Stabilization of solid lipid nanoparticles with glycyrrhizin

This study investigated the influence of saponin glycyrrhizin on the formation and stability of solid lipid nanoparticles. The hypothesis was that glycyrrhizin facilitates the formation of stable crystalline lipid particles due to its molecular characteristics and slows down polymorphic transition. Tristearin solid lipid nanoparticles emulsified with glycyrrhizin at varying concentrations at pH 7 and 3 were generated by hot high-pressure homogenization. The influence of glycyrrhizin on the physical stability and crystallization behavior of solid lipid nanoparticles was evaluated by dynamic and static lights scattering, electrophoretic light scattering, optical microscopy, visual observations, and differential scanning calorimetry. The results showed that glycyrrhizin formed nanosized solid lipid nanoparticles at both pH 7 and 3. The glycyrrhizin concentration and the pH applied governed the crystallization behavior and the polymorphic stability as well as the physical appearance. Overall, glycyrrhizin showed remarkable ability to stabilize solid lipid nanoparticles against polymorphic transition over time. These results are relevant for food, pharmaceutical, and cosmetic industries to form stable carrier systems by using natural plant-based saponins as surfactants.


Introduction
Saponins are an interesting group of natural plant-based surfactants that offer means to stabilize colloidal carriers. One of such colloidal carrier systems is solid lipid nanoparticles (SLN) that are used for delivery of bioactive compounds to reduce their degradation, enhance their bioavailability, and modulate controlled release [1][2][3]. SLN can be described as crystallized oil-in-water emulsions. Their properties and stability are governed by lipid and surfactant compositions, particle size, but also by the encapsulated bioactive compounds [1,2]. The main instability mechanism in solidified lipid particles is polymorphic transition that describes the changes in molecular rearrangements and packing of the lipid crystals, especially those from saturated triacyclglycerols, from α-to β′-to β-subcell crystals that is driven by thermodynamics [4][5][6]. These changes typically lead to particle aggregation or gelation due to the inability of the surfactant layer to maintain the α-subcell crystal configuration or the lack of insufficient continuous phase surfactants unable to adsorb fast to the polymorphing particle surfaces [7,8]. Therefore, the selection of appropriate surfactant type and concentration is essential for stabilization of SLN [1,2].
So far, mainly saponins from Quillaja saponaria Molina tree have been investigated in forming and stabilizing SLN and partially crystalline particles denoted as nanostructured lipid carriers [9,10]. Quillaja saponins alone and in combination with high-melting phospholipids have shown to be excellent in stabilizing the solidified lipid particles against polymorphic transition [9,10]. Therefore, we hypothesized that formation and stabilization of SLN is possible also with other types of saponins due to the basic structural similarities.
For this purpose, we selected saponin glycyrrhizin. Chemically, saponins are glycosides comprising a hydrophobic triterpene or steroidal aglycone to which one to three hydrophilic sugar chains of different lengths and compositions are attached to [11,12]. This amphiphilic structure makes the saponins highly functional that is displayed by their surface activity, technofunctionality (e.g., ability to act as emulsifiers, and foaming agents), and various biological activities [11,13]. Glycyrrhizin (or glycyrrhizic acid) is a triterpenoid monodesmosidic saponin found in licorice root (Glycyrrhiza glabra L.). It comprises a hydrophobic 18β-glycyrrhetinic acid aglycone and a hydrophilic di-glucuronic acid chain [14]. It is used as a natural sweetener, and as a medicinal plant due to its biological and pharmacological properties such as anti-microbial, antiviral, anti-inflammatory, antioxidant, and anticancer activities [15,16]. Recent studies have shown that glycyrrhizin and its aglycone can form and stabilize oil-in-water emulsions [17,18]. Based on this, we hypothesized that glycyrrhizin will form and stabilize also SLN. To our knowledge, there are no previous studies on investigating the formation of purely glycyrrhizin stabilized SLN or their stability over time.
Therefore, the aim of this study was to investigate the influence of glycyrrhizin at various concentrations on the physical and polymorphic stability of tristearin SLN at pH 3 and 7 during a three-week storage at 20 °C. The investigation focused on pH 7 and 3 due to the unusual characteristics of glycyrrhizin. The solubility and structural assembly of glycyrrhizin is based on its weak acid character that depends on the pH as depicted by its dissociation constants at 3.98 (pK a1 ), 4.62 (pK a2 ), and 5.17 (pK a3 ) [19]. At pH ≤ 4.5, glycyrrhizin is insoluble in water [14,20]. At pH 4.5-6.0, glycyrrhizin molecules can assemble to rod-like micelles and fibrils at concentrations of around 0.5-0.9% [14,21]. At pH ≥ 7, glycyrrhizin exist as monomers, and the predominantly dissociated carboxylic acid groups in the anionic form improve its solubility [14].

SLN preparation
SLN were prepared using a hot high-pressure homogenization method. Tristearin (10% w/w) as the lipid phase was melted at 85 °C in a water bath and then mixed with a hot (85 °C) surfactant solution (90% w/w). The mixture was first pre-emulsified using an ultra-turrax (Labworld, Staufen, Germany) at 24,000 min −1 for 2 min and then homogenized using a Microfluidizer Processor M-110EH-30 (Microfluidics Corp., Newton, MA, USA) equipped with a H10Z ceramic interaction chamber (diameter: 100 μm) at 500 bar with 5 cycles. The microfluidizer was pre-heated to prevent crystallization upon homogenization. The final emulsion was cooled down in an ice bath for 1 h to induce crystallization, and then the samples were stored at 20 °C.

Particle size
The surface (d 32 ) and volume (d 43 ) -based mean particle sizes and particle size distributions of SLN were assessed using a static light scattering device (Horiba LA-950, Retsch Technology GmbH, Haan, Germany) at 25 °C. Refractive indices of 1.54 and 1.33 were used for tristearin and buffer, respectively. The hydrodynamic diameter (d H , z-average) and polydispersity index (PDI) of SLN was measured using a dynamic light scattering device (NanoZS, Malvern Instruments, Worcestershire, UK) at 25 °C. The samples were diluted to 1:100 using the sodium phosphate buffer at appropriate pH to avoid multiple scattering effects.

ζ-Potential
The ζ-potential of SLN was evaluated using an electrophoretic light scattering device (NanoZS, Malvern Instruments, Worcestershire, UK) at 25 °C. The samples were diluted to 1:100 using the sodium phosphate buffer at appropriate pH.

Microstructure and visual observation
The microstructure of SLN was assessed using an optical microscope with an A-Plan objective with a numerical 1 3 aperture of 0.65 and phase contrast of Ph2 at magnification of 40x (Carl Zeiss MicroImaging GmbH, Jena, Germany). For microscopic imaging, at least four images of each sample were recorded by using a Canon Powershot G10 digital camera (Canon, Tokyo, Japan) with the following settings: Mode dial AV, white balance tungsten, flash off, digital zoom 1.7x, exposure compensation + 1, and aperture > F5.6 with a manual focus. Additionally, photographic images of the samples in test tubes were taken.

Differential scanning calorimetry (DSC)
The thermal behavior of SLN was analyzed by using a differential scanning calorimetry device DSC 8500 (Perkin Elmer, Shelton, CT, USA). Approximately 10 mg of sample was placed in an aluminum pan and hermetically sealed. The samples were then heated from 20 to 85 °C and cooled down to 10 °C at 20 °C/min. An empty sealed aluminum pan was used as a reference. The onset and peak temperatures as well as melting and crystallization enthalpy changes (ΔH in J/g) were obtained from the DSC curves by integrating the endothermic and exothermic peaks by using Pyris™ software of the instrument (Perkin Elmer, Shelton, CT, USA). The polymorphic stability was calculated from the ratio of melting enthalpy change of α-subcell crystals and crystallization enthalpy change: ΔH m(α) /ΔH C .

Statistical analysis
All measurements were performed in triplicates, and expressed as means and standard deviations.

Influence of glycyrrhizin on physical stability of SLN
First, the influence of glycyrrhizin concentration (0.25-2.0% w/w) on the physical stability of 10% (w/w) tristearin SLN at pH 3 and 7 was evaluated during 3 weeks of storage at 20 °C. The SLN produced with glycyrrhizin at pH 3 and 7 were nanosized (d H = 176-224 nm; d 32 = 148-212 nm) with monomodal (PDI < 0.2) particle size distributions directly after preparation (Tables 1, 2). This was also verified by the static light scattering data showing monomodally distributed particle sizes (Online Resource Fig. S1). The mean particle diameters of SLN produced at pH 3 reduced slightly with increasing the glycyrrhizin concentration from 0.25 to 0.5% (Tables 1, 2), however, increasing the glycyrrhizin concentration further did not lead to smaller particle sizes. At pH 7, on the other hand, the surfactant concentration had no major impact on the mean particle sizes (Tables 1, 2). Increasing concentrations of surfactants has been shown to reduce the initial SLN sizes until a sufficient surface coverage has been achieved [22]. All the SLN at pH 3 and 7 were also homogenous liquids and did not show any visible aggregation or phase separation on the day of the preparation ( Fig. 1) that was also verified by the optical microscopy imaging (data not shown). It should be noted that the freshly prepared SLN at pH 3 were visually observed to have higher viscosities than at pH 7.
At the lowest concentration of 0.25% of glycyrrhizin, the SLN at pH 7 showed the initial formation of slightly smaller particle sizes (d H = 187 nm; d 32 = 175 nm) than at pH 3 (d H = 224 nm, d 32 = 212 nm) (Tables 1, 2). The volumebased particle diameter data also showed the presence of small particles (d 43, pH 7 = 260 ± 63 nm; d 43, pH 3 = 265 ± 7 nm) ( Table 2) with monomodal distributions (Online Resource   S1). Over the 21-day storage time, however, the d 43 -values of SLN stabilized with 0.25% glycyrrhizin at pH 3 and 7 increased slightly (Tables 1, 2), indicating some instability. This corroborated with the particle size distribution data showing that some larger particles around 20-40 μm were present (Online Resource Fig. S1). At pH 7, SLN stabilized with 0.25% glycyrrhizin remained in a liquid form during the storage test, whereas at pH 3 they gelled after 14 days ( Fig. 1). At 0.5% glycyrrhizin, the SLN at pH 7 remained physically stable upon storage (Tables 1, 2, Online Resource Fig. S1) that was also confirmed by optical microscopy imaging (data not shown). On the other hand, SLN at pH 3 showed major physical instability as depicted by the increases in particle sizes and polydispersity (Tables 1, 2, Online Resource Fig. S1) and gelled after 14 days (Fig. 1). This was also corroborated with optical microscopy imaging showing aggregation of the samples at this time point (Fig. 2). At 1.0-2.0% glycyrrhizin, the particle sizes of SLN at both pH 3 and 7 remained nanosized and monomodally distributed over the 21-day of storage (Tables 1, 2, Online Resource Fig. S1). The only difference was that the SLN stabilized with 1.0-2.0% glycyrrhizin remained liquid at pH 7 ( Fig. 1). At pH 3, on the other hand, only the SLN stabilized with 1.0% glycyrrhizin remained liquid, whereas the SLN stabilized with 1.5 and 2.0% glycyrrhizin gelled after 14 days (Fig. 1). Overall, the ζ-potential measurements revealed that all the SLN at pH 7 and 3 were negatively charged. Some small variations were detected in the net charges as a function of glycyrrhizin concentrations, showing that increasing the glycyrrhizin concentration resulted in slightly more net negatively charged particles at pH 7 (ζ 0.25-0.5% = -52 ± 3 mV; ζ 1.0% = -59 ± 1 mV; ζ 1.5% = -65 ± 1 mV; ζ 2.0% = -71 ± 1 mV) as well as at pH 3 (ζ 0.25-1.0% = -28 ± 1 mV; ζ 1.5-2.0% = -35 ± 1 mV). Similarly negative charges (ζ = -24--55 mV) and increasingly negative charges with increasing glycyrrhizin concentrations have also been reported in oil-in-water emulsions [17]. This magnitude of charges at any glycyrrhizin concentration should have provided enough electrostatic repulsion between the SLN to stabilize them. This indicated that the observed particle size increases in some samples were not based on electrostatic attraction. Further insights into this mechanism will be discussed in section 'Influence of glycyrrhizin on the polymorphic stability of SLN'.
The differences in the net negative charges of SLN at pH 7 and 3 can be explained by the different dissociation behavior of the carboxylic acid groups in glycyrrhizin [19,23]. These carboxylic acids exist in fully ionic forms at pH > 7, in both ionic and non-ionic forms between pH 2-6, and fully non-ionic forms < pH 2 [19]. This is why glycyrrhizin becomes less soluble upon lowering the pH [14]. This also explains the higher initial viscosity of the SLN at pH 3 compared to SLN at pH 7, as a low enough solubility is required to enable network formation [24]. In physical gels, where the molecular interactions are a combination of hydrophobic, electrostatic, and hydrogen bonding, the gelation time depends on the molecular rearrangements before reaching an equilibrium state [25]. This could be a reason why gelation was observed in SLN at pH 3 after 14 days of storage.
Saha et al. [26] reported that glycyrrhizin in aqueous solutions began to form fibrils at 0.025% (w/w), whereas a practically full conversion from the monomeric forms into fibrous structures occurred at 0.25% (w/w), and the glycyrrhizin solution became highly viscous. Increasing the glycyrrhizin concentration to 0.3% (w/w) resulted in formation of a hydrogel. This self-assembly of glycyrrhizin molecules into long, right-hand twisted anisotropic (2.5 nm thickness with 9 nm periodicity) fibrils was due to the hydrophobic aglycones aligning laterally with the diglucuronic acid chains positioning themselves to the water phase [26]. Therefore, the role of glycyrrhizin fibrils in the eventual gelation of the SLN samples at pH 3 should be taken into consideration. The glycyrrhizin concentration used in this study (0.25-2.0%) is close or above the minimum gelation concentration observed for aqueous glycyrrhizin (0.3%) [26]. Nevertheless, in dispersed systems, the concentration of surfactant in the aqueous phase decreases as a high amount is adsorbed to the particle interfaces. Therefore, the gelation occurs at a higher concentration than in purely water-based system. This was observed in oil-in-water emulsions (60% olive oil, pH not given) prepared at elevated temperature (80 °C) where the emulsions remained liquid at 0.5% glycyrrhizin concentration, but formed soft gels at 1% glycyrrhizin concentration after cooling down and storing at 25 °C for 12 h due to the formation of glycyrrhizin fibrils [27]. It should be noted though that the oil droplet concentration also affects the viscosity of the emulsions.
On the other hand, the fibrillar structure formation of glycyrrhizin has been demonstrated to occur at pH 5 to 6 [14]. This indicates that the formation of glycyrrhizin fibrils at pH 3 is unlikely as the glycyrrhizin molecules have very low water solubility in the more acidic environment. Nevertheless, the glycyrrhizin solution was heated up before the SLN formation, thus leading to improved solubility, which may have induced self-assembly of some fibrillar structures. However, it is not known if the formation of fibrils is possible, especially when glycyrrhizin molecules have less charged groups at lower pH values [19]. This needs to be confirmed in future studies. Nevertheless, as the SLN samples were cooled down and stored at room temperature directly after preparation, the glycyrrhizin molecules in the continuous phase would have become insoluble again. Therefore, it is likely that the gelation was due to this insolubility as explained above. At pH 7, no gelation due to the formation of glycyrrhizin fibrils is possible, because the glycyrrhizin molecules exist as monomers [14].

Influence of glycyrrhizin on thermal behavior of SLN
In this series of experiments, we investigated the influence of glycyrrhizin (0.25-2.0% w/w) on the melting and crystallization behavior of 10% (w/w) tristearin SLN at pH 3 and 7 during 21 days of storage at 20 °C.

Melting behavior
The DSC heating thermographs showed the presence of two endothermic peaks at ~ 54-56 °C and ~ 68-71 °C in SLN emulsified with glycyrrhizin regardless of the used surfactant concentration, pH, or storage time (Fig. 3a, b) which correspond to the melting of α-and β-subcell crystals of the bulk tristearin, respectively [4]. These melting temperatures (T m ) of the tristearin SLN stabilized with glycyrrhizin were ~ 2-5 °C lower than observed in bulk tristearin (T m(α) = 52 °C; T m(β) = 73 °C) [28,29]. This agrees with previous literature [28,30]. Upon heating, a small exothermic peak at ~ 61-63 °C at pH 7 and at ~ 60-63 °C pH 3 was also observed at all surfactant concentrations during storage ( Fig. 3a-b) that can be explained by a recrystallization event of α-to β-subcell crystals [31,32]. The DSC heating thermographs of the SLN stabilized with glycyrrhizin showed that the α-subcell crystals were the predominant crystal species compared to the β-subcell crystals (Fig. 3a, b). The polymorphic stability of the SLN over time is discussed in section 'Influence of glycyrrhizin on the polymorphic stability of SLN'.
Minor crystallization events close or slightly below the crystallization temperature of bulk lipids has been observed in other studies, mainly in interfacial layers containing saturated phospholipids [7,9,29,[33][34][35][36]. These studies indicated that the phospholipids with their lipid tails may interact with the triacylglycerol molecules and solubilize them to become a part of the interfacial layer and subsequently co-crystallize at higher temperatures. Nevertheless, the exact mechanism remains unclear. The glycyrrhizin molecule with its hydrophobic triterpenoid structure, however, does not enable solubilization of the tristearin molecules on the interfacial layer. This has been also indicated for bile salts with a hydrophobic steroidal structure [33] and for Quillaja saponins with a hydrophobic triterpene backbone [9]. The monodesmosidic saponins such as glycyrrhizin typically adopt a side-on configuration where the aglycone orients itself to the lipid phase and the hydrophilic sugar chain protrudes to the aqueous phase [37]. The short diglucuronic acid moiety in glycyrrhizin underlines its low hydrophilicity with hydrophobic surfaces dominating over the polar surfaces, thus limiting glycyrrhizin's solubility. The low solubility as well as the charge properties contribute to glycyrrhizin's poor ability to efficiently pack on the interfaces, leading to formation of inelastic layers [17,37]. Therefore, the minor crystallization events are most likely due to the presence of small number of flocculated or coalesced lipid particles that crystallize at higher temperatures [7,33,35,38]. Increased particle sizes or aggregation were only observed in SLN at pH 3 at low glycyrrhizin concentrations (0.25-0.5%) and at pH 7 at 0.25% glycyrrhizin concentration (Table 1, 2). At 0.25% glycyrrhizin concentration, the ΔH-values of these small exothermic peaks at ~ 44-48 °C did not increase over time at both pH 3 (ΔH t0 = -10.3 ± 1.3 J/g; ΔH t21d = -14.5 ± 4.3 J/g) and pH 7 (ΔH t0 = -8.1 ± 0.5 J/g; ΔH t21d = -8.3 ± 0.8 J/g). However, as the d 43 -values increased over time (Table 2), this indicated that only some flocculation occurred. For the SLN emulsified with 0.5% glycyrrhizin at pH 3, on the other hand, increased ΔH-values for the exothermic peaks at ~ 48 °C were detected after 14 (ΔH t14d = -37.4 ± 6.8 J/g) and 21 days of storage (ΔH t21d = -89.0 ± 5.6 J/g), indicating the presence of aggregated particles that coalesced upon heating. This explanation would also correspond with the DSC data showing that less (pH 7) or none (pH 3) of these small exothermic events were observed at higher glycyrrhizin concentrations (≥ 1%), indicating that the interface was more saturated with the surfactant and thus the particles were more physically stable. At 0.25-1.0% glycyrrhizin, the two major crystallization events in SLN at pH 7 were 15 °C (T C ≈ 37 °C) and 22-23 °C (T C ≈ 30-33 °C) lower (Fig. 3c) than reported for bulk tristearin (T C ≈ 52 °C) [28]. At 1.5-2.0% glycyrrhizin, these crystallization temperatures were 14 °C (T C ≈38 °C) and 19 °C (T C ≈ 30-33 °C) lower than that of the bulk lipid (Fig. 3c). At pH 3, the final crystallization temperatures were 23 °C (T C ≈ 29 °C) and 24 °C (T C ≈ 28 °C) lower for SLN emulsified with 0.25% and 0.5-2.0%, respectively, (Fig. 3d) than for the bulk lipid. The lower crystallization temperature of SLN compared to that of the bulk lipid is due to the fact that the lipids are confined within the small-sized dispersed particles that remain in a supercooled state [3,39]. This is because a certain activation energy has to be overcome to induce crystallization in dispersed systems [3,39]. In bulk lipids, the crystallization is initiated via heterogenous nucleation due to the presence of impurities. In dispersed lipids, the number of impurities in the overall lipid particles is very low, and therefore they crystallize via homogenous nucleation upon reaching a low enough temperature [40]. Based on this, all the SLN at pH 3 showing a single exothermic peak (Fig. 3d) crystallized via homogenous nucleation. This is similar to tristearin SLN stabilized with Quillaja saponins that crystallized at 29 °C [9].
On the other hand, the two major exothermic peaks detected at higher temperatures for SLN at pH 7 than at pH 3 ( Fig. 3c) indicated that the crystallization was initiated via heterogenous nucleation. In dispersed lipids, heterogenous nucleation has been shown to occur via templating effect of the interfacial surfactant layer [9,10,28] and in the presence of excess of surfactant aggregates or reversed micelles in the aqueous phase [41]. This has typically occurred in the presence of emulsifiers such as saturated phospholipids that crystallize at a temperature close to that of the melting point of the saturated carrier lipid [9,10,29,33,42]. However, previous studies on Quillaja saponin stabilized SLN (pH 7) do not show such a templating effect [9,10]. Quillaja saponins, however, are a group of highly amphiphilic compounds with different structures comprising typically bidesmosidic triterpenes such as quillaic acid aglycones with varying sugar chain lengths and compositions [43,44]. They form highly elastic interfaces most likely due to strong hydrogen bonding [37,45], whereas the glycyrrhizin interfaces are inelastic [17,37]. Moreover, the successive exothermic events between ~ 30 and ~ 38 °C at pH 7 (Fig. 3c) suggest crystallization via heterogenous nucleation with different particle size distributions, where the larger SLN crystallize first and the smaller ones crystallize at lower temperature. It should be noted that the polydispersity index data from the dynamic light scattering measurements (Table 1) reporting initially monodisperse particle sizes (PDI ≤ 0.2) does not exclude the presence of differently sized nanoparticles within the particle population. For a perfectly uniform population of particles the PDI would be 0.0, which is not the case here. The presence of differently sized nanoparticle populations, even within the narrow and monomodally distributed particle sizes, ranged from around 0.05 μm up to 1000 μm (Fig.  S1). This would therefore facilitate crystallization via heterogenous nucleation of differently sized nanoparticles at different temperatures as suggested. Previous studies have shown that decreasing the particle size of SLN leads to lower crystallization temperatures [22,46]. At pH 7, crystallization of the SLN occurred at a higher temperature when higher concentrations of glycyrrhizin (1.5-2.0%) were used (Fig. 4). At pH 3, however, the crystallization temperatures were not dependent on the surfactant concentration (Fig. 4). Stepwise increases in crystallization temperatures with increasing surfactant concentrations have been previously reported [41]. At higher surfactant concentrations, the interfaces become fully saturated and more tightly packed and any unadsorbed surfactant molecules remained in the continuous phase, thereby altering the temperatures at which heterogenous nucleation occurs [34,41].

Influence of glycyrrhizin on the polymorphic stability of SLN
To gain insights into the polymorphic stability of the SLN stabilized with glycyrrhizin during the 3-week storage test at 20 °C, the ratio of melting enthalpy change of α-subcell crystals (ΔH m(α) ) over the crystallization enthalpy change (ΔH C ) was calculated (Fig. 5). A high ratio depicts that the lipid crystals are predominantly in their α-subcell crystal configuration, whereas a low ratio describes a polymorphic transition to β-subcell crystals.
The results showed that the all the freshly prepared SLN at pH 3 and 7 had lipid crystals composed of purely α-subcell crystals (Fig. 5). Over time, SLN stabilized with glycyrrhizin retained a remarkable ratio of the α-subcell crystals, indicating that glycyrrhizin was very efficient in slowing down polymorphic transition. Only at low glycyrrhizin concentrations, SLN showed more polymorphic instability: At pH 7, the lowest polymorphic stability was observed in SLN stabilized with 0.25% glycyrrhizin after 21 days (Fig. 5a). This also explains the increased particle sizes and polydispersity detected over time (Table 1, 2, Online Resource Fig. S1) as discussed in section 'Influence of glycyrrhizin on physical stability of SLN'. At pH 3, SLN stabilized with 0.5% glycyrrhizin showed a rapid polymorphic transition from α-to β-subcell crystals after 14 days (Fig. 5b) that coincided with the detection of large particle aggregates in the samples (Table 1,  at pH 3 showed the second lowest ΔH m(α) /ΔH C -ratio after 21 days of storage, also demonstrating some polymorphic transition (Fig. 5b). Again, the polymorphic transition concurred with the presence of some larger particle sizes (Tables 1, 2, Online Resource Fig. S1). Polymorphic transition from α-to β-subcell crystals involves a shape change from a spherical particle into a platelet-shape that increases the surface area of particles [8,36,47]. This forms particles with uncovered hydrophobic surfaces that aggregate through hydrophobic interactions and may even lead to gelation [7,8]. Such aggregation is especially prevalent, if the concentration of the surfactant in the continuous phase is low upon polymorphic transition and the surfactant molecules do not cover the exposed lipid surfaces fast enough. Increasing the surfactant concentration in the aqueous phase has been shown to stabilize SLN and prevent aggregation [34]. This was also observed in this study, showing that higher glycyrrhizin concentrations were better at stabilizing the SLN against polymorphic transition (Fig. 5).
Overall, such high polymorphic stability as observed in this study (Fig. 5) has been previously shown only in tristearin SLN stabilized with Quillaja saponins alone and Quillaja saponin-high-melting lecithin combinations [9,10]. Even combination of bile salt and high-melting lecithins in stabilizing the tristearin SLN interface have been shown to lead to a faster polymorphic transition, although still retaining the physical stability of the particles [33].

Key insights
Based on the data, glycyrrhizin was very efficient in forming initially nanosized SLN at all concentrations (0.25-2.0%) at both pH 3 and 7 (Table 1, 2). Over time, however, the SLN stabilized with lower glycyrrhizin concentrations were less stable than the ones emulsified with higher surfactant concentrations. This indicated that the surface coverage of the SLN at low concentrations was not high enough to fully prevent polymorphic transition from α-subcell crystals to more thermodynamically stable β-subcell crystals. Moreover, the lack of sufficient amount of continuous phase surfactant at low concentrations prevented further adsorption of the molecules to the uncovered surfaces upon polymorphic transition, thus leading to particle size increases, flocculation and even gelation.
Interfacial stabilization by glycyrrhizin can be attributed to hydrophobic interactions via van der Waals forces and electrostatic repulsion [17]. The hydrophobic interactions are favored because glycyrrhizin has a low surface activity as the hydrophobic surfaces of the molecule dominate over the polar surfaces [14,17]. The negative charges generated by the carboxylic acid groups in the glycyrrhizin molecules [19,23] contribute to electrostatic repulsive forces between the particles. The pH plays a major role for the molecular characteristics of glycyrrhizin that can be attributed to its dissociation behavior [19,23]. The following section describes the key insights of tristearin SLN stabilization using glycyrrhizin (0.25-2.0%) at pH 3 and 7:

At pH 7
• Glycyrrhizin molecules exist mainly as highly soluble monomers [14]. • Glycyrrhizin acted as a template to induce heterogenous interfacial crystallization of the SLN at 39-40 °C (Fig. 3c). The two-fold crystallization event observed at 39-40 °C and 33-35 °C (Fig. 3c) suggested a distribution of particles that started crystallizing at different temperatures with larger particles crystallizing at higher temperatures and the smaller sized particles crystallizing at lower temperature. • At 0.25% glycyrrhizin, the initially nanosized SLN showed some physical instability after 14 days of storage as some larger particle sizes were detected (Tables 1-2, Online Resource Fig. S1) that were attributed to polymorphic transition from α-to β-subcell crystals in some of the tristearin particles (Fig. 5a). • At 0.5-2.0% glycyrrhizin, SLN remained physically stable during the three-week storage time (Tables 1-2, Online Resource Fig. S1) as the higher concentrations of glycyrrhizin effectively reduced polymorphic transition of the tristearin crystals (Fig. 5a). • The liquid appearance (Fig. 1) throughout the storage time was due to the predominantly good polymorphic stability of the SLN (Fig. 5a), and to the monomeric form of glycyrrhizin molecules at pH 7 [14] preventing any network formation and subsequent gelation.

At pH 3
• Glycyrrhizin molecules are insoluble in water [14]. • Glycyrrhizin was not able to act as a template to induce heterogenous nucleation and the SLN crystallized via homogenous nucleation at 28-29 °C (Fig. 3d). • At 0.25-0.5% glycyrrhizin, the SLN formed gels (Fig. 1) as they were polymorphically unstable (Fig. 5b). • At 1.0% glycyrrhizin, the SLN were physically and polymorphically stable during the 3-week storage time (Tables 1, 2, Online Resource Fig. S1, Fig. 5b). The dispersions remained in a liquid form (Fig. 1), although the viscosity increased after 2 weeks of storage as observed visually, indicating the impact of time kinetics on molecular reassembly [25]. The liquid character persisted because the concentration of the insoluble glycyrrhi-zin molecules remaining in the aqueous phase was low enough to prevent gelation even after 3 weeks of storage. • At 1.5-2.0% glycyrrhizin, the SLN were physically and polymorphically stable (Tables 1, 2, Online Resource Fig. S1, Fig. 5b). The dispersions remained as kinetically stable liquids until they gelled after 2 weeks of storage (Fig. 1). This may be attributed to the excess of insoluble glycyrrhizin molecules in the continuous phase that gelled after reaching a more thermodynamically stable state.

Conclusions
This study examined the influence of glycyrrhizin at varying concentrations on the physical stability and crystallization behavior of solid lipid nanoparticles made with tristearin as a carrier lipid at pH 7 and 3. The investigation showed that glycyrrhizin at a high enough concentration formed kinetically stable nanosized solid lipid nanoparticles with remarkable polymorphic stabilities at both pH 7 and 3, thereby confirming the hypothesis. Interestingly, the crystallization behavior differed depending on the pH applied. At pH 7, solid lipid nanoparticles emulsified with glycyrrhizin crystallized via heterogenous interfacial nucleation. At pH 3, they crystallized via homogenous nucleation. The stabilization mechanism of amphiphilic glycyrrhizin on the particle interface has been suggested to base on mainly on the hydrophobic interactions due to the more hydrophobic nature of the molecule, but also on its charge properties resulting in electrostatic repulsion between the particles. The reason for the templating effect of glycyrrhizin at pH 7 compared to pH 3 is not clear, although the different solubility and molecular characteristics may be a likely cause. Ongoing research on glycyrrhizin on stabilization of solid lipid nanoparticles at pH 5 may give some new insights into this. Moreover, the reason for the high polymorphic stability of saponins remains unclear, especially because glycyrrhizin forms inelastic interfaces that would suggest that they would be prone to polymorphic transition. It seems that the predominantly hydrophobic nature of the glycyrrhizin molecules on the hydrophobic tristearin particle interface leads to thermodynamically favorable interactions that are not easily broken down. This has to be looked into in future studies. Overall, this study contributes to a better understanding of saponins in stabilizing crystallized lipid particles. Furthermore, the exceptional ability of saponins such as glycyrrhizin to slow down polymorphic transition offers opportunities to form stable solid lipid (nano) particles in food, pharmaceutical and cosmetic fields.