Preparation and Characterisation of Nobiletin-Loaded Nanostructured Lipid Carriers

The objective of this manuscript was to investigate and optimise the potential of nanostructured lipid carriers (NLCs) as a carrier system for nobiletin (NOB), which was prepared by high-pressure homogenisationmethod. Additionally, this study was focused on the application of NOB-loaded NLC (NOB-NLC) in functional food. Response surface method with a three-level Box–Behnken design was validated through analysis of variance, and the robustness of the design was confirmed through the correspondence between the values measured in the experiments and the predicted ones. Properties of the prepared NOB-NLC, such as Zaverage, polydispersity, entrapment efficiency, zeta potential, morphology, and crystallinity, were investigated. NOB-NLC exhibited a spherical shape with a diameter of 112.27 ± 5.33 nm, zeta potential of −35.1 ± 2.94mV, a polydispersity index of 0.251 ± 0.058, and an EE of 81.06% ± 6.02%. Results fromX-ray diffraction and differential scanning calorimetry ofNOB-NLC reviewed that theNOB crystal might be converted to an amorphous state. Fourier transform infrared spectroscopic analysis demonstrated that chemical interaction was absent between the compound and lipid mixture in NOB-NLC.


Introduction
Nobiletin (5,6,7,8,3  ,4  -hexametoxyflavone, NOB) is a citrus polymethoxylated flavone that possesses an extensive range of health benefits among others, including antioxidant activity, glucose regulation, and fat metabolism [1].It was previously reported that nobiletin could inhibit tumor angiogenesis by regulating Src, FAK, and STAT3 signalling through PXN in ER+ breast cancer cells [2] and also showed its ameliorative effects toward ischemia-reperfusion injury by suppressing the function of Kupffer cells after liver transplantation in rats [3].Furthermore, nobiletin has been found to have positive effects of nobiletin on propofol-mediated antiinflammatory as well as having neuroprotective effect [4].It may represent beneficial drug candidates for the treatment and prevention of Alzheimer's and Parkinson's disease because of its naturally occurring phytochemicals [5].Similar to other flavonoids, NOB exists widely in nature and is found mainly in various citrus peels.Therefore, a citrus processing waste can be potentially converted into a valuable functional ingredient.However, NOB exhibits poor water solubility because all hydroxyl groups in the structure are methoxylated, leading to low oral bioavailability, thus limiting the practical uses of NOB [6].
Delivery systems using lipid-based bioactive components as promising carriers are getting increasing attention nowadays because of their potential to increase solubility and improve bioavailability of water-insoluble and/or lipophilic bioactive components [7].NLCs are second-generation lipid nanoparticles distinguished from SLN as described by Müller et al. [8,9].NLCs possess the advantages related to SLNs, such as feasibility for encapsulation of hydrophobic and hydrophilic compounds and the capability to impart controlled and sustained release characteristics over free-state compound.Furthermore, NLCs have the ability to protect compounds from chemical, physical, oxidative, or light degradation; are of extremely low toxicity; and are also of low cost [10,11].NLCs are derived from oil-water emulsions and SLNs by preparing a liquid lipid through a matrix of a solid lipid.Compared with SLNs, this matrix has a less-ordered crystalline nanostructure that imparts higher loading capacity for the compound, overcomes the drawback of the compound expulsion from the lipid core, and now is popularly studied and used in pharmaceutical, cosmetic, and food products [12,13].In spite of this, there has been no report about the application of NLC on delivery of NOB.Quite a few examinations have been undertaken to improve the healthful potential of NOB for oral administration by choosing appropriate delivery system, including chitosan-based microemulsions loaded with NOB for intravenous injection [14], self-nanoemulsifying delivery systems [15], NOB-loaded chitosan nanoparticles [16], self-assembling NOB proliposomes [17], and NOB-loaded solid lipid nanoparticles (SLNs) [18].According to recent researches, nanoemulsion-filled hydrogel could achieve the sustained release and absorption of nobiletin and prevent its precipitation in the gastrointestinal tract [19].Taken together, these previous findings provided the foundation and evidence for further research on NOB-loaded NLC.
The objectives of this research were as follows: firstly, to study the suitability and the effectiveness of NLCs as a new type carrier system to encapsulate NOB and, secondly, to investigate the characteristics of the prepared NLC.To perform and optimise the formulations, a Box-Behnken design (BBD) of three factors and three levels was used, and highpressure homogenisation technology was chosen by considering industrial needs [12,20].Furthermore, the following characteristics of NLC system were fully examined: particle size, polydispersity index (PDI), zeta potential, entrapment efficiency (EE), morphology and crystallinity through dynamic light scattering (DLS), high-performance liquid chromatography method (HPLC), transition electron microscopy (TEM), differential scanning calorimetry (DSC), Fourier transform infrared (FT-IR), and X-ray diffraction (XRD).

Preparation of NOB-NLC.
NOB-NLC was prepared through melt-emulsification technique and high-pressure homogenisation [21].Briefly, 3 mL ethanol was added as cosolvent to the lipid phase, which was evenly composed of NOB, solid lipid, liquid lipid, and lecithin.A total of 100 mL each of aqueous phase, including surfactant, and lipid phase was heated to 75 ± 3 ∘ C in a water bath.Subsequently, the lipid matrix was injected into the aqueous matrix, and the mixture phase was stirred at 500 rpm for 10 min and then mixed by a Lab high-shear homogeniser SRH-S300 (Siehe Industry, Shanghai, China) at 10000 rpm for 2 min at the same temperature.Afterwards, the coarse emulsion was homogenised using a nanohomogeniser machine AH100D (ATS Engineering Inc., Vancouver, Canada) under 800 bar for 10 cycles.Finally, the resulting nanoemulsion was allowed to cool down to room temperature to form NOB-NLC.

Box-Behnken Design for Optimisation of NOB-NLC.
A response surface method (RSM) based on BBD was employed, and a 17-run design with three factors and three levels, which included 12 factorial points and 5 axial points, was carried out in the present work [22].Meanwhile, Design Expert5 8.0.5 software (Stat-Ease Inc., Minneapolis, USA) was applied to determine the response patterns and then to establish models, which are shown below: ( As shown in (1), the measured responses () were mean particle size ( 1 ) and EE ( 2 ).The three independent variables selected were emulsifier-to-lipid ratio ( 1 ), lecithin-to-Poloxamer 188 ratio ( 2 ), and liquid lipid-to-solid lipid ratio ( 3 ) [23].For each factor, high, medium, and low levels were represented by codes 1, 0, and −1, respectively (Table 1). 0 is the intercept;  1 ,  2 , and  3 are the linear coefficients;  12 ,  13 , and  23 are the interaction coefficients; and  11 ,  22 , and  33 are the quadratic coefficients.
2.4.-Average, PDI, and Zeta Potential Characterisation.average, PDI, and zeta potential of NOB-NLC were analysed through DLS using a Zeta Analyser ZEN-3600 (Malvern Instruments, Worcester, UK).Sample measurements were taken in triplicate at 25 ∘ C with a fixed scattering angle of 173 ∘ .The collected data were then analysed through the manufacturer-supplied software from Malvern Instruments [24].

Entrapment Efficiency and High-Performance Liquid
Chromatography Measurement.The EE of NOB-NLC was determined by measuring free NOB in the aqueous phase, which was separated using ultrafiltration centrifugation [21,25].A total of 500 L NOB-NLC was added to the upper chamber of Centrifugal Filter Unit 100 kDa (Millipore Co., Boston, USA) and then centrifuged at 3000 rpm for 20 min at 4 ∘ C through UniCen MR (Herolab, Germany) [26].The filtrate was dissolved in methanol and analysed through HPLC using Dionex UltiMate 3000 (Sunnyvale, USA) with C18 column.Chromatographic analysis was performed at 30 ∘ C and 333 nm using ultrapure water and acetonitrile (55 : 45 v/v) as the mobile phase, with injection volume of 20 L and a flow rate of 1.0 mL/min.Acetonitrile and ultrapure water were previously eluted through a 0.22 m membrane filter.The EE was calculated through the following equation [27]: where  total is the mass of initial NOB used for the assay and  free is the mass of free NOB in the aqueous phase.

Differential Scanning Calorimetry Measurement.
The thermograms of freeze-dried NOB-NLC, blank NLC, and NOB were analysed using DSC Q20 (TA Instruments, New Castle, USA). 3 mg of the samples was placed into standard aluminum pans, and the temperature was then increased from 25 ∘ C to 240 ∘ C at a heating or cooling scanning rate of 10 ∘ C min −1 .During thermal scans, dry nitrogen was used as effluent gas and an empty pan as a reference for the analysis [25,26,28].

Formulation Optimisation of NOB-NLC.
Based on the BBD design, 17 different formulations were carried out to determine the most critical independent variables, and the responses are summarised in Table 2.The final parameters that most intensely affected the properties of NOB-NLC were identified to be emulsifier-to-lipid ratio ( 1 ), lecithin-to-Poloxamer 188 ratio ( 2 ), and liquid lipid-to-solid lipid ratio ( 3 ).The polynomial equations were statistically validated through analysis of variance, and a summary of the results of regression analysis of responses is demonstrated in Table 3.The 3D graphs of the response surface plotted to represent the effects of the independent factors on measured responses of mean particle size and EE are shown in Figures 1 and 2.

Effect on Mean
Particle Size.The mean particle size ( 1 ) varied from 95.7 nm (Formulation ( 13)) to 186.325 nm (Formulation (2)), whereas the average was 139.96 nm as shown in Table 2.The effect of independent variables on mean particle size has been presented through the quadratic equation as follows: ( Equation ( 3) revealed negative effects on the variables  1 ,  2 , and  3 on  1 , as indicated by the negative value before the variable in the quadratic equation.Higher emulsifier-tolipid ratio may promote formation of smaller particles and provide higher stabilisation of nanosystem because of the reduction in interfacial tension between the lipid matrix and the hydrophilic phase [31].The results showed that as liquid lipid-to-solid lipid ratio increased, particle size decreased from 175.39 nm (Formulation (6)) to 96.93 nm (Formulation (8)), which can be attributed to the increased amount of solid lipids that tend to facilitate the aggregation of particles.
The 3D surface plots for the effect of interaction between  1 - 2 ,  1 - 3 , and  2 - 3 on particle size are shown in Figures 1(a)-1(c).As shown in Table 3, all independent factors have significant effect on the mean particle size of NOB-NLC ( < 0.05), particularly the liquid lipid-to-solid lipid ratio ( 3 and  3 2 ).Furthermore, the interaction term of emulsifier-to-lipid ratio and lecithin-to-Poloxamer 188 ratio surfactant ( 1  2 ) is also significant ( < 0.05), which is consistent with the analysis 3D plot of response surface (Figure 1(a)).In addition, the correlation coefficient ( 2 ) was 0.9427, indicating that merely 5.73% of total variables could not be illustrated by the model.The coefficient of variation was at 8.9%, and the low value suggested that the observed values were consistent with the predicted ones.The signal-to-noise ratio of the model was evaluated by "Adeq Precision" [32], with the value of 9.208 (>4), which indicated the sufficiency of model with mean particle size [33].

Effect on Entrapment Efficiency.
In the case of EE, the value ranged from 59.35% (Formulation ( 8)) to 88.72%   (Formulation ( 17)) with the selected levels of variables, whereas the average was found to be 76.30%.The following quadratic equation showed the effect of various independent factors on EE: The positive effect of variables  1 ,  2 , and  3 on EE could be seen in ( 4) as the positive value before these variables, as shown in the quadratic equation.Higher emulsifier-tolipid ratio could induce an increase in EE; the reason might be the availability of adequate emulsifier, which made the compound stay within the lipid particles and/or on the surface of the particles.Therefore, the liquid lipid-to-solid lipid ratio directly affects EE, which can be ascribed to the availability of higher amount of liquid lipid for compound entrapment.This can also be caused by the amphipathy of lecithin.Moreover, the interaction terms  1 - 2 and  2 - 3 exhibited a negative effect on particle size, whereas  1 - 3 revealed positive effects on EE of NOB-NLC [34].
The 3D response surface graphs, which depict the interaction effects of  1 - 2 ,  1 - 3 , and  2 - 3 on EE, are presented in Figures 2(a)-2(c).The model  value less than 0.05 and the -value of 10.65 indicated the significance of the model for EE.Additionally,  values of linear coefficient of  3 , interaction term coefficient of  1  2 , and quadratic term coefficients of  1 2 and  3 2 were all above 0.05, implying that they were significant on EE of NOB-NLC.Coefficient of correlation ( 2 = 0.932) and the coefficient of variation were at a low value (5.28%), which suggested a good relevance between predicted and experimental values.The model could be employed to navigate the design space demonstrating that the value of "Adeq Precision" is 9.341 [33].

Optimisation and Validation.
The optimisation of NOB-NLC preparation was chosen based on the prerequisite of fulfilling the minimum mean particle size and maximising the value of EE by applying the optimum formulation point of numerical method generated by Design Expert software.The optimised composition was selected as the emulsifierto-lipid ratio of 1.76, lecithin-to-Poloxamer 188 ratio of 1.35, and liquid lipid-to-solid lipid ratio of 0.95, which were suggested to attain the prerequisites of the optimised NOB-NLC formulation.
These observed responses of mean particle size (112.27± 5.33 nm) and EE (81.06 ± 6.02%), presented by the optimised NOB-NLC formulation, and the predicted value of mean particle size (103.18nm) and EE (84.28%), generated by the Design Expert software, were found to be in good agreement.This finding demonstrates that the optimised formulation was reliable in predicting the response of the NOB-NLC system.

Surface Characterisation.
The optimised NOB-NLC was homogeneous and slightly bluish, as shown in Figure 3(a).The particles exhibited a size of 112.27 ± 5.33 nm, with a PDI of 0.251 ± 0.058, an EE of 81.06 ± 6.02%, and zeta potential of −35.1 ± 2.94 mV.Studies of receiver biases suggest that the open endothelial gaps were in a range from 100 nm to 1000 nm [35], and particles of small size could trigger a long blood circulation as a free compound.The nanoparticles with a size of 100-300 nm have been found suitable to distribute in a tumor and easily transport through blood brain barriers [36].PDI values below 0.3 indicated a narrow size distribution, excellent dispersity, and high homogeneity of nanoparticles [37].The zeta potential played an important role in evaluating the stability of a nanoparticle system; high zeta potential can prevent particles from aggregating through electric repulsion.Currently, if the zeta potential is below −30 mV or above +30 mV, the systems will be considered stable.Furthermore, the powerful statistical obstruction and beneficial static repulsion will contribute to additional stability [38,39].
TEM images of NOB-NLC (Figure 3(b)) revealed that the optimised nanoparticles were well-identified as spherical, uniform, and nonadherent and were approximately 120 nm in diameter with a dense core (Figure 3(c)), which is accordant with the data estimated through DLS method [40].Moreover, the particle size measured through DLS, called hydrodynamic size, was found to be larger than that of the TEM results, called actual size.These differences are hypothesised to be attributed to different size measurement methods, as the solvent layer attached to the particle is measured in DLS [36].Different morphologies have been reported in much of the research in NLC; nevertheless, spherical morphology that has been discovered holds a greater proportion [12].Previous study indicated that the differences of compositions of lipid and surfactants or photographic equipment employed may result in the distinction in morphology of NLC [41].

Solid-State Characterisation of NOB-NLC Systems: DSC
Data. Figure 6 shows the modulated DSC scans of the freezedried NOB-NLC, blank NLC, and NOB.A sharp endothermic peak was observed in the crystalline NOB at 137.03 ∘ C, which was in concert with its melting point (Figure 4(c)).The melting of blank NLC peaked at 48.73 ∘ C with a small shape (Figure 4(a)), and a lower melting peak of NOB-NLC appeared at 50.58 ∘ C in the thermogram (Figure 4(b)).In the NOB-NLC formulations, the NOB melting could not be detected, which could be attributed to the existence of NOB in the amorphous state or in a molecularly dispersed state within the lipid matrix.Even if a small fraction of NOB is present in NOB-NLC as undissolved material in the crystalline state, the relatively low compound loading of NOB in NOB-NLC would render NOB-NLC improbable to detect the melting event of any such fraction using a DSC-based technique [42].This phenomenon can also be ascribed to the interference of NOB and the surfactants in NOB-NLC crystallisation [43].Furthermore, according to Gibbs-Thomson equation, particle size also has a conspicuous effect on the melting endotherm of lipid particles in nanoscale range [44].Different lipid nanoparticles melted at different temperatures and led to peak broadening and a change in melting transition to lower temperature on the basis of this equation, as was different from that of mass lipid [36].NOB-NLC, which was further confirmed using DSC.The slight shift and decrease of absorption bands of NOB in NOB-NLC can be ascribed to a transformation in molecular environment and intermolecular interactions accompanied by dispersed compound molecule in lipid matrix [36].
3.6.Solid-State Characterisation of NOB-NLC Systems: XRD Data.XRD experiments were performed to investigate the changes of the microstructure in the lipid crystallisation process (Figure 6).The XRD spectra of NOB in Figure 6(c) showed distinct and intense peaks at 2 scale, indicating the high crystalline structure of NOB (i.e., three sharp peaks at 2 = 7.68 ∘ , 11.47 ∘ , and 11.86 ∘ ; five medium-intensity peaks at 2 = 8.99 ∘ , 10.19 ∘ , 12.46 ∘ , 15.28 ∘ , and 18.84 ∘ ; and several lowintensity peaks at 2 = 5.97 ∘ , 9.64 ∘ , and 13.42 ∘ ).Characteristic peaks of NOB were absent from the XRD pattern of NOB-NLC (Figure 6(b)), indicating that the NOB molecule lost its crystalline nature and formed an amorphous complex with the NLC matrix.This discovery corresponds to the result from the DSC analysis, providing evidence of the phase transition of NOB from crystalline to amorphous during processing [45].

Conclusion
NOB-loaded NLCs were successfully developed through a high-pressure homogenisation technique and optimised by seventeen-run, three-factor, three-level BBD.The lecithin-to-Poloxamer 188 ratio and the liquid lipid-to-solid lipid ratio were crucial determinants for NOB-NLC production as these factors influenced the physical properties of NOB-NLC.The optimised NOB-NLC manifested a size of 112.27 ± 5.33 nm with a polydispersity of 0.251 ± 0.058, an EE of 81.06 ± 6.02%, and zeta potential of −35.1 ± 2.94 mV.Particle size and TEM study confirmed nanosized discrete spherical globules with a smooth surface area.DSC and XRD confirmed the transformation of the crystalline nature of NOB into an amorphous nature, which played an important role in the enhancement of absorption rate, followed by NOB bioavailability.FT-IR analysis confirmed the efficient entrapment of NOB in NOB-NLCs.Thus, this present research demonstrated that the optimised formulation has potential in delivering NOB and other lipophilic compounds.Future research should focus on the appraisal of the practical potential of the optimum formulation in food-relevant models and actual food systems during production and storage.

Table 1 :
Variables with respective coded levels of the Box-Behnken design for the preparation and optimization of NOB-NLC.

Table 2 :
Observed response in Box-Behnken design for preparation and optimization of NOB-NLC formulations with predicted values generated by Design Expert software.

Table 3 :
Statistical analysis results of lack of fit test and model summary for mean particle size and entrapment efficiency.