Contribution of M-cells and other experimental variables in the translocation of TiO2 nanoparticles across in vitro intestinal models

Accepted Manuscript Contribution of M-cells and other experimental variables in the translocation of TiO2 nanoparticles across in vitro intestinal models

Joan Cabellos, Camilla Delpivo, Elisabet Fernández-Rosas, Socorro Vázquez-Campos, Gemma Janer PII: DOI: Reference:

S2452-0748(16)30126-4 doi: 10.1016/j.impact.2016.12.005 IMPACT 46

To appear in:

NANOIMPACT

Received date: Revised date: Accepted date:

29 September 2016 1 December 2016 19 December 2016

Please cite this article as: Joan Cabellos, Camilla Delpivo, Elisabet Fernández-Rosas, Socorro Vázquez-Campos, Gemma Janer , Contribution of M-cells and other experimental variables in the translocation of TiO2 nanoparticles across in vitro intestinal models. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Impact(2016), doi: 10.1016/j.impact.2016.12.005

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Contribution of M-cells and other experimental variables in the translocation of TiO2 nanoparticles across in vitro intestinal models Joan Cabellos1, Camilla Delpivo1, Elisabet Fernández-Rosas1, Socorro Vázquez-Campos1, Gemma Janer1 1

PT

Leitat Technological Center, Terrassa (Barcelona), Spain

RI

The prediction of nanomaterial (NM) intestinal absorption is a key step in the evaluation of risks after oral exposure. The Caco-2 monolayer model is widely used to predict the

SC

permeability of substances across the intestinal barrier. The addition of M-cells should improve the relevance of the model for NMs, since transcytosis through M-cells is considered

NU

to play a major role in the intestinal absorption of particles. A Caco-2/Raji B co-culture was established to induce the transformation of enterocytes into M-cells. The model was characterized by monitoring the transepithelial electrical resistance (TEER) values and by

MA

scanning electron microscope (SEM) imaging. The translocation of TiO2 nanoparticles (NPs) was evaluated in the Caco-2/M-cell system and compared to that in a Caco-2 monolayer

D

model. In addition, two different co-incubation periods (co-culture day 4 vs. co-culture day 7) and two different NP dispersion protocols (probe sonication for different periods and

PT E

amplitude) were used to evaluate whether these experimental variables could influence translocation. The exposure to TiO2 NPs did not cause changes in the morphology or TEER values of the cell monolayers. A three-way ANOVA for the titanium levels in the basal

CE

compartment showed no interactions between the variables tested, but a significant contribution to the overall variance of the data by the cell model (p<0.05). The levels of

AC

titanium recorded in the basal compartment were significantly higher (p<0.05) than the background titanium levels only in the presence of M-cells. This work supports a relevant role for M-cells in nanoparticle absorption. However, a standardized protocol for the Caco-2/M-cell co-culture model should be established and well characterized before this can be used to compare translocation data among NMs, particularly if tested in different laboratories.

1

ACCEPTED MANUSCRIPT

NU

SC

RI

PT

Graphical abstract:

MA

Keywords: Caco-2, Raji B, M-cell, oral absorption, TiO2 nanoparticles

A Caco-2/M-cell co-culture was obtained after co-incubation of Caco-2 cells and Raji B

PT E



D

Highlights:

cells. 

TiO2 NP at 100 µg/mL did not cause changes in morphology or in the TEER values of the Caco-2 or Caco-2/M-cell models.

The inclusion of M-cells increased the translocation of TiO2 NPs.



Two sonication protocols led to different aggregate sizes, but did not result on statistically

CE



AC

significant differences in translocation rates.

2

ACCEPTED MANUSCRIPT 1. Introduction The recent rise in the use of manufactured products containing nanomaterials has increased the human exposure to particles in the nanometric range (<100 nm in at least one dimension). Inhalation is the main route of exposure to nanomaterials in occupational settings and in some consumer applications and has received a lot of attention. Besides inhalation exposure, oral exposure to nanoparticles also deserves attention due to the use of nanomaterials in the food

PT

sector (additives, colorants, etc). Some metal oxide compounds, such as TiO2 (E171), synthetic amorphous silica (E551), or iron oxide (E172) have been used as food additives (colouring or

RI

anticaking agents) for several decades (WHO TRS 557, 1974; SCF, 1975; WHO TRS 773, 1986). The safety evaluation of E171 have assumed that they would be free of toxic effects based on

SC

the lack of significant absorption and tissue storage in several species including humans (EFSA, 2004). And the evaluation of E172 by the JECFA Committee was also based on the absence of

NU

information on physiological absorption and storage of iron following the use of these pigments as food additives (WHO TRS 557, 1974; and WHO TRS 631, 1978). The food colorant TiO2 (E171) has a target particle size around 200-300 nm, which provides its white colour.

MA

However, the size distribution of the E171 material is wide and a fraction of the particles is in the nano range: 10-15% of the particles according to Peters et al. (2014) or 36% of the particles

D

according to Weir et al. (2012) would have sizes below 100 nm. Also in the case of E551, there is evidence that shows that a proportion of the particles have sizes that fall in the nano range

PT E

(Dekkers et al., 2011; Athinaravanan et al., 2014). Understanding the rate of absorption of these and other nanomaterials is a critical step in their

CE

safety evaluation. The estimation of the rate of absorption of nanomaterials in in vivo oral studies is challenging due to low absorption rates, rapid biodistribution from blood into multiple tissues, and several analytical limitations to quantitatively assess the presence of

AC

nanomaterials in tissues. However, advances in analytical techniques suggest that, although at a very low rate, intestinal absorption of nanoparticles does occur (e.g., Hillyer and Albrecht, 2001; Wang et al., 2007; Li et al., 2012; Schleh et al., 2012; Baek et al., 2012; Brun et al., 2014; Janer et al., 2014; MacNicoll et al., 2015). In addition to these technical difficulties, concerns exist on the use of experimental animals for the potential multitude of nanomaterial variations that can be generated (e.g., in terms of size, shape, coating, etc). Therefore, there is a need to establish robust in vitro models that can predict in vivo oral absorption or, at least, can be used to compare the relative absorption of different NMs. 3

ACCEPTED MANUSCRIPT When grown in monoculture in a cell insert, Caco-2 cells differentiate as polarized enterocytes. The Caco-2 model has a high predictive value for organic chemicals, covering diffusion and active transport through some membrane transporters, and in particular it is widely used by the pharmaceutical industry to predict oral absorption (Artursson et al., 2001). This model has also been used to evaluate the potential intestinal translocation of nanomaterials (Sonavane et al., 2008; Koeneman et al., 2010; Al-Jubory and Handy, 2013; Kenzaoui et al., 2012; Reix et al., 2012; Antunes et al., 2013; Jin et al., 2013; Janer et al., 2014; Zhu et al., 2016; Liu et al., 2016).

PT

Nevertheless, neither diffusion nor membrane transporters are considered the main pathways by which nanomaterials can be absorbed. The main pathways for absorption of particles across

RI

intestinal barriers are considered to be: 1) transcytosis across enterocytes and mostly across

SC

M-cells located in the Peyer’s patches (Powell et al., 2010); 2) paracellular transport through the tight junctions, only in the case of very small particles (< 5nm) (Ruenraroengsak et al., 2010); and 3) transport across degrading enterocytes (Hillyer and Albrecht, 2001; Volkheimer,

NU

1993). Therefore, the Caco-2 monolayer system would be missing the potential transcytosis across M-cells. Although these cells represent only a small proportion of the cells lining in the

MA

intestine, they have an important role in the absorption of (micro)particles (Florence et al., 1995; Kerneis et al., 1997), and it is to be expected that they are also a key player in the

D

intestinal absorption of nanomaterials (des Rieux, et al., 2007). Previous studies have shown that the in vitro co-culture system of Caco-2 cells with Raji B cells

PT E

(B lymphocytes derived from a human Burkitt cell lymphoma) induces the M-cell phenotype on Caco-2 cells (Kerneis et al., 1997; Gullberg et al., 2000; Martinez-Argudo et al., 2007). This in vitro model has been used to evaluate the translocation of pathogenic microorganism and

CE

peptide drugs (Kernéis et al., 1997; Martinez-Argudo et al., 2007; Albac et al., 2016), antigen delivery in the field of development of oral vaccines (Kim et al., 2010; Rochereau et al., 2013),

AC

and recently the permeability of nanoparticles (des Rieux et al., 2005; des Rieux et al., 2007; Bouwmeester et al., 2011, Loo et al., 2012; Schimpel et al., 2014; Brun et al., 2014, MacNicoll et al., 2015).

Based on the relevance of E171 intake (Weir et al., 2012; Rompelberg et al., 2016), with its corresponding fraction of nano-sized particles, TiO2 nanoparticles were selected as model NPs for this study. Another advantage of these materials for the translocation experiments is that, in comparison to other type of nanomaterials, TiO2 nanoparticles have very low dissolution rates. This facilitates the interpretation of the experimental results.

4

ACCEPTED MANUSCRIPT The translocation of TiO2 through Caco-2 differentiated cell monolayers has been previously evaluated and low translocation rates have been reported (Koeneman et al., 2010; Fisichella et al., 2012a; Janer et al., 2014). In addition, two recent studies evaluated the translocation of TiO2 NPs using the Caco-2/Raji B model (Brun et al., 2014; MacNicoll et al., 2015). Brun et al. (2014) showed intracellular accumulation of Ti in the Caco-2/Raji B co-culture by particleinduced X-ray emission, followed by analyses by µXRF and XAS that proved that titanium detected in cells was from anatase TiO2-NPs and not from dissolved Ti ions. In their study, the

PT

accumulation of Ti in the Caco-2 monoculture was insignificant. The differences between the two models were further confirmed by the evaluation of cross-sections of the cell layers by

RI

TEM, that showed the presence of TiO2 NP in the basolateral side of Caco-2/Raji B co-cultures

SC

but not in Caco-2 monocultures. However, the level of translocation could not be quantitatively evaluated in this study. MacNicool et al. (2015) evaluated translocation of TiO2 nanoparticles of different sizes on the Caco-2/Raji B model, and titanium levels in the basal

NU

compartment were always below their detection limits. However, the authors did not investigate the translocation of the particles through the inserts in the absence of a cellular

MA

layer. Depending on the aggregation of the particles in the test system (and possibly other factors such as the characteristics of the insert), the insert itself may prevent that

D

nanoparticles reach the basal compartment.

PT E

Inorganic NM produced at an industrial scale are usually provided as aggregated powdered form that needs to undergo a dispersion process before it can be tested in liquid media. Different types of dispersion methods (i.e.: agitation, sonication, wet milling…) can be employed with the aim to obtain nanoparticle suspensions as stable as possible during the

CE

experiments. Depending on the delivered amount of energy, the dispersion process breaks the pristine powdered NM in aggregates of different sizes, leading to a different specific surface

AC

area and agglomerate density, which in turn can modify NM dosimetry. In fact, the (primary and aggregate) size of a NM is one of the main extrinsic factor that can determine whether NM dispersions are stable or not over time. Therefore, the different NP aggregate size distribution obtained applying different dispersion protocols might be one of the factors contributing to some of the inconsistencies in in vitro toxicity data reported in the literature for some nanomaterials (Jiang et al., 2009; Vranic et al., 2016). However, the impact of different dispersion methods on toxicity or biokinetics of nanomaterials have been seldomly investigated (Hartmann et al., 2015).

5

ACCEPTED MANUSCRIPT The goals of this work are: 1) to obtain and characterize a Caco-2/M-cell co-culture model; 2) to compare the translocation of TiO2 NP in the Caco-2/M-cell model vs. that in a Caco-2 cell model; and 3) to evaluate the impact of some experimental variables (co-culture duration and nanoparticle dispersion protocol) on the resulting nanoparticle translocation.

PT

2. Materials and Methods 2.1. Nanoparticles

RI

The TiO2 NPs used were obtained from L’Urederra (Spain) in powder form. The synthesis process

SC

and the methods used for their physicochemical characterization were previously described (Janer et al., 2014). They are spherical with a diameter of 18 ± 8 nm and polymorph of anatase (79%) and rutile (21%). Previous BET analyses showed a specific surface area of 89.759 m2/g,

NU

with a corresponding particle diameter of 12.9 nm and a porosity of 0.0372 cm3/g (Janer et al., 2014).

MA

2.2. Cell culture

Caco-2 and Raji B cells were obtained from the ATTC. Caco-2 cells were grown in Dulbecco’s

D

modified Eagle medium (DMEM; Sigma-Aldrich, USA) and Raji B cells in RPMI medium (Sigma-

PT E

Aldrich, USA), both supplemented with 10% fetal calf serum (FCS; Biowest, France), 1% glutamine (Lonza, Spain) and antibiotics (1% penicillin and 1% streptomycin; Lonza, Spain). For maintenance, Caco-2 cells were cultured in flasks and sub-cultured once a week when they

CE

reached around 80% confluence. Raji B cells were grown in suspension and sub-cultured periodically, depending on their density. Caco-2 cells were differentiated to a monolayer of polarized cells in 3 µm transwell inserts (PET membrane, BD Falcon, USA) both for the Caco-2

AC

and Caco-2/Raji B conditions (see below). Cells were incubated at 37°C under 5% CO2 and water saturated atmosphere. 2.3. M-cell differentiation Caco-2 cells were seeded in the transwell inserts at 5x105 in 0.3 mL and grown for 14 days with medium changes every two or three days. On day 14 of culture, Raji B cells (4.44 x 105 cells in 0.9 mL in DMEM:RPMI (1:1)) were added to the basolateral compartment and left for 48 hours. Afterwards, the cell media was changed and Raji B cells were reseeded every 24 hours, to control the density of the Raji B cells and to avoid depletion of media nutrients. Two co-culture durations were tested. In the first one, Raji B cells were reseeded until co-culture day 4 (day 18 of Caco-2 6

ACCEPTED MANUSCRIPT culture), and later on cell media was continued to be changed daily until the end of the experiment (day 20 of Caco-2 culture), when inserts were collected and evaluated by SEM. In the second one, Raji B cells were reseeded until co-culture day 7 (day 21 of Caco-2 culture), and later on cell media was continued to be changed daily until the end of the experiment (day 23 of Caco2 culture), when inserts were collected and evaluated by SEM. Caco-2 cells were cultured in DMEM, but from the onset of the co-culture with Raji B cells, DMEM:RPMI (1:1) was used for the

PT

basal compartment. The transepithelial electric resistance (TEER) was daily evaluated during the whole co-culture

RI

period and until the end of the experiments. TEER was measured using a Millicel®ers reader (Merck Millipore, Germany) coupled to electrodes from World Precision Instruments Inc. (USA),

SC

after equilibration of cell culture plates at room temperature.

NU

2.4. SEM

Caco-2 monolayers cultured with and without Raji B cells were preserved in buffered glutaraldehyde (2.5%; pH 7.4) followed by serial dehydration through a series of ethanol

MA

increasing concentrations (30%, 50%, 70%, 90% and 100%). Samples were treated with Critical Point Drying (CPD) method and gold coated using a sputter coater (108 auto sputter coater,

D

Cressington Scientific Instruments Ltd., UK). Cell layers were examined using a scanning electron microscope (SEM) (JSM-6010 LV, JEOL Ltd., Japan) with an InTouchScopeTM operation system

PT E

(V1.06, JEOL Ltd., Japan). The frequency of M-cells in the cell layer was evaluated by counting cells with an M-morphology (reduced number, shorter and irregular microvilli) in around 400

and 7 days).

CE

cells for each combination of cell model (Caco-2 and Caco-2/Raji B) and co-culture duration (4

AC

2.5. Transport studies

To obtain the stock dispersion, the TiO2 nanoparticles were dispersed in MiliQ water (at 2 mg/mL) by probe sonication using a Labsonic sonicator (VCX750 Ultrasonic Cell Disrupter, SONICS Vibracell, USA). Two sonication protocols were evaluated: 1) Sonication Protocol A: One minute of sonication in continuous mode at 40% amplitude using a 3mm x 80 mm tip (BBraun, Spain) and 2) Sonication Protocol B: Fifteen minutes of sonication in continuous mode at 20% amplitude using a 6mm x 80 mm tip (BBraun, Spain). The energy applied during the sonication protocols A and B was calculated to be 5.4 and 20.5 J, respectively. The stock dispersion was diluted in cell culture media to obtain a 0.1 mg/mL dispersion that was applied on the apical side of the transwell inserts at the end of the co-culture period (Figure 1). This concentration was 7

ACCEPTED MANUSCRIPT selected because it did not induce toxicity in our previous experiments with Caco-2 cells (Janer et al., 2014). To evaluate whether nanomaterials could cross the polymeric membrane of the inserts, the same dispersion was also applied to the apical side of transwell inserts that had not been seeded with cells. Twenty-four and 48 hours after exposure, the basal compartment was collected to determine the translocation of TiO2 across the inserts, in the presence and in the absence of a cell

PT

monolayer. TEER values were measured at 24 and 48 hours after the exposure to evaluate the integrity of the cell layers. At the end of the experiment transwell inserts with cells were fixed

RI

and processed for SEM observation.

SC

In order to evaluate the relevance of the co-culture duration on the nanoparticle translocation, cells were exposed to the nanoparticles after the two co-culture protocols, as illustrated in Figure

NU

1. 2.6. Aggregate size distribution by DLS

MA

The aggregate size distribution of the nanoparticle dispersions in cell culture media was determined by DLS using a Malvern Zetasizer nano ZS (UK). Before size measurements, the dispersions were vortex-mixed to ensure their homogenization. Z-average diameter (i.e.,

D

mean intensity weighted mean diameter) and polydispersity index (PdI; i.e., broadness of the

PT E

size distribution calculated from the Cumulants Analysis) were obtained from DLS data expressed by intensity, in backscattering detection mode (scattering angle of 173°) and setting measurement duration on automatic. After 2 min of temperature equilibration at 25 °C, 1 ml

CE

volume was subjected to ten consecutive measurements and the average was reported. For samples characterized by a multimodal size distribution, the position of each peack was assigned based on the Distribution Analysis. In case that more than one size population were

AC

present, the relative amount of each peak was estimated from the volume size distribution, which was calculated applying the Mie theory and assuming a nanomaterial refractive index of 2.41 and an absorbance of 0.001. 2.7. ICP-MS analysis The Ti content was analyzed by ICP-MS (7500ce, Agilent, USA), monitoring the

47

Ti mass.

Preliminary experiments had shown that the salt content of the cell media used in the assays considerably decreased the sensitivity of the analysis by ICP-MS. Due to this reason a dialysis step (24-hour dialysis in a membrane with a molecular weight cut-off of 3500 Da, equivalent to < 2 nm, from Spectrum Labs, USA) was conducted prior to analysis. 8

ACCEPTED MANUSCRIPT Before ICP-MS analysis, an acidic digestion (hydrofluoric acid (HF) and nitric acid (HNO3) 1:20; Sigma-Aldrich, USA) with an analytical microwave (Mars, CEM; USA) was performed. This process results in the complete dissolution of the nanoparticles into metal ions. The temperature ramp used was 15 minutes to 190 °C and hold 10 min, and the RF power: 400, 800 or 1000 w depending on the number of samples. The quantification was performed by interpolation from a calibration curve prepared from a commercial certified titanium standard (1000 ppm Ti, Sigma-

PT

Aldrich, USA). The stock dispersions used in the exposure experiments were also dialyzed, acid digested and

RI

analysed to evaluate their titantium content.

SC

The limit of detection (LOD) and the limit of quantification (LOQ) were calculated based on the Ti background levels from blank digested samples used in each set of experiments analysed: LOD = averageblank + 3 x SDblank, LOQ = averageblank + 10 x SDblank. The calculated LOD ranged between

NU

0.007 and 0.016 and the LOQ ranged between 0.018 and 0.042 mg/L.

MA

2.8. Statistical analysis

All results are presented as mean and standard error of three independent experiments, each of them including two technical replicates (i.e., two cell inserts). For the graphical and statistical

D

analysis, titanium levels that were detected but were below the limit of quantification (LOQ)

PT E

were assigned an arbitrary value corresponding to the average between LOQ and LOD. Undetected values were assigned an arbitrary value corresponding to 0.5 x LOD. For the statistical analyses, titanium levels were log-transformed and a one-way ANOVA followed by

CE

post-hoc tests was used to compare exposed groups versus unexposed controls (Dunnett’s test) or to compare all groups (Tukey post hoc tests). A three-way ANOVA was performed to evaluate if any of the variables included in the study (cell model, co-incubation time, and sonication

AC

protocol) had a significant contribution to the overall variance in the data, and to identify possible interactions among the variables. All graphs and statistical analyses were done using Graphpad Prism® software (Prism 6.02 for Windows, La Jolla, USA), except for the three-way ANOVA that was performed in SPSS (SPSS Statistics for Windows version 17.0, Chicago, USA). Statistical significance was set at a p value < 0.05.

3. Results 3.1. Characterization of the Caco-2/M cell model

9

ACCEPTED MANUSCRIPT The morphology of Caco-2 cell monolayers co-incubated with Raji B cells for 4 and 7 days was observed and compared to Caco-2 cell monocultures by scanning electron microscopy (SEM). The Caco-2 cells were characterized by the presence of dense and relatively regular microvilli, whereas microvilli of the M cells were much fewer in number, irregular, and shorter (see Figure 2). Incidence of M-cells were of 2.6 and 2.0% for the 4-day and 7-day co-incubation periods. A low incidence (0.6 ± 0.2%) of cells with morphology consistent with M-cells were observed in Caco-2 monolayers that had not been co-cultured with Raji B cells. In some occasions, cells with

PT

such morphology were clearly undergoing cell division (Supplementary material S1).

RI

The TEER was used to evaluate the integrity of the Caco-2 cell and Caco-2/M-cell monolayers. A transient decrease of the TEER was observed after the inclusion of Raji B cells into the basal

SC

compartment. The decrease was marked during the first three days of co-incubation and the TEER recovered afterwards and reached the same levels as in the Caco-2 monoculture system.

NU

This trend did not depend on the length of the co-incubation system, and a similar pattern was observed for the 4 and 7-days set-ups (Figure 3).

MA

3.2. Characterization of nanomaterial dispersions

The two sonication protocols employed led to differences in the aggregate particle size of the

D

TiO2 dispersions, as evaluated by DLS (Figure 4). For sonication protocol A, the Z-average values obtained right after preparation of the working solution at 0.1 mg/mL were around 1.5-fold

PT E

larger than those obtained for sonication protocol B (338 nm vs. 229 nm, respectively). Particle size distribution curves showed a main size population peak centered at 396 and 295 nm,

CE

for sonication protocol A and B respectively. In both sonication protocols, the peak in the intensity distribution appeared broad and for protocol B (at t=0h) the main peak presented a

AC

shoulder at sizes lower than 100 nm. The intensity distribution is very sensitive to the presence of large aggregates (light intensity is proportional to the sixth power of particle diameter). Therefore, the volume distribution is also provided (Figure 4B) in order to elucidate whether the broad peaks in the intensity distribution were due to the presence of multiple peaks and to estimate their relative amount. The volume distribution curves obtained showed two main peaks. Applying protocol A, the relative amount of the first peak (at size 100nm) and the second peak (at 500-700 nm) was evaluated to be roughly 42 vol. % and 58 vol. %, respectively. Whereas for protocol B, the relative amount of the first and the second peaks was evaluated to be roughly 83 vol. % and 14 vol. %, respectively. Therefore, the differences in the Z-average parameter were attributed to a higher 10

ACCEPTED MANUSCRIPT fraction of aggregates of bigger sizes when sonication protocol A was applied, and a better dispersion with smaller aggregate sizes when sonication protocol B was applied (Figure 4B). This is consistent with the fact that the amount of energy provided by protocol A was around 4-fold lower than by protocol B (5.4 and 20.5 J, respectively). Both dispersions were relatively stable over time and Z-average values obtained after 24 hours were slightly lower than those recorded at time 0 (301 nm and 179 nm for sonication protocol A

PT

and B, respectively, representing a 11% and 22% decrease in the Z-average values over 24 h, respectively). This slight decrease was attributed to the precipitation of a fraction of the big

RI

aggregates during the temperature equilibration step, resulting in an apparent smaller Z-average

SC

size. 3.3. Translocation experiments

NU

The TEER values indicate that the integrity of the cell layer was not affected by the exposure of titanium dioxide nanoparticles at 100 µg/mL (see Figure 5). This is also supported by the SEM images of the monolayers, where no obvious changes in their morphology were identified

MA

(Figure 6). Some NP aggregates/agglomerates are observed in the images (Figure 6). The levels of titanium detected in the basal compartment of the wells with cell-free insert

D

membranes during the 48 hour-period (sum of values at 24h and 48h) corresponded

PT E

approximately to half of the titanium added to the apical compartment. This indicates that the titanium dioxide nanoparticles could move through the cell-free insert membranes. No differences between the two sonication protocols were observed: 50.6 ± 1.5 % and 52.6 ± 2.9 %

CE

of the titanium in the basal compartment for sonication protocol A and B, respectively. These values are slightly below the expected concentration if a totally free diffusion across the membrane would take place, which would be 75% considering that the volume in the basal

AC

compartment is 0.9 mL and that the volume in the apical compartment is 0.3 mL. The titanium levels in the basal compartment of wells with unexposed cells and exposed Caco-2 cells were very low, often below the detection limits of the analytical method. The exposed groups were compared to the unexposed controls using one-way ANOVA with Dunnett post-hoc tests. The average titanium levels observed in the basal compartment were statistically higher than in the unexposed controls when Raji B cells were present in the 4-day coculture period, regardless of the sonication protocol (p< 0.05) (Figure 7A). In order to evaluate the effect on titanium dioxide translocation of the different variables investigated in the study (i.e. cell model, sonication protocol, and co-culture duration) a three-way ANOVA was 11

ACCEPTED MANUSCRIPT performed and did not show any significant three- or two-way interactions. Cell model was identified as a significant source of variance (p<0.05), but no significant contribution to the variance was found for the sonication protocol or the co-incubation period. Since cell model included only two groups (Caco-2 and Caco-2/Raji B), it can be derived from these analyses that titanium levels were significantly higher in the Caco-2/Raji B model than in the Caco-2 model. Considering that according to this analysis the co-incubation period and the sonication protocol were not statistically significant factors, the data was pooled for the Caco-2 and Caco-

PT

2/Raji B exposed to TiO2 (Figure 7B). A one-way ANOVA on these data confirmed that the titanium levels in the basal compartment of the Caco-2/Raji B exposed cells were significantly

SC

RI

higher than those in Caco-2 exposed cells and in unexposed cells (Figure 7B).

NU

4. Discussion

There is a need to establish a robust in vitro system to predict systemic exposure after oral exposure to nanomaterials. In the present study, an in vitro Caco-2 / M-cell permeability model

MA

was developed and used to investigate the contribution of M-cells in the translocation of TiO2 NP. In addition, two different nanoparticle dispersion protocols and two co-incubation periods

D

were compared in terms of TiO2 NP translocation.

PT E

Previous reports had described the development of a Caco-2/M cell model (Gullberg et al., 2000; Martinez-Argudo et al., 2007; des Rieux et al., 2007; Sae-Hae et al., 2010; Masuda et al., 2011; Bouwmeester et al., 2011; Rochereau et al., 2013; Loo et al., 2012; Araújo and Sarmento, 2013;

CE

Antunes et al., 2013). However, these reports differ in the experimental conditions used to induce the M cell phenotype (e.g., onset of the co-culture, co-culture duration, density of cells seeded, use of inverted insert, etc). In addition, the methodology is generally described with

AC

rather limited detail. For example, there is rarely information on whether Raji B cells are reseeded during the co-culture period or how often cell media is renewed. Taking into consideration that these cells grow in suspension, these are important parameters. The cell media used during the co-cultures are also not always reported. Again this is not a trivial decision considering that Caco-2 cells and Raji B cells are commonly grown on different cell media. This study reports with detail the methodology used. In addition, it investigates the impact of the duration of the co-incubation period, and shows no remarkable differences between co-culture durations of 4 and 7 days.

12

ACCEPTED MANUSCRIPT SEM studies showed that a similar proportion of Caco-2 cells differentiated into M-cells in the two co-culture set-ups of this study (2.6 vs. 2.0% for the 4-day and 7-day co-cultures, respectively). The fraction of cells converted to M-cells was rather small, but this is consistent with previous reports on this model (e.g., Liang et al., 2002), and with the frequency of M-cells reported in human small intestine, i.e.: 3 to 10% of the follicular-associated epithelium cells (Cuvelier et al., 1994; Jepson et al., 1998; Giannasca et al., 1999), corresponding to <1% of Mcells in the whole intestinal surface. Some studies introduced modifications of the method

PT

(inversion of inserts) to increase the differentiation rate, reaching values of 23% (des Rieux et al.,

RI

2007). However, such steps are tedious and introduce variability across laboratories. The co-culture duration in previous reports ranges from 2 (Sae-Hae et al., 2010) to 7 days (Araújo

SC

et al., 2013), with most studies using an intermediate duration of 4 or 5 days (Gulberg et al., 2000; des Rieux et al., 2007; Masuda et al., 2011; Bouwmeester et al., 2011; Rochereau et al.,

NU

2011; Antunes et al., 2013). Our first co-culture duration (4 days) was selected to represent a commonly used duration in previous reports, whereas the second was selected to allow the recovery of the TEER values prior to the translocation experiments. The decrease in TEER values

MA

observed after the onset of the co-incubation had been previously described in other studies (Rochereau et al., 2013, des Rieux et al., 2007, Araújo and Sarmento, 2013, Schimpel et al.,

D

2014). This could be related to the reorganization of the cell layer during the dedifferentitation of some enterocyte-phenotype cells and the later differentiation into M-cell phenotype (Gullberg,

PT E

2005). In contrast to previous reports, we observed a consistent recovery of TEER values after three or four days of co-incubation, reaching the initial TEER levels around co-incubation day 6. There are several differences in the experimental design that could explain why previous studies

CE

had not reported such recovery in TEER values. For example, some of the previous reports started the co-culture at a much earlier stage than we did (Araújo and Sarmento, 2013;

AC

Rochereau et al., 2013), others used inverted inserts (des Rieux et al., 2007), or only reported final TEER values at the end of their experiment, a time point at which TEER values may have not yet totally recovered (Schimpel et al., 2014). While conducting this study we had to reject several assays because of no proliferation of Raji B cells in the basal compartment and/or abnormal evolution of TEER values at the onset of the coculture. The accurate control of the Raji B cell density in the stock culture and associated proliferation rate did improve the success of the assays. Reproducibility issues with this assay have been previously reported (Martinez-Argudo et al., 2007). These authors explained that controlling the Caco-2 passage was a relevant factor to improve reproducibility of the assay. Reproducibility issues with the assay may also be one of the reasons behind the methodological 13

ACCEPTED MANUSCRIPT differences between different labs and the relatively low number of publications with this model, despite of its extremely high interest in several areas (e.g., nanomaterials, vaccines). The evaluation of cellular morphology by SEM offers a rather clear way to identify the presence of M-cells in the co-culture. However, this technique is relatively costly and time consuming. The identification of M-cells by means of fluorescence molecular markers would be an approach of easiest implementation by most research laboratories. However, the molecular markers that

PT

have been proposed so far (such as occludin antibodies, actin related markers, expression of enzymes like phosphatase alkaline, M-cell receptors and their ligands, etc.) show only

RI

qualitative differences between M-cells and Caco-2 cells and cannot be used to unequivocally

SC

identify M-cells.

Our study shows a statistically significant translocation of TiO2 nanoparticles in the Caco-2/Raji B co-culture model, but not in the Caco-2 monoculture model. The lack of significant translocation

NU

through the Caco-2 cell model is in line with previous reports that have also shown undetectable or very limited translocation of TiO2 NP across the Caco-2 cell barrier by ICP-MS (Janer et al.,

MA

2014; Song et al., 2015). This limited translocation is further supported by studies looking at internalization of TiO2 NP by differentiated Caco-2 cells. No internalization of TiO2 by differentiated Caco-2 cells was detected by TEM in the studies by Fisichella et al. (2012a) and by

D

Song et al. (2015). Brun et al. (2014) did observe TiO2 NP entrapped in cytoplasmic vesicles

PT E

nearby the apical membrane of exposed Caco-2 cells by TEM. But they did not observe complete translocation of NP (Brun et al., 2014). Koeneman et al. (2010) also reported the presence of TiO2 inside cells and underneath them by scanning confocal microscopy, although in a very low

CE

percentage of cells (approximately 0.4% at the concentration of 100 µg/mL). They also quantified titanium levels after exposing cells to different TiO2 NP concentrations and reported that 14.4% of the titanium was present in the basolateral compartment (measured by graphite furnace

AC

atomic absorption spectroscopy) at the exposure concentration of 100 µg/mL. However, they did not report the titanium concentration in unexposed wells, and it is unclear up to which extend the concentration of titanium reported differed from background levels. The increase in titanium levels in the basolateral chamber of the Caco-2/Raji B model supports a relevant role of M-cells in the intestinal translocation of TiO2 NP, as it had been previously suggested by Brun et al., 2014. In their study, intracellular agglomerates of TiO2 NP were more frequently observed in the Caco-2/Raji B co-culture than in the Caco-2 model. In addition, some agglomerates were observed between the basolateral Caco-2/Raji B cell membrane and the surface of the transwell (Brun et al., 2014), providing evidence that translocation has occurred. 14

ACCEPTED MANUSCRIPT MacNicoll et al. (2015) also investigated translocation of TiO2 NP through Caco-2/Raji B cocultures. They did not observe translocation of TiO2 particles of different sizes (<25 nm to 5 µm) by single particle ICP-MS (sp-ICP-MS), despite a rather low detection limit (<1 µg/L; McNicoll et al., 2015). The reasons for these differences are unclear, but might be related to the fact that they used an inverted co-culture model to induce differentiation of Caco-2 cells into M-cells. This is the first study using inorganic nanomaterials that has quantitatively compared

PT

translocation between the Caco-2 and the Caco-2/Raji B models. A high variability was observed in the measured titanium levels, but on average the titanium values in the basal compartment

RI

were 5-fold higher in the Caco-2/Raji B model than in the Caco-2 model. This value is likely to underestimate the real differences in translocation, because the titanium levels measured in the

SC

basolateral compartment of the Caco-2 model did not significantly differ from the titanium background levels. In previous reports with organic micro- and nanoparticles, a higher

NU

translocation (2- to 50-fold, depending on the study) was observed when M-cells were present than in the Caco-2 monoculture model (des Rieux et al.,2007; Rochereau et al., 2013; Martinez-

MA

Argudo et al., 2007; Schimpel et al., 2014).

We did not observe decreases in the TEER values in the cells exposed to the TiO2 NPs. This indicates that the TiO2 NP did not disrupt the tight junctions of the cell monolayers. This is

D

consistent with previous reports showing that exposure to similar TiO2 NP concentrations (50 to

PT E

250 µg/mL) did not decrease TEER values in Caco-2 (Koeneman et al., 2010; Brun et al., 2014; Jones et al., 2015) and in Caco-2/Raji B cell monolayers (McNicoll et al., 2015). Indeed, in some of these reports, the TEER values were somewhat higher in exposed cell monolayers than in control

CE

cells.

Besides the lack of effects on TEER values, no morphological effects induced by TiO2 NP were

AC

observed by SEM. Other studies had previously investigated the effects of TiO2 NP exposure on Caco-2 cell morphology with conflicting results (Koeneman et al., 2010; Fisichella et al., 2012a; Faust et al., 2014). Similar to our observations, Fisichella et al. (2012a) reported intact Caco-2 cell morphology after exposure to 100 µg/mL coated TiO2 NP for 72h. In contrast, Koeneman et al. (2010) and Faust et al., (2014) reported disruption of the microvillar organization of Caco-2 cells after TiO2 NP and E171 exposure, respectively. Koeneman et al. (2010) observed that microvilli decreased in number, no longer stood erect, and were no longer cylindrically shaped at TiO2 NP concentrations as low as 10 µg/mL after 24h incubation. At higher concentrations (1000 µg/mL) the microvilli appeared to be reabsorbed by the cells (Koeneman et al., 2010). Faust et al. (2014) reported that food grade TiO2 at 350 ng/mL caused disruption of the erect microvilli morphology 15

ACCEPTED MANUSCRIPT as well as a reduction in the total number of microvilli. The different types of TiO2 NP used in these studies may partly explain the reported differences on cell morphology effects, as suggested in letters to the editor exchanged by the two research groups (Faust et al., 2012; Fisichella et al., 2012b). Another apparent difference between these studies is the type of Caco-2 cells used. In our study and in that of Fisichella et al. (2012a), common Caco-2 cells were used, whereas a special clone of Caco-2: the brush border expressing clone C2BBe1 was used in the two studies reporting morphological effects (Koeneman et al., 2010 and Faust et al., 2012).

PT

According to the ATCC, this clone was selected “on the basis of morphological homogeneity and exclusive apical villin localization, and are more homogeneous than the parental Caco-2 cell line

RI

with respect to BB expression”. The SEM images for unexposed controls shown by Fisichella et al.

SC

(2012a) and Faust et al., (2014) are considerably different. Indeed, the images for unexposed controls shown by Fisichella et al. (2012a) are rather similar to the exposed cells in Faust et al. (2014). In our study we observed both type of morphologies in the control preparations (see

NU

Figure 2B and Figure S2 in the Supplementary material) and in the exposed preparations (wee Figure 6A). The SEM morphology of the Caco-2C2BBe1 cell line may be more homogeneous than of

MA

the parental Caco-2 cell line and facilitate the identification of subtle morphological effects. Nevertheless, marked morphological effects such as those shown by Koeneman et al., (2010) after an acute exposure to 1000 µg/mL and after a chronic exposure to 10 µg/mL would have

D

definitely been identified in our study. Further studies with systematic and objective

PT E

morphological evaluations are needed to understand the effects of nanoparticle exposure on intestinal cell morphology. The doses used in our study (100 µg/mL equivalent to 33 µg/mL) are more than two orders of magnitude above the upper range of estimated human exposure

CE

(McCracken et al., 2016). But the doses that were associated to morphological effects in the study by Faust et al. (2014) are much closer these estimated human exposures.

AC

The physiological relevance of the Caco-2/M-cell model could be further improved by including mucus producing cells. When HT29-MTX cells (to mimic Goblet cells) are added to the Caco-2 cell model, the presence of the mucus producing cells disrupts the enterocytes tight junctions, increasing the permeability of the overall cell layer (Walter et al., 1996; Behrens et al., 2001). In addition, nanoparticles may interact with the mucus layer, changing their retention time in the intestine surface and modulating their uptake. Some reports have used the Caco-2/HT29-MTX and/or the Caco-2/HT29-MTX/Raji B model to investigate intestinal absorption of polystyrene nanoparticles, showing that cell internalization and translocation are modulated by the presence of mucus producing cells (Mahler et al., 2012; Walczak et al., 2015).

16

ACCEPTED MANUSCRIPT Another factor that is expected to modulate oral absorption of nanomaterials is their aggregated size. For inorganic nanomaterials of industrial use, which are commonly provided in powder form, a dispersion step is needed prior to their incorporation into a test system. Differences in NP dispersion have been suggested to be one of the causes of discrepancies among different in vitro and in vivo studies investigating toxicity of nanoparticles (Vranic et al., 2016). Some initiatives have proposed a sonication method as a standard protocol to facilitate comparison among different toxicity studies (e.g., the EU FP7 project NanoReg, Taurozzi et al., 2013,

PT

Hartmann et al., 2015). Although having a ‘default’ sonication method to be used in all studies can facilitate comparison of data, it could also lead to artifacts in the assessment of specific

RI

nanoparticles for which the method is too aggressive or is not sufficient to achieve a stable

SC

suspension (Hartmann et al., 2015). Therefore, it might be more appropriate to establish clear criteria to optimize and standardize dispersion methods for each combination of nanoparticle and matrix, and, at the same time, design studies to understand the real consequences of using

NU

different dispersion methods on the final toxicity results. Indeed, a limited number of studies have investigated up to which extend the differences in dispersion methods affect the toxicity

MA

results (Magdolenova et al., 2012; Carrière et al., 2014). Our study evaluated the impact of two different sonication protocols on the size of the TiO2 nanoparticle aggregates, and on their consequent translocation in two in vitro models. One of the sonication protocols we used

D

(sonication protocol B) followed the protocol selected by the Nanopolytox EU FP7-funded project

PT E

(15 min sonication with a 6 mm horn at 20% amplitude), while sonication protocol A used a smaller size horn, a shorter duration and a higher amplitude to deliver around 4-fold higher energy (1 min sonication with a 3 mm horn at 40% amplitude). The two sonication protocols led

CE

to differences in the NM-aggregate size distribution, which did not result on statistically significant differences in the observed translocation of TiO2 NPs. The relatively high variability in the titanium levels observed might have precluded the identification of differences in

AC

translocation between groups.

4.1. Conclusions M-cells are considered to have an important role in the translocation of nanoparticles through the intestinal barrier. Consistently with previous reports for other nanoparticles (mostly organic), we observed a higher translocation of TiO2 NPs through a Caco-2/M-cell model than in a Caco-2 monoculture model. Indeed, the Ti levels in the basolateral compartment of the Caco-2 monoculture model did not differ from background Ti levels. The Caco-2/M-cell co-culture model 17

ACCEPTED MANUSCRIPT has potential to become the in vitro model of choice to predict the intestinal absorption of nanomaterials. However, there are still a limited number of studies that have used this model and usually methodological details reported on the model itself are scarce and not standardized across studies. We have shown that the duration of the co-culture period had no relevant impact on the observed nanoparticle translocation through the model, but it is still unclear how other parameters differing among studies could affect the results (e.g., inverted vs. non-inverted inserts, onset of the co-culture, cell density). Therefore, an in depth understanding of the co-

PT

culture model and the role of its different parameters, followed by the establishment of a standardized protocol with tools to characterize the resulting co-culture are needed in order to

RI

compare data from different laboratories. In addition to the particularities of the model, the

SC

impact of experimental variables during the preparation of nanomaterial dispersions also need

NU

to be taken into account.

Aknowledgements

MA

We would like to thank Dr. B. Guerrero for the ICP-MS analyses and Dr. J.L. Muñoz for the calibration of the sonicator tips and calculation of the energy delivered by the sonication

D

protocols. The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/2007-2013) under the GUIDEnano Project

CE

References

PT E

(Grant Agreement No. 604387).

Albac S, Schmitz A, Lopez-Alayon C, d'Enfert C, Sautour M, Ducreux A, Labruère-Chazal C, Laue

AC

M, Holland G, Bonnin A, Dalle F. Candida albicans is able to use M cells as a portal of entry across the intestinal barrier in vitro. Cell Microbiol. 2016; 18(2): 195-210. Al-Jubory AR, Handy RD. Uptake of titanium from TiO(2) nanoparticle exposure in the isolated perfused intestine of rainbow trout: nystatin, vanadate and novel CO(2)-sensitive components. Nanotoxicology. 2013; 7(8): 1282-301. Antunes F, Andrade F, Araújo F, Ferreira D, Sarmento B. Establishment of a triple co-culture in vitro cell models to study intestinal absorption of peptide drugs. Eur J Pharm Biopharm. 2013; 83(3): 427-35. Araújo F, Sarmento B. Towards the characterization of an in vitro triple co-culture intestine cell model for permeability studies. Int J Pharm. 2013; 458(1): 128-34. 18

ACCEPTED MANUSCRIPT Artursson P, Palm K, Luthman K. Caco-2 monolayers in experimental and theoretical predictions of drug transport. Adv Drug Deliv Rev. 2001; 46(1-3): 27-43. Athinarayanan J, Periasamy VS, Alsaif MA, Al-Warthan AA, Alshatwi AA. Presence of nanosilica (E551) in commercial food products: TNF-mediated oxidative stress and altered cell cycle progression in human lung fibroblast cells. Cell Biol Toxicol. 2014; 30(2): 89-100. Baek M, Chung HE, Yu J, Lee JA, Kim TH, Oh JM, Lee WJ, Paek SM, Lee JK, Jeong J et al. 2012. Pharmacokinetics, tissue distribution, and excretion of zinc oxide nanoparticles. Int J

PT

Nanomedicine 7:3081-97.

Bannunah AM, Vllasaliu D, Lord J, Stolnik S. Mechanisms of nanoparticle internalization and

RI

transport across an intestinal epithelial cell model: effect of size and surface charge.

SC

Mol Pharm. 2014; 11(12): 4363-73.

Behrens I, Stenberg P, Artursson P, Kissel T. Transport of lipophilic drug molecules in a new mucus-secreting cell culture model based on HT29-MTX cells. Pharm Res. 2001; 18(8):

NU

1138-45.

Bouwmeester H, Poortman J, Peters RJ, Wijma E, Kramer E, Makama S, Puspitaninganindita K,

MA

Marvin HJ, Peijnenburg AA, Hendriksen PJ. Characterization of translocation of silver nanoparticles and effects on whole-genome gene expression using an in vitro intestinal epithelium coculture model. ACS Nano. 2011; 5(5): 4091-103.

D

Brun E, Barreau F, Veronesi G, Fayard B, Sorieul S, Chanéac C, Carapito C, Rabilloud T,

PT E

Mabondzo A, Herlin-Boime N, Carrière M. Titanium dioxide nanoparticle impact and translocation through ex vivo, in vivo and in vitro gut epithelia. Part Fibre Toxicol. 2014; 11:13.

CE

Carrière M, Pigeot-Remy S, Casanova A, Dhawan A, Lazzaroni JC, Guillard C, Herlin-Boime N. Impact of Titanium Dioxide Nanoparticle Dispersion State and Dispersion Method on Their Toxicity Towards A549 Lung Cells and Escherichia coli Bacteria.J. Translational

AC

Toxicology, 2014, 1, pp.10-20. Cuvelier CA, Quatacker J, Mielants H, De Vos M, Veys E, Roels HJ. M-cells are damaged and increased in number in inflamed human ileal mucosa. Histopathology. 1994; 24(5): 417-26. Dekkers S, Krystek P, Peters RJ, Lankveld DP, Bokkers BG, van Hoeven-Arentzen PH, Bouwmeester H, Oomen AG. Presence and risks of nanosilica in food products. Nanotoxicology. 2011; 5(3): 393-405. des Rieux A, Fievez V, Théate I, Mast J, Préat V, Schneider YJ. An improved in vitro model of human intestinal follicle-associated epithelium to study nanoparticle transport by M cells. Eur J Pharm Sci. 2007; 30(5): 380-91. 19

ACCEPTED MANUSCRIPT des Rieux A, Ragnarsson EG, Gullberg E, Préat V, Schneider YJ, Artursson P. Transport of nanoparticles across an in vitro model of the human intestinal follicle associated epithelium. Eur J Pharm Sci. 2005; 25(4-5): 455-65. EFSA, 2004. Question N° EFSA-Q-2004-103. The EFSA Journal 163:1-12. Eskandari S, Varamini P, Toth I. Formulation, characterization and permeability study of nano particles of lipo-endomorphin-1 for oral delivery. J Liposome Res. 2013; 23(4): 311-7. Faust JJ, Zhang W, Koeneman BA, Chen Y, Capco DG. Commenting on the effects of surface

PT

treated- and non-surface treated TiO(2) in the Caco-2 cell model. Part Fibre Toxicol. 2012; 9: 42.

RI

Faust JJ, Doudrick K, Yang Y, Westerhoff P, Capco DG. Food grade titanium dioxide disrupts

SC

intestinal brush border microvilli in vitro independent of sedimentation. Cell Biol Toxicol. 2014; 30(3): 169-88.

Fisichella M, Berenguer F, Steinmetz G, Auffan M, Rose J, Prat O. Intestinal toxicity evaluation

NU

of TiO2 degraded surface-treated nanoparticles: a combined physico-chemical and toxicogenomics approach in caco-2 cells. Part Fibre Toxicol. 2012a; 9:18.

MA

Fisichella M, Bérenguer F, Steinmetz G, Auffan M, Rose J, Prat O. Reply to comment on Fisichella et al. (2012), "Intestinal toxicity evaluation of TiO2 degraded surface-treated nanoparticles: a combined physico-chemical and toxicogenomics approach in Caco-2

D

cells" by Faust et al. Part Fibre Toxicol. 2012b; 9: 39.

PT E

Florence AT, Hillery AM, Hussain N, Jani PU. Factors affecting the oral uptake and translocation of polystyrene nanoparticles: histological and analytical evidence. J Drug Target. 1995; 3(1): 65-70.

CE

Giannasca PJ, Giannasca KT, Leichtner AM, Neutra MR. Human intestinal M cells display the sialyl Lewis A antigen. Infect Immun. 1999; 67(2): 946-53. Gitrowski C, Al-Jubory AR, Handy RD. Uptake of different crystal structures of TiO₂

AC

nanoparticles by Caco-2 intestinal cells. Toxicol Lett. 2014; 226(3): 264-76. Gullberg E, Leonard M, Karlsson J, Hopkins AM, Brayden D, Baird AW, Artursson P. Expression of specific markers and particle transport in a new human intestinal M-cell model. Biochem Biophys Res Commun. 2000; 279(3): 808-13. Erratum in: Biochem Biophys Res Commun 2001; 281(4): 1063. Gullberg, E. Particle transcytosis across the human intestinal epithelium. Thesis. 2005. http://publications.uu.se/theses/spikblad.xsql?dbid=4780. Hartmann NB, Jensen KA, Baun A, Rasmussen K, Rauscher H, Tantra R, Cupi D,Gilliland D, Pianella F, Riego Sintes JM. Techniques and Protocols for Dispersing Nanoparticle

20

ACCEPTED MANUSCRIPT Powders in Aqueous Media-Is there a Rationale for Harmonization? J Toxicol Environ Health B Crit Rev. 2015; 18(6): 299-326. He B, Lin P, Jia Z, Du W, Qu W, Yuan L, Dai W, Zhang H, Wang X, Wang J, Zhang X, Zhang Q. The transport mechanisms of polymer nanoparticles in Caco-2 epithelial cells. Biomaterials. 2013; 34(25): 6082-98. Hillyer JF, Albrecht RM. Gastrointestinal persorption and tissue distribution of differently sized colloidal gold nanoparticles. J Pharm Sci. 2001; 90(12): 1927-36.

PT

Janer G, Mas del Molino E, Fernández-Rosas E, Fernández A, Vázquez-Campos S. Cell uptake and oral absorption of titanium dioxide nanoparticles. Toxicol Lett. 2014; 228(2): 103-

RI

10.

SC

Jepson MA, Clark MA. Studying M cells and their role in infection. Trends Microbiol. 1998 Sep;6(9):359-65.

Jiang JK, Oberdorster G, Biswas P. Characterization of size, surface charge, and agglomeration

NU

state of nanoparticle dispersions for toxicological studies. Nanoparticle Research. 2009; 11: 77–89.

MA

Jin X, Zhang ZH, Li SL, Sun E, Tan XB, Song J, Jia XB.. A nanostructured liquid crystalline formulation of 20(S)-protopanaxadiol with improved oral absorption. Fitoterapia. 2013; 84: 64-71.

D

Jones K, Morton J, Smith I, Jurkschat K, Harding AH, Evans G. Human in vivo and in vitro studies

PT E

on gastrointestinal absorption of titanium dioxide nanoparticles. Toxicol Lett. 2015 Mar 4;233(2):95-101.

Kenzaoui BH, Vila MR, Miquel JM, Cengelli F, Juillerat-Jeanneret L. Evaluation of uptake and

CE

transport of cationic and anionic ultrasmall iron oxide nanoparticles by human colon cells. Int J Nanomedicine. 2012; 7: 1275-86. Kerneis S, Bogdanova A, Kraehenbuhl JP, Pringault E. Conversion by Peyer's patch lymphocytes

AC

of human enterocytes into M cells that transport bacteria. Science. 1997; 277(5328): 949-52.

Kim SH, Seo KW, Kim J, Lee KY, Jang YS. The M cell-targeting ligand promotes antigen delivery and induces antigen-specific immune responses in mucosal vaccination. J Immunol. 2010; 185(10): 5787-95. Koeneman BA, Zhang Y, Westerhoff P, Chen Y, Crittenden JC, Capco DG. Toxicity and cellular responses of intestinal cells exposed to titanium dioxide. Cell Biol Toxicol. 2010; 26(3): 225-38

21

ACCEPTED MANUSCRIPT Li CH, Shen CC, Cheng YW, Huang SH, Wu CC, Kao CC, Liao JW, Kang JJ. Organ biodistribution, clearance, and genotoxicity of orally administered zinc oxide nanoparticles in mice. Nanotoxicology. 2012; 6(7): 746-56. Lin IC, Liang M, Liu TY, Ziora ZM, Monteiro MJ, Toth I. Interaction of densely polymer-coated gold nanoparticles with epithelial Caco-2 monolayers. Biomacromolecules. 2011; 12(4): 1339-48.

oral insulin delivery. Drug Deliv. 2016; 23(6): 2015-25.

PT

Liu C, Shan W, Liu M, Zhu X, Xu J, Xu Y, Huang Y. A novel ligand conjugated nanoparticles for

Loo Y, Grigsby CL, Yamanaka YJ, Chellappan MK, Jiang X, Mao HQ, Leong KW. Comparative

SC

models. J Control Release. 2012; 160(1): 48-56.

RI

study of nanoparticle-mediated transfection in different GI epithelium co-culture

MacNicoll A, Kelly M, Aksoy H, Kramer E, Bouwmeester H, Chaudhry Q. A study of the uptake and biodistribution of nano-titanium dioxide using in vitro and in vivo models of oral

NU

intake. Journal of Nanoparticle Research. 2015; 17:66.

Magdolenova Z, Bilaničová D, Pojana G, Fjellsbø LM, Hudecova A, Hasplova K, Marcomini A,

MA

Dusinska M. Impact of agglomeration and different dispersions of titanium dioxide nanoparticles on the human related in vitro cytotoxicity and genotoxicity. J Environ Monit. 2012; 14(2): 455-64.

D

Mahler GJ, Esch MB, Tako E, Southard TL, Archer SD, Glahn RP, Shuler ML. Oral exposure to

12;7(4):264-71.

PT E

polystyrene nanoparticles affects iron absorption. Nat Nanotechnol. 2012 Feb

Martinez-Argudo I, Sands C, Jepson MA. Translocation of enteropathogenic Escherichia coli

CE

across an in vitro M cell model is regulated by its type III secretion system. Cell Microbiol. 2007; 9(6): 1538-46. Masuda K, Kajikawa A, Igimi S. Establishment and Evaluation of an in vitro M Cell Model using

AC

C2BBe1 Cells and Raji Cells. Biosci Microflora. 2011; 30(2): 37-44. McCracken C, Dutta PK, Waldman WJ. Critical assessment of toxicological effects of ingested nanoparticles. Environ Sci Nano. 2016; 3(2): 256-282. Peters RJ, van Bemmel G, Herrera-Rivera Z, Helsper HP, Marvin HJ, Weigel S, Tromp PC, Oomen AG, Rietveld AG, Bouwmeester H. Characterization of titanium dioxide nanoparticles in food products: analytical methods to define nanoparticles. J Agric Food Chem. 2014; 62(27): 6285-93. Powell JJ, Faria N, Thomas-McKay E, Pele LC. Origin and fate of dietary nanoparticles and microparticles in the gastrointestinal tract. J Autoimmun. 2010; 34(3): J226-33.

22

ACCEPTED MANUSCRIPT Reix N, Parat A, Seyfritz E, Van der Werf R, Epure V, Ebel N, Danicher L, Marchioni E, Jeandidier N, Pinget M et al. In vitro uptake evaluation in Caco-2 cells and in vivo results in diabetic rats of insulin-loaded PLGA nanoparticles. Int J Pharm. 2012; 437(1-2): 213-20. Rochereau N, Drocourt D, Perouzel E, Pavot V, Redelinghuys P, Brown GD, Tiraby G, Roblin X, Verrier B, Genin C, Corthésy B, Paul S. Dectin-1 is essential for reverse transcytosis of glycosylated SIgA-antigen complexes by intestinal M cells. PLoS Biol. 2013; 11(9): e1001658.

PT

Rompelberg C, Heringa MB, van Donkersgoed G, Drijvers J, Roos A, Westenbrink S, Peters R, van Bemmel G, Brand W, Oomen AG. Oral intake of added titanium dioxide and its

RI

nanofraction from food products, food supplements and toothpaste by the Dutch

SC

population. Nanotoxicology. 2016; 13: 1-11.

Ruenraroengsak P, Cook JM, Florence AT. Nanosystem drug targeting: Facing up to complex realities. J Control Release. 2010; 141(3): 265-76.

NU

Sae-Hae K, Seo KW, Kim J, Lee KY, Jang YS. The M cell-trageting ligand promotes antigen delivery and induces antigen-specific immune responses in mucosal vaccination. J

MA

Immunol. 2010; 185: 5787-5795.

SCF (Scientific Committee for Food), 1975. Reports of the Scientific Committee for Food: first series,

p.

17.

Available

online:

D

http://ec.europa.eu/food/fs/sc/scf/reports/scf_reports_01.pdf

PT E

Schimpel C, Teubl B, Absenger M, Meindl C, Fröhlich E, Leitinger G, Zimmer A, Roblegg E. Development of an advanced intestinal in vitro triple culture permeability model to study transport of nanoparticles. Mol Pharm. 2014; 11(3): 808-18.

CE

Schleh C, Semmler-Behnke M, Lipka J, Wenk A, Hirn S, Schaffler M, Schmid G, Simon U, Kreyling WG. Size and surface charge of gold nanoparticles determine absorption across intestinal barriers and accumulation in secondary target organs after oral

AC

administration. Nanotoxicology. 2012; 6(1): 36-46. Sonavane G, Tomoda K, Sano A, Ohshima H, Terada H, Makino K. In vitro permeation of gold nanoparticles through rat skin and rat intestine: effect of particle size. Colloids Surf B Biointerfaces. 2008; 65(1): 1-10. Song ZM, Chen N, Liu JH, Tang H, Deng X, Xi WS, Han K, Cao A, Liu Y, Wang H. Biological effect of food additive titanium dioxide nanoparticles on intestine: an in vitro study. J Appl Toxicol. 2015; 35(10): 1169-78. Taurozzi JS, Hackley VA, Wiesner MR. A standardised approach for the dispersion of titanium dioxide nanoparticles in biological media. Nanotoxicology. 2013; 7(4): 389-401. Volkheimer G. [Persorption of microparticles]. Pathologe. 1993; 14(5): 247-52. 23

ACCEPTED MANUSCRIPT Vranic S, Gosens I, Jacobsen NR, Jensen KA, Bokkers B, Kermanizadeh A, Stone V, BaezaSquiban A, Cassee FR, Tran L, Boland S. Impact of serum as a dispersion agent for in vitro and in vivo toxicological assessments of TiO(2) nanoparticles. Arch Toxicol. 2016 Feb 12. Walczak AP, Kramer E, Hendriksen PJM, Tromp P, Helsper JPFG, van der Zande M, Rietjens IMCM, Bouwmeester H. Translocation of differently sized and charged polystyrene nanoparticles in in vitro intestinal cell models of increasing complexity. Nanotoxicology

PT

2015; 9(4): 453–61.

Walter E, Janich S, Roessler BJ, Hilfinger JM, Amidon GL. HT29-MTX/Caco-2 cocultures as an in

RI

vitro model for the intestinal epithelium: in vitro-in vivo correlation with permeability

SC

data from rats and humans. J Pharm Sci. 1996; 85(10): 1070-6.

Wang J, Zhou G, Chen C, Yu H, Wang T, Ma Y, Jia G, Gao Y, Li B, Sun J et al. Acute toxicity and biodistribution of different sized titanium dioxide particles in mice after oral

NU

administration. Toxicol Lett. 2007; 168(2): 176-85.

Weir A, Westerhoff P, Fabricius L, Hristovski K, von Goetz N. Titanium dioxide nanoparticles in

MA

food and personal care products. Environ Sci Technol. 2012; 46(4): 2242-50. WHO Technical Report Series, No. 557, 1974. Evaluation of certain food additives (Eighteenth report of the Joint FAO/WHO Expert Committee on Food Additives). Available at

D

http://www.who.int/foodsafety/publications/jecfa-reports/en/

PT E

WHO Technical Report Series, No. 631, 1978. Evaluation of certain food additives and contaminants (Twenty-second report of the Joint FAO/WHO Expert Committee on Food Additives). Available at http://www.who.int/foodsafety/publications/jecfa-

CE

reports/en/

WHO Technical Report Series, No. 773, 1986. Evaluation of certain food additives and contaminants (Twenty-ninth report of the Joint FAO/WHO Expert Committee on Food

AC

Additives). Available at http://www.who.int/foodsafety/publications/jecfa-reports/en/ Zhu S, Chen S, Gao Y, Guo F, Li F, Xie B, Zhou J, Zhong H. Enhanced oral bioavailability of insulin using PLGA nanoparticles co-modified with cell-penetrating peptides and Engrailed secretion peptide (Sec). Drug Deliv. 2016; 23(6): 1980-91.

24

ACCEPTED MANUSCRIPT Figures

Figure 1. Schematic representation of the process to obtain the Caco-2/M-cell model and of the experimental protocol to evaluate translocation of TiO2 NPs.

RI

PT

Figure 2. Representative SEM images of Caco-2 monocultures (A & B) and SEM images including M-cells obtained from Caco-2 co-cultured with Raji B cells for four (C) and seven days (D). Caco-2 cells show dense microvilli at their apical surface and contrast with M-cells (M) with only residual and irregular microvilli.

NU

SC

Figure 3. Decrease of transepithelial resistance (TEER) during M cell differentiation in the two different co-culture duration protocols tested. The TEER is expressed as the percentage vs. the TEER value on day 14. TEER values on day 14 ranged from 1460 to 2070 Ω/cm2. Results are the mean and standard error of three independent experiments.

PT E

D

MA

Figure 4. Hydrodynamic diameter distribution of TiO2 NP suspensions in cell culture medium obtained by two sonication protocols at time 0 and at time 24 hours. The intensity distribution (A) and the volume distribution (B) are presented. The values are the mean of three independent experiments.

CE

Figure 5. Effect of TiO2 NP exposure on TEER measurements. The TEER is expressed as the percentage vs. the TEER value on day 14. TEER values on day 14 ranged from 1430 to 2070 Ω/cm2. Results are the mean of 3 independent experiments and bars represent standard error.

AC

Figure 6. SEM images after 48h of exposure to TiO2 NPs (sonication protocol A) of Caco-2 monoculture (A and B for TiO2 NP exposure initiated on day 18; C and D for TiO2 NP exposure initiated on day 21) and Caco-2/Raji B co-cultures (E & F for 4-day and 7-day co-cultures, respectively). The arrows point to structures that, based on their morphology, are likely to be TiO2 NPs.

Figure 7. Levels of titania in the basal compartment 48h after exposure to control cell culture media (‘unexposed cells’) and to 100µg/mL TiO2 nanoparticles. A) Levels of titania are presented for the unexposed wells and for exposed wells using two different sonication protocols (A and B in the X axis) and different Caco-2/Raji B co-incubation periods and the corresponding days in a Caco-2 monoculture. Results are mean values and standard error from three independent experiments. *Denotes statistically significant differences vs. unexposed 25

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

cells (One-way ANOVA followed by Dunnett post-hoc tests, p < 0.05). B) Levels of titania are presented for unexposed wells, exposed Caco-2 wells, and exposed Caco-2/Raji B wells, pooling data for different sonication protocols and co-incubation periods. *Denotes statistically significant differences between groups (One-way ANOVA followed by Tukey posthoc tests, p < 0.05).

26

ACCEPTED MANUSCRIPT Figure 1.

Sonication Protocol A: 3mm x 80 mm probe Time: 1 minute Amplitude: 40%

Protocol B: 6mm x 80 mm probe Time: 15 minute Amplitude: 20%

19

20

21

22

23

RI

18

PT

18 or 21 days

Mono-culture Caco-2

Transwell Ø 3 µm

SC

Mono-culture Caco-2

Exposure to NPs (48h)

SEM evaluation

TEER evaluation

14 days

14

15

16

17

15

16

17

18

19

20

18

19

20

21

22

23

NU

Mono-culture Caco-2

14

ICP-MS analyses

MA

Raji B

CE

PT E

D

Transwell Ø 3 µm

AC

Co-culture Caco-2/Raji B

Mono-culture Caco-2

14 days

27

ACCEPTED MANUSCRIPT Figure 2.

(B)

(C)

(D)

M

AC

CE

PT E

D

MA

NU

SC

M

RI

PT

(A)

28

ACCEPTED MANUSCRIPT Figure 3.

% T E E R ( d a y 1 4 o f c u lt u r e )

125

100

75

50

25

125

100

75

50

M o n o c u lt u r e C a c o -2 C o -c u lt u r e C a c o -2 / R a ji B c e lls

25

0

0 12

14

16

18

20

22

12

14

16

18

20

22

24

26

D ays

CE

PT E

D

MA

NU

SC

RI

D ays

AC

% T E E R ( v s . d a y 1 4 o f c u lt u r e )

C o - c u lt u r e p e r io d

150

PT

C o - c u lt u r e p e r io d

150

29

ACCEPTED MANUSCRIPT Figure 4.

(A) 12

PT

8

6 4

RI

Intensity (%)

10

SC

2

10

100

NU

0

1000

Size (nm)

MA

(B) 14

D

12

PT E

8

CE

6

4 2 0

AC

Volume (%)

10

10

100

1000

Size (nm) Sonication A, t=0h

Sonication B, t=0h

Sonication A, t=24h

Sonication B, t=24h

30

ACCEPTED MANUSCRIPT Figure 5.

(A )

(B )

E x p o su re to N P s

E x p o su re to N P s

C a c o -2 C o n tro l

100

75

50

25

C a c o - 2 T iO 2 _ S o n ic a t io n A

125

C a c o - 2 T iO 2 _ S o n ic a t io n B 100

C a c o - 2 / R a ji B C o n t r o l 75

C a c o - 2 / R a ji B T iO 2 _ S o n ic a t io n A C a c o - 2 / R a ji B T iO 2 _ S o n ic a t io n B

50

25

0 14

16

18

20

22

12

14

16

D ays

18

20

22

24

26

PT E

D

MA

NU

SC

D ays

CE

12

RI

0

PT

% T E E R ( d a y 1 4 o f c u lt u r e )

C o - c u lt u r e p e r io d

125

AC

% T E E R ( d a y 1 4 o f c u lt u r e )

C o - c u lt u r e p e r io d

150

150

31

ACCEPTED MANUSCRIPT Figure 6.

(B)

(C)

(D)

D

MA

NU

SC

RI

PT

(A)

(F)

AC

CE

PT E

(E)

32

ACCEPTED MANUSCRIPT Figure 7. (A)

(B) *

1 .0

* T i (µ g /m L )

0 .8 0 .6 0 .4

0 .0

ce

lls

Ca

-2 co

ex

RI

ed

po

Ca

se

co

d

-2 /

R

x B e a ji

po

se

d

AC

CE

PT E

D

MA

NU

SC

Un

ex

s po

PT

0 .2

33