Nonsense-mediated RNA decay (NMD) is a highly conserved pathway originally discovered by virtue of its ability to recognize and degrade abnormal mRNAs harboring nonsense and frameshift mutations 1, 2, 3. These mutations generate premature termination codons (PTCs) that are recognized as aberrant by the NMD pathway. The ability of NMD to degrade PTC-bearing mRNAs can be protective, as the PTCs lead to the translation of truncated, potentially deleterious dominant-negative proteins 1, 4, 5, 6. Subsequently, it was discovered that NMD degrades subsets of normal RNAs, with loss or disruption of NMD leading to dysregulation of 5–20% of the normal transcriptome in species spanning the phylogenetic scale 7, 8, 9. This discovery raised the possibility that the function of NMD extends beyond quality control, a notion supported by scores of subsequent studies showing that NMD factors are critical for many fundamental processes, including development, differentiation, cell cycle regulation, cell survival, the integrated stress response, and autophagy 2, 10, 11. Disruptions of NMD are also implicated in causing a variety of disorders, including neurodevelopmental disorders, intellectual disability (ID), and malignancy 3, 11.

NMD is well-studied at the biochemical level, with over 20 proteins known to be involved in this pathway (Figure 1) 1, 2, 3. Three of these proteins — UPF1, UPF2, and UPF3 — are present in all eukaryotes and considered to be core NMD factors [12]. UPF1 is an ATP-dependent RNA helicase recruited to all mRNAs but that specifically remains stably bound to NMD-target mRNAs by an unknown mechanism [13]. UPF1 is thought to initiate the NMD mechanism through forming a complex with a protein kinase dedicated to NMD, SMG1, as well as the translation termination factors eRF1 and eRF3 2, 14. UPF1 also binds to UPF2, which triggers the former to undergo a conformational change that activates its RNA helicase activity [15]. UPF2 also serves as a scaffolding protein by forming a bridge between UPF1 and the third NMD core protein: UPF3. Vertebrates harbor two UPF3 gene paralogs: UPF3A and UPF3B (also called ‘UPF3’ and ‘UPF3X’, respectively). UPF3A is an enigmatic protein whose primary function(s) remain to be determined 16•, 17•, 18•, 19. UPF3B is essential for the decay of a subset of NMD-target mRNAs, and thus, UPF3B is generally considered to be a ‘NMD branch’ factor 17•, 18•, 19, 20. UPF3B interacts with the exon junction complex (EJC), a NMD-promoting protein complex that binds just upstream of exon–exon junctions formed after RNAs are spliced in the nucleus [21]. The EJC consists of the proteins eIF4AIII, Y14 [RBM8A], MAGOH; this complex is transported with spliced mRNAs to the cytoplasm, where accessory proteins are recruited. One of the accessory proteins that is sometimes recruited to this complex is a fourth EJC protein — CASC3 — which was recently shown to form a UPF2-independent protein complex with AKT that functions in an alternative branch of the NMD pathway regulated by AKT-inducing stimuli, such as insulin [22].

The position of the EJC on a spliced RNA is considered to be a major signal that determines whether or not that RNA is degraded by NMD [23]. EJCs are recruited just upstream of most exon–exon junctions after RNA splicing in the nucleus. Evidence suggests that these EJCs are stripped off of RNAs after the entry of RNAs into the cytoplasm by translocating ribosomes, thereby preventing NMD from being activated by the EJC. However, RNAs that contain a stop codon upstream of at least one exon–exon junction (i.e. in a middle exon) would be expected to fail to be completely stripped of EJCs because the translocating ribosomes would terminate translation at the stop codon and thus could not displace the downstream EJC(s). This predicts that such RNAs (i.e. those with the main open reading frame [ORF] stop codon in a middle exon) would be degraded by NMD, and, indeed, this has been demonstrated to be the case for many mRNAs 15, 23. RNAs that do not abide by the downstream exon-exon junction rule are those that instead abide by the ‘50–55 nucleotide boundary rule’, a rule that arose from the finding that most RNAs with a stop codon ≤50–55 nucleotides upstream of the last exon–exon junction escape NMD [23]. This boundary rule likely stems from ‘the footpad’ size of the EJC and its position ∼20 nt upstream of exon–exon junctions. Of note, not all RNAs are degraded by NMD via the EJC. For example, long 3′UTRs and short ORFs upstream of the main ORF can also trigger 2, 24.

This and other data suggest that NMD is not a single linear pathway but instead is comprised of different ‘branches’ dependent on different factors that act on different target RNAs. This has interesting implications [25]. For example, this mechanistic diversity enables NMD to potentially differentially regulate a broad range of physiological processes. It also has potential therapeutic implications, as it means that manipulation of only a single branch of NMD is likely to cause less side effects than manipulation of the entire NMD pathway.

One mechanism by which NMD is initiated is via the phosphorylation of UPF1 by SMG1. Phosphorylated UPF1 binds to the NMD factors, SMG5 and SMG7, which, in turn, recruit deadenylases that lead to mRNA decay through exonucleolytic attack 26, 27. Phosphorylated UPF1 also binds to an endonuclease dedicated to NMD — called SMG6 — that has specificity for NMD-target mRNAs [28]. These two mechanisms are interconnected, as recruitment of SMG5-SMG7 to phosphorylated UPF1 is required to activate SMG6’s endonucleolytic activity [29]. SMG6 can also interact with UPF1 in a phosphorylation-independent manner that is critical for NMD [30].

In this review, we focus on the known and hypothesized roles of NMD in the nervous system (Figure 1). Most of the progress in this regard has been made in mice, so we highlight these studies. When the information is available, we relate NMD-deficient mutant mice phenotypes to the phenotypes of NMD-deficient humans and lower organisms. Because the molecular mechanisms by which NMD acts in neural development have also been studied in some detail in neural stem cells (NSCs) and neural progenitor cells (NPCs), we also review this topic. We note that several excellent reviews on the biological roles of NMD, including in the nervous system, have previously been published 10, 11, 31, 32. In our review, we highlight neurocentric papers published since these reviews but also cover key earlier papers.