Toxicity of Metal Oxide Nanoparticles: Looking through the Lens of Toxicogenomics

Abstract

The impact of solubility on the toxicity of metal oxide nanoparticles (MONPs) requires further exploration to ascertain the impact of the dissolved and particulate species on response. In this study, FE1 mouse lung epithelial cells were exposed for 2-48 h to 4 MONPs of varying solubility: zinc oxide, nickel oxide, aluminum oxide, and titanium dioxide, in addition to microparticle analogues and metal chloride equivalents. Previously published data from FE1 cells exposed for 2-48 h to copper oxide and copper chloride were examined in the context of exposures in the present study. Viability was assessed using Trypan Blue staining and transcriptomic responses via microarray analysis. Results indicate material solubility is not the sole property governing MONP toxicity. Transcriptional signaling through the 'HIF-1α Signaling' pathway describes the response to hypoxia, which also includes genes associated with processes such as oxidative stress and unfolded protein responses and represents a conserved response across all MONPs tested. The number of differentially expressed genes (DEGs) in this pathway correlated with apical toxicity, and a panel of the top ten ranked DEGs was constructed (Hmox1, Hspa1a, Hspa1b, Mmp10, Adm, Serpine1, Slc2a1, Egln1, Rasd1, Hk2), highlighting mechanistic differences among tested MONPs. The HIF-1α pathway is proposed as a biomarker of MONP exposure and toxicity that can help prioritize MONPs for further evaluation and guide specific testing strategies.

Keywords: BMC modelling; canonical pathways; enrichment analysis; nanomaterials; nanotoxicology; omics; potency ranking.

Figure A1

Schematic of the ‘HIF-1α Signaling’ IPA canonical pathway. A legend is available in Figure A2.

Figure A2

Legend for IPA canonical pathway schematics.

Figure 1

Percent viable cell density of FE1 cells following 2, 24, and 48 h of exposure to MONPs, MOMPs, and metal chlorides compared to time-matched medium controls. Error bars indicate the standard deviation (n = 3–4). Graphs were labeled based on the type of metal oxide. Yellow: MONPs. Blue: MOMPs. Green: metal chloride salts. MO: metal oxide. MeCl: metal chloride. Statistical significance against time-matched medium controls was determined through a 2-way ANOVA with a Dunnett’s post-hoc. *: p < 0.05. **; p < 0.01. ***; p < 0.001.

Figure 2

Total number of differentially expressed genes following 2–48 h exposure to MONPs, MOMPs, and metal chloride salts. Graphs were labeled based on the type of metal oxide. Yellow: MONPs. Blue: MOMPs. Green: metal chloride salts. MO: metal oxide. MeCl: metal chloride.

Figure 3

Total number of perturbed IPA canonical pathways following 2–48 h exposure to MONPs, MOMPs, and metal chloride salts. Graphs were labeled based on the type of metal oxide. Yellow: MONPs. Blue: MOMPs. Green: metal chloride salts. MO: metal oxide. MeCl: metal chloride. Significantly enriched pathways were combined in the case of ZnO MPs from enrichment of both the low fold-change and high fold-change datasets (see Section 4).

Figure 4

Hierarchical clustering of metal oxide and metal chloride-induced canonical pathway perturbation The clustering was conducted on the z-scores of significantly enriched pathways. The red dashed line indicates where the dendrogram was cut to produce groupings. Each grouping was numbered and highlighted with a colored box.

Figure 5

Perturbation of the ‘HIF-1α Signaling’ pathway by MONPs, MOMPs, and metal chloride samples from 2–48 h. (a) Number of DEGs, pathway coverage ratio (# DEGs/total genes in the pathway), p-value, and z-score for all samples for which this pathway was enriched. (b) Heatmap showing clustering of samples and DEGs associated with the pathway. Lines indicate groups formed from the sample dendrogram. (c) Percent viable cell density plotted against the coverage ratio of the ‘HIF-1α Signaling’ pathway for samples where the pathway was significantly enriched. A Spearman’s correlation was conducted, with the resulting correlation coefficient and p-value displayed and a trendline fit to the graph.

Figure 6

Heatmap showing the differential expression of the top-10-ranked DEGs from the ‘HIF-1α Signaling’ pathway across all samples for which this pathway is significantly enriched. The values in each cell indicate the fold change over the control. The left-most DEG is the top-ranked, whereas the right-most DEG is the bottom-ranked. Blank cells indicate no significant differential expression. Red: increased fold change. Green: decreased fold change.

Figure 7

BMDS-based BMC modeling of viable cell density reduction following (a) 24 h and (b) 48 h exposure to metal oxides and metal chlorides, with (c) differences in potency for each metal variety. Benchmark response: 0.5 (hybrid risk). The ‘x’ indicates the BMC. Left and right bars indicate lower and upper 95% confidence intervals around the BMC, respectively. Potencies for different compounds are considered equivalent if the confidence intervals overlap.

Figure 8

The transcriptomic-based BMC determined through BMDExpress2 BMC modeling for (a) 2 h, (b) 24 h, and (c) 48 h metal oxide and metal chloride exposures, with (d) differences in potency for each metal variety. Benchmark response: 1 (standard deviation). The ‘x’ indicates the BMC. Left and right bars indicate lower and upper 95% confidence intervals around the BMC, respectively. Potencies for different compounds are considered equivalent if the confidence intervals overlap.