Avocado-derived polyols for use as novel co-surfactants in low energy self-emulsifying microemulsions

Avocado (Persea americana Mill.; Lauraceae) seed-derived polyhydroxylated fatty alcohols (PFAs) or polyols (i.e., avocadene and avocadyne) are metabolic modulators that selectively induce apoptosis of leukemia stem cells and reverse pathologies associated with diet-induced obesity. Delivery systems containing avocado polyols have not been described. Herein, natural surface active properties of these polyols are characterized and incorporated into self-emulsifying drug delivery systems (SEDDS) that rely on molecular self-assembly to form fine, transparent, oil-in-water (O/W) microemulsions as small as 20 nanometers in diameter. Mechanistically, a 1:1 molar ratio of avocadene and avocadyne (i.e., avocatin B or AVO was shown to be a eutectic mixture which can be employed as a novel, bioactive, co-surfactant that significantly reduces droplet size of medium-chain triglyceride O/W emulsions stabilized with polysorbate 80. In vitro cytotoxicity of avocado polyol-SEDDS in acute myeloid leukemia cell lines indicated significant increases in potency and bioactivity compared to conventional cell culture delivery systems. A pilot pharmacokinetic evaluation of AVO SEDDS in C57BL/6J mice revealed appreciable accumulation in whole blood and biodistribution in key target tissues. Lastly, incorporation of AVO in SEDDS significantly improved encapsulation of the poorly water-soluble drugs naproxen and curcumin.

leads to their spontaneous self-assembly at the oil-water interface where the polar portion of the polyol aligns towards the aqueous phase and the apolar aliphatic chain aligns towards the oil phase, which collectively modifies film curvature. Second, the self-assembly of polyols at the interface increases the elasticity of the interfacial film (referred to as the bending moduli) [15][16][17] .
Microemulsions as delivery-vehicles for cosmetic, food and pharmaceutical formulations are of great interest due to their low energy requirements (spontaneous formation), thermodynamic stability, low viscosity and high solubilization capacity 18,19 . However, translation of microemulsions into practice is limited due to high surfactant concentrations typically required, the need for co-surfactants (exceeding 10%), and low drug loading efficiency.
In this study, we tested the ability of avocadene, avocadyne and their mixtures to spontaneously form O/W nano and microemulsions when incorporated with polysorbates and medium chain triglyceride oils. Using differential scanning calorimetry (DSC) and powder X-ray diffraction (XRD) we explored the unique physical properties of avocado polyols that enable their molecular self-assembly in SEDDS. Further, we examined if avocado polyol SEDDS have enhanced bioactivity in vitro and if they accumulate in blood and key target tissues when delivered orally to mice. Lastly, we tested the ability of AVO to act as a novel co-surfactant and its impact on droplet size of poorly water-soluble drugs encapsulated in SEDDS.

Results
Avocado polyol physical characterization studies. Purification of avocado seed ethyl acetate extracts yielded a fraction pure in avocadene and avocadyne (Fig. 1A) at a 1:1 (mol:mol) ratio (AVO). Commercially available standards of pure avocadene and avocadyne were purchased for physical characterization studies and also used to generate samples with a 3:1 or 3:2 molar ratio of avocadene:avocadyne. DSC showed avocadyne had a significantly higher melting point than avocadene, as previously reported by Kashman et al. who used a Fisher-Johns apparatus 1 (Fig. 1B). Interestingly, AVO had the lowest peak melting point suggesting that it may be a eutectic composition of avocadene and avocadyne (Fig. 1C). A eutectic composition refers to a binary mixture of two structurally distinct solids that exhibit a depressed melting point relative to either pure component 20 . Importantly, these trends were unaltered when melting onset temperatures for all avocado polyol samples were assessed (data not shown). Furthermore, AVO exhibits a lower value for ΔS m (the ratio of melting enthalpy (ΔH m ) to the melting point (T m ) referred to as entropy of melting) compared to pure avocadyne or 3:1 or 3:2 mixtures (Fig. 1D). Collectively, the lower melting point and entropy of AVO in comparison to the other avocado polyol samples is suggestive of a more disordered or liquid-like crystal. Eutectic mixtures identified in other research areas also have low melting points and entropies which is explained by significant differences in intermolecular interactions that introduce greater molecular mobility or disorder [21][22][23] .
X-ray powder diffraction was performed to determine the molecular arrangement of avocado polyol crystal structures. Measurements in the small and wide angle X-ray scattering regions (SAXS and WAXS, respectively) provide information on the longitudinal and lateral packing of molecules in a crystalline structure, respectively 24 . Diffraction patterns for all avocado polyols in the SAXS region displayed a lamellar crystal arrangement of similar size with characteristic reflections associated with the [001] molecular plane and its higher orders, i.e., [002], [003], [004], etc., evident at the expected ratio of spacings (1:1/2:1/3:1/4:1, and so on) (Supplementary Fig. 3A-E). However, avocadyne lamellae were generally larger and two species of 31 and 40 Angstroms were evident in the [001] plane suggesting the presence of two polymorphs or two geometric isomers. The WAXS patterns for avocadene and mixtures of avocadene and avocadyne were very similar, suggesting that the long hydrocarbon chains of the fatty alcohols pack laterally in similar fashion or in the same polymorphic form. Avocadyne however, displayed a different WAXS patterns than the rest of the samples, suggesting a different unit cell structure. Indexing and unit cell identification were not attempted due to concerns over crystal purity. The observed eutectic behaviour of AVO was attributable to its domain size (i.e., the persistence length and the physical thickness of the nanocrystals in the [001] crystal planes' direction). This analysis showed that the domain size of AVO was noticeably smaller compared to all other avocado polyol samples whereas avocadyne exhibited the largest domain size (Fig. 1D). The relationship between small domain sizes and lower melting points has widely been reported for both eutectic 25,26 and monotectic systems 27 . Furthermore, our XRD analysis on structurally related odd-numbered carbon lipids (i.e., 1-heptadecanol, heptadecanoic acid and 16-heptadecynoic acid) showed larger lamellae (SAXS diffraction peaks), greater domain sizes, and different lateral packing of hydrocarbon chains (WAXS diffraction peaks) compared to avocado polyols (Supplementary Fig. 4A-C).

SEDDS development and characterization.
Our previous work has shown that AVO and its individual constituents have limited aqueous solubility. Thus, we evaluated if existing O/W self-emulsifying systems, commonly used in food and pharmaceutical formulations 18,28,29 , could be used to easily incorporate avocado polyols. Supplementary Table 1 shows a list of formulations that were tested for self-emulsification. Of the SEDDS tested, NeobeeM5 or coconut oil in combination with Tween 80 or Cremophor EL (CrEL), at an oil/surfactant ratio of 1:1, were the most suitable SEDDS, when diluted 10 folds in phophate-buffered saline (PBS) ( Supplementary  Fig. 5). Higher weight ratios of NeobeeM5 to Tween 80 or CrEL (greater than 3:1) generally resulted in larger droplet sizes with higher polydispersity (Supplementary Fig. 5). The cytotoxicity of 1:1 NeobeeM5-CrEL SEDDS (mean droplet diameter of ~80 nm) compared to NeobeeM5-Tween 80 SEDDS (mean droplet diameter of ~200 nm) was significantly greater in AML (OCI-AML-2) and non-AML cell lines (INS-1 (832/13), Caco-2, and HepG2) ( Supplementary Fig. 6). As such, only 1:1 NeobeeM5-Tween 80 was further studied as the chosen SEDDS system.
Avocado polyols self-assembled at the O/W interface of the chosen SEDDS. Avocado polyols (1-20 mg) were pre-dissolved in 100 µL of the oil/surfactant phase by heating to 75 °C for 2 hr, after which 900 µL water phase (PBS) was added directly to the oil phase and vortexed for 30 sec (Fig. 2A). This method of preparation was necessary to produce consistent emulsion droplet sizes; another method where hot oil phase was added to PBS under constant stirring did not produce consistent results. When 1-2% (w/w) AVO was incorporated into SEDDS, significant reduction in turbidity (as characterized by an increase in transparency) and mean droplet size were observed where 2% (w/w) AVO reduced SEDDS droplet diameter below 25 nm (Fig. 2B). Ambient temperature polydispersity and droplet size measurements increased for 1.5-2% (w/w) AVO containing SEDDS over four weeks compared to control SEDDS (Fig. 2C). Cryo-transmission electron microscopy (TEM) images of the control and 2% (w/w) AVO emulsions highlighted the drastic reduction in mean droplet size caused by AVO (Fig. 2D). Similarly, AVO significantly reduced droplet diameter of 1:1 coconut oil-Tween 80 SEDDS ( Supplementary Fig. 7A,B) and 1:1 NeobeeM5-CrEL SEDDS ( Supplementary Fig. 7C,D) in a concentration dependent manner. Hydrophilic co-surfactants like glycerol, propylene glycol and ethanol have also been reported to penetrate the interfacial film and decrease droplet size of O/W emulsions 15,16,30 ; however, no such effects were observed in the chosen SEDDS system (data not shown). Collectively, this highlights AVO as a novel and effective co-surfactant with potential to be utilized in other delivery systems.
Next, we incorporated pure avocadene or avocadyne into SEDDS to determine their individual effects. Avocadene in SEDDS behaved similar to AVO where 0.5-2% (w/w) avocadene caused a significant reduction in droplet diameter compared to control (Fig. 4A,B). Surprisingly, avocadyne did not exhibit the same behaviour as AVO or avocadene in the chosen SEDDS system and microemulsions did not form when water phase was added to hot oil phase containing 1-20 mg of avocadyne (Fig. 4C,D). However, fine transparent microemulsions for 1-2% (w/w) avocadyne were observed after application of heat (75 °C for 3-5 minutes) and keeping samples at 37 °C. www.nature.com/scientificreports www.nature.com/scientificreports/ DSC was also used to evaluate the melting temperatures of avocado polyols when incorporated in only the oil/ surfactant phase (at a concentration of 200 mg/mL) of the SEDDS (1:1 NeobeeM5-Tween 80). This analysis also revealed that avocadyne exhibited larger thermal and entropic parameters compared to all other polyol samples ( Supplementary Fig. 8A,B). Collectively, differences in crystal structure and packing as well as solubility of avocado polyols in the chosen oil phase are integral factors that impact self-assembly.

Stability of avocado polyol SEDDS.
Investigation on the mechanism of destabilization of avocado polyol SEDDS revealed that a combination of Ostwald ripening and coalescence take place as determined by regression analysis on the cube of mean droplet radius (r 3 ) (Ostwald ripening model) and inverse of the square of mean droplet radius (1/r 2 ) (coalescence model) 30 (Supplementary Table 2). Particle size distribution (PSD) analysis over time provided additional insight into the kinetics of the two destabilization mechanisms 31 . Both PSD peak broadening (suggestive of coalescence occurring at a faster rate) and sharpening (suggestive of Ostwald ripening following first order kinetics) was observed over time in a concentration independent manner for all polyol SEDDS (Supplementary Figs. 9A-C and 10A-F). In contrast, further dilution of all the described avocado polyol SEDDS www.nature.com/scientificreports www.nature.com/scientificreports/ (up to 1000 folds in PBS) slowed droplet size growth significantly (data not shown); a finding widely reported for micro and nano-self-emulsifying systems 32,33 . PBS was chosen as the aqueous phase for all SEDDS characterization studies to ensure no changes to osmotic and pH balance in vitro and in vivo 34 . Zeta potential values for all SEDDS prepared in PBS at 10-1000 fold dilutions were not significantly different (i.e., less than −6 mV) (data not shown). Thermodynamic stability was assessed for freshly prepared avocado polyol SEDDS using centrifugation and freeze-thaw tests. AVO SEDDS (up to 2% (w/w)) did not flocculate or cream after ultracentrifugation ( Supplementary Fig. 9D). AVO SEDDS freeze at −20 °C, but an isotropic emulsion is reformed after thawing at 37 °C ( Supplementary Fig. 9E). Bright field microscopy of a six-month aged 2% (w/w) AVO SEDDS showed large crystals (a combination of AVO, oil and surfactant) which disappeared after heating (45 °C) for 3-5 min ( Supplementary Fig. 9F) suggesting that a destabilized polyol microemulsion can undergo self-emulsification with heating. Polarized light microscopy confirmed that all avocado polyol SEDDS were isotropic colloidal dispersions void of lamellar liquid crystals ( Supplementary Fig. 11), as no birefringence was observed between crossed polarizing plates 35 . These properties were observed for all freshly prepared avocado polyol SEDDS. www.nature.com/scientificreports www.nature.com/scientificreports/ In vitro activity of avocado polyol SEDDS. Previously, it has been reported that AVO induces mitochondria-mediated death in AML cells 5,6 where AVO was delivered using a conventional cell culture vehicle of dimethyl sulfoxide (DMSO). Solubility of long aliphatic chain avocado polyols in DMSO is likely a limiting factor for in vitro cellular delivery, thus we directly compared the cytotoxicity of avocado polyols in AML cells when delivered using DMSO or SEDDS (Fig. 5A-E; Table 1). All avocado polyols delivered in SEDDS lowered AML cell viability with doubled potency compared to DMSO delivery.
Avocado polyols selectively exert cytotoxicity in AML cells while sparing non-AML cells 5,36 . Hence, we tested the cytotoxicity of avocado polyols (delivered in DMSO or SEDDS) in non-AML, adherent, cells such as INS-1 (832/13), C2C12 myotubes, Caco-2, and HepG2 (Fig. 6A-E). Notably, all avocado polyols delivered in DMSO showed no activity in any of the non-AML cell lines even at high concentration ranges between 15-29 µg/mL. However, avocado polyols delivered in SEDDS at the supraphysiological concentration of 29 µg/mL (>50 µM) showed some reductions in viability in INS-1 (932/13), C2C12 myotubes and HepG2 cells suggesting enhanced delivery. The greater in vitro potency of avocado polyol SEDDS in AML cell lines demonstrated here will enable more effective dosing of these compounds in future in vitro and in vivo studies. After 72 hours, cell viability was measured by the MTS assay. For avocadyne (A), avocadene (B), avocatin B (C), and 3:1 avocadene-avocadyne (D), 20 mg/mL (2% w/w) SEDDS were freshly prepared and used for serial dilution. For 3:2 avocadene-avocadyne (E), 15 mg/mL (1.5% w/w) SEDDS were freshly prepared and used. Data represent logarithmic transformation of avocado polyol concentrations (µg/mL) and cell viability data that was fit to a nonlinear four-parameter logistic curve to determine inhibitory concentration 50 (IC50). All data represents mean ± SEM from three independent experiments performed in triplicate.  www.nature.com/scientificreports www.nature.com/scientificreports/ pilot in vivo pharmacokinetic study. The in vivo nutritional significance, bioavailability, and biodistribution of long chain fatty alcohols is not well studied 3,37 . The pharmacokinetics of a 100 mg/kg (body weight (b.w.)) oral bolus dose containing 20 mg/mL AVO SEDDS was examined in mice. Non-compartmental pharmacokinetic analysis showed that the maximum concentration (Cmax) of avocadyne and avocadene in whole blood was 1687.90 and 1544.83 ng/mL, respectively (Table 2). Despite similar maximal plasma concentrations (Fig. 7A), exposure to avocadene seemed greater than avocadyne due to its higher area under the curve (AUC 0-24h ), however more time points and are needed to decipher statistically significant differences in all pharmacokinetic parameters. Furthermore, avocadyne and avocadene were detectable in various metabolically active tissues 24 h from oral administration, accumulating most in the liver and pancreas (Fig. 7B). This pilot pharmacokinetic study also allowed for the development and validation of a bioanalytical method that has desirable selectivity, linearity, extraction recovery, accuracy and precision for the quantitation of AVO in mouse whole blood and tissue matrices (Supplementary Table 3).
Our previous work reports the positive effects of AVO in a mouse model of diet-induced obesity and utilized the AVO SEDDS described in this work to deliver an oral dose of 100 mg AVO/kg (body weight (b.w.)) twice weekly for five weeks to obese mice on a high fat diet 10 . In a prior unpublished study, we utilized the same AVO SEDDS to deliver 100 mg AVO/kg (b.w.) twice weekly for 13 weeks to mice on standard and high fat diet. In both studies, repeated oral dosing of AVO SEDDS for a prolonged period of time in lean and obese mice showed no adverse effects as assessed by onset of physical abnormalities, excessive weight loss and significant changes in endpoint complete blood count data. Although more tolerability data (changes in liver, muscle, kidney and heart enzymes) is required to completely evaluate the safety of AVO SEDDS, it is likely the low concentrations of excipients in the SEDDS are advantageous from a safety perspective.

Other applications of avocado polyol SEDDS. SEDDS are often commercialized in solid dosage forms
where the oil phase and drugs are adsorbed onto a solid carrier and used to produce tablets, capsules etc., for oral consumption 29 . The oil phase for both control and AVO SEDDS adsorbed easily onto the solid carrier neusilin 38 . These solid-SEDDS self-assembled into expected droplet size distributions upon dissolution in water ( Supplementary Fig. 12) suggesting the described SEDDS are scalable and suitable for common commercial  Table 2. Avocatin B non-compartmental pilot pharmacokinetic analysis. Data are shown as mean ± S.D., N = 3 in each group. Cmax denotes maximum concentration of bioactive in blood; Tmax denotes time at which Cmax occurs; Kel denotes elimination rate constant; t1/2 denotes half-life of bioactive in blood; AUC0_t denotes area-under-the-curve (AUC) from time 0-24 hr; AUC0_inf denotes AUC from time 0 to infinity. Different superscript letters within each row indicates significant differences (P < 0.05) between avocadene and avocadyne, two-tailed Student's t-test (n = 3). www.nature.com/scientificreports www.nature.com/scientificreports/ dosage forms.The ability of AVO to enhance encapsulation of poorly water soluble compounds such as naproxen and curcumin was also tested. In a previous study, naproxen (3 mg/mL) encapsulated in 1:1 NeoBeeM5-Tween 80 SEDDS had a mean droplet size of 240 nm 28 ; however, the method of SEDDS preparation was different than described herein. In this study, 5 mg/mL naproxen encapsulated in AVO-SEDDS was observed to have a mean droplet size of 32 nm compared to naproxen in control SEDDS which was 229 nm (Fig. 8A,B). Similarly, 5 mg/mL curcumin encapsulated in AVO-SEDDS showed a mean droplet size of 52 nm compared to 297 nm in control SEDDS (Fig. 8C,D). Both naproxen and curcumin control and AVO SEDDS destabilized over the course of 3 days; however, higher dilutions (above 1:100) slowed this effect to several weeks (data not shown). Finally, curcumin has an established in vitro bioactivity profile in AML cells 39 ; thus, we delivered curcumin to OCI-AML-2 cells in a DMSO or SEDDS formulation (Fig. 8E). Expectedly, curcumin in AVO-SEDDS had greater potency compared to curcumin/AVO in DMSO or curcumin in control SEDDS (Fig. 8E and Table 3).

Discussion
Avocado derived avocadene and avocadyne are prevalent in the cosmetic, agricultural and food industries. Their diverse medicinal properties warrant additional research that evaluates these natural compounds as novel therapies for human diseases. This study highlights strong links between the physical properties of avocado polyols extracted from avocado seeds to: i) their behaviour in SEDDS, ii) in vitro bioactivity, and iii) in vivo pharmacokinetics. The physical characterization of avocadene, avocadyne and their binary mixtures through DSC and XRD reveals valuable new data providing key insights for future large-scale formulation studies. The eutectic phase behaviour of AVO is advantageous for its successful incorporation into simple, low-energy, bio-compatible SEDDS. Eutectic mixtures are shown to improve drug dissolution/solubility and bioavailability [40][41][42] ; however, these systems employ structurally unrelated molecules (e.g. drug and excipients) to form eutectics. In contrast, avocadene and avocadyne are two structurally similar natural compounds that form a eutectic mixture at a molar ratio of 1:1 which translates to their efficient incorporation into a microemulsion-based delivery system. Mechanisms of eutectic formation are generally attributed to differences in i) molecular size and geometry, ii) polymorphic state, and iii) external fields (cooling and shear rates), which collectively leads to the formation of a more disordered crystal packing 25 . The XRD data presented here confirmed that molecular size and potentially polymorphic state contribute to the eutectic formation seen in AVO.
The incorporation of small amounts of AVO into SEDDS resulted in drastic reductions in droplet diameter without adding energy or a co-surfactant. This is strongly indicative of AVO self-assembling at the oil-water interface altering the interface curvature and film flexibility. The co-surfactant effects of avocado polyols in MCT oil-high HLB surfactant SEDDS correlated well with their physical properties (notably melting temperatures, enthalpies and entropy as well as crystal structure and thickness). Avocadene was demonstrated to be more compatible in SEDDS than avocadyne. The heat sensitive nature of avocadyne SEDDS was an unexpected finding which requires further evaluation to link its unique physical properties to extent of self-assembly. Furthermore, all avocado polyol SEDDS showed signs of destabilization at concentrations greater than or equal to 2% (w/w) suggesting that higher polyol concentrations may penetrate the interfacial film in such a way that pushes the spontaneous curvature of the surfactant film past the hydrophilic-lipophilic balance point 30,43 .
All avocado polyol SEDDS showed enhanced in vitro potency compared to DMSO delivery in AML cell lines, while maintaining selective toxicity. We hypothesize SEDDS enabled increased accumulation of polyols in molecular targets (predominantly mitochondria) compared to DMSO. However, further in vitro studies that test the solubility, release profile 44 , and cellular uptake 45 of avocado polyols in SEDDS versus conventional cell culture solvents are required to elucidate the mechanisms by which SEDDS increase in vitro potency. Since the absorption and cellular uptake of long chain fatty alcohols in biological systems is not well understood 37,46 , the pharmacokinetic data presented in this work provides impetus for further research on avocado polyol cellular uptake and absorption. While promising pharmacokinetics and biodistribution was obtained, a full-scale pharmacokinetic study (a minimum of 6-8 time points) is required to deduce absolute bioavailability of AVO delivered in SEDDS (ratio of AUCs obtained from oral and intravenous bolus doses of AVO SEDDS) and generate kinetic biodistribution data in all key tissues including lungs, skeletal muscle and spleen.
Encapsulation of curcumin in AVO-SEDDS highlights the novel use of AVO as both a bioactive and co-surfactant that improves bioactivity and encapsulation of bioactives. However, further experiments are required to determine if AVO has similar effects when curcumin and other anti-cancer drugs are encapsulated in emulsion-based delivery systems different to the one described here.
Lastly, all experimental data in the present study does not account for the potential effects of stereoisomers of avocadene and avocadyne 47 . Avocado polyols utilized in this study were extracted from avocado seed using column chromatography that cannot resolve all four possible stereoisomers of avocadene from those of avocadyne. The chemical synthesis of all 4 stereoisomers of avocadene and avocadyne has recently been reported 48 which will allow future work to determine i) which stereoisomers are naturally occurring in the avocado, and ii) how different stereoisomers of avocadene and avocadyne behave in physical and biological systems described in this work.

Methods
Avocado polyol extraction, purification and analytical characterization. AVO

Preparation and characterization of emulsions.
Formulations tested for SEDDS characteristics consisted of either long or medium chain triglycerides and surfactants/co-surfactants with varying HLB values, at various oil-to-surfactant weight ratios (9:1, 5:1, 1:1) (see Supplementary Table 2 for a detailed list of excipients tested). Each formulation was prepared in 2 ml polypropylene centrifuge tubes, vortexed and heated at 75 °C for 5 minutes. The heated oil was then diluted 10-fold with the immediate addition of water phase (PBS) and vortexed for 30 sec. All formulations were visually analyzed for macroscopic appearance (characterized as clear, opalescent or milky appearance). The mean droplet size (Z-average in nm), polydispersity index (PDI), and zeta potential for each formulation was determined using dynamic light scattering (Zetasizer Nano ZS, Malvern Instruments, Malvern, U.K.). A refractive index of 1.33 was used for the aqueous phase, and refractive indices for each oil/surfactant mix was determined using a table top refractometer (Zeiss Abbe, NY, USA). Droplet size was measured on three independently prepared formulations and averaged from three readings per formulation, whereas zeta potential was analyzed on two independent formulations, with each measurement averaged from four readings per formulation. All measurements were carried out at 25 °C except for avocadyne SEDDs where the zetasizer compartment temperature was set to 37 °C to prevent destabilization.
Formulations chosen for further studies were selected based on their self-emulsification properties, small droplet size (below 300 nm), low PDI (below 0.4), and no requirement for co-surfactants. Formulations of avocado polyol-SEDDS contained 1-20 mg powder of each polyol added to 100 µL of a 1:1 mixture of oil and surfactant which was heated to 75 °C for 2 hr in an incubator (MaxQ 4450, Barnstead/Lab-Line, IL, USA). After dissolution of polyols in the oil phase, 900 µL water phase (PBS) was added to the oil phase and vortexed for 30 s. Visual appearance of fresh polyol emulsions were noted followed by droplet size, PDI and zeta potential measurements on day zero (immediate measurement on fresh emulsion), day 3 and weekly for 4 weeks. Poorly water-soluble drugs naproxen and curcumin were also encapsulated in SEDDS. Briefly, naproxen or curcumin were added to oil phase (1:1 NeoBeeM5-Tween 80) at a concentration of 50 mg/mL with or without 100 mg/mL AVO and heated overnight at 75 °C and then the oil phase was diluted 10 folds with PBS and vortexed for 30 s. Droplet size and PDI were then measured as described above.
Cryo-transmission electron microscopy. Transmission electron microscopy (TEM) was conducted at the Advanced Imaging Center in the University of Guelph. Approximately 5 μl of emulsion was placed on a carbon grid with a perforated carbon film (Canemco & Marivac, Quebec, Canada). Filter paper was used to blot the excess liquid. The grid was then immediately dipped into liquid ethane and transferred to a cryo-holder for direct observation at −176 °C on a FEI Tecnai G2 F20 energy-filtered Cryo-TEM (FEI Corp., Hillsboro, Oregon, USA) operated at 200 kV in low dose mode, equipped with a Gatan 4k CCD camera (Gatan Inc., Pleasanton, California, USA).
Optical and polarized light microscopy. Optical and polarized light microscopy was performed using an optical microscope model BX60 (Olympus Optical Co., Tokyo, Japan). Images were captured (using 20× objective lens) with a model DP71 digital camera (Olympus Optical Co., Tokyo, Japan) using the cellSens (version 1.0) software. 20-50 µL of emulsion samples were placed on a microscope slide and micrographs were obtained using either brightfield or polarized light. Polarized light microscopy was performed on emulsions to confirm absence of crystalline phases (i.e., confirm optical isotropy).
Thermodynamic stability of SEDDs. Emulsion stability was assessed by centrifugation at 21.1 x g for 15 min and phase separation, creaming or flocculation was observed. The effect of temperature on SEDDS was also evaluated using freeze-thaw tests where formulations were subject to 6 freezing-heating cycles (−20 °C for 24 hr followed by 37 °C for 24 hr) after which visual observations and droplet size measurements were performed. DMSO (µg/mL) SEDDS (µg/mL) Curcumin 8.55 ± 0.10 a 5.83 ± 0.19 b Curcumin + AVO 3.75 ± 0.18 a 1.08 ± 0.13 b Table 3. IC50 values of curcumin delivered via DMSO or in SEDDS to OCI-AML-2 cells. IC50 was calculated using non-linear regression model (logarithmic inhibitor vs. normalized response-variable slope). Different superscript letters within each row indicate significant differences (P < 0.0001) between treatments, two-way ANOVA with Sidak's post hoc test (n = 3). (2020) 10:5566 | https://doi.org/10.1038/s41598-020-62334-y www.nature.com/scientificreports www.nature.com/scientificreports/ cell culture and in vitro cytotoxicity. Cell culture and in vitro cytotoxicity assay methods are described in section 1 of Supplementary Information.
In vivo pilot pharmacokinetic study. In vivo study design, procedures and pharmacokinetic data analysis is described in section 1 of Supplementary Information. All animal studies were carried out according to the regulations of the Canadian Council on Animal Care and with the approval of the University of Guelph, Animal Care Committee.

Preparation of solid-SEDDS. Solid-SEDDS preparation is described in section 1 of Supplementary
Information. Statistical data analysis. Unless otherwise stated, DSC results are presented as mean ± S.D., droplet size and PDI measurements for SEDDS are presented as mean ± SEM, in vitro results are presented as mean ± SEM, and in vivo results are presented as mean ± SD. Data were analyzed with GraphPad Prism 6.0 (GraphPad Software, USA) using one or two-way ANOVA with Tukey's or Dunnett's post hoc analysis for between group comparisons. Standard student's t-tests were also used where appropriate. P < 0.05 was accepted as being statistically significant where *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.