TiO₂ Nanoparticles and Commensal Bacteria Alter Mucus Layer Thickness and Composition in a Gastrointestinal Tract Model

Abstract

Nanoparticles (NPs) are used in food packaging and processing and have become an integral part of many commonly ingested products. There are few studies that have focused on the interaction between ingested NPs, gut function, the mucus layer, and the gut microbiota. In this work, an in vitro model of gastrointestinal (GI) tract is used to determine whether, and how, the mucus layer is affected by the presence of Gram-positive, commensal Lactobacillus rhamnosus; Gram-negative, opportunistic Escherichia coli; and/or exposure to physiologically relevant doses of pristine or digested TiO₂ NPs. Caco-2/HT29-MTX-E12 cell monolayers are exposed to physiological concentrations of bacteria (expressing fluorescent proteins) and/or TiO₂ nanoparticles for a period of 4 h. To determine mucus thickness and composition, cell monolayers are stained with alcian blue, periodic acid schiff, or an Alexa Fluor 488 conjugate of wheat germ agglutinin. It is found that the presence of both bacteria and nanoparticles alter the thickness and composition of the mucus layer. Changes in the distribution or pattern of mucins can be indicative of pathological conditions, and this model provides a platform for understanding how bacteria and/or NPs may interact with and alter the mucus layer.

1. Introduction

Nanoparticles (NPs), deriving from the Greek word “nano” meaning extremely small, are defined as clusters of atoms ranging from 1 to 100 nm that serve numerous purposes in consumer products. Many types of food contain nanoparticles, whether they are organic, inorganic, intentionally or unintentionally added during the manufacturing process. Nanoparticles, specifically metal oxide nanoparticles, are often added to food to improve or enhance food texture, flavor, and/or nutrient stability.^[1] As nanomaterials are incorporated in storage containers and food packaging, NP-food cross contamination is expected.^[2] Human ingestion of nanoparticles through food, food packaging, and other consumer products is therefore nearly unavoidable.^[3] Titanium dioxide (TiO₂) NPs are widely used in consumer food products due to whitening pigment and opacifying characteristics.^[4,5] The highest concentration of TiO₂ NP fractions are found in the coating constituent of sweets, candies, and chewing gums.^[4,6] Previous studies have estimated that daily exposure to TiO₂ NPs via ingestion ranges between 1 and 2 mg TiO₂ kg⁻¹ b.w. in U.S. children under 10 years of age, and 0.2–0.7 mg TiO₂ kg⁻¹ b.w. in U.S. adults.^[4,7,8] However, these estimations have been refined recently and the range of exposure was reported to be between 5.5 and 10.4 mg kg⁻¹ b.w. in children depending on the exposure scenario.^[8] Referred to as food grade nanoparticle in Europe, E171 can contain ≈43% of TiO₂ NPs, and has been found to be absorbed in the gastrointestinal (GI) tract at low levels.^[6,8,9]

Models of the GI tract have been used as a means to understand the potential toxic or functional effects of exposure to NPs as a food additive.^[1,5,6,10] The small size of NPs allows them to cross cellular barriers and can lead to free radical formation in tissues and oxidative damage to the cells that line the GI tract or the mucus layer.^[11,12] The GI tract, particularly the small intestine, plays a major role in ingestion, digestion, and absorption of nutrients, and the elimination of non-digested foods and microbial products.^[1,13–15] The mucosal surface of the GI tract is a complex environment of epithelial cells and immune cells, and is colonized by microbiota.^[16] Covered by mucin-secreting goblet cells and other substances that make up the GI mucus layer, the intestinal epithelium serves as the first line of defense against attachment and invasion of microorganisms, including commensals and pathogens.^[13,16,17] Commensal bacteria normally fail to reach intestinal epithelial cells since they are trapped in the mucus layer and often discarded through peristaltic movement.^[13] The major goblet cell mucins in the small and large intestine are MUC2 proteins, which are gel-forming secretory mucins localized in clusters.^[18–20] Goblet cells synthesize and secrete bioactive molecules such as secretory and membrane associated mucins, trefoil peptides, resistin-like molecule beta (RELMβ), and Fc-γ binding protein (Fcgbp); Paneth cells secrete antimicrobial peptides (beta defensin and lysozymes); and enterocytes secrete secretory IgA; which are all components of mucus.^[13] Epithelial mucins are further classified into neutral and acidic grouped sialo- and sulfomucins depending on the sialic acid or ester sulfate radicals with hexosamine.^[21] Neutral mucins are neutral glycoproteins with immunologically active carbohydrate and protein containing substances, whereas acidic mucins are composed of sulfo/sialomucins.^[21–23] A few studies have been performed on the intestinal mucus lining with the intent to better understand the role it plays as a protective barrier, and to determine how it is affected by bacteria present in the gut or by ingested food additives.^[19,24,25]

The interaction between the human intestine and its microbiota is mediated by an intervening mucus layer that is secreted by the intestinal epithelial cells along the apical surface.^[26] Similar to the gut microbiota, mucus thickness also varies throughout the length of the intestine. Studies have identified that the GI mucus forms a single, easily removable layer in the small intestine and a double layer firmly attached to the colon epithelium.^[20,24,27] These mucus layers are organized around the MUC2 mucin to form a large, net-like polymer that is permeable to gases, water and nutrients, but impermeable to most microorganisms.^[19] Mucus lubricates the GI epithelium to facilitate protection from excessive mechanical stress by dietary components, corrosive action of gastric acid, and from proteolysis by digestive enzymes.^[28] It also prevents infection by binding bacteria or viruses through specific interactions with carbohydrate chains. Despite being smaller than the average mucus mesh pore size, the majority of particles administered orally are not able to be transported through the GI mucus and/or they are likely to be trapped by mucus via adhesive forces (hydrophobic interactions) or are rapidly eliminated.^[29,30] It has previously been found that Norwalk (38 nm) and human papilloma virus (HPV; 55 nm) are able to penetrate the mucus layers at a rate identical to water.^[30] Additionally, it has been demonstrated that TiO₂ NPs were able to permeate the mucus layer into the underlying tissue, regardless of their size and hydrophilicity/hydrophobicity.^[19] A number of in vivo studies further suggest that nanoparticles are able to enter the epithelium through M-cells, since no mucus is secreted in the region surrounding these cells and particulate uptake by the mammalian GI tract mainly occurs via Peyer’s patches.^[31] These previous findings strongly establish that NPs are able to diffuse through low viscosity pores in the mucin fiber matrix, assuming they do not adhere to the mucus, because small molecules are unable to form polyvalent adhesive bonds with the mucus gel.^[29–31]

The human GI tract contains trillions of microorganisms, including bacteria, viruses, and fungi.^[32,33] These microorganisms form the intestinal microbiota, which plays a crucial role in GI tract physiologic functions such as gastric secretion, gut motility, mucosal permeability and barrier function, visceral sensitivity, and mucosal blood flow.^[34,35] The gut microbiota aids in food digestion, and plays a critical role in the absorption of nutrients from some indigestible foods.^[36] Additionally, bacteria present in the gut microbiota inhabit exposed mucosal surfaces and have been shown to inhibit pathogenic bacteria colonization by producing antimicrobial compounds, which also protect the intestinal epithelium from the harmful effects of pathogens.^[32,37] However, epithelial invasion by pathogens or certain commensal bacteria can occur and disturb the barrier’s function, thus increasing gut permeability.^[33] This primarily occurs following bacterial production of specific mucin-degrading enzymes that release monosaccharides attached to the mucin glycoproteins.^[18] Previous studies have shown that commensal bacteria such as Lactobacillus sp. were likely to increase the production of MUC2 and MUC3 mucins and hinder the attachment of enteropathogenic Escherichia coli, indicating that enhanced mucus layers together with the presence of Lactobacillus sp. provides protection against invasion by pathogens.^[13,17] One example is L. rhamnosus GG, a Gram-positive bacterium that behaves as a probiotic to promote better health, leading to an increase in the production of mucins to stimulate intestinal epithelial cell barrier response.^[32]

Due to the symbiotic relationship between bacteria and the human intestinal cells, it is important to assess the impact of NPs on the gut microbiota. Commensal bacteria have a major impact on immunophysiological functions by inducing protective responses that prevent colonization and invasion of pathogens.^[38] Previous in vitro studies, performed to provide insights into mechanisms that may be overshadowed by in vivo systems, have shown that toxicity effects in both Gram-positive and Gram-negative bacterial species can be dependent of the photocatalytic activity of TiO₂ NPs.^[39] The photocatalyticmediated toxicity of TiO₂ NPs can also inhibit the growth of commensal bacteria in an in vitro insect model in the absence of light, but not in vivo.^[37] In vitro microbiome models and in vivo studies have shown that exposure to food-grade TiO₂ NPs (E171, 25 nm) can induce compositional shifts in microbiota. Although the shift in abundance of bacterial species did not display measurable physiological changes in the GI tract in one study, a separate study demonstrated that TiO₂ did promote biofilm formation of commensal bacteria, resulting in an alteration in the release of bacterial metabolites of mucin gene expression by the gut epithelial cells.^[40,41] In addition, it has also been found that silver NPs have the potential to disrupt the intestinal microbiome.^[19] Minimal and non-statistically significant changes in GI microbiota following exposure to TiO₂ have also been observed in mice.^[39] Studies have reported that exposing E. coli to TiO₂ NPs resulted in an induction of oxidative stress, of changes on the bacterial wall morphology, and changes in the membrane permeability and osmotic balance regulation.^[42] While it is unclear what factors led to these in vivo–in vitro discrepancies, these studies highlight how the presence of ingested nanoparticles can amplify or negate changes in the microbiota.

The primary objective of this study is to understand how exposure to TiO₂ NPs at physiologically relevant doses and the presence of Gram-positive L. rhamnosus (beneficial) or the Gram-negative E. coli (opportunistic) affect the production and composition of the intestinal mucus layer. Throughout the GI tract, there are mucus cells filled by both neutral and non-sulfated acidic mucins. Changes in the distribution or pattern of acidic and neutral mucins are involved in inflammation and cancer.^[43,44] While enterocyte-like (Caco-2) intestinal epithelial cells possess membrane-bound mucins, they do not produce secreted mucins such as MUC2 and MUC5AC, the main secreted gel forming mucins in the GI tract.^[45] Therefore an in vitro model of the GI tract with cocultures of Caco-2 and mucus-producing cells (HT29-MTX-E12), representing the mucus-secreting goblet cells of the intestine, were utilized to evaluate properties of the mucus following exposure to TiO₂ NPs, bacteria or both. This model provides a platform for understanding changes in the mucus layer, understanding disease pathogenesis, and/or studying therapeutic options such as the manipulation of the gut microbiota to address and diagnose chronic inflammatory gut diseases.

2. Results

2.1. NP Characteristics and Behavior in Dispersants

The nanoparticles used in this study (30 nm anatase TiO₂ nanoparticles, US3498US, Research Nanomaterials, Inc., Houston, TX) were previously characterized.^[7] Transmission electron microscopy analysis showed that TiO₂ NPs sizes were within manufacturer’s range (Table 1).^[7,46] Briefly, primary sizes of TiO₂ were demonstrated to range between 20 and 40 nm, which was found to be consistent with manufacturer-reported values. The behavior of TiO₂ NPs in suspension was assessed through dynamic light scattering (DLS) using a Zetasizer Nano ZS90 (Malvern Panalytical, Ltd.). TiO₂ NPs ranged from 332 ± 42 nm in water and 355 ± 70 nm in serum free DMEM which suggested the formation of aggregates in solution.^[7,46] A tendency toward agglomeration was shown with the size of agglomerates increasing based on dispersant (polydispersity index [Pdi]: 0.34–0.49 in 18 MΩ water and 0.42–0.54 in DMEM).^[7] The hydrodynamic diameter of digested TiO₂ NPs was 220.9 ± 44.8 nm in digested dispersants (Table 1). The PdI of digested TiO₂ NPs, generated by the Malvern software, was 0.509 ± 0.073. The digested TiO₂ NPs had a ζ potential of −13.0 ± 1.8 mV (Table 1). In contrast, simulated gastrointestinal tract fluid reduced the hydrodynamic sizes of TiO₂ NPs when compared with culture medium. Porcine pepsin and pancreatin/bile enzymes in the digestion solution may affect the detection and measurement of NPs properties at any NP concentration. The simulated gastrointestinal tract fluid increased the ζ potentials of digested TiO₂ NPs when compared with undigested NP in culture medium. The digested TiO₂ NPs have large, negative ζ potentials, meaning that they strongly repel each other to prevent aggregation. A high ζ potential, especially more negative than −30 mV reflects the stability of the dispersant system.^[47] The sizes measured by DLS were larger than the primary particle diameters shown in TEM images because the NP agglomerates when suspended in water.^[7]

Table 1.

Characterization of digested and undigested TiO₂ NPs in DMEM and in simulated gastrointestinal tract fluid. Hydrodynamic size (d_h,z-average), polydispersity index (PdI), and zeta potential (ζ) of 30 nm TiO₂ NPs concentrations were measured.

Concentration	μg mL−1	0.14
	Particle mL−1	2.3 × 109
Undigested NP in serum free DMEM[48]	dh,z-average [nm]	874 ± 93
	PdI	0.611 ± 0.042
	ζ [mV]	−23.4 ± 0.5
Digested NP	dh,z-average [nm]	221 ± 45
	PdI	0.509 ± 0.073
	ζ [mV]	−13 ± 1.8
Min [dh,z-average, nm]		Max [dh,z-average, nm]
Min [PdI]		Max [PdI]
Min [ζ, mV]		Max [ζ, mV]

Data are shown as mean ± SEM. (p < 0.05), n ≥ 3 independent measurements.

2.2. Evaluation of Mucus production/Staining Assays

Cocultures of Caco-2/HT29-MTX-E12 grown in either a 24-well plate, coverslips within a 6-well plate or on a polycarbonate membrane for 2 weeks at a ratio of 3:1, to represent small intestinal epithelium, developed a visible and continuous mucus layer. Secreted mucus along the barrier formed by the monolayer was evaluated using alcian blue (AB) and periodic acid-Schiff (PAS) staining. AB is a histochemical dye that stains glyco-polysaccharide and muco-polysaccharide or acidic mucosubstances. AB staining indicated that acidic mucins are spread throughout the cell monolayer to serve as a protective barrier prior to and after exposure to TiO₂ NPs. Monoculture of Caco-2 cells do not produce mucus and showed only background levels of staining when compared to staining of Caco-2/HT29-MTX-E12 conditions (Figure 1A). Monocultures of HT29-MTX-E12 and coculture of Caco-2/HT29-MTX-E12 showed more intense staining when exposed to TiO₂ compared to that of unexposed cells (Figure 1B,C). To quantitatively assess the amount of acidic mucosubstance, the amount of blue color generated from the dye was measured and it was shown that there was an increase in intensity following exposure to pristine TiO₂ NPs alone and pristine TiO₂ NPs combined with bacteria (Figure 1D,F). An increase in intensity was also observed following exposure to digest and L. rhamnosus with digest compared to negative controls with DMEM only (Figure 1G). An increase in intensity was observed following coexposure to pristine TiO₂ NPs and E. coli (Figure 1D). However, no significant changes were observed following coexposure to digested TiO₂ NPs and E. coli (Figure 1E).

Figure 1.

Acidic mucins stained with Alcian blue and imaged (20× magnification). A) Caco-2 monolayer, B) HT29-MTX-E12 monolayer, and C) Caco-e/HT29-MTX-E12 monolayer or quantified based on alcian blue color intensity in in vitro model (Caco-2/HT29-MTX-E12) following individual exposure to undigested TiO₂ NPs, bacteria or coexposure to either D,F) E. coli or L. rhamnosus, and E,G) digested TiO₂ NPs coexposed to either bacteria, respectively. Results are expressed relative to each column as mean standard error, and considered significantly different with *p < 0.05, n = 12 using a one-way ANOVA (Analysis of Variance) with Tukey’s post-test. Bars that do not share any letters are significantly different. CT – DMEM only, DG – digest.

Neutral and sialic acid containing mucosubstances were then quantified using PAS, and higher levels of neutral and/or sialic acid containing mucosubstances were detected in Caco-2/HT29-MTX-E12 exposed to TiO₂ NPs, both pristine and digested. PAS results showed that neutral and/or sialic acid containing mucosubstances secretions significantly increased in the presence of bacteria (with or without digest), stomach digest, pristine TiO₂ NPs and following coexposure to non-digested TiO₂ NPs and L. rhamnosus (Figure 2A–D). However, there was a significant decrease following coexposures to digested TiO₂ NP and E. coli (Figure 2B).

Figure 2.

Quantification of neutral mucins based on periodic acid-Schiff staining method following exposure to TiO₂ NPs and/or bacteria. A,C) Undigested and B,D) digested conditions showing different alterations in mucus secretion. A one-way ANOVA (Analysis of Variance) with Tukey’s post-test was used to determine statistical differences between the different means with p < 0.05. Bars that do not share any letters are significantly different.

2.3. Histochemical Analysis of Mucus Secretion

The combination of alcian blue and PAS was used to differentiate acidic mucins from neutral mucins present in the monolayer of Caco-2/HT29-MTX-E12. Sections were cut and stained with standard alcian blue, pH 2.5 followed by PAS. Acidic mucins were stained deep blue while neutral mucins had a bright magenta color. When both type of mucins were present, the color was a dark purple which can be seen in the displayed image of the monolayer of mucus producing HT29-MTX-E12 (Figure 3A). Cocultures of Caco-2/HT29-MTX-E12 were also stained following different exposure conditions using both staining methods and the average of overall mucus layer thickness was quantified (Figure 3B,C and Table 2). The mucus thickness increased following both exposure to bacteria and coexposure to both E. coli or L. rhamnosus and TiO₂ compared to unexposed cells. However, coexposure to L. rhamnosus and TiO₂ NPs reinstated a decreased mucus thickness, albeit still increased compared to controls. In contrast, coexposure to E. coli and TiO₂ NPs further increased the mucus thickness, indicating that the interaction of TiO₂ NPs with the bacteria varies.

Figure 3.

Cross-sectional images of HT29-MTX-E12 cell monolayer A) stained with AB/PAS showing both acidic and neutral mucins with an average mucus thickness 8.8976 ± 0.241 μm. Blue color represents Acidic mucins (AC), Magenta color – Neutral mucins (NC) and deep dark purple – both mucins types and sialic acid containing mucosubstances together. B,C) Quantification of mucus thickness following coexposure to digested TiO₂ NPs with E. coli or L. rhamnosus.

Table 2.

Representation of average mucus thickness following exposure to digested TiO₂ NPs and/or bacteria.

Treatment	Average mucus thickness [μm]	SEM
Control (DMEM only)	2.388	0.052
E. coli	4.008	0.134
L. rhamnosus	3.226	0.125
Digest	8.93	0.176
Digested TiO2	1.884	0.048
Digested L. rhamnosus	14.38	0.357
Digested E. coli	11.95	0.258
Digested TiO2 & L. rhamnosus	5.810	0.084
Digested TiO2 & E. coli	12.61	0.208

Results represented as mean ± standard error mean (SEM). Bars that do not share letters are significantly different according to one-way ANOVA with Tukey’s post-test p < 0.05.

2.4. Visualization/Quantification of Mucus by Confocal Laser Scanning Microscopy

To further visualize and quantify the mucus layer of the intestinal epithelial barrier, the samples were analyzed using confocal microscopy. Bacteria were transformed with mCherry to express red fluorescence. With the use of different fluorochromes at particular excitations, the distinction between various barrier structures including the mucus layer was made possible. Hoechst and WGA were used to stain the nucleus and the mucins secreted, respectively (Figure 4). The various components of the in vitro model with mucus layer were present on the monolayer surface while entrapped bacteria were present within the monolayer. Z-scans of a wide area of the monolayer were performed and many transversal cuts were obtained to better distinguish the mucus layer and bacteria interactions from the other components of the cells (Figure 5). The images obtained from the scan confirm the interaction observed in the quantifying methods of AB/PAS, specifically that an increase in the mucus layer was observed in the presence of bacteria alone and bacteria combined with digested TiO₂ NPs when compared to unexposed cells. However, exposure to TiO₂ NPs alone did not result in a drastic change in the mucus layer.

Figure 4.

Confocal images (orthogonal views) of Caco-2/HT29-MTX-E12 in vitro barrier. Z-scans were performed where monolayer was stained with appropriate fluorochromes. Confocal microscope was used to differentiate the components of in vitro model: A) control, D) digested NPs, E) digested NPs and E. coli, and F) digested NPs and L. rhamnosus. Mucus in green, and cell nucleus in blue, both E. coli and L. rhamnosus in expressed in red (B,C,E,F), and NPs reflection in white over the mucus layer (D) (scale bar = 20 μm).

Figure 5.

Transversal cuts of Caco-2/HT29-MTX-E12 in vitro barrier. Further analysis of the Z-scans was performed to study the interaction between TiO₂ and/bacteria with the mucus layer following a 4 h exposure. A) barrier and DMEM only B) barrier and digested TiO₂ NPs, C) barrier, E. coli D) digested TiO₂ NPs and E. coli E) barrier, L. rhamnosus F) digested TiO₂ NPs and L. rhamnosus. Mucus is stained in green, cell nucleus in blue and bacteria in red. (scale bar = 20 μm). NP was not reflective enough to be seen due to the intensity of mucus stain. White circles indicate bacteria entrapped in the mucus layer.

3. Discussion

While the production and the use of nanomaterials or nanosized particles as food additives are growing, information on their potential effects on human health, particularly the mucus layer lining the GI tract, are still lacking. This is because the complex interaction between living systems and nanoparticles are not fully understood. Such complexity derives from the ability of particles to bind and interact with biological matter while transforming their surface characteristics based on their external environment.^[10] While TiO₂ NPs have been found to be absorbed at low levels in the GI tract, their exponential production and use in foods have raised concerns about potential toxicity due to chronic ingestion.^[49] Nano-TiO₂, produced at a large scale for applications in many consumer products, have been classified as chemically and biologically inert without photoactivation.^[50,51] Nano-TiO₂ becomes highly reactive under ultraviolet irradiation and produces reactive oxygen species (ROS) which can lead to strong antibacterial activity.^[52,53] The mechanism of nanotoxicity for TiO₂ NPs has been suggested to be a production of ROS, which can alter proteins, RNA, and DNA composition.^[41] In contrast, some studies have reported few effects of TiO₂ NPs following exposure to mammalian systems and these conclusions could be due to different NPs dispersion states in those experiments.^[53,54] Previous studies with the identical TiO₂ particles, concentrations, and exposure times showed that exposure of L. rhamnosus to TiO₂ NPs did not result in a loss of bacterial viability.^[46] Additionally, experiments performed with TiO₂ of different sizes (100% rutile 17 nm, and rutile/anatase mixture M [37 and 9 nm] and P25 [48 and 55 nm]) demonstrated that, independently of their size, NPs were toxic and resulted in a decrease of E. coli viability at concentrations above 10 ppm.^[42,53] This bacteria toxicity in response to TiO₂ NPs could be a result of the adsorption of NPs to the bacterial surface due to a gap of surface charge, highly oxidative radicals inducing ROS, a potential penetration of smallest NPs or abundance of aggregates in NPs suspensions.^[42,53]

In the small intestine, secretion of gastrointestinal mucus by goblet cells (modeled in this study by HT29-MTX-E12) forms a gel that adheres at the mucosal surface to protect intestinal epithelial cells against shear forces that occur during the digestion process.^[55] The GI mucus forms a single easily removable layer and a dense inner layer firmly attached to the epithelium not allowing bacteria to penetrate.^[18,56] These mucus layers are organized around the highly glycosylated MUC2 mucin to form a net like structure that traps and transports debris and bacteria, lubricates, and lowers mechanical stress on the epithelium. By binding with bacteria and pathogenic viruses, mucus aids in digestion to immobilize enzymes near the epithelium surface, facilitating hydrolysis and absorption of nutrients.^[28] Through this net like structure, several bacteria of the intestinal microbiota are able to penetrate and use the mucus glycans as a source of carbon and energy.^[18] These properties emphasize the important role of mucus in human nutrition and health. Both the intestinal microbiota and the mucus play an important role in digestive health. Despite its classification as a major player in human nutrition and health, the microbiota have often been neglected in food toxicology in the context of nanoparticles, either separately or together.^[19] Decades ago, data pertaining to the interaction between the mucus and TiO₂ NPs using in vitro coculture model of Caco-2/HT29 was still in its infancy.^[6,57] As of today, however, studies have been conducted to further understand this interaction. Through experimental studies, researchers have reported that the translocation and accumulation of TiO₂ in the mucus layer were facilitated by the different factors such as exposure time and particle concentration.^[8,57] It has also been shown that TiO₂ NPs permeate the mucus layer and penetrate into the epithelial tissues in ex vivo studies performed on porcine buccal mucosa.^[19] The findings in this study are significant because they shed light on the effects of acute exposure to TiO₂ NPs and the implications on overall intestinal health. These effects include thickness of the mucus layer, and changes in mucin composition (amounts of neutral and acidic mucins and proportions of sulfated and sialylated mucins). Possible toxicitydriven health conditions such as inflammatory bowel diseases, ulcerative colitis and Crohn’s disease are generally linked to an alteration in mucosal integrity.^[18] With studies showing the correlation between oral exposure to NPs and damages in the GI tract lining, it is important that physical and chemical properties of the GI tract are thoroughly assessed including the intestinal mucus layer.

Previously, in vivo studies performed on rats to determine the effects of induced ROS on colonic mucins showed that high levels of ROS significantly reduced the mucus barrier function by causing a decrease in mucus thickness by up to 50%.^[58] With such reduction in the colonic mucus barrier, an increased levels of oxidant stress and decreased levels of antioxidant defense are witnessed, particularly in patients with active ulcerative colitis.^[59,60] As MUC2 is the major secretory mucin synthesized and secreted by goblet cells, a wide array of bioactive factors, including cholinergic agonist, hormones (neuropeptides), microbes and microbial products and toxins, inflammatory cytokines, and reactive oxygen and nitrogen species have triggered massive secretion of MUC2 mucin by exocytosis in stimulated secretion.^[13,61,62] Excessive increase in mucin production has also been observed in patients with cystic fibrosis, whereas production of abnormally glycosylated mucins have been found in patients with familial polyposis coli, ulcerative colitis, and colon cancer.^[43] Similarly, previous work has shown that TiO₂ NPs induce ROS in an in vitro intestinal model, and it is possible that any decrease observed in the thickness of the mucus following exposure to TiO₂ NPs is due to ROS.^[7,46] There are difficulties in the evaluation of the effects of NPs on the gut in vivo due to differences in species and variability in mucus composition. Additional studies focusing on physicochemical properties of mucins and the microbiota will be needed to further look into and determine changes in mucus layer function caused by ROS.

Studies performed on germ-free rats have shown that positive influence on mucus thickness and composition was exerted by the presence of microflora throughout the GI tract.^[63,64] Cellular damage to the biological systems has been reported to be caused by several nanoparticles as they accumulate within the system. These biological changes in the human system are mainly brought about by concentration, electrical charges, particle-size distribution, and interfacial characteristics.^[12,65] Thus, toxicity assessments of NP ingestion and bacteria interactions are important since TiO₂ NPs have been classified as potentially carcinogenic for humans through inhalation.^[41] In the current study, cells exposed to L. rhamnosus and E. coli showed an increased mucus thickness and changes in secretion of both neutral and acidic mucins. These results include increases in thickness of the mucus layer in the presence of bacteria or bacteria/TiO₂ NPs (Figure 3B,C). Exposure to E. coli led to an increase of the secretion of neutral mucins when alone or digested. These findings show that the presence of undigested TiO₂ NPs alone or with bacteria has an impact on the mucus layer function and can cause mucus secretion to be increased or decreased in the gut depending on exposure. Bacterial components recognized by mucus secreting cells also lead to an increase in secretion and production of mucins in order to fight against the presence of possible pathogens. Some invasive E. coli strains are known to be associated with Crohn’s disease, mainly because they possess a protease that is involved in mucins degradation and therefore decrease mucus viscosity.^[18] Nanosized TiO₂ has been reported to be toxic to E. coli in the presence of sunlight as it leads to the production of ROS.^[52] TiO₂ NPs can induce ROS and DNA damage in E. coli, which can produce a protein that can induce mucus release and enhance bacterial colonization of the mucosa.^[20,66] This protein belongs to a family of molecules that appear to modulate both adherence to epithelial cells and intestinal colonization. However, studies demonstrating whether TiO₂ NPs directly affect such proteins by inducing ROS or DNA damage are lacking.

A decrease in the thickness of the mucus layer could affect its integrity by making it permeable to pathogens and allowing the pathogens to penetrate and infiltrate the intestinal epithelium and lead to inflammation.^[18,24] The rheological properties of mucus have been proven to be pH dependent, mucus it is a viscous solution at neutral pH and a gel in acidic conditions.^[67] Since the mucus layer of the intestinal epithelial surface consists primarily of neutral mucins, an alteration in the function of the mucosa as a protective barrier can be indicative of a decrease in neutral mucins concentration. In the current study, it was necessary to stain the cell cocultures with the combined AB/PAS method to determine if there were changes in pattern or distribution of both mucin types which could be indicative of pathology.^[13,62] Studies that have compared germ-free mice and conventionally raised animals demonstrated that microflora have significant effects on mucus composition and thickness. Furthermore, germ-free mice had a higher percentage of neutral and sulfated mucins with a relatively thinner mucus layer.^[20] Additionally, certain intestinal microbes have been shown to induce mucin changes during infection in the GI tract by inducing the production of acidic mucins at the cost of neutral mucins, therefore changing its viscoelastic properties.^[67] In this work, we found that the coexposure to TiO₂ NPs and bacteria yields different results in each type of mucosubstances secretion as well as mucus thickness compared to individual exposure. Upon coexposure to L. rhamnosus with pristine TiO₂ NPs, an increase in both neutral and acidic mucins was observed, which is indicative of barrier integrity enhancement to act as a trap to prevent NPs from reaching and penetrating the intestinal epithelial surfaces. On the contrary, coexposure to L. rhamnosus and digested TiO₂ NPs led to no change of the neutral and acidic mucins secretions. The mucus thickness following exposure to L. rhamnosus and digested TiO₂ NPs was slightly increased when compared to controls but decreased when compared to L. rhamnosus only. This may be due to the fact that bacteria are able to induce mucin degradation by expressing specific proteins through enzymatic activities to attach to mucins glycoproteins.^[18] Coexposure to digested TiO₂ NPs and E. coli decreased neutral mucosubstances, possibly due to the secretion of proteins involved in mucin degradation and delivery of toxins to the cell surface, but increased mucus layer thickness when compared with all other conditions. The presence of microbiota in germ-free rats has been shown to influence thickness and composition of mucins. Particularly probiotic agents such as Lactobacillus species, can simulate mucin secretion in the intestine, improving pathogen resistance.^[17,63] These findings validate results in this study and the increase in mucin secretion or mucus thickness observed in the presence of L. rhamnosus. Once impaired, the mucus barrier becomes susceptible to certain bacteria species that are able to access the epithelial surface to cause inflammation, thus, it is critical to maintain the integrity of the mucus layer through commensal residents.

4. Conclusion

In this work, it was shown that exposure to TiO₂ NPs and commensal or opportunistic bacteria resulted in alterations to the thickness of the mucus layer in an in vitro GI tract model. It was found that there is an increase of mucus secretion in the presence of bacteria, which changes significantly upon coexposure to NPs, suggesting that exposure to both commensal bacteria, opportunistic bacteria, and NP impacts the mucus layer. Independent exposure to physiologically relevant concentrations of digested TiO₂ NPs resulted in a decrease of the mucus thickness. When compared to negative controls, coexposure to digested TiO₂ NPs and E. coli or L. rhamnosus resulted in a significant increase of mucus thickness. Higher amounts of acidic mucins were found in monolayers coexposed to pristine TiO₂ NPs and bacteria, while coexposure to digested TiO₂ NPs and bacteria showed no significant changes. An increase in neutral mucins was observed in the presence of both E. coli and L. rhamnosus, which was reversed when coexposed with TiO₂ NPs. These results are significant as changes in the distribution or pattern of acidic and neutral mucins are indicative of certain pathological conditions. Overall, the in vitro model developed in this study provides a platform to enable the further understanding of the changes in the mucus layer, disease pathogenesis, and to develop novel therapeutic options such as the manipulation of the gut microbiota to address and diagnose chronic inflammatory gut diseases and show that interaction of ingested TiO₂ NPs with commensal bacteria and mucus layer can alter gut homeostasis.

5. Experimental Section

All chemicals, enzymes, and hormones were purchased from Sigma-Aldrich (St Louis, MO, USA) unless otherwise indicated. All culture plates, flasks, tubes, and pipette tips used for cell culture were purchased from Corning Life Sciences. Glassware used in sample preparation and analysis was washed, soaked in 10% hydrochloric acid and rinsed with 18 MΩ de-ionized (DI) water. All other reagents were prepared in 18 mΩ water or 1× phosphate-buffer saline in water.

Nanoparticle Dose and Preparation:

This study is designed to expose an in vitro model of the GI tract to nanoparticles commonly used as food additives at physiologically relevant doses. The titanium dioxide (TiO₂) nanoparticles (30 nm, anatase, US3498US, Research Nanomaterials, Inc., Houston, TX) doses and characterization were previously described in detail.^[7] Table 1 shows physicochemical properties of primary particles and behavior in dispersants. Once dispersed, TiO₂ NPs were serially diluted in DMEM supplemented with 10% v/v heat-inactivated fetal bovine serum (HI FBS, ThermoFisher Scientific, Waltham, MA, USA) or subjected to an in vitro digestion prior to measurements of in vitro exposures. A dose of 0.14 μg mL⁻¹ was used for all in vitro experiments and all exposure times were 4 h. The acute exposure of 4 h is related to digestion time. It is the time it takes for the contents of the small intestine to transit.^[7] This selected exposure time did not affect the viability of the cells, as barrier function is necessary to complete functional studies and is a better representation of particle settling following short-term digestion. The diameter of 30 nm, anatase TiO₂ is a model of TiO₂ NPs used as food additive. The exposure time and NPs diameter were equivalent to the ones used in previous studies.^[7]

Nanoparticles Characterization:

NP agglomerates were broken down by 30 min sonication (VWR symphony Ultrasonic Cleaners, RF-48 W, 35 kHz operation frequency), and then diluted to final doses (shown in Table 1). TiO₂ NPs were sonicated in 18 MΩ DI water and dispersed at 28 mg mL⁻¹ in 5 mL of solute in 20 mL, 28 mm diameter borosilicate glass scintillation vials, by indirect sonication with an Ultrasonic Cleaner (VWR symphony Ultrasonic Cleaners, RF-48 W, 35 kHz operation frequency, Radnor, PA). The sample container with the suspension was placed in the sonication bath for 30 min on continuous mode operation with temperature starting at 21 °C. The system was calibrated by the calorimetric calibration method described by Taurozzi, whereby the power delivered to the sample was determined to be 0.128 W.^[7,68] The delivered sonication energy was measured from the calibrated delivered power over the continuous 30 min time period and determined to be 230 J. After sonication and dilution, the NP dispersants were used for determining NPs properties or exposure to the in vitro model. The properties of NPs, including distributions of digested NPs sizes, average ζ potential, and PdI, were measured with DLS using a Zetasizer Nano ZS (Malvern Instruments, Inc., Southborough, MA), with Malvern disposable polycarbonate folded capillary cells (DTS1070).

Cell Culture:

A coculture of human colon carcinoma Caco-2, and mucus producing HT29-MTX-E12 cell lines were used to mimic the human small intestinal epithelial layer. The Caco-2 cell line was obtained from the American Type Culture Collection (Manassas, VA, USA) at passage 20 and experiments were conducted at passage 30–40. The human colon adenocarcinoma mucus-secreting cell line HT29-MTX-E12 was purchased from Sigma-Aldrich (Saint Louis, MO, USA) at passage 57 and was used in experiments at passage 65–75. Both cell lines were grown in Dulbecco’s Modified Eagle Medium and 4.5 g L⁻¹ glucose, l-glutamine (DMEM, Gibco) supplemented with 10% v/v heat-inactivated fetal bovine serum (HI FBS, ThermoFisher Scientific, Waltham, MA, USA). Cells were maintained at 37 °C under 5% CO₂ and culture medium was changed every other day. For all experimental studies, cells were stained with trypan blue, counted using a hemocytometer, resuspended at a ratio of 3:1 Caco-2 to HT29-MTX. Cells were seeded at a density of 100 000 cells cm⁻² onto polycarbonate (0.4 μm), pore size (1.12 cm²), Transwell inserts, or at a density of 50 000 cells cm⁻² onto 24 well plates or (22 × 22 mm) glass coverslips within six well plates (Corning Life Sciences, Kennebunk, ME, USA) coated with rat tail Type I collagen (BD Biosciences, San Jose, CA) at 8 μg cm⁻². Previous work has shown that Caco-2/HT29-MTX monolayers seeded at 3:1 and grown for two weeks developed a (2–10 μm) thick mucus layer that covers the cell monolayer completely.^[69] All experiments were performed on fully differentiated cells 14 days post-seeding.

Bacterial Cultures:

Caco-2/HT29-MTX and bacteria triculture methods have been described previously.^[46] Lactobacillus rhamnosus GG:pMF440 and E. coli ATCC 11775, and E. coli 10798:pMF440 were inoculated and grown in brain heart infusion media (BHI, Becton, Dickinson, Sparks, MD) in glass tubes for 48 and 24 h, respectively. pMF440 is a broad host range plasmid containing the mCherry gene. Both microorganisms were kept at 37 °C and 5% CO₂. L. rhamnosus was grown in a static environment and E. coli in agitation (≈150 rpm). Prior to inoculation, the concentration of bacteria was estimated by determining the optical culture density and using a calibration curve. Culture was then serially diluted, in DMEM, to a concentration of 10³ CFU mL⁻¹ (≈263 CFU cm⁻²). Previous work has shown that bacterial viability is unaffected following a 4 h exposure to 0.14 μg mL⁻¹ TiO₂ NP.^[46]

In Vitro Digestion/Acute Exposure:

For the in vitro simulated digestions experiments, a detailed 3-phase (mouth, stomach, small intestine) gastrointestinal tract (GIT) simulator adapted from previous studies was used.^[70] Briefly, TiO₂ nanoparticles were dispersed in 18 MΩ sterile water in a scintillation vial (VWR International, LLC) and sonicated for 30 min using a bath sonicator (VWR symphony Ultrasonic Cleaners, RF-48 W, 35 kHz operation frequency). In the mouth phase, a simulated mixture of salts solution (0.1 mL of 1 m KCl, 0.3 mL of 1 m CaCl₂, and 1.4 mL of NaCl) was added for 1–5 min. To initiate gastric digestion, the pH was adjusted to pH 2.0 with 1 m HCl, 1 mL of pepsin solution (0.1 g in 0.1 m HCl) was added, and the solution was rocked on a rocker for 1 h at 55 oscillations per min at 37 °C to mimic the digestion phase in the stomach. Following gastric incubation, the pH of the sample (also referred to as the “digest”) was raised to 5.5–6 with 1 m NaHCO₃ and mixed with 5.25 mL of simulated intestinal fluid with bile salts/porcine pancreatin (0.05 g of porcine pancreatin and 0.3 g of porcine bile extract in 0.1 m NaHCO₃ for 2 h). The pH was then adjusted to 6.9–7.0 with 1 m NaHCO₃ and the final volume was increased to 20 mL with 140 mm NaCl, 5 mm KCl. The samples were then considered digests or digested NP to help differentiate from pristine (non-digested) TiO₂.

For acute 4 h exposure, cells were grown to confluency and allowed to differentiate for 14 days post-seeding while changing the culture medium at least three times per week. Fully differentiated cells were exposed to a total volume of 0.5 mL (for 24 well plates and 12 well plate Transwells) or 2 mL (for 6 well plates) of prepared digested TiO₂ or TiO₂-free and bacteria-free digest, bacteria and/or pristine TiO₂ diluted in DMEM for 4 h. To recreate these physiologically relevant doses on a surface area basis in vitro, the amount of ingested NP was divided by the surface area of the human small intestine (2 × 10⁶ cm²), which is described as the medium or average dose. High doses were two orders of magnitude higher than medium dose (Table 1). The properties of digested NP, including distributions of digested NP sizes, average ζ potential, and PdI, were measured with DLS using a Zetasizer Nano ZS (Malvern Instruments, Inc., Southborough, MA), with Malvern disposable polycarbonate folded capillary cells (DTS1070).^[7] In vitro dosage used was estimated and rationalized according to the estimated daily human intake of NPs and the surface area of human small intestinal epithelium. TiO₂ NP intake has previously been estimated to be 0.2–0.7 mg kg⁻¹ b.w. per day for Western adults and 36% of the NPs is in the nanoscale, meaning that ≈25 mg of nano-TiO₂ is consumed daily.^[4] Doses in this study were deemed relevant to reflect real-life exposure from a single meal and equivalent to the high dose used in previous work.^[7] The acute dosage seems to be a more accurate exposure estimate as ingested materials move through GI tract. The limitation of added TiO₂ NPs is set to 1% of the overall food weight in the United States.^[71]

Mucus Layer Characterization:

A modification of well-established Alcian Blue methods were used to visualize and quantify the mucus layer coverage, and quantify the acidic mucosubstances.^[55,72,73] Cocultures of Caco-2/HT29-MTX were performed on well plates rather than inserts for these assays. Following acute exposure to NP and/or bacteria, the cells were carefully washed with PBS and fixed with 4% paraformaldehyde (PFA) in 1× PBS, pH 4.0 for 30 min at room temperature. The fixing solution was then replaced with alcian blue (pH 2.5) dissolved in acetic acid to visualize acidic mucosubstances attached to the cells. Following incubation at RT for 30–40 min in the dark, cells were rinsed until supernatant was clear and images were captured using brightfield microscope (brand name). After images were taken, the intensity of the alcian blue was also quantified by lysing the stained cells with acetic acid and depositing the supernatant following centrifugation in 96-well plates. Absorbance was measured at 630 nm with a plate reader (Gen5, BioTek Synergy Instruments, Inc., Vermont, USA). A PAS assay was also performed to quantify secretion of neutral mucins and sialic acid containing mucosubstances. Following exposure to bacteria and nanoparticles, cells were lysed with 10% Triton-X 100 in 1× PBS and triplicate of aliquots (30 μL) of each cell lysate were then transferred onto 96-well plates where 120 μL of 0.6% periodic acid was added to the lysate containing wells and incubated at RT for 90 min under little agitation; followed by 100 μL of Schiff reagent for 40 min in the dark without agitation. Plates were then read at 550 nm to measure absorbance using a plate reader. Standard curves for both assays were prepared using stomach porcine mucin with increasing concentrations (0–3500 μg mL⁻¹).

Histology of Mucus Secreted by HT29-MTX-E12 Cells:

Cocultures of Caco-2 and HT29-MTX-E12 were seeded onto polycarbonate, 0.4 μm pore size, 12-mm Transwell inserts (VWR, Radnor, PA, USA) at a density of 100 000 cells cm⁻² for histochemical evaluation. After 14 days post-seeding, culture medium was removed and cells were exposed to physiologically relevant doses of TiO₂ and bacteria for 4 h; fixed with 4% PFA in 1× PBS, pH 4.0 for 30 min at room temperature and rinsed with PBS. HistoGel (Thermo Scientific, Richard-Allen Scientific, Kalamazoo, MI, USA) was liquified by heating to 60 °C and roughly 0.5 mL was added to the apical side of the polycarbonate membrane. Once HistoGel had cooled and solidified, the membranes were removed with a disposable scalpel (Electron Microscopy Sciences, Hatfield, PA) and embedded in paraffin. Thin sections were transversely cut and stained with combined alcian blue (pH 2.5; 8GX, Alfa Aesar, Wardwill, MA) and PAS to visualize the mucus layer thickness and acidic, neutral and sialic acid containing mucosubstances attached to the monolayer.

Confocal Laser Scanning Microscopy of Bacteria and TiO₂ Interaction with Mucus Layer from Coculture (Caco-2/HT29-MTX-E12):

Secretion of mucus by the coculture and its interaction with bacteria distribution was observed at a concentration of 10³ CFU mL⁻¹ using laser confocal microscopy (LSM 880, Carl Zeiss Microscopy, LLC) equipped with an oil immersion objective (PL APO 63x/1.40 oil DIC, Carl Zeiss Microscopy, LLC). In vitro cocultures were seeded onto sterile microscope glass coverslips coated with 8 μg cm⁻² of collagen Type I in 6 well plates at a density of 50 000 cells cm⁻². After 2 weeks of culture and exposure to TiO₂ and/or bacteria, cells were fixed with 4% PFA for 30 min at room temperature and rinsed with PBS. The cell membrane, nucleus and mucus layer were stained using fluorochromes Cellmask Deep Red plasma membrane stain (C10046, Life Technologies, USA), Hoechst 33342, and wheat germ agglutinin (WGA, Alexa Fluor 488 conjugate, Invitrogen) diluted in 1× PBS for 15 min in the dark at concentrations of 1/1000, 1/500, and 1/200, respectively. Samples were then rinsed, airdried and coverslips were removed from 6 well plates to be mounted on a glass slide with ProLong Gold (Thermofisher Scientific, USA) and dried overnight in the dark. Z-stack images were taken the following day and images were processed using Zen and ImageJ software (Cornell, Ithaca, USA).

Statistical Analysis:

Results are expressed as mean ± standard error. Analyses of mucosubstances composition in the in vitro model were performed using the software package GraphPad Prism version 5.0 for Windows (GraphPad Software, San Diego, California, USA). Kolmogorov-Smirnov and D’Agostino and Pearson omnibus normality test were performed prior to one-way ANOVA (Analysis of Variance) with Tukey’s post-test to assess for statistical differences between the various means of each series of experiments. Differences between means were considered statistically not significant when the p-value was >0.05. Mucus thickness was measured (50 measurements per section/3–6 sections per sample/3 samples per condition) using IBM SPSS Statistics version 25’s one-way ANOVA post hoc test Duncan with p < 0.05.