Interface tuning and stabilization of monoglyceride mesophase dispersions: Food emulsifiers and mixtures efficiency

Journal of Colloid and Interface Science 496 (2017) 26–34

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science journal homepage: www.elsevier.com/locate/jcis

Regular Article

Interface tuning and stabilization of monoglyceride mesophase dispersions: Food emulsifiers and mixtures efficiency Sébastien Serieye, Fabienne Méducin, Irena Miloševic´ 1, Ling Fu, Samuel Guillot ⇑ Interfaces, Confinement, Matériaux et Nanostructures (ICMN), Université d’Orléans, CNRS, UMR 7374, 1b rue de la Férollerie, CS 40059, 45071 Orléans Cedex 2, France

g r a p h i c a l a b s t r a c t

a r t i c l e

i n f o

Article history: Received 25 October 2016 Revised 15 January 2017 Accepted 17 January 2017 Available online 27 January 2017 Keywords: Cubosomes Hexosomes Micellar cubosomes Emulsified microemulsions Food emulsifier Monolinolein Lyotropic liquid crystals Limonene

a b s t r a c t Several food surfactants were examined as possible efficient emulsifiers for liquid crystalline monolinolein-based particles and as alternative choices to the non-food-grade emulsifier conventionally used PluronicÒ F127. We described a food emulsifiers’ toolbox, investigating their ability to emulsify mesophases (stabilization capacity, particle size, zeta potential) and their impact on internal nanostructures (from swelling to drastic modifications). Among the selected surfactants, sucrose stearate (S1670) was found to be the best candidate for replacing in a long term F127 as an efficient stabilizer of lipid particles. The emulsification performed by mixing F127 with S1670 or sodium caseinate (NaCas), and S1670/ NaCas helped to discriminate their respective role in the particles and so their efficiency for the stabilization. In case of S1670 as co-emulsifier no strong structural modification was observed, while using F127 (25 wt% NaCas) an unexpected hexagonal mesophase was highlighted in self-assemblies. The evolution of zeta potentials by varying the mesophase and the emulsifier also informed about the distribution of cosurfactants in the particles. We thus reported submicronic nanostructured systems (from 100 to 350 nm) that were fully food-grade and possibly contained limonene, with a surface charge from
70 to
5 mV. Ó 2017 Elsevier Inc. All rights reserved.

1. Introduction ⇑ Corresponding author. E-mail address: [email protected] (S. Guillot). Present address: Powder Technology Laboratory, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland. 1

http://dx.doi.org/10.1016/j.jcis.2017.01.059 0021-9797/Ó 2017 Elsevier Inc. All rights reserved.

Amphiphilic lipids such as monoglycerides are known to selfassemble when mixed with water. They form inverted mesophases leading to high interfacial area materials with both hydrophilic and hydrophobic subspaces. Hence, such mesophases can be used as

S. Serieye et al. / Journal of Colloid and Interface Science 496 (2017) 26–34

reservoirs of active compounds of various hydrophobicities. The aqueous part embedded in the continuous hydrophobic medium may be organized in continuous cubic networks, hexagonal phase with water nano-channels or inverse micelles organized in a cubic arrangement. Among nanostructured binary mixtures, those based on monoolein [1–4], monolinolein [5] or phytantriol [6,7] are extensively studied. A fluid isotropic reversed micellar phase L2, a reversed hexagonal phase H2, lamellar and reversed bicontinuous  and Ia3d  space groups) may be formed cubic phases V2 (Pn3m depending on temperature, water content or external pressure [8]. Mesophases are thermodynamically stable systems. When swollen by water they can be dispersed in a continuous aqueous phase as particles ideally keeping the structure of the bulk phase with excess of water [5]. Particles of submicron size containing V2 phases (cubosomes) [9], the H2 phase (hexosomes) [10], the reversed micellar cubic phase I2 (micellar cubosomes) [11,12] and the L2 phase (emulsified micro emulsions - EME) [13,14] are reported. Below room temperature, the emulsification is often easily obtained by adding an apolar component (oil) to the binary system; an oil/monoglyceride ratio is used to relate to the internal structure [14,15]. The destabilization of dispersions is prevented by using emulsifiers. In many studies, the synthetic triblock copolymer F127, consisting in two poly(ethylene oxide) blocks (PEO) separated by a poly(propylene oxide) block (PPO), is used for this purpose. The hydrophobic moieties (PPO) of the polymer are adsorbed at the surface of the particles, whereas the hydrophilic ones (PEO) build a corona that sterically stabilizes the dispersion. Emulsifiers may however interact with the internal structures of particles. Even in the case of F127 [16–18], a weak interaction is found for low F127 contents but colloidal stability is thus strongly reduced. At larger contents, the internal structure may be drastically modified by incorporating F127 [19]. In this context, screening of Pluronics performance was carried out to optimize the stability of the emulsified cubic phases [20]. New classes of steric stabilizers (PEO stearate) for lyotropic liquid crystal dispersions of monoolein and phytantriol were also recently screened [21,22]. The possibility of integrating functional molecules (drugs, foods, aroma) in those nanostructured particles suggests many potential applications like delivery systems [23,24]. Particularly they are promising systems in food industry to improve solubilization and protection of active ingredients, to be used as reactors, to control aroma release or to create structured food products [25,26]. In this context, structural transformations of these particles are notably investigated under conditions of digestion [27]. Although F127 is very efficient for stabilizing lipid mesophases, it cannot be employed for some direct applications because it is not foodgrade. This drawback is overcome by searching other emulsifiers to effectively stabilize such particles. Dextran and hydrophobically modified starches are reported to disperse monoolein-based cubosomes [28]. Cellulose derivatives are tentatively used for the replacement of F127 again for stabilizing cubosomes [29,30], as well as mixtures of F127/b-casein used for emulsifying cubosomes and hexosomes [31]. By using hydrophobically modified ethyl hydroxyethyl cellulose the long-term stability as obtained with F127 is not ensured [29], whereas it is demonstrated for only 30 days with hydroxypropyl methyl cellulose acetate succinate [30]. Polysorbate 80 is used to help for the stabilization of glycerol dioleate/diglycerolmonooleate sponge-like particles for months (long-term size data not shown) [32]. The anionic surfactant citrem also stabilizes non cubic particles for three weeks [33]. Partially hydrolyzed lecithin is also used for the emulsification of monolinolein/vitamin E acetate mesophases but a poor degree of structural order is observed [34]. Moreover, the use of F127 often induces undesired vesicles in increasing amount in the sample

27

with larger proportion of polymer [9,10]; the substitution of F127 by other emulsifiers could solve this problem [31]. Our main goals are to propose other emulsifiers being able to efficiently stabilize nanostructured lipid dispersions and to offer complete food-grade systems. In this study, neutral or charged surfactants with low or high molecular weight are used: sucrose stearate and oleate, sodium stearoyl lactylate, sodium caseinate, whey protein isolate, lecithin, and Tween 80. We first tested the ability of those food emulsifiers to disperse and stabilize various mesophases. To specify the framework of possible applications using such lipid-based particles, we tested their stability with time. Dispersions were probed by small angle X-ray scattering in order to determine dispersed mesophases; in case we described any structural modification due to an interaction between emulsifiers and mesophase compounds. The influence of the food emulsifier content on the size and overall charge of particles was observed for different kinds of mesophase. We also studied size variations and internal structure modifications with emulsifier mixtures at different ratios. The study was conducted at room temperature. 2. Experimental 2.1. Materials DimodanÒ U/J (DU) is a commercial-grade form of monolinolein and is supplied by DANISCO A/S (Braband, Denmark). It contains 96% distilled monoglycerides, of which 62% are monolinoleate. R(+)-limonene is purchased from Fluka (purity > 96%). PluronicÒ F127 (PEO99-PPO67-PEO99) is provided by BASF. Sucrose stearate (S1670) has a monoester content of 75% and a purity of stearic acid of 70%; sucrose oleate (OWA-1570) has a monoester content of 70% and a purity of oleic acid of 70%. The sugar esters S1670 and OWA1570 are gifts from Mitsubishi-Kagaku Foods Corporation. Sodium stearoyl-2 lactylate (SSL) is given by Dr. Straetmans Chemische Produkte GmbH. Sodium caseinate salt from bovine milk (NasCas) is purchased from Sigma Aldrich. Whey protein isolate (WPI, ProlactaÒ 90) is kindly provided by BBA Lactalis Industries where proteins represent 90% of the dry matter. Polysorbate 80 (TweenTM 80) is from Croda Uniquema. Lecithin (PhospholiponÒ 85G from soybean) is kindly provided by Lipoid GmbH and contains a minimum of 85% phosphatidylcholine. F127, S1670, Tween 80 and OWA1570 are neutral emulsifiers while SSL, WPI and NaCas are anionic emulsifiers at neutral pH. Pure lecithin is zwitterionic, however PhospholiponÒ may contain fatty acids with negative charges. The molecular structures of those materials are gathered in Fig. 1. Copolymer F127 usually stabilizes macroemulsions by steric repulsion while charged surfactants should stabilize by electrostatic repulsion. Milk proteins (NaCas and WPI) benefit both stabilizing properties. All chemicals are used without further purification. 2.2. Sample preparation Ultrapure water (deionized water at 18.2 MX cm from a Millipore Milli-Q device) is used for the preparation of all the aqueous dispersions. The mixture forming the dispersed phase is first preDU pared. The d weight ratio d ¼ DUþoil  100 represents the percentage of Dimodan U in the dispersed phase. R-(+)-limonene is the oil added at room temperature to DU in order to tune the type of mesophase. The emulsifier is separately dissolved into deionized water. The emulsifier/mesophase content is defined by the b weight ratio b ¼ Emulsifier  100. Except the study conducted to DUþoil determine the emulsifier content effect on the size, b is kept constant at about 8.

28

S. Serieye et al. / Journal of Colloid and Interface Science 496 (2017) 26–34

Fig. 1. Molecular structures of (a) Monolinolein, (b) Limonene, (c) F127, (d) S1670, (e) OWA-1570, (f) SSL, (g) Phosphatidylcholine, (h) Tween 80.

Finally, both phases are mixed and the emulsification of the raw mixture is realized by fragmentation using ultrasonic waves (Vibracell 75115 ultrasound tip working at 20 kHz), without external cooling, for 8 min at 100 W in pulse mode (1 s on and 1 s off). Following this procedure we ensured of no substantial degradation of the lipid [5]. Mesophases at a given d are dispersed in the aqueous continuDUþoil ous phase at a mass fraction of / ¼ DUþoilþF127þwater . The dispersions are stored at ambient temperature. In case of emulsifier mixtures, Sub that represents the mass fraction of we define the ratio a ¼ SubþF127 the emulsifier (Sub: NaCas or S1670) that substitutes F127 in the total mass of emulsifiers; such a ratio is also defined in case of NaCas . S1670 substitution by NaCas as a ¼ NaCasþS1670

2.3. Dynamic light scattering and zeta potential measurements A laser granulometer (Zetasizer Nano ZS90 from Malvern) working at k = 633 nm is used for the measurement of the size dispersions. Dynamic light scattering (DLS) measurements are carried out at fixed scattering angle of 90° at 25 °C. Correlation functions are automatically treated by an inverse Laplace transform giving rise to the hydrodynamic size distributions. The most representative size of the particles is taken as the mode (maximum value of the peak) derived from the size number distribution, which is of log-normal shape. Each mode value is derived from the average of 3 measurements; each measure includes 11 runs of 10 s each. This average gives rise to the error bars on the mode shown in the figures. The width of the lognormal size distribution can be calculated through asymmetrical error bars (see procedure in [35]) but is out of interest in this study. The samples are diluted in order to avoid multiple scattering, knowing that the dilution of such emulsified mesophase particles do not affect their size; the viscos-

ity of the samples is therefore that of water. Samples are diluted in 10 mM NaCl according to the International Standard ISO 13321:1996 in order to suppress the double layer contribution to the hydrodynamic radius of charged particles. The zeta potential of the lipid particles is determined by using the Zetasizer Nano ZS90.

2.4. Small angle X-ray scattering measurements The internal structure of the particles is determined by the small angle X-ray scattering technique (SAXS). The scattered intensity is collected as a function of the wave vector q ¼ 4kp sinðh=2Þ with h the scattering angle and k the beam wavelength. The interpretation of the scattering data is based on the indexation of the Bragg peak positions displayed in the intensity spectra with the corresponding (h, k, l) Miller indices. This fully determines the structures or the space groups involved, and makes possible to calculate the mean lattice parameter a of the ordered structure from the interplanar distance. The lattice parameter is given by pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 2 2 a ¼ 2qp h þ k þ l and a ¼ q4ppffiffi3 h þ k for cubic and 2Dhexagonal structures respectively. As for the L2 phase, a broad peak is displayed that only gives a characteristic distance which corresponds to the mean size of microdomains. SAXS measurements were collected on different synchrotron beamlines under the following conditions: at DELTA (Dortmund ELecTron Accelerator, Germany) with a beam energy of 10 keV (5  109 photons/s), k ¼ 1:24 Å, a q range from 0.3 to 3.5 nm
1 and an exposure time of 600 s; at MAX-lab (National electron accelerator laboratory for synchrotron radiation research, Lund, Sweden) with a beam energy of 10 keV (1011 photons/s), k ¼ 1:24 Å, a q range from 0.1 to 3 nm
1 and an exposure time of 500 s; at HASYLAB (HAmburger SYnchrotronstrahlungsLAbor, Germany) with a

29

S. Serieye et al. / Journal of Colloid and Interface Science 496 (2017) 26–34

beam energy of 8 keV (109 photons/s), k ¼ 1:50 Å, a q range from 0.2 to 2.5 nm
1 and an exposure time of 800 s; at ESRF (European Synchrotron Radiation Facility, Grenoble, France) with a beam energy of 10 keV, k ¼ 1:24 Å, a q range from 0.1 to 3 nm
1 and an exposure time of 500 s. 3. Results and discussion 3.1. Emulsification of the DU/water swollen phase 3.1.1. Particle size stability with time For simplicity, we initially tested the ability of the food emulsifiers to be efficient stabilizers for DU-based dispersions without limonene. Thus we worked at d 100, value at which we expected to obtain bicontinuous cubosomes. The stability of these dispersions was investigated by following the long-term size evolution of the particles over time by DLS (Fig. S1). Samples of / = 1 wt% dispersed phase at d 100 emulsified at b 8 by triblock copolymer (F127), sucrose stearate (S1670) and sodium caseinate (NaCas) are shown. As expected, the good colloidal stability using F127 was confirmed: an average size value over the 135-day period (maximum duration of the experiment) was derived at 180 ± 4 nm. In the same duration, the size of particles stabilized by S1670 was also found stable at 160 ± 3 nm. Thus, S1670 appeared to be a good candidate to substitute F127 as a longterm stabilizer of the DU-based particles. NaCas stabilized the dispersion only for 40 days with a size of 150 ± 4 nm; then the size grew rapidly and was not measurable anymore after 90 days by DLS due to large particles settled in the solution that produced no scattering. No antibacterials were added in our preparations, so that the miniemulsion destabilization through this abrupt change in size could be a consequence of NaCas degradation upon bacterial growth. Table 1 gathers data on the temporal stability limit and the size of d 100 dispersions using various food emulsifiers. Size results mainly demonstrated that DU-based particles, possibly nanostructured, were difficult to stabilize over a wide period of time. Indeed except S1670, which was comparable to F127 in term of stability, none of the chosen emulsifiers could stabilize the lipid dispersion for more than two months. NaCas and lecithin had an intermediate efficiency of stabilization in the range of 40– 55 days; however lecithin stabilized smaller particles (125 nm). Although the stability was less than F127 they could be utilized in short use applications (e.g., vectorization, cytotoxicity [36]) or embedded in a polymer matrix [24]. Finally the sucrose oleate (OWA-1570), Tween 80 and WPI were obviously not suitable for lastingly stabilizing monolinolein/water particles. We did not focus in this study on the destabilization process of the dispersions. 3.1.2. Particle size vs. emulsifier content In the most favorable cases of stabilization at b 8, the particles obtained had submicron sizes from 125 to 180 nm depending on the chosen emulsifiers. However the size could be tuned by varying the emulsifier content (Fig. S2). As for normal macroemulsions, an increased amount of emulsifiers at constant energy input in the emulsification process of lipid-based particles generally led to smaller sizes [18,35]. Among the rather good stabilizers, F127, S1670 and lecithin offered a wide range of particle size by varying b: from 1 up to 17, the size decreased from 295 down to 165 nm for F127, from 245 down to 145 nm for S1670, and from 174 down to 100 nm for lecithin. Using WPI, the size however did not decrease as much and then stabilized. Indeed WPI stabilized particles mostly by steric repulsion. The primary structure of WPI is known to be globular, the hydrophobic amino acids are gathered in the center of the globular structure while hydrophilic amino acids are at the outer face. At a water/oil interface, the hydrophobic part

Table 1 Stability of the dispersions and size (mode) of d 100 particles for the different emulsifiers tested at b = 8; / = 1%. Emulsifier

Average size (nm)

Stability (days)

Pluronic F127 S1670 NaCas WPI Lecithin SSL Tween 80 OWA-1570

180 160 150 155 125 115 100 100

>135 >135 40 7 55 20 7 1

of the protein is exposed to the oil part of the interface and the hydrophilic parts to the aqueous phase [37]. WPI is denatured to an almost planar structure; the interface is partially covered and the molecules cannot find a tight molecular arrangement. However this WPI layer seems to prevent additional adsorption of proteins at the water/oil interface, which leads to a constant particle size. The size using SSL reached so small values (60 nm) that the internal phase was probably not cubic. 3.1.3. Liquid crystal phase determination In addition to the colloidal stability, we must ensure that those food surfactants can stabilize a lipid structure. Thus we determined the possible internal structure within particles at d 100 by SAXS measurements. The swollen DU/water mesophase is known to be  type at room temperature [5]. F127 may integrate a V2 of Pn3m the dispersed mesophase, whose cubic symmetry then evolves into  Im3m; it is known d 100 dispersions show mixtures of particles of both cubic types [19]. As shown in Fig. 2, scattered intensities of 1% dispersions stabilized by pure F127 (a- left or right), pure S1670 (eright) and pure NaCas (e- left) at b 8 showed Bragg peaks that revealed an internal structure (Bragg peak locations for characteristic curves are given in Fig. S3). As expected from a previous measurement with monolinolein [18], we deduced the dispersed phase  stabilized by pure F127 (a = 0) as V2 with a coexistence of Pn3m  symmetries (Fig. S3). In this study we used DU as the and Im3m lipid and the lattice parameters of the cubic phases were of 9.6 nm and 12.9 nm respectively. The use of S1670 and NaCas led  space group) with lattice parameters of to a V2 phase (pure Pn3m 11.3 and 9.7 nm respectively. Lecithin could also emulsify a V2 phase, as did WPI with a lattice parameter of 9.7 nm (Fig. S4). Conversely the SAXS pattern with SSL as the emulsifier did not show any peak demonstrating that SSL was not able to stabilize the V2 phase. Thus the production of bicontinuous cubosomes was feasible by using most of the selected food emulsifiers. Except using F127, the emulsified V2 phases had the same symmetry but values of the lattice parameter might vary with the emulsifier type; this indicated that a fraction of emulsifier might affect the internal structure. This effect should depend on the emulsifier affinity to the lyotropic liquid crystalline phase. Several emulsifiers could be used to stabilize mesophase dispersions, but their structure could affect the emulsion size or charge, its stability and the inner structure. 3.2. Emulsification of DU/limonene/water particles 3.2.1. Mesophase determination R-(+)-limonene was used as an apolar additive to tune the type of mesophase and thus to easily obtain hexosomes, micellar cubosomes and EMEs at ambient temperature [15,38]. We tried to emulsify such non-cubic particles with the selected food emulsifiers. The particles containing limonene had stable sizes for 15 days. We only focused here on the ability of the chosen

30

S. Serieye et al. / Journal of Colloid and Interface Science 496 (2017) 26–34

Fig. 2. SAXS curves from d 100 particles emulsified by NaCas/F127 mixtures (left; 16 days after preparation) and S1670/F127 mixtures (right; 8 days after preparation); / = 1%, b = 8.

stabilizers to emulsify the lipid/limonene mesophases and their effect on the internal structure. SAXS measurements were performed on dispersions containing limonene at different d values; a description of the emulsified mesophases found within the particles is given for various emulsifiers in Fig. 3. Mesophases did not occur below d 20; normal emulsions were produced because the amount of lipid was not sufficient to structure the dispersed phase. Above this limit, the emulsification of H2, I2 and L2 phases was achieved by most of the selected compounds. Nevertheless, we observed some boundary shifts corresponding to different extent of the mesophases in d for the various emulsifiers. In most of studies F127 was used for its low interaction with the internal phase therefore we used it for comparison. From size measurements, the use of SSL as the emulsifier already raised doubt about its effectiveness to help to stabilize a V2; SAXS measurements clearly emphasized that lipid particles produced at d 100 and b 8 were unstructured. However SAXS measurements showed that SSL could emulsify mesophases containing limonene, and at d 95 hexosomes were found instead of bicontinuous cubosomes when stabilized by F127 (Fig. 3). The extents of the I2 and H2 phases in the phase diagram were also relatively large (at least from d 30 to d 95) compared to all other emulsifiers. By using S1670 instead of F127, phase domains seemed shifted towards smaller d: S1670 stabilized hexosomes at d 60 instead of micellar cubosomes. In contrast, lecithin seemed to favor negative mean interfacial curvature so that the emulsification of mesophases could occur in less extended d range than for F127. Whereas F127 stabilized the L2 phase at d 40,  symmetry. Structural NaCas and WPI formed the I2 phase of Fd3m details of DU/limonene mesophases were gathered in Table 2 and were found dependent on the emulsifier used. These data will be used in the discussion part to compare the behavior of the emulsifiers at interfaces with or without limonene. 3.2.2. DU/limonene particle size We also compared the size of DU/limonene dispersions as a function of d after 1 day preparation for the various emulsifiers. A general trend of the particles size vs. d could be observed (Fig. 4): overall, as d increased the particle size decreased at constant dispersion energy. The sharp decrease observed around d 20 is a signature of the transition from the normal emulsions range to emulsified mesophases indicating a sufficient quantity of structuring lipid in the droplets to obtain the L2 phase. Above d 20, we observed an overall decrease in size. The L2 phase is fluid and usually provides largest particle sizes; moreover the storage moduli of V2 and I2 phases are similar but larger than for the intermediate H2 phase [39,40]. The overall decrease showed particle sizes were

Fig. 3. Type of mesophases dispersed at 1% by various food-grade emulsifiers after 1 day preparation, derived from SAXS patterns, depending on the lipid/limonene ratio d; b = 8.

Table 2 Lattice parameters and characteristic distances (in nm) of DU/limonene identical mesophases emulsified at / = 1%; b = 8. The lack of data indicates a different mesophase preventing any possible comparison. Emulsifier

d 30 – L2

d 50 – Fd3 m

d 85 – H2

F127 S1670 NaCas WPI Lecithin SSL

5.15 5.96 5.26 5.76 – –

17.01 19.02 17.39 17.64 – 17.08

6.20 6.72 6.44 6.22 6.22 6.15

probably not impacted by the viscoelastic properties of the mesophase. As expected, proteins (WPI and NaCas) produced the largest size dispersions.

3.2.3. Zeta potentials of DU/limonene particles Thereby, emulsifiers could vary the particle size according to their nature or concentration; they could also change the surface charge of particles as shown in Fig. S5. Zeta potentials were found all negative for the tested emulsifiers in a large range from 0 down to -70 mV. Neutral emulsifiers, i.e. F127 and S1670, gave particles with the lowest negative charge densities (e.g., 10–15 mV at d

S. Serieye et al. / Journal of Colloid and Interface Science 496 (2017) 26–34

31

Fig. 4. Dispersion size after 1 day preparation as a function of d using F127, S1670, NaCas, WPI and SSL as emulsifiers; / = 1%, b = 8.

Fig. 5. Lattice parameters of emulsified mesophases at d 100 stabilized by NaCas/ F127 and S1670/F127 mixtures; / = 1%, b = 8.

100). Conversely, particles stabilized by negatively charged emulsifiers (SSL, lecithin, proteins) exhibited the highest negative potentials. SSL carries one charge per molecule; this small surfactant can produce dense interfacial layers and thus may induce the largest particle charge density. We also modulated the DU content at constant oil phase concentration by varying d. As the percentage of DU in the dispersed phase was increased, the surface charge was found to decrease in absolute values for all emulsifiers except F127 for which an increase by 5 mV was observed.

was observed for d below 40. Thus, in case of a F127/S1670 mixture, the size value is still fixed by F127, that ensured the steric stabilization, even in small proportion. As in the previous case, the mesophase particles stabilized by NaCas had a larger size than those using pure S1670 for all d values (Fig. 7, right). A similar trend to F127 was observed by adding S1670 to NaCas: mixtures of emulsifiers at a = 0.75 and 0.5 produced roughly the same sizes than with pure NaCas; thus NaCas would fix the particle size at those a ratios. However, the size variation was more continuous in that case since we got intermediate particle sizes at a = 0.25 for all d values. At this intermediate a range, S1670 would participate to the stabilization as a co-surfactant.

3.3. Dispersions with mixtures of emulsifiers 3.3.1. Effect on the internal structure of DU/water particles We described the gradual replacement of F127 by NaCas or S1670 in connection with a possible effect on the internal structure of particles dispersed at 1%. SAXS intensities as a function of the wave vector q were collected from d 100 dispersions for different NaCas/F127 and S1670/F127 ratios (Fig. 2). Starting with pure   coexisF127 (a = 0) and a dispersed V2 phase with a Pn3m/Im 3m tence, the increasing content of NaCas led to a V2 phase of single  symmetry above a = 0.5. From 0.5 to 1, the lattice parameter Pn3m was found to increase with a until the same value than for the  particles stabilized by F127 was reached (Fig. 5). InterestPn3m ingly we discovered an intermediate regime below a = 0.5 (see SAXS pattern at a = 0.25) where the NaCas/F127 mixture stabilized the H2 phase at d 100. The progressive substitution of F127 by  cubosomes and the Pn3m  cubic lattice S1670 led also to pure Pn3m parameter increased continuously from 9.6 to 11.3 nm. However, we still observed at a = 0.25 the coexistence of both symmetries in the V2 phase; we did not observe any H2 phase at the chosen a ratios with S1670. The same experiment was done substituting S1670 by NaCas  cubosomes; the (Fig. 6). All dispersions were composed of Pn3m lattice parameter decreased continuously from 11.3 down to 9.7 nm when NaCas content was increased. 3.3.2. Size of DU/limonene particles with mixtures of emulsifiers We also tested mixtures of high and low molecular weight emulsifiers, F127/S1670 and NaCas/S1670, in order to determine their relative contribution at the outer interface of the droplets. After 1 day of preparation, structured particles solely stabilized by F127 (a = 1) had larger sizes than using pure S1670 (a = 0) for all d values (Fig. 7, left). By increasing the S1670 content, sizes did not evolve much; even up to a = 0.75, a rather small decrease

3.4. Discussion Time stability results using pure S1670 and its capacity of emulsifying all types of mesophases showed that S1670 was an efficient substitute to F127 as a stabilizer of monolinolein/oil/water dispersions. As for F127, weak zeta potentials were found with S1670 and the stabilization of dispersions could only be viewed as a steric one. S1670 is a low molecular weight emulsifier that can induce a good coverage of the surface of particles. A dense corona made of sucrose head groups can be created, ensuring the steric stabilization. Nevertheless S1670 did not only stabilize the particles but was also involved in the mesophase. Indeed, the progressive substitution of F127 by S1670 led to dispersions with only one type of cubosomes, and with a lattice parameter that significantly  cubosomes stabilized by F127 withincreased from 9.6 nm (Pn3m out any incorporation) up to 11.3 nm (cubosomes stabilized and swollen by S1670). F127 can incorporate the V2 phase, however  The small molecule by drastically changing the structure to Im3m. S1670 could more easily integrate a mesophase without any modification of its structure type. The 1.7 nm increase of the lattice parameter at d 100 was compatible with the swelling induced by the incorporation of S1670 molecules at the lipid/water interface, i.e. the additional presence of two hydrophilic sugar head groups on both sides of the water subspace. This affinity for the lipid mesophase resulted in structural shifts observed in d compared to F127. Like using F127, the DU/limonene/water bulk mesophase in excess of water should be I2 [15]. By using S1670, the mesophase emulsified at d 60 was H2 that showed sucrose ester molecules modified the mean interfacial curvature of oil/water interfaces. The large hydrophilic S1670 head groups localized at those interfaces would then induce a less negative mean curvature, giving rise

32

S. Serieye et al. / Journal of Colloid and Interface Science 496 (2017) 26–34

Fig. 6. SAXS curves from d 100 particles emulsified by NaCas/S1670 mixtures (left), and corresponding lattice parameters (right); / = 1%, b = 8.

Fig. 7. Dispersion size after 1 day preparation as a function of d on substituting F127 with S1670 (left), and S1670 with NaCas (right); / = 1%, b = 8.

to the transition towards the H2 phase. Dealing with other types of mesophase including limonene, we found for each d a significant increase in the lattice parameters (Table 2). This showed that S1670 was involved in all types of lipid mesophases. S1670 is not only an alternative for the steric stabilization but is also able to complement F127. We demonstrated for all mesophases that the dispersion size was mainly controlled by the presence of F127. Even at a = 0.75, the sizes were found similar to the particles only stabilized by F127. This gave information about the respective contribution of emulsifiers to the particles when mixed: however incomplete [41] the coverage of the outer surfaces was ensured by F127, which fixed the particle size. Its adsorption was favored by its high molecular weight; the steric stabilization was thus ensured by the PEO corona. S1670 might split for one part in the holes between F127 molecules at the interface and for the other main part included in the mesophase. In the meantime F127 left the V2 phase when decreasing its relative content; this  symmetry for a was reflected by the disappearance of the Im3m above 0.25. For small F127 contents, S1670 behaved as a cosurfactant and drove the particle size. For further development, we could take advantage of such mixtures of emulsifiers; indeed S1670 being organized in a more compact way at interfaces, the outer layer might be improved in terms of permeability [42]. Cubosomes stabilized by pure NaCas were less stable in time than F127 or S1670 (40 days) indicating NaCas alone was not as efficient as F127 as emulsifier, probably due to protein degradation. NaCas could emulsify all the mesophases including limonene. The bicontinuous cubic structure stabilized by NaCas and the lat-

tice parameter were found similar to the F127 stabilized particles  type with 9.7 nm). This proved that pure NaCas did not inte(Pn3m grate the V2 phase at this b content, and would be available at the external interface for stabilization. When NaCas emulsified nonbicontinuous cubic mesophases we however observed a small increase of the lattice parameter compared to F127 for all mesophases (Table 2). NaCas might thus interact with the internal phase since limonene is added to the system. Despite the absence of internalization of NaCas in the V2 phase at d 100, its interaction with the outer surface of the particle was different than F127. If we compare their substitution with S1670, we showed S1670 was already involved in tuning the particle size at equal NaCas/ S1670 content, which was observed only above a = 0.75 using F127. This difference pointed out that the NaCas interfacial coverage was not as packed as F127 for the same weight content. Mixtures of NaCas/F127 also showed that the cubic internal structure could be modified. The first evidence was the unexpected observation of the H2 phase at d 100 that should only be due to a specific interaction between both emulsifiers with the lipid. The second was the measurement of smaller cubic lattice parameters than the lipid/water bulk one at intermediate ratios from 10.1 to 11.3 nm (Fig. 5). This interesting point should be investigated in a further study. In case of NaCas/S1670 mixtures, we found again a swelling of the mesophase due to the incorporation of the S1670 in the structure (Fig. 6, right). The substitution of S1670 by NaCas that should  type. As the stay at the outer surface only stabilized a V2 of Pn3m S1670 contribution to the internal structure was reduced, the

S. Serieye et al. / Journal of Colloid and Interface Science 496 (2017) 26–34

lattice parameter decreased and reached obviously that one of the bulk mesophase, thus stabilized by pure NaCas settled at the surface of particles. Concerning the emulsification by SSL at b 8, no Bragg peaks were detected on the SAXS pattern from the DU/SSL/water mixture (d 100). Thus the 20 days stabilized particles were probably not nanostructured miniemulsions. By the addition of limonene the emulsification of non-cubic lipid mesophases by SSL was possible. However no significant differences in the lattice parameters at d 50 and d 85 compared with F127 were observed, demonstrating that SSL did not participate to the internal phase of non-bicontinuous cubic particles. Lipid dispersions obtained with SSL were strongly negatively charged, i.e. the head group of the emulsifier was obviously ionized and the Na+ ions were released in the solution. Electrostatic interactions between surfactant head groups favor an increase of the mean interfacial curvature towards zero. The possibility to obtain vesicles instead of cubosomes when oleic acid is ionized by tuning the pH is known; this phenomenon is combined with a decrease in size [43]. The same effect could be expected by increasing the SSL content at d 100 resulting in dispersions composed of DU/SSL vesicles. This would be consistent with the sharp decrease in size measured by DLS above b 8 at d 100 (Fig. S2). Lecithin and WPI could emulsify all lipid mesophases, however they stabilized them during different time ranges. Like NaCas, WPI emulsified the V2 phase without integrating it (lattice parameter of 9.7 nm). This was also true for the H2 phase but we observed an effect of WPI on the discontinuous cubic lattice parameter (0.63 nm increase) and the characteristic distance of the L2 phase (0.61 nm more). Thus globular proteins from WPI seemed to easier interact with the internal phase when micellar phases are involved. The degree of interaction of lecithin with the internal lipid mesophase was only compared for the H2 phase. We obtained the same lattice parameter than with F127 meaning lecithin stayed at the outer surface of the particle. Zeta potential measurements showed an influence of the lipid content on the resulting charge of the particles (Fig. S5). Using F127, it was found to increase in absolute value when d increased. Because F127 is neutral, this behavior has to be correlated with the DU content in the particles. Dimodan U is a mixture with a small content of free fatty acids which increased with d. The negative charge at the surface of the particles would thus be explained by this fraction of ionized fatty acids of DU leading to zeta potentials from
5 to
10 mV. Concerning S1670, particles exhibited stronger negative charge surfaces reduced by an increasing DU content. S1670 is mainly composed by neutral monoesters which can be hydrolyzed in water leading to the presence of carboxylic acids in the system, and thus to additional negative charges. Consequently, the observed potential in the order of
25 mV for the lower d could be justified. As more DU was added, the surface charge density was reduced. In the DU mixture, the neutral lipids could act as co-surfactants rather than the ionized fatty acids because of electrostatic repulsions. The increasing adsorption of the neutral DU fraction could explain the decreasing charge density of the particles for larger d. In the d range of this study, the zeta potential difference with S1670 was about only 10 mV, certainly due to the good coverage of the interface by this small surfactant. As expected, dispersions stabilized by charged emulsifiers showed strong negative surface charges. At low lipid fractions, the outer surface was mainly covered by the negatively charged emulsifiers. As before, the electrostatic repulsions favored the adsorption of uncharged surfactant lipids at the outer surface of the particles. Therefore the zeta potential decreased in absolute value with more DU. However the order of magnitude was emulsifier-dependent: small surfactants (SSL, lecithin) that provide

33

tightly packed arrangement at interfaces showed the smallest difference at about 15 mV like S1670. Conversely, in the case of proteins, with a less packed coverage at the surface, the neutral fraction of DU could act as an efficient co-surfactant resulting in larger charge difference (25–30 mV).

4. Conclusions Many potential applications can be found for nanostructured lipid dispersions in cosmetics, food science, or medicine [23,24,27]. A lot of work is now dedicated to find new classes of efficient emulsifiers for those lipid mesophases [20–21]. To be useful for food applications, it was necessary to formulate and study the complex system with approved FDA products. However most of studies focused on the characterization of their internal structure stabilized by a non-food grade emulsifier F127, deemed not to interact with the mesophase. A few studies used other surfactants to overcome this problem but the long-term stabilization as does F127 was not ensured [29,30,33]. Based on their structure, their charge and their amount, a bench of useful commercially available food grade surfactants has been selected and we hypothesized that they may emulsify and stabilize these liquid crystal mesophases as efficiently as F127. Like F127, we expected that some of our stabilizers will interact with our phases in some conditions and aimed at clearly evidencing this effect. On the basis of these hypotheses, selected food grade surfactants were compared through their capacity to emulsify cubic and non-cubic lipid mesophases. This study showed that except SSL without limonene, they all could disperse monolinolein internal structures. The only surfactant capable to lastingly emulsify those particles, as did F127, was the sucrose stearate S1670. The determination of the structure (SAXS), the size (light scattering) and the surface charge (zeta potentials) allowed describing interaction with the internal structure and location of emulsifiers and mixtures. Emulsifiers were found to strongly interact with the internal structure or only to swell it. We evidenced for mixtures of sodium caseinate and F127 that an unexpected H2 phase could be observed without limonene. This pointed out a little is known so far about the interaction between the emulsifier and lipid interfaces inside the particles. Future work should focus on developing effective strategies to evaluate the stabilization and interface permeability vs. the emulsifier content in details in order to optimize the toolbox. Overall, these results are the first step towards the design of nanostructured lipid particles as nanocontainers for delivery systems to encapsulate, protect, and deliver a large variety of hydrophobic and hydrophilic molecules-based on their singular internal structure in foods.

Acknowledgements We gratefully acknowledge the Dortmund Electron Storage Ring Facility (DELTA, Dortmund, Germany) for the provision of the BL9 line and especially Christian Sternemann, Martin Schroer and Christoph Sahle for their great support. We acknowledge the European Synchrotron Radiation Facility for provision of synchrotron radiation facilities and we would like to thank Cyrille Rochas for his assistance in using beamline CRG D2AM BM02. Parts of this research were also carried out at the light source DORIS II at DESY, a member of the Helmholtz Association (HGF); we would like to thank all beamline staff for assistance in using beamline A2. Another acknowledgement goes to the MAX IV Laboratory and especially the staff of the MAX II SAXS beamline I911-SAXS.

34

S. Serieye et al. / Journal of Colloid and Interface Science 496 (2017) 26–34

Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcis.2017.01.059.

References [1] C. Czeslik, R. Winter, G. Rapp, K. Bartels, Temperature-dependent and pressuredependent phase behavior of monoacylglycerides monoolein and monoelaidin, Biophys. J. 68 (1995) 1423–1429. [2] J. Briggs, H. Chung, M. Caffrey, The temperature-composition phase diagram and mesophase structure characterization of the monoolein/water system, J. Phys. II France 6 (1996) 723–751. [3] H. Qiu, M. Caffrey, The phase diagram of the monoolein/water system: metastability and equilibrium aspects, Biomaterials 21 (2000) 223–234. [4] J. Clogston, J. Rathman, D. Tomasko, H. Walker, M. Caffrey, Phase behavior of a monoacylglycerol (Myverol 18–99K)/water system, Chem. Phys. Lipids 107 (2000) 191–220. [5] L. de Campo, A. Yaghmur, L. Sagalowicz, M.E. Leser, H. Watzke, O. Glatter, Reversible phase transitions in emulsified nanostructured lipid systems, Langmuir 20 (2004) 5254–5261. [6] J. Barauskas, T. Landh, Phase behavior of the phytantriol/water system, Langmuir 19 (2003) 9562–9565. [7] Y.-D. Dong, I. Larson, T. Hanley, B.J. Boyd, Bulk and dispersed aqueous phase behavior of phytantriol: effect of vitamin E acetate and F127 polymer on liquid crystal nanostructure, Langmuir 22 (2006) 9512–9518. [8] A. Yaghmur, M. Kriechbaum, H. Amenitsch, M. Steinhart, P. Laggner, M. Rappolt, Effects of pressure and temperature on the self-assembled fully hydrated nanostructures of monoolein-oil systems, Langmuir 26 (2010) 1177– 1185. [9] J. Gustafsson, H. Ljusberg-Wahren, M. Almgren, K. Larsson, Cubic lipid-water phase dispersed into submicron particles, Langmuir 12 (1996) 4611–4613. [10] J. Gustafsson, H. Ljusberg-Wahren, M. Almgren, K. Larsson, Submicron particles of reversed lipid phases in water stabilized by a nonionic amphiphilic polymer, Langmuir 13 (1997) 6964–6971. [11] A. Yaghmur, L. de Campo, S. Salentinig, L. Sagalowicz, M.E. Leser, O. Glatter, Oilloaded monolinolein-based particles with confined inverse discontinuous cubic structure (Fd3m), Langmuir 22 (2006) 517–521. [12] I. Martiel, S. Handschin, W.-K. Fong, L. Sagalowicz, R. Mezzenga, Oil transfer converts phosphatidylcholine vesicles into nonlamellar lyotropic liquid crystalline particles, Langmuir 31 (2015) 96–104. [13] E. Pilman, K. Larsson, E. Tornberg, Inverse micellar phases in ternary systems of polar lipids/fat/water and protein emulsification of such phases to w/o/wmicroemulsion-emulsions, J. Disper. Sci. Technol. 1 (1980) 267–281. [14] A. Yaghmur, L. de Campo, L. Sagalowicz, M.E. Leser, O. Glatter, Emulsified microemulsions and oil-containing liquid crystalline phases, Langmuir 21 (2005) 569–577. [15] A. Salonen, S. Guillot, O. Glatter, Determination of water content in internally self-assembled monoglyceride-based dispersions from the bulk phase, Langmuir 23 (2007) 9151–9154. [16] T. Landh, Phase behavior in the system pine oil monoglycerides-poloxamer 407-water at 20 C, J. Phys. Chem. 98 (1994) 8453–8467. [17] M. Nakano, A. Sugita, H. Matsuoka, T. Handa, Small-angle X-ray scattering and 13 C NMR investigation on the internal structure of ‘‘cubosomes”, Langmuir 17 (2001) 3917–3922. [18] S. Guillot, S. Salentinig, A. Chemelli, L. Sagalowicz, M.E. Leser, O. Glatter, Influence of the stabilizer concentration on the internal liquid crystalline order and the size of oil-loaded monolinolein-based dispersions, Langmuir 26 (2010) 6222–6229. [19] L. Sagalowicz, M. Michel, M. Adrian, P. Frossard, M. Rouvet, H.J. Watzke, A. Yaghmur, L. de Campo, O. Glatter, M.E. Leser, Crystallography of dispersed liquid crystalline phases studied by cryo-transmission electron microscopy, J. Microsc. (Oxford, U. K.) 221 (2006) 110–121. [20] J.Y.T. Chong, X. Mulet, L.J. Waddington, B.J. Boyd, C.J. Drummond, Steric stabilization of self-assembled cubic lyotropic liquid crystalline nanoparticles: high throughput evaluation of triblock polyethylene oxide-polypropylene oxide-polyethylene oxide copolymers, Soft Matter 7 (2011) 4768–4777.

[21] J.Y.T. Chong, X. Mulet, L.J. Waddington, B.J. Boyd, C.J. Drummond, High throughput discovery of novel steric stabilizers for cubic lyotropic liquid crystal nanoparticle dispersions, Langmuir 28 (2012) 9223–9232. [22] J.Y.T. Chong, X. Mulet, B.J. Boyd, C.J. Drummond, Accelerated stability assay (ASA) for colloidal systems, ACS Comb. Sci. 16 (2014) 205–210. [23] A. Yaghmur, O. Glatter, Characterization and potential applications of nanostructured aqueous dispersions, Adv. Colloid Interf. Sci. 147–148 (2009) 333–342. [24] I. Miloševic´, S. Guillot, M. Tadic´, M. Duttine, E. Duguet, K. Pierzchala, A. Sienkiewicz, L. Forró, M.-L. Saboungi, Loading and release of internally selfassembled emulsions embedded in a magnetic hydrogel, Appl. Phys. Lett. 104 (2014) 043701. [25] L. Sagalowicz, M.E. Leser, H.J. Watzke, M. Michel, Monoglyceride self-assembly structures as delivery vehicles, Trends Food Sci. Technol. 17 (2006) 204–214. [26] J. Ubbink, A. Burbidge, R. Mezzenga, Food structure and functionality: a soft matter perspective, Soft Matter 4 (2008) 1569–1581. [27] S. Salentinig, L. Sagalowicz, M.E. Leser, C. Tedeschi, O. Glatter, Transitions in the internal structure of lipid droplets during fat digestion, Soft Matter 7 (2011) 650–661. [28] P.T. Spicer, W.B. Small, M.L. Lynch, J.L. Burns, Dry powder precursors of cubic liquid crystalline nanoparticles (cubosomes), J. Nanopart. Res. 4 (2002) 297– 331. [29] M. Almgren, J. Borné, E. Feitosa, A. Khan, B. Lindman, Dispersed lipid liquid crystalline phases stabilized by a hydrophobically modified cellulose, Langmuir 23 (2007) 2768–2777. [30] M. Uyama, M. Nakano, J. Yamashita, T. Handa, Useful modified cellulose polymers as new emulsifiers of cubosomes, Langmuir 25 (2009) 4336–4338. [31] J. Zhai, L. Waddington, T.J. Wooster, M.-I. Aguilar, B.J. Boyd, Revisiting b-casein as a stabilizer for lipid liquid crystalline nanostructured particles, Langmuir 27 (2011) 14757–14766. [32] J. Barauskas, A. Misiunas, T. Gunnarsson, F. Tiberg, M. Johnsson, ‘‘Sponge” nanoparticle dispersions in aqueous mixtures of diglycerol monooleate, glycerol dioleate, and polysorbate 80, Langmuir 22 (2006) 6328–6334. [33] C. Nilsson, K. Edwards, J. Eriksson, S.W. Larsen, J. Ostergaard, C. Larsen, A. Urtti, A. Yaghmur, Characterization of oil-free and oil-loaded liquid-crystalline particles stabilized by negatively charged stabilizer citrem, Langmuir 28 (2012) 11755–11766. [34] L. Sagalowicz, S. Guillot, S. Acquistapace, B. Schmitt, M. Maurer, A. Yaghmur, L. de Campo, M. Rouvet, M. Leser, O. Glatter, Influence of vitamin E acetate and other lipids on the phase behavior of mesophases based on unsaturated monoglyceride, Langmuir 29 (2013) 8222–8232. [35] I. Miloševic´, V. Mauroy, H. Dabboue, S. Serieye, F. Warmont, J.-P. Salvetat, M.-L. Saboungi, S. Guillot, Synthesis and size control of polystyrene nanoparticles via ‘‘liquid crystalline” nanoemulsion, Micropor. Mesopor. Mater. 120 (2009) 7–11. [36] S. Murgia, A.M. Falchi, M. Mano, S. Lampis, R. Angius, A.M. Carnerup, J. Schmidt, G. Diaz, M. Giacca, Y. Talmon, M. Monduzzi, Nanoparticles from lipid-based liquid crystals: emulsifier influence on morphology and cytotoxicity, J. Phys. Chem. B 114 (2010) 3518–3525. [37] H.B. Wijayanti, N. Bansal, H.C. Deeth, Stability of whey proteins during thermal processing: a review, Compr. Rev. Food Sci. Food Saf. 13 (2014) 1235–1251. [38] S. Guillot, C. Moitzi, S. Salentinig, L. Sagalowicz, M.E. Leser, O. Glatter, Direct and indirect thermal transitions from hexosomes to emulsified microemulsions in oil-loaded monoglyceride-based particles, Colloids Surf. A 291 (2006) 78–84. [39] R. Mezzenga, C. Meyer, C. Servais, A.I. Romoscanu, L. Sagalowicz, R.C. Hayward, Shear rheology of lyotropic liquid crystals: a case study, Langmuir 21 (2005) 3322–3333. [40] M. Pouzot, R. Mezzenga, M. Leser, L. Sagalowicz, S. Guillot, O. Glatter, Structural and rheological investigation of Fd3m inverse micellar cubic phases, Langmuir 23 (2007) 9618–9628. [41] S. Stolnik, B. Daudali, A. Arien, J. Whetstone, C.R. Heald, M.C. Garnett, S.S. Davis, L. Illum, The effect of surface coverage an conformation of poly(ethylene oxide) (PEO) chains of poloxamer 407 on the biological fate of model colloidal drug carriers, Biochim. Biophys. Acta 1514 (2001) 261–279. [42] J.-B. Bezelgues, S. Serieye, L. Crosset-Perrotin, M.E. Leser, Interfacial and foaming properties of some food grade low molecular weight surfactants, Colloids Surf. A 331 (2008) 56–62. [43] S. Salentinig, L. Sagalowicz, O. Glatter, Self-assembled structures and pKa value of oleic acid in systems of biological relevance, Langmuir 26 (2010) 11670– 11679.