Objective: Preeclampsia is a common and serious complication of pregnancy, posing a threat to maternal and fetal safety due to the lack of effective biomarkers and treatment strategies. This study aimed to identify potential biomarkers that can be used to predict preeclampsia and identify the molecular mechanisms of preeclampsia pathogenesis and drug prediction at the transcriptome level.
Methods: We analyzed differential expression genes (DEGs) in preeclampsia and non-preeclampsia groups in the GSE75010 dataset, cross-linking with extracted inflammatory response-related genes to obtain differentially expressed inflammation-related genes (DINRGs). Enrichment analysis and protein-protein interaction (PPI) networks were constructed to understand the functions and enrichment pathways. Machine learning models were used to identify key genes associated with preeclampsia and build a nomogram in the training set, which was validated in the validation set. The R package RcisTarget was used to predict transcription factors, and Cytoscape was used to construct miRNA-mRNA pathways, which could identify the molecular mechanisms. Then, we conducted molecular docking of the obtained key genes INHBA (inhibin subunit beta A), OPRK1 (opioid receptor kappa 1), and TPBG (trophoblast glycoprotein), as well as predicted transcription factors with drug molecules. Additionally, the CIBERSORT method explored the differences in immune cell infiltration between preeclampsia and non-preeclampsia samples based on the GSE75010 dataset.
Results: A total of 69 DINRGs associated with preeclampsia patients were screened. INHBA, OPRK1, and TPBG were the key genes based on machine learning models. A nomogram for prediction was further constructed, and the receiver operating curves (ROCs) showed good performance. Based on the transcriptome level of key genes, we proposed that RELA-miR-548K/miR-1206-TPBG may be a potential RNA regulatory pathway regulating the progression of early preeclampsia. Molecular docking suggested the effectiveness of curcumin in the treatment of preeclampsia. Additionally, regulatory T cells (Tregs) and resting mast cells were significantly different between the two groups.
Conclusion: In summary, we identified three key inflammation-associated genes, namely INHBA, OPRK1, and TPBG, which can be used as potential genetic biomarkers for preeclampsia prediction and treatment, and established a nomogram as a predictive model. Additionally, we provided insights into the mechanisms of preeclampsia development at the transcriptome level and performed corresponding drug predictions.
Keywords: curcumin; inflammation; machine learning; preeclampsia; regulatory pathways.
Copyright © 2022 Wang, Li and Zhao.