Titanium Dioxide Nanoparticle Ingestion Alters Nutrient Absorption in an In Vitro Model of the Small Intestine

Abstract

Ingestion of titanium dioxide (TiO₂) nanoparticles from products such as agricultural chemicals, processed food, and nutritional supplements is nearly unavoidable. The gastrointestinal tract serves as a critical interface between the body and the external environment, and is the site of essential nutrient absorption. The goal of this study was to examine the effects of ingesting the 30 nm TiO₂ nanoparticles with an in vitro cell culture model of the small intestinal epithelium, and to determine how acute or chronic exposure to nano-TiO₂ influences intestinal barrier function, reactive oxygen species generation, proinflammatory signaling, nutrient absorption (iron, zinc, fatty acids), and brush border membrane enzyme function (intestinal alkaline phosphatase). A Caco-2/HT29-MTX cell culture model was exposed to physiologically relevant doses of TiO₂ nanoparticles for acute (four hours) or chronic (five days) time periods. Exposure to TiO₂ nanoparticles significantly decreased intestinal barrier function following chronic exposure. Reactive oxygen species (ROS) generation, proinflammatory signaling, and intestinal alkaline phosphatase activity all showed increases in response to nano-TiO₂. Iron, zinc, and fatty acid transport were significantly decreased following exposure to TiO₂ nanoparticles. This is because nanoparticle exposure induced a decrease in absorptive microvilli in the intestinal epithelial cells. Nutrient transporter protein gene expression was also altered, suggesting that cells are working to regulate the transport mechanisms disturbed by nanoparticle ingestion. Overall, these results show that intestinal epithelial cells are affected at a functional level by physiologically relevant exposure to nanoparticles commonly ingested from food.

1. Introduction

Engineered nanoparticles have become a commonly ingested material, and the effects of nanoparticles on gastrointestinal (GI) health and function are not well understood. Engineered nanoparticles (NP) exhibit specific physiochemical properties including unique optical effects, melting points, conductivity, ionization potential, electron affinity, magnetism, surface energy, reactivity, and potentially biological effects when compared to the bulk materials with the same mass dose^1–3. The unique properties of NP stem from size, and the scale-dependent changes in the ratio of surface area to volume dramatically affect NP behavior. Volume decreases with size and the proportion of atoms at the particle surface increases, meaning that the number of atoms localized at the surface exponentially increases as the size decreases. This can increase the chemical reactivity and catalytic behavior per unit mass, and can alter absorption and excretion rates in biological systems such as DNA, proteins, and cell membranes^1–4. Without a thorough understanding of the biological behavior of NP, it is impossible to predict the risks associated with NP exposure, and each new nanomaterial must be subject to health and safety assessment.

NP are increasingly used in food and food packaging applications, and companies are not required to seek regulatory approval before launching products containing nanosized ingredients made from approved bulk materials ⁵. There are currently 150–600 nano foods and 400–500 nano food packaging applications containing nanotechnology-derived food ingredients, additives, supplements, and contact materials that are commercially available⁶. The dietary consumption of NPs in developed countries is estimated at more than 10¹² particles/day, consisting mainly of titanium dioxide (TiO₂) and mixed silicates⁶. The European Food Safety Authority (EFSA), commissioning the Joint Research Centre (JRC), prepared an inventory of currently used and reasonably foreseen applications of nanomaterials in agriculture and food or feed production, and TiO₂ is the main type of engineered nanomaterial added to food⁷. TiO₂ NP exists in processed foods such as candies and chewing gums, and is primarily used as whitening agent due to its brightness, high refractive index, resistance to discoloration, and dispersion in water as a fairly stable colloid. Approximately 36% of food-grade TiO₂ (E171) are less than 100 nm in at least one dimension⁸. Personal care products, like toothpastes and some sunscreens, contain 1% to over 10% titanium by weight^8–10. Human GI exposure to nanoparticles can occur in several ways. Nano-food ingredients, additives, and supplements from food packaging and contact materials that migrate into food can be ingested. Following consumption, nano-TiO₂ materials can enter the environment by treated effluent discharged to surface waters or biosolid application to agricultural land, for example, which can contribute to human exposure via drinking water or the food chain ^2,8. Due to the prevalence of nano-TiO₂ human ingestion is nearly unavoidable, which highlights the importance of studying the effects of TiO₂ NP ingestion.

In this study the two most common types of intestinal epithelial cells, absorptive and goblet, were represented by Caco-2 and HT29-MTX cells^11,12. Caco-2 cells, which are derived from colonic epithelial adenocarcinoma cells, differentiate into a polarized, enterocyte-like epithelial barrier; express microvilli and tight junctions (TJ); and are capable of paracellular, transcellular, active, and transcytotic transport^13,14. Caco-2 cells also express all of the major amino acid, electrolyte, fatty acid, sugar/carbohydrate, iron, and zinc uptake, storage, transport, and carrier proteins^15–20. The HT29-MTX cells are a subpopulation of HT29 human colonic adenocarcinoma cells selected for resistance to methotrexate (MTX), and mimic mucus secreting goblet-like cells²¹. When seeded at a ratio of 75% Caco-2 to 25% HT29-MTX and cultured for two weeks, a mucus layer that completely covers the cell monolayer and is 2–10 μm thick is formed²². This in vitro mucus layer is ~2/3 of the thickness of the duodenal firmly adherent mucus layer in humans (15 μm)²³.

Previous NP ingestion studies with this in vitro model²⁴ showed that following exposure to 10⁹ 50 nm carboxylated polystyrene nanoparticles/cm², iron transport, which is representative of iron transfer into the bloodstream, was significantly lowered. Exposure to the same size and concentration of NP also affected iron absorption in an in vivo chicken model. Ferritin analysis, divalent metal transporter 1 (DMT1) and nuclear factor kappa-light-chain-enhancer of activated B cells (NFκB1) gene expression, and histological examination demonstrated that the changes in iron absorption were not due to changes in transport protein expression or an inflammatory response, but instead, that exposure to NP significantly increased the overall villi volume. These experiments showed that data from the in vitro system correlates well with in vivo models, and that nanoparticle consumption alters intestinal function.

The overall goal of the current study is to determine if acute or chronic exposure to TiO₂ NP, which are commonly added to food and food packaging, affects small intestinal function. The interaction of 30 nm TiO₂ NP with Caco-2/HT29-MTX monolayers was evaluated by examining molecular, functional, and structural characteristics of the cells including nutrient transporter gene expression, reactive oxygen species (ROS) generation, Fe, Zn, or fatty acid absorption, alkaline phosphatase enzyme activity, TJ functionality, and microvilli structure. Transepithelial nutrient permeability is regulated by amino acid, electrolyte, fatty acid, sugar/carbohydrate, and mineral protein transporters, and is a highly regulated process²⁵. Overall, these results show that exposure to physiologically relevant concentrations of TiO₂ NPs can have molecular, functional, and structural effects on intestinal epithelial cells.

2. Materials and Methods

30nm TiO₂ NP were purchased by US Research Nanomaterials, Inc. (Houston, TX). All culture flasks, plates, tubes, and pipette tips used for culturing cells were purchased from Corning (Corning, NY). All other chemicals, enzymes, and hormones were purchased from Sigma Aldrich (St. Louis, MO) unless otherwise stated. Glassware used in sample preparation and analysis was washed, soaked in 10% hydrochloric acid and 10% nitric acid overnight, and rinsed with 18 MΩ water to avoid iron or zinc contamination. All reagents were prepared in 18 MΩ water.

2.1. Nanoparticle Dose Calculations

In vitro doses of NP were formulated to represent potential real-life exposure. The total intestinal surface area is approximately 2×10⁶ cm^{2 26} and the daily intake of nano-TiO₂ has been estimated to be 10¹²–10¹⁴ NP per day, which is approximately 10¹¹–10¹³ particles per meal²⁷. Ingesting 10¹³ NP exposes the small intestine to 10⁶ particles/cm². If 10¹¹ or 10¹³ particles are ingested, the dose to the duodenum is approximately 10⁸ or 10¹⁰ particles/cm², respectively. The duodenum is the first section of the small intestine, the site where most nutrient absorption occurs, and has approximately 900 cm² of absorbing surface area^28,29. Supplemental Table 1 describes the low, medium, and high concentrations of 30 nm TiO₂ nanoparticles used for this study. When 100 μL of these solutions were added to 0.33 cm² cell monolayers, the concentrations were 10⁶ particles/cm² (low), 10⁸ particles/cm² (medium), and 10¹⁰ particles/cm² (high) for acute exposures. Chronic doses were three times (3X) the acute doses, representing the NP consumed in one day instead of one meal.

TiO₂ NP powder was weighed in a polystyrene weighing dish, and dispersed in sterile 18 MΩ water. Solutions were mixed uniformly in a sterile tube using a Thermolyne Mixer (Maxi Mix II, Type 37600) for 1 min, and then serially diluted to the concentrations shown in Supplemental Table 1. Nanoparticles solutions were placed in a sonicator (VWR® symphony^™ Ultrasonic Cleaners, RF-48 W) for 30 min to break down NP agglomerates. The aqueous dispersions were then measured and used for in vitro model exposures.

2.2. Nanoparticle Characterizations

The distributions of TiO₂ NP sizes and average ζ-potentials were measured with a Zetasizer Nano ZS (Malvern Instruments Inc, Southborough, MA). Measurements were performed in Malvern disposable polycarbonate folded capillary cells with gold plated beryllium-copper electrodes (DTS1070), which were rinsed with ethanol, 18 MΩ water, and sample dispersions before readings. The Refractive Index (RI) value of TiO₂ is 2.42, and water RI is 1.33. Sample viscosities refer to the viscosity of water (0.8872 cP), and the dielectric constant of water is 78.5. The samples were equilibrated in the instrument chamber for 120 s, and measured at 25°C.

The TiO₂ NPs moved randomly in dispersants via Brownian Motion, and the size (hydrodynamic diameter, d_h,z-average) of the NP determined the speed of movement. The translational diffusion coefficient of particles and the intensity fluctuations in the scattered light were expressed in hydrodynamic diameter by Dynamic Light Scattering (DLS). Polydispersity index (PdI), generated by the Malvern software, is dimensionless and refers to the range of hydrodynamic diameter distribution. If PdI>0.5, NPs are polydisperse (polymodal) distributions ³⁰; PdI<0.1, represents monodisperse distributions ³¹.

A potential exists between the surfaces of the TiO₂ NPs and the dispersants, and the charge measurement is expressed in terms of ζ-potential.

The magnitude of the ζ-potential reflects the stability of the colloidal system. As the NPs in dispersants have a large negative or positive ζ-potential, they strongly repel each other to prevent particle aggregation. A solution is normally considered stable if the ζ-potential is more positive than +30mV or more negative than −30mV ³².

2.3. Transmission Electron Microscopy

The primary size and morphology of nano-TiO₂ dispersed in 18 MΩ water, DMEM and MEM were evaluated using transmission electron microscopy (TEM) on a JEOL JEM-2100F. Samples were diluted to 1.4×10⁻⁴ mg/ml from a stock dispersion of 14 mg/ml. A drop of sample was loaded on an ultrathin 400 mesh copper TEM grid (Ted Pella, Inc) with a plastic transfer pipette. The grids were allowed to air-dry overnight before imaging.

2.4. Cell culture

The human colon carcinoma Caco-2 cell line was purchased from the American Type Culture Collection (Manassas, VA, USA) at Passage 17 and used in experiments at Passage 70–75. The HT29-MTX cell line was kindly provided by Dr. Thécla Lesuffleur of INSERM U560 in Lille, France, at Passage 11 and used in experiments at Passage 40–45. Both cell types were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco® Thermo Fisher Scientific, Waltham, MA USA) with 10% (v/v) heat inactivated fetal bovine serum (HI-FBS, Gibco® Thermo Fisher Scientific, Waltham, MA USA). In experimental studies, Caco-2 and HT29-MTX were stained with trypan blue, counted with a hemocytometer, and resuspended at ratios of 75:25 (Caco-2/HT29-MTX), at a density of 100,000 cells/cm² onto polycarbonate, 0.4-μm pore size, 0.33 cm² membrane, 24-well Transwell® inserts (Corning Life Sciences) coated with rat tail Type I collagen (BD Biosciences, San Jose, CA, USA) at 8 μg/cm² for 1 hour at room temperature ³³. Cells were grown for 16 days to form a monolayer in a 37°C incubator with 5% CO₂, and culture medium was changed every 2 days. On the day before the in vitro exposure experiments, DMEM was removed and the monolayers were rinsed with Phosphate-buffered saline (PBS), and cultured overnight in very low iron and zinc minimal essential medium (MEM, Gibco®) supplemented with 10 mM PIPES, 4 mg/L hydrocortisone, 5 mg/L insulin, 5 μg/L selenium, 34 μg/L triiodothyronine, 1% antibiotic– antimycotic solution and 20 μg/L epidermal growth factor at pH 7.0 ³⁴.

2.5. Transepithelial Electrical Resistance

Transepithelial electrical resistance (TER) of the monolayers was measured every three days after seeding in Transwell® inserts, and it was also measured before and after NP exposure with the EVOM2 and Endohm-6 chamber from World Precision Instruments (Sarasota, FL). The Endohm chamber was soaked in 70% ethanol for 15 min, 2 mL of sterile 100 mM KCl solution was added to the chamber, and the chamber was then connected to the EVOM2 which had a voltage reading between 0–20 mV. The old KCL solution was then removed and 600 μL fresh KCl was added into the Endohm chamber. A sterilized Calicell with 200μL KCL solution was inserted to the chamber and the Ohm reading was calibrated. After removing the Calicell, the Endohm chamber was rinsed with sterile 18 MΩ water three times, and then equilibrated with 2mL serum free DMEM for 15 minutes. After replacing the old medium with 600μL fresh serum free medium, the TER of every sample was measured three times at three different insert positions. The culture Transwells were taken out of incubator 5 minutes prior to measurements. Monolayer with TER values between 200 and 300 Ω×cm² were treated with nano-TiO₂ ³⁵.

2.6. Acute and Chronic Exposure to NPs

Cells were rinsed once with PBS and then 600μL of MEM was placed into the basolateral chamber of the Transwells and 100μL culture medium containing low, medium, or high doses of TiO₂ NPs was placed into the apical chamber. In acute exposures the NP solutions were made in low mineral MEM. Cells were incubated at 37°C and 5% CO on a rocking shaker (Laboratory Instrument Model RP-50, Rockville, MD) at 6 oscillations/minute for 4 hours. In chronic exposures the NP solutions were made fresh every day in DMEM+10% HI-FBS, and 100 μL was loaded into apical inserts. The basolateral medium and nano-TiO₂ solutions were changed daily for 5 days in chronic exposure treatments.

2.7. ⁵⁸Fe and ⁶⁷Zn uptake and transport

Mineral transport experiments were performed immediately after nanoparticle exposure. Stable isotope (⁵⁸Fe, ⁶⁷Zn) was added to the cells as 10 μM Fe(II)-ascorbate or 10 μM Zn(II)-ascorbate in serum free, very low mineral concentration MEM. The iron or zinc experimental medium was prepared immediately before use by combining ⁵⁸Fe or ⁶⁷Zn and 200 μL of 100 mM ascorbic acid (pH2). The molar ratio of Fe/Zn: ascorbic acid was 1:20. The mineral solutions sat at room temperature for 10 minutes, then 334 μL of 1.5 M NaCl was added, followed by 10 mL MEM. 100 μL of the mineral transport medium was added to the apical chamber immediately after nanoparticle exposure. Cells were incubated at 37°C and 5% CO₂ on a rocking shaker for 2 hours. The basolateral culture medium at bottom chamber was collected into a sterile 1.5 mL centrifuge tube, 10 μl HNO₃ was added, and the samples were stored at 4°C until ⁵⁸Fe and ⁶⁷Zn quantification with Inductively Coupled Plasma-Mass Spectroscopy (ICP-MS).

2.8. Inductively Coupled Plasma Mass Spectrometry Measurement

Stable isotope ratios in culture medium were determined via ICP-MS (ICAP Model 61E Trace Analyser; Thermo Jarrell Ash Corporation, Franklin, MA, USA). Samples were wet ashed with HNO₃ and HClO₄. Next 4 mL of 60/40 volume, double distilled 70% HNO₃/HClO₄ mixture and 0.25 mL of 40 mg/L Yttrium was added as an internal standard to each sample. Samples were incubated overnight at room temperature, then heated to 120°C in an aluminum heating block for 2 hours. If a sample did not clarify with this treatment, 0.25ml of concentrated nitric acid was added for further digestion until the temperature reached 195°C. Next, 20 mL of DI water was added, tubes were vortexed, and the solution was transferred into new glass tubes. Concentrations of ⁵⁶Fe and ⁵⁸Fe and ⁶⁵Zn and ⁶⁷Zn were determined. ⁷²Ge was added using a mixing-T just prior to the sample entering the nebulizer, serving as an internal standard. Hydrogen at 5 mL/min was used as a reaction gas to remove polyatomic interferences of Ar-O+ at mass 56. ⁵⁸Fe counts were corrected for ⁵⁸Ni using the natural abundance ratios of 2.59 (⁵⁸Ni/⁶⁰Ni). The concentrations of ⁶⁷Zn were determined similarly. The element concentrations were drift corrected and normalized using Yttrium as an internal standard. Data output from ICP-MS is expressed as mg/kg (ppm).

2.9 Fatty Acid Uptake

Caco-2/HT29-MTX cells cultured for 16 days in 96-well plates were immediately rinsed with 200μL cold medium after acute and chronic exposure. Cellular uptake studies of free fatty acids were performed using fluorescent BODIPY® 500/510 C1, C12 (4, 4-Difluoro-5-Methyl-4-Bora-3a, 4a-Diaza-s-Indacene-3-Dodecanoic Acid, ThermoFisher)^36–38. Stock solutions were prepared in 5mmol/L ethanol solution, and stored at −20°C. The analogs were added to the culture medium to obtain a final concentration of 50 μmol/L in DMEM, 50μL to each well, and labeling was performed for 10 min. After 10 min the medium was quickly replaced by analog-free medium, and the cells cultured for an additional 1 h at 37°C and 5% CO₂. Fluoresce in each well was measured using a with a fluorescent plate reader (Biotek Synergy 2, Winooski, VT, excitation/emission, 490/530).

2.10. Alkaline Phosphatase Activity Assay

TiO₂ exposure also significantly altered brush border membrane enzyme functionality. Monolayers were seeded into 24 well plates and exposed to medium or high concentrations of 30 nm TiO₂ for acute (4 hours) or chronic (5 days) time periods. Following exposure, cells were washed with 0.5 mL of PBS and then sonicated for 5 minutes at room temperature in 0.2 mL PBS. To recover the cell lysate each well was scraped into individual 1.5 mL centrifuge tubes.

The alkaline phosphatase (AP) assay detects the presence of alkaline phosphatase activity by using p-nitrophenyl phosphate (pNPP) as the substrate. The pNPP solution was made by dissolving one Tris Buffer tablet and one pNPP tablet (Sigma Aldrich, St. Louis MO) in 5mL of 18 MΩ water. AP hydrolyzes pNPP to p-nitrophenol, which turns bright yellow based on the concentration present. 25 μL of cell lysate solution from each tube was added to each well of a 96-well plate. 85 μL of the pNPP solution was then added to the wells. The plate was then incubated at room temperature for 1 hour. The absorbance was read on a plate reader at 405 nm to measure to concentration of p-nitrophenol.

The Bradford assay was used to determine the total cell protein concentration. 5 μL of cell lysate was added to a 96 well plate. 250 μL of Bradford Reagent was then added to each well. After incubating for 15 minutes at room temperature, absorbance was read at 595 nm using a plate reader. For each assay, a standard curve was created with p-nitrophenol (for the AP assay) or bovine serum albumin (BSA, for the Bradford assay) to calculate the unknown concentrations of p-nitrophenol or protein.

2.11. Scanning Electron Microscopy

Caco-2/HT29-MTX cells were seeded into 6-well plates containing sterilized cover slips coated with 8 μg/cm² rat tail Type I collagen and cultured for 16 days. The monolayers were acutely (4 hours) or chronically (5 days) exposed to TiO₂ NP in DMEM. The samples were then fixed in 4% paraformaldehyde, then rinsed by phosphate-buffered saline (PBS), dehydrated using an ethanol gradient (50, 75, 95,100 and 100%), transferred to hexamethyl disilizane (HMDS) and dried overnight (1:2 HMDS: Ethanol, 2:1 HMDS: Ethanol, 100% HMDS). Samples were then mounted, carbon coated, and viewed using a Zeiss Supra 55 Scanning Electron Microscope (Oberkochen, Germany) at 5k eV.

2.12. Gene Expression

Following NPs exposure and mineral transport studies, cells were processed for gene expression analysis. Total RNA was extracted using a Qiagen RNeasy Mini Kit. After exposure to NP, cells on Transwell inserts were released from the membrane with a cell scraper and provided lysis buffer. The cell lysate was homogenized with a QIAshredder column (Qiagen) and the RNA was isolated according to the manufacturer’s instructions. Purified RNA was eluted in water and stored at −80°C. RNA was reverse transcribed to cDNA using the SuperScript III RT-PCR kit with oligo(dT) primer (Invitrogen). Primer sequences are shown in Supplemental Table 2. The genes encoding DcytB, DMT1, HEPH, FPN1, FABP, FABP2, ZnT1, ZIP1, SGLT-1, IL-8, TNF-α, and NFκB were analyzed. Real-time polymerase chain reaction (RT-PCR) was performed in a MiniOpticon Real-Time PCR Detection System (Biorad). The 20 μL PCR mixtures consisted of 10 μL of POWER SYBR Green PCR Master Mix (Applied Biosystems, Carlsbad, CA), 7 μL of water, and 1 μL of each primer that was added to 1 μL of the cDNA samples. All reactions were performed in duplicates and under the following conditions: 95°C for 3 min, 50 cycles of 95°C for 60 s, 54°C for 15s, and 72°C for 30s. After the cycling process was completed, melting curves was determined from 65.0°C to 95.0°C with increment 0.5°C for 5s in order to ensure amplification of a single product. Expressions of the different genes were normalized to the expression of GADPH and compared with unexposed controls.

2.13. Reactive Oxygen Species Generation

Following NP exposure and mineral transport, 5 μmol/L of CellROX® Reagent (Thermo Fisher Scientific, excitation/emission 485/520) in MEM solution was added to the cells and incubated for 30 minutes at 37°C. Then medium was then removed, cells were rinsed three times with PBS, and ROS generation was measured with a fluorescent plate reader.

2.14. Immunocytochemistry

Cells exposed to CellROX® Reagent were fixed in 4% paraformaldehyde (PFA). 100 μL PFA was added to top well of plate for 1 hour. After fixing, cell monolayers were rinsed with PBS and 100 μL 0.1% solution of Triton X-100 in PBS was added to the top well, 600 μL to the bottom well for 5 minutes to permeabilize the cells. After the 5 minutes, an aliquot of 100 μL 5% Bovine Serum Albumin (BSA) in PBS solution was added to the top well and 600 μL to bottom well, and plates were rotated for 1 hour on a rocking platform. Cells were next incubated for 2 hours with 25 μL of 1:100 dilution of mouse anti-occludin primary antibody (Thermo Fisher Scientific), and then 2 hours with 25 μL 1:100 dilution of Alexa Fluor 568 goat anti-mouse secondary antibody (Thermo Fisher Scientific). The experiment was performed at room temperature, and cells were rinsed with PBS between each step. DNA was stained with 25 μL1:1000 solution of DRAQ5 (Thermo Fisher Scientific) in PBS for 30 minutes in the dark. After rinsing the cells with 18MΩ water, the membranes were removed and mounted on glass slide with ProLong Gold mounting medium (Thermo Fisher Scientific), and allowed to sit overnight in the dark. Finally, slides were sealed before imagining with a Leica TCS SP5 confocal microscope.

2.15. Statistics

All measurements were made at least 3 times for 2 separate experiments (n = 6) for each treatment. Results are expressed as mean ± standard error. Data was analyzed with the GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA). A one-way ANOVA with Tukey’s posttest or an unpaired Student’s t-test was used to compare differences between means, and data was transformed when necessary to obtain equal sample variances. Differences between means were considered significant at p < 0.05.

3. Results

3.1 Nanoparticle Characterization

High-resolution TEM analysis (Figure 1A) demonstrated that the sizes of primary TiO₂ NP ranged from 20 to 40 nm, which were similar to the sizes given by the manufacturer. As shown in Supplemental Table 1, DLS measurement of hydrodynamic sizes (d_h,z-average) of TiO₂ NP were in the range of 300–2000 nm in dispersants. The much larger hydrodynamic diameter suggests that TiO₂ NP were aggregated, which is shown in the lower magnification TEM images that displayed the coexistence of larger TiO₂ agglomerates (Figure 1B). The d_h,z-average of TiO₂ NPs in MEM were larger than in 18 MΩ water and DMEM. PdI values (Supplemental Table 1) in 18 MΩ water (0.34–0.49) and DMEM (0.42–0.54) were smaller than in MEM, which varied between 0.54 and 0.71. TiO₂ NPs in 18 MΩ water and DMEM therefore have relatively monodisperse size distributions and greater NP aggregation contributed to higher PdI values in MEM dispersant. Using lower concentration (1×10⁻³ mg/ml) of NP dispersions for DLS measurements helped to improve measurement accuracy ³⁹.

Figure 1.

Transmission electron microscopy. Titanium dioxide (TiO₂) nanoparticles in water dispersant at 1.2×10⁻³ mg/ml, 20nm scale bar (A); nano-TiO₂ in water dispersant at 2.5×10⁻² mg/ml, 50nm scale bar (B); and TiO₂ nanoparticles in DMEM dispersant at 1.2×10⁻³ mg/ml, 50 nm scale bar (C).

Environmental pH and ions may also alter the surface charge density of nano-TiO₂ ^32,40. The pH values decreased from ~6.5 in 18 MΩ Water to 3.5 in 18 MΩ Water-⁵⁸Fe(II)/⁶⁷Zn(II)-ascorbate. The NP ζ-potential changed from −18mV in 18 MΩ water to 5mV in 18 MΩ water-⁵⁸Fe(II)/⁶⁷Zn(II)-ascorbate with the consequence of NP aggregation, as the electrostatic repulsion, which prevents aggregation, was inferior to the van der Waals forces that universally attract particles together ⁴¹. The ζ-potentials of TiO₂ NP in MEM-⁵⁸Fe(II)/⁶⁷Zn(II)-ascorbate were in the range from −10mV to −13mV, which was statistically the same as ζ-potentials in MEM without Fe and Zn. This means that the TiO₂ NP did not bind a significant amount of Fe or Zn to the surface.

3.2 Tight Junction Functionality

TJs are continuous, belt-like and apical-most adhesive junctional complexes around mammalian epithelial cells ²⁵. Staining for the TJ protein occludin and measuring the transepithelial resistance (TER) are two common methods for evaluating epithelial monolayer integrity and TJ functionality. The Caco-2 and HT29-MTX monolayer confluence was quantitatively measured with TER, the magnitude of which represented the intactness of the monolayer. If nano-TiO₂ disrupted the monolayer integrity by either killing cells or breaking TJs, the TER values would decrease proportionally to the amount of damage to the epithelial layer. In acute exposures the nano-TiO₂ solution was applied once at the beginning of the experiment, and following NP exposure the TER values did not have any significant differences from unexposed controls (Figure 2A). The acute doses of TiO₂ NPs also had no effect on occludin expression (Supplemental Figure 1). In contrast, in monolayers exposed chronically to NP, after a 2-day exposure to low, medium, and high doses of nano-TiO₂ particles the TER dropped significantly below the control value (Figure 2B). A decrease in TER indicates that the doses of TiO₂ NPs were increasing the permeability of the TJs and allowing more passive diffusion between cells. An increased permeability is a sublethal toxic outcome, but interrupted barrier function of the intestinal epithelium can allow materials to diffused freely into the blood circulation from the intestinal lumen ⁴². The TER values of chronically exposed monolayers decreased to ~160Ω×cm² following a medium dose 5 day chronic exposure, but remained above 200Ω×cm² under high and low exposures and therefore did not cause epithelial cells to completely lose barrier function.

Figure 2.

Transepithelial Electrical Resistance (TER) and reactive oxygen species generation (ROS) of Caco-2/HT29-MTX monolayers following acute and chronic exposure to TiO₂ nanoparticles (NP). TER after a 4-hour exposure to low, medium, or high doses of TiO₂ NP (A). TER measurements after chronic exposure low, medium, or high doses of TiO₂ NP, medium containing NP was changed very day and TER was measured every 24 hours. * denotes significant differences according to a one-way ANOVA with Tukey’s posttest (p < 0.05, n = 12) (B). ROS production in response to of acute (C and E) and chronic (D and F) doses of TiO₂ with ⁵⁸Fe (C and D) and ⁶⁷Zn (E and F). Data shown is mean±SEM, * denotes significance according to an unpaired Student’s t-test, p < 0.05, n = 6. Confocal microscopy of the in vitro epithelium immunofluorescently stained for occludin (red), which is an integral plasma-membrane protein located at the tight junctions ⁹¹ after acute chronic exposure to an acute high TiO₂ dose with ⁶⁷Zn (H). The control (G) was exposed to ⁶⁷Zn, but not NP. DNA (blue) is stained with DRAQ5, ROS (green) is stained with CellROX® Reagent. Scale bars are 20 μm.

Staining for occludin proteins showed continuous ring appearance between adjacent cells before incubation with NPs (Supplemental Figure 1). After acute and chronic exposure, the staining for occludin proteins did not appeared discontinuous, which indicated that there were no openings of cell TJs in the monolayer. Comparatively, chronic TiO₂ NP exposure enlarged the gaps between cells or interrupted the TJ integrity (Supplemental Figure 1D, F).

3.3. Reactive Oxygen Species Formation

TiO₂ NP exposure resulted in the formation of oxidative stress in intestinal epithelial cells. The mean fluorescent signal intensities of ROS indicator significantly increased, indicating ROS is produced in response to acute or chronic TiO₂ exposure (Figure 2C–F). ROS can damage macromolecules, including DNA, oxidize polyunsaturated fatty acids in lipids and amino acids in proteins, break the structures of specific enzymes⁴³, and can also play an important role during the induction of apoptotic cell death ^44,45.

3.4. Mineral Transport and Uptake

Mineral uptake, which is the amount of nutrient taken up into the cells, and mineral transport, which is representative of nutrient absorption into the bloodstream, are sensitive to NP exposure at realistic food doses. Following acute exposure, TiO₂ NP significantly decreased iron transport at medium and high doses, significantly decreased zinc transport at a medium dose, and significantly decreased iron uptake at a low dose (Figure 3A,C,E,G). Following chronic exposure, TiO₂ NP significantly decreased iron transport at medium and high doses, significantly decreased zinc transport at a medium dose, nutrient transport, Acute NP exposure significantly decreased Fe transport at medium and high doses, and chronic medium doses significantly decreased Fe transport (Figure 3A, C). Zn transport was significantly decreased following an acute exposure to a medium TiO₂ NP dose (Figure 3B, D). Nutrient uptake was not strongly affected by NP exposure (Figure 3E,F,G,H). Low, acute doses of TiO₂ NP significantly decreased iron uptake into the Caco-2/HT29-MTX monolayers and medium, chronic doses of TiO₂ NP significantly decreased zinc uptake. High chronic doses of TiO₂ NP significantly increased zinc uptake into the monolayers.

Figure 3.

Nutrient transport and brush border membrane expression. ⁵⁸Fe transport (A and B) or uptake (E, F), ⁶⁷Zn transport (C and D) or uptake (G, H), fluorescent BODIPY® 500/510 C1, C12 lipid uptake (I and J), and alkaline phosphatase activity (K and L) in response to acute (A, C, E, G, I, K) and chronic (B, D, F, H, J, L) doses of TiO₂. Data is mean±SEM. * denotes significance according to an unpaired Student’s t-test, p < 0.05, n ≥6.

3.5. Intestinal Fatty Acid Uptake

Dietary fat supplies 30–40% of the average Western adult’s daily caloric intake. Fat (primarily long-chain triglycerides) is hydrolyzed by pancreatic lipases to monoglycerides and free fatty acids that are emulsified with bile acids to form mixed micelles⁴⁶. Fatty acids are absorbed by intestinal enterocytes by both passive diffusion and protein-facilitated transfer. One protein that may play an important role in fatty acid absorption is fatty acid–binding protein (FABP), which is expressed more heavily in the intestinal villi when compared with crypts⁴⁷. Fluorescent fatty acid analogs were taken up by Caco-2/HT29-MTX monolayers from the apical Transwell chamber. Acute exposure to TiO₂ did not significantly alter fatty acid uptake by cells, but a high, chronic dose of TiO₂ NP significantly decreased fatty acid uptake (Figure 3I, J).

3.6. Alkaline Phosphatase

TiO₂ NP exposure also significantly altered brush border membrane enzyme functionality. Figure 3K shows that acute medium and high TiO₂ NP exposure significantly increased intestinal alkaline phosphatase (IAP) activity. Figure 3L shows that chronic exposure to medium and high doses of TiO₂ also significantly increases IAP activity, and that IAP activity for chronically exposed cells is ~70% higher than acutely exposed cells.

3.7. Gene expression

Nano-TiO₂ exposure significantly decreased gene expression of the Fe transport proteins Dcytb, DMT1, HEPH, and FPN1, which were downregulated 0.44±0.03, 0.12±0.02, 0.21±0.04, and 0.62±0.05 fold, respectively, in response to chronic doses with ⁵⁸Fe (Figure 4C). DMT1 gene expression significantly decreased (0.31±0.08 fold) and Dcytb gene expression significantly increased (3.18±0.55 fold) following TiO₂ NP exposure with ⁶⁵Zn (Figure 4D). TiO₂ NPs significantly downregulated gene expressions of ZIP1 to 0.44±0.06 fold in response to chronic exposure to all three doses with ⁵⁸Fe (Figure 4C). Gene expression of ZnT1 was decreased to 0.33±0.03 fold following a medium, chronic dose of with ⁵⁸Fe (Figure 4C). Nano-TiO₂ particles significantly increased gene expressions of FABP1 and FABP2 to 2.11±0.29 and 2.62±0.39 fold respectively in response to a medium, chronic dose of TiO₂ NP with ⁵⁸Fe (Figure 4C). FABP2 gene expression was significantly upregulated 2.09±0.52 fold following chronic exposure to a low dose with ⁶⁵Zn (Figure 4D). Nutrient and fatty acid transport pathways and TJ protein components are summarized in Figure 5.

Figure 4.

Gene expression in response to of acute (A, B) and chronic (C, D) doses of TiO₂ with ⁵⁸Fe (A, C) and ⁶⁷Zn (B, D). Bars that do not share any letters are significantly different according to a one-way ANOVA with Tukey’s posttest (p < 0.05).

Figure 5.

Molecular nutrient transporters and brush border enzymes in the intestinal epithelium. Cell membrane proteins (DcytB, DMT1, HEPH, FPN1, ZnT1, ZIP1, SGLT-1, L-FABP, I-FABP) transporting nutrient through the intestinal epithelial cells. The transepithelial electrical resistance comes from junctions between epithelial cells and is composed of tight junction resistance (R_tj), intercellular space resistance (R_ics), apical membrane resistance (R_a); basolateral membrane resistance (R_b)⁹². Tight junctions are constructed by integral transmembrane proteins—claudins, occludin and junctional adhesion molecules (JAM-A, -B and -C) between the lateral intercellular space, and are fixed on actomyosin cytoskeleton filaments through scaffolding proteins, for example zonula occludens (ZO-1, ZO-2, and ZO-3) ⁹³. Intestinal alkaline phosphatase (IAP) regulates absorption of lipids across the apical intestinal epithelial membrane⁸⁸.

Proinflammatory gene expression was also analyzed. TiO₂ NPs upregulated expression of IL-8 following chronic NP exposure (3.59±0.56 fold, Figure 4 C and D). Nano-TiO₂ significantly increased TNFα and NFκB1 gene expression (5.20±1.15, 2.56±0.47 fold increase, respectively) was upregulated following chronic exposure (Figure 4 C, D).

3.8 Microvilli Structure

Figure 6 shows scanning electron microscopy images of Caco-2/HT29-MTX monolayer untreated controls (Figure 6A–D), following acute exposure to medium (Figure 6 E,F) or high (Figure 6 I, J) doses of TiO₂ NP, or following chronic exposure to medium (Figure 6 G,H) or high (Figure 6 K,L) doses of TiO₂ NP at varying magnification. Similar to results reported by Koeneman et al.⁴⁸, both acute and chronic nanoparticle exposure resulted in a decrease in absorptive cell microvilli.

Figure 6.

Scanning electron microscopy (SEM). SEM images of control (A–D), acutely exposed to medium (E, F) or high (I, J) doses of nano-TiO₂, or chronically exposed medium (G, H) or high (K, L) doses of nano-TiO₂. The microvilli on the apical surface of in vitro model was exposed to TiO₂ NPs acutely (column 1 and 2) and chronically (column 3 and 4). Scalebars are 5 μm in panels A, E, I, C, G, K (magnification 10K), and 1 μm in panels B, F, J, D, H, L (magnification 40K).

4. Discussion

TiO₂ NP powder dispersed in water or medium resulted in NP agglomeration, even following sonication. The mean NP hydrodynamic diameter of the TiO₂ NPs was 300–500 nm (Supplementary Table 1) in water and medium, which is larger than endocytic vesicles (approximately 50–150 nm) and therefore likely partitioned the TiO₂ NP to the apical surface of the monolayers⁴⁹. Efforts to measure TiO₂ transport to the basolateral chamber with ICP-MS showed no TiO₂ transport across the Caco-2/HT29-MTX monolayers (data not shown). This agrees with work by Brun et al. who showed that, in vitro, TiO₂ nanoparticles are only transported by monolayers containing M cells⁵⁰. A study by Jani et al. showed that 500 nm TiO₂ NP administered to female Sprague Dawley rats could translocate to the liver, spleen, lung and peritoneal tissues after oral gavage daily for 10 days⁵¹, although the doses used were significantly higher (12.5 mg/kg) than those used in this study. Fisichella et al. found that doses of TiO₂ up to 100 μg/mL do not affect the viability of Caco-2 cells and determined that the particles are unlikely to enter the body via oral routes, although pristine NP did induce ROS production after a 72-hour exposure ⁵².

Because of the high surface free energy, TiO₂ NPs in biological medium absorb biological components, primarily proteins and amino acids ⁵³. In particular, proteins bind to the NPs surface and construct the protein corona or biological coating⁵⁴. Doses of nano-TiO₂ particles in DMEM dispersant were coated by cell culture medium components (Figure 1C). Supplementary Table 1 shows that the average diameter of NP in DMEM+10% FBS is much smaller when compared with TiO₂ NP in serum free MEM. Previous work has shown that FBS at a typical concentration of 10% (v/v) improves TiO₂ NP dispersion, and that the NP “hard” protein corona can change based on the protein content of the solute^33,41,55. Acute exposure studies were performed in serum-free, low mineral MEM, while the chronic exposure studies required DMEM+10% HI-FBS. The chronically exposed cells may therefore have been exposed to improved doses of NP with a “hard” corona, and this may have contributed to the greater effects seen due to chronic NP exposure.

TJs reside at the apical-most region of the paracellular space and polarize the intestinal epithelium into apical and basolateral regions by maintaining asymmetry in protein and lipid composition. As a selective/semipermeable paracellular barrier, dynamic, multiprotein TJ complexes restrict the ion, most solute, antigen, microorganism, and toxin transport through the intercellular space and prevent unwanted materials from entering systemic circulation ²⁵. The effects of TiO₂ NPs on the opening of TJs (measured by reduction in TER) was evaluated in the monolayers following acute and chronic exposure to three NP doses. TiO₂ NPs did not disrupt junctional complexes following acute exposures, as TER values were stable, and TER remained statistically the same between controls and TiO₂ NP-exposed cultures (Figure 2A). Chronic TiO₂ NP exposure significantly decreased TER to 150 Ω×cm², but most barrier function of the monolayers was maintained (Figure 2B). Occludin contributes to the barrier function of the TJs and possibly to the formation of aqueous pores within TJ strands⁵⁶. Occludin staining showed that the morphology and lateral intercellular spaces in cell monolayers were not detectably changed following NP exposure, even at high doses and chronic exposures (Supplementary Figure 1).

Doses used in this study were relevant to real-life exposures from a single meal. These exposures were able to induce ROS generation (Figure 2C–H) and, following chronic dosing, significantly increased intestinal permeability (Figure 2B). Similar to ROS results in this study, De Angelis et al. found that TiO₂ NPs induced a high level of ROS formation in a dose-dependent manner in Caco-2 cells after 6 h, but did not induce IL-8 release or cytotoxicity ⁵⁷. Long-term (two months) exposure to low concentrations (1–50 μg/mL) of TiO₂ NPs did not affect viability of A549 alveolar epithelial cells, but initiated oxidative damage to DNA and resulted in accumulated intracellular NPs ⁵⁸. The reduction in TER in this study was likely related to ROS formation, as cellular oxidative stress has been shown to activate redox-responsive signaling that disrupts TJ proteins⁵⁹. Oxidative stress causes inflammation in gastrointestinal epithelial cells by producing proinflammatory cytokines such as IL-8 and TNF-α⁶⁰. TNF-α is a major mediator of inflammation that activates NF-κB ⁶¹. The proinflammatory mechanism induced by TNF-α can increase intestinal epithelial TJ permeability and contribute to intestinal inflammation as seen in Crohn’s disease, for example⁶². The TER of filter-grown Caco-2 monolayers has been shown to display concentration-dependent decreases in permeability with TNF-α treatment⁶³. The TNF-α-induced permeability correlated with NFκB activation, which in turn altered TJ protein localization and downregulated zonula occludens (ZO-1) proteins⁶³. IL-8, TNF-α, and NFκB gene expression were upregulated following chronic TiO₂ NP exposure in the current study (Figure 4C, D), which shows the potential for even low doses of NP to induce inflammatory signaling with repeated exposure.

Almost all absorption and chemical digestion occur in the small intestine, which is divided into three unequal sections: the duodenum, the jejunum, and the ileum⁶⁴. The duodenum is the first section of the small intestine and the site where most iron absorption occurs⁶⁷. Intestinal iron absorption is regulated with body iron requirements and decreases when body iron stores increase^65,66. Most dietary nonheme iron exists in form of ferric iron (Fe³⁺) complexes in physiological conditions, whereas intestinal epithelium efficiently absorbs ferrous iron Fe²⁺ form, which is very unstable and quickly oxidized to ferric iron. Specialized transmembrane electron transport systems have evolved to reduce ferric Fe³⁺ form to ferrous iron Fe²⁺ form at the brush border extracellular surface ⁶⁷. Dcytb (duodenal cytochrome b) on the apical membrane plays an important role in ascorbate-dependent reduction of non-heme iron in the gut prior to uptake by the ferrous-iron transporter, DMT1^67–69. Cellular iron is stored in the intracellular iron storage protein, ferritin, in order to both maintain large amounts of iron in a compact and bioavailable form in solution and protect the cell from oxidative damage ⁷⁰. Ferrous iron that reaches the basolateral membrane is oxidized back to ferric iron by the trans-membrane ferroxidase, hephaestin (HEPH), and then transported into the circulation by the iron export protein, ferroportin (FPN1), which co-localizes with HEPH in the basolateral membrane ^65,67,68. Dcytb, DMT1, HEPH, and FPN1 were downregulated in response to chronic medium and high doses of TiO₂ with ⁵⁸Fe (Figure 4C). The downregulation could be in response to the high iron content in the transport study medium or, for Dcytb and DMT1, because of the decrease in microvilli, which is where these proteins are expressed⁷¹. Iron transport was significantly affected by medium and high doses for both acute and chronic exposures (Figure 3A, B). Iron uptake, however, was only significantly reduced at low, acute doses of TiO₂ NP (Figure 3E). The alterations in nutrient transport were due to structural changes in the cell microvilli. TiO₂ exposure decreased the number of microvilli, which decreased the surface area available for nutrient transport (Figure 6). Koeneman et al.⁴⁸ has also reported that TiO₂ NPs (10 μg/mL) altered the microvillar organization on the apical surface of the epithelium following a 10-day treatment with TiO₂ NP⁴⁸. Iron uptake was less affected because in the Caco-2 model the iron transport phase is rate limiting⁷², meaning that iron uptake occurs much faster than transport and is therefore less sensitive to cell alterations.

The jejunum has the highest rate of zinc absorption^73,74. Zinc cannot cross biological membranes by simple diffusion because it is a highly-charged, hydrophilic ion and specialized mechanisms exist for its cellular uptake and release. Two families of mammalian zinc transporters regulate zinc levels in cytoplasm and intracellular. ZnT proteins, as members of the action diffusion facilitator family, move zinc across the lipid bilayer of the plasma membrane from the cytoplasm to the lumen of organelles or the extracellular space, and ZIP family is responsible for increasing intracellular zinc levels by either transporting the metal from the extracellular space or organelle lumen into the cytoplasm ^75–79. ZnT1 was the first zinc transporter cloned at the basolateral membrane of intestinal epithelial cells in the upper portion of the villus in the duodenum and jejunum. ZnT1 is regulated according to the amount of zinc in the diet, and mediates delivery of dietary zinc to the circulation ^75,78–80. The Zip1 protein is located in proximity to the apical microvilli in differentiated Caco-2 cells and it is responsible for intracellular Zn accumulation and transport⁷⁶. As an intracellular sensor to regulate zinc homoeostasis, over expression of Zip1 results in upregulated uptake of Zn ^76,81. Zip1 and ZnT1 gene expression was significantly decreased only following chronic TiO₂ NP exposure in medium with Fe added (Figure 4C). This is consistent with studies by Yasuno et al. and Dufner-Beattie et al. that have shown that zinc transport is not correlated with zinc transporter mRNA expression in rats, and that ZIP gene expression is not regulated by dietary zinc^82,83. Zinc transport across the cell monolayers was significantly decreased only by acute and chronic exposure to medium doses of TiO₂ NP (Figure 3C, D). Zinc uptake was significantly decreased by chronic medium doses of TiO₂ NP, and significantly increased by chronic high doses of TiO₂ NP (Figure 3H). The increase in zinc uptake at high TiO2 NP doses for chronic exposures was likely due to the increase in ROS formation (Figure 2F) and resulting inflammation. Proinflammatory conditions have been shown to increase Zn absorption^84,85, which allowed the cells to compensate for the loss of microvilli due to TiO₂ exposure

Cytoplasmic fatty acid binding proteins (FABPs) mediate fatty acid uptake and trafficking in small intestinal enterocytes. Two kinds of FABPs are largely present in the absorptive intestinal cells; liver-type FABP (LFABP; FABP1), and intestinal FABP (IFABP; FABP2)⁸⁶. FABP1 and FABP2 were upregulated by medium dose chronic TiO₂ NP exposure with Fe (Figure 4C), and FABP2 was upregulated by low dose chronic TiO₂ NP exposure with Zn (Figure 4D). A high, chronic dose of TiO₂ NP resulted in a significant decrease in fatty acid (Figure 3J), likely because of the decreased surface area available for absorption due to damaged microvilli.

TiO₂ exposure also significantly altered brush border membrane enzyme functionality. Alkaline phosphatases are enzymes present in all tissues. In the small intestinal epithelium, IAP is responsible for the proper breakdown and adsorption of nutrients⁸⁷. Recent work has shown that IAP regulates the absorption of lipids such as fatty acids across the apical intestinal epithelial membrane. Inhibition of IAP in mice on a high fat diet showed increased lipid transcytosis, visceral fat accumulation, and hepatic steatosis⁸⁸. Many studies have shown evidence that IAP is a gut mucosal defense factor. For example, IAP can modulate the pH of the intestinal epithelium surface in the duodenum, helping to prevent acid-induced duodenal injury by regulating bicarbonate secretion⁸⁹. Following gut injury IAP is upregulated and prevents bacterial lipopolysaccharides from crossing the gut mucosal barrier⁹⁰. In this study both medium and high doses of TiO₂ for acute and chronic exposures resulted in significantly increased IAP activity (Figure 4K, L).

5. Conclusions

TiO₂ exposure significantly affected Fe and Zn nutrient transport, fatty acid uptake, IAP activity, and tight junction functionality. Gene expression and ROS formation analysis showed NP exposure changed the expression levels of nutrient transport proteins and induced proinflammatory signaling. TiO₂ NP exposure decreased the number of intestinal microvilli, which decreased the surface area available for nutrient absorption. Overall, the results from this study indicate that intestinal epithelial cells are affected at a functional level by physiologically relevant exposure to TiO₂ NPs commonly ingested from food.

Highlights.

• Ingestion of TiO₂ nanoparticles is nearly unavoidable.

• Intestinal epithelial cells were acutely and chronically exposed to 30 nm TiO₂.

• TiO₂ exposure affected nutrient absorption and enzyme activity.

• Nanoparticle exposure induced a significant decrease in absorptive microvilli.

• Intestinal cells are affected functionally by commonly ingested nanoparticles.