Novel Approach to Screen Endocrine-Disrupting Chemicals via Endocrine-Enhanced Reduced Human Transcriptome

Endocrine-disrupting chemicals (EDCs) can interfere with multiple pathways and trigger different modes of action. Thus, the traditional EDC in vitro screening processes often require a battery of bioassays to cover multiple target pathways. Here we developed an endocrine-enhanced reduced human transcriptome (ERHT) focused on hormone receptor signaling induced by the EDCs regulating specific genes. ERHT was developed based on 1200 prioritized genes covering 110 endocrine-related biological pathways across eight potential adverse outcomes. The ability of this approach to identify EDCs was derived from machine learning of 1068 dose-dependent transcriptome profiles and enhanced by quantifying chemical-induced critical pathway responses, and thus, it demonstrated excellent classification performance (AUC = 0.84 ± 0.03) in internal cross-validation. We ultimately applied this approach to known EDCs and inactive substances to validate the reliability of this approach. Through external validation on 210 chemicals, the extrapolation accuracy exceeded 80%, demonstrating the outstanding practical performance of this approach. Meanwhile, the pathway responses induced by the same chemical were consistent with the experimental results reported by multiple sequencing platforms, highlighting the robustness of this approach. The above results demonstrate that this approach can provide novel insights for EDCs’ high-throughput screening and comprehensive toxic mechanisms through biological pathways.内分泌干扰物 (Endocrine-disrupting chemicals，EDCs) 可干扰多条信号通路并引发不同作用模式。因此，传统的内分泌干扰物体外筛选流程通常需要依托一系列生物检测实验，以覆盖多类靶标通路。本研究开发了内分泌增强型精简人类转录组 (ERHT)，其聚焦于由调控特定基因的内分泌干扰物所诱导的激素受体信号通路。ERHT基于1200个优先筛选基因开发，这些基因覆盖了8类潜在不良结局相关的110条内分泌相关生物学通路。该方法识别内分泌干扰物的能力源自对1068份剂量依赖性转录组图谱的机器学习分析，并通过量化化学物质诱导的关键通路响应得到强化；因此，其在内部交叉验证中展现出优异的分类性能（AUC = 0.84 ± 0.03）。本研究最终将该方法应用于已知内分泌干扰物与无活性物质，以验证其可靠性。通过对210种化学物质的外部验证，该方法的外推准确率超过80%，彰显了其出色的实际应用性能。与此同时，同一化学物质诱导的通路响应与多组测序平台报道的实验结果相一致，凸显了该方法的稳健性。上述结果表明，该方法可为内分泌干扰物的高通量筛选及基于生物学通路的全面毒理机制研究提供全新视角。