(274c) Genomic Toxicity Assessment of Engineered Nanomaterials

In this paper, we performed a simpler, faster, and more reliable assay procedure (compare to microarray approach), for toxicity assessment of seven NMs: nano-silver (nAg) and nano-titanium dioxide anatase (nTiO2_a), nano-titanium dioxide rutile (nTiO2_r), carbon black(CB), Fullerene, single-wall carbon nano-tube (SWCNT), SWCNT with H2O2 oxidized (SWCNT_OX). We employed a whole-cell-array library consisting of 91 recombinated E.coli K12 strains with transcriptional GFP fusions covering most known stress response gene (Detailed information of method and data analysis are available in our previous study) [1].

The result showed very dynamic and complex toxin-induced real time gene expression across the stress genes examined, with most of the genes exhibiting different patterns and varying magnitudes of transcription activities over time (as show in Figure 1).

Figure 1 Real-time (temporal) gene expression profiles of 91 stress genes involved in different stress functional categories in exposure to nAg 10mg/L. Color bar left top: natural log of induction factor (lnI). (Red spectrum colors indicate up-regulation, green spectrum colors indicate down-regulation); X-axis: time in minutes (the first data point shown is at 15 minutes after exposure due to moving average); Y-axis: list of genes.

The temporal change in gene expression level reflected the dynamic of the cellular response system and the time sequence for a particular set of gene to be involved, which may depend on the system-level multiple gene activation, signaling pathways and the roles and time-sequence in which they are involved in the stress-response mechanism [2]. And the gene expression profiles seem to be not only chemical-specific but also concentration-dependent indicating that the molecular level genomic activities are very sensitive to not only the type of toxin but also the level of toxin in exposure.

Figure 2 Candidate biomarkers for nAg and nTiO2_a exposure. oxyR, cls, and cspB are the potential bio-markers for nAg exposure (left) and mutT, sodB pbpG are the three potential biomarkers for n_TiO2_a exposure (right).

Compound--specific signature gene expression profile offers the possibility to uncover novel biomarkers of exposure and predict the presence of a class of contaminants, especially the emerging contaminants such as NMs for which biomarkers are not presently available. Based on the gene expression results, we can propose some genes as potential bio-markers based on the following guidelines a) genes that show chemical-specific and concentration-dependent expression pattern; b) genes that are more related to MOA of a contaminant rather than those general stress or function genes. We screened and selected genes that seem to show significant alteration in their expression level and exhibited concentration-dependent patterns as candidate biomarkers for nAg and n_TiO2_a exposure. oxyR, cls and cspB are shown to be the potential bio-markers for nAg exposure and mutT, sodB, pbpG are the three potential bio-markers for n_TiO2_a exposure. Of course, the specificity, sensitivity and reliability of these suggested candidate biomarkers require further investigation and evaluation.

In addition, we observed that the NMs at lower concentration tend to induce more chemical-specific toxicity response, while at higher concentrations, more general global response dominates. The information-rich real time gene expression results allowed for identification of potential biomarkers that can be employed for specific toxin detection and monitoring applications. To link the toxicogenmic results with regulatory benchmarks and conventional toxicity assessment endpoints, we determined the Non Observed Transcriptional Level (NOTEL) values for the NMs and compared them with other established toxicity assessment endpoints such as Biological Oxidative Damage (BOD)[3]. To overcome the challenge of linking multi-dimensional toxicogenomic results with regulatory benchmarks and conventional toxicity assessment endpoints, we proposed a new ToxicoGenomic Response Indicator (TGRI) using a mathematical manipulation that incorporate both the number and level of genes with altered expression with the time of maximum expression level. The TGRI correlated well with those established endpoints (BOD and SSA) indicating that it can be potentially employed as a regulatory benchmark and toxicity assessment endpoints. The results seem to indicate that these toxicogenomics-derived parameters are consistent with established endpoints, therefore can be potentially used as regulatory benchmarks and toxicity assessment endpoints.

Figure 3 The schematic diagram in left shows the approach to generate the new index (TGRI). Right figure shows comparison of new index with biological oxidative damage (BOD) and Non Observed Transcriptional Level (NOTEL). The correlation coefficience between TGRI and BOD is 0.974 (p-value=0.001), 0.947 (p-value=0.004) between TGRI and NOTEL^-1

In our results, most NMs were found to cause oxidative stress as well as cell membrane and transportation damage. But the difference in the specific genes with altered expression upon the exposure to the NMs, suggested the different toxicity mechanisms.

Figure 4 TGRI of each gene functional category for all the several NMs, redox stress, drug resistance and detoxicification all belong to oxidative damage.

nTiO2_r and fullerene has very low toxicity, nTiO2_a, nAg and CB show high toxic to DNA and protein. CB exhibit highest oxidative toxic. SWCNT induced very low level of drug resistance.

All the carbon NMs induced the specific superoxide degradation pathway that is more associated with preventing DNA damage, rather than the other one which is more effective in protecting cytoplasmic superoxide sensitive enzymes. In addition, they all seem activated the multidrug efflux protein. The result indicates CB is slightly more toxic than SWCNT, and they are both more toxic than fullerenes, suggesting varying toxicity depending on geometric dimensions of nanocarbon. Most expressed genes belong to detoxicification and redox stress as a result of oxidative radical generated by the NMs. In contrast to more concentration-dependent pattern showed by metal NMs we observed before, these carbon NMs showed less concentration-dependent pattern.

Several NMs cause DNA damage, however, differences were observed in the genes and their magnitude of alterations involved in the DNA repair system, nTiO2_a induced SOS response via previously identified pathway, nAg at high concentration cause some DNA damage. and CB seemed to induce DNA repair via a pathway different from SOS. Carbon NMs, as also construct by element carbon, may be similar to CB, has a different DNA repair pathway rather than SOS response.

Then maximum lnI as maximum induction factor for each genes were compiled for hierarchical clustering (HCL) analysis that is done by MultiExperiment View (MeV). Distance metric in the two analyses was set as Euclidean distance and linkage method was set as average linkage clustering.

Result show in figure 4, obviously, most metal NMs cluster together, and carbon NMs cluster together. Although CB, fullerene and SWCNT have different structures, as they are all formed by element carbon, there maybe some common toxicity mechanism among them when compared to other NMs.

Figure 5 HCL cluster of all the NMs at all the concentration examined based on the maximum lnI of all the genes.

NMs at low concentrations with non-observable transcriptional level effect clustered together. NMs at high concentrations that led to more global level stress response seem to cluster together. These confirmed the hypothesis that transcription profile has high sensitivity and "conservativeness (signature)".

Our results revealed more detailed transcriptional information on the toxic mechanism of these NMs and led to a better understanding of the MOA of metal and carbon NMs. So our result demonstrated that the proposed prokaryotic toxicogenomic approach using whole-cell arrays can be potentially applied as a feasible method for quantitative toxicity assessment, for large number of emerging contaminants screening and for compound-specific biomarkers identification.

References

[1] Onnis-Hayden, A.; Weng, H. F.; He, M.; Hansen, S.; Ilyin, V.; Lewis, K.; Gu, A. Z., Prokaryotic Real-Time Gene Expression Profiling for Toxicity Assessment. Environ. Sci. Technol. 2009, 43, (12), 4574-4581.

[2]

Keseler, I. M.; Bonavides-Martinez, C.; Collado-Vides, J.; Gama-Castro, S.; Gunsalus,

R.

P.; Johnson, D. A.; Krummenacker, M.; Nolan, L. M.; Paley, S.; Paulsen, I. T.; Peralta-Gil, M.; Santos-Zavaleta, A.; Shearer, A. G.; Karp, P. D., EcoCyc: A comprehensive view of Escherichia coli biology. Nucleic Acids Res. 2009, 37, D464D470.

[3] Bello, D., Hsieh, Shu-Feng, Schmidt, Daniel and Rogers, Eugene, Nanomaterials properties vs. biological oxidative damage: Implications for toxicity screening and exposure assessment. Nanotoxicology 2009, 3, (3), 13.