Assessment of the Influence of Crystalline Form on Cyto-Genotoxic and Inflammatory Effects Induced by TiO2 Nanoparticles on Human Bronchial and Alveolar Cells

Titanium dioxide nanoparticles (TiO2NPs) are increasingly used in consumer products, industrial and medical applications, raising concerns on their potential toxicity. The available in vitro and in vivo studies on these NPs show controversial results. Crystalline structure is the physicochemical characteristic that seems to influence mainly TiO2NPs toxicity, so its effect needs to be further studied. We aimed to study whether and how crystalline form influences potential cyto-genotoxic and inflammatory effects induced by two commercial TiO2NPs (TiO2-A, mainly anatase; TiO2-B, mainly rutile) in human alveolar A549 and bronchial BEAS-2B cells exposed to 1–40 µg/mL. Cell viability (WST-1), membrane damage (LDH release), IL-6, IL-8 and TNF-α release (ELISA) and direct/oxidative DNA damage (fpg-comet assay) were evaluated. Physicochemical characterization included analysis of crystalline form (TEM and XRD), specific surface area (BET), agglomeration (DLS) and Z-potential (ELS). Our results show that TiO2-A NPs induce in BEAS-2B cytotoxicity and a slight inflammation and in A549 slight oxidative effects, whereas TiO2-B NPs induce genotoxic/oxidative effects in both cell lines, revealing different toxicity mechanisms for the two tested NPs. In conclusion, our study confirms the influence of crystalline form on cellular response, also demonstrating the suitability of our in vitro model to screen early TiO2NPs effects.


Introduction
Nano-Titanium dioxide (nano-TiO 2 ) has become the most highly used nanoparticle (NP) worldwide [1], occurring in daily products such as paints, plastics, papers, inks, foods, pharmaceuticals, cosmetics [2] and many personal care products [3]. Thus, due to their widespread use and the consequent unintentional exposure, the concern about the potential toxicity of TiO 2 nanoparticles (TiO 2 NPs) is raising. Exposure to TiO 2 nanoparticles can occur during manufacturing, handling and use [4,5] involving consumers, workers, researchers of Laboratories involved in the development and production of nanomaterials (NMs) [6].
Titanium dioxide is the naturally occurring oxide of titanium and has different crystalline structures: rutile, the most common natural and stable form [1,7], anatase, the most commercially used type [4], brookite, the rarest form [7]; there is also a fourth amorphous form named "titanium dioxide (B)" [8]. In its nano-form, TiO 2 belongs to the category of metallic NPs [9]. The International Agency for Research on Cancer (IARC) has reclassified TiO 2 as a Group 2B carcinogen (possibly carcinogenic to humans) on the basis of inadequate evidence in humans and sufficient evidence in experimental animals that fine TiO 2 particles cause cancer [8].
account that global DNA hypomethylation has been associated with several diseases and cancer, the authors suggested that epigenetic endpoints should be considered in the safety assessment of TiO 2 NPs in various consumer products [27]. Uboldi et al. [24] in a study on mouse fibroblasts exposed to pure anatase and rutile TiO 2 NPs, evidenced only for rutile TiO 2 NPs a cytotoxic effect in terms of dose-dependent reduction of the clonogenic activity. Similarly, rutile nano-TiO 2 appeared to be slightly clonogenic and genotoxic, whereas anatase TiO 2 NPs did not induce any significant neoplastic or genotoxic effect. Additionally, as indicated by the quantification of the in vitro uptake of TiO 2 NPs, they showed that the internalization was independent of the crystalline form but size dependent, as nano-TiO 2 were taken up more than their respective bulk materials [24].
In the present work we aimed to study whether and how crystalline form influences potential cyto-genotoxic and inflammatory effects induced by two commercial TiO 2 NPs (A, 79% anatase; B, 81% rutile) in human lung cells. In particular, here we report the results obtained testing the rutile nano-form, TiO 2 -B, and compare them with those observed on the anatase crystalline form, TiO 2 -A, previously published [10]. To do this, we used the same well-characterized experimental approach (methods and cellular in vitro model) we developed to study the potential toxicity of other nanomaterials such as pristine and functionalized carbon nanotubes [28,29], and cobalt oxide NPs [30], in addition to the mentioned anatase titanium dioxide NPs [10]. The cellular model involves the use of human alveolar and bronchial epithelial cells (A549 and BEAS-2B) that represent the main lung cells and the most used cell types in inhalation toxicity studies [28].

Nanoparticles
Two commercially available titanium (IV) oxide nano-powders, product number 677,469 (TiO 2 -A) and product number 634,662 (TiO 2 -B), were purchased from Sigma-Aldrich (St. Louis, MO, USA). As specified by the supplier, the purity of TiO 2 -A was up to 97% and 1% Mn was present in the sample as dopant; the purity of TiO 2 -B was 99.5% with trace metals basis.

Nanoparticles Characterization
For characterization, we prepared a stock solution (2 mg/mL) of TiO 2 NPs in ultrapure sterile water, then we vortexed it for 1 min and sonicated for 5 min to disperse NPs. NPs diameter and shape, and agglomerate sizes in water and cell culture media (RPMI-1640 and BEGM) were analyzed by an Energy Filtered Transmission Electron Microscopy (EFTEM) (FEI TECNAI 12G2 Twin; Termo Fisher Scientific-FEI Company, Hillsboro, OR, USA), at an accelerating voltage of 120 kV, equipped with an electron energy filter (Gatan Image Filter, BioFilter model; Gatan Inc., Pleasanton, CA, USA) and a Peltier cooled charge-coupled device based slow scan camera (Gatan multiscan camera, model 794IF; Gatan Inc., Pleasanton, CA, USA). We acquired conventional and High Resolution TEM (HRTEM) micrographs and performed the elemental and structural analysis by Electron Energy Loss Spectroscopy (EELS) and Nanobeam Electron Diffraction (NED). HRTEM, EELS and NED showed the crystalline nature of nanoparticles. The image analysis was performed by Digital Micrograph (Gatan Inc., Pleasanton, CA, USA) and ImageJ (National Institutes of Health, Bethesda, MD, USA).
NPs hydrodynamic diameter and zeta potential were analyzed by Dynamic Light Scattering (DLS) and Electrophoretic Light Scattering (ELS) (Zetasizer nano ZS, Malvern, UK). Suspension stability was evaluated measuring the hydrodynamic diameter of the TiO 2 NPs in each culture medium at the beginning of exposure (t0) and after 24 h (t24).
The Specific Surface Area (SSA) of samples was obtained by N 2 adsorption at 77 K using the Brunauer-Emmett-Teller (BET) method and a Quantachrome Nova 2200 Surface Area Analyzer (Quantachrome Instruments, Boynton Beach, FL, USA).
Cells were seeded into a 24 well culture plate (15.6 mm well diameter) at a density of 8 × 10 4 cells per well and were grown for 24 h before exposure at 37 • C in 5% CO 2 and humidified atmosphere. On the day of exposure, cell media were replaced with fresh media.

Exposure
At the time of exposure, the stock solutions in water (2 mg/mL) were vortexed 1 min and then sonicated 5 min (Branson 2510; Branson Ultrasonics Corporation, Danbury, CT, USA) to better disperse the NPs. Then, a working solution (1 mg/mL) in complete culture medium (RPMI-1640 or BEGM) was prepared and sonicated in two 5-min steps with a 30-s pause, before being quickly added to the wells to the final concentrations of 1, 5, 10, 20, and 40 µg/mL. During exposure, cells were maintained at 37 • C in 5% CO 2 and humidified atmosphere. Three independent experiments were conducted in three separate times.

Cell Viability
The cell viability of cells exposed for 24 h to TiO 2 NPs was assessed by the colorimetric Premix WST-1 Cell Proliferation Assay System (Takara Bio Inc., Shiga, Japan). The WST-1 is assumed to be reliable to evaluate NM cytotoxicity since does not show interference with NPs including TiO 2 as previously reported [33,34].
WST-1 is a water-soluble tetrazolium salt that is reduced by mitochondrial dehydrogenases in cells to a water-soluble formazan dye. The amount of the formazan dye is directly proportional to the number of living cells.
The assay was performed according to our previous study [10]. Briefly, after exposure, the media were removed and cells were rinsed twice with Phosphate Buffered Saline (PBS) to avoid any possible interference in light absorption due to TiO 2 NPs. Then, according to the manufacturer's guidelines, fresh culture medium plus PreMix WST-1 (10:1 ratio) were added to each well and plates were incubated for 3 h at 37 • C in the dark. At the end of incubation, 200 µL of supernatant in duplicate were transferred in an optically clear 96-well flat bottom plate and absorbance [A] was measured at 450 nm using a spectrophotometric microplate reader (Wallac Victor 2; Perkin Elmer, Boston, MA, USA). Unexposed cells were used as negative control and cells exposed to 1% Triton X-100 (Sigma-Aldrich, St. Louis, MO, USA) were used as positive control. A blank well was set up with only fresh medium plus Premix WST-1 (10:1 ratio) and used as background. After subtracting background absorbance, the cell viability (% of control) was calculated as follows:

Membrane Damage
To determine the membrane damage of cells exposed for 30 min, 2 h and 24 h to TiO 2 NPs, the lactate dehydrogenase (LDH) assay (Cytotoxicity Detection Kit; Roche Diagnostics, Mannheim, Germany) was used. LDH assay is a colorimetric test performed to quantify cell membrane damage and is based on the reduction of tetrazolium salt to formazan by the LDH. The LDH is an enzyme present in all cells and it is rapidly released from the cytosol of damaged cells into the cell culture supernatant.
This assay was performed as previously [10], according to the manufacturer's guidelines. Briefly, the reaction mixture plus the culture supernatant (1:1 ratio) were transferred in triplicate to the wells of an optically clear 96-well flat bottom plate and incubated for 30 min at T amb (15-25 • C), protecting the plate from light. Blank (only culture medium), negative controls (cells not exposed) and positive controls (cells exposed for the same exposure times to 1% Triton X-100) (Sigma-Aldrich, St. Louis, MO, USA) were arranged. At the end of incubation, absorbance was measured at 490 nm using a spectrophotometric microplate reader (iMark; Bio-Rad, Milan, Italy).
To take into account the possible interference in light absorption due to TiO 2 NPs, a parallel set of experiments was carried out without cells; then, the absorbance of each concentration of TiO 2 NPs (without cells) was subtracted from that of the corresponding sample (with cells).
Finally, after subtracting background absorbance, the percentage of cytotoxicity was calculated as follows:

DNA Damage
To assess the genotoxic/oxidative DNA damage of cells exposed for 2 and 24 h to TiO 2 NPs we used the Fpg (Formamido-pyrimidine DNA glycosylase; Sigma-Aldrich, USA) modified comet assay which allows to evaluate direct and oxidative DNA damage simultaneously.
The Fpg is a DNA base excision repair enzyme that removes the oxidized guanines and some alkylated bases, generating strand breaks, and is used to detect the oxidative damage by comet assay [35].
The previously described protocol [36] was used. Briefly, after exposure, the cell suspensions from each experimental point are mixed with agarose and set on two Gelbond films (Sigma-Aldrich, USA) to evaluate direct and oxidative DNA damage at the same time. Unexposed cells are used as negative control and cells exposed for 30 min to 100 µM hydrogen peroxide (H 2 O 2 ) are used as positive control. After lysis, one set of Gelbond films is incubated with Fpg enzyme to evaluate oxidative DNA damage and the other one only with buffer to detect direct DNA damage. This is followed by an electrophoresis in alkaline buffer performed in the dark. It is important to avoid any further DNA damage due to the possible photo-activation of TiO 2 NPs by laboratory light exposure as reported for P25 TiO 2 NPs by Petersen et al. (2014) who suggest to perform comet assay of TiO 2 exposed cells in the dark to avoid potential artifacts [37].
After ethidium bromide staining, 100 randomly chosen comet images per slide are acquired by fluorescence microscopy (Axioplan 2 Imaging; Carl Zeiss, Göttingen, Germany) at 200× magnification and then analyzed by a specific image analysis software (Delta Sistemi, Rome, Italy).
DNA damage is calculated as the mean % tail DNA from 100 comets. For each experimental point, direct DNA damage is calculated, in comets after buffer incubation, as the ratio of % tail DNA of exposed cells to % tail DNA of negative control; oxidative DNA damage (Fpg-sensitive sites) is calculated by subtracting the % tail DNA with buffer incubation from the % tail DNA with enzyme incubation.
The manufacturer's guidelines were followed. Briefly, after exposure, the cell supernatant was collected and stored at −20 • C until use. Samples, including negative controls (cells not exposed to TiO 2 NPs), appropriate cytokine standards, blank and control samples were tested in duplicate. The absorbance was measured at 450 nm and quantified with a microplate absorbance reader (iMark; Bio-Rad, Milan, Italy).

Statistical Analysis
Three independent experiments were conducted and data are expressed as mean ± standard deviation (SD). Statistical significance of the data was analyzed by Non-parametric Kruskal-Wallis test followed by post-hoc analysis via the Bonferroni correction or Dunnett's T3 test. p-values ≤ 0.05 and ≤0.01 were considered significant. Table 1 shows the physicochemical characteristics of TiO 2 -A and TiO 2 -B and the testing methods used. TEM and XRD analyses revealed that TiO 2 -A is mainly made up of anatase with a spherical shape, whereas TiO 2 -B is mainly in the rutile crystalline form with irregulartetragonal shape. The tested NPs have similar SSA values but slightly different primary NP diameters, with TiO 2 -A diameter smaller than TiO 2 -B. (diam Zav nm) (20 µg/mL) 224.6 (t24) 282.0 (t24) agglomerate/aggregate size (diam Zav nm) (100 µg/mL) 186.0 (t0) 456.0 (t0) TEM  High Resolution TEM (HRTEM), elemental and structural analysis by Electron ergy Loss Spectroscopy (EELS) and Nanobeam Electron Diffraction (NED) of TiO2-A w previously reported [10].

Nanoparticle Characterization
The crystalline nature of the TiO2-B NPs was confirmed by the following TEM te niques: HRTEM image (Figure 2a High Resolution TEM (HRTEM), elemental and structural analysis by Electron Energy Loss Spectroscopy (EELS) and Nanobeam Electron Diffraction (NED) of TiO 2 -A were previously reported [10].
The crystalline nature of the TiO 2 -B NPs was confirmed by the following TEM techniques: HRTEM image (Figure 2a,b) revealed the lattice planes allowing the measurement of the plane distance (0.36 nm); elemental analysis of the TiO 2 -B NPs by EELS (Figure 2c) showed the splitting of the titanium L 2 and L 3 core loss peaks, not observable in amorphous Ti; NED revealed the very regular diffraction spot pattern typical of rutile crystalline form (Figure 2d).
In both water and the culture media the TiO 2 NPs dissolved forming agglomerate/aggregate with Zav diameters smaller in TiO 2 -A than in TiO 2 -B. The agglomerate/aggregate sizes are greater in BEGM medium than in RPMI and in culture media than in water for both TiO 2 . DLS analysis shows a Z potential slightly more negative in BEGM than in RPMI and a much more negative value in water than in culture media for both TiO 2 ( Table 1). The agglomerate/aggregate sizes of TiO 2 -B in water and BEGM decrease over time (t0-t24) due to the sedimentation of large agglomerates. The agglomerate/aggregate sizes of TiO 2 -A, showed in Table 1, remain almost similar over time (t0-t24), suggesting suspension stability and low sedimentation. In both water and the culture media the TiO2NPs dissolved forming agglomerate/a gregate with Zav diameters smaller in TiO2-A than in TiO2-B. The agglomerate/aggrega sizes are greater in BEGM medium than in RPMI and in culture media than in water both TiO2. DLS analysis shows a Z potential slightly more negative in BEGM than in RP and a much more negative value in water than in culture media for both TiO2 (Table  The agglomerate/aggregate sizes of TiO2-B in water and BEGM decrease over time (t t24) due to the sedimentation of large agglomerates. The agglomerate/aggregate sizes TiO2-A, showed in Table 1, remain almost similar over time (t0-t24), suggesting suspe sion stability and low sedimentation. Figure 3 shows the results of WST-1 assay and demonstrates that in A549 cells bo TiO2NPs do not induce any cytotoxic effect, although there is a very slight viability redu tion at the highest concentration (Figure 3a). In BEAS-2B cells, viability decreases diff ently for the two titania: TiO2-A NPs induce a dose-dependent viability reduction from µg/mL; TiO2-B NPs induce a moderate viability reduction at 5 µg/mL and 10 µg/mL (F ure 3b).  Figure 3 shows the results of WST-1 assay and demonstrates that in A549 cells both TiO 2 NPs do not induce any cytotoxic effect, although there is a very slight viability reduction at the highest concentration (Figure 3a). In BEAS-2B cells, viability decreases differently for the two titania: TiO 2 -A NPs induce a dose-dependent viability reduction from 10 µg/mL; TiO 2 -B NPs induce a moderate viability reduction at 5 µg/mL and 10 µg/mL (Figure 3b).   Figure 4 shows the results of LDH assay performed after 30 min, 2 h, and 24 h of exposure of A549 and BEAS-2B cells to the NPs. Both TiO2NPs induce a slight, early and  Figure 4 shows the results of LDH assay performed after 30 min, 2 h, and 24 h of exposure of A549 and BEAS-2B cells to the NPs. Both TiO 2 NPs induce a slight, early and transient increase of LDH release in both cell types. Furthermore, we found no effect on membrane integrity after 24 h exposure (Figure 4e Figure 4 shows the results of LDH assay performed after 30 min, 2 h, and 24 h of exposure of A549 and BEAS-2B cells to the NPs. Both TiO2NPs induce a slight, early and transient increase of LDH release in both cell types. Furthermore, we found no effect on membrane integrity after 24 h exposure (Figure 4e,f). In particular, TiO 2 -A NPs induce an increase in cytotoxicity at the highest concentration after 30 min and 2 h exposure in A549 (Figure 4a,c); in BEAS-2B TiO 2 -A induces a LDH release from 10 µg/mL, but statistically significant only at the highest concentration, after 30 min exposure and a peak at 10 µg/mL after 2 h exposure (Figure 4b,d). TiO 2 -B NPs causes a peak of LDH release after 30 min exposure at 10 µg/mL in A549 (Figure 4a) and after 2 h exposure at 5 µg/mL in BEAS-2B (Figure 4c).

DNA Damage (2 h and 24 h)
TiO 2 -A NPs induce direct and oxidative DNA damage only after 2 h exposure at 40 µg/mL in A549 (Figure 5a,b,e,f); they induce neither direct nor oxidative DNA damage in both cell types after 24 h exposure (Figure 5c,d,g,h). On the other hand, TiO 2 -B NPs induce direct DNA damage after 2 h exposure at the highest concentrations in both cell types and an oxidative DNA damage at the highest concentrations statistically significant only in A549 (Figure 5a,b,e,f). After 24 h exposure, the direct DNA damage induced by TiO 2 -B is dose dependent from 5 µg/mL, with a peak at 20 µg/mL, in A549 cells and at the highest concentrations in BEAS-2B cells (Figure 5c,d). The oxidative DNA damage induced by TiO 2 -B after 24 h exposure is statistically significant only at the highest concentration in A549 cells (Figure 5g,h), whereas in BEAS-2B cells increased slightly starting from 10 µg/mL (Figure 5f). In particular, TiO2-A NPs induce an increase in cytotoxicity at the highest concentration after 30 min and 2 h exposure in A549 (Figure 4a,c); in BEAS-2B TiO2-A induces a LDH release from 10 µg/mL, but statistically significant only at the highest concentration, after 30 min exposure and a peak at 10 µg/mL after 2 h exposure (Figure 4b,d). TiO2-B NPs causes a peak of LDH release after 30 min exposure at 10 µg/mL in A549 (Figure 4a) and after 2 h exposure at 5 µg/mL in BEAS-2B (Figure 4c).

DNA Damage (2 h and 24 h)
TiO2-A NPs induce direct and oxidative DNA damage only after 2 h exposure at 40 µg/mL in A549 (Figure 5a,b,e,f); they induce neither direct nor oxidative DNA damage in both cell types after 24 h exposure (Figure 5c,d,g,h). On the other hand, TiO2-B NPs induce direct DNA damage after 2 h exposure at the highest concentrations in both cell types and an oxidative DNA damage at the highest concentrations statistically significant only in A549 (Figure 5a,b,e,f). After 24 h exposure, the direct DNA damage induced by TiO2-B is dose dependent from 5 µg/mL, with a peak at 20 µg/mL, in A549 cells and at the highest concentrations in BEAS-2B cells (Figure 5c,d). The oxidative DNA damage induced by TiO2-B after 24 h exposure is statistically significant only at the highest concentration in A549 cells (Figure 5g,h), whereas in BEAS-2B cells increased slightly starting from 10 µg/mL (Figure 5f).

Cytokine Release (2 h and 24 h)
TiO2-A NPs induced in A549, a significant increase in the release of IL-6 at 5 µg/mL after 2 h exposure and a decrease in the release of IL-8 at the highest concentration after 2 h and 24 h exposure; we found no effect on the release of TNF-α after 2 h and 24 h exposure (Figure 6a,c). No statistically significant increase of cytokine release was detected in BEAS 2B (Figure 6b,d).
TiO2-B NPs induced an increase of IL-6 release in A549 cells after 2 h exposure at 10 and 40 µg/mL whereas no statistically significant increase of release of any of the three cytokines tested was found in BEAS-2B cells (Figure 6e

Cytokine Release (2 h and 24 h)
TiO 2 -A NPs induced in A549, a significant increase in the release of IL-6 at 5 µg/mL after 2 h exposure and a decrease in the release of IL-8 at the highest concentration after 2 h and 24 h exposure; we found no effect on the release of TNF-α after 2 h and 24 h exposure (Figure 6a,c). No statistically significant increase of cytokine release was detected in BEAS-2B (Figure 6b,d).

Cytokine Release (2 h and 24 h)
TiO2-A NPs induced in A549, a significant increase in the release of IL-6 at 5 µg/mL after 2 h exposure and a decrease in the release of IL-8 at the highest concentration after 2 h and 24 h exposure; we found no effect on the release of TNF-α after 2 h and 24 h exposure (Figure 6a,c). No statistically significant increase of cytokine release was detected in BEAS-2B (Figure 6b,d).
TiO2-B NPs induced an increase of IL-6 release in A549 cells after 2 h exposure at 10 and 40 µg/mL whereas no statistically significant increase of release of any of the three cytokines tested was found in BEAS-2B cells (Figure 6e-h).

Discussion
The toxicity of TiO2NPs has been evaluated in numerous in vitro and in vivo studies. Although anatase and rutile crystalline forms, which have different electrical and optical properties, are used in different applications, their potential ability to induce more or less toxicity still needs to be clarified. The final aim of this study was to understand if and how the crystalline form influences the potential toxicity of TiO2NPs on lung that represents one of the main target organs of occupational exposure to TiO2NPs. With this purpose, we used two types of commercial TiO2NPs that differed mainly for their crystalline form and we compared their potential ability to induce cytotoxic, genotoxic, and inflammatory effects on human alveolar and bronchial cells. TiO2-A exposure results, have already been published by our Laboratory [10], therefore in this work we compare the results of that study with those obtained testing rutile form TiO2-B.
The findings, summarised in Table 2, show the different toxic effects of the tested NPs, with high cytotoxicity, in terms of viability reduction, for TiO2-A in BEAS-2B cells, which responded to the injury of this kind of NPs with a dose-dependent trend, reaching

Discussion
The toxicity of TiO 2 NPs has been evaluated in numerous in vitro and in vivo studies. Although anatase and rutile crystalline forms, which have different electrical and optical properties, are used in different applications, their potential ability to induce more or less toxicity still needs to be clarified. The final aim of this study was to understand if and how the crystalline form influences the potential toxicity of TiO 2 NPs on lung that represents one of the main target organs of occupational exposure to TiO 2 NPs. With this purpose, we used two types of commercial TiO 2 NPs that differed mainly for their crystalline form and we compared their potential ability to induce cytotoxic, genotoxic, and inflammatory effects on human alveolar and bronchial cells. TiO 2 -A exposure results, have already been published by our Laboratory [10], therefore in this work we compare the results of that study with those obtained testing rutile form TiO 2 -B.
The findings, summarised in Table 2, show the different toxic effects of the tested NPs, with high cytotoxicity, in terms of viability reduction, for TiO 2 -A in BEAS-2B cells, which responded to the injury of this kind of NPs with a dose-dependent trend, reaching about 40% of viable cells at the highest concentrations; differently, TiO 2 -B induced a lower decrease of cell viability in the same cell line.  This particular result confirms the higher susceptibility of bronchial BEAS-2B cells to cytotoxic effects and highlights the need to use more than one kind of cells when we have to screen the toxicity of new materials such as nanomaterials. TiO 2 -A induced also very early membrane damage in both cell lines, higher in BEAS-2B cells, which seems to correlate with the higher effects on cell viability. TiO 2 -B induced earlier membrane damage in A549 than in BEAS-2B cells whose membranes resulted damaged only after 2 h exposure. This different behaviour in membrane damage induction and consequently in cell viability reduction could be explained by the specific physicochemical characteristics (crystalline form, spherical shape and smaller NP diameter) that influence the lower tendency of TiO 2 -A compared to TiO 2 -B to agglomerate, with consequent smaller agglomerates that may cross directly and faster the membrane damaging it. Instead, larger TiO 2 -B agglomerates could enter the cell more slowly by an endocytosis mechanism.
The comparison between the two tested NPs on the induced genotoxic effects demonstrated that TiO 2 -A was able to induce early direct DNA damage only in A549 cells at the highest concentration, whereas TiO 2 -B induced DNA damage, after 2 and 24 h exposure, in both the tested cell lines. Differently, oxidative DNA damage was induced only in A549 cells after 2 h of exposure for TiO 2 -A and at both the times of exposure for TiO 2 -B. The lower genotoxicity of TiO 2 -A in respect to TiO 2 -B could be explained by the fact that uptaken agglomerates of TiO 2 -A NPs, entered the cells through non-endocytic pathways, are too large to cross nucleus membrane and directly damage DNA. A small TiO 2 -A NPs fraction (the smallest agglomerates) could co-localize in the nuclei and other organelles, such as mitochondria, and induce genotoxic/oxidative effects. Therefore, only at the highest exposure concentration the TiO 2 -A NPs are in sufficient amount to induce early DNA damage in more susceptible A549 cells as previously hypothesized [10].
Both tested TiO 2 NPs induced an early and transient IL-6 cytokine release only in A549 cells that seems to correlate with the oxidative DNA damage found after 2 h exposure.
The analysis of agglomeration of the tested NPs performed by both TEM and DLS demonstrated very larger agglomerates of TiO 2 -B compared to those of TiO 2 -A in both culture media at the concentrations used to expose A549 and BEAS-2B cells. In addition, in BEGM medium used for BEAS-2B cultures, TiO 2 -B aggregates resulted larger than in the RPMI medium used for A549 cultures. These differences could explain the observed toxic effects. As reported by Veranth et al. [38], the tendency to agglomerate and sediment, also determined by the composition of specific culture medium and by the presence of serum proteins (as in complete RPMI), influences the biological effects, as NMs agglomerates sediment more easily and interact with the cells more rapidly and at higher effective concentrations than the non-agglomerated NMs [39,40]. It leads to greater toxic effects. The higher sedimentation of larger TiO 2 -B NP agglomerates could so explain the higher genotoxicity of rutile form in respect to anatase. The DNA damage induced by TiO 2 -B could be generated through an indirect mode of action probably through intra-cytoplasmatic oxidants induction linked to the higher photocatalytic properties of rutile TiO 2 NPs that make it able to generate ROS on their surface, as also reported by Fenoglio et al. [41]. The ability to induce genotoxic effects by ROS induction has been demonstrated for several kind of NPs [42]. However, other physicochemical properties such as size, mono/polydispersity or surface charge could play a synergistic role in the differences between the rutile and anatase nano-TiO 2 toxicity mechanisms [24].
To date it is very difficult to find more than one study performed on the same NPs and an experimental protocol useful to make a comparison of the obtained results, therefore there is still a lack of information about the potential toxicity of NMs.
The comparison of the cytotoxicity of anatase and rutile form of TiO 2 NPs performed by Sayes et al. [43] on A549 cells, demonstrated a higher cytotoxicity for anatase compared to rutile and an increased release of IL-8 at concentrations higher than those used in our study. Higher toxicity of anatase compared to rutile in A549 cells was reported by De Matteis et al. [44], who also demonstrated the different degradation of the two crystalline forms in different conditions of pH and sunlight exposure. The Authors concluded that the higher toxicity of anatase is due to its major ionization.
Yu et al. [26] found a lower toxicity of anatase compared to rutile NPs in terms of viability reduction on macrophages. Uboldi et al. [24] demonstrated that rutile titania was slightly more toxic than anatase on Balb 3T3/mouse fibroblasts. Falck et al. [45] demonstrated that anatase reduces cell viability more than rutile on BEAS-2B cells.
The higher susceptibility of A549 cells to genotoxic effects induced by TiO 2 -B NPs, found in the present study and evident after 24 h exposure already at low concentrations, confirms what we found in our previous study for TiO 2 -A NPs [10] and in our other previous studies using the same experimental model [30,46]. The transformed A549 cells react to NM injury by direct and oxidative DNA damage because these cells are unaffected by cytotoxicity and more susceptible to genotoxic effects, probably because of lower capability to repair DNA damage by inactivation of both NER (Nucleotide Excision Repair) and BER (Base Excision Repair) [47].
Bronchial epithelial cells represent the first defence against airborne particulate matter and it may explain their stronger cytotoxic response in comparison with alveolar cells [48]. Normal BEAS-2B cells, respond to injury killing damaged cells (probably by apoptosis as reported by Park et al. 2008 [49] and Shi et al. 2010 [50]).
The different sensitivity of A549 and BEAS-2B cells suggests the need to use both cell types together with several and complementary toxicity end-points to evaluate the NM-induced biological effects.
Moreover, we highlight the need to evaluate the dispersion of tested NMs in the specific culture media. All together these factors contribute to induce the toxicity of NMs.
In conclusion, our findings confirm the influence of physicochemical characteristics and particularly of the crystalline form on cellular response to TiO 2 NPs, particularly regarding cytotoxic, genotoxic/oxidative effects and inflammation, with higher cytotoxicity for anatase exposure compared to rutile on bronchial cells and with rutile more genotoxic than anatase on both alveolar and bronchial cells. In addition, this study shows the suitability of our experimental system to screen toxic effects of nanomaterials with different properties.