Quantitative Structure-Activity Relationship Models for Predicting Inflammatory Potential of Metal Oxide Nanoparticles

Abstract

Background: Although substantial concerns about the inflammatory effects of engineered nanomaterial (ENM) have been raised, experimentally assessing toxicity of various ENMs is challenging and time-consuming. Alternatively, quantitative structure-activity relationship (QSAR) models have been employed to assess nanosafety. However, no previous attempt has been made to predict the inflammatory potential of ENMs.

Objectives: By employing metal oxide nanoparticles (MeONPs) as a model ENM, we aimed to develop QSAR models for prediction of the inflammatory potential by their physicochemical properties.

Methods: We built a comprehensive data set of 30 MeONPs to screen a proinflammatory cytokine interleukin (IL)-1 beta ($\text{IL-}1\beta$) release in THP-1 cell line. The in vitro hazard ranking was validated in mouse lungs by oropharyngeal instillation of six randomly selected MeONPs. We established QSAR models for prediction of MeONP-induced inflammatory potential via machine learning. The models were further validated against seven new MeONPs. Density functional theory (DFT) computations were exploited to decipher the key mechanisms driving inflammatory responses of MeONPs.

Results: Seventeen out of 30 MeONPs induced excess $\text{IL-}1\beta$ production in THP-1 cells. In vivo disease outcomes were highly relevant to the in vitro data. QSAR models were developed for inflammatory potential, with predictive accuracy (ACC) exceeding 90%. The models were further validated experimentally against seven independent MeONPs ($\text{ACC}=86\%$). DFT computations and experimental results further revealed the underlying mechanisms: MeONPs with metal electronegativity lower than 1.55 and positive $\zeta \text{-potential}$ were more likely to cause lysosomal damage and inflammation.

Conclusions: $\text{IL-}1\beta$ released in THP-1 cells can be an index to rank the inflammatory potential of MeONPs. QSAR models based on $\text{IL-}1\beta$ were able to predict the inflammatory potential of MeONPs. Our approach overcame the challenge of time- and labor-consuming biological experiments and allowed for computational assessment of MeONP inflammatory potential by characterization of their physicochemical properties. https://doi.org/10.1289/EHP6508.

Figure 1.

Schematic workflow of metal oxide nanomaterials (MeONPs) library construction, predictive modeling, mechanism interpretation, and experimental validation.

Figure 2.

Fold changes of $\text{IL-}1\mathrm{\beta }$ production ($F{C}_{\text{IL-}1\mathrm{\beta }}$) in metal oxide nanomaterials (MeONPs)-treated THP-1 cells. THP-1 cells were exposed to 0, 3.1, 6.2, 13, 25, 50, 100, and $200\mathrm{\mu }\mathrm{g}/\mathrm{mL}$ MeONPs for 24 h. $\text{IL-}1\mathrm{\beta }$ levels in supernatants were quantified by ELISA. $F{C}_{\text{IL-}1\mathrm{\beta }}$ was calculated by Equation 3. The $F{C}_{\text{IL-}1\mathrm{\beta }}$ was expressed as the mean of three replicates and added in the heatmap.

Figure 3.

Pulmonary inflammation of six selected metal oxide nanomaterials (MeONPs) in mice. (A) Cytokine production ($\text{IL-}1\mathrm{\beta }$ and MCP-1) in BALF, and (B) H&E staining of lung tissues after 40 h exposure to MeONPs. C57Bl/6 mice ($n=6$) were exposed to ${\mathrm{La}}_2{\mathrm{O}}_3$, ${\mathrm{Gd}}_2{\mathrm{O}}_3$, ${\mathrm{Co}}_3{\mathrm{O}}_4$, ZnO, ${\text{TiO}}_2$, and ${\text{WO}}_3$ at $2\mathrm{mg}/\mathrm{kg}$ by oropharyngeal instillation. Quartz was used as positive control to treat animals ($5\mathrm{mg}/\mathrm{kg}$). After 40 h, animals were sacrificed to measure $\text{IL-}1\mathrm{\beta }$ and MCP-1 production in BALF by ELISA. The lung tissues were fixed for H&E staining (three sections for each mouse). Normal distribution was confirmed by Kolmogorov-Smirnov test ($\text{significance}>0.05$). *$p<0.05$ compared with vehicle control by two-tailed Student’s t-test.

Figure 4.

C4.5 decision tree for predicting the inflammatory potential of metal oxide nanomaterials (MeONPs). The root and interior nodes are drawn in rectangles with splitting descriptors inside, and the splitting criteria are under the rectangles. The leaf nodes are presented in oval circles, which are marked as “YES” and “NO,” indicating the nanomaterials in the leaf nodes are predicted to be inflammatory potential and noninflammatory potential, respectively.

Figure 5.

Performance of the continuous model. (A) Plot of experimentally determined data (x-axis) vs. predicted fold change of $\text{IL-}1\mathrm{\beta }$ production ($F{C}_{\text{IL}-1\mathrm{\beta }}$) values (y-axis). The straight solid line represents perfect agreement between experimental and calculated values. Squares represent values predicted for the metal oxides (MeONPs) from the training set; triangles represent MeONPs from the test set. The distance of each symbol from the solid line corresponds to its deviation from the related experimental value. The dotted lines showing the range encompassing 90% of the predictions. (B) Model applicability domain: Williams plot of standardized residuals (${\mathrm{\sigma }}^{*}$) vs. leverage values ($h_i$) for $F{C}_{\text{IL}-1\mathrm{\beta }}$. MeONPs having $h_i>h^{*}$ or $\left|\mathrm{\sigma }\right|>3$ should be identified as outliers.

Figure 6.

Periodic table of predicted inflammatory potential of metal oxide nanomaterials (MeONPs). Sixty-six MeONPs were plotted according to the metal element position in the periodic table. The number in each color square represents MDEI. *, experimental data is unavailable. The predicted value is not shown when the experimental value is available.

Figure 7.

Adsorption energy (|$E_{\text{ad}}$|) of protons onto metal oxide nanomaterials (MeONPs) with different metal atom electronegativity (${\mathrm{\chi }}_{me}$). Bar heights represent calculated values of |$E_{\text{ad}}$|. MeONPs with higher ${\mathrm{\chi }}_{me}$ (black) have much weaker proton adsorption than those with lower ${\mathrm{\chi }}_{me}$ (red). Normal distribution was confirmed by Kolmogorov-Smirnov test ($\text{significance}>0.05$). *$p<0.05$ compared with ip-MeONPs by two-tailed Student’s t-test.

Figure 8.

Proposed schematic image of inflammatory mechanisms by metal oxide nanomaterials (MeONPs). (A) Endocytosis: MeONPs with a positive $\mathrm{\zeta }\text{-potential}$ were most internalized by THP-1 cells and lysosomes. (B) Proton sponge effect. MeONPs with metal atom electronegativity $\le 1.55$ tend to trigger a proton sponge effect, followed by lysosome damages, leakage of lysosomal contents and excess $\text{IL-}1\mathrm{\beta }$ production. (C) Release of toxic ions.