Back

Functional Genomic Analyses Reveal transfer RNA Modification Enzymes as Risk Genes for Bipolar I Disorder and Schizophrenia

Bingwu, L.; Wang, O.

2026-01-27 psychiatry and clinical psychology
10.64898/2026.01.26.26344826 medRxiv
Show abstract

Enzyme-mediated RNA modifications provide a versatile and dynamic regulatory layer that fine-tunes gene expression programs throughout brain development and in mature neural function. More than 100 RNA modification proteins (RMPs) have been identified to drive the dynamics of RNA modifications; yet how genetic variation in human RMP genes contributes to the risks of neurological disorders has not been systematically examined. Here we have curated 123 RMPs corresponding to 31 RNA modifications on 6 RNA biotypes, and integrated genome-wide association studies (GWAS) with gene-based association testing, as well as QTL-informed approaches such as transcriptome-wide association studies (TWAS), summary data-based Mendelian randomization (SMR), and Bayesian colocalization, to systematically prioritize RMP genes underlying risk for neuropsychiatric disorders. Across multiple complementary testings, NSUN2, encoding an RNA cytosine-5 methyltransferase, emerged as a reproducibly associated locus for bipolar I disorder (BIPI) and schizophrenia (SCZ). The results show NSUN2 as a neuroprotective protein; shared cis-regulatory variations influencing both isoform-specific NSUN2 expression and disease risk. Several RMP genes, including NSUN6, TYW5, TRMT61A, and QTRT1, showed significant gene-level associations with BIPI or SCZ, implicating tRNA modification pathways more broadly. Together, these findings indicate that common genetic risks for BIPI and SCZ converge on regulatory variation affecting RNA modification machinery, nominating dysregulation of tRNA as a molecular pathway contributing to psychiatric disease susceptibility.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.