Toxicity of orally administered food‐grade titanium dioxide nanoparticles

This year, France banned the application of titanium dioxide nanoparticles as a food additive (hereafter, E171) based on the insufficient oral toxicity data. Here, we investigated the subchronic toxic responses of E171 (0, 10, 100, and 1,000 mg/kg) and tried to elucidate the possible toxic mechanism using AGS cells, a human stomach epithelial cell line. There were no dose‐related changes in the Organisation for Economic Cooperation and Development test guideline‐related endpoints. Meanwhile, E171 deeply penetrated cells lining the stomach tissues of rats, and the IgM and granulocyte‐macrophage colony‐stimulating factor levels were significantly lower in the blood from rats exposed to E171 compared with the control. The colonic antioxidant protein level decreased with increasing Ti accumulation. Additionally, after 24‐h exposure, E171 located in the perinuclear region of AGS cells and affected expression of endoplasmic reticulum stress‐related proteins. However, cell death was not observed up to the used maximum concentration. A gene profile analysis also showed that immune response‐related microRNAs were most strongly affected by E171 exposure. Collectively, we concluded that the NOAEL of E171 for 90 days repeated oral administration is between 100 and 1,000 mg/kg for both male and female rats. Additionally, further study is needed to clarify the possible carcinogenesis following the chronic accumulation in the colon.


| INTRODUCTION
Owing to its beneficial properties (i.e., bright white, poorly water soluble, and inactive), titanium dioxide (TiO 2 ) particles have been used to manufacture a wide range of products including cosmetics, skin care products, paints, and building materials as a pigment (called as Pigment White 6 or CI77891) for about 100 years (European Union, 1994;Weir, Westorhoff, Fabricius, Hristovski, & von Goetz, 2012). Market survey data also suggested that paints, varnishes, paper, and plastics account for 80% of the global TiO 2 consumption (Brandessence Market Research and Consulting Pvt. Ltd, 2020). In addition, the Joint Food and Agriculture Organization TiO 2 (called as E171 in EU) to food products without the requirement of ingredient label disclosure (US FDA, 2020). Thus, TiO 2 particles have been extensively used in the production of various foodstuffs such as chocolates, candies, chewing gum, ice cream, donuts, confectionery, and beverages, and they have been also incorporated into toothpaste and pharmaceutical products (Dorier et al., 2019).
Meanwhile, the application of TiO 2 particles in the food industry has been a major public concern for a long period of time regarding the direct exposure, especially in children who tend to like eating sweets, and this concern has been amplified with the growth of the nanoindustry. In addition, the International Agency for Research on Cancer (IARC) classified inhaled TiO 2 nanoparticles (NPs) as a Group 2B (potential human carcinogen), particularly for workers (IARC, 2006). France, which worked as a lead sponsor in a sponsorship program of the Organisation for Economic Cooperation and Development (OECD) working party for manufactured nanomaterials, also reviewed the health effects of food-grade TiO 2 NPs (hereafter, E171) in 2017 (The French Agency for Food, Environmental and Occupational Health & Safety, 2020), and France's National Institute for Agricultural Research (INRA) and its colleagues reported that orally dosed E171 may cross the intestinal wall and translocate to other organs (or tissues) via the bloodstream (Bettini et al., 2017). They demonstrated that the adverse health effects of E171 are attributable to the absorption of the nanoscale but not the microscale particle fraction and that E171 disturbs the homeostasis of the immune system. Similarly, about 36% of the particles obtained from a single source of E171 were less than 100 nm in dimension, indicating a potentially significant dietary exposure to TiO 2 NPs (Tassinari et al., 2014). In addition, chronic intake of foodstuffs containing E171 initiated and promoted the expansion of preneoplastic lesions in the colon and induced a slight inflammatory microenvironment in the mucosa (Dorier et al., 2019). Therefore, France banned the use of TiO 2 as a food additive in 2020 based on inadequate evidence to guarantee its safety to humans. In this study, our objective was to identify the possible adverse health effects of ingested E171. We also investigated the potential toxic mode of action of E171 in AGS cells, a human stomach epithelial cell line.

| Animals and housing
Five-week-old male and female specific pathogen-free Sprague Dawley rats (40 rats per sex and five rats per cage) were obtained from ORIENT BIO Inc. (Seongnam-si, Gyeongi-do, Korea) and maintained in stainless wire cages (255 × 465 × 200 mm) under controlled environment (12/12 h light/dark cycle with 150-300 lx, temperature of 23 ± 3 C, relative humidity of 50 ± 10%, and air ventilation of 10-20 times/h). Food (PMI Nutrition International, St. Louis, MO, USA) was ad libitum given with tap water, and wood chews were provided for animal welfare. The rats were randomly assigned to one of four groups (0, 10, 100, or 1,000 mg/kg) via Pristima v. 7.4 (Xybion Medical Systems Corporation, Lawrenceville, NJ, USA). The dose levels were selected based on a previous 4-week dose range finding study and 13-week repeated study of a different TiO 2 material (P25) in rats (Heo et al., 2020). According to OECD test guideline no. 408, E171 (10 rats/sex/dose) was administered daily by oral gavage for 90 days. The control group received equal volumes of DW, and the rats were euthanized using isoflurane on necropsy. The experimental design was reviewed and assessed by the Association for the Assessment and Accreditation of Laboratory Animal Care International and the Institutional Animal Care and Use Committee of the Korea Institute of Toxicology.

| Clinical observations and blood analysis
The health status of all rats was observed daily during the study period. The type, time of occurrence, and severity of abnormal symptoms were recorded with the Pristima v. 7.4 system. The rats were weighed upon arrival, before randomization, weekly during pretreatment, before dosing during treatment, and before necropsy. Food consumption was recorded weekly and calculated as g/rat/day. Urinalysis was performed during the treatment period on all surviving animals. The urine was collected for 17 h before necropsy and its volume, specific gravity (SG), color, pH, and protein (PRO), ketone body (KET), occult blood (BLD), glucose (GLU), bilirubin (BIL), nitrite (NIT), and urobilinogen (

| Macroscopic and microscopic findings
Forty-two tissues were taken from all rats at necropsy. The tissues were weighed, and relative organ weights were calculated using the body weights measured at necropsy. The eyes with optic nerves were fixed in Davidson's fixative solution, and all other tissues were preserved in 10% (v/v) neutral buffered formalin. The tissues were embedded in paraffin, sectioned, stained with hematoxylin and eosin, and examined under microscope (Olympus BX53, Olympus America, USA).

| Accumulation of E171
AGS cells (no. 21739), a human stomach epithelial cell line, was purchased from Korea Cell Line Bank and maintained in 37 C incubator with humidified atmosphere of 5% CO 2 using RPMI1640 media containing 10% fetal bovine and 1% penicillin/streptomycin. The colons obtained from rats administered the maximum dose were

| Trace element determination
The tissues (colons, spleens, and kidneys) were digested in 70% (v/v) nitric acid solution using a microwave digestion system (Milestone, Sorisole, Italy). Finally, concentrations of trace elements in tissues were measured by inductively coupled plasma mass spectrometry (ICP-MS) at the Korean Basic Science Institute (Seoul, Korea).
Lastly, the images were captured with an inverted phase-contrast fluorescence microscope (IX51, Olympus, Tokyo, Japan).

| Enzyme-linked immunosorbent assay
Serum was made by centrifugation (3,000 rpm, 15 min) of whole blood, and the blood level of immunoglobulin (Ig)A, IgE, IgG (KOMA Biotech, Seoul, Korea), IgM, and granulocyte-macrophage colony-stimulating factor (GM-CSF) (eBioscience, San Diego, CA, USA) was measured according to manufacturer's instruction. Finally, absorbance (450 nm) was measured using a 96-well microplate reader (PerkinElmer, Waltham, MA, USA), and the concentration in each sample was calculated based on the corresponding standard curve.

| Gene profiling analysis
The AGS cells were incubated with or without E171 (40 μg/ml) for 24 h to evaluate the effects of E171 on the gene profile. Briefly, mRNA was prepared using TRIzol solution (Invitrogen) and the microarray analysis was conducted at Macrogen (Seoul, Korea) using an Affymetrix ® human 2.0ST gene chip (Illumina, San Diego, CA, USA).
The data were summarized and normalized by the robust multiaverage (RMA) method in Affymetrix ® Power Tools. The results of the gene-level RMA analysis were exported to perform a differentially expressed gene (DEG) analysis. For each DEG set, a hierarchical cluster analysis was conducted using complete linkage and Euclidean distance. Gene enrichment and functional annotation analyses of significant probe lists were performed via GO (http://geneontology. org) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (www. genome.jp/kegg/). All data analyses and DEG visualizations were performed in R v. 3.3.3 (www.r-project.org).

| Statistical analysis
Statistical significance of data were analyzed using Prisitima v. 7.4.
Multiple comparison methods, including Bartlett's test, one-way analysis of variance followed by post hoc Dunnett's test, Kruskal-Wallis (H) test followed by post hoc Dunn's rank sum test, and Fisher's exact test, were used to compare the data of the various treatment groups with those of the control. A p < 0.05 level was considered to be significant.

| Characterization of E171
E171 had diameter of about 150 nm and a single anatase phase (PDF card no. 21-1272) in DW (Figure 1), and the diameter and shape were not notably altered either in AGF or cell culture media. In addition, in the same condition, E171 had highly crystalline structures with a 0.35-mm lattice spacing corresponding to the (101) plane of the anatase phase ( Figure 1A). Meanwhile, the average hydrodynamic T A B L E 1 Changes in food consumption after treatment with E171 during the test period (unit: g)  Note. Data are represented as mean ± SD. *Significant differences from control group (p < 0.05).
T A B L E 6 Summary of microscopic findings at control and 1,000 mg/kg/day dosed rat for 90 days diameters ( Figure 1B) and the average surface charge values were quite different in the DW, AGF, and cell culture media ( Figure 1C).

| Clinical observations
There were no treatment-related deaths. Abnormal clinical signs such as fur loss, scabs, scars, scratch wounds, swelling, and discolored urine were observed in the E171-dosed rats, but these effects were not dose dependent. Whereas body weight did not significantly differ among groups (Table S1), a significant increase in food consumption was observed only in male rats dosed at 1,000 mg/kg (Table 1).

| Blood analysis
As shown in Table 2, the proportion of lymphocytes in WBC slightly decreased in male rats administered the maximum E171 dose. There were no dose-related changes in any hematological (Table 2) and biochemical (Table 3) parameters, absolute organ weight (Table 4), and organ weight relative to body weight (Table 5).

| E171 accumulation in the stomach wall
There are no remarkable changes in macroscopic finding (data not shown), and dose-related histopathological lesions were also not detected (Table 6 and Figure S1). Meanwhile, we found E171 accumulation in the stomach wall of rats administered with 1,000 mg/kg of E171 for 90 days (Figures 2 and S2).

| The main routes of excretion
TiO 2 has low water solubility but can be dissolved in the acidic conditions such as gastric juice and lysozyme. NPs can also penetrate into cells in the bloodstream, and the damaged red blood cells are eliminated via the spleen. In the current study, we measured Ti concentrations in the colons, kidneys, and spleens harvested from all rats at necropsy. Importantly, the Ti concentration clearly increased only in the colons of both sexes administered with 1,000 mg/kg of E171 compared with the control, indicating that the colon is the main excretion route (Table 7). In addition, TEM images revealed that E171 accumulates in the cytosol and nuclei of various cells comprising the colon tissue and forms lamella-like structures (Figure 3). Ti accumulation can also affect the distribution of elements cross-binding with it or participating in the antioxidant response, and we found here that the colonic zinc (Zn) concentration increases in female rats exposed to E171 compared with the control. In addition, SOD proteins play a central role in inhibiting xenobiotic-induced oxidative damage and subsequent apoptosis (Fukai & Ushio-Fukai, 2011); thus, we assessed the effects of E171 on expression of SOD-1, SOD-2, and cytochrome C protein in the colonic tissues of rats in control and the maximum-dosed group.
Interestingly, the expression of SOD-1 and SOD-2 proteins was clearly downregulated in the colonic tissues of both sexes and female rats, respectively. However, that of cytochrome C protein was not significantly different between groups (Figure 4).

| Effects of E171 on systemic immune response
Given that the proportion of lymphocytes in WBC decreased in rats administered with the maximum dose compared with control, we measured the GM-CSF and immunoglobulin concentrations in the blood. Importantly, the levels of GM-CSF (female) and IgM (both sexes) significantly reduced in rats administered with 1,000 mg/kg of  (Figure 7). In addition, the expression levels of various microRNAs and unknown genes were markedly altered in the E171-exposed cells relative to the control (Table 8).
More interestingly, cell death was not observed even at the highest concentration tested (40 μg/ml).

| DISCUSSION
The potential risks of nanoscale particles on the environment and human health have been continuously issued along with the great importance in future industry, and thus, nanotechnology has often been compared with a double-edged sword (Kashanian, Habibi-Rezaei, Bagherpour, Seyedarabi, & Moosavi-Movahedi, 2017;Patni & Bhatia, 2008;Solaiman et al., 2019). Meanwhile, all substance is potentially harmful to human health when it accumulates at sufficiently high concentrations, which can disturb biological homeostasis, as was first expressed by Paracelsus, a Swiss physician. Furthermore, nanoscale particles have unique physicochemical properties that differ from those of the bulk forms of the same materials. Therefore, although available information for the microscale particles is enough, biostability, interactions with biological systems, biodistributions, health effects, and the possible toxic mechanism should be carefully re-evaluated for the nanoscale particles (Fadeel & Garcia-Bennett,-2010). Here, we found that E171 is insoluble in DW, a vehicle used for dosing, and the physicochemical properties were not substantially altered in AGF or cell culture media. In addition, when orally dosed 10, 100, and 1,000 mg/kg to rats for 90 days in accordance with an OECD test guideline (OECD, 2018), any significant tissue damage was not found even in the maximum dose. Meanwhile, contrary to the 0.1 F I G U R E 6 Transmission electron microscope images of AGS cells exposed to E171 (40 μg/ml) for 24 h.  (Cui et al., 2012). A 100-day repeated dosed E171 also promoted microinflammation and initiated preneoplastic lesions in the colons, and it altered the expression of genes involved in innate and adaptive immune response and oxidative stress (Blevins et al., 2019). In the current study, Ti concentrations clearly increased in the colonic tissues, altering the tissue level of the antioxidant protein (SOD-1 and SOD-2). We also found that the proportion of lymphocytes in WBC was clearly lower in rats exposed to E171 compared with the control and that the GM-CSF and IgM levels notably reduced in the blood of rats in the same group. GM-CSF regulates myelopoiesis in physiological steady state and modulates immunity under inflammatory conditions including autoimmune disease (Becher, Tugues, & Greter, 2016;Bhattacharya et al., 2015).
In addition, IgM is the antibody that is produced mainly in the spleen in response to initial antigen exposures. Hence, we hypothesize that part of E171 may be dissolved under the acidic conditions of the stomach and that the rest may form aggregates with diet or other particles. Also, it may affect antioxidant capacity being resorbed during the stay in the colon (Lomer, Thompson, & Powell, 2002;Park, Yoon, Choi, Yi, & Park, 2009;Proquin et al., 2017). In addition, we hypothesize that chronic E171 intake might impair host's defense function against foreign bodies.
The reassessment of E171 as a putative carcinogen may be crucial in the determination and establishment of its safety (Armand et al., 2016;Falck et al., 2009;Kang, Kim, Lee, & Chung, 2008;Warheit, Brown, & Donner, 2015). Accumulated clinical and empirical evidence has demonstrated that TiO 2 NPs induce ER stress by promoting oxidative stress. In addition, chronic ER stress may be associated with tumor development by triggering inflammatory responses (Lin, Jiang, Chen, Zhao, & Wei, 2019), and it could also be involved in immunosuppression (Salminen, Kaarniranta, & Kauppinen, 2020). In this study, we found that E171 did not affect the expression of SOD-1 and SOD-2 proteins in AGS cells. In addition, E171 penetrated the cells comprising the colon tissue and localized to the perinuclear regions of AGS cells. Meanwhile, dead cells were not observed even in cells treated at the maximum concentration (40 μg/ml).
Furthermore, the expression of ER stress-related proteins increased in E171-treated cells compared with control, and microarray analysis demonstrated that expression of several microRNAs is the most affected following exposure to E171. In particular, the expression of microRNA 3908 was the most downregulated. More interestingly, many of the affected genes were those whose function is unknown.
MicroRNAs are noncoding RNAs that are involved in posttranscriptional regulation by affecting both the stability and translation of mRNA. Previous studies have suggested that microRNA 3908 inhibits cancer progression by inducing apoptosis (Liu, Chen, & Zhang, 2017) and that bitter-taste receptor genes (such as TAS2R50) can be involved in progression of colorectal neoplasia (Schembre, Cheng, Wilkens, Albright, & Marchand, 2013). In addition, SOD-2 transforms toxic mitochondrial superoxide into nontoxic products, inhibiting apoptosis, and ferritin heavy chain also protects cells against free radical accumulation (Mumbauer, Pascual, Kolotuev, & Hamaratoglu, 2019). Furthermore, ER stress can initiate pyroptosis and inflammasome formation (Lebeaupin et al., 2015). Therefore, further study is required to elucidate the adverse health effects following chronic accumulation in the stomach and colons (Proquin et al., 2017).
In conclusion, we suggest that NOAEL of 90-day repeated orally dosed E171 is between 100 and 1,000 mg/kg for both sexes of rats and that further study is needed to clarify the possible carcinogenesis following the chronic accumulation in the colons.