A Model To Predict Nanoparticle Toxicity

As nanoparticles increasingly make their way into consumer products and the environment, toxicologists want to understand their effects on human health. Ideally, they’d like to develop models that predict a material’s toxicity based on its chemical properties. Now a new study has a reported the first such model for metal oxide nanoparticles that uses the materials’ electrical and solubility properties. Metal oxide nanoparticles are semi conducting materials that drive oxidation and reduction reactions in devices such as fuel cells and electronics. Previous studies have linked the materials to health problems, such as inflammation in the lungs of welders who inhale fumes containing the compounds. Andre Nel, of the University of California, Los Angeles, wondered if the particles’ electrical properties linked to their toxicity. The reactivity of each metal oxide material depends on its band gap, the energy needed for it to pick up an electron from another compound or material, or to donate one. If adsorbed by a cell, Nel figured, nanoparticles would certainly encounter molecules also engaged in oxidation and reduction reactions. He and his team thought that if a nanoparticle’s band gap matched the energies required to drive these cellular reactions, the materials could disrupt the cell’s well-regulated oxidation and reduction reactions. Biochemists have known that such disruptions lead to cellular damage and inflammation. To test their hypothesis, the researchers studied 24 metal oxide nanomaterials. Based on the materials’ band gaps and known cellular oxidation-reduction energies, the team predicted that six were likely to be toxic: TiO2, Ni2O3, CoO, Cr2O3, Co3O4, and Mn2O3. They then exposed human bronchial and mouse blood cells to each of the 24 nanoparticles. The researchers found that five of the six predicted metal oxides—all but TiO2—caused toxic effects in the cells, including reducing cell survival by as much as 80% compared to cells not exposed to nanoparticles. Two materials not predicted by the team’s model, CuO and ZnO, were also toxic. Unlike most of the other nanoparticles tested, both of these metal oxides dissolve in water to release toxic metal ions, the researchers say. Based on these results, Nel says that the toxicity model will have to include not only information about a material’s band gap, but also its solubility. Next, the team tested the nanoparticles’ effects on live animals’ health by forcing mice to inhale suspensions of the materials. The five metal oxides that caused toxicity to cells also caused inflammation in the animals’ lungs: Numbers of white blood cells and levels of cytokines, two signs of inflammation, were at least fivefold as high in animals exposed to the five metal oxides as in mice exposed to the other particles. Wolfgang Kreyling, of Helmholtz Centre Munich, calls the study “a really good first attempt” to predict nanoparticle toxicity. He says that its strength is that it connects cell toxicity to animal data. But this model probably won’t work for everything, he says: Each preparation of nanoparticles is unique, with its own sizes, shapes, and purity levels—properties that could dramatically influence toxicity. Nel hopes industry and government agencies will use his team’s model to make decisions about which materials to test in detailed, animal-based analyses. By eliminating time-intensive blind screens for toxicity, he says the model could make safety assessment an integral part of its design, “rather than a post-hoc cleanup exercise.”

Chemical & Engineering News, 23 April 2012 ;http://pubs.acs.org/cen/news ;