Characterizing mechanisms of toxicological action for data-poor priority pollutants in Canada using novel model systems and transcriptomic methodologies

我们的环境中有数以千计的化学品从未接受过毒理测试(称为“数据贫乏”)。造成这种情况的根本原因是依赖于不高吞吐量或不经济的传统测试。这些测试也没有揭示告知物种相关性和个体易感性的毒理学机制。转录组学(全球基因表达谱)为了解和预测毒理学效应提供了丰富的信息来源。转录组学尚未被充分利用，用于监管决策和风险评估。目的我的研究和培训计划的总体目标是使化学毒性试验的监管评估现代化。我们的目标是开发和应用实用的方法来有效地识别毒理学机制并从转录数据中预测危害。具体目标我们将在人体体外模型中应用高通量转录组学和生物信息学，以解决数据贫乏的高度关注的化学物质中的知识缺口。项目将侧重于两个化学品家族，即全氟和多氟烷基物质和双酚，因为它们的广泛使用以及这些家族中大量缺乏数据的化学品。这项工作将分为三个阶段：阶段1a-假设生成。深入挖掘现有和新的基因组数据集，以破译毒性机制并识别毒性的物理化学驱动因素；阶段1b-方法开发。应用/开发机器学习工具，根据转录图谱预测毒理学效应，以提高数据解释的效率和重复性；第二阶段--假设检验和方法验证。进行验证实验，以确认第一阶段、第三阶段-发现中的分子、生化和表型预测。应用单细胞RNA测序来确定细胞类型和个体特定的机制、潜力和影响。该项目每年资助5名研究生，共培养4名硕士、3名博士、2名硕士、3名博士。该计划提供了一个独特的培训场所，重点是将研究创新转化为监管决策、政策和实践。新颖性和意义这项工作将提供有关缺乏数据的化学品毒性的关键信息，并揭示如何使用新模型和人工智能来预测毒理效应。分析方法可以应用于绿色化学开发管道，以减少令人遗憾的更换的可能性。这项工作将影响与优先化学品类别和在监管决策中使用基因组数据有关的政策。这些工具提供了新的方法，增加了分析吞吐量，节省了化学评估的成本/时间，这对行业和监管机构产生了影响。重要的是，该培训计划将为加拿大培养下一代毒物学家和风险评估专家。

There are thousands of chemicals in our environment that have never been subject to toxicological testing (referred to as `data-poor'). The root cause of this is reliance on traditional tests that are not high-throughput or economical. These tests also do not reveal toxicological mechanisms that inform species relevance and individual susceptibility. Transcriptomics (global gene expression profiling) provides a rich source of information to understand and predict toxicological effects. Transcriptomics has yet to be fully harnessed for use in regulatory decision-making and risk assessment. Objectives The overarching goal of my research and training program is to modernize regulatory evaluation of chemical toxicity testing. We aim to develop and apply pragmatic approaches to efficiently identify toxicological mechanisms and predict hazards from transcriptomic data. Specific objectives We will apply high-throughput transcriptomics and bioinformatics in human in vitro models to address knowledge gaps in data-poor chemicals of high concern. Projects will focus on two chemical families, per- and poly-fluorinated alkyl substances and bisphenols, because of their widespread use and the large number of data-poor chemicals within these families. The work will be broken into three phases: Phase 1a - Hypothesis generation. Deeply mine existing and new genomic data sets to decipher mechanisms of toxicity and identify physico-chemical drivers of toxicity; Phase 1b - Methods development. Apply/develop machine-learning tools to predict toxicological effects from transcriptomic profiles to increase the efficiency and reproducibility of data interpretation; Phase 2 - Hypothesis testing and methods validation. Conduct validation experiments to confirm molecular, biochemical and phenotypic predictions from Phase 1; Phase 3 - Discovery. Apply single-cell RNA-sequencing to identify cell type- and individual-specific mechanisms, potencies, and effects. The program supports 5 graduate students each year, for a total of 4 MSc, 3 PhD completed, and 2 MSc, 3 PhD initiated. The program provides a unique training ground focused on translation of research innovation into regulatory decision-making, policy and practice. Novelty and significance The work will provide critical information on the toxicity of data-poor chemicals and reveal how new models and artificial intelligence can be used to predict toxicological effects. The analytical approaches can be applied in green-chemistry development pipelines to reduce the probability of regrettable replacements. The work will impact policy related to priority chemical categories and use of genomic data in regulatory decision-making. The tools provide new approaches that increase analytical throughput and save cost/time for chemical evaluations, which impacts industry and regulatory agencies. Importantly, the training program will produce the next-generation of toxicologists and risk assessment specialists for Canada.