Constructing Selenium Nanoparticles with Enhanced Storage Stability and Antioxidant Activities via Conformational Transition of Curdlan

Selenium nanoparticles (SeNPs) are among the emerging selenium supplements because of their high bioactivity and low toxicity. However, bare SeNPs are prone to activity loss caused by aggregation and sedimentation. This study aims to stabilize SeNPs with curdlan (CUR), a polysaccharide, to maintain or even enhance their biological activity. Herein, the stable SeNPs were constructed via the unique conformational transition of CUR induced by alkali-neutralization (AN) pretreatment. The physicochemical properties and structures of the prepared SeNPs were characterized by dynamic light scattering (DLS), UV-visible spectroscopy, Fourier-transform infrared spectroscopy (FTIR), X-ray diffraction (XRD), transmission electron microscopy (TEM), and free-radical-scavenging activity assays. The results show that most SeNPs are stabilized within the triple helix of CUR that has been pretreated with high-intensity AN treatment. These amorphous, small-sized (average size was 53.6 ± 17.7 nm), and stabilized SeNPs have significantly enhanced free-radical-scavenging ability compared to the control and can be well-stabilized for at least 240 days at 4 °C. This work indicates that CUR, as a food additive, can be used to well-stabilize SeNPs by AN pretreatment and provides a facile method to prepare and enhance the stability and bioactivity of SeNPs via triple-helix conformational transition.


Introduction
Selenium plays a key role in regulating immunity and sperm activity [1], and selenium levels in the body can affect the occurrence of inflammation and diseases [2]. A recent analysis on the relationship between COVID-19 cases and the regional selenium status of China has revealed that the pathogenicity and lethality of COVID-19 are higher in regions with low selenium levels (except for Wuhan), indicating the potential role of selenium in the prevention and treatment of viral infections [3]. Among numerous selenium supplements, selenium nanoparticles (SeNPs) reportedly have lower toxicity and higher biological activity than organic selenium (e.g., selenomethionine and Se-methyl selenocysteine) and inorganic selenium (e.g., sodium selenite (Na 2 SeO 3 )) [4][5][6][7]. SeNPs can be synthesized chemically [8] or by using physical procedures [9] or can be obtained by biological ways [10]. The main method of preparing SeNPs is the chemical reduction in selenium salts, such as Na 2 SeO 3 and sodium selenosulfate [11][12][13][14]. Synthesis of SeNPs by redox reaction of Na 2 SeO 3 and ascorbic acid is a relatively simple method. The reduction in selenite with ascorbic acid can be described by the general reaction as follows [15].

Alkali-Neutralization Pretreatment of CUR
Alkali-neutralization (AN) pretreatment was performed according to our previous work with slight modifications [34]. In brief, 0.6 g of CUR powder was first slowly added into the 30 mL of ultrapure water under stirring magnetically for 12 h at 25 • C; then, the aqueous suspension of CUR was filtered and the filtrate was collected for subsequent steps. The filtrate was divided into two equal portions, and the NaOH solutions were separately added to make the final NaOH concentrations 0.1 and 1.0 mol/L. The mixture was stirred magnetically at 25 • C for 6 h to induce the denaturation of CUR. Finally, the curdlan dispersion was neutralized (pH 7) and immediately used for the preparation of CUR-SeNP composites.

Preparation of CUR-SeNPs
CUR-SeNPs were prepared according to the following procedure. The reaction of sodium selenite with ascorbic acid was carried out as described by Li, Li, Wong, Chen, Zhang, Liu, and Zheng [35] with slight modifications. In brief, the CUR solutions with different intensities of AN pretreatment were immediately separately mixed with 250 µL of 0.1 mol/L Na 2 SeO 3 aqueous solution, and magnetically stirred for 1 min. Then, 500 µL of 0.2 mol/L ascorbic acid aqueous solution was separately added into the mixtures to start the synthesis reaction of SeNPs, and the resulting mixtures were stirred for 24 h in the dark at 25 • C. At the end of the reaction, the mixtures were dialyzed against ultrapure water for 48 h, and the water was changed every 12 h. After the necessary analysis of the dialyzed sample, it was freeze dried with a SCIENTZ 18N (Xinzhi, China) freeze dryer and stored at 4 • C and 25 • C. The samples were named CUR-SeNPs-L and CUR-SeNPs-H according to the intensity of AN pretreatment. Control sample without CUR was prepared as above and named SeNP control. Control samples without SeNPs were prepared as described previously and named CUR-L and CUR-H according to the intensity of AN pretreatment.

Hydrodynamic Diameter Distribution and Zeta Potential Measurements
The hydrodynamic-diameter distribution and zeta potential of the CUR-SeNP composites were measured by dynamic light scattering (DLS) with a ZETASIZER NANO ZS90 (Malvern Panalytical, Malvern, UK) at 25 • C [36]. The diluted sample solution was injected into the cuvette along the inner wall with a pipette, and the liquid surface height did not exceed the maximum height specified by the instrument, and then the cuvette was placed into the test port and covered with the cuvette, and the test chamber cover was closed for measurement. A PCS1115 cuvette was selected for the hydrodynamic-diameter distribution measurement. The refractive and absorption indices were set at 2.92 and 0.30, respectively. The equalization time of all samples was set as 90 s. A DTS1070 cuvette was selected for zeta potential measurement, and the number of zeta runs was set to 12.

UV-vis Absorption Spectroscopy
About 2 mL of the diluted sample was drawn into a quartz cuvette and transferred into a corrected UV-2501(PC)S (Shimadzu, Tokyo, Japan) UV-vis absorption spectrometer. The scanning wavelength range was 200-800 nm with a resolution of 1 nm [36].

Fourier-Transform Infrared (FTIR) Spectroscopy
The KBr-disk method was used for FTIR analysis [18]. Briefly, the lyophilized sample was suitably ground to powder and mixed with dried KBr powder for grinding and drying. Then, the mixture was transferred into the mold for a 3 min hydraulic treatment to obtain a test piece, which was immediately tested in a TENSOR II (Bruker, Billerica, MA, USA) FTIR spectrometer. The FTIR spectra of the samples were recorded in the scanning wavenumber range of 4000-400 cm −1 and at a resolution of 0.5 cm −1 . Each sample was scanned 16 times. The results were analyzed by an OPUS 8.5 (Bruker, Billerica, MA, USA).

X-ray Diffraction (XRD)
The lyophilized sample was suitably ground to powder and measured by a D/MAX 2500 V (Rigaku, Tokyo, Japan) X-ray diffractometer using Bragg-Brentano (θ, 2θ) geometry and Cu-Kα radiation. The operational parameters were as follows: scanning range of 5-80 • ; scanning speed of 4 • /min; and the voltage and current were 40 kV and 40 mV, respectively [27]. The results were analyzed using a Jade 5.0 (MDI, Livermore, CA, USA).

Transmission Electron Microscopy (TEM)
The samples were observed with a HT-7700 (Hitachi, Tokyo, Japan) transmission electron microscope [37]. In a typical procedure, before the observation, the sample was dripped onto a copper grid using a syringe and air dried at room temperature in an ultra-clean bench. Accelerating voltage set to 100 kV. Mean particle size of SeNPs in electron microscopy photographs was calculated by Nano Measurer 1.2.5 software (Xujie, Shanghai, China).
2.9. Assay of Reactivity of CUR-SeNP Composites with Free Radicals 2.9.1. Hydroxyl-Radical-Scavenging Activity Assay Hydroxyl-radical-scavenging activity was determined by the procedure described by Mao, Zou, Feng, Wang, Zhao, Ye, Zhu, Wu, Yang, and Wu [38] with slight modifications. In a typical procedure, 1.0 mL of 9.0 mmol/L FeSO 4 aqueous solution, 1.0 mL of 9.0 mmol/L salicylic acid ethanol solution, and 1.0 mL of sample were mixed. After adding 1.0 mL of 9.0 mmol/L H 2 O 2 , the mixture was incubated for 30 min at 37 • C. The absorbance was measured at 510 nm. The hydroxyl-radical-scavenging ability was calculated using the following equation: where A 1 , A 2 , and A 0 represent the absorbance in the presence of sample, the absorbance in the absence of H 2 O 2 (water instead), and the absorbance in the absence of sample (water instead), respectively.

Superoxide-Radical-Scavenging Activity Assay
Superoxide-radical-scavenging activity was determined by the procedure described by Li [39] with slight modifications. In a typical procedure, 50 µL of 60 mmol/L pyrogallol solution (1 mmol/L HCl, 37 • C) was mixed with 2900 µL of Tris-HCl buffer (0.05 mol/L, pH 7.4, 37 • C) containing 1 mmol/L Na 2 EDTA. The absorbance was measured at 325 nm every 30 s for 10 min. The superoxide-radical-scavenging ability was calculated using the following equation: Superoxide-radical-scavenging ability (%) = (∆A 0 − ∆A 1 )/∆A 0 × 100 (2) where ∆A 0 represents the increment in A 325nm of the mixture without a sample, and ∆A 1 represents the increment in A 325nm of the mixture with a sample.

Statistical Data Analysis
All determinations were performed in duplicate, and the results are presented as the mean ± standard deviation. Statistical analysis was conducted using IBM SPSS 20.0 software. An analysis of variance (ANOVA) using Turkey's multiple-range test was performed for mean comparison at p < 0.05.

Hydrodynamic-Diameter Distribution and Zeta Potential of CUR-SeNPs
Generally, at a given selenium concentration, the SeNPs with smaller particle size exhibit a more yellowish and more transparent appearance [20,24,36]. As shown in Figure 1a, both CUR-SeNPs were more yellowish in color and more transparent than the SeNP control, indicating that the SeNPs in both composites were stabilized and had a smaller particle size. The bottom photos of the samples were also obtained ( Figure 1b). It was clear that dark red particles appeared at the bottom of the SeNP control, indicating that slight aggregation and sedimentation of SeNPs had occurred. In addition, we found a few translucent, small, and colored particles at the bottom of the CUR-SeNPs-L sample, which may be due to re-formation of triple-helix aggregates (THAs) entrapping some SeNPs under the low-intensity AN treatment. Three samples were also irradiated with a 650 nm laser to determine the light transmission (Figure 1b), and all of them showed optical pathways. According to the Tyndall phenomenon, it indicated to some extent that the average particle size of all three samples was less than 650 nm. As shown in Figure 2a, the maximum absorption wavelengths (λ max ) of the three samples were 497, 258, and 264 nm, respectively, and their absorbances started to increase significantly from 600 nm, which was consistent with the typical UV-vis spectroscopic pattern of SeNPs [40]. The introduction of SeNPs seems to change the UV absorption of CUR, which is caused by the superposition of the absorption spectra of CUR and SeNPs. As shown in Figure 2a, CUR-L and CUR-H have the same λ max (266 nm), although they are very different in conformation [34]. However, the particle sizes of SeNPs in CUR-SeNPs-L (45.4 ± 21.1 nm) and CUR-SeNPs-H (53.6 ± 17.7 nm) were different, as supported by our subsequent TEM results. Based on a previous study on the relationship between the particle size of SeNPs and their absorption spectral characteristics [40], the SeNPs with larger particle size have the larger λ max . For example, the λ max of SeNPs with an average particle size of 70.9 ± 9.1 nm is approximately 270 nm, while the λ max of SeNPs with an average particle size of 101.6 ± 9.8 nm is approximately 300 nm. Based on this relationship, it is reasonable to assume that the λ max of SeNPs in both CUR-SeNPs-L and CUR-SeNPs-H should be less than 266 nm, and the λ max of the former is smaller. Therefore, the larger SeNPs in CUR-SeNPs-H resulted in a weaker blueshift (266 nm to 264 nm) in λ max , while CUR-SeNPs-L had a stronger blueshift (266 nm to 258 nm).
Superoxide-radical-scavenging ability (%) = (ΔA0 − ΔA1)/ΔA0 × 100 (2) where ΔA0 represents the increment in A325 nm of the mixture without a sample, and ΔA1 represents the increment in A325 nm of the mixture with a sample.

Statistical Data Analysis
All determinations were performed in duplicate, and the results are presented as the mean ± standard deviation. Statistical analysis was conducted using IBM SPSS 20.0 software. An analysis of variance (ANOVA) using Turkey's multiple-range test was performed for mean comparison at p < 0.05.

Hydrodynamic-Diameter Distribution and Zeta Potential of CUR-SeNPs
Generally, at a given selenium concentration, the SeNPs with smaller particle size exhibit a more yellowish and more transparent appearance [20,24,36]. As shown in Figure  1a, both CUR-SeNPs were more yellowish in color and more transparent than the SeNP control, indicating that the SeNPs in both composites were stabilized and had a smaller particle size. The bottom photos of the samples were also obtained ( Figure 1b). It was clear that dark red particles appeared at the bottom of the SeNP control, indicating that slight aggregation and sedimentation of SeNPs had occurred. In addition, we found a few translucent, small, and colored particles at the bottom of the CUR-SeNPs-L sample, which may be due to re-formation of triple-helix aggregates (THAs) entrapping some SeNPs under the low-intensity AN treatment. Three samples were also irradiated with a 650 nm laser to determine the light transmission (Figure 1b), and all of them showed optical pathways. According to the Tyndall phenomenon, it indicated to some extent that the average particle size of all three samples was less than 650 nm. As shown in Figure 2a, the maximum absorption wavelengths (λmax) of the three samples were 497, 258, and 264 nm, respectively, and their absorbances started to increase significantly from 600 nm, which was consistent with the typical UV-vis spectroscopic pattern of SeNPs [40]. The introduction of SeNPs seems to change the UV absorption of CUR, which is caused by the superposition of the absorption spectra of CUR and SeNPs. As shown in Figure 2a, CUR-L and CUR-H have the same λmax (266 nm), although they are very different in conformation [34]. However, the particle sizes of SeNPs in CUR-SeNPs-L (45.4 ± 21.1 nm) and CUR-SeNPs-H (53.6 ± 17.7 nm) were different, as supported by our subsequent TEM results. Based on a previous study on the relationship between the particle size of SeNPs and their absorption spectral characteristics [40], the SeNPs with larger particle size have the larger λmax. For example, the λmax of SeNPs with an average particle size of 70.9 ± 9.1 nm is approximately 270 nm, while the λmax of SeNPs with an average particle size of 101.6 ± 9.8 nm is approximately 300 nm. Based on this relationship, it is reasonable to assume that the λmax of SeNPs in both CUR-SeNPs-L and CUR-SeNPs-H should be less than 266 nm, and the λmax of the former is smaller. Therefore, the larger SeNPs in CUR-SeNPs-H resulted in a weaker blueshift (266 nm to 264 nm) in λmax, while CUR-SeNPs-L had a stronger blueshift (266 nm to 258 nm).    Table 1. As shown in Table 1, the z-average size of the SeNP control, CUR-SeN and CUR-SeNPs-H was 159.1, 230.1, and 280.6 nm, respectively, indicating that CU creased the overall Dh. As shown in Figure 2b,c (original graphs can be found in Fi S1), the Dh distribution by intensity and volume demonstrated this pattern with incre clarity. The P3 peaks with an average Dh of 4737 nm and 4978 nm (by intensity) in C SeNPs-L and CUR-L confirmed the presence of THAs in the solution. The size of T was reported to be in the micrometer scale [41]. In CUR-SeNPs-L and CUR-SeNPs-H peaks could be observed, which were not present in their control samples without Se indicating that the P1 peaks were generated by free SeNPs. Therefore, after confirm the reasons for the formation of P1 peak and P3 peak in the Dh distribution, it is reason that the P2 peak was attributed by independent triple helices (ITHs) dissociated THAs. Furthermore, as shown in Table 1, the average Dh-P2 (both by intensity and volu of both CUR-SeNP samples was higher than that of their control samples without Se indicating that SeNPs were bound to ITHs. However, the binding between SeNPs CUR was different in the two CUR-SeNP samples. As we described in the introduc CUR has different conformational transitions induced by different intensities of AN treatment. In the formation of CUR-SeNPs-L, SeNPs cannot be wrapped inside the through the reaggregation of ITHs to THAs, so SeNPs can only be attached to the ou of the ITHs or be trapped inside the re-aggregated THAs [34]. Comparatively, in the mation of CUR-SeNPs-H, the ITHs could be reformed by the rewinding of random allowing more SeNPs to be stabilized inside the triple helix by chemical interaction  Table 1. As shown in Table 1, the z-average size of the SeNP control, CUR-SeNPs-L, and CUR-SeNPs-H was 159.1, 230.1, and 280.6 nm, respectively, indicating that CUR increased the overall D h . As shown in Figure 2b,c (original graphs can be found in Figure S1), the D h distribution by intensity and volume demonstrated this pattern with increased clarity. The P3 peaks with an average D h of 4737 nm and 4978 nm (by intensity) in CUR-SeNPs-L and CUR-L confirmed the presence of THAs in the solution. The size of THAs was reported to be in the micrometer scale [41]. In CUR-SeNPs-L and CUR-SeNPs-H, P1 peaks could be observed, which were not present in their control samples without SeNPs, indicating that the P1 peaks were generated by free SeNPs. Therefore, after confirming the reasons for the formation of P1 peak and P3 peak in the D h distribution, it is reasonable that the P2 peak was attributed by independent triple helices (ITHs) dissociated from THAs. Furthermore, as shown in Table 1, the average D h-P2 (both by intensity and volume) of both CUR-SeNP samples was higher than that of their control samples without SeNPs, indicating that SeNPs were bound to ITHs. However, the binding between SeNPs and CUR was different in the two CUR-SeNP samples. As we described in the introduction, CUR has different conformational transitions induced by different intensities of AN pretreatment. In the formation of CUR-SeNPs-L, SeNPs cannot be wrapped inside the ITHs through the reaggregation of ITHs to THAs, so SeNPs can only be attached to the outside of the ITHs or be trapped inside the re-aggregated THAs [34]. Comparatively, in the formation of CUR-SeNPs-H, the ITHs could be reformed by the rewinding of random coils, allowing more SeNPs to be stabilized inside the triple helix by chemical interaction and physical encapsulation during the rewinding. This conclusion was confirmed by the larger D h-P2 (both by intensity and volume) of CUR-SeNPs-H compared to CUR-SeNPs-L.  (Figure 2d, and original graph can be found in Figure S2). The zeta potential difference between CUR-SeNPs-L and CUR-SeNPs-H also confirmed that the SeNPs in CUR-SeNPs-H were encapsulated inside the triple helices, because the SeNPs were entrapped in the cavities of the triple helix, shielding the partial negative charges of SeNPs and leading to smaller negative potentials.

Molecular Structure and Crystalline Structure of CUR-SeNPs
FTIR and XRD are effective methods for analyzing the structural characteristics of CUR-SeNPs. Figure 3a shows the FTIR spectra of CUR-L, CUR-H, CUR-SeNPs-L, and CUR-SeNPs-H, respectively. In the case of CUR-L, the stretching or vibrations corresponding to the main peaks were as follows: 3385 cm −1 (stretching vibration of -OH), 2881 cm −1 (stretching vibration of C-H), 1606 cm −1 (absorption peak of the adsorbed water blending vibration), 1002 cm −1 (stretching vibration of C-O), and 890 cm −1 (β-glycosidic linkage) [34]. The differences in the absorption peaks of these stretching vibrations between the samples are detailed in Table 2.
physical encapsulation during the rewinding. This conclusion was confirmed b larger Dh-P2 (both by intensity and volume) of CUR-SeNPs-H compared to CUR-SeN  Figure S2). The zeta potential difference between SeNPs-L and CUR-SeNPs-H also confirmed that the SeNPs in CUR-SeNPs-H were e sulated inside the triple helices, because the SeNPs were entrapped in the cavities triple helix, shielding the partial negative charges of SeNPs and leading to smaller tive potentials.

Molecular Structure and Crystalline Structure of CUR-SeNPs
FTIR and XRD are effective methods for analyzing the structural characteris CUR-SeNPs. Figure 3a shows the FTIR spectra of CUR-L, CUR-H, CUR-SeNPs-L CUR-SeNPs-H, respectively. In the case of CUR-L, the stretching or vibrations sponding to the main peaks were as follows: 3385 cm −1 (stretching vibration of -OH cm −1 (stretching vibration of C-H), 1606 cm −1 (absorption peak of the adsorbed blending vibration), 1002 cm −1 (stretching vibration of C-O), and 890 cm −1 (β-glyc linkage) [34]. The differences in the absorption peaks of these stretching vibratio tween the samples are detailed in Table 2.   There is a characteristic absorption peak of the β-conformation at 890 cm −1 in all spectra, indicating the presence of curdlan in all samples. The -OH stretching vibration peaks of both CUR-SeNP samples were shifted to the lower wavenumbers (3379 cm −1 and 3376 cm −1 ) compared to that of AN-pretreated CURs (3385 cm −1 ), indicating that SeNPs were stabilized by the interaction between Se atoms and hydroxyl groups, and this result was similar to that of the other THP-SeNP composites [23,24]. From the electronegativity of H (χ = 2.20), O (χ = 3.44), and Se (χ = 2.55) atoms, it can be inferred that the strength of the H···Se hydrogen bonding interaction is lower than the intra-or intermolecular H···O hydrogen bonding of the CUR, which was confirmed by a recent molecular simulation study that reported the important role of H···Se hydrogen bonding in the formation of functional materials [42]. Therefore, we attributed the red shift of the -OH stretching vibration peak to Se···H hydrogen bonding. Furthermore, the larger wavenumber shift of CUR-SeNPs-H suggested that more SeNPs were stabilized compared to CUR-SeNPs-L, which was consistent with the previous findings of the hydrodynamic-diameter distribution. For the -CH and C-O stretching vibration peaks, the shift was identical for both CUR-SeNP samples, indicating that the interaction between SeNPs and -OH groups also affected the -CH and C-O stretching vibration [23], but the difference in the AN pretreatment intensity did not affect the shift of this peak. The absorption peak around 1606 cm −1 was attributed to the adsorbed water blending vibration; the slightly lower wavenumber of CUR-H indicated its higher water solubility, and the addition of SeNPs had no obvious effect on this peak.
Overall, the FTIR results showed that the SeNPs were stabilized through the hydrogen bonding interactions, and the different positions of broad peaks in the spectra of the two CUR-SeNPs samples (3379 cm −1 in CUR-SeNPs-L and 3376 cm −1 in CUR-SeNPs-H) indicated the different structures of the CUR-SeNPs prepared at different AN pretreatment intensities. Figure 3b shows the XRD patterns of SeNP control, CUR-L, CUR-H, CUR-SeNPs-L, and CUR-SeNPs-H, respectively. No narrow diffraction peaks were observed in the pattern of SeNPs, indicating that the SeNPs prepared in this work were amorphous. A broad peak around 2q = 20 • was observed in all samples containing CUR, which is characteristic of CUR [43]. A low-intensity narrow diffraction peak at around 2q = 28 • was observed in the pattern of CUR-L but did not appear in the other patterns. We think this peak was caused by THAs, and its disappearance in the pattern of CUR-SeNPs-L indicated that SeNPs interfered with the re-formation of THAs. In addition, no other narrow diffraction peaks were observed in all samples, indicating that the introduction of CUR did not change the amorphous state of SeNPs.

Surface Micromorphology of CUR-SeNPs
CUR-SeNPs were observed using TEM to more intuitively determine if the SeNPs were encapsulated inside the triple helices. Figure 4a shows the photograph of the SeNP control. The aggregation of SeNPs was serious without the stabilization of CUR, which was obviously not conducive to the stability and storage of SeNPs. The average particle size of the SeNP control was 162.2 ± 17.4 nm, which was similar to the average hydrodynamic diameter of the SeNP control in Table 1. Figure 4b shows the photograph of CUR-SeNPs-L. Many free SeNPs were observed, and many large flake-like substances were believed to be water-insoluble THAs. A few SeNPs were observed inside the THAs, and these SeNPs were believed to have been trapped inside the THAs during the re-aggregation of THAs. The average particle size of individual SeNPs in CUR-SeNPs-L was 45.4 ± 21.1 nm, which was slightly smaller than the average hydrodynamic diameter because of the hydrated layer on the surface of SeNPs.
SeNPs-L. Many free SeNPs were observed, and many large flake-like substances were b lieved to be water-insoluble THAs. A few SeNPs were observed inside the THAs, a these SeNPs were believed to have been trapped inside the THAs during the re-aggreg tion of THAs. The average particle size of individual SeNPs in CUR-SeNPs-L was 45.4 21.1 nm, which was slightly smaller than the average hydrodynamic diameter because the hydrated layer on the surface of SeNPs.  Figure 4c shows the photograph of CUR-SeNPs-H. THAs were rarely present. T SeNPs showed a worm-like distribution instead of the large agglomerated distribution the control, and the SeNPs were more clearly divided from one another. Moreover, t SeNPs did not show a very regular spherical feature, which may be caused by squeezi during the rewinding of the triple helix. Therefore, we believe that the SeNPs were enca sulated in the triple helix, and the worm-like structure of triple helix determined the sp tial distribution of SeNPs. In addition, as shown in Figure 4, the average particle size individual SeNPs in CUR-SeNPs-H was 53.6 ± 17.7 nm, which was also slightly smal than the previous average hydrodynamic diameter, and the ranking of SeNP particle siz in the three samples was exactly the same as that of Dh corresponding to the P1 peak Table 1.
Overall, the particle size of individual SeNPs stabilized by CUR was significan lower, which was conducive to enhancing the biological activity of SeNPs [24]. Althou the particle size of individual SeNPs in CUR-SeNPs-H was slightly larger than that CUR-SeNPs-L, the former SeNPs were encapsulated inside the triple helix and it did n contain insoluble THAs, which was more favorable for long-term stabilization.

Reactivity of CUR-SeNPs with Reactive Oxygen Species
While different AN pretreatments affected the structure of CUR-SeNPs, the biolo cal activity of SeNPs in both composites was unclear. The physiological activity of SeN is usually associated with reactive oxygen species (ROS). Song, Chen, Zhao, Sun, Che, a Leng [44] suggested that the Se 0 ↔Se 4+ cycle is the mechanism of the anti-tumor activity SeNPs. Therefore, the reactivity of CUR-SeNPs with ROS can reflect the biological activ of SeNPs to some extent. The forms of intracellular ROS in human cells are generally co sidered to be hydroxyl radicals, superoxide anion radicals, and hydrogen peroxide. The fore, to better simulate the reaction of CUR-SeNPs with ROS in human cells, the reactiv of CUR-SeNPs with hydroxyl radicals and superoxide anion radicals was determined.
The reactivity of CUR-SeNPs with hydroxyl radicals is shown in Figure 5a. The sca enging ability was already about 100% at an ascorbic acid concentration higher than mg/mL (Vc concentrations were omitted on the horizontal coordinate in Figure   Figure 4. TEM photographs of (a) SeNP control, (b) CUR-SeNPs-L, and (c) CUR-SeNPs-H with their calculated average particle size of SeNPs. Figure 4c shows the photograph of CUR-SeNPs-H. THAs were rarely present. The SeNPs showed a worm-like distribution instead of the large agglomerated distribution in the control, and the SeNPs were more clearly divided from one another. Moreover, the SeNPs did not show a very regular spherical feature, which may be caused by squeezing during the rewinding of the triple helix. Therefore, we believe that the SeNPs were encapsulated in the triple helix, and the worm-like structure of triple helix determined the spatial distribution of SeNPs. In addition, as shown in Figure 4, the average particle size of individual SeNPs in CUR-SeNPs-H was 53.6 ± 17.7 nm, which was also slightly smaller than the previous average hydrodynamic diameter, and the ranking of SeNP particle sizes in the three samples was exactly the same as that of D h corresponding to the P1 peak in Table 1.
Overall, the particle size of individual SeNPs stabilized by CUR was significantly lower, which was conducive to enhancing the biological activity of SeNPs [24]. Although the particle size of individual SeNPs in CUR-SeNPs-H was slightly larger than that in CUR-SeNPs-L, the former SeNPs were encapsulated inside the triple helix and it did not contain insoluble THAs, which was more favorable for long-term stabilization.

Reactivity of CUR-SeNPs with Reactive Oxygen Species
While different AN pretreatments affected the structure of CUR-SeNPs, the biological activity of SeNPs in both composites was unclear. The physiological activity of SeNPs is usually associated with reactive oxygen species (ROS). Song, Chen, Zhao, Sun, Che, and Leng [44] suggested that the Se 0 ↔Se 4+ cycle is the mechanism of the anti-tumor activity of SeNPs. Therefore, the reactivity of CUR-SeNPs with ROS can reflect the biological activity of SeNPs to some extent. The forms of intracellular ROS in human cells are generally considered to be hydroxyl radicals, superoxide anion radicals, and hydrogen peroxide. Therefore, to better simulate the reaction of CUR-SeNPs with ROS in human cells, the reactivity of CUR-SeNPs with hydroxyl radicals and superoxide anion radicals was determined.
The reactivity of CUR-SeNPs with hydroxyl radicals is shown in Figure 5a. The scavenging ability was already about 100% at an ascorbic acid concentration higher than 0.2 mg/mL (Vc concentrations were omitted on the horizontal coordinate in Figure 5), indicating the validity of this hydroxyl-radical-scavenging capacity assay. The scavenging ability of CUR-SeNPs-L and CUR-SeNPs-H for hydroxyl radicals was higher than that of the SeNP control at any selenium concentration. The enhanced scavenging ability of CUR-SeNPs is not due to the reaction of CUR itself with radicals, as CUR has been reported to have very weak antioxidant properties [45]. We believed that the size of SeNPs and the hydrogen bonding interactions between SeNPs and CUR hydroxyl groups were the main reasons for the enhanced scavenging ability. First, there is a negative correlation between the antioxidant activity of SeNPs and their particle size [12], i.e., the smaller the particle size, the stronger the antioxidant activity, because SeNPs with small particle size have a higher specific surface area, which makes them more likely to react with radicals. Secondly, the introduction of hydroxyl groups on the surface of SeNPs can enhance the interaction between CUR-SeNPs and the radicals. For example, the free-radical-scavenging ability of chitosan-stabilized SeNPs was reported to be higher than that of carboxymethyl chitosanstabilized SeNPs at the same concentration (the particle size difference of SeNPs in the two samples was very small) because there were more C 6 primary hydroxyl groups converted to carboxylate groups in carboxymethyl chitosan molecules compared to chitosan, which may reduce the total number of hydroxyl groups [46]. In our work, the FTIR results confirmed that more Se···H hydrogen bonds were formed between SeNPs and hydroxyl groups in CUR-SeNPs-H. Although the TEM results and the DLS results showed that the particle size of SeNPs in CUR-SeNPs-H was slightly larger than that of SeNPs in CUR-SeNPs-L, the hydroxyl-radical-scavenging ability of CUR-SeNPs-H was still larger than that of CUR-SeNPs-L, indicating that Se···H hydrogen bond number is more important for the radical-scavenging ability of CUR-SeNPs than the size of SeNPs.
indicating the validity of this hydroxyl-radical-scavenging capacity assay. The scavenging ability of CUR-SeNPs-L and CUR-SeNPs-H for hydroxyl radicals was higher than that of the SeNP control at any selenium concentration. The enhanced scavenging ability of CUR-SeNPs is not due to the reaction of CUR itself with radicals, as CUR has been reported to have very weak antioxidant properties [45]. We believed that the size of SeNPs and the hydrogen bonding interactions between SeNPs and CUR hydroxyl groups were the main reasons for the enhanced scavenging ability. First, there is a negative correlation between the antioxidant activity of SeNPs and their particle size [12], i.e., the smaller the particle size, the stronger the antioxidant activity, because SeNPs with small particle size have a higher specific surface area, which makes them more likely to react with radicals. Secondly, the introduction of hydroxyl groups on the surface of SeNPs can enhance the interaction between CUR-SeNPs and the radicals. For example, the free-radical-scavenging ability of chitosan-stabilized SeNPs was reported to be higher than that of carboxymethyl chitosan-stabilized SeNPs at the same concentration (the particle size difference of SeNPs in the two samples was very small) because there were more C6 primary hydroxyl groups converted to carboxylate groups in carboxymethyl chitosan molecules compared to chitosan, which may reduce the total number of hydroxyl groups [46]. In our work, the FTIR results confirmed that more Se···H hydrogen bonds were formed between SeNPs and hydroxyl groups in CUR-SeNPs-H. Although the TEM results and the DLS results showed that the particle size of SeNPs in CUR-SeNPs-H was slightly larger than that of SeNPs in CUR-SeNPs-L, the hydroxyl-radical-scavenging ability of CUR-SeNPs-H was still larger than that of CUR-SeNPs-L, indicating that Se···H hydrogen bond number is more important for the radical-scavenging ability of CUR-SeNPs than the size of SeNPs. Furthermore, the concentration-scavenging ability of all samples showed a concentration-dependent increase, and the corresponding half-inhibitory concentration (IC50) was approximately 6.58 μg/mL (selenium concentration). The reactivity of CUR-SeNPs with superoxide anion radicals is shown in Figure 5b. The pattern between the samples was basically the same as that of the hydroxyl-radical-scavenging ability assay. The main difference was that the reactivity of the CUR-SeNPs with superoxide anion radicals was not as good as that with hydroxyl radicals, similar to the results reported by Zhang, Zhai, Zhao, Ren, and Leng [47]. Overall, the antioxidant activity of CUR-SeNPs-H was better than that of CUR-SeNPs-L. Figure 6 shows the change in the Dh of CUR-SeNPs-H with storage time. As described in the previous section, peak P1 and P2 in the Dh distribution were assigned to free SeNPs Furthermore, the concentration-scavenging ability of all samples showed a concentrationdependent increase, and the corresponding half-inhibitory concentration (IC 50 ) was approximately 6.58 µg/mL (selenium concentration). The reactivity of CUR-SeNPs with superoxide anion radicals is shown in Figure 5b. The pattern between the samples was basically the same as that of the hydroxyl-radical-scavenging ability assay. The main difference was that the reactivity of the CUR-SeNPs with superoxide anion radicals was not as good as that with hydroxyl radicals, similar to the results reported by Zhang, Zhai, Zhao, Ren, and Leng [47]. Overall, the antioxidant activity of CUR-SeNPs-H was better than that of CUR-SeNPs-L. Figure 6 shows the change in the D h of CUR-SeNPs-H with storage time. As described in the previous section, peak P1 and P2 in the D h distribution were assigned to free SeNPs and ITHs that contain some SeNPs, respectively. Thus, the changes in their corresponding D h are shown separately. At 4 • C (Figure 6a), CUR-SeNPs-H exhibited good storage even at 240 days because no significant changes occurred in the D h . At 25 • C (Figure 6b), from the first day to 180th day, the D h of free SeNPs increased significantly from 81.83 ± 13.39 nm to 126.37 ± 15.86 nm, and the D h of combination of SeNPs and triple helix increased significantly from 403.52 ± 90.73 nm to 629.92 ± 60.61 nm. The accelerated aggregation of SeNPs explained these pattern differences at varied temperatures [48]. Although the storage stability of CUR-SeNPs-H at 25 • C was insufficient, it can maintain good stability at 4 • C for 240 days, and to our knowledge, the storage stability of CUR-SeNPs-H is better than that of most other polysaccharide-SeNP composites [17]. As previously described, CUR-SeNPs-H could co-stabilize SeNPs through CUR molecule wrapping and Se···H hydrogen bonding interactions. We believe that this stable triplex wrapping is the main reason that makes CUR-SeNPs-H more stable than other polysaccharide-SeNP composites.

Storage Stability of CUR-SeNPs
Dh are shown separately. At 4 °C (Figure 6a), CUR-SeNPs-H exhibited good storage even at 240 days because no significant changes occurred in the Dh. At 25 °C (Figure 6b), from the first day to 180th day, the Dh of free SeNPs increased significantly from 81.83 ± 13.39 nm to 126.37 ± 15.86 nm, and the Dh of combination of SeNPs and triple helix increased significantly from 403.52 ± 90.73 nm to 629.92 ± 60.61 nm. The accelerated aggregation of SeNPs explained these pattern differences at varied temperatures [48]. Although the storage stability of CUR-SeNPs-H at 25 °C was insufficient, it can maintain good stability at 4 °C for 240 days, and to our knowledge, the storage stability of CUR-SeNPs-H is better than that of most other polysaccharide-SeNP composites [17]. As previously described, CUR-SeNPs-H could co-stabilize SeNPs through CUR molecule wrapping and Se···H hydrogen bonding interactions. We believe that this stable triplex wrapping is the main reason that makes CUR-SeNPs-H more stable than other polysaccharide-SeNP composites.

Construction Mechanism of CUR-SeNPs
Based on all characterization results, the construction mechanisms of CUR-SeNPs under different AN pretreatments were derived. As shown in Figure 7, when the intensity of AN pretreatment was low (CNaOH/HCl = 0.1 mol/L), large-sized THAs first disaggregated to small-sized THAs and some ITHs under alkali treatment. After neutralization treatment and initiation of the synthetic reactions of SeNPs, some SeNPs were trapped inside the THAs through the re-aggregation of ITHs while the remaining SeNPs failed to be stabilized and thus dispersed in water. When the intensity of AN pretreatment was high (CNaOH/HCl = 1.0 mol/L), THAs first disaggregated and denatured to random coils under alkali treatment. After neutralization and initiation of the synthetic reactions of SeNPs, most SeNPs were stabilized and encapsulated inside the triple helices through the rewinding of ITHs, whereas a few SeNPs were stabilized on the outside of the triple helices or dispersed in water. Obviously, the AN pretreatment intensity determined whether or not SeNPs can be effectively encapsulated inside the triple helix, which was crucial to the stability of SeNPs.

Construction Mechanism of CUR-SeNPs
Based on all characterization results, the construction mechanisms of CUR-SeNPs under different AN pretreatments were derived. As shown in Figure 7, when the intensity of AN pretreatment was low (C NaOH/HCl = 0.1 mol/L), large-sized THAs first disaggregated to small-sized THAs and some ITHs under alkali treatment. After neutralization treatment and initiation of the synthetic reactions of SeNPs, some SeNPs were trapped inside the THAs through the re-aggregation of ITHs while the remaining SeNPs failed to be stabilized and thus dispersed in water. When the intensity of AN pretreatment was high (C NaOH/HCl = 1.0 mol/L), THAs first disaggregated and denatured to random coils under alkali treatment. After neutralization and initiation of the synthetic reactions of SeNPs, most SeNPs were stabilized and encapsulated inside the triple helices through the rewinding of ITHs, whereas a few SeNPs were stabilized on the outside of the triple helices or dispersed in water. Obviously, the AN pretreatment intensity determined whether or not SeNPs can be effectively encapsulated inside the triple helix, which was crucial to the stability of SeNPs.

Conclusions
All results confirmed the hypothesis that different AN pretreatments induced different self-assembly behaviors of CUR, which in turn resulted in prepared CUR-SeNPs with

Conclusions
All results confirmed the hypothesis that different AN pretreatments induced different self-assembly behaviors of CUR, which in turn resulted in prepared CUR-SeNPs with various structures, stabilities, and dispersion properties. At a low AN pretreatment intensity (C NaOH/HCl = 0.1 mol/L), only the disaggregations/aggregations of THAs were present, so most SeNPs were free in water and a few SeNPs were trapped in THAs. At a high AN pretreatment intensity (C NaOH/HCl = 1.0 mol/L), the ITHs dissociated from THAs undergo further unwinding-rewinding, so most SeNPs were encapsulated inside the cavity during the rewinding of ITHs. From the latter route, SeNPs were better stabilized and dispersed by Se···H hydrogen bonding interactions and physical wrapping of a triple helix, and the CUR-SeNPs had a better antioxidant activity.
Overall, this study constructed the CUR-SeNPs and revealed the mechanism of CUR-SeNPs prepared through different self-assembly behaviors of the triple helix. In addition to providing evidence to the hypothesis initially proposed, this finding can also help broaden the application of CUR and SeNPs. In addition, we believe that the triple-helix self-assembly of CUR will have more potential in nutrition fortification and supplementation research.
In the future, further research and analysis are needed in the following areas, some of which we have already undertaken: (1) The underlying mechanism of the Se···H hydrogen bonding interaction between SeNPs and CUR (or other polysaccharides); (2) The maximum loading of SeNPs in CUR helices and methods for their detection; (3) Further investigation of other biological activities of CUR-SeNPs, such as anti-tumor activity, immunomodulatory activity, etc.