Profiling Tox21 Compounds That Activate p53 Signaling After Metabolic Activation

Although increasing numbers of chemicals are used in commerce, most of them have little-to-no toxicity information. To address this data gap, the Toxicology in the 21^st Century (Tox21) used a battery of assays to evaluate environmental chemicals in a quantitative high throughput screening (qHTS) platform. Most compounds are metabolized in the body, but the majority of in vitro HTS assays do not have metabolic capability due to the lack of metabolic enzymes in cell lines. To compensate for this, introduction of metabolic capacity through external methods such as adding the S9 mix is commonly used in toxicity testing. However, the working concentration of S9 (i.e., usually used at 10%) is toxic to the cells. Thus, it requires additional washing steps that are not suitable for HTS. We used the CellSensor p53RE beta-lactamase reporter assay as a model to incorporate metabolic capability into qHTS assays in a 1536-well format. In order to supply external metabolism in a homogenous assay format, we optimized the assay with rat liver microsomes (RLM) or human liver microsomes (HLM). The working concentrations are 5.6% and 2.6% for RLM and HLM, respectively, which did not cause cytotoxic effects. The Tox21 10K compound library was screened using the p53RE assay under three conditions: without microsomes, with RLM, or with HLM. The p53RE assay with RLM or HLM identified 73 and 6  additional compounds, respectively, which were not identified in the p53RE assay without microsomes. We next tested if the induced activity was due to metabolic activation using the p53RE assay with heat-attenuated microsomes or the p53RE assay without NADPH, a co-factor of major metabolic enzymes. Forty-four of the 73 compounds were less potent under these conditions compared to the p53RE assay with RLM, suggesting that these compounds need to be metabolized to become bioactive. On the other hand, none of the six compounds treated with HLM were less potent under these conditions. Then, the 44 compounds were further examined in the metabolic intrinsic clearance assay. The results showed that 36 compounds were metabolized by rat liver microsomes. Finally, the intrinsic clearance was compared between RLM and HLM using metabolism prediction software. We found that RLM showed higher intrinsic clearance than HLM for most of the tested compounds, consistent with our experimental results that RLM identified more compounds than HLM did. Taken together, our study indicates that this approach can identify compounds that will not be detected by standard screening methods that lack metabolic capability.