Heating-Induced DMPC/Glycyrrhizin Bicelle-to-Vesicle Transition: A X-Ray Contrast Variation Study

Abstract

In this study, we investigated the conversion of lipid bicelles into vesicles in the case of a system composed of the phospholipid 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) and the saponin glycyrrhizin in the presence of sucrose. Glycyrrhizin is a biosurfactant present in the licorice root and possesses a triterpenic hydrophobic backbone and a hydrophilic headgroup built from two sugar molecules. The aim of this study is to determine the initial bicelle size at temperatures below the lipid’s main phase transition temperature T_m and, based on these results, characteristics of the temperature-induced bicelle-to-vesicle transition. Moreover, the influence of the heating rate on this transition is followed. The general picture concluded from photon correlation spectroscopy and small angle X-ray scattering was confirmed by additional imaging with cryogenic transmission electron microscopy. Small angle X-ray scattering was especially used to determine size parameters of the existing structures. To enhance the contrast for X-rays, a buffer containing 25 wt% sucrose was used. It was found that larger vesicles were formed from smaller precursor particles and that monodisperse precursors are required for formation of very monodisperse vesicles upon temperature increase. At high glycyrrhizin contents and above a critical heating rate of ∼5°C min⁻¹, the polydispersity of these vesicles is decoupled from both parameters, glycyrrhizin content and heating rate. However, the vesicle size stays tunable by the glycyrrhizin content and increases upon increasing the glycyrrhizin concentration. Therefore, vesicles of defined size and with a rather low polydispersity of ∼12–14% can be formed.

Significance

Bicelles or nanodisks are frequently used for structure determination of membrane proteins via, e.g., NMR spectroscopy. In this study, we report on bicelle formation in mixtures of a phospholipid (1,2-dimyristoyl-sn-glycero-3-phosphocholine) and the saponin glycyrrhizin, which is the rim-stabilizing compound. To manipulate the X-ray scattering contrast, the study is performed in the presence of 25 wt% sucrose. After determination of the bicelle size at temperatures below the lipid’s main phase transition temperature (T_m), we report on the fully reversible, heating-induced reorganization of these bicelles into well-defined unilamellar vesicles at T > T_m. Therewith, we highlight the possibility of producing vesicles with defined size and polydispersity by choosing an appropriate initial bicelle size, heating rate, and buffer composition.

Introduction

Amphiphiles with different shapes tend to self-assemble in aqueous solution, but the shape of the formed aggregates strongly depends on the geometrical properties of the amphiphile (1). Because the molecular shape of phospholipids is rather cylindrical, these molecules tend to assemble in flat structures. Most classical detergents have a rather conical structure and therefore tend to form aggregates along curved surfaces (1, 2, 3). If both kinds of amphiphiles are mixed, depending on their ratio, different kinds of structures occur. At high phospholipid share, lamellar aggregates are present, which transform with increasing detergent share into micellar aggregates (1,4). In the transition region between both defined structures (vesicles and micelles), a structure involving characteristics of both lamellar and micellar structures occur (5,6). These structures are usually called bicelles (7, 8, 9) but sometimes also nanodisks (10). The latter name is rather common for similar structures built by phospholipids surrounded by specific proteins (11, 12, 13). In bicellar structures, long-chain lipids form a discrete lipid bilayer, which is surrounded by a belt of detergent molecules (or short-chain lipids) (8,14). This belt protects the hydrophobic lipid chains from the hydrophilic solvent. Bicelles are self-assembled structures and are therefore intrinsically monodisperse (15). Moreover, bicelles are alignable in an external magnetic field under certain conditions (7,15,16). Proteins can be incorporated into the aligned bicelles and therefore gain a net molecular alignment. This property is, e.g., used for protein structure determination via NMR because the resolution can be highly improved in comparison with randomly oriented samples. Bicelles, moreover, mimic a quite realistic membrane because of the intact lipid bilayer incorporating the protein (7,8).

The existence of bicellar structures is dependent on the lipid/detergent ratio (4,6). Because of the different free molecule concentrations of the lipid and the surfactant in solution, dilution causes a removal of the detergent from the bicellar rim, and a structural transition into vesicular structures occurs (1,17,18). Removal of surfactant leads to an increase in the rim line tension, whereby coalescence of bicelles, and therefore a structural growth, is induced. Folding of larger lamellar sheets can, in the following, lead to formation of closed vesicles (19). Besides dilution, this transition can be induced by surfactant extraction or temperature variation (17). The method of temperature variation is especially suitable to study the characteristics of the underlying transition because of the reversibility of the process. Other groups found a strong dependence of the structural transition on the kinetics of the surfactant removal, which is dependent on the heating rate. It was found that an increase in heating rate leads to more monodisperse vesicles, whereas at lower rates, the formation of extended structures, e.g., lipid sheets, is favored (1,17,20). Moreover, it was observed that smaller and therefore more stable bicelles form bigger vesicles because folding into a vesicle occurs at a larger size of the vesicle sheet (19).

The system investigated in this study is composed of the phospholipid 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) and the saponin glycyrrhizin in the presence of 25 wt% sucrose in the aqueous solution. The molecular structures of both compounds, the lipid and the saponin, are depicted in Fig. 1. Saponins are natural surfactants that share the same basic structure and include a large variety of different molecules (21, 22, 23). Because of their amphiphilicity, saponins have various applications in medicine and industry (21,22,24,25). The hydrophobic backbone of saponins is built by a steroid or triterpene, to which a different number of hydrophilic sugar chains is attached (21,26). In the case of glycyrrhizin, which is more correctly called glycyrrhizinic acid (27), the hydrophobic backbone is built by a triterpene called glycyrrhetinic acid (28,29). At the C3 position, a sugar chain with two glucuronic acid molecules is bound (see Fig. 1). The acidic group at the C20 position of glycyrrhetinic acid significantly influences the amphiphilicity of the whole molecule depending on the pH value. Whereas glycyrrhizin exhibits a clear critical micelle concentration up to pH 6, at a neutral pH value, glycyrrhizin does not self-assemble in discrete aggregates (28). This is most probably due to the deprotonated acidic groups bound to the backbone at opposite sites, which cause the loss of the clear amphiphilic structure and moreover induce repulsion effects between different glycyrrhizin molecules (30). Whereas glycyrrhizin is badly soluble at low pH value, the solubility is strongly increased at neutral pH value (28). In water, the formation of fibrillar networks with a uniform thickness of 25 Å was found (31).

Figure 1.

Molecular structures of (a) the phospholipid DMPC and (b) the saponin glycyrrhizin. The hydrophilic molecular parts are shown in purple, and the hydrophobic part in black. To see this figure in color, go online.

As naturally occurring substance glycyrrhizin can be extracted from the roots of Glycyrrhiza glabra, also known as licorice (32). It is 30–50 times sweeter than glucose (27,29,33). Moreover, it exhibits a low toxicity and is therefore used as a sweetener. However, the recommended daily consumption is less than 0.229 mg glycyrrhizin/kg body weight/day (34). In addition, glycyrrhizin is known to have several pharmacological actions, such as antiinflammatory, antimicrobial and antiviral, antioxidative, and antitumor activity (29,35). Also, the enhancement of therapeutic effects of various drugs by the formation of inclusion complexes is known (36). De Groot and Müller-Goymann described the surface activity of glycyrrhizin as quite low compared to other saponins (37), and Wojciechowski et al. found only weak interactions with lipid membranes (38,39). On the contrary, MD simulation for a 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) bilayer by Selyutina et al. suggested that glycyrrhizin most likely gets incorporated into the lipid bilayer and induces thinning of the lipid membrane (40). Furthermore, for erythrocyte cells, an incorporation and, at higher glycyrrhizin concentrations, an extraction of cholesterol from the cell membrane was demonstrated. The membrane elasticity was strongly altered even in the presence of low amounts of glycyrrhizin (36). The enhancement of the effect of drugs has been attributed to a higher permeability of the cell membranes, e.g., by pore formation induced by glycyrrhizin (36,41). Pore formation was not confirmed by a MD simulation study conducted by Shelepova et al. (42), and most likely, a different mechanism has to be the reason for the enhanced membrane permeability.

We expect that addition of high shares of glycyrrhizin (around a lipid/saponin ratio of 1:1) causes a solubilization of the DMPC lipid membrane into smaller bicellar structures. Therefore, most likely, DMPC molecules form the lipid bilayer of the bicellar structures, and glycyrrhizin rim stabilizes this bilayer. The bicelle formation of a system involving DMPC and the saponin aescin was reported in previous studies (43, 44, 45), but in general, bicelle formation involving saponins is quite unexplored. A system reported in literature, which comes closest to the system investigated in this work, involves different phospholipids and bile salts like (tauro)cholate, glycochenodeoxycholic acid or 3-[(3-cholamidopropyl)dimethylammonio]-2-hydroxy-1-propanesulfonate (17,20,46,47). Bile salts and saponins share a similar cyclic hydrophobic backbone, but because of the sugar groups attached, saponins possess a much larger headgroup. In these investigated systems, at high bile salt concentration, disk-like structures were found. By dilution-induced removal of bile salt molecules from the bicelle rim, a transformation into either extended lamellar sheets or closed vesicles, which might show perforations coated by bile salt molecules, takes place (47,48).

A quite high amount of the sugar sucrose (25 wt%) is added to the lipid-saponin system in this work. Addition of sucrose to the continuous phase changes the electron density and therefore the contrast of the solvent. Therefore, a manipulation of the X-ray scattering contrast can be achieved, and, depending on the sugar amount, different features of the scattering structure can be highlighted (49,50). In the case of the lipid system investigated in this work, an enhancement of the contrast between the lipid membrane and the solvent is expected upon sucrose addition (17,51,52). According to the literature, addition of high amounts of sugars can lead to modified lipid membrane properties. Among others, a change in the spontaneous curvature of liposomes (53) and a reduction of the membrane bending modulus (54) are reported after addition of sugars. Morandi et al. recently investigated the effect of high sucrose concentrations (up to 1.5 M) on lipid bilayers composed of DPPC (55). They showed that the main phase transition temperature of the lipid (T_m) is only slightly influenced and that the main influence is detected at T > T_m. Whereas at T < T_m, the sucrose molecules stay excluded from the bilayer’s interfacial region, they can penetrate into it at T > T_m (55). For the lipid DMPC, similar results were found by Fabrie et al. (56). Moreover, Kiselev et al. reported that the polydispersity of DMPC small unilamellar vesicles (SUVs) decreases in the presence of sugar (57). It should be mentioned that the structural picture of the interaction of sugar and lipid is not fully understood and that different, partially contradicting results are presented in the literature. Because the temperature-induced reorganization of bicelles to vesicles was successfully studied by Lesieur et al. in the presence of sucrose (40 wt%) before, conservation of the general process upon sucrose addition is expected (17).

Our work focuses on the completely reversible temperature-induced bicelle-to-vesicle transition of mixtures of DMPC and glycyrrhizin in the presence of a 25 wt% sucrose buffer solution. Before the mentioned transition is focused, the behavior of sucrose-free and sucrose-containing samples with similar glycyrrhizin content is briefly compared. At low temperature, sucrose does not change the obtained structures and indeed serves only as a contrast-changing agent. However, at high temperature, the structure formation is changed by the sucrose. In the sucrose-containing system, the influence of the heating rate, especially on the resulting structures, is investigated upon heating from low (10°C) to high temperature (50°C). At low temperature (T < T_m), the formation of small bicellar structures is suggested. Increasing the temperature well above T_m, which is around 24°C for DMPC (43), most likely leads to formation of closed vesicles. The overall size of the structures involved at low and high temperature is determined by photon correlation spectroscopy (PCS) and the general structural picture is confirmed by cryogenic transmission electron microscopy (cryo-TEM). The shape, size, and polydispersity of the resulting structures are investigated by small angle X-ray scattering (SAXS). Addition of sucrose increased the scattering contrast significantly and therefore drastically shortened the measurement time.

Materials and Methods

Chemicals and sample preparation

The phospholipid DMPC was obtained from Lipoid (Ludwigshafen, Germany; ≥99% purity). The saponin glycyrrhizin (used as ammonium salt, ≥95%, Chemical Abstracts Service (CAS) number: 53956-04-0), and chloroform were purchased from Sigma-Aldrich (Munich, Germany). The sugar sucrose was obtained from Merck Chemicals (Darmstadt, Germany).

All samples were prepared in a 50 mM phosphate buffer with a sucrose amount of 25 wt% (c_sucrose ≈ 0.8 M). For comparison, samples in a pure buffer solution in the absence of sucrose were also prepared. The pH value of the buffer was adjusted to a value of 7.4. The lipid mass concentration in all samples was fixed to a value of 15 g ⋅ L⁻¹. Glycyrrhizin contents x_glycyrrhizin between 35 and 70 mol% were investigated. The glycyrrhizin content is thereby defined with respect to the lipid concentration in solution (x_glycyrrhizin = n_glycyrrhizin/(n_DMPC + n_glycyrrhizin)).

For the preparation of DMPC-glycyrrhizin mixtures with different glycyrrhizin contents, first, DMPC was dissolved in chloroform and dried afterwards using a rotary evaporator to gain a thin lipid film. To remove chloroform residues, the lipid film was stored over night at 60°C. Afterwards, the lipid film was rehydrated with a sucrose-containing aqueous glycyrrhizin buffer solution at the final concentration. Dissolving glycyrrhizin in water decreases the pH value significantly, and the solubility at low pH values is bad. The solubility of glycyrrhizin is increased by adding a small amount of a concentrated sodium hydroxide solution, and therefore, the pH value of the glycyrrhizin solution was adjusted to a value of 7.4. The samples were subjected to five consecutive freeze-thaw cycles (in liquid nitrogen and warm water) and were afterwards frozen for storage.

Cryo-TEM

Cryo-TEM was used to get a real space impression of the structures formed by DMPC and glycyrrhizin in the presence of a sucrose-containing buffer. Therefore, the sample with 70 mol% glycyrrhizin was maintained at temperatures of 15 and 50°C before freezing and imaged afterwards. A JEOL JEM-2200FS electron microscope (JEOL, Freising, Germany) equipped with a cold field emission electron gun was used for cryo-TEM imaging. The sample was applied to a grid by a Leica blotting and plunging device operated at the respective temperature (Leica EM GP; Leica Mikrosysteme Vertrieb, Wetzlar, Germany). For freezing, the samples were plunged into liquid ethane, which was cooled with liquid nitrogen to achieve sufficiently fast cooling. The grids were afterwards transferred to a cryo transfer and tomography holder (Fischione Model 2550; E.A. Fischione Instruments, Pittsburgh, PA). The cryo-TEM images were recorded digitally by a bottom-mounted camera (Gatan OneView; Gatan, Pleasanton) and processed with a digital imaging processing system (Digital Micrograph, Version 3.21, GMS 3; Gatan, Pleasanton, CA). For the image of the sample maintained at 15°C before freezing, the microscope was operated at an acceleration voltage of 200 kV, and the sample was applied to Lacey Carbon Film coated grids (200 Mesh, Cu; Science Services, Munich, Germany). At a temperature of 50°C, the microscope was operated at an acceleration voltage of 80 kV, and TEM holey carbon grids (Qantifoil R2/1, 200 mesh; Plano, Wetzlar, Germany) were used.

PCS

By PCS, the size of particles diffusing in solution can be obtained. Therefore, the time-dependent intensity change of light scattered by a solution of particles at a specific angle of 2θ is recorded. Usually, the scattering angle is converted to the scattering vector q, which allows us to bring different scattering experiments to a common scale (Eq. 1).

$$q=\frac{4\pi n}{\lambda }\times \text{sin}\left(\theta \right)$$ (1)

In this equation, n stands for the refractive index, and λ is the wavelength of the radiation used. The time-dependent intensity fluctuation is analyzed by generation of the time intensity autocorrelation function. This function is converted into the respective field correlation function of the scattered light. In the simplest case, this function decays as a single exponential, and the relaxation rate Γ_T(q) directly relates the diffusion coefficient of the particle in solution D_T, with the square of the scattering vector q² (Eq. 2; (58)).

$${\Gamma }_{\text{T}\left(\text{q}\right)}=D_{\text{T}}{q}^2$$ (2)

The diffusion coefficient D_T is converted to the hydrodynamic radius R_H of the particle by the Stokes-Einstein equation (Eq. 3). R_H describes the size of a spherical particle, including a solvent shell with the same diffusive properties as the scattering object (58). For a correct calculation of R_H from D_T, the absolute temperature T and especially the dynamic viscosity of the solvent η have to be taken into account. k_B is the Boltzmann constant.

$$D_T=\frac{k_{B}T}{6\pi \eta R_H}$$ (3)

PCS measurements were performed to investigate the glycyrrhizin content- and temperature-dependent evolution of the hydrodynamic radius R_H. The samples were therefore measured at low temperature (10°C) and at high temperature (50°C). All samples were stored at a temperature of 4°C before use. For the measurement at high temperature, the samples were directly placed in the decaline bath of the PCS setup, which was preheated to a temperature of 50°C. This results in a very fast heating of the sample. The samples were measured in conventional NMR tubes and centrifuged at 10,000 rpm for 20 min before filling to remove dust particles or aggregates from the sample.

A 3D light scattering goniometer setup (LS Instruments, Fribourg, Switzerland) equipped with an HeNe laser with λ = 632.8 nm (1145P; JDSU, Milpitas, CA) was used for the PCS measurements. The scattered light intensity was recorded in modulated three-dimensional correlation mode. The autocorrelation function was generated with a multiple-τ-digital correlator (correlator.com) and recorded for scattering angles between 40 and 110° in steps of 5°. The obtained data were analyzed with the program CONTIN (59).

To correctly determine the hydrodynamic radius R_H from the diffusion coefficient D_T by Eq. 3, the viscosity of the sucrose-containing buffer was determined at both temperatures, 10 and 50°C, using an Ubbelohde viscosimeter with a capillary constant of k = 0.03174 m²s⁻² (Schott Geräte, Mainz, Germany). The obtained kinematic viscosity multiplied with the density yields the dynamic viscosity (η_buffer,10°C = 3564 μPa ⋅ s and η_buffer,50°C = 1289 μPa ⋅ s). The density itself was measured with a DMA 4500 density meter from Anton Paar (Graz, Austria).

SAXS

SAXS is often used to determine structural parameters of vesicular and bicellar structures (17,20,47,60,61). The head-tail contrast normally seen by X-rays thereby allows us to accurately determine the proportions of a lipid’s head and tail part. In general, the total scattering intensity of a particle depending on the scattering vector magnitude q at an angle of 2θ (see Eq. 1) is given by

$$I\left(q\right)=N\times {\left(\Delta XSLD\right)}^2\times V^2\times P\left(q\right)\times S\left(q\right).$$ (4)

Here, N describes the number of particles, V the scattering volume of the particle, P(q) the form factor of the particles, and S(q) the structure factor of the solution. ΔXSLD = XSLD_particle − XSLD_solvent is the scattering length density difference between the particle and the solvent, which for X-rays is mostly dependent on the electron density of the particles investigated. In this case, the membrane’s scattering length density profile is modified by the presence of sucrose in the aqueous buffer. A conventional head-tail contrast is not present anymore because the scattering length density of the surrounding solvent is matched with the lipid’s head. This phenomenon is presented in Fig. S5 for DMPC SUVs in aqueous buffer in the presence and absence of 25 wt% sucrose.

From small angle scattering data, structural information about the scattering object can be obtained by a variety of methods. Here, general information about the overall size and shape of the structures present under varying conditions is first obtained using the indirect Fourier transformation (IFT) method. This technique was introduced by O. Glatter and is a model-independent method (62,63). It yields the pair distance distribution function (p(r)), whose shape is related to the shape of the scattering particle. In this method, the scattering data are fitted by a set of Fourier-transformed cubic spline functions by which the p(r) function is approximated (64). Additionally, the radius of gyration of the particle R_G can be obtained from the p(r) function (see Eq. 5).

$$R_G=\sqrt{\frac{\int r^{2}p\left(r\right)dr}{2\int p\left(r\right)dr}}$$ (5)

Besides the model-independent IFT approach, model-dependent fitting can yield specific structural parameters of the scattering particles. This requires a general idea of the shape of the structures involved. In this work, a cylinder model (used to describe a disk-like shape) as well as a core-shell sphere model, both implemented in the program SASView (65), are used. Equations describing the single models can be found in the Supporting Materials and Methods and in the SASView documentation. Each model requires information about the scattering length density XSLD of the scattering particle as well as the solvent, which is calculated on the basis of their molecular volumes and element composition (see Results and Discussion) (66).

By SAXS, the influence of different heating rates on the temperature-induced transition in DMPC-glycyrrhizin mixtures was investigated. The samples were measured on an in-house SAXS system (XEUSS; Xenocs, Sassenage, France) equipped with a CuK_α source (λ = 1.541 Å, GeniX Ultra Low Divergence; Xenocs) and a Pilatus 300K hybrid pixel detector (Dectris, Baden Deattwil, Switzerland). A sample-to-detector distance of 2.7 m was used to cover a q-range from ∼6 × 10⁻³ to 0.12 Å⁻¹. The samples were measured at 10 and 50°C, whereas for heating from 10 to 50°C, different heating rates between k = 30°C min⁻¹ and k = 0.5°C min⁻¹ were used. Data acquisition time was around 2–3 h per temperature. The two-dimensional data were analyzed using the Foxtrot software (V3.3.4.) (67). The samples were measured in a flow-through Kapton capillary (1 mm; GoodFellow, Bad Nauheim, Germany) positioned in a Linkam stage (Linkam Scientific, Tadworth, UK). The scattering of the sample was normalized with respect to incident intensity, sample thickness, acquisition time, transmission, and background. The data were brought to absolute scale using glassy carbon type 2 as standard (68). After normalization, the data were treated by the dynamic rebin formalism implemented in SAXSutilities to improve statistics at high q-values (minimal steps: 1, minimal Δq: 0.005 Å⁻¹ (69)).

Results and Discussion

General phase behavior and influence of added sucrose on the DMPC-glycyrrhizin system

In this study, the structural transitions of mixtures of DMPC and glycyrrhizin in a concentration range from 35 to 70 mol% were investigated. Samples with the respective glycyrrhizin amounts were prepared in the absence and presence of 25 wt% of the sugar sucrose. The sugar was added to enhance the scattering contrast for X-rays. First, the visual appearance of samples with different glycyrrhizin contents (between x_glycyrrhizin = 20 and 70 mol%) is depicted in Fig. S1 a in the absence and Fig. S1 b in the presence of sucrose at temperatures of 4 and 50°C (after fast heating and subsequent cooling). Optical transparent samples are obtained at glycyrrhizin content higher than 35 mol% in the absence and higher than 30 mol% glycyrrhizin in the presence of sucrose. This shift in phase boundary directly indicates an influence of added sucrose on the DMPC membrane solubilization properties of the saponin glycyrrhizin. In the presence of sucrose, samples with glycyrrhizin contents of 35 and 40 mol% glycyrrhizin still exhibit a slightly bluish color, indicating incomplete conversion of the lipid bilayer to very small, monodisperse, and self-assembled bicelles. Irrespective of the presence of sucrose, all samples are bluish at high temperature (50°C), and a temperature-driven structural rearrangement also remains in the presence of sucrose in the bulk solution. Similar appearance of all samples after cooling down again to 4°C indicates complete reversibility of this structural rearrangement. This reversibility is further proven by SAXS measurements (see Fig. S8) and is discussed in detail later in this work. Only samples with x_glycyrrhizin ≥ 35 mol% are considered in this work because in this study, the temperature-driven dependence of the structural rearrangement on the initial bicelle size in the presence of sucrose shall be investigated.

Before the structural rearrangement in the presence of sucrose is focused, the influence of added sucrose to the system will be elucidated. As indicated by the optical appearance of the samples, the DMPC membrane gets completely solubilized by the addition of high amounts of glycyrrhizin. The size of the resulting self-assembled structures was estimated by PCS in the absence and presence of sucrose for x_glycyrrhizin ≥ 35 mol% at a temperature of 10°C, well below the lipid’s main phase transition temperature T_m. The corresponding hydrodynamic radii R_H are depicted in Fig. S2, and the radii obtained in the presence of sucrose are additionally shown in Fig. 2. Also, this parameter expresses a shift in the solubilization boundary. Whereas for the sample with a glycyrrhizin content of 40 mol%, an R_H-value of 1191 ± 42 Å is obtained in the absence of sucrose, with 259 ± 3 Å, this value is already much smaller in the presence of sucrose. At glycyrrhizin contents higher than 40 mol%, more similar R_H-values are obtained in the presence and absence of sucrose. In the absence of sucrose, the bicellar structures appear slightly smaller, with R_H-values between 118 ± 1 and 121 ± 1 Å, in comparison with the samples with sucrose, with R_H-values between 129 ± 2 and 141 ± 1 Å. Hence, at a temperature well below T_m, an influence of added sucrose to the buffer solution is visible, but the general aspect of membrane solubilization by the addition of high amounts of glycyrrhizin is maintained.

Figure 2.

R_H-values for mixtures of DMPC with different contents of glycyrrhizin at 10 and 50°C in the presence of 25 wt% sucrose. Whereas at low temperature, R_H decreases with increasing x_glycyrrhizin because of a higher amount of saponin, this behavior is inverted at high temperature. To see this figure in color, go online.

As a basis for the elucidation of the structural rearrangements in the sucrose-containing DMPC-glycyrrhizin system, the T_m-value of pure DMPC SUVs was determined in the presence and absence of sucrose. Such vesicles with a defined size were produced by extrusion through a membrane with a pore size of 500 Å. The temperature-induced density and refractive index change of the lipid membrane was used to obtain T_m. The respective recorded optical densities as a function of temperature are shown in Fig. S3 a. After referencing to the constant contributions of the gel-like β- and the liquid-crystalline α-phase, a degree of conversion Θ(T) is calculated (Fig. S3 b). From this function, T_m-values for samples in the absence and presence of sucrose were calculated by derivation (Fig. S3 c). Values of $T_{\text{m},{\text{H}}_2\text{O}}$ = 24.6 ± 0.1°C in the absence and $T_{\text{m},{\text{H}}_2\text{O},25\text{wt\%}\text{sucrose}}$ = 24.1 ± 0.1°C in the presence of 25 wt% were acquired. Hence, no significant influence of the addition of sucrose in the respective amount is observed. In literature, an increase of the value of T_m by ∼2°C by addition of 60 wt% sucrose is reported (52,57) However, other studies report that the addition of sucrose does not have a significant impact on T_m of the lipid (55,56). In this study, too, no significant influence of the addition of sucrose on DMPC’s T_m was observed, and the phase transition of DMPC seems to be mainly unaffected by the addition of the sugar.

To finally determine the influence of added sucrose at temperatures above T_m of the lipid system, exemplary SAXS data were recorded for a sample containing a glycyrrhizin content of 50 mol% in a temperature range between 10 and 50°C in steps of 5°C. The corresponding data are shown in Fig. S4 a in the absence and in Fig. S4 b in the presence of sucrose. A prominent membrane-contrast-altering effect of the addition of sucrose becomes visible. This effect is highlighted in Fig. S5, in which the scattering profiles of unilamellar DMPC vesicles with defined size produced by extrusion are compared in the absence and presence of 25 wt% sucrose. Sucrose molecules assemble near the lipid headgroups (55) and, in glycyrrhizin-containing samples, presumably also in the region of the glycyrrhizin headgroups, whereby the head-to-tail contrast of the lipid membrane gets lost and the continuous phase (the solvent) adopts the XSLD of the lipid’s headgroup parts. For this reason, a clear minimum in the sucrose-containing sample is not visible, and the lipid membrane can be described by a one-shell contrast.

For the data presented in Fig. S4, at low temperatures below the lipid’s T_m, scattering curves characteristic for small, bicellar structures are obtained in the absence of sucrose in the buffer solution (43,44). Also, the curves obtained in the presence of sucrose are characteristic for bicellar structures, as further elaborated in this work (see Fig. 5). Upon heating of the samples, significant structural changes are observed at temperatures above T_m in both cases: the presence and absence of sucrose. In the absence of sucrose, the system tends to form correlated membrane structures at temperatures above 25°C, which become visible from the correlation signal appearing at a q-value of ∼0.1 Å⁻¹. This corresponds to a lamellar repeat distance of ∼60 Å, if the relation d = 2πq⁻¹ is used to calculate this distance. Therefore, this system is quite similar to a system composed of DMPC as the lipid and the saponin aescin (66). Also there, at certain aescin contents, the formation of correlated membrane structures is observed at temperatures well above the lipid’s T_m. The formation of correlated membrane structures was explained to arise from a strong interaction of the aescin’s headgroups, which facilitates a bridge formation between two separated membrane fragments (66). This exact bridge formation seems not to be possible anymore in the sucrose-containing system, which, at temperatures well above the lipid’s T_m, forms closed unilamellar vesicles instead (see Fig. S4 b). The existence of these unilamellar vesicles is further elucidated by cryo-TEM (see Fig. 3) and SAXS (see Fig. 7) in the following sections of the manuscript. We explain this observation by unspecific interactions of the sugar residues of the glycyrrhizin molecules with the sucrose molecules in the bulk. Therefore, specific interactions between the glycyrrhizin residues are avoided, which leads to the formation of structures other than correlated lipid membranes.

Figure 5.

(a) SAXS data of samples with different glycyrrhizin contents x_glycyrrhizin at a temperature of 10°C with fits with a cylinder model from SASView (65). (b) Fit parameters for the radius R_cylinder and the height of the cylinder H_cylinder are shown. To see this figure in color, go online.

Figure 3.

Cryo-TEM images of a sample containing 70 mol% glycyrrhizin maintained at temperatures of (a) 15°C and (b) 50°C before freezing in the presence of 25 wt% sucrose in the buffer. Whereas the sample appears disk-like at low temperature, vesicular structures are found at high temperature. To see this figure in color, go online.

Figure 7.

Normalized p(r) functions obtained from IFT model fits to the SAXS data measured at a temperature of 50°C after heating samples with different glycyrrhizin contents with rates of (a) k = 0.5°C min⁻¹ and (b) k = 30°C min⁻¹. To see this figure in color, go online.

Structural rearrangement of DMPC-glycyrrhizin aggregates in the presence of sucrose

Sample characterization with cryo-TEM

The SAXS data recorded for the sample with 50 mol% sucrose already indicated certain particle shapes of the underlying self-assembled DMPC-glycyrrhizin structures, and cryo-TEM was used to visualize these structures at both low and high temperatures. Therefore, the sample with a glycyrrhizin content of 70 mol% was maintained at temperatures of 15 and 50°C, frozen, and imaged afterwards. Pictures for both temperatures are presented in Fig. 3. The structures present at a temperature of 15°C should be comparable to the ones present at 10°C because a significant structural transition is only expected at temperatures well above the lipid’s T_m (compare Fig. S4). At this temperature, monodisperse and disk-like particles with an average thickness of 53 ± 21 Å and an average length of 208 ± 50 Å are identified and highlighted by red boxes in Fig. 3 a. Distributions of both size parameters are shown in Fig. S6, a and b, and the size parameters reported before result from approximation with Gaussian functions. Cryo-TEM images of DMPC/dihexanoylphosphatidylcholine (DHPC) bicelles by other groups (7,70) look similar, and therefore, the presence of bicellar structures in this case is verified.

At 50°C (T > T_m) (Fig. 3 b), the structures observed can be identified as unilamellar vesicles with an average diameter of 208 ± 85 Å. Again, this average size parameter was determined by fitting a Gaussian to the diameter distribution (see Fig. S6 c). The vesicles appear quite polydisperse, and some of them show deformations and are not completely spherical. However, the latter observation might be an artifact due to the presence of the high sugar amount in the buffer. Because of the usage of a transmission method, the specific polydispersity corresponds to the actual particle distribution. Additionally, the polydispersity will, in the following, be analyzed by SAXS. Similar results concerning the aggregate size and polydispersity are not necessarily expected because the process of maintaining the sample at a high temperature and freezing the small sample volume afterwards is prone to artifacts like, for instance, evaporation effects. Nevertheless, cryo-TEM confirmed the structural picture of small, disk-like structures at low temperature (T < T_m) that convert into vesicular structures at high temperature (T > T_m).

Global size determination by PCS

To gain information about the glycyrrhizin-content-dependent size evolution at low temperature and, moreover, the temperature-dependent size evolution upon fast heating of the samples, PCS measurements were performed at both temperatures, at 10 and 50°C. Therefore, first the particle diffusion coefficient D_T is calculated from the slope of the linear dependence of the relaxation rate Γ_T(q) and the square of the scattering vector q² (see Fig. S7). By Eq. 3, the hydrodynamic radius R_H of the particles is computed, which is shown in Fig. 2 for both temperatures. As already elaborated in the comparison of sucrose-free and sucrose-containing samples, at low temperature, the size of the structures decreases with increasing x_glycyrrhizin, and only minor changes are observed between contents of 50 and 70 mol% (R_H ≈ 129 ± 2–141 ± 1 Å). Because of a higher amount of saponin, a bigger bicelle rim surface can be stabilized, and smaller structures occur. Upon temperature increase, the evolution of R_H is inverted. R_H increases with increasing x_glycyrrhizin, which indicates that bigger particles are formed from smaller precursor particles present at low temperature. Such a behavior was also predicted by Nieh et al. (19). The size decrease observed for samples with 35 and 40 mol% glycyrrhizin might be explainable by a folding of an extended lipid sheet into a vesicular structure, which leads to a shape with a smaller R_H-value. This scenario will be investigated in more detail by SAXS.

Particle morphology by SAXS: model-dependent fitting of SAXS data at 10°C

First, the x_glycyrrhizin-dependent evolution of the size of the bicellar structures at low temperature (10°C) is investigated by SAXS. Scattering data obtained at this temperature and at different x_glycyrrhizin are shown in Fig. 5 a. Because of the absent head-to-tail contrast, a simple cylinder model (65) is used to approximate the scattering data and to thereby describe the bicellar, disk-like structure. A schematic of the model used is depicted in Fig. 4 a, and mathematical equations describing the model can be found in the Supporting Materials and Methods or in the SASView model documentation (65). In Fig. 4 a, it becomes clearly visible that because of the contrast variation by sucrose addition, only the hydrophobic interior of the bicelle contributes to the scattering signal. In this model, the lipid-built disk size is represented by the cylinder radius R_cylinder, and the membrane thickness is described by the cylinder height H_cylinder. Both the X-ray scattering length density XSLD of the solvent and the cylinder (the lipid-glycyrrhizin mixture) were calculated before fitting and were fixed during the fitting procedure. The XSLD-values of the pure compounds DMPC and glycyrrhizin were calculated on the basis of the molar volume of each compound. Therefore, the temperature-dependent value for DMPC was derived from a study of Nagle and Wilkinson (71), and the temperature-independent one for glycyrrhizin was calculated with ChemSketch (72). A more detailed description of the XSLD calculation can be found elsewhere (66). XSLD-values for the pure compounds, glycyrrhizin-content-corrected ones, and the ones for the buffer solutions can be found in Tables S1 and S2. Additional parameters in the fitting procedure are a constant background B and a scaling factor S. x_glycyrrhizin-dependent cylinder fits to the scattering data and fit results for R_cylinder and H_cylinder are presented in Fig. 5, a and b, respectively. Additionally, all fit parameters are summerized in Table S3.

Figure 4.

Schemes for model-dependent fitting of SAXS data matched with sucrose. In all cases, mainly the hydrophobic part of the lipid membrane is visible for the X-rays, and the solvent matches the hydrophilic membrane portion. (a) Cylinder model is used to describe the bicellar shape at a temperature of 10°C, which is reduced to a cylindrical shape because of the contrast matching. (b) Core-shell sphere model describing the vesicular shape at a temperature of 50°C is shown. Again, because of contrast matching, the conventional three-shell lipid membrane can be represented by one shell corresponding to the hydrophobic membrane portion. Parameters: R_cylinder, cylinder radius; H_cylinder, cylinder height (equal the thickness of the hydrophobic membrane part); R_core, inner vesicle radius; t_shell, thickness of hydrophobic membrane part; SLD_cylinder, scattering length density of the cylinder (equal to the hydrophobic membrane part); SLD_shell, scattering length density of the shell of the vesicle (equal to the hydrophobic membrane part); SLD_solvent, scattering length density of the solvent (depicted in gray); SLD_core, scattering length density of the solvent in the inner part of the vesicle (depicted in light gray). To see this figure in color, go online.

In all cases, R_cylinder was successfully determined. As expected from prior PCS measurements, R_cylinder decreases with increasing x_glycyrrhizin. Because of the absence of the beginning of a plateau at low q for both samples with the lowest x_glycyrrhizin, the resulting errors for the determined radii of 383.2 ± 28.2 Å for 35 mol% and 327.4 ± 17.2 Å for 40 mol% are quite large in comparison to the ones obtained at x_glycyrrhizin > 40 mol%, which are around 4–5 Å. For H_cylinder, more uncertain results were obtained, which is most probably due to the absence of a clear membrane contribution in the scattering signal. If this parameter was treated as an adjustable parameter, for some samples, unphysically small values of ∼5 Å were obtained. Therefore, in the fitting procedure, H_cylinder was initially set to a value of 20 Å, which is approximately the size of the hydrophobic DMPC membrane part. After optimization of all parameters (including B and S), H_cylinder was optimized in the last step and remained at the mentioned value (see also Table S3). Moreover, we like to mention that the choice of H_cylinder did not significantly influence the results for R_cylinder. Comparison of the fit parameters of the sample with 70 mol% glycyrrhizin with the size parameters obtained from cyro-TEM shows a concordance of the values derived from both methods. Especially, the diameter or length of the structures with values of 220 ± 7 Å from SAXS and 208 ± 50 Å from cryo-TEM are in good agreement. The bicelle height with 20 ± 0.5 Å appears smaller in comparison to the one derived from cryo-TEM (53 ± 21 Å). This is most probably due to the modified membrane contrast seen by SAXS, which is especially important with respect to the apparent membrane thickness.

Moreover, to confirm the reversibility of the temperature-dependent structural rearrangement in the sucrose-containing DMPC-glycyrrhizin system, exemplary SAXS data were recorded for the sample containing 70 mol% glycyrrhizin after heating and subsequent cooling (see Fig. S8). Therefore, different heating rates were used to heat the sample from 10°C up to 50°C. These heating rates correspond to the ones listed in Fig. 8: cycle #1 corresponds to the measurement before the first heating, and cycle #8 to the one after having heated with the slowest heating rate of 0.5°C min⁻¹. The blue solid lines again describe fits with the cylinder model (compare Fig. 4 a). Corresponding fit parameters are listed in Table S4. The shape of the scattering curves obtained after repeated heating and cooling looks very similar, and the parameter R_cylinder moreover confirms the reversibility of the structural transition. Except for the data taken before the first heating (cycle #1), within the error, this parameter stays constant at ∼108 Å.

Figure 8.

Core-shell sphere fits to SAXS data of samples with different x_glycyrrhizin at a temperature of 50°C (solid red lines). The samples were heated from 10°C with different rates k to a temperature of 50°C (numbers in gray, right). For better comparison, the data are scaled by constant factors (left). To see this figure in color, go online.

Particle morphology by SAXS: model-independent evaluation of SAXS data at 50°C

The main aim of this study is to elucidate the effect of the heating rate on the temperature-induced formation of self-assembled DMPC-glycyrrhizin structures. Based on the cryo-TEM results and additional SAXS data in the Supporting Materials and Methods (see Fig. S4), the formation of unilamellar vesicles is expected by an increase of the temperature to T > T_m. SAXS curves for samples with different x_glycyrrhizin were recorded at a temperature of 50°C after heating up the samples from 10°C with different heating rates k between 0.5 and 30°C min⁻¹. At first, data only for the slowest (k = 0.5°C min⁻¹) and the fastest rates (k = 30°C min⁻¹) are investigated by the model-independent IFT method (62,64). Data obtained for all heating rates used are later analyzed by model-dependent fitting, and model-independent fitting is first utilized to confirm the shape of the structures obtained at both extremes, slow and fast heating. The IFT method allows determination of the shape and dimensions of the scattering particle (73). The scattering data, together with IFT fits, are shown for both heating rates in Fig. 6, a and b, respectively. For the fits of the resulting p(r) functions, 15 cubic B-splines up to the maximal distance d_max are used. The corresponding p(r) functions are presented in Fig. 7 a for k = 0.5°C min⁻¹ and Fig. 7 b for k = 30°C min⁻¹.

Figure 6.

SAXS data for samples with different x_glycyrrhizin at a temperature of 50°C obtained after heating from 10°C with rates of (a) k = 0.5°C min⁻¹ and (b) k = 30°C min⁻¹. Solid lines are IFT model fits to the scattering data. To see this figure in color, go online.

For both heating rates k, an increase of the particle size with increasing x_glycyrrhizin is observed. This is in concordance with the results obtained from PCS measurements. By comparison with p(r) functions of different shapes (73, 74, 75), the existence of spherical particles, most likely unilamellar vesicles, is demonstrated. In particular, the inclination of the maximum of the p(r) function toward higher distances r indicates the presence of a hollow sphere and hence a vesicular structure (73). The inclination is more pronounced for a heating rate of 30°C min⁻¹, which directly indicates the formation of more monodisperse structures at higher heating rates. Moreover, when comparing the p(r) functions for samples with different x_glycyrrhizin at constant k, formation of less polydisperse structures is observed with increasing x_glycyrrhizin. In this case, the p(r) function for the sample containing 35 mol% glycyrrhizin at a heating rate of k = 0.5°C min⁻¹ resembles a compact sphere rather than a hollow sphere. When taking a look at the polydispersity of the sample, which is determined by fitting to a value of ∼80% in the following (see Figs. 8 and 9), this observation clearly arises from this high polydispersity value. In summary, the IFT analysis shows that not only the initial particle size at low temperature but also the heating rate used to reach higher temperatures seem to play an important role in the formation mechanism of self-assembled vesicles.

Figure 9.

Core-shell sphere fit results for x-ray data recorded for different x_glycyrrhizin and k. In (a) and (b), the inner radius R_core and its polydispersity σ(R_core) are plotted as depending on k, whereas both parameters are plotted as depending on x_glycyrrhizin in (c) and (d). To see this figure in color, go online.

An additional parameter derivable from the IFT method is the radius of gyration R_G (see Eq. 5). In Fig. S9, R_G-values obtained from data recorded at the fastest heating rate of k = 30°C min⁻¹ are compared to R_H-values derived from PCS measurements, in which the sample was also heated very quickly by placing the cold sample into the thermostatted PCS setup. Comparison of R_G and R_H shows, as mentioned before, an increase in both parameters with increasing x_glycyrrhizin. However, more interesting is the ratio between both R_G and R_H, which is also shown as R_G/R_H in Fig. S9 (see right axis). Neglecting the data point at x_glycyrrhizin = 35 mol%, R_G/R_H decreases from a value of ∼1 to a value of ∼0.72. Whereas an R_G/R_H ratio of 1 represents a spherical shell or hollow sphere, a ratio of 0.78 stands for a solid sphere (76, 77, 78, 79). Hence, the vesicles obtained upon temperature increase seem to get denser by increasing x_glycyrrhizin and resemble a rather compact sphere. Such behavior was not expected and might be explainable by the very low polydispersity of the vesicles, which has a strong influence on the parameter R_G/R_H (77,79). Another possibility to explain this unexpected behavior is altered hydration properties in the presence of very high amounts of glycyrrhizin, which may lead to an increased value in R_H and therefore a decreased value of R_G/R_H.

Particle morphology by SAXS: model-dependent fitting of SAXS data at 50°C: determination of vesicle size and polydispersity

To analyze x-ray data obtained for the whole x_glycyrrhizin content range, model-dependent fits were applied to the scattering data recorded after heating from 10 to 50°C with heating rates between k = 0.5–30°C min⁻¹. Because of the presence of SUVs (see Figs. 3 b and 7) and the absence of a clear membrane contribution (compare Fig. S5), a core-shell sphere model from SASView (65) was employed to describe the scattering data (see Fig. 8, a–e). A schematic of the model is shown in Fig. 4 b, and a mathematical description can be found in the Supporting Materials and Methods or in the model documentation of SASView (65). All parameters obtained from the fitting procedure are listed in Tables S5–S7 for the different glycyrrhizin contents. Again, related to the contrast, only one shell was sufficient to represent the scattering data, and this model was applied instead of the vesicle model (65) to regard slightly different XSLDs of the SUV’s interior and the continuum (the solvent) (66,80). The size parameter obtained from this fitting procedure is the core radius of the SUV (R_core) and the polydispersity corresponding to this parameter (σ(R_core); compare Fig. 4 b). Based on the results obtained from cylinder fits at 10°C, the membrane thickness t_shell was set to a constant value of 20 Å and was not adjusted by fitting. As for the cylinder model, unsystematic and, for some samples, very small values in t_shell of ∼5 Å were obtained if this parameter was treated as adjustable. Again, the choice of this value did not influence the results for the important parameters R_core and σ(R_core) significantly. Values for the XSLDs of the membrane composed of DMPC and glycyrrhizin, as well as the one for the solvent, were again calculated before fitting (see Tables S1 and S2) (66). All calculated XSLD-values were used as constant parameters in the fitting procedure, and only the one for the SUV interior was optimized in the last step, but it remained very close to the initial value (see Tables S5–S7). XSLD_solvent and XSLD_core differed at most in the second decimal, but the usage of different values for both parameters was necessary to achieve an overlap of the experimental data with the fit function, especially in the low-q region. This is important to guarantee a meaningful determination of the SUV size. We want to mention that because of the temperature-induced removal of glycyrrhizin from the self-assembled structures, the exact composition of the SUV membrane is not known. However, because of the presence of only one XSLD difference (between the membrane and the interior or continuum), changing the membrane’s XSLD results mainly in a change of the scaling factor S (describing the position of the calculated fit function on the intensity or y axis) and not in the shape of the resulting approximation. Results for both fit parameters R_core and σ(R_core) are shown in Fig. 9, dependent on the heating rate k (in Fig. 9, a and b) and on the glycyrrhizin content x_glycyrrhizin (in Fig. 9, c and d).

When first taking a look at the dependence of R_core on different k (Fig. 9 a), it can be seen that x_glycyrrhizin clearly dictates the size of the SUVs formed upon temperature increase and that R_core increases with increasing x_glycyrrhizin. For a heating rate of 10°C min⁻¹, for instance, R_core increases from 148.7 ± 2.4 Å at 35 mol% glycyrrhizin to 321.8 ± 1.1 Å at 70 mol% glycyrrhizin. Hence, formation of larger structures at higher x_glycyrrhizin can be attributed to a decreased membrane flexibility, whereby folding into a closed vesicle occurs in a later state of the reorganization process, resulting in a bigger SUV size (19). Additionally, at a higher glycyrrhizin concentration, a larger lipid fragment can be stabilized by the rim-covering glycyrrhizin molecules, and finally, the critical lipid disk size before closure into a SUV structure results from a balance between the line tension and the bending energy of the membrane fragment (81). With R_core-values between ∼150 and 350 Å, the self-assembled vesicles have a very small size, which is similar to the sizes obtained by Lesieur et al. for a system composed of DMPC and sodium cholate (17).

Whereas, for example, for the sample with 60 mol% glycyrrhizin at k < 5°C min⁻¹, a decrease of R_core from 287.8 ± 1.1 Å (at 0.5°C min⁻¹) to 268.3 ± 0.8 Å (at 2.5°C min⁻¹) with increasing k is observed, R_core becomes independent from k at k > 5°C min⁻¹ at constant x_glycyrrhizin. For the sample with 60 mol% glycyrrhizin, the R_core-values range from 265.8 ± 0.9 Å (at 5°C min⁻¹) to 258.2 ± 0.2 Å (at 30°C min⁻¹). This indicates that k ≈ 5°C min⁻¹ is a critical heating rate, and increasing k above this value does not have a significant effect on the SUV formation process anymore. The dependence of the corresponding polydispersity σ(R_core) as a function of k is depicted in Fig. 9 b. Because the best fits obtained do not capture all minima occurring in the experimental data, values for σ(R_core) are probably overestimated. Nevertheless, we think that a comparison of the values obtained gives a reliable statement about the evolution of this parameter as a function of the heating rate k as well as the glycyrrhizin content x_glycyrrhizin. In concordance with the literature (17,20), σ(R_core) decreases with increasing x_glycyrrhizin, and very high σ(R_core)-values of ∼40–80% are obtained for the samples with x_glycyrrhizin = 35 and 40 mol%, whereas σ(R_core) decreases with increasing k. σ(R_core)-values for higher x_glycyrrhizin (50–70 mol%) are nearly identical and are much smaller (∼12–20%) compared to the ones obtained for 35 and 40 mol% glycyrrhizin. Relating this observation to the results obtained from PCS (see Fig. 2) and from SAXS measurements at low temperature (see Fig. 5), it can be concluded that highly monodisperse SUVs are only obtainable from very small and monodisperse precursor bicelles. This is in line with the study of Nieh et al. (19). In the case of larger precursor particles, presumably mainly a closure of the sheet-like structure into a vesicle occurs, which seems to lead on the one hand to smaller particles but, on the other hand, to more polydisperse particles. In this case, a folding into SUVs occurs at an early stage in the transition because of the low amount of surfactant covering the bicelle rim and an increased rim line tension (19,82). Both parameters, R_core and σ(R_core), are moreover plotted as depending on x_glycyrrhizin in Fig. 9, c and d, respectively. When comparing both panels, it can be seen that for the samples with lower glycyrrhizin content (35 and 40 mol%), R_core is quite unaffected by varying k, whereas the polydispersity decreases strongly with increasing k. A fast temperature increase seems therefore to lead to a rapid closure of the precursors, resulting in more monodisperse SUVs. At slow heating rates, maybe partial coalescence effects before SUV closure lead to a broader particle distribution and consequently to a higher σ(R_core). For samples with higher glycyrrhizin contents (50–70 mol%), again, an influence of k on R_core and σ(R_core) is mainly visible for k < 5°C min⁻¹. Whereas the particle size R_core stays tunable by x_glycyrrhizin, σ(R_core) reaches a minimal value of ∼12–14% and is independent of k (and x_glycyrrhizin at x_glycyrrhizin > 50 mol%).

Conclusions

This work focuses on the temperature-induced bicelle-to-vesicle transition in mixtures of DMPC and glycyrrhizin (with amounts between 35 and 70 mol%) in the presence of 25 wt% sucrose. The use of sucrose leads to a significant simplification of the scattering curves due to changes in contrast. Moreover, the bicelle-to-vesicle transition is only possible in the sucrose-containing system and the conditions used in the presented experiments. First, the system was characterized at a low temperature (10°C, T < T_m) by cryo-TEM, PCS, and SAXS. Self-assembled disk-like structures or bicelles were identified, and as expected, the size of these structures decreases with increasing x_glycyrrhizin. A sucrose-free system with similar glycyrrhizin content showed a similar behavior, and only a small shift in solubilization boundary to lower glycyrrhizin contents was observed. Cryo-TEM imaging of a sample with 70 mol% glycyrrhizin maintained at a temperature of 50°C (T > T_m) before freezing indicates formation of closed SUVs after very fast heating. Because of the formation of specific glycyrrhizin-glycyrrhizin interactions, the formation of correlated membrane stacks was observed in the absence of sucrose. In the presence of sucrose, unspecific interactions between the sucrose in the buffer solution and the glycyrrhizin heads seem to avoid specific glycyrrhizin-glycyrrhizin interactions, and therefore, a strong influence of the addition of sucrose was demonstrated at T > T_m. PCS measurements of the sucrose-containing system show an increasing overall SUV size with increasing x_glycyrrhizin, which denotes the formation of larger SUVs from smaller precursor particles. Shape, size, and finally, also polydispersity of the SUVs are determined by SAXS measurements at high temperature for several heating rates k and x_glycyrrhizin. Calculation of p(r) functions indicates formation of very monodisperse SUVs for high k (k = 30°C min⁻¹), whereas much more polydisperse ones are formed at low k (k = 0.5°C min⁻¹). As also seen in PCS, the SUV size increases with increasing x_glycyrrhizin independent of k. At lower x_glycyrrhizin (35 and 40 mol%), small but very polydisperse SUVs are formed, and the polydispersity is strongly affected by the heating rate k. The precursor particles are comparably large, and a closure to SUVs most likely occurs in an early state of the conversion process. A slow k may allow coalescence processes before SUV closure, which can lead to a nearly unaffected mean size but strongly increasing polydispersity. For higher glycyrrhizin contents (x_glycyrrhizin = 50–70 mol%), vesicles with a minimal polydispersity of ∼12–14% are formed irrespective of x_glycyrrhizin and, above ∼5°C min⁻¹, also irrespective of k. In the mentioned case, the SUV size stays tunable by x_glycyrrhizin. In summary, the system presented in this study behaves in terms of temperature-induced bicelle-to-vesicle transition similar to well-known systems composed of phospholipids and bile acids. Hence, it can be concluded that introduction of a huge hydrophilic headgroup, in comparison with bile salts, does not show a significant influence on the bicelle-to-vesicle transition process, and SUVs with defined size and polydispersity can be obtained by choosing the appropriate x_glycyrrhizin and heating rate k.

Editor: Tommy Nylander.

Author Contributions

C.D. designed the research, prepared the samples, and performed PCS and SAXS measurements. Y.H. conducted cryo-TEM experiments. C.D. and T.H. discussed the results and wrote the article.