Determination of Two Differently Manufactured Silicon Dioxide Nanoparticles by Cloud Point Extraction Approach in Intestinal Cells, Intestinal Barriers and Tissues

Food additive amorphous silicon dioxide (SiO2) particles are manufactured by two different methods—precipitated and fumed procedures—which can induce different physicochemical properties and biological fates. In this study, precipitated and fumed SiO2 particles were characterized in terms of constituent particle size, hydrodynamic diameter, zeta potential, surface area, and solubility. Their fates in intestinal cells, intestinal barriers, and tissues after oral administration in rats were determined by optimizing Triton X-114-based cloud point extraction (CPE). The results demonstrate that the constituent particle sizes of precipitated and fumed SiO2 particles were similar, but their aggregate states differed from biofluid types, which also affect dissolution properties. Significantly higher cellular uptake, intestinal transport amount, and tissue accumulation of precipitated SiO2 than of fumed SiO2 was found. The intracellular fates of both types of particles in intestinal cells were primarily particle forms, but slowly decomposed into ions during intestinal transport and after distribution in the liver, and completely dissolved in the bloodstream and kidneys. These findings will provide crucial information for understanding and predicting the potential toxicity of food additive SiO2 after oral intake.


Introduction
Amorphous silicon dioxide (SiO 2 ) is widely applied to the food industry as a food additive-for example, as a thickener, anticaking agent, carrier for fragrances and flavors, chillproofing agent in beer, or filter aid [1,2]. As a direct additive, the United States Food and Drug Administration regulates that the level of SiO 2 cannot exceed 2% by weight of the food, whereas it can be added in the amount necessary to obtain its intended functionality as an indirect additive [3]. It is registered as food additive E551 in the European Union, and its maximum levels in dried, powdered foods and tablet foods are authorized at 10 g/kg and quantum satis level, respectively [4]. An acceptable daily intake for SiO 2 is not currently specified [5,6]. Until now, food additive SiO 2 has been used for a long time without any adverse effects on human health at its levels of usage. However, the European Food Safety Authority (EFSA) suggests re-evaluating the toxicity of E551, due to the presence of nanosized SiO 2 . A nanomaterial is defined as a material with more than 50% of its particles having at least one external dimension in the size range of 1-100 nm, based on number size distribution [7]. The EFSA report and other studies demonstrated that nanoparticles (NPs) of less than 100 nm were present in E551, although the amounts of NPs differed between the materials tested and analytical methods [8][9][10]. Recently, it was reported that E551 can be considered to be a food additive intentionally produced to nanosized materials [11]. However, current specifications do not include the particle size distribution nor the percentage of NPs in SiO 2 [3,8]. 2 of 19 Concerns about the potential toxicity of SiO 2 have been raised because some in vitro studies have suggested that SiO 2 NPs can cause inflammatory responses of the intestinal wall [12,13] and affect small intestinal function [14]. Neurotoxicity of SiO 2 NPs was also reported in neuronal cells and zebrafish [15][16][17]. However, there is no evidence that these toxicological findings are comparable to in vivo animal models and humans. Indeed, the no-observed-adverse-effect level (NOAEL) of SiO 2 NPs was reported to be more than 2000 mg/kg [18]. Meanwhile, SiO 2 is known to not be rapidly decomposed into Si ions under acidic or biological conditions [19]; it undergoes in vitro and in vivo gradual degradation, forming silicic acid (ortho-, meta-, di-, and trisilicates) via hydrolysis [20,21]. Therefore, the determination of the fates of food additive SiO 2 in commercial products and biological systems is of importance, and can provide information about the presence and amount of NPs, aggregates, or dissolved forms, contributing to the understanding of its potential toxicity.
Another factor to be considered is that the physicochemical properties and biological responses of SiO 2 can be also influenced by the manufacturing method. Indeed, commercially available food additive SiO 2 is manufactured by two different methods: One is precipitated SiO 2 , produced by the reaction of metal silicate solutions and sulfuric acid, resulting in white precipitates [22]. The other is fumed SiO 2 -also known as pyrogenic SiO 2 -which is produced by the flame pyrolysis of silicon tetrachloride (SiCl 4 ), or from quartz sand vaporized in an electric arc at 3000 • C [23,24]. It was reported that the dissolution and physicochemical properties of SiO 2 NPs differ between the two manufacturing methods [25,26]. Hence, the fates of precipitated and fumed SiO 2 NPs in the body, which have not been clearly demonstrated yet, might be also different.
In the present study, we developed a cloud point extraction (CPE) approach to separate SiO 2 NPs captured in Triton X-114 (TX-114)-based micelles from biomatrices. TX-114 is a cost-effective surfactant, and forms micelles at room temperature. CPE methods using TX-114, ethylenediaminetetraacetic acid, or 8-hydroxyquinoline were used to analyze trace NPs under aqueous conditions or minerals in tissues, resulting in their separation in the precipitated, surfactant-rich phase [27,28]. In this case, pretreatments to digest organic matrices were essentially required prior to CPE application. However, pretreatment with acids at high temperatures can decompose or dissolve NPs into small molecules or ions, thereby inhibiting NP detection as an intact particle form. No attempt has been made to detect SiO 2 NPs in complex biosystems using a CPE approach. Optimization of the CPE method depending on matrix type, such as cell and tissue matrices, was carried out in order to determine the fates of SiO 2 NPs in intestinal cells, intestinal barriers, and tissues after oral administration in rats. Two differently manufactured food additive SiO 2 NPsprecipitated and fumed types-were used to investigate the effects of the manufacturing method on the particles' fates in biological systems.

Characterization
The constituent particle sizes and size distributions of two differently manufactured SiO 2 particles were measured via scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Figure 1 demonstrates that the constituent particle sizes of precipitated and fumed SiO 2 particles were 16 ± 4 and 14 ± 4 nm, respectively, with round shapes. No significant difference in average size between the two particles was found (p > 0.05). The dynamic light scattering (DLS) results in distilled water (DW) show that the particle fraction of fumed SiO 2 was smaller (43% smaller than 100 nm) than that of precipitated SiO 2 (96% larger than 200 nm), based on number % ( Table 1). The Z-average diameters of precipitated SiO 2 (375 ± 1 nm) were larger than those of fumed SiO 2 (156 ± 1 nm), indicating that the former was more aggregated than the latter in DW. On the other hand, the specific surface areas of precipitated and fumed SiO 2 particles, as measured by the Brunauer-Emmett-Teller (BET) method, were 167 ± 1 and 177 ± 1 m 2 /g, respectively ( Table 1). The zeta potential values of both SiO 2 particles as a function of pH are presented in Figure S1A, showing more negative charges for precipitated SiO 2 than for fumed SiO 2 , at pH levels ranging from 2 to 5. The isoelectric points (IEPs) of precipitated SiO 2 and fumed SiO 2 were pH 1.53 and 1.92, respectively. The hydrodynamic diameters of SiO 2 particles were also measured as a function of pH, showing high aggregation of both particles at low pH (below 3.0) ( Figure S1B). Smaller hydrodynamic diameters were also found for fumed SiO 2 than those for precipitated SiO 2 at pH 2.0-7.4 ( Figure S1B). nm), indicating that the former was more aggregated than the latter in DW. On the other hand, the specific surface areas of precipitated and fumed SiO2 particles, as measured by the Brunauer-Emmett-Teller (BET) method, were 167 ± 1 and 177 ± 1 m 2 /g, respectively ( Table 1). The zeta potential values of both SiO2 particles as a function of pH are presented in Figure S1A, showing more negative charges for precipitated SiO2 than for fumed SiO2, at pH levels ranging from 2 to 5. The isoelectric points (IEPs) of precipitated SiO2 and fumed SiO2 were pH 1.53 and 1.92, respectively. The hydrodynamic diameters of SiO2 particles were also measured as a function of pH, showing high aggregation of both particles at low pH (below 3.0) ( Figure S1B). Smaller hydrodynamic diameters were also found for fumed SiO2 than those for precipitated SiO2 at pH 2.0-7.4 ( Figure S1B).

Method Validation and Quantitative Analysis in Cells or Tissue Matrices
The accuracy and precision of the analytical procedure for SiO2 were explored using Si standard solution, precipitated SiO2, fumed SiO2, and SiO2 spiked with human intestinal Caco-2 cells or liver tissue, after predigestion with nitric acid (HNO3)/hydrofluoric acid (HF) in a microwave digestion system, followed by inductively coupled plasma atomic emission spectroscopy (ICP-AES) analysis. Indeed, HF is effective at dissolving SiO2 into 1 Specific surface areas, as measured by the Brunauer-Emmett-Teller (BET) method. Different lowercase letters (a,b) indicate significant differences between precipitated SiO 2 and fumed SiO 2 (p < 0.05).

Method Validation and Quantitative Analysis in Cells or Tissue Matrices
The accuracy and precision of the analytical procedure for SiO 2 were explored using Si standard solution, precipitated SiO 2 , fumed SiO 2 , and SiO 2 spiked with human intestinal Caco-2 cells or liver tissue, after predigestion with nitric acid (HNO 3 )/hydrofluoric acid (HF) in a microwave digestion system, followed by inductively coupled plasma atomic emission spectroscopy (ICP-AES) analysis. Indeed, HF is effective at dissolving SiO 2 into Si ions [29], while HNO 3 is widely applied to digest organic matrices [10,29]. Table 2 shows that the recoveries (%) of Si standard solution, pristine food additive SiO 2 , and SiO 2 in spiked cells or liver tissues ranged from 96.1 to 105.0%, with repeatability (relative standard deviation, RSD%) of 1.54-5.12%. The limit of detection (LOD) and limit of quantification (LOQ) values of SiO 2 in cells or tissue matrices were highly sensitive, showing 0.36-0.52 SiO 2 µg/g of matrix and 1.08-1.58 SiO 2 µg/g of matrix, respectively. The root-mean-square error (RMSE) values were also small, indicating good quality of the quantitative results obtained.

Solubility
The dissolution properties of both types of SiO 2 particles were evaluated in distilled and deionized water (DDW) and 0.5% ethanol solution according to Organization for Economic Co-operation and Development (OECD) test guideline (TG) 105 [30]. The solubility of precipitated and fumed SiO 2 particles in DDW was 0.29 and 0.31%, respectively, and slightly increased in 0.5% ethanol solution (0.49 and 0.54% for the former and the latter, respectively) ( Figure 2A). Extremely low dissolutions (< 0.1%) of both particles were found in artificial lysosomal fluid (ALF), with the same tendency obtained in DDW and ethanol, and the solubility did not increase even after 24 h ( Figure 2B). The solubility of fumed SiO 2 was significantly higher than that of precipitated SiO 2 in DDW (Figure 2A), 0.5% ethanol solution (Figure 2A), and ALF ( Figure 2B). On the other hand, the solubility of both particles was~0.19% at 6-24 h in 4% bovine serum albumin (BSA) solution, without significant differences between the two types of particles (p > 0.05) ( Figure 2B). The solubility of precipitated SiO 2 and fumed SiO 2 increased to~0.81 and~0.65%, respectively, in minimum essential medium (MEM), showing significantly higher solubility for the former than the latter (p < 0.05) ( Figure 2B). Low solubility of SiO 2 particles was also measured in in vitro digestion fluids-such as saliva, gastric fluids, and intestinal fluids-and the highest solubility was found in the intestinal fluid, compared with saliva or gastric fluid ( Figure 2C). When three consecutive steps of digestion fluids, such as saliva followed by gastric and intestinal fluids, were applied, the solubility of precipitated SiO 2 and fumed SiO 2 was 2.8 and 2.0%, respectively, with significant difference (p < 0.05) ( Figure 2D). The same tendency was also obtained using ex vivo rat-extracted gastrointestinal (GI) fluids, showing 2.4 and 1.8% solubility for precipitated SiO 2 and fumed SiO 2 , respectively ( Figure 2E).
The hydrodynamic diameters of SiO 2 particles were also checked in DDW, ALF, 4% BSA solution, MEM, and three consecutive steps of in vitro digestion solutions, where their solubility was evaluated. Figure 2F shows that the hydrodynamic diameters of precipitated SiO 2 were significantly larger than those of fumed SiO 2 in DDW and ALF, whereas fumed SiO 2 had statistically larger hydrodynamic diameters than precipitated SiO 2 in MEM and the three consecutive steps of digestion solutions. No statistical difference in the hydrodynamic diameters between the two types of particles was found in 4% BSA solution (p > 0.05). The smallest and largest hydrodynamic diameters of SiO 2 particles were measured in MEM and the three steps of digestion fluids, respectively. SiO2 in MEM and the three consecutive steps of digestion solutions. No statistical difference in the hydrodynamic diameters between the two types of particles was found in 4% BSA solution (p > 0.05). The smallest and largest hydrodynamic diameters of SiO2 particles were measured in MEM and the three steps of digestion fluids, respectively.

Optimization of CPE for SiO 2
The CPE for two differently manufactured types of SiO 2 particles was optimized using TX-114 at pH 3.0. The micelle formation of particles in TX-114 easily occurs at the IEP, but a slightly greater pH than the IEPs (1.53 and 1.92 for precipitated SiO 2 and fumed SiO 2 , respectively, Figure S1) was chosen. Indeed, the addition of sodium chloride (NaCl) during the CPE process at a pH higher than the IEP elevates the zeta potential values of NPs [31], which reduces the electrostatic repulsion between negatively charged NPs in TX-114-based micelles, thereby resulting in a net zero interaction and enhancing phase separation [31,32]. Hence, the particles as intact forms are precipitated in TX-114-based micelles, and Si ions released from the particles are present in the supernatants after centrifugation. As shown in Figure 3A, the hydrodynamic diameters of two differently manufactured SiO 2 types in TX-114-rich precipitates obtained via CPE were similar to those of pristine SiO 2 , without significant differences (p > 0.05). This result was obtained after the digestion of organic matrices with HNO 3 and 1 mL of hydrogen peroxide (H 2 O 2 ). The ICP-AES analysis shows that the recoveries (%) as intact particle forms were 96.3 and 92.2% for precipitated SiO 2 and fumed SiO 2 , respectively ( Figure 3B). Meanwhile, the concentrations of Si ions released from the particles during the CPE procedure, found in the supernatants, were extremely low, showing 0.9 and 0.6% for precipitated SiO 2 and fumed SiO 2 , respectively. Total detected Si levels in the precipitates and supernatants by CPE were 96.9 and 92.8% for precipitated SiO 2 and fumed SiO 2 , respectively. fumed SiO2 under the same conditions (p < 0.05). Different lowercase letters (a,b,c) indicate significant differences between experimental conditions (p < 0.05).

Optimization of CPE for SiO2
The CPE for two differently manufactured types of SiO2 particles was optimized using TX-114 at pH 3.0. The micelle formation of particles in TX-114 easily occurs at the IEP, but a slightly greater pH than the IEPs (1.53 and 1.92 for precipitated SiO2 and fumed SiO2, respectively, Figure S1) was chosen. Indeed, the addition of sodium chloride (NaCl) during the CPE process at a pH higher than the IEP elevates the zeta potential values of NPs [31], which reduces the electrostatic repulsion between negatively charged NPs in TX-114-based micelles, thereby resulting in a net zero interaction and enhancing phase separation [31,32]. Hence, the particles as intact forms are precipitated in TX-114-based micelles, and Si ions released from the particles are present in the supernatants after centrifugation. As shown in Figure 3A, the hydrodynamic diameters of two differently manufactured SiO2 types in TX-114-rich precipitates obtained via CPE were similar to those of pristine SiO2, without significant differences (p > 0.05). This result was obtained after the digestion of organic matrices with HNO3 and 1 mL of hydrogen peroxide (H2O2).
The ICP-AES analysis shows that the recoveries (%) as intact particle forms were 96.3 and 92.2% for precipitated SiO2 and fumed SiO2, respectively ( Figure 3B). Meanwhile, the concentrations of Si ions released from the particles during the CPE procedure, found in the supernatants, were extremely low, showing 0.9 and 0.6% for precipitated SiO2 and fumed SiO2, respectively. Total detected Si levels in the precipitates and supernatants by CPE were 96.9 and 92.8% for precipitated SiO2 and fumed SiO2, respectively.  The CPE was also optimized with SiO 2 -spiked Caco-2 cells or liver tissue. In this case, we focused on the effect of ultrasonic homogenization-which was used to lyse the cells or tissues-on the inert size distribution of NPs. Figure 4 demonstrates that the recoveries (%) of both SiO 2 particle forms, obtained via CPE in Caco-2 cells or liver tissue, ranged from 91.6 to 97.8%, although 1.5-6.7% of Si ions were found to be released from the particles after the CPE and lysis procedure. Slightly higher Si releases from the particles in SiO 2 -spiked liver cells than those in SiO 2 -spiked Caco-2 cells were also observed after CPE and lysis. The CPE was also optimized with SiO2-spiked Caco-2 cells or liver tissue. In this case, we focused on the effect of ultrasonic homogenization-which was used to lyse the cells or tissues-on the inert size distribution of NPs. Figure 4 demonstrates that the recoveries (%) of both SiO2 particle forms, obtained via CPE in Caco-2 cells or liver tissue, ranged from 91.6 to 97.8%, although 1.5-6.7% of Si ions were found to be released from the particles after the CPE and lysis procedure. Slightly higher Si releases from the particles in SiO2-spiked liver cells than those in SiO2-spiked Caco-2 cells were also observed after CPE and lysis.

Intracellular Uptake and Intestinal Transport Fates of SiO2 Particles
The uptake and intracellular fates of two differently manufactured types of SiO2 particles were evaluated in Caco-2 cells after incubation for 2, 6, and 24 h, and determined by applying the optimized CPE. The highest uptake amounts of the two differently manufactured SiO2 particle types were detected at 2 h post-incubation, and decreased with incubation time ( Figure 5). Significant differences in uptake amount between precipitated SiO2 and fumed SiO2 were found, showing higher uptake of the former than the latter (p < 0.05). SiO2 particles inside cells were determined via CPE to be primarily present in particle form, regardless of incubation time or manufactured type. About 95% of both types of SiO2 was found in particle form at 2 h, and this figure decreased to ~88% at 24 h, without significant differences between the manufactured types (p > 0.05), whereas the proportion of Si ions increased as incubation time increased.

Intracellular Uptake and Intestinal Transport Fates of SiO 2 Particles
The uptake and intracellular fates of two differently manufactured types of SiO 2 particles were evaluated in Caco-2 cells after incubation for 2, 6, and 24 h, and determined by applying the optimized CPE. The highest uptake amounts of the two differently manufactured SiO 2 particle types were detected at 2 h post-incubation, and decreased with incubation time ( Figure 5). Significant differences in uptake amount between precipitated SiO 2 and fumed SiO 2 were found, showing higher uptake of the former than the latter (p < 0.05). SiO 2 particles inside cells were determined via CPE to be primarily present in particle form, regardless of incubation time or manufactured type. About 95% of both types of SiO 2 was found in particle form at 2 h, and this figure decreased to~88% at 24 h, without significant differences between the manufactured types (p > 0.05), whereas the proportion of Si ions increased as incubation time increased. The intestinal transports and fates of SiO2 particles were evaluated using in vitro 2D Caco-2 monolayer and 3D follicle-associated epithelial (FAE) models. Caco-2 monolayer and FAE models represent the intestinal tight junction barrier and microfold (M) cells in Peyer's patches, respectively. The permeability of Lucifer yellow-a fluorescent dye that is only transported paracelluarly-was 2.3 and 2.4% in Caco-2 monolayer and FAE models, respectively, indicating well-established intestinal barriers [33,34]. The results show that the two types of particles were determined to be transported through both Caco-2 monolayer and M cells, but more massive transports by M cells were found ( Figure 6). Total transported amounts of both kinds of SiO2 particles increased as incubation time increased, and significantly higher transport of precipitated SiO2 compared with fumed SiO2 was found in both models. Transport amounts (%) of precipitated SiO2 and fumed SiO2 at 6 h were ~1.0 and ~0.7% by Caco-2 monolayers, and ~2.1 and ~1.5% by the FAE model, respectively. It is worth noting that the recoveries (%) of SiO2 particles in both apical and basolateral sides ranged from 98.7 to 101.6% in all cases, supporting the accuracy of the analyzed transport levels (Table S1).
When the fate of SiO2 was determined by CPE, most SiO2 particles were present as particle forms at 2 h, and the portion of Si ions increased up to ~49% at 6 h in a Caco-2 monolayer model, regardless of manufactured type. On the other hand, ~65% of Si ions from precipitated SiO2 and fumed SiO2 were found at 2 and 6 h in an FAE model. The intestinal transports and fates of SiO 2 particles were evaluated using in vitro 2D Caco-2 monolayer and 3D follicle-associated epithelial (FAE) models. Caco-2 monolayer and FAE models represent the intestinal tight junction barrier and microfold (M) cells in Peyer's patches, respectively. The permeability of Lucifer yellow-a fluorescent dye that is only transported paracelluarly-was 2.3 and 2.4% in Caco-2 monolayer and FAE models, respectively, indicating well-established intestinal barriers [33,34]. The results show that the two types of particles were determined to be transported through both Caco-2 monolayer and M cells, but more massive transports by M cells were found ( Figure 6). Total transported amounts of both kinds of SiO 2 particles increased as incubation time increased, and significantly higher transport of precipitated SiO 2 compared with fumed SiO 2 was found in both models. Transport amounts (%) of precipitated SiO 2 and fumed SiO 2 at 6 h were~1.0 and~0.7% by Caco-2 monolayers, and~2.1 and~1.5% by the FAE model, respectively. It is worth noting that the recoveries (%) of SiO 2 particles in both apical and basolateral sides ranged from 98.7 to 101.6% in all cases, supporting the accuracy of the analyzed transport levels (Table S1).
When the fate of SiO 2 was determined by CPE, most SiO 2 particles were present as particle forms at 2 h, and the portion of Si ions increased up to~49% at 6 h in a Caco-2 monolayer model, regardless of manufactured type. On the other hand,~65% of Si ions from precipitated SiO 2 and fumed SiO 2 were found at 2 and 6 h in an FAE model.

Oral Distribution and Fates of SiO 2 Particles in Tissues
The distribution and fates of SiO 2 particles were evaluated after a single-dose oral administration (300 or 2000 mg/kg) in rats. The doses were chosen based on the previous report, showing that the NOAEL of SiO 2 NPs was more than 2000 mg/kg [18]. Two different doses (300 and 2000 mg/kg) were used to investigate whether the dose affects the oral particle/ionic forms of SiO 2 particles. The timepoints for tissue distribution after oral administration of 300 or 2000 mg/kg were set at 2 and 10 h, based on our previous research [35,36]. The CPE model developed was applied to the gastric fluid, liver, blood, and kidneys after oral administration in rats, and Si levels were analyzed in the precipitates and supernatants via CPE. As shown in Figure 7, extremely low amounts (less than 0.2%) of SiO 2 particles were dissolved in the gastric fluid, with no significant differences between administered doses or manufactured type (p > 0.05). When SiO 2 particles were distributed in the liver through intestinal and hepatic portal vein transport, the majority of the fates (~81%) of precipitated SiO 2 and fumed SiO 2 were ionic forms, regardless of manufactured type. Moreover, particulate forms were detected in the liver in a dose-dependent manner, based on more particle forms being detected at 2000 mg/kg than at 300 mg/kg. When SiO 2 particles were absorbed into the bloodstream, greater proportions of both types of SiO 2 were present as ionic forms, and then completely dissolved into Si ions in the kidneys. A slightly greater proportion of particulate forms of precipitated SiO 2 compared with fumed SiO 2 was found in the blood. Almost 100% of fumed SiO 2 was present in ionic forms after entering systemic circulation and distribution in the kidneys.

Oral Distribution and Fates of SiO2 Particles in Tissues
The distribution and fates of SiO2 particles were evaluated after a single-dose oral administration (300 or 2000 mg/kg) in rats. The doses were chosen based on the previous report, showing that the NOAEL of SiO2 NPs was more than 2000 mg/kg [18]. Two different doses (300 and 2000 mg/kg) were used to investigate whether the dose affects the oral particle/ionic forms of SiO2 particles. The timepoints for tissue distribution after in a dose-dependent manner, based on more particle forms being detected at 2000 mg/kg than at 300 mg/kg. When SiO2 particles were absorbed into the bloodstream, greater proportions of both types of SiO2 were present as ionic forms, and then completely dissolved into Si ions in the kidneys. A slightly greater proportion of particulate forms of precipitated SiO2 compared with fumed SiO2 was found in the blood. Almost 100% of fumed SiO2 was present in ionic forms after entering systemic circulation and distribution in the kidneys.

Discussion
The biological fate and toxicity of NPs can differ according to their physicochemical properties, which are mainly related to different manufacturing methods. The results of physicochemical characterization of the most widely applied food additive SiO2 particles-precipitated, and fumed forms-demonstrate different characteristics; the constituent particle sizes of precipitated SiO2 and fumed SiO2 were similar, with no significant difference (p > 0.05, Figure 1), but present as aggregates, as shown in the SEM/TEM images ( Figure 1) and DLS results ( Table 1). The hydrodynamic diameters of fumed SiO2 in DW were much smaller than those of precipitated SiO2, as indicated by the

Discussion
The biological fate and toxicity of NPs can differ according to their physicochemical properties, which are mainly related to different manufacturing methods. The results of physicochemical characterization of the most widely applied food additive SiO 2 particlesprecipitated, and fumed forms-demonstrate different characteristics; the constituent particle sizes of precipitated SiO 2 and fumed SiO 2 were similar, with no significant difference (p > 0.05, Figure 1), but present as aggregates, as shown in the SEM/TEM images ( Figure 1) and DLS results ( Table 1). The hydrodynamic diameters of fumed SiO 2 in DW were much smaller than those of precipitated SiO 2 , as indicated by the high proportion of the fraction less than 100 nm and small Z-average size, contributing to high BET surface area for fumed SiO 2 ( Table 1). The zeta potential values of both particles were negative, but slightly more negative charge was found for precipitated SiO 2 compared with fumed SiO 2 , leading to a lower IEP for the former (pH 1.53) than for the latter (pH 1.92) ( Figure S1).
The solubility of SiO 2 particles differed based on their manufacturing method and fluid type. Fumed SiO 2 had significantly higher dissolution properties in DDW, 0.5% ethanol, and ALF than did precipitated SiO 2 , and the solubility increased in the order of ALF < 4% BSA solution < DDW < 0.5% ethanol < MEM < digestion fluids, in all cases ( Figure 2). On the other hand, the solubility of precipitated SiO 2 was higher than that of fumed SiO 2 in MEM, in vitro digestion fluids, and ex vivo rat-extracted GI fluids ( Figure 2B-E). This discrepancy can be explained by the DLS results obtained in each fluid tested, showing larger hydrodynamic diameters of precipitated SiO 2 in DDW and ALF, but smaller diameters in MEM and the three steps of digestion solutions, compared to fumed SiO 2 ( Figure 2F). Hence, aggregation of SiO 2 under test conditions can result in low solubility, related to its small specific surface area and low reactivity. The different solubility depending on fluid type can be explained by several factors; interaction between SiO 2 and various components of MEM-such as 10% fetal bovine serum (FBS), amino acids, vitamins, and salts-can increase the solubility to some extent, as observed by high dissolution in MEM ( Figure 2B). Indeed, the role of cell culture medium as a dispersant of NPs has been well reported [37,38], which is consistent with the smallest hydrodynamic diameters of SiO 2 particles being observed in MEM ( Figure 2F). The effect of 4% BSA-which reflects albumin level in the plasma [39]-on solubility seems to be minor when the solubility between DDW and 4% BSA solution is compared (Figure 2A,B). It is interesting to note that the highest solubility of SiO 2 particles was found in the three consecutive steps of digestion fluids, where their high aggregation was found ( Figure 2D-F); this result may be related to the effect of pH, as demonstrated by the high aggregation of SiO 2 particles at acidic pH due to their IEPs at low pH ( Figure S1A). It was demonstrated that electrostatic repulsive forces between SiO 2 particles are minimized, whereas van der Waals attractive forces increase, at the IEP-inducing aggregation [40], which is also seen in Figure S1B. High dissolution properties of SiO 2 under alkaline conditions, as observed in intestinal fluid ( Figure 2C), can contribute to its high solubility in three steps of digestion fluids, where duodenal and bile conditions are reflected (~pH 8.2, Table S3) [41]. Meanwhile, the higher solubility of precipitated SiO 2 particles than of fumed SiO 2 particles is in good agreement with the report demonstrating that precipitated SiO 2 has more silanol groups (Si-OH) per unit surface than fumed SiO 2 , resulting in high solubility [26]. It is worth noting that the solubility in an in vitro three consecutive steps of digestion model ( Figure 2D) was highly correlated with ex vivo results obtained using rat-extracted GI fluids, showing about 2.4 and 1.8% solubility for precipitated SiO 2 and fumed SiO 2 , respectively ( Figure E). These results suggest that most SiO 2 particles are not dissolved, and present as particle forms under GI conditions.
All validation parameters of quantitative analytical methods for SiO 2 particles in biomatrices-such as linearity, low RMSE, high recovery (%), low repeatability (RSD %), and sensitive LOD and LOQ values-reveal the accuracy and precision of the quantitative methods, including digestion procedure and ICP-AES analysis ( Table 2). This is also supported by the high recoveries (93-102%) of total Si levels from Si ions and SiO 2 particles obtained via CPE (Figures 3 and 4). The reliability of the CPE to detect intact SiO 2 was confirmed by the statistically similar hydrodynamic diameters of pristine SiO 2 before and after CPE ( Figure 3A), and by the high recoveries (92-96%) as particle forms in TX-114based precipitates obtained via CPE ( Figure 3B). The recoveries of SiO 2 particles in the precipitates obtained by CPE ranged from 92 to 98% in SiO 2 -spiked biomatrices (Caco-2 cells and rat liver) (Figure 4), supporting the efficacy of the CPE method. Slightly increased levels of Si ions (3.7-6.7%) were detected in SiO 2 -spiked liver ( Figure 4B), probably resulting from the ultrasonic homogenization process for tissue lysis prior to CPE application. Hence, the tissue fates of SiO 2 after oral administration were evaluated considering the proportions of Si ions versus SiO 2 in SiO 2 -spiked liver, obtained via CPE ( Figure 4B).
The intracellular fates of precipitated SiO 2 and fumed SiO 2 in Caco-2 cells, determined by CPE, were mainly particle forms, even after 24 h of incubation, but~95% of particle forms at 2 h decreased to~89% at 24 h, without significant differences between the manufactured types (p > 0.05, Figure 5), suggesting slow decomposition of SiO 2 inside cells. The uptake amounts of both particles decreased as incubation time increased, implying possible exocytosis or excretion of the decomposed particles. Meanwhile, the uptake amounts of precipitated SiO 2 were significantly higher than those of fumed SiO 2 at 2-24 h, which seems to be associated with the smaller hydrodynamic diameters of the former than the latter in the three steps of in vitro digestion solutions ( Figure 2F). Indeed, intracellular organelles such as endosomes and lysosomes have acidic pH and many enzymes for the degradation of molecules [42,43]. This result suggests that small particles can be more easily and massively taken up by cells [44]. Indeed, energy-dependent endocytosis of SiO 2 particles has been well reported [45]. The intestinal transport amounts of precipitated SiO 2 through the Caco-2 monolayer and M cells were also significantly higher than those of fumed SiO 2 (Figure 6), indicating efficient intestinal transport of SiO 2 with small hydrodynamic diameters. Total transport (%) of precipitated SiO 2 and fumed SiO 2 by both the Caco-2 monolayer and FAE models at 6 h was~3.0 and~2.2%, respectively (Table S1), which is in good agreement with our in vivo oral absorption [46]. It should be noted that energy-dependent transport of SiO 2 particles through intestinal barriers has been demonstrated [8,46]. However, intestinal transport fates of SiO 2 were different depending on the types of intestinal transport models. The two differently manufactured particle types were mainly present as particles at 2 h, and~51% particle forms were found at 6 h in a Caco-2 monolayer model ( Figure 6A), indicating slow dissolution during transportation through intestinal tight junction barriers. Conversely, about 63% and 67% of Si ions were detected at 2 and 6 h, respectively, in the FAE model ( Figure 6B), implying relatively fast dissolution of SiO 2 during intestinal transport via M cells. This may be related to the role of lymphocyte Raji B cells in the FAE model, which are involved in the immune system [47].
After oral administration of two SiO 2 particles in rats, significantly higher amounts of precipitated SiO 2 were found in all organs analyzed, compared with those of fumed SiO 2 (Figure 7), which is consistent with the cellular uptake ( Figure 5) and intestinal transport ( Figure 6) amounts. Both types of particles were not dissolved in the gastric fluid, but~81% of ionized forms were detected in the liver regardless of administered doses, indicating that the majority of the oral fates of both SiO 2 types were ionic forms in the liver. On the other hand, about 4 and 7% particle forms of precipitated SiO 2 were found in the blood at 300 and 2000 mg/kg ( Figure 7A-C), respectively, whereas less than 1% of particles remained intact in the blood after the oral administration of fumed SiO 2 , regardless of the administered dose ( Figure 7B-D). No particle forms were detected in the kidneys in all cases. Taken together, SiO 2 particles were present as intact particle forms in the stomach, and slowly dissolved into Si ions during intestinal cellular uptake and transport. They were then more decomposed after distribution in the liver and entering the circulation system, and finally excreted as Si ions in the kidneys. Indeed, SiO 2 undergoes hydrolysis to form silicic acid, and in vivo biodegradation of SiO 2 to silicic acid was demonstrated [20,36]. The metabolic pathway involved in the degradation or decomposition of SiO 2 particles in the tissues and cells remains to be elucidated in the near future.

Cell Culture
Human intestinal epithelial Caco-2 cells and non-adherent human Burkitt lymphoma Raji B cells were supplied by Korean Cell Line Bank (Seoul, Korea). Caco-2 cells and Raji B cells were cultured in MEM and RPMI 1640 medium, respectively, containing 10% FBS, 100 units/mL of penicillin, and 100 µg/mL of streptomycin under 5% CO 2 atmosphere at 37 • C.

Animals
Seven-week-old female Sprague Dawley (SD) rats weighing around 130-170 g were purchased from Koatech Co. (Pyeongtaek, Gyeonggi-do, Korea). The animals were housed in plastic laboratory animal cages in a ventilated room, maintained at 20 ± 2 • C and 60% ± 10% relative humidity with a 12 h light/dark cycle. Water and commercial laboratory complete food for rats were available ad libitum. The animals were environmentally acclimated for 7 days before experimental treatments. All animal experiments were performed in accordance with the guidelines established by the Animal and Ethics Review Committee of Seoul Women's University (SWU IACUC-2020A-8). Hydrodynamic diameters and zeta potentials of SiO 2 particles were measured via DLS and electrophoretic light scattering (ELS), respectively, using a Zetasizer Nano system (Malvern Instruments, Worcestershire, UK). For DLS and ELS analysis, 1 mL of SiO 2 stock suspensions (1 mg/mL) was placed in disposable plastic cuvettes and measured at room temperature.

Characterization
The specific surface area was determined via the BET method, using N 2 gas adsorptiondesorption isotherms (at 77 K) of the particles. The moisture in the particles was removed by drying at 100 • C for 2 h prior to analysis, and then measured with a surface area analyzer (TriStar II 3020, Micromeritics, Norcross, GA, USA).

Digestion of Organic Matrices, Microwave Digestion, and ICP-AES Analysis
SiO 2 particles were quantified by measuring total Si contents using ICP-AES (JY2000 Ultrace, HORIBA Jobin Yvon, Longjumeau, France), with Si standard solutions of different concentrations. To digest organic materials in the supernatants after solubility or CPE experiments, digestion was performed with 10 mL of ultrapure HNO 3 and 1 mL of H 2 O 2 at 180 • C until the solution was colorless and completely evaporated. The same digestion procedure was applied to measure the hydrodynamic diameters of SiO 2 in the precipitates via CPE. For quantitative analysis of the precipitates containing SiO 2 via CPE, the samples were transferred to perfluoroalkoxy microwave digestion vessels, and 6 mL of 70% HNO 3 and 1 mL of 40% HF were added. The samples were digested for 55 min at 1600 W in a microwave digestion system (ETHOS EASY, Milestone Srl, Sorisole, Italy), by irradiation at 120, 160, and 210 • C for 15, 10, and 30 min, respectively, followed by holding for 1 min. After digestion, all samples were diluted to appropriate volumes with DDW, shaken manually, and analyzed via ICP-AES with a radiofrequency power of 1000 W and an argon flow rate of 0.02 mL/min in the nebulizer. Method blanks were determined by performing the same procedure in the absence of Si samples.

Validation of Quantitative Analytical Method
The method for quantitative analysis was validated by evaluating the LOD, LOQ, linearity, recovery (%), accuracy, precision, and matrix effects according to the International Conference on Harmonization (ICH, Q2B) guidance [48]. All validation parameters were evaluated with Si standard solution, precipitated and fumed SiO 2 , and SiO 2 -spiked matrices (cells and liver), at SiO 2 concentrations of 0, 2.5, 5, 25, and 50 µg/g (corresponding to Si concentrations of 0, 1.2, 2.3, 12, and 23 µg/g, respectively). The LOD and LOQ were calculated according to the following equations: LOD = 3.3 × σ/S; and LOQ = 10 × σ/S (σ: standard deviation of the response; S: slope of the calibration curve). The linearity of the quantification was evaluated as a coefficient of determination values (R 2 ). The accuracy was evaluated by the recovery (%) of known added SiO 2 at different concentrations. The precision was determined by repeatability and expressed as RSD% from measurements of five replicates [49]. Uncertainty of the obtained calibration curves was validated by calculating RMSE values [50].

In Vitro Dissolution Properties of SiO 2
The solubility of SiO 2 particles in water was evaluated according to the flask method of OECD TG 105 water solubility [30]. SiO 2 particles (50 g/L) were dispersed in DDW and preincubated with shaking at 30 • C for 24 h to achieve the saturation equilibrium, followed by shaking at 20 • C for 24, 48, and 72 h. The supernatants were then collected via ultracentrifugation (16,000× g) for 15 min, filtered through a syringe filter (pore size 0.22 µm), and quantitative analysis of the dissolved Si from SiO 2 was performed using ICP-AES (JY2000 Ultrace, HORIBA Jobin Yvon). If the differences in Si concentrations measured in the samples at 20 • C for 24, 48, and 72 h were less than 15%, the concentrations were used for the solubility values.
The solubility of SiO 2 was investigated in ALF, the composition of which is summarized in Table S2 [51]. SiO 2 particles (1 mg/mL) were dispersed in ALF and incubated for designated times (2, 6, and 24 h) at 37 • C. The solubility of SiO 2 was also assessed in the presence of 4% BSA and MEM under the same conditions. After incubation, the samples were centrifuged at 16,000× g for 15 min, and the supernatants were subjected to ICP-AES analysis.
The solubility of SiO 2 was also evaluated in an in vitro three steps of digestion model consisting of simulated saliva, gastric fluids, and intestinal fluids, as described by Peters et al. (Table S3) [9]. All simulated digestion fluids were prepared on the day of the experiment, and the fluids were preheated to 37 • C for at least 2 h. For dissolution experiments in each digestion fluid, SiO 2 particles (1 mg/mL) were dispersed in simulated saliva, gastric fluids, and intestinal fluids, and incubated for 5 min, 2 h, and 2 h, respectively, on a head-over-head rotator at 37 • C. For consecutive digestion experiments, SiO 2 particles (1 mg/mL) were dispersed in simulated saliva (6 mL) for 5 min at 37 • C, followed by sequential gastric digestion for 2 h at 37 • C by the addition of gastric fluid (12 mL). Then, intestinal digestion was further performed for 2 h at 37 • C by the addition of duodenal (12 mL) and bile (6 mL) fluids to the suspensions. After each digestion step, the samples were centrifuged at 16,000× g for 15 min, and the Si contents in the supernatants were measured using ICP-AES (JY2000 Ultrace, HORIBA Jobin Yvon).

Ex Vivo Dissolution Properties of SiO 2 in Rat-Extracted GI Fluids
Rat gastric and intestinal fluids were prepared as described by Lee et al. [52]. Briefly, the stomachs and small intestines of rats were collected and washed with saline, and then the gastric and intestinal fluids were extracted. The supernatants of the gastric and intestinal fluids were obtained by centrifugation at 16,000× g for 15 min at 4 • C, and were used for ex vivo solubility experiments. Ex vivo digestion was performed in the same manner as described in the in vitro three steps of digestion model (2:3 v/v gastric:intestinal). For solubility in each digestion fluid, SiO 2 particles (1 mg/mL) were dispersed in the rat-extracted gastric and intestinal fluids, respectively, and incubated for 2 h at 37 • C. For consecutive digestion experiments, SiO 2 particles were dispersed in the gastric fluid (0.5 mL) at a concentration of 1 mg/mL and incubated for 2 h at 37 • C. The suspensions were then combined with the intestinal fluid (0.75 mL) and incubated for a further 2 h at 37 • C. After each digestion step, the samples were centrifuged at 16,000× g for 15 min, and Si concentrations in the supernatants were analyzed using ICP-AES (JY2000 Ultrace, HORIBA Jobin Yvon).

Optimization of CPE for SiO 2
The CPE method for SiO 2 was established with the stock solutions of food additive SiO 2 (100 µg/mL), in the absence of matrices. The SiO 2 solutions (7 mL) were added to conical-bottomed glass centrifuge tubes, and the pH was adjusted to 3.0 with diluted HNO 3 . Then, 0.5 mL of 5% (w/v) TX-114 and 0.75 mL of 0.2 M NaCl were added to the samples. After dilution to 10 mL with DDW, the mixtures were incubated for 30 min at 45 • C and then centrifuged for 5 min at 2500× g at 25 • C to facilitate phase separation. The precipitates of SiO 2 particles in the TX-114-rich phase and the supernatants of the upper aqueous phase were collected for Si quantification via ICP-AES (JY2000 Ultrace, HORIBA Jobin Yvon). The recoveries (%) of SiO 2 particles in intact particle forms in the precipitated TX-114-rich phase were checked.
The CPE method for SiO 2 particles in cells was optimized using SiO 2 -spiked Caco-2 cells. Briefly, the cells (1 × 10 6 cells) were resuspended in 1 mL of DDW and transferred to conical-bottomed glass centrifuge tubes. Then, 1 mL of SiO 2 stock solution (100 µg/mL) was added to the cell suspension. The prepared SiO 2 -spiked cells were sonicated four times for 10 s at 150 W on ice using an ultrasonic homogenizer (Sonics & Materials Inc., Newtown, CT, USA). After dilution to 7 mL with DDW, the same CPE procedure for food additive SiO 2 was applied, as described above.
The CPE method for SiO 2 particles in tissues was optimized using SiO 2 -spiked rat liver. The rat liver was collected and chopped with scissors, and 0.1 g of the chopped tissues was transferred to conical-bottomed glass centrifuge tubes. After adding 1 mL of SiO 2 stock solution (100 µg/mL) to the chopped tissues, the volume was adjusted to 7 mL with DDW. The SiO 2 -spiked tissues were homogenized on ice using an ultrasound homogenizer (Sonics & Materials Inc.), and the same CPE procedure applied to food additive SiO 2 was performed, as described above.

Cellular Uptake and Intracellular Fates of SiO 2
The cells were plated at a density of 1 × 10 6 cells/well and incubated with SiO 2 particles (500 µg/mL) for 2, 6, and 24 h. After incubation, the cells were washed three times with DPBS and harvested with a scraper. Then, the cells were centrifuged for 1 min at 3000 rpm at 4 • C, and resuspended in 1 mL of DDW to determine the intracellular fate of SiO 2 via CPE. Briefly, the suspended cells (1 mL) were added to conical-bottomed glass centrifuge tubes and lysed as described in Section 4.9. Then, the volume was adjusted to 7 mL with DDW, and the same CPE procedure for SiO 2 was applied (see Section 4.9). Cells in the absence of SiO 2 particles were used as controls. Si concentrations in the precipitates and supernatants, as measured by CPE, were analyzed using ICP-AES (JY2000 Ultrace, HORIBA Jobin Yvon).

Intestinal Transport and Fates of SiO 2
A Caco-2 monoculture model was used to evaluate the transport of particles through the intestinal epithelial tight junction barrier. After Matrigel ® diluted in serum-free DMEM was coated on a Transwell ® insert for 1 h, the supernatants were removed, and the inserts were washed with serum-free DMEM. Caco-2 cells (4.5 × 10 5 cells/well) were seeded on the upper insert and grown for 21 days until the transepithelial electrical resistance (TEER) value reached more than 300 Ω cm 2 . Then, the apical medium of the monolayers was replaced by medium containing SiO 2 particles (500 µg/mL), and incubation was continued for 2 and 6 h. The pH of the apical and basolateral media was 7.4.
The FAE model, mimicking the M cells of Peyer's patches in the intestinal epithelium, was prepared as previously described [53,54]. Briefly, Caco-2 cells (1 × 10 6 cells/well) were seeded on the upper insert and grown for 14 days. Raji B cells (1 × 10 6 cells/well) in DMEM were added to the basolateral inserts and co-cultured for 5 days (150-200 Ω cm 2 ). Then, the apical medium of the monolayers was replaced by medium containing SiO 2 particles (500 µg/mL), and incubation was continued for 2 and 6 h. The CPE method was applied by adding 1 mL of apical and basolateral solutions to conical-bottomed glass centrifuge tubes and diluting to 7 mL with DDW. The same CPE procedure for SiO 2 was applied (see Section 4.9). Si quantifications in the precipitates and supernatants, obtained via CPE, were performed using ICP-AES (JY2000 Ultrace, HORIBA Jobin Yvon).
The permeability of Lucifer yellow, a marker of tight junction integrity, was checked as described previously [55,56]. The apical and basolateral sides of the Caco-2 monolayer and FAE models were washed with HBSS, and then the apical media were replaced with Lucifer yellow solution (300 µg/mL in HBSS). After incubation for 1 h at 37 • C, the apical and basolateral media were collected, and fluorescence was measured with a fluorescence microplate reader at excitation and emission wavelengths of 430 nm and 540 nm, respectively. The permeability (%) was calculated according to the following formula: Permeability (%) = (apical medium − blank)/(basolateral mediam − balnk) × 100

In Vivo Oral Distribution and Fate Determination of SiO 2 in Tissues
Two groups of eight-week-old female SD rats (n = 4 per group) received a single dose of 300 or 2000 mg/kg of SiO 2 particles by oral gavage, and one additional group of three rats received an equivalent volume of DW as controls. One gender (female) was used based on our previous research, showing no effect of gender on tissue distribution [36]. Tissue samples of blood, gastric fluid, kidneys, and liver were collected at time to peak concentration (300 mg/kg, 2 h; 2000 mg/kg, 10 h) after euthanasia by CO 2 [35,57]. The tissue samples were stored at −80 • C until analysis.
The kidney and liver tissues were chopped with scissors. Then, 0.1 g of the chopped tissues was transferred to conical-bottomed glass centrifuge tubes and homogenized (Sonics & Materials Inc.) on ice in 7 mL of DDW. The biofluids (0.1 g), such as blood and gastric fluid, were directly transferred to conical-bottomed glass centrifuge tubes, and the volume was adjusted to 7 mL with DDW. The same CPE procedure for food additive SiO 2 was applied (see Section 4.9). Si levels in the precipitates and supernatants, as obtained via CPE, were analyzed by ICP-AES (JY2000 Ultrace, HORIBA Jobin Yvon).

Statistical Analysis
Results were presented as means ± standard deviations. One-way analysis of variance with Tukey's test was performed using the SAS Ver.9.4 (SAS Institute Inc., Cary, NC, USA) to determine the significances of intergroup differences. Statistical significance was accepted for p values of <0.05.

Conclusions
The constituent particle sizes of precipitated SiO 2 and fumed SiO 2 particles were similar, but different levels of solubility were found depending on the hydrodynamic diameters and aggregate states in the biofluid type. TX-114-based CPE for SiO 2 was developed and optimized in intestinal cells and rat liver for its fate determination. The results show that the majority of the fates of SiO 2 inside intestinal cells were particle forms, but slowly decomposed into ions during intestinal transport through the Caco-2 monolayer and M cells, regardless of manufactured type. Most SiO 2 particles remained intact in the stomach, but ionic forms were primarily found in the liver and blood, and completely dissolved into ions in the kidneys. These findings will provide crucial information for understanding and predicting the potential toxicity of food additive SiO 2 after oral intake. Further study is required in order to elucidate the metabolic pathway involved in the degradation or decomposition of SiO 2 particles in the tissues.