Among the more than 150 chemically distinct modifications that have been identified in cellular RNA, 5-methylcytosine — abbreviated m5C — has long occupied an ambiguous position in the landscape of epitranscriptomics. Present in transfer RNAs, messenger RNAs, and several non-coding RNA species, m5C is deposited by a family of enzymes whose biological importance has become progressively clearer over the past decade. Chief among them is NSUN2, a human RNA methyltransferase whose dysregulation has been linked to cancers of multiple tissue types, to intellectual disability, and to a spectrum of other neurological conditions. Despite this clinical significance, the molecular logic governing which of the countless cytidines distributed across the transcriptome NSUN2 chooses to modify has remained poorly understood — a gap that a study published in Nature on 27 May 2026 by Jacob Canepa, Victor M. Ruiz-Arroyo, Netanya Schlamowitz, and Yunsun Nam at the University of Texas Southwestern Medical Center now begins to close.
The challenge of understanding NSUN2 substrate selectivity is rooted in a paradox of versatility. The enzyme modifies cytidines across structurally and functionally disparate RNA classes, suggesting that its recognition mechanism cannot rely solely on nucleotide sequence. Using cryo-electron microscopy, the investigators captured NSUN2 in complex with RNA substrates at multiple stages of its catalytic cycle — before, during, and after the methyl transfer reaction. These structural snapshots, resolved at near-atomic resolution, revealed that NSUN2 does not rely primarily on primary sequence to identify its targets. Instead, the enzyme reads the three-dimensional architecture of the RNA molecule: it recognizes double-stranded RNA stem structures that flank the modification site, making intimate contacts with the phosphate backbone and with specific base-pair geometries that are largely independent of the underlying nucleotide sequence. The modification site itself must be presented at the 5′ end of the first stem of a characteristic dual-stem structure, and must be embedded within a short sequence motif designated CNNRR (where N denotes any nucleotide and R a purine), a combination of shape and sequence that effectively constitutes the molecular address to which NSUN2 is directed.
Beyond substrate recognition, the structural data illuminates a broader mechanistic picture. The investigators found that NSUN2 undergoes a series of conformational rearrangements as it progresses through successive steps of its catalytic cycle, adopting distinct active-site geometries that appear to be tightly coupled with catalytic competence. The enzyme forms a covalent intermediate with its RNA substrate — an obligatory step in cytosine methylation by S-adenosylmethionine-dependent methyltransferases — and the structures capture both this covalent adduct state and its resolution. Collectively, these mechanistic details help explain a longstanding puzzle: how NSUN2 can modify cytidines in environments as different as the anticodon loop of a transfer RNA and the coding region of a messenger RNA, while maintaining a degree of selectivity that prevents promiscuous, deleterious modification of transcripts that share only superficial sequence similarity.
The therapeutic implications of this structural framework are considerable. NSUN2 is overexpressed in several cancer types, where elevated m5C methylation of specific messenger RNA targets has been proposed to promote oncogenic translation programs; conversely, NSUN2 loss-of-function mutations give rise to autosomal recessive intellectual disability by disrupting tRNA modification in neurons. Understanding the structural determinants of substrate recognition thus creates a molecular foundation upon which inhibitors can be rationally designed — molecules capable of disrupting the specific protein-RNA interface identified in the study rather than merely targeting the enzyme’s catalytic machinery indiscriminately. The derivation of a minimized NSUN2 substrate — a synthetic dual-stem RNA construct bearing the CNNRR motif that recapitulates the preferred features of a natural NSUN2 target — is itself a tool of immediate practical utility for screening campaigns.
Several important caveats temper these prospects. The cryo-EM structures, while of high resolution, were obtained with a limited set of substrate RNAs, and it remains to be established whether the dual-stem recognition principle operates identically across the full range of NSUN2 targets in vivo — particularly for mRNA substrates, which adopt far more dynamic and contextually variable conformations than the relatively rigid tRNA molecules that formed the basis of the present analysis. The relationship between m5C deposition and downstream molecular phenotype also remains incompletely resolved: in most cellular contexts, the identities of the proteins that read and interpret m5C marks on mRNA have not been unambiguously defined, leaving a gap between the structural mechanism elucidated here and the gene-regulatory consequences that make NSUN2 clinically relevant. These are limitations the authors acknowledge, and they underscore how much remains to be learned. Nonetheless, the study represents a conceptually significant advance: by establishing that NSUN2 substrate identity is encoded primarily in RNA structure rather than primary sequence, it rewrites the rules by which the epitranscriptomic code is read — and opens, in doing so, a new chapter in the pharmacology of RNA-modifying enzymes.
Leave a Reply