The development of various chemical methods has enabled scientists to decipher the distribution features and biological functions of RNA modifications in the past decade. In addition to modifying non-coding RNAs such as tRNAs and rRNAs, N⁶-methyladenosine (m⁶A) has been proven to be the most abundant internal chemical modifications on mRNAs in eukaryotic cells, and is also the most widely studied mRNA modification to date. Extensive studies have repeatedly demonstrated the important functions of m⁶A in various biological conditions, ranging from embryonic organ development to adulthood organ function and pathogenesis. Unlike DNA methylation which is relatively stable, the reversible m⁶A modification on mRNAs is highly dynamic and easily influenced by various internal or external factors, such as cell type, developmental stage, nutrient supply, circadian rhythm, and environmental stresses. 

In this Account, we reviewed our previous findings on the site selectivity mechanisms regulating m⁶A formation, as well as the physiological roles of m⁶A modification in cerebellum development and long-term memory consolidation. In our initial efforts to profile m⁶A in various types of mouse and human cells, we surprisingly found that the sequence motifs surrounding m⁶A sites were often complementary with the seed sequences of miRNAs. By manipulating the abundance of the miRNA biogenesis enzyme Dicer or individual miRNAs, or mutating miRNA sequences, we were able to reveal a new role of nucleus localized miRNAs, which is to guide the m⁶A methyltransferase METTL3 to bind to mRNAs and to promote m⁶A formation. As a result, we partially answered the question of why only a small proportion of m⁶A motifs within an mRNA could have m⁶A modification at a certain time point. We further explored the functions of m⁶A modification in regulating brain development and brain functions. We found that cerebellum had the most severe defects when knocking out Mettl3 in developing mouse embryonic brain, and revealed that the underlying mechanisms could be attributed to aberrant mRNA splicing and enhanced cell apoptosis under m⁶A deficit conditions. On the other hand, knocking out Mettl3 in postnatal hippocampus did not cause morphological defects in mouse brain, but impaired the efficacy of long-term memory consolidation. Under learning stimuli, formation of m⁶A modifications could be detected on transcripts encoding proteins related to dendrite growth, synapse formation, and other memory related functions. Loss of m⁶A modifications on these transcripts would result in translation deficiency and reduced protein production, particularly in the translation of early response genes, therefore would compromise the efficacy of long-term memory consolidation. Interestingly, excessive training sessions or increased training intensity could overcome such m⁶A deficiency related memory defects, which is likely due to the longer turnover cycle and the cumulative abundance of proteins throughout the training process. In addition to revealing the roles of m⁶A modification in regulating long-term memory formation, our work also demonstrated an effective method for studying memory efficacy. As the lack of appropriate model for studying memory efficacy has been a long-lasting problem in the field of neural science, our hippocampus-specific postnatal m⁶A knockout model could also be utilized to study other questions related to memory efficacy.