Optimization and characterization of poly(lactic-co-glycolic acid) nanoparticles loaded with astaxanthin and evaluation of anti-photodamage effect in vitro

Astaxanthin is a xanthophyll carotenoid with high beneficial biological activities, such as antioxidant function and scavenging oxygen free radicals, but its application is limited because of poor water solubility and low bioavailability. Here, we prepared and optimized poly(lactic-co-glycolic acid) (PLGA) nanoparticles loaded with astaxanthin using the emulsion solvent evaporation technique and investigated the anti-photodamage effect in HaCaT cells. The four-factor three-stage Box–Behnken design was used to optimize the nanoparticle formulation. The experimental determination of the optimal nanoparticle size was 154.4 ± 0.35 nm, the zeta potential was 22.07 ± 0.93 mV, encapsulation efficiency was 96.42 ± 0.73% and drug loading capacity was 7.19 ± 0.12%. The physico-chemical properties of the optimized nanoparticles were characterized by dynamic light scattering, scanning electron microscopy, transmission electron microscopy, Fourier-transform infrared spectroscopy, X-ray diffraction, differential scanning calorimetry and thermo-gravimetric analyser. In vitro study exhibited the excellent cell viability and cellular uptake of optimized nanoparticles on HaCaT cells. The anti-photodamage studies (cytotoxicity assay, reactive oxygen species content and JC-1 assessment) demonstrated that the optimized nanoparticles were more effective and safer than pure astaxanthin in HaCaT cells. These results suggest that our PLGA-coated astaxanthin nanoparticles synthesis method was highly feasible and can be used in cosmetics or the treatment of skin diseases.

FH, 0000-0002-2048-3981; LY, 0000-0002-4640-1016 Astaxanthin is a xanthophyll carotenoid with high beneficial biological activities, such as antioxidant function and scavenging oxygen free radicals, but its application is limited because of poor water solubility and low bioavailability. Here, we prepared and optimized poly(lactic-co-glycolic acid) (PLGA) nanoparticles loaded with astaxanthin using the emulsion solvent evaporation technique and investigated the anti-photodamage effect in HaCaT cells. The four-factor threestage Box-Behnken design was used to optimize the nanoparticle formulation. The experimental determination of the optimal nanoparticle size was 154.4 ± 0.35 nm, the zeta potential was 22.07 ± 0.93 mV, encapsulation efficiency was 96.42 ± 0.73% and drug loading capacity was 7.19 ± 0.12%. The physico-chemical properties of the optimized nanoparticles were characterized by dynamic light scattering, scanning electron microscopy, transmission electron microscopy, Fouriertransform infrared spectroscopy, X-ray diffraction, differential scanning calorimetry and thermo-gravimetric analyser. In vitro study exhibited the excellent cell viability and cellular uptake of optimized nanoparticles on HaCaT cells. The antiphotodamage studies (cytotoxicity assay, reactive oxygen species content and JC-1 assessment) demonstrated that the optimized nanoparticles were more effective and safer than pure astaxanthin in HaCaT cells. These results suggest that our properties of the optimized AST-PLGA NP were characterized, and the cellular uptake effect of nanoparticles was assessed using HaCaT cells. Furthermore, the in vitro photodamage protective effect of optimized AST-PLGA NP and astaxanthin were evaluated.

Preparation of poly(lactic-co-glycolic acid) loaded with astaxanthin nanoparticles
The nanoparticles were prepared following the emulsion solvent evaporation technique [27]. Briefly, an accurately weighed amount of PLGA and astaxanthin were dissolved in a 1 ml mixture of dichloromethane and acetone (dichloromethane: acetone = 3 : 2) under probe sonication. This organic phase was slowly added to bovine serum albumin (BSA) 1% (w/v) solution and sonicated together. The final emulsion was maintained under magnetic stirring at room temperature overnight. After centrifuging at 14 000 rpm for 40 min, the PLGA loaded with astaxanthin nanoparticles were washed three times with deionized water. The unloaded nanoparticles (PLGA NP) were obtained following the same procedure without astaxanthin. The optimized preparation method through response surface experiment design is as follows.

Response surface methodology experiment design
In order to explore optimal formulation condition, we employ the RSM to evaluate the relationship between factors and responses using a minimum number of experiments and use BBD to analyse the results [28]. In this study, the experiment was employed with four factors, three levels and 29 runs for the optimization study using DESIGN-EXPERT 8.0.6 software. The concentration of PLGA (X A ), the concentration of astaxanthin (X B ), water volume (X C ) and sonication time (X D ) were selected as independent variables and they were set at low, medium and high levels, respectively (table 1). Then, different batches were prepared with different independent variables at different levels, and responses like size, drug loading (DL) capacity and encapsulation efficiency (EE) were obtained. The data were analysed using the DESIGN-EXPERT software, and polynomial model equations were given as follows: where Y is the individual response factor or dependent variable; β0-β11 are regression coefficients; and X A , X B , X C and X D are the independent variables.

Evaluation of the encapsulation efficiency and drug loading capacity
The EE and DL of AST-PLGA NP were determined by a UV-visible spectrophotometer (UV-2600, Shimadzu, Japan). One milligram lyophilized AST-PLGA NP was accurately weighed, and astaxanthin in AST-PLGA NP was extracted by adding 1 ml dichloromethane: ethanol (1 : 10, v : v). The above extract was vortexed and centrifuged at 10 000 rpm for 5 min and the supernatant collected, which was measured at the wavenumber of 478 nm. The EE and DL were calculated using the following equation:

Morphologic analysis of nanoparticles
The shape and size of the nanoparticles were observed using a scanning electron microscopic (SEM) (Merlin, Carl Zeiss AG, Germany) at an acceleration voltage of 5 kV. A drop of 1 mg ml −1 sample was placed on a wafer, dried overnight and then coated with platinum for the experiment. The surface morphology of nanoparticles was further characterized by transmission electron microscopy (TEM).

Fourier-transform infrared spectroscopy
The Fourier-transform infrared (FTIR) spectra of astaxanthin, free PLGA nanoparticles and AST-PLGA NPs were recorded by a Thermo Scientific Nicolet Nicolet Nexus FTIR spectrometer (Thermo-Nicolet, Gaisburg, MD, USA) in the ranges of 400-4000 cm −1 for chemical analysis of functional groups. Samples were ground together with KBr (1 : 50, w/w).

X-ray diffraction patterns
A certain amount of astaxanthin, free PLGA NP and AST-PLGA NP were scanned on a powder X-ray diffractometer (XTV160H, X-TEK, U.K) with an angle of 2θ from 5 to 60°under ambient temperature.

Thermal analysis
The thermal properties of astaxanthin, PLGA NP and AST-PLGA NP were measured using a differential scanning calorimetry (DSC) and thermo-gravimetric analyser (TGA twice with phosphate buffered saline (PBS) and harvested using trypsin-EDTA (0.25% v/v). The results were recorded using a BD C6 flow cytometry (FACS) with a 488 nm laser. For each experiment, a total of 10 000 cells were gated per sample and each sample was performed in triplicate. Confocal laser scanning microscopy: laser confocal microscopy (Ti-E A1, Nikon, Japan) was used to qualitatively analyse the localization of fluorescent nanoparticles in HaCaT cells. The cells were seeded at 5 × 10 4 cells well −1 in 24-well plates containing glass coverslips. After incubation with fluorescent nanoparticles for a certain period of time, the cells were washed gently with PBS three times and fixed with 4% paraformaldehyde for 20 min. Finally, cells were stained with 4,6-diamidino-2 phenylindole (DAPI) for 5 min.

Ultraviolet B irradiation and phototoxicity measurement
Cells were irradiated with UVB from two fluorescent Philips lamps (280-370 nm) with a peak at 312 nm and measured by a UVB radiometer with a sensor. Cells were pretreated with or without astaxanthin, and AST-PLGA NP for 12 h, and exposed to the dose of 30 mJ cm −2 irradiation. Cells were rinsed by PBS once and irradiated under a thin layer of PBS. After UVB radiation, cells were incubated with the drug or free media for another 24 h immediately. The control group was treated to the same procedures without exposing to UVB lamps. The protection of phototoxicity of astaxanthin and AST-PLGA NP was assessed by the MTT method.

Assessment of intracellular reactive oxygen species
The fluorescence dye 2 0 ,7 0 -dichlorodihydrofluorescein diacetate (DCFH-DA) (Sigma-Aldrich, USA) was used to investigate intracellular ROS. HaCaT cells were seeded at 1.5 × 10 5 cells well −1 in 12-well plates overnight. The above irradiation method was carried on until after the cells reached 80% confluence. Finally, the cells were incubated with 10 µM DCFH-DA in serum-free medium for 30 min, and the results were analysed by FACS and confocal laser scanning microscopy.

Measurement of mitochondrial membrane potential
Mitochondrial membrane potential (ΔΨm) was measured by a mitochondrial membrane potential assay kit with JC-1 (C2006, Beyotime Biotechnology, Jiangsu, China) according to the manufacturer's kit instructions. In brief, after the irradiation treatment mentioned above, cells were washed with PBS and treated with JC-1 at 37°C for 20 min in dark conditions. Then the results of the fluorescence intensity were recorded by FACS and confocal laser scanning microscopy. The value of mitochondrial membrane potential was calculated using the following equation:

Statistical analysis
The experiments were repeated three times separately and the results are presented as the means ± s.e.m. The statistical significance of the results was evaluated using an independent t-test for the comparison of two samples. A p-value of less than 0.05 was considered statistically significant. 3. Results and discussion 3.1. Formulation and optimization of astaxanthin-loaded poly(lactic-co-glycolic acid) nanoparticles by Box-Behnken design BBD can study the effects and interactions between all factors and responses with the least number of experiments [29]. In our study, the four-factor, three-level experiment design was analysed with three responses to explore the best formulation. The higher and lower limits were chosen in pilot studies. The results of the whole design contained 29 experiment runs with five central points, displayed in

. Effect of formulation variable on encapsulation efficiency
We used the quadratic model to analyse the variance of the BBD experiment. ANOVA results for EE response are shown in table 3. The full regression model was significant ( p = 0.0004) with insignificant lack of fit (F = 0.67; p = 0.72), which meant that it was excellent fitting to the corresponding response and a quadratic second-order polynomial model was fitted with the equation as follows: From the equation, a positive value indicated a synergistic effect on optimization in the regression model, whereas a negative value indicated an antagonistic effect [30]. Factors A, AB, AD, BC, A 2 , C 2 and D 2 had a positive effect on the EE. However, factors B, C, D, AC, BD, CD and B 2 showed passive influence on EE. In table 3, the R 2 values and the adjusted R 2 values were 0.8772 and 0.7545, respectively, and the value of the coefficient of variation (CV) was 3.45%, which suggested that the fitted model was in reasonable agreement.

Effect of formulation variable on drug loading
High DL capacity in nanoparticles can decrease drug dosing frequency and side effects [31]. ANOVA results of DL are shown in table 4. The fitted model was extremely significant (F = 61.47 and p < 0.0001). The lack of fit was not significant (0.6) and a quadratic second-order polynomial model was fitted with the equation as follows:

Effect of formulation variable on size
It is well known that the size and PDI of nanoparticles are very important features, which may influence toxicity, stability and biodistribution of drug delivery [32]. The size of nanoparticles was seen as a response, ANOVA results are shown in table 5. The fitted model was highly significant ( p = 0.0005 and F = 6.77). Meanwhile, the lack of fit was not significant which means it is a good fit to the corresponding response, and a quadratic second-order polynomial model was fitted with the equation as follows: Size(Y 3 Þ ¼ 183:93 þ 3:98X A À 5:2X B À 70:97X C þ 0:73X D À 9:230X A X B þ 32:38X A X C À 6:45X A X D þ 0:78X B X C À 21:26X B X D À 20:48X C X D þ 9:4X 2 A À 18:47X 2 B þ 42:21X 2 C þ 8:14X 2 D : ð3:3Þ In the model, the R 2 values and the adjusted R 2 values were 0.8713 and 0.7425, respectively, in reasonable agreement. Among the factors, A, D, AC, BC, A 2 , C 2 and D 2 had a positive effect on the size of nanoparticles, and water volume (C) had the highest significant effect on the size with p < 0.0001. The change of water volume actually changed the ratio of the W/O phase. When the surfactant (BSA) concentration remained unchanged and increased the water concentration, the particle size decreased. These results were consistent with the previous study [28].

Optimization and model validation
The optimized PLGA NP loaded with astaxanthin formulation was performed based on the maximum value of EE and DL, meanwhile minimizing the particle size by using the prediction method of DESIGN-EXPERT 8.0.6 software. The optimized formation composition is shown in table 6. PLGA NP loaded with astaxanthin was synthesized at the optimal conditions three times to verify. The experimentally observed values for the three responses are close to the predicted values with low percentages of the relative error to ensure the validity and reproducibility of the model [33]. Figure 1 shows the linear relationship between the experimental value of the response and the predicted value. The more uniformly the point is close to the 45°line, the better the model fits [34]. The synthesised method of optimized PLGA NP loaded with astaxanthin was prepared in the remaining experiments.

Characterization of the optimized poly(lactic-co-glycolic acid) nanoparticles loaded with astaxanthin
The mean particle size of optimized AST-PLGA NP was 154.4 ± 0.35 nm, and the PDI was 0.16 ± 0.032, while the PLGA NP without astaxanthin using the same method was 147.5 ± 1.48 nm and 0.256 ± 0.012 with narrow size distribution, respectively (figure 2a). The size of AST-PLGA NP was slightly larger  AST-PLGA NP were 96.42 ± 0.73% and 7.19 ± 0.12%, respectively. It is reported that the EE and DL are the significant parameters because they are directly related to the administration of the nanoparticle's quantity [35]. The EE and DL results suggest that the properties of the optimized AST-PLGA NP was excellent. SEM micrographs of astaxanthin, blank PLGA NP and AST-PLGA NP are shown in figure 2b-d, respectively. It could be seen that pure astaxanthin was an irregularly lumpy crystal. After synthesizing, nanoparticles using the optimized method, PLGA NP and AST-PLGA NP showed small, smooth and globular particles without drug crystal presentation, which suggested that they can more easily penetrate into cells than pure astaxanthin. Besides, the results of TEM further validate the morphology of the nanoparticles.

Fourier-transform infrared analysis of the optimized astaxanthin-loaded poly(lactic-coglycolic acid) nanoparticles
To investigate whether astaxanthin was successfully loaded in PLGA nanoparticles, the FTIR experiment was carried out. As shown in figure 3, the astaxanthin presented characteristic absorption peaks at 2922 cm −1 , 1650 cm −1 , 1551 cm −1 and 976 cm −1 , which are attributed to −CH 3 stretching, the CvO stretching vibration, the stretching vibration of CvC in the aromatic ring and C-H in the C and C conjugate system, respectively [36,37]. All bands of blank PLGA NP could be found in the spectra of AST-PLGA NP. Compared with the spectra of AST-PLGA NP and astaxanthin, it was obvious that specific functional groups of astaxanthin also appeared in AST-PLGA NP. The results demonstrated that astaxanthin was successfully encapsulated into PLGA nanoparticles physically. 3.4. X-ray diffraction analysis of the optimized astaxanthin-poly(lactic-co-glycolic acid) nanoparticles As shown in figure 3b, the characteristic crystalline peaks of astaxanthin occurred at 10.7°, 14.2°, 16.4°, 18.2°, 20.4°and 24.7°, respectively, which meant that the astaxanthin still maintained the crystalline form. Besides, the blank PLGA NP and AST-PLGA NP showed a wider diffraction peak around 11.4°to 25.9°, which is similar to a previous report [38]. Otherwise, as for AST-PLGA NP, the sharpness peaks of astaxanthin disappeared in the X-ray diffraction (XRD) spectra, which suggested that astaxanthin was successfully encapsulated in an amorphous form. These results are similar to a previous report [18].
3.5. Thermal analysis of the optimized astaxanthin-loaded poly(lactic-co-glycolic acid) nanoparticles The results of DSC thermograms are shown in figure 3c. The endothermic peaks of astaxanthin, pure PLGA NP and AST-PLGA NP are observed clearly. The curves of astaxanthin (figure 3c, blue line) showed a sharp endothermic peak at 230°C [39]. However, AST-PLGA NP (figure 3c, red line) only displays a sharp endothermic peak around 350°C like pure PLGA NP (figure 3c, black line) [40]. The disappearance of the astaxanthin characteristic peaks in AST-PLGA NP proved that astaxanthin was successfully wrapped into polymer nanoparticles in an amorphous state. Figure 3d shows the TGA curves of astaxanthin, blank PLGA NP and AST-PLGA NP respectively. There were two stages of weight loss. The weight loss of the first stage up to 100°C may be the physical adsorption by water in those test compounds [41]. The mass loss of astaxanthin and AST-PLGA NP at 256°C were around 4.8% and 3.9%, respectively, while about 93.3% and 90.7% was a loss at 550°C, respectively. In other words, these data further indicated the presence of astaxanthin in the PLGA nanoparticles [42].

Cellular uptake study
To demonstrate that PLGA nanoparticles were uptake by HaCaT cells, FITC was chosen as a model fluorescent probe to perform a quantitative and qualitative analysis of cellular uptake of the nanoparticles [43]. The results of FACS could quantitatively analyse the uptake of fluorescent nanoparticles by HaCaT cells as shown in figure 4a,b. The cellular uptakes of FITC-loaded PLGA nanoparticles at a concentration of 100 µg ml −1 were time-dependent. After incubation with cells for 0 h, 1 h, 2 h, 4 h and 8 h, the fluorescence intensity of the fluorescent nanoparticles were 2214 ± 48.31; 5989 ± 65.02; 7521 ± 39.01; 10 007 ± 76.43; 18 758 ± 76.41, respectively. The intracellular uptake of FITCloaded PLGA nanoparticles in HaCaT cells was visualized using a laser scanning confocal microscope. FITC-loaded PLGA nanoparticles (green) were distributed around the nucleus (blue) which proved that nanoparticles can be taken up by the cells through the cell membrane (figure 4c). After fluorescent nanoparticles were treated with HaCaT cells for 1 h, the green fluorescence intensity was very weak. However, the green fluorescence intensity was obvious at 8 h. These results confirmed that the cellular uptake of PLGA nanoparticles increased in a time-dependent way.

Cell viability and photoprotective studies
The in vitro cytotoxicity of astaxanthin, blank PLGA NP and AST-PLGA NP in HaCaT cells were evaluated using MTT assay ( figure 5a,b). The HaCaT cells treated with astaxanthin did not cause any significant decrease in cell viability, while at 48 h, cell viability of 2, 4 and 8 µM was less than 95%. So, we chose 0.4, 0.8 and 1 µM for later experiments. As previously reported [18], the biocompatibility of PLGA NP and AST-PLGA NP had lower cytotoxicity at high concentration at 24 and 48 h.
As shown in figure 5c,d, the photo protective effects of astaxanthin and AST-PLGA NP on HaCaT cells under UVB irradiation of 30 mJ cm −2 were evident. The results displayed that the cell viability of irradiated cells was significantly reduced by 42.53% ( p < 0.005) compared with control cells, which means that the irradiation model was appropriate [44]. The data showed astaxanthin at 1 µM and 1 µM AST-PLGA NP increased cell viability about 16.10% and 18.14%, respectively, indicating that AST-PLGA NP possessed increased UV protection compared to free astaxanthin.  UVB radiation can cause a mass of ROS, which induces oxidative stress and consequently accounts for cellular compensation which ultimately leads to programmed cell death [45]. In our study, ROS levels in HaCaT cells was obviously increased after 30 mJ cm −2 of UVB irradiation compared with control cells (by 297.7 ± 14.46%). As shown in figure 6a, astaxanthin pre-treatment decreased the intracellular ROS generation and reduced DCF fluorescence intensity; the dose-response effect was observed. The influence on ROS level of AST-PLGA NP had a similar result, which indicated AST-PLGA NP did not change the antioxidant properties of astaxanthin. Furthermore, the results of the laser scanning confocal microscope for astaxanthin and AST-PLGA NP testifies that the ROS level was quenched and varies with concentration. In a word, astaxanthin and AST-PLGA NP can both protect HaCaT cells from UVB-induced oxidative stress by scavenging free radicals.

Measurement of mitochondrial membrane depolarization
Mitochondrial membrane depolarization, which caused ΔΨm to lose, is regarded as a hallmark of the early and irreparable stage of cell apoptosis [46]. Red fluorescence stands for the mitochondrial aggregated form of JC-1 because of a high ΔΨm, while green fluorescence represents the monomeric form of JC-1 owing to low ΔΨm [47]. Variation in ΔΨm leads to mitochondrial dysfunction and

Conclusion
In this research, we have successfully developed the PLGA nano drug delivery system loaded with astaxanthin using the emulsion solvent evaporation technique. The four-factor and three-level BBD optimized the parameters and obtained the optimal process conditions. With this, we synthesized AST-PLGA NP with the concentration of PLGA 10 mg ml −1 , the concentration of astaxanthin 0.81 mg ml −1 , water volume 3 ml and sonication time 0.5 min. This synthesized method optimized the AST-PLGA NP with an EE of 96.42 ± 0.73%; DL capacity of 7.19 ± 0.12% and size of 154.4 ± 0.35 nm, indicating the good drug delivery capacity of AST-PLGA NP. In addition, the results of SEM and TEM showed that the nanoparticles were discrete and spherical in shape and displayed a good size distribution. At the same time, FTIR studies confirmed that astaxanthin had been successfully loaded into PLGA NP. XRD and DSC studies demonstrated that astaxanthin existed in the form of dispersed amorphous or disordered crystals in the molecular state while formed a solid solution state in the polymer matrix. PLGA NP were biocompatible and could be taken up by HaCaT cells in a time-dependent manner. In particular, AST-PLGA NP showed better antioxidant activity compared to pure astaxanthin in the UVB radiation photodamage model of HaCaT cells. Furthermore, AST-PLGA NP resists the photodamage in HaCaT cells by reducing ROS levels and restoring mitochondrial membrane potential. Therefore, these results may provide a new means to solve the problem such as high hydrophobicity and poor chemical stability of astaxanthin and demonstrate the therapeutic application of PLGA-encapsulated astaxanthin nanoparticles in skin diseases. All together, the AST-PLGA NP possesses the potential application in the field of cosmetics and might be considered as a potential therapeutic component for skin disease.