Comprehensive In Vitro Toxicity Testing of a Panel of Representative Oxide Nanomaterials: First Steps towards an Intelligent Testing Strategy

Nanomaterials (NMs) display many unique and useful physico-chemical properties. However, reliable approaches are needed for risk assessment of NMs. The present study was performed in the FP7-MARINA project, with the objective to identify and evaluate in vitro test methods for toxicity assessment in order to facilitate the development of an intelligent testing strategy (ITS). Six representative oxide NMs provided by the EC-JRC Nanomaterials Repository were tested in nine laboratories. The in vitro toxicity of NMs was evaluated in 12 cellular models representing 6 different target organs/systems (immune system, respiratory system, gastrointestinal system, reproductive organs, kidney and embryonic tissues). The toxicity assessment was conducted using 10 different assays for cytotoxicity, embryotoxicity, epithelial integrity, cytokine secretion and oxidative stress. Thorough physico-chemical characterization was performed for all tested NMs. Commercially relevant NMs with different physico-chemical properties were selected: two TiO2 NMs with different surface chemistry – hydrophilic (NM-103) and hydrophobic (NM-104), two forms of ZnO – uncoated (NM-110) and coated with triethoxycapryl silane (NM-111) and two SiO2 NMs produced by two different manufacturing techniques – precipitated (NM-200) and pyrogenic (NM-203). Cell specific toxicity effects of all NMs were observed; macrophages were the most sensitive cell type after short-term exposures (24-72h) (ZnO>SiO2>TiO2). Longer term exposure (7 to 21 days) significantly affected the cell barrier integrity in the presence of ZnO, but not TiO2 and SiO2, while the embryonic stem cell test (EST) classified the TiO2 NMs as potentially ‘weak-embryotoxic’ and ZnO and SiO2 NMs as ‘non-embryotoxic’. A hazard ranking could be established for the representative NMs tested (ZnO NM-110 > ZnO NM-111 > SiO2 NM-203 > SiO2 NM-200 > TiO2 NM-104 > TiO2 NM-103). This ranking was different in the case of embryonic tissues, for which TiO2 displayed higher toxicity compared with ZnO and SiO2. Importantly, the in vitro methodology applied could identify cell- and NM-specific responses, with a low variability observed between different test assays. Overall, this testing approach, based on a battery of cellular systems and test assays, complemented by an exhaustive physico-chemical characterization of NMs, could be deployed for the development of an ITS suitable for risk assessment of NMs. This study also provides a rich source of data for modeling of NM effects.


Fig. O. Interference of Titanium dioxide nanomaterials (NM-103 and NM-104) with LDH assay (A and B) and ELISA (C and D)
A significant interference with the LDH assay in a dose-dependent manner was observed. However at the doses used in our studies the maximum percent change induced by both nanomaterials was around 3%. Values of interference with the ELISA were always below 1%.
For WST-1 assay (data not shown), both NM-103 and NM-104, control experiment wells containing NMs and cells only were shown to have minimal absorbance readings, thus it was concluded that there was no interference from TiO2 on the assay.
For WST-8 assay the interferences with TiO 2 used for the cytotoxicity testing shown in Figures S6-S8 in File S3 were performed as described previously (Kroll et al. 2012). The used concentrations were below the observed interfering concentrations.

Fig. P. Interferences of TiO 2 (NM-103 and NM-104) with resazurin and NRU assays
The fluorescence read at the 60 min time point in the absence of cells was not significantly influenced by TiO 2 NMs and < 0.5% of the minimal value obtained in the presence of cells. It is concluded that TiO 2 NMs do not interfere with the resazurin method through autofluorescence (A and B). The supplementation of medium with TiO 2 NMs did not appreciably quench resazurinderived fluorescence (C and D). For the NRU assay (E and F), the results indicate that addition of neutral red solution with TiO 2 (NM-103 or NM-104). NMs decrease OD values and showing a binding of neutral red to the NMs. However, binding is not dose dependent, suggesting that only a fraction of the dye solution (which is not homogeneous) is adsorbed to the NMs. Neutral Red binding to TiO 2 NMs should remove the dye to the solution and, hence, lower its uptake by the cells, factitiously lowering viability signal. Alternatively, given that NMs enter cells, dye binding could increase its accumulation in the cells and, hence, increase viability signal. Actually, results of neutral red assay indicate smaller viability decreases compared with the resazurin assay. Data are means ± SD of 6 independent determinations in two separate experiments. Statistical analysis was performed using one-way ANOVA followed by Bonferroni post-hoc test. *p< 0.05, ***p< 0.001. In the estimation of interference of WST-8 (CCK) assay, NMs were directly added to the test system without cells. The absorbance of OD450 was measured with or without CCK solution stain.
The difference between the regular test procedures to this no-cell WST-8 test assay was that most NMs in the incubation medium being washed away before CCK solution added in the regular ones.
There was only about 10 percent of the cell viability over-estimation at 80 μg/ml TiO 2 treatment condition in no-cell test, and even much lower at other lower exposure concentrations. Therefore, the actual test interference of those residue NMs was quite low and not likely to influence the results significantly.

Fig. R. Interference of Zinc oxide nanomaterials (NM-110 and NM-111) with LDH assay (A and B) and ELISA (C and D)
The results showed that NM-110 and NM-111 do not interfere with the LDH assay as the values of interference were always below 0.25%. The evaluation of NMs-assay interferences showed that none of the ZnO NMs interferes with the ELISA assay.
For the WST-1 assay (data not shown), both NM-110 and NM-111 shown to have minimal absorbance readings, thus it was concluded that there was no interference from ZnO in this assay. binding of neutral red to the NMs. However, binding is not dose dependent, suggesting that only a fraction of the dye solution (which is not homogeneous) is adsorbed to the NMs. Neutral Red binding to ZnO NMs apparently decreases the free dye concentration in the solution and, hence, lowers its uptake by the cells, factitiously lowering viability signal. Alternatively, given that NMs enter cells, dye binding could increase its accumulation in the cells and, hence, increase viability 6 signal. Actually, results of neutral red assay indicate smaller viability decreases compared with the resazurin assay. Data are means ± SD of 10 independent determinations in two separate experiments. Statistical analysis was performed using one-way ANOVA followed by Bonferroni post-hoc test. *p< 0.05, **p< 0.01.

ELISA (C and D)
The results showed that both NMs  do not interfere with the LDH assay as the values of interference were always below 0,25%. Also none of the SiO2 NMs interferes with the ELISA.
WST-1 assay (data not shown) -for both NM-200 and NM-203, control wells containing NM and wells containing cells only were shown to have minimal absorbance readings, thus it was concluded that there was no interference from the SiO 2 . 8

Fig. U. Interferences of SiO 2 NMs with resazurin (A-D) and NRU (E-F) assays
The fluorescence read at the 60 min time point was not significantly influenced by SiO2 NMs and < 0.5% of the minimal value obtained in the presence of cells. It is concluded that SiO 2 NMs do not interfere with the resazurin method through autofluorescence (A and B) while the addition of SiO 2 NMs in culture medium did not appreciably quench resazurin-derived fluorescence (C and D).
Regarding the assay-NMs interferences with NRU assay (E and F), the results indicate that the supplementation of neutral red solution with SiO 2 NMs did not influence the OD values, indicating the absence of significant dye binding. Data are means ± SD of 10 independent determinations in two separate experiments. Statistical analysis was performed using one-way ANOVA followed by Bonferroni post-hoc test.