3.1. Emulsion Design
The formulation of W/O inverted emulsions involved the optimization of several variables such as: the PLs mixture, the organic continuous phase, the PLs’ concentration, and the amount of water.
Among the frequently used PLs, such as PCs, which cannot act as emulsifiers themselves [
25,
26], the chosen PLs mixture (POPE-POPG at 70/30 mol/mol), besides being in agreement with our bioinspiration purpose, should potentially behave properly as an emulsion stabilizer. As a matter of fact, this mixture exhibits specific physico-chemical properties suitable for this aim, such as: oil-solubility, low HLB value and phase behavior. Indeed, both lipids are highly soluble in organic solvents (e.g., chloroform and methanol), but slightly soluble in water. PEs are reported to have HLB values around 6, while PGs’ HLB values are even lower [
27,
28]; hence their mixtures are expected to have HLB values which would promote their use as inverted emulsion stabilizers. Moreover, the PLs mixture employed is characterized by a thermotropic phase transition of the lipid microscopic arrangement from a gel phase to a fluid liquid crystalline phase at 18 °C [
29]. Therefore, at 25 °C (i.e., our experimental temperature), PLs are in a fluid liquid crystalline phase thus potentially providing a major stability by wrapping the dispersed emulsion droplets in a viscous liquid crystalline matrix [
30].
The choice of the organic phase, namely the dispersion medium of the inverted emulsion, is extremely important and is governed by the balance between the solubility of the lipid and the stability of the inverted emulsion. Generally, alkanes with chains shorter than 10 carbons lead to the formation of highly polydisperse microemulsions where it is not possible to control with reasonable accuracy the droplet size [
9]. Taking this into account, we decided to use dodecane. Aiming at a more bioinspired choice of the organic phase, we also tested squalene, a natural compound which is extracted either from shark liver oil or vegetable oil. Squalene, a 30-carbon atom triterpene, is a highly viscous liquid which has been reported to stabilize W/O emulsion [
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20,
21,
22,
23,
24,
25,
26,
27,
28,
29,
30,
31].
The choice of the amount of water (i.e., the dispersed phase) is also crucial since it significantly affects emulsion stability. At high water contents (i.e., for volume fraction higher than 1%) phospholipids can partition between water and oil [
9]. The heavier water phase, consisting of big emulsion droplets, would quickly set at the bottom. Consequently, the droplets would coalesce, thus leading to fast emulsion destabilization. Therefore, in order to avoid a rapid coalescence, our emulsions contained no more than 0.5 vol % of water.
PLs concentration influences both emulsion stability and droplets size. It has been proved that at low PLs concentration (i.e., lower than their CAC), PLs are expected to stabilize smaller water droplets [
10]. Therefore, to provide stable nanoemulsions in our experiments PLs concentration was always fixed at 0.05 mg mL
−1, that is below our PLs CAC. Indeed, for PG and PE the CAC is generally in the micromolar range [
32,
33]. In our systems, the amount of PLs was in the order of a few nanomoles per liter.
As a last step in the emulsion design process, interfacial tension measurements have been carried out in order to optimize the preparation protocol. The obtained γ values are shown in
Table 1. Spontaneous emulsification (i.e., thermodynamic stability) occurrs only when interfacial tension is as low as 0.01 mN m
−1 [
34]. Inspection of the table shows that, even though the lipid mixtures are effective in lowering the γ values for both the considered oil/water pairs, still the values are three orders of magnitude higher than 0.01 mN m
−1. Thus, the investigated nanoemulsion can be only kinetically stable and a significant energy supply is needed for the emulsification process. For this reason, we decided to use a sonication by a tip probe for quite a long time (20 min) in an ice bath.
3.2. Dynamic Light Scattering
Water/dodecane inverted emulsions stabilized by POPE-POPG have been initially investigated. Droplet dimension distribution was evaluated by DLS immediately after sonication and over time. DLS provides information about the scatterers’ motion in terms of diffusion coefficient, from which it is possible to calculate the hydrodynamic radius by applying the Stokes–Einstein equation Equation (1). The results are reported in
Figure 2a where the scattering intensity distribution has been normalized by assuming a dependency of the mass (
M) of the droplets with respect to their hydrodynamic radius as
M RH3. As can be seen, a bimodal distribution is observed, with one population being significantly smaller than the other. The dispersed phase is mostly distributed into small droplets, whose radii is lower than 100 nm. This observation is in agreement with recent findings showing that, when the concentration of PLs is far lower than their CAC, they form a single monolayer at the interface giving rise to smaller droplets less prone to coalescence during emulsification [
10]. However, the presence of a second population of larger droplets clearly points to an unsatisfactory emulsification. Indeed, the water/dodecane POPE-POPG emulsion exhibited a very low stability: a macroscopic phase separation occurred in a few minutes, most likely driven by the fast coalescence of the bigger emulsion droplets. This does not allow the measurements to be carried out over time.
Facing this instability problem, we tested two alternative approaches. The first strategy consists in optimizing the lipid mixture by adding specific stabilizing components. The second one consists in changing the oil-phase.
By following the first approach, we redesigned the emulsifier system by adding a small amount of a glycosylated surfactant to the original lipid mixture with the aim to provide a more viscous external coating for the water droplets, thus preventing their coalescence. We chose Span 80, one of the most diffuse W/O emulsifiers. 5% (mol/mol) of Span 80 was added to the lipid mixture during the film preparation while the lipid concentration was still fixed at 0.05 mg mL
−1. Once formulated, the water/dodecane POPE-POPG-Span 80 emulsion was then characterized. The droplet size distribution was determined by DLS. The hydrodynamic radii distribution was measured immediately after sonication (t
0) and over time (three times a step of 30 min). The results at t
0 and after 1 h are shown in
Figure 2b. The system is characterized by a single monomodal distribution hinged at about 110 nm (black line). The observed population is unchanged after 1 h (red line). The same result was obtained even after 90 min (data not shown). Indeed, there was no clear phase separation for at least two hours. Thus, Span 80 greatly improves emulsion droplet dispersion as well as stability. This result has to be ascribed to the specific Span 80 features. Its low HLB value of 4.3 [
35] along with its high solubility in organic solvents makes it suitable for W/O inverted emulsion stabilization. Presenting a relatively small size, it rapidly diffuses at the interface creating a compact superficial coating. The long hydrocarbon chain stretches out from the absorbed interfacial layer, pointing towards the continuous phase, and creates a strong steric barrier. In this way, as two droplets approach each other, coalescence is hindered by the unfavored interpenetration of the lipid layers, thereby resulting in entropic repulsion [
36].
Even though this first strategy proved to be effective in enhancing emulsion stability, Span 80 inevitably changes the absorbed lipidic layer composition at water droplets boundary thus compromising the declared bioinspiration purpose. As an alternative approach, we changed the dispersion medium substituting squalene to dodecane. In this case the emulsions were formulated without adding any stabilizers like Span 80. Measurements have been carried out with the same frequency as before and the effects on the emulsification process (i.e., droplets’ size and emulsion stability) of squalene as a continuous phase are presented in
Figure 2c. One monomodal distribution is recorded at t
0, namely immediately after sonication (black line) centered at 100 nm. This population is still observed after one hour (red line) and also after 90 min from t
0 (data not shown); even in this case, no phase separation was detected for at least two hours. It is clear that the different continuous phase has strongly influenced the emulsification results. The increased stability has to be ascribed to the higher viscosity of the squalene (12 cP) [
23] compared to the dodecane one (1.34 cP) [
22]. The water droplets’ diffusion in the dispersion medium is lowered by the increased viscosity of the continuous phase, as described by equation 1. A low diffusion coefficient reduces the number of collisions, so the rate of coalescence becomes smaller [
9], thus leading to major stability over time.
Overall, both adopted strategies led to the formation of stable W/O emulsion whose droplets size distribution was fixed in the range of 100 nm. In other terms, stable W/O Pls nanoemulsions have been successfully formulated.
3.3. Electron Paramagnetic Resonance
The addition of Span 80 to the original POPE-POPG lipid matrix as well as the substitution of the continuous phase (squalene in lieu of dodecane) made W/O PLs emulsions stable enough to be characterized by means of EPR spectroscopy.
EPR measurements were carried out to analyze the microstructure of the lipid layer adsorbed at the boundary of the water droplets, by incorporating in it phosphatidylcholines spin-labeled at the different positions of the sn-2 chain (n-PCSL, with n = 5, 7, 10, 14). EPR analysis was performed both on the emulsions and on vesicles of the same lipid composition, eventually including Span 80. PLs vesicles, prepared with a well-established method (see
Section 2.4), were chosen as a reference system. Indeed, while EPR is a well-consolidated approach to characterize lipid bilayers [
37], it has been only seldom used for PLs reverse micelles [
38]. To the best of our knowledge this is the first time that the same method is used for lipid monolayers stabilizing emulsion droplets. Each probe provides different information: 5-PCSL monitors the region just underneath the hydrophilic interface of the lipid layer (or bilayer), while in the case of 14-PCSL the reporter group is either directed towards the oil phase or deeply embedded in the membrane hydrophobic core, in the case of emulsions or lipid bilayers, respectively [
39].
First, n-PCSL spectra in POPE-POPG-Span 80 water/dodecane emulsion and in POPE-POPG-Span 80 vesicles were recorded, see
Figure 3a. For all the labels, no evidence of superimposed spectra arising for the label partitioning between different environments is observed, thus confirming the labelled lipids reside exclusively at the W/O interface, participating to the Pls adsorbed monolayer.
Focusing our attention on the results obtained for the water/dodecane emulsion (black solid lines in
Figure 3a) we note that the 5-PCSL spectrum shows an evident anisotropic lineshape. The low field signal is split in two peaks, which, however, are partially overlapped, while at the high field region, two separate minima can be clearly identified. In the case of 7-PCSL the two peaks at low field coalesce in one maximum with a shoulder on the left-hand-side, while in the high field region an unresolved extremely broad signal is detected. The spectra of 10 and 14-PCSL show only three broad signals typical of a slow, isotropic motion of the label. The lines become narrower, shifting from 10 to 14-PCSL. In summary, the spectra of n-PCSL embedded in the POPE-POPG-Span80 monolayer adsorbed at the water/dodecane interface of the inverted emulsion exhibit a gradient of the anisotropy of the label motion in going from the lipid headgroups to the tail termini. This reflects an increasing rotational freedom of the acyl chain segments when moving from the constrained water/oil interface to the continuous apolar environment.
It is interesting to compare these spectra with those recorded for n-PCSL spin-labels embedded in POPE-POPG-Span 80 vesicles (dotted red lines in
Figure 3a). In this case, the 5-PCSL spectrum shows a marked anisotropic lineshape. The signal splitting both in the high and low field region is clearly resolved. The anisotropy is preserved in the 7-PCSL spectrum and still well distinguishable for 10-PCSL. In the 14-PCSL spectrum, the two peaks at low field are no longer resolved, thus coalescing in a single maximum, while at high field region a partial anisotropy is still detectable. Overall, the n-PCSL spectra observed in the bilayer show an evident mobility gradient in going from the membrane interface to the inner core. The comparison of the spectra shows that the acyl chains of the PLs absorbed at the boundary of the emulsion water droplets, with respect to the same molecules organized in bilayers, are far more capable of rotational motion. Indeed, the anisotropy is less marked at all label positions. This indicates that they are involved in a less tightly self-organized structure. Further details can be achieved by a quantitative analysis of the spectra. A parametrization of the n-PCSL spectra was realized determining the order parameter,
S, and the hyperfine coupling constant,
a’N. These parameters can be evaluated from the distance (in G) between maxima and minima of the spectra (see
Figure 3a) using the following equations:
S, which express the angular amplitude of the motion, is a measure of the local orientational ordering of the labeled segment of the lipid tail with respect to the normal to the absorbed monolayer/bilayer surface, while
a’N is an index of the micropolarity experienced by the nitroxide radical [
40].
Figure 4a,b show the dependence of
S and
a’N on the label position along the chain, n, for n-PCSL in POPE-POPG-Span 80 water/dodecane emulsion (black circles, solid black line) and POPE-POPG-Span 80 vesicles (red circle, dotted red line). Inspection of the figure reveals an
S decreasing trend for both the emulsion and the vesicle suspension. The
S values obtained for the adsorbed monolayer are always lower than those relative to the lipid bilayer. This quantitatively demonstrates a higher acyl chain mobility due to a looser structuring of the lipids in the former case. It is interesting to observe that in the adsorbed monolayer the
S decrease is steeper close to the lipid headgroups while in the bilayer the largest S drop between 10-PCSL and 14-PCSL. Indeed, the lipid packing of phospholipid chains in a bilayer produces a strong reduction of their mobility, while the adsorbed monolayer presents a much more disordered organization. This result clearly highlights the strict interconnection between the lipid organization, at the microscopic level, and the aggregate morphology, at the mesoscopic level.
The a’N trends also show a decrease for both emulsion and vesicles. This is consistent with the a’N physical meaning, this parameter being related to the micropolarity of the local environment in which the reporter group is embedded. Indeed, it can be seen that a’N decreases as the nitroxide group is shifted along the chain, corresponding to a deeper localization either in the oil continuous phase or in the hydrophobic interior of the phospholipid bilayer. However, it is interesting to observe that the overall a’N variation is much larger in the emulsion than in the membrane. Indeed, for 10- and 14-PCSL, the a’N are very low, indicating that the microenvironment of the lipid monolayer stabilizing emulsion droplets is much more hydrophobic than the interior of the lipid bilayers. At the same time, the polarity gradient across the lipid layer is much steeper in the emulsion than in the membrane, particularly close to the lipids headgroups. This difference has to be connected to the different lipid self-organization. Because of the great conformational freedom of the PLs acyl chains, the hydrophobic section of the adsorbed monolayer resembles an apolar liquid almost impermeable to water molecules. Moreover, the dodecane molecules can establish hydrophobic interaction with the phospholipid acyl chains, thus being able to solubilize in the Pls monolayer. This leads to a much more hydrophobic environment experienced by the nitroxide labels in the layer interior and, at the same time, to a disordered tail packing. Interestingly, the higher disorder favors the penetration of water molecules close to the headgroup region, as highlighted by the higher a’N observed for 5-PCSL.
n-PCSL spectra in POPE-POPG water/squalene emulsion and in POPE-POPG vesicles were also recorded—see
Figure 3b. Inspection of the figure shows that substituting the dispersion medium (squalene in lieu of dodecane) as well as changing the emulsifier mixture composition (removing Span 80) does not cause dramatic changes in n-PCSL spectra with respect to the previously analyzed system. The only noteworthy difference is a significant variation of the 14-PCSL spectrum in the emulsion. Indeed, while the two peaks at low field are still not resolved, in the high field region a first peak is followed by a broad, partially unresolved, signal. This indicates that for 14-PCSL in POPE-POPG water/squalene emulsion a partial anisotropy is still preserved, thus suggesting a more restricted motion of the phospholipids chain termini. This restriction likely derives from a partial intercalation of the bulky squalene molecules among the tails. A similar evidence is not detectable in the 7-PCSL spectrum, thus indicating the squalene entering in the monolayer to be limited to its oil-exposed boundary. On the other hand, for n-PCSL spin-labels incorporated in POPE-POPG vesicles (dotted red lines in
Figure 3b), no difference has been observed with respect to the POPE-POPG-Span 80 vesicles, the spectra being perfectly superimposable. This implies that inclusion of Span 80 does not interfere or perturb the lipid microscopic organization of the bilayer.
Even for this system, the spectra were quantitatively analyzed by the parametrization approach. The dependence of
S and
a’
N on the label position along the chain for the n-PCSL spin-labels in POPE-POPG water/squalene emulsion (black triangle, solid black line) and POPE-POPG-Span 80 vesicles (red triangles, dotted red lines) is reported in
Figure 4a,b. Analysis of the figures leads to conclusions qualitatively similar to those reached for the previously analyzed system. Indeed, even for POPE-POPG water/squalene emulsion the
S values show a steep decreasing trend while
a’
N exhibits a larger gradient across the lipid layer than in the membrane. However, significant differences arise by quantitatively comparing the
S and
a’
N values.
The
S trends are slightly different. In the presence of squalene as dispersion medium, moving the probing group down along the acyl chains,
S immediately decreases but then tends to stabilize, maintaining in the case of 14-PCSL a higher value than observed using dodecane as the oil. This is most likely caused by the insertion of the bulky and sterically hindered squalene molecules among the lipid tail termini, which leads to a decrease in the chain termini mobility [
41]. However, because of its molecular structure, squalene is not able to be enter into the monolayer interior. The latter consideration is corroborated by literature data showing that squalene allows lipid self-aggregates to be oil-free [
9]. Even in the lack of oil penetration, the monolayer interior is extremely hydrophobic. Indeed, the
a’
N values obtained for the POPE-POPG water/squalene emulsion are lower than those relative to the POPE-POPG-Span 80 water/dodecane emulsion. This evidence could be ascribed to the disordered organization of the lipids (confirmed by the low S values observed for 7-PCSL), forming a liquid-like apolar environment.
It has to be remarked that the values of S and a’N of the membranes in the presence and absence of Span 80 are perfectly matching, thus giving a quantitative confirmation of Span 80 incorporation without disturbing the PLs’ local structure.