Liquid chromatography high-resolution mass spectrometry (LC-HRMS) shows great potential for expanding our understanding of relevant unknown chemical components present within complex environmental mixtures. This study identified potentially endocrine active components within Minnesota lakewater by prioritizing LC-HRMS features uniquely present at sunfish spawning habitats where male fish showed signs of estrogen agonism. Porewater samples from four locations within the same lake were analyzed using liquid chromatography tandem quadrupole time of flight mass spectrometry (LC-QToF/MS) with positive (ESI+) and negative (ESI−) electrospray ionization. Plasma vitellogenin concentrations of captured male sunfish was used to designate sites as either endocrine active (ACT; 2 sites) or reference (REF; 2 sites). Assuming unique chemical presence at active sites contributed to endocrine activity, features at significantly higher intensities (p-value < 0.05, t-value > t-critical, log-fold change > 0.1; equal variance t-test of log2 transformed data) in ACT sites were then compiled into a suspect search list for feature identification. Adducts and isotopes of prioritized features were deprioritized using pattern recognizing algorithms using mass, retention time, and intensity. Feature identities were reported according to established confidence metrics using spectral libraries and elemental composition algorithms. This LC-HRMS approach identified a number of features omitted by targeted analysis with higher relative abundances in ACT sites, including plant essential oils, fatty acids, and mycotoxins. Multivariate analysis determined whether features were either present at both sites (AB) or unique to individual ACT sites (A or B). Detection frequency across datasets indicated bias in feature prioritization influenced by the chosen sampling method and sample acquisition mode. The majority of features prioritized by this workflow remain tentatively identified or unidentified masses of interest, reflective of current limitations in shared spectral libraries for soft ionization analyses. Strategies similar to this workflow have the potential to reduce bias in database-driven toxicological prioritization frameworks.

液相色谱高分辨率质谱法（LC-HRMS）具有巨大的潜力，可以扩大我们对复杂环境混合物中存在的相关未知化学成分的了解。这项研究通过优先考虑在雄性鱼类显示雌激素激动迹象的翻车鱼产卵栖息地中独特存在的LC-HRMS特征，确定了明尼苏达州湖水中潜在的内分泌活性成分。使用液相色谱串联四极杆飞行时间质谱（LC-QToF / MS）分析正离子（ESI +）和负离子（ESI-）电喷雾电离的来自同一湖中四个位置的孔隙水样品。捕获的雄性翻车鱼的血浆卵黄蛋白原浓度被指定为内分泌活性部位（ACT； 2个部位）或参考部位（REF； 2个部位）。p值<0.05，t值> t临界，对数倍变化> 0.1; 等方差t然后，将ACT站点中的log2转换数据的测试）编译为可疑搜索列表以进行特征识别。使用质量，保留时间和强度的模式识别算法对优先特征的加合物和同位素进行优先排序。使用光谱库和元素组成算法，根据已建立的置信度指标报告特征身份。这种LC-HRMS方法可识别出目标分析中ACT部位相对丰度较高的许多特征，这些特征包括植物精油，脂肪酸和霉菌毒素。多变量分析确定特征是否同时存在于两个位点（AB）或对于单个ACT位点而言是唯一的（A或B）。整个数据集的检测频率表明，特征优先级的偏差受所选采样方法和样本采集模式的影响。该工作流程优先考虑的大多数功能仍是暂时确定的或未确定的感兴趣质量，反映了共享频谱库中用于软电离分析的当前限制。与该工作流程类似的策略有可能减少数据库驱动的毒理学优先级划分框架中的偏见。