You are here:
A Data-Driven Selection of the Cell Lines with Maximal Biological Diversity for High Throughput Transcriptomics-Based Toxicity Testing
Citation:
You, D., C. Willis, Woodrow Setzer, J. Harrill, AND N. Sipes. A Data-Driven Selection of the Cell Lines with Maximal Biological Diversity for High Throughput Transcriptomics-Based Toxicity Testing. North Carolina Society of Toxicology, Durham, NC, September 17 - 23, 2020. https://doi.org/10.23645/epacomptox.25749486
Impact/Purpose:
The US EPA Center for Computational Toxicology and Exposure and the National Toxicology Program are engaged in a collaborative project as part of the Tox21 Consortium focused on developing data-driven methods for selection of complementary cell lines for high-throughput bioactivity screening. Cell types throughout the human body are morphologically heterogeneous, have different functions, express different genes and may have differing sensitivities or responses to xenobiotic chemicals. Therefore, in the context of high-throughput chemical screening, there is no single human-derived cell type that could be used to characterize the variety of transcriptional, phenotypic and/or toxic responses to xenobiotics that may occur in the human body. One way to address this heterogeneity is to test chemicals across multiple cell types. However, when faced with limited resources and the capacity to test only a few cell lines from the thousands that are available in the research community, it may be beneficial to objectively select cell lines that cover complementary (e.g. non-overlapping) signaling pathways or domains of biology. This abstract demonstrates a novel method for data-driven cell line selection based on basal gene expression in cancer and artificially-immortalized cell lines. This information would be of interest to scientists using or contemplating the use of cell-based high-throughput screening technologies for chemical hazard evaluation.
Description:
This abstract is for a presentation at the North Carolina Society of Toxicology (NCSOT) Annual Meeting. High-throughput transcriptomics (HTT) can be an efficient tool to screen large sets of chemicals in vitro, determine transcriptional point-of-departures, and investigate toxicological modes of actions. However, no single cell line can capture the full array of biological responses associated with chemical exposures. Here, we developed a data-driven strategy to select a cell line panel that maximally cover biological targets and allow for complimentary and non-redundant HTT-based chemical screening. Using publicly available microarray data, we assessed the diversity of >1,000 human cancer cell lines of various organ lineage. Euclidean distances calculated on baseline gene expression data showed that cell lines presented with diverse transcriptomic profiles. The key contributing factor for such diversity was tissue origin. Using a “content maximization” approach, we selected a limited number of cell lines to maximize the content, or a generalized high-dimensional volume, of biological diversity in gene expression. The diversity of the selected cell lines was re-assessed in conjunction with 13 hTERT-immortalized cell lines using baseline transcriptome data generated on the TempO-Seq platform. Diversity among cancer cell lines reflected by TempO-Seq data was similar to that observed with microarray data, suggesting that relative similarity between cell lines remain consistent regardless of HTT platform types. In addition, Gene Set Variation Analysis revealed that TempO-Seq data of most cell lines correlated well to gene expression profiles of their respective organ lineage. Lastly, we observed that immortalized cell lines were more similar to each other than to cancer cell lines, regardless of tissue origin. These results enabled us to select a final set of “diverse” cell lines that will be further tested with chemicals to assess biological responses. In summary, we demonstrated a systematic method to choose cell lines that can optimize the coverage of biological space in future HTT-based chemical screening. This abstract does not reflect USEPA policy.
URLs/Downloads:
DOI: A Data-Driven Selection of the Cell Lines with Maximal Biological Diversity for High Throughput Transcriptomics-Based Toxicity Testing
NCSOT_POSTERPITCH33_DYOU_FINAL.PDF (PDF, NA pp, 5352.713 KB, about PDF)