Abstract
The dynamic regulation of RNA modifications, collectively termed the epitranscriptome, serves as a critical layer of post-transcriptional control during human development. Among over 170 known modifications, N6-methyladenosine (m6A) is the most abundant and well-characterized, influencing mRNA stability, translation, and splicing. However, bulk sequencing methods often obscure the cell-type-specific nuances essential for understanding complex developmental transitions. This study utilizes emerging single-cell epitranscriptomic technologies, including scDART-seq and Rho-seq, to map m6A and other modifications at single-cell resolution across human embryonic stem cell-derived lineages. Our methodology integrated single-nucleotide resolution mapping with advanced machine learning frameworks to identify modification sites in low-input samples. We observed that m6A landscapes are highly heterogeneous across distinct progenitor populations, with specific enrichment in genes associated with lineage commitment and cell cycle regulation. Furthermore, we identified a significant interplay between m6A and non-m6A modifications, such as pseudouridine and 5-methylcytosine, suggesting a coordinated epitranscriptomic code that governs cellular plasticity. Analysis of human preimplantation states and subsequent organogenesis models revealed that m6A writers, such as WTAP, and readers, such as Prrc2a, exhibit stage-specific expression patterns that correlate with the methylation status of key developmental transcripts. These findings provide a comprehensive atlas of the human epitranscriptome during early development, offering insights into the molecular mechanisms of cellular differentiation and the potential role of RNA modifications in developmental disorders. This high-resolution mapping represents a significant advancement in our ability to decipher the transcriptomic systems that steer human development.