Functional and mechanistic approaches to improve neurotoxicity predictions

Abstract

Current assessments of neurotoxicity (NT) and developmental NT (DNT) using animal models lack throughput and predictivity. These central drawbacks need to be overcome by the development of functional and mechanistic in vitro methods that increase human relevance and throughput for safety assessment.Paper #1 demonstrates the development of an in vitro assay that specifically detects neurite damage. This test, called NeuriTox test, is based on the conditionally immortalized human dopaminergic neuronal cell line LUHMES. The NeuriTox test was converted to a high-throughput version. In a proof-of-concept study, a chemical library containing 80 substances was screened. Test performance parameters were quantified, a prediction model and standard operating procedures (SOP) were developed and submitted to DB-ALM, the database for alternative test methods at the European Centre for Validation of Alternative Methods. Further on, the test was evaluated, together with the U.S. Environmental Protection Agency (US EPA), as one of the best currently available in vitro assays for neurotoxicity assessment.Paper #2 focused on the metabolic background of the LUHMES cells used for the NeuriTox test. We quantified the rearrangements that occur between the immature proliferative and the mature, differentiated stage. By applying transcriptomics, proteomics, metabolomics and fluxomics, we discovered multiple rearrangements on all investigated biological layers. The most important changes occurred on the metabolic level. A dependency on extracellular glutamine was found for proliferating cells, but not for mature neurons. Proliferating cells also displayed a pronounced glycolytic metabolism with no spare glycolytic and mitochondrial capacities. Therefore, they were more sensitive for inhibitors of mitochondrial metabolism than mature neurons, while no sensitivity difference was observed for non-specific toxicants. Thus, we unraveled a toxicant sensitivity dependent on the differentiation-stage.In paper #3, the knowledge on the dominant glycolytic metabolism of the immature LUHMES cells was used to improve the sensitivity of the NeuriTox test to mitochondrial toxicants. By exchanging the medium sugar from glucose (Glc) to galactose (Gal), strong metabolic reprogramming of LUHMES cells was achieved without affecting functional parameters, as e.g. neurite outgrowth kinetics. The performance of the Glc-Gal NeuriTox test was examined with 50 compounds. It showed high sensitivity and specificity to detect both, mitochondrial toxicants and neurotoxicants. For the mitochondrial toxicants, an independent hit follow-up assay was developed to specify the mitochondrial mode of action, on the level of respiratory chain complex impairment.To address an important problem about in vitro screen data obtained here and in databases, the discussion suggests a universal concept for probability based hazard assessment. It overcomes the binary hazard classification of test substances as toxic vs non-toxic.