You are here:
Development of omcis biomarkers for estrogen exposure using mRNA, miRNA and piRNAs
Citation:
Toth, Greg, J. Martinson, D. Bencic, D. Lattier, M. Kostich, AND A. Biales. Development of omcis biomarkers for estrogen exposure using mRNA, miRNA and piRNAs. AQUATIC TOXICOLOGY. Elsevier Science Ltd, New York, NY, 235:105807, (2021). https://doi.org/10.1016/j.aquatox.2021.105807
Impact/Purpose:
We developed an mRNA-based omics classifier that is able to accurately discriminate estrogen exposed vs. non-exposed larval fathead minnow over a broad range of concentrations likely to be found in real-world exposures. We further identify small non-coding RNAs that are responsive to estrogenic exposure and demonstrate that small ncRNA are a viable pool of RNA for biomarker development.
Description:
The number of chemicals requiring risk evaluation exceeds our capacity to provide the underlying data using traditional methodology. This has led to an increased focus on the development of novel approach methodologies. This work aimed to expand the panel of gene expression-based biomarkers to include estrogens, to identify training strategies that maximize the range of applicable concentrations, and to evaluate the potential for two classes of small non-coding RNAs (sncRNAs), microRNA (miRNA) and piwi-interacting RNA (piRNA), as biomarkers. To this end larval Pimephales promelas (96 hpf +/- 1h) were exposed to 5 concentrations of 17- ethinylestradiol (0.12, 1.25, 2.5, 5.0, 10.0 ng/L) for 48 h. For mRNA-based biomarker development, RNA-seq was conducted across all concentrations. For sncRNA biomarkers, small RNA libraries were prepared only for the control and 10.0 ng/L EE2 treatment. In order to develop an mRNA classifier that remained accurate over the range of exposure concentrations, three different training strategies were employed that focused on 10 ng/L, 2.5 ng/L or a combination of both. Classifiers were tested against an independent test set of individuals exposed to the same concentrations used in training and subsequently against concentrations not included in model training. Models trained on 10 ng/L EE2 performed poorly when applied to lower concentrations, whereas models trained using either the 2.5 ng/L or combination treatments performed similarly well across all concentrations tested except for 0.125 ng/L. With the exception of the 10 ng/L treatment, few mRNA differentially expressed genes (DEG) were observed, however, there was marked overlap of DEGs across treatments. Overlapping DEGs have well established linkages to estrogen and several of the 81 DEGs identified in the 10 ng/L treatment have been previously utilized as estrogenic biomarkers (vitellogenin, estrogen receptor-). Following multiple test correction, no sncRNAs were found to be differentially expressed, however, both miRNA and piRNA classifiers were able to accurately discriminate control and 10 ng/L exposed organisms with AUCs of 0.83 and 1.0 respectively. We have developed a highly discriminative estrogenic mRNA biomarker that is accurate over a range of concentrations likely to be found in real-world exposures. We have demonstrated that both miRNA and piRNA are responsive to estrogenic exposure, suggesting the need to further investigate their regulatory roles in the estrogenic response
URLs/Downloads:
DOI: Development of omcis biomarkers for estrogen exposure using mRNA, miRNA and piRNAs