**4. Discussion**

This systematic review of animal studies found that TiO2 dietary exposure might increase or decrease abundance in specific bacterial species, even if an overall impact on bacterial α-diversity has not been clearly demonstrated. Moreover, this review highlights that TiO2 exposure could lead to perturbations in intestinal metabolism, gu<sup>t</sup> barrier integrity, and gu<sup>t</sup> immunity.

The limited effect of TiO2 exposure on α-diversity of the gu<sup>t</sup> microbiota was found in the majority of included studies. This could be explained by the short duration of the interventions, not exceeding three months. The lack of effects of different dietary interventions on gu<sup>t</sup> microbiota diversity has been shown in previous systematic reviews investigating the effects of dietary patterns or dietary interventions—such as dietary fiber interventions or probiotics interventions—on gu<sup>t</sup> microbiota [52]. Long-term studies are required to assess this hypothesis. In regards to bacterial abundances, in various included studies [35,39,44,49], significant compositional changes are reported after TiO2 exposure compared with controls. TiO2 exposure could lead to an alteration of the Firmicutes/Bacteroides ratio, a depletion of *Lactobacillus,* and enrichment of Proteobacteria [40,50]. Interestingly, these microbial variations are also found in studies investigating the effect of other food nanoparticles such as nano-Ag, ZnO, and SiO2 exposure [53]. *Lactobacillus* is a genus wellknown to produce SCFAs, metabolites involved in host metabolism, while Proteobacteria might be overrepresented in inflammatory intestinal and extra-intestinal diseases. Indeed, this observed dysbiosis is also a hallmark of inflammatory bowel disease, colorectal cancer, or chronic metabolic disorders such as obesity [54].

The intestinal microbiota plays a key role in gastrointestinal functions such as the digestion and fermentation of indigestible polysaccharides, differentiation of the intestinal epithelium, and the maintenance of mucosal barrier integrity, including mucus characteristics. Mucus is a viscoelastic gel that separates the intestinal epithelium from the gu<sup>t</sup> lumen. It consists of water and mucins, lipids, as well as epithelial and globets cells. Goblet cells are localized in the intestinal crypts and secrete proteins such as muc-2 (encoded by MUC2 gene). Intestinal bacteria influence the shaping of the mucus regulating LPS and SCFAs. Indeed, SCFAs—mainly butyrate—stimulate muc-2 protein production and influence mucus quality. Numerous studies [55–57] demonstrated that germ-free mice, comparing with conventionalized mice, were provided with an underdeveloped intestinal epithelium with decreased mucus production, intestinal epithelial cell differentiation, and villus thickness. These alterations could lead to an increased permeability allowing the passage of harmful intraluminal microorganisms and microbial toxins. These bidirectional interactions between gu<sup>t</sup> microbiota composition and gu<sup>t</sup> barrier functions could be impaired with TiO2 exposure. Indeed, in some included studies [35,45,49], TiO2 exposure could be associated with a reduction of SCFAs, a decrease of goblet cells and crypts, a reduction of mucus production with a lower MUC2 expression. These in vivo findings confirmed the results of in vitro studies demonstrating that TiO2 NPs could alter microvilli structure and epithelial integrity [19,24]. Particularly, in vivo and ex vivo, TiO2 NPs can cross the regular ileum and follicle-associated epithelium and alter the paracellular permeability of the ileum and colon epithelia, which is a sign of integrity alteration [58]. However, three studies [8,37,48] did not show significant changes in terms of epithelium permeability, SCFAs levels, and mucus barrier impairment. Considering the TiO2 dose exposure of the studies, we can hypothesize that these discrepancies could be due to dose exposure and healthy conditions of the animals at baseline.

TiO2 NPs also could interact with gu<sup>t</sup> immunity. Indeed, a majority of included studies have assessed associations between TiO2 exposure and increased biomarkers of intestinal inflammation such as increased interleukins levels. Recent in vitro studies [19,27] found that TiO2 NPs could stimulate the production of pro-inflammatory cytokines. Moreover, in vivo, the number of T reg cells decreased after 100 days of TiO2 exposure [8]. T reg cells are well-known to limit gu<sup>t</sup> inflammatory responses and prevent food allergy development [59]. Thus, long-term TiO2 exposure could have an immunosuppressive role by limiting the production of T reg cells. Interestingly, there are significant changes in terms of IL production, significantly aggravated in obese mice treated with TiO2 compared with non-obese mice [35,51]. This shows that TiO2 could exacerbate intestinal inflammation in mice affected by metabolic diseases such as obesity. Mu et al. [44] analyzed the effect of TiO2 NPs on DSS-induced chronic colitis in mice showing that DSS-induced chronic colitis worsened by chronic TiO2 NPs exposure with a reduction of immune cells such as CD4 + T

cells and Tregs. Further studies are required to deepen the effects of TiO2 NPs on immunity responses, and specifically on the gu<sup>t</sup> microbiota immune axis.

Overall, TiO2 exposure can raise concerns if we consider the cocktail effects of daily consumption of the different food additives. Indeed, other NPs present in food, emulsifiers, and artificial sweeteners have also dysbiotic effects on gu<sup>t</sup> microbiota [60]. This cocktail effect raises particular concerns since the quantity of food additives is not detailed in the ingredient list, making impossible the calculation of the daily quantity of food additives. For example, chewing one piece of chewing gum can result in an intake of 1.5–5.1 mg of TiO2 NPs [61]. These concerns are even more important in children. Indeed, candies, gums, desserts, and beverages—products containing the highest levels of TiO2 NPs—are consumed two to four times higher for children than for adults [3]. A Dutch survey estimated a mean TiO2 NPs intake of 2.16 (2.13–2.26) mg/kg bw per day among children aged two to six years old, and a mean of 0.55 (0.52–0.58) among people aged 7–69 years old, with toothpaste, candy, coffee creamer, fine bakery wares, and sauces mostly contributing to the TiO2 daily intake [62]. Childhood is a key development time for the shape of the microbiota that can have considerable consequences in later life [63]. Although TiO2 consumption has considerably increased in the last few decades in Western countries and despite dietary composition having an impact on gu<sup>t</sup> and overall health [64], the possible impacts of long term effects of TiO2are still poorly understood.

This systematic review has some limitations. Although the majority of included studies have used rodent models, the methods of administration (gastric gavage, addition to drinking water, addition to food), TiO2 doses, and exposure durations differ between studies and do not allow pooling results. Thus, since some studies detect only a limited impact on the microbiota, others reporting various significant changes, it remains difficult to reach firm conclusions. Another limitation are the very high doses used in animal studies compared to the estimated daily intake in humans. Indeed, the amount of TiO2 consumed is estimated to 1 mg of TiO2/kg bw/day in adults in the United Kingdom and Germany, while the ingested quantity can exceed 3 mg of TiO2/kg bw/day in children [3,65]. Thus, the results from animal studies cannot be directly extrapolated to humans. Furthermore, only 15% of the 16S rRNA sequence dataset for the mouse microbiota are shared with humans [66]. Since randomized controlled studies are unethical, the use of germ-free mice inoculated with the human microbiota could be feasible to elucidate the impact of TiO2 NPs on gu<sup>t</sup> bacteria that colonize the human intestine. Moreover, different dietary patterns such as HFD or high fiber diet should be evaluated to compare the impact on TiO2 NPs in healthy individuals with those in poor health.
