Meadows, D.; Hargis, H. L.; Ellis, A.; Mangone, M.

MicroRNAs (miRNAs) are important post-transcriptional regulators that play key roles in development, cell identity, and disease. While significant progress has been made in identifying miRNA targets, the mechanisms governing miRNA strand selection during biogenesis remain poorly understood. Here we investigate the hypothesis that miRNA strand selection is not random but follows specific thermodynamic and regulatory rules that vary across tissues and developmental stages. Using Caenorhabditis elegans as a model system, we employed a novel high-throughput qPCR-based approach (HiTmiSS) to quantify miRNA strand selection throughout development and within somatic tissues, coupled with a novel machine learning model trained with our validated wet bench data, that can predict miRNA strand selection with high accuracy. When applied to vertebrates, our ML model has identified specific conserved features with important evolutionary implications for miRNA biogenesis. Our findings indicate that miRNA strand preference is widespread but varies dynamically, with a subset of miRNAs switching strand usage depending on developmental stage or tissue types. Furthermore, our analysis suggests that thermodynamic asymmetry alone does not fully explain strand selection, implicating additional cellular factors. This work provides a comprehensive framework for studying miRNA biogenesis, offering insights into regulatory principles governing strand selection and its broader implications in gene regulation, development, and disease across metazoans.