1
Scientific RepoRts | 7:40373 | DOI: 10.1038/srep40373
www.nature.com/scientificreports
Food-grade TiO
2
impairs intestinal
and systemic immune homeostasis,
initiates preneoplastic lesions
and promotes aberrant crypt
development in the rat colon
Sarah Bettini
1
, Elisa Boutet-Robinet
1
, Christel Cartier
1
, Christine Coméra
1
, Eric Gaultier
1
,
Jacques Dupuy
1
, Nathalie Naud
1
, Sylviane Taché
1
, Patrick Grysan
2
, Solenn Reguer
3
,
Nathalie Thieriet
4
, Matthieu Réfrégiers
3
, Dominique Thiaudière
3
, Jean-Pierre Cravedi
1
,
Marie Carrière
5,6
, Jean-Nicolas Audinot
2
, Fabrice H. Pierre
1
, Laurence Guzylack-Piriou
1
&
Eric Houdeau
1
Food-grade titanium dioxide (TiO
2
) containing a nanoscale particle fraction (TiO
2
-NPs) is approved as
a white pigment (E171 in Europe) in common foodstus, including confectionary. There are growing
concerns that daily oral TiO
2
-NP intake is associated with an increased risk of chronic intestinal
inammation and carcinogenesis. In rats orally exposed for one week to E171 at human relevant levels,
titanium was detected in the immune cells of Peyer’s patches (PP) as observed with the TiO
2
-NP model
NM-105. Dendritic cell frequency increased in PP regardless of the TiO
2
treatment, while regulatory T
cells involved in dampening inammatory responses decreased with E171 only, an eect still observed
after 100 days of treatment. In all TiO
2
-treated rats, stimulation of immune cells isolated from PP
showed a decrease in Thelper (Th)-1 IFN-γ secretion, while splenic Th1/Th17 inammatory responses
sharply increased. E171 or NM-105 for one week did not initiate intestinal inammation, while a 100-
day E171 treatment promoted colon microinammation and initiated preneoplastic lesions while also
fostering the growth of aberrant crypt foci in a chemically induced carcinogenesis model. These data
should be considered for risk assessments of the susceptibility to Th17-driven autoimmune diseases and
to colorectal cancer in humans exposed to TiO
2
from dietary sources.
Titanium dioxide (TiO
2
) is a naturally occurring metal oxide and is one of the ve engineered nanomaterials
most commonly used in daily consumer products, including food
1
. e TiO
2
food additive, referred to as E171
in the European Union (EU), is commonly used as a whitening and brightening agent in confectionary (candies
and chewing gum), white sauces and icing
1–3
. e Food and Drug Administration approved the use of food-grade
TiO
2
in 1966 with the stipulation that TiO
2
levels must not exceed 1% of the food weight
4
. In Europe, the current
EU Directive 94/36/EC authorizes the use of E171 in foodstus without establishing an acceptable daily intake
level by the Joint FAO/WHO Expert Committee on Food Additives, based on TiO
2
absorption considered to be
very low
5
. Nevertheless, the common use of E171 leads to signicant levels of daily dietary intake of nanopar-
ticulate matter among humans
1
. Indeed, E171 batches show broad size distributions of TiO
2
primary particles
1
Toxalim (Research Centre in Food Toxicology), Université de Toulouse, INRA, ENVT, INP-Purpan, UPS, Toulouse,
France.
2
Luxembourg Institute of Science and Technology (LIST), Materials Research and Technology (MRT),
Advanced Instrumentation for Ion Nano-Analytics (IANA), L-4362 Esch-sur-Alzette, Luxembourg.
3
Synchrotron
SOLEIL, F-91192 Gif-sur-Yvette, France.
4
French Agency for Food, Environmental and Occupational Health and
Safety (ANSES), F-94701 Maisons-Alfort, France.
5
Université Grenoble-Alpes, INAC-LCIB, Laboratoire Lésions des
Acides Nucléiques, 17 rue des Martyrs, F-38000 Grenoble, France.
6
CEA, INAC-SCIB, Laboratoire Lésions des Acides
Nucléiques, 17 rue des Martyrs, F-38000 Grenoble, France. Correspondence and requests for materials should be
addressed to L.G.-P. (email: laurence.guzylack@inra.fr) or E.H. (email: eric.houdeau@inra.fr)
Received: 13 June 2016
Accepted: 06 December 2016
Published: 20 January 2017
OPEN
www.nature.com/scientificreports/
2
Scientific RepoRts | 7:40373 | DOI: 10.1038/srep40373
(diameters of 30 to 400 nm), with up to 36% of particles falling below 100 nm in one dimension, i.e., nanoparticles
(TiO
2
-NPs)
1–3
. TiO
2
-NPs have been easily isolated from food products such as chewing gum
6
. Human exposure
analyses on foods consumed among American and British populations report that children under the age of
10 present the highest exposure level compared to adults (1–3 vs. 0.2–1 mg TiO
2
/kg of body weight (BW)/day,
respectively)
1
. However, the oral route for TiO
2
remains poorly investigated among toxicological testing studies,
in contrast to issues of dermal contact or inhalation, i.e., the main routes for occupational exposure
1,7
. In addi-
tion, studies on the gastrointestinal (GI) uptake and eects of TiO
2
have been primarily conducted based on NP
models such as P25 Aeroxide
®
, which is referenced in the nanoparticle repository of the Joint Research Centre
(JRC) (Ispra, Italy) as NM-105; in contrast to E171, these NP models are strictly nanosized
8–12
. Although most
studies agreed for limited intestinal absorption of TiO
2
in rats and humans
13–15
, a facilitated passage of TiO
2
-NPs
through microfold cells (M-cells) lining the Peyer’s patches (PP) has been demonstrated in vitro and in vivo
8,16
.
In humans, TiO
2
particles of dietary origin have been found in the PP of patients suering from inammatory
bowel disease (IBD)
17
including infants
18
, and potent inammasome activation has been reported in vitro using
TiO
2
-NPs
19
. ese studies point to possible contributions to chronic inammatory processes in the gut if TiO
2
particles accumulate in the cells of the PP through chronic dietary exposure, and this remains to be explored in
vivo with the E171 food additive at relevant exposure levels for humans. Furthermore, TiO
2
has been classied
by the International Agency for Research on Cancer (IARC) as a possible human carcinogen in Group 2B aer
inhalation
20
on the basis that inhaled or intra-tracheally administered nano- and ne-sized TiO
2
induces lung
cancer in rats
21
. Given the increasing number of commercial foods containing the TiO
2
additive, in vivo experi-
ments are required to determine whether chronic exposure to food-grade TiO
2
particles may present risks of IBD
and/or carcinogenesis in the exposed gut on a daily basis. In the present study, we examine the tissue distribution
and immunotoxicity of E171 food-grade TiO
2
orally administered over 7 days to rats at 10 mg/kg of BW/day in
comparison to the NM-105 (i.e., P25) referent OECD nanomaterial. e patterns of intestinal inammation,
preneoplastic lesion development and colonic aberrant crypt foci (ACF) promotion were assessed in rats with or
without dimethylhydrazine (DMH)-induced carcinogenesis following oral E171 treatment at the same dosage
delivered over 100 days.
Results
Food-grade TiO
2
particles cross the gut barrier and reach the liver without altering intestinal
permeability or causing DNA damage in Peyer’s patches. Analysis of the particle size and crystal
form of food-grade TiO
2
shows that our E171 batch is a representative commercially sourced TiO
2
food addi-
tive for our oral toxicity study (SupplementaryFig.S1 and SI section). Confocal and uorescence reection
microscopy methods were used to examine the fate of TiO
2
along the gut-liver axis in rats that were orally given
ultrasonicated E171 particles in water. We rst studied the dispersion state of TiO
2
particles recovered from the
luminal content of the jejunum and colon 4 h aer a single dose of E171 was delivered. In comparison to the
initial bolus, TiO
2
particles did not reagglomerate in vivo when transiting along the gut (SupplementaryFig.S2).
Upon absorption, light-diracting TiO
2
particles were found in the PP along the small intestine as well as in the
colonic mucosa and liver of rats orally given E171 for 7 days but not in the controls (Fig.1a and complementary
TEM images in SupplementaryFig.S3). In the same rats, to ensure that the light scattering particles are primarily
TiO
2
, we used µ XRF for Ti element detection. As expected, Ti was detected in the gut lumen (i.e., corresponding
to the residual bolus of E171 given to the rats) and PP (Fig.1b) as well as in colon mucosa (Fig.1c). In addition, Ti
was found in the liver, with the highest density found close to the portal vein sinus, which collects blood from the
intestine (Fig.1d). Finally, we assessed whether oral exposure to E171 aected gut permeability in vivo, thereby
facilitating particle absorption as a result of barrier disruption. No signicant change in epithelial paracellular
permeability to
51
Cr-EDTA that was orally given to rats was observed in the E171 group in comparison to the
controls (1.41 ± 0.08 vs. 1.63 ± 0.09% of total radioactivity recovered in 24 h urine samples, respectively; P = 0.1).
To compare the subcellular distribution of Ti elements in rats orally given E171 or TiO
2
-NP model NM-105,
we executed nanoscale secondary ion mass spectrometry (nanoSIMS) imaging with a beam size of 80–100 nm,
allowing for the high-resolution mapping of the distribution of TiO clusters
22
in rats orally dosed for 7 days. No Ti
signal was detected in PP tissue sections of the control rats (Fig.2), while Ti was found in the PP of all TiO
2
-treated
rats, with similar distribution patterns between the NM-105 and E171 TiO
2
sources (Fig.2). e highest Ti den-
sity was found in the central zones of the PP, which are rich in immune cells (SupplementaryFig.S4a,b). In addi-
tion to Ti in the cytoplasm, Ti-rich regions were identied in the nuclei of PP cells and were closely associated
with the phosphorus-positive chromatin (Fig.2 and SupplementaryFig.S4b). Due to the nuclear translocation
of Ti, the genotoxicity of both TiO
2
compounds was evaluated (see SI section). No increase in DNA damage was
detected in PP cells of the E171- and NM-105-treated rats (SupplementaryFig.S4c,d).
Food-grade TiO
2
particles aect dendritic cell frequencies and T cell populations in the Peyer’s
patches and cause imbalances in intestinal and systemic immune responses. Resident dendritic
cells (DC) in gut sample antigens from the lumen, and have important implications for tolerance and immune
defences
23
. We rst evaluated frequency of DC in the TiO
2
-treated rats, namely the CD11b/c
+
CD103
+
MHC-II
+
DC, which are pivotal for immune tolerance as they induce regulatory T cells (Tregs)
24,25
. Aer 7 days of oral
exposure, both NM-105 and E171 induced a signicant increase in DC frequency in PP (Fig.3a) without aecting
the spleen at the systemic level (not shown). Aer chronic E171 treatment, early eects on DC in the PP were
found to be transient, as they were not detected in rats exposed for 100 days through drinking water (Fig.3a).
Regarding Tregs, the NM-105 nanomaterial had no effect on PP after 7 days of oral exposure (Fig.3b)
while the same duration of treatment with the E171 additive led to a signicant decrease in this cell subset
(i.e., CD4
+
CD25
+
FoxP3
+
) that was still observed in PP aer 100 days of exposure (Fig.3b,d). Interestingly, we
found that decreased levels of Tregs appeared concomitantly with a decrease in CD4
+
CD25
+
T helper () cells,
www.nature.com/scientificreports/
3
Scientific RepoRts | 7:40373 | DOI: 10.1038/srep40373
indicating failure of cell expansion (Fig.3c,e). To determine whether TiO
2
particles directly mediated T cell
depletion, cells isolated from PP of untreated rats were exposed ex vivo to E171 particles or NM-105 TiO
2
-NPs,
and cell viability and proliferation were compared. A dose-dependent cytotoxic and anti-proliferative eect on
the T cells was observed, and this eect was found to be more pronounced with E171 compared to the NM-105
TiO
2
-NP model (SupplementaryFig.S5).
We then compared the eects of orally administering NM-105 and E171 particles to rats for 7 days on mucosal
inammation and immune cell responses in PP and the spleen. We did not detect any change in myeloperoxidase
(MPO) activity, a marker of neutrophil inltration, or in the content of basal cytokines (i.e., tumour necrosis fac-
tor (TNF)-α interleukin (IL)-10, IL-1β , interferon (IFN)-γ and IL-17) in mucosa of the small and large intestine
relative to the control rats (SupplementaryTableS1). To study ex vivo immune cell responses, total immune cells
were isolated from PP and the spleen and then cultured with anti-CD3/CD28 antibodies to induce cytokine secre-
tion into the culture media. In the PP, all TiO
2
materials attenuated inammatory IFN-γ secretion relative to the
controls while the IL-17 response remained unchanged (Fig.4a). In the spleen, both NM-105 and E171 elicited a
potent 1/17 immune response through increased production of IFN-γ and IL-17 (Fig.4b).
Food-grade TiO
2
particles initiate and promote preneoplastic lesion formation in the colon and
induce mucosal low-grade inammation. We rst explored the promotion of preneoplastic lesions (i.e.,
ACF) in vivo in rats treated with DMH to initiate colon carcinogenesis. Rats were exposed to food-grade TiO
2
in
drinking water at 200 µ g and 10 mg/kg of BW/day for 100 days, i.e., at doses approximating human dietary levels
for adults and children
1
. e number and size of ACF (i.e., the number of lesions and the number of aberrant
crypts per lesion) and the number of total aberrant crypts per colon were examined in a double-blind study. E171
treatment at 10 mg/kg of BW/day signicantly increased the total number of aberrant crypts per colon as well as
the number of large ACF per colon (i.e., more than three aberrant crypts per ACF) (Fig.5a,b) relative to the con-
trol and 200 µ g/kg of BW/day groups. Despite an increasing trend at the highest dose, no signicant dierence in
the number of ACF per colon was observed between the groups of rats (Fig.5c). To explain the growth-promoting
eects on colonic preneoplastic lesions, we tested whether E171 dierentially aects the viability of normal or
preneoplastic cells through the comparative cytotoxicity of food-grade TiO
2
particles on nonmutated (Apc+ /+ )
cells and genetically dened preneoplastic (Apc Min/+ ) cells using an MTT assay. At the two concentrations
tested, we found that 24 h exposure to E171 was more cytotoxic to Apc+ /+ than to Apc Min/+ cells (Fig.5d),
hence providing an in vitro rationale for the selection of preneoplastic cells in early stages of carcinogenesis.
Figure 1. Tissue distribution of E171 particles in the rat intestine and liver aer 7 days of oral exposure.
(a) Confocal images of PP, colon and liver tissue sections from control and E171-treated rats showing tissue
autouorescence in red and light-scattering TiO
2
particles in green (arrowheads) (scale bars 10 µ m). (b–d) µ XRF
mapping of Ti distribution (red pixels) in PP (b), colon (c), and liver (d) tissue sections. In (b), the two le panels
show Ti distribution in PP vs. the luminal side (lum) in higher resolution maps (10X) (scale bars 100 µ m). In (c), note
the presence of Ti overlaying iron (Fe)-rich epithelial cells lining the colonic mucosa (dashed line in the right panel)
and Ti distribution in the mucosa (muc) (scale bars 100 µ m). In (d), a tissue section from the liver in a Ti-rich area
close to the portal vein sinus is shown (scale bar 50 µ m).
www.nature.com/scientificreports/
4
Scientific RepoRts | 7:40373 | DOI: 10.1038/srep40373
We also determined whether chronic E171 treatment at 10 mg/kg of BW/day may initiate the spontane-
ous development of ACF in normal rats, i.e., without the induction of carcinogenesis by DMH. No ACF were
observed in the colons of the control rats (Fig.6a). Conversely, in the E171 group, 4 of the 11 animals spon-
taneously developed one to three ACF per colon (Fig.6a). ree of the 4 rats developed lesions of 1 to 3 aber-
rant crypt(s) per ACF, and 1 rat developed a severe lesion of 12 aberrant crypts (Fig.6b). Interestingly, cytokine
assays showed moderate but signicant increases in TNF-α (+ 26%, P < 0.05), IL-8 (+ 45%, P < 0.01), and IL-10
(+26%, P < 0.05) in the colonic mucosa of E171-treated rats relative to the controls (Fig.6c). Western blotting for
caspase-1 did not show cleaved caspase-1 in the colons of E171-treated rats relative to control animals (Fig.S6),
indicating the absence of inammasome activation into the mucosa; accordingly, no signicant change in the
downstream caspase-1 eectors IL-1β and IL-18 (Fig.6c) were found in our experimental setting using a low
dose of E171.
Discussion
Titanium dioxide, which is manufactured as a food ingredient (and referred to as E171), is ingested daily as mixed
nano- and submicron-sized particles in the human diet
1
. While recent reports based on NP models show that
TiO
2
-NPs translocate through the intestinal epithelia, no in vivo study has been carried out to investigate the tis-
sue distribution of food-grade TiO
2
particles along the gut and whether the nanoscale fraction of E171 particles
presents a specic risk via the oral route. Our study shows that ultrasonicated E171 particles prepared in water
before oral administration to rats did not reagglomerate in vivo in the intestinal lumen. Transepithelial passage
occurred in the jejunum and the colon aer one week of treatment, and the titanium (Ti) reached the liver, exhib-
iting systemic absorption of E171 as previously reported based on TiO
2
-NP models
13,14,26,27
. is rst indicates
that the daily consumption of E171-containing food may constitute a persistent source for the systemic passage
of TiO
2
-NPs and that particles sequestered into the gut mucosa represent an unexplored topic for in vivo toxic-
ity assessments of food-grade TiO
2
. No change in intestinal permeability was observed, indicating that particle
absorption did not result from a loss of epithelial barrier integrity aer TiO
2
treatment. e relatively low density
of Ti signals in the liver suggested limited hepatic retention aer one week of daily dosing. is is in accordance
with previous oral studies using NP models, wherein low Ti levels (< 0.03 to 0.2 µ g Ti/g of tissue) were detected in
the liver and were not found to accumulate aer 5 days of daily oral exposure at a similar dose
14
or aer 13 weeks
of treatment at higher doses (> 250 mg/kg of BW/day)
13
.
Figure 2. NanoSIMS analyses of subcellular Ti distribution in PP aer 7 days of oral exposure to NM-105
or E171. e raster size was set to 20 × 20 µ m
2
. NanoSIMS images for the elemental distributions of carbon-
nitrogen
12
C
14
N (green), phosphorus
31
P (blue), and titanium oxide
48
Ti
16
O (red) and the merged image of
31
P
and
48
Ti
16
O (blue/red) on ultra-thin sections of PP (scale bars, 5 µ m). e image overlay (P + Ti) shows Ti-rich
zones in the nuclei of cells in PP (arrowheads).
www.nature.com/scientificreports/
5
Scientific RepoRts | 7:40373 | DOI: 10.1038/srep40373
In humans, TiO
2
absorption into the bloodstream has been recently reported for healthy volunteers
orally given a single dose of a pharmaceutical/food-grade TiO
2
formulation
28
. In this pilot study, because
lumen-to-blood TiO
2
passage was found within 2 h aer ingestion, the authors concluded that particle uptake
is limited to the small intestine. Using rats exposed daily for one week, we provide evidence for the occurrence
of TiO
2
absorption not only in the small intestine but also in the colon. It is likely that the slow transit time in
the large intestine is responsible for TiO
2
accumulation in the colonic lumen aer repeated oral intake, forming
a reservoir that could favour local absorption by epithelial cells. Importantly, the colon epithelium in rats and
humans is a region rich in mucus-producing goblet cells, and a recent study using human Caco-2/HT29-MTX
cell co-culture (i.e., a model of goblet cells) clearly showed that TiO
2
-NPs are preferentially entrapped by cells in
the HT29-MTX co-culture model compared to Caco-2 cells cultured alone as regular enterocytes
8
.
In the human-focused study using a single dose
28
, lumen-to-blood TiO
2
translocation began early in the small
bowel but peaked 6 h aer ingestion. is delayed passage is thought to be a result of PP uptake due to an avid
capture of TiO
2
particles by antigen-presenting M-cells lining the dome of the PP. A facilitated translocation
pathway for TiO
2
-NPs has been demonstrated in vitro using a cell model of follicle-associated epithelium mim-
icking M-cells
8
. In our in vivo study, µ XRF and nanoSIMS clearly showed Ti internalization in PP cells of rats
orally exposed to food-grade TiO
2
. Given that the high resolution of nanoSIMS images (i.e., 80–100 nm) allows
subcellular cartography, it is noteworthy that E171 titanium reached not only the cytoplasm of PP cells but also
the nucleus. Similar Ti internalization was observed with pure nanoparticulate TiO
2
matter (i.e., NM-105), sug-
gesting that the nanoscale particle fraction of TiO
2
in the E171 additive also distributes to immune cells following
uptake by PP. Using Raman imaging, high levels of nuclear TiO
2
-NP uptake have been recently reported in vitro
Figure 3. Frequency of dendritic, regulatory T and cells in Peyer’s patches following oral TiO
2
. Rats
were orally exposed to 10 mg/kg of BW/day with NM-105 (grey bars), E171 (black bars) or vehicle (white bars)
for 7 days by daily gastric gavage or for 100 days through the drinking water. e average frequency of DC cells
(a), Tregs (b), and cells (c) in PP (n = 10 to 11 rats/group); representative eects of chronic E171 treatment
for 100 days on Treg and cell populations based on FoxP3 and CD25 expression by CD4
+
T cells (d) and
CD25 expression in CD4+ T cells (e). All data are expressed as proportions and are written as the mean ± s.e.m.
*P < 0.05, **P < 0.01, ***P < 0.001 vs. the control: one-way ANOVA followed by Tukey’s multiple comparison
test for the 7-day treatment and a Student’s t-test for the 100-day treatment.