included in Table 1, for each system under study, the R_e^sol values, i.e., the effective detergent/lipid ratios above which all the lipid occurs in the form of lipid-detergent mixed micelles. These values were obtained following the same procedures as for R_e^sat. Comparable data in the literature are scarce and usually scattered. Heerklotz (3) has collected many of them. In particular, we have been unable to find previously published data for the onset of solubilization of bilayers in both the gel and fluid states for the same lipid composition, except for the data by Schnitzer et al. (9) for DPPC, however in their case the lipid was in the form of sonicated vesicles, and the detergent effects on this kind of vesicles are particularly complex (33). Nyholm and Slotte (34) estimated R_e^sat = 0.2 for the C₁₆SM/Triton X-100 system, and R_e^sat = 0.29 for DPPC/Triton X-100. The former value, but not the latter one, is in agreement with our data in Table 1. In fact, Patra et al. (7) and Schnitzer et al. (9) had shown that, both above and below the gel-fluid transition temperature, R_e^sat and R_e^sol may exhibit large temperature-dependent changes in a narrow range of temperature. The previous authors (7,9) had observed that a minimum detergent concentration was required to solubilize bilayers close to their gel-fluid temperature. Schnitzer et al. (9) interpret this as a result of two opposing factors, namely the lipid spontaneous curvature and void energy contributions. Thus, the observed differences between laboratories may be due to small operational changes and the absolute figures of the values for DPPC in Table 1 are best interpreted within the context of the present study. R_e^sat values for egg PC estimated by different authors range between 0.64 and 0.88 (35–37) and Table 1. Comparison of R_e^sat and R_e^sol values for a given system (Table 1) does not lead to an easily recognizable pattern. In several cases R_e^sol is ∼threefold R_e^sat, e.g., SM (gel), DMPC (fluid), or egg PC (gel and fluid). However, for DPPC (gel and fluid) R_e^sol/R_e^sat ≈1.3, whereas the corresponding value for SM (fluid) or DMPC (gel) is closer to ≈6. This means that the different saturability of the bilayers by detergent is probably not the only factor explaining why (full) solubilization of bilayers in the fluid and gel states can require very different amounts of detergents.

Numerous detergents including Triton X-100 cause a decrease in the gel-fluid transition temperature of phospholipids (freezing point depression) (5,25,26,38). In principle this should lead to a considerable decrease in the partition constant when going from the fluid to the gel phase, however this is not observed, or the changes are not significant enough, when nonionic surfactant are used (5,29). Currently, the origin of the different saturabilities of gel and fluid phases is unclear. One possibility is that the detergent does not mix well with lipids in the gel phase, so that it accumulates at lattice defects and fragments of gel bilayer surrounded by detergent instead of true micelles (i.e., more or less spherical structures containing molecular mixtures of lipid and detergent) are formed. Such a possibility was suggested for the solubilization of Halobacterium purple membranes that exist in the bilayer crystalline state (39). This would mean that saturation occurs through a different mechanism in the fluid and in the gel phases, as supported by the data in Funari et al. (22). This would also be illustrated by Fig. 3 (lower spectrum) in the present work, in which no lamellar patterns are detected, although there should still be plenty of the coexisting lamellar phases. In light of this one could immediately predict that R_e^sat of a gel phase would be considerably lower than that of a fluid phase, as the results in this work have quantitatively shown. Moreover, it is possible that in the fluid state the lipid molecules are flexible enough to compensate for the different intrinsic curvature of the detergent, and that this allows the bilayer to accommodate a higher proportion of detergent before breaking down into mixed micelles. These ideas should be susceptible to experimental testing.

Table 1 also contains D_w data, i.e., the equilibrium concentration of detergent in buffer. These values are obtained from the intercept of the straight lines in Figs. 2 and 4 at [lipid] = 0 (12). In all cases D_w data are below the critical micellar concentration of the detergent (≈0.25 mM), in agreement with data in the literature (2,3).

In summary, the previous results are conclusive in showing that different detergent concentrations are required to saturate lipid bilayers in the gel and fluid state. This is an explanation, but not necessarily the only one, of the fact that bilayers in the gel state are solubilized by lower detergent concentrations than those in the fluid lamellar phase. Further investigations will be required to examine the micellization steps in the solubilization of gel and fluid bilayers to shed further light on this phenomenon.