Nanomaterial categorization by surface reactivity: A case study comparing 35 materials with four different test methods

Abstract Nanomaterials (NMs) can be manufactured in plenty of variants differing in their physicochemical properties. Functional assays can be highly useful to cope with the enormous variability by supporting prioritization and categorization. Oxidative potential (OP) seems to be in particular important in this context and different assays are available. However, their reliability and predictivity are not well-characterized. This study compares four different test methods for measuring NM OP. Reactive oxygen species (ROS) generation was measured on a set of 35 different materials, all extensively characterized with respect to physicochemical properties and most of them with respect to toxicity. Different acellular assays were applied, namely electron spin resonance (ESR) spectroscopy using CPH spin probe and DMPO spin trap, and the ferric reduction ability of serum (FRAS) assay. In addition, protein carbonylation as a marker for oxidative protein damage was analyzed in NRK-52E cells. All assays were assessed individually for their predictivity compared to established toxicological endpoints. We also aimed to identify the optimal assay combination using multivariate logistic regression and other statistical measures. BET surface area-based doses were more suitable to relate surface reactivity to toxicity. In addition, normalization to the deposited dose was advantageous for cellular assays as it improved the predictivity for in vitro as well as in vivo toxicity. The carbonylation assay, potentially in combination with ESR (DMPO spin trap) or FRAS assay, led to the best predictive performance. In summary, we propose a testing strategy for NM OP and demonstrated the applicability in an extended case study on 35 materials. This work is an important contribution towards reliable grouping and testing strategies for NMs.