Nonsense-Mediated Decay
Nonsense-mediated mRNA decay (NMD) is a eukaryotic translation-coupled surveillance and gene-regulatory pathway that degrades transcripts harboring premature termination codons upstream of exon-junction complexes, including alternatively spliced poison exons.
Nonsense-mediated mRNA decay (NMD) and poison exons constitute an integrated eukaryotic RNA quality-control and post-transcriptional homeostatic gene-regulation pathway that couples nuclear pre-mRNA splicing to cytoplasmic translation termination. Discovered by Regis Losson and François Lacroute in Saccharomyces cerevisiae (1979) and by Lynne Maquat and colleagues in human β0-thalassemia erythroblasts (1981), nonsense-mediated decay (NMD) selectively identifies and destroys messenger RNAs (mRNAs) containing a premature termination codon (PTC) (UAA, UAG, or UGA) before they can be translated in bulk into C-terminally truncated polypeptides that could exert dominant-negative or proteotoxic gain-of-function effects.
Beyond its protective role against somatic and germline nonsense or frameshift mutations, between 10% and 20% of normal mammalian physiological transcripts are regulated via Alternative Splicing Coupled to Nonsense-Mediated Decay (AS-NMD), also known as regulated unproductive splicing and translation (RUST). Central to AS-NMD are poison exons—evolutionarily ultraconserved cassette exons (or alternative 5′/3′ splice sites) whose inclusion during alternative splicing introduces an in-frame PTC or shifts the downstream reading frame, targeting the mature mRNA for rapid cytoplasmic destruction by the UPF1–SMG1–SMG6 machinery. Splicing factors of the serine/arginine-rich (SR) and heterogeneous nuclear ribonucleoprotein (hnRNP) families use poison exons as negative-feedback homeostats to buffer their own cellular protein concentrations, while pathogenic mutations that activate cryptic poison exons in haploinsufficient neurodevelopmental genes (SCN1A, SYNGAP1, CHD2) are major causes of genetic epilepsies and targets of therapeutic antisense oligonucleotides (ASOs).
The Exon-Junction Complex (EJC) and the 50–55 Nucleotide Rule
Deposition of the Core EJC Upstream of Splice Junctions
In vertebrate and plant cells, the primary molecular landmark that distinguishes a premature termination codon (PTC) from a physiological normal termination codon (NTC) is the Exon-Junction Complex (EJC). During nuclear transcription and pre-mRNA splicing, the spliceosome deposits a multiprotein EJC core onto the mRNA tightly bound 20 to 24 nucleotides upstream (−20 to −24 nt) of every newly ligated exon–exon junction, independent of nucleotide sequence.
The tetrameric EJC core consists of:
- eIF4A3 (DDX48): A DEAD-box RNA helicase clamped onto the sugar-phosphate backbone of the single-stranded mRNA in an ATP-bound closed conformation.
- MAGOH (or MAGOHB) and Y14 (RBM8A): An obligate heterodimer that binds eIF4A3 and inhibits its intrinsic ATPase activity, locking the EJC onto the mRNA at the −24 nt position throughout nuclear export.
- CASC3 (BTZ / MLN51): A cytoplasmic-enriched scaffold protein that stabilizes the eIF4A3–MAGOH–Y14 clamp and bridges the EJC to the peripheral NMD factor UPF3B (UPF3X).
The Pioneer Round of Translation and the 50–55 Nucleotide Boundary
As the spliced mRNA is exported through the nuclear pore complex into the cytoplasm, its exons are decorated with EJCs spaced 20–24 nt upstream of each exon–exon boundary. During the initial (or subsequent) rounds of translation elongation, the 80S ribosome carries the ribosome-associated protein PYM1 and mechanically displaces every EJC within the coding sequence (CDS) as it translocates along the open reading frame.
In >95% of mammalian protein-coding genes, the physiological normal termination codon (NTC) resides inside the final (last) exon—or within the last 50 to 55 nucleotides of the penultimate exon. Because the eukaryotic 80S ribosome footprint extends roughly 15–18 nucleotides downstream of the decoding A-site and the EJC sits 20–24 nucleotides upstream of the exon–exon junction, a terminating ribosome that arrests at a stop codon located less than 50–55 nucleotides upstream of the final exon–exon junction sterically dislodges the final EJC before termination occurs. Consequently, normal translation termination takes place on an mRNA completely stripped of downstream EJCs, allowing Cytoplasmic Poly(A)-Binding Protein 1 (PABPC1) on the 3′ poly(A) tail to stimulate efficient peptide release by eukaryotic release factors eRF1 and eRF3.
Conversely, if a nonsense mutation, frameshift indel, or poison exon introduces a stop codon located ≥50 to 55 nucleotides upstream of the final exon–exon junction, the 80S ribosome stalls at the PTC before reaching the downstream EJC. At least one intact EJC (eIF4A3–MAGOH–Y14–UPF3B–UPF2) remains bound downstream on the 3′ side of the terminating ribosome, triggering NMD activation (Nagy & Maquat's 50–55 nucleotide rule). A secondary, EJC-independent ("faux 3′ UTR") NMD pathway also operates across eukaryotes (and predominates in intron-poor S. cerevisiae): when a transcript possesses an abnormally long 3′ untranslated region (3′ UTR > 800–1,000 nt), the increased physical distance prevents 3′-bound PABPC1 from interacting with eRF3 at the terminating ribosome, allowing UPF1 accumulating along the long 3′ UTR to trigger decay without requiring a downstream EJC.
| NMD / Splicing Component | Human Gene Symbol(s) | Biochemical Activity & Domain Architecture | Mechanistic Role in NMD or Poison Exon Regulation | Associated Human Genetic Disorder |
|---|---|---|---|---|
| Core EJC Clamp | EIF4A3, RBM8A (Y14), MAGOH | DEAD-box ATPase clamped at −20 to −24 nt upstream of exon junctions | Marks splice history; retains downstream UPF3B/UPF2 if stop codon is ≥50–55 nt upstream | RBM8A (Y14): Thrombocytopenia-absent radius (TAR) syndrome; EIF4A3: Richieri-Costa-Pereira syndrome |
| Central NMD Helicase | UPF1 (RENT1) | Superfamily 1 (SF1) 5′→3′ ATP-dependent RNA helicase + N-terminal CH domain + SQ motifs | Binds eRF1/eRF3 in SURF; phosphorylated by SMG1; unwinds mRNP for decay | Somatically mutated in pancreatic adenosquamous carcinoma |
| EJC–SURF Bridge | UPF2, UPF3B (UPF3X), UPF3A | MIF4G domains (UPF2) and RRM domain (UPF3B) | Bridges downstream EJC to UPF1 CH domain, releasing UPF1 autoinhibition | UPF3B loss-of-function: X-linked intellectual disability & schizophrenia |
| PIKK Kinase Complex | SMG1, SMG8, SMG9 | Phosphatidylinositol 3-kinase-related protein kinase (PIKK) | Phosphorylates [S/T]Q motifs in N- and C-termini of UPF1 upon DECID formation | SMG9 deficiency: heart and brain malformation syndrome |
| NMD Effector Nucleases | SMG6, SMG5, SMG7 | SMG6: C-terminal PIN endonuclease domain; SMG5–SMG7: 14-3-3-like phosphoserine binders | SMG6 cleaves mRNA at the PTC; SMG5–SMG7 recruit CCR4–NOT deadenylase & DCP2 | Depletion causes embryonic lethality via accumulation of AS-NMD isoforms |
| Poison Exon Regulators | SRSF1–SRSF12, TRA2B, PTBP1 | RNA-recognition motif (RRM) + arginine/serine (RS) rich domains | Bind exonic splicing enhancers (ESEs) in ultraconserved poison exons for autoregulation | Pathogenic cryptic poison exon inclusion in SCN1A (Dravet syndrome) |
Molecular Cascade of NMD: SURF Assembly, SMG1 Phosphorylation, and SMG6 Cleavage
When an 80S ribosome arrests at a PTC ≥50–55 nt upstream of an EJC, mRNA degradation proceeds through a four-stage biochemical cascade:
- Formation of the SURF Complex at the Terminating Ribosome: Delayed peptide release at the PTC allows the SURF complex—consisting of the PIKK-family protein kinase SMG1 (bound to its regulatory subunits SMG8 and SMG9), the central Superfamily 1 RNA helicase UPF1, and the eukaryotic release factors eRF1 and eRF3—to assemble at the ribosomal A-site. In the SURF complex, UPF1 is held in an autoinhibited conformation by its N-terminal cysteine-histidine-rich (CH) domain, and SMG1 kinase activity is suppressed by SMG8–SMG9.
- Conversion to the DECID Complex via EJC Bridging: The SURF-bound ribosome interacts with the downstream EJC via UPF2 and UPF3B, forming the Decay-Inducing (DECID) complex. Binding of the C-terminal region of UPF2 to the CH domain of UPF1 induces a large conformational rotation that unlocks both UPF1 5′→3′ ATPase/helicase activity and SMG1 kinase activity.
- Hyperphosphorylation of UPF1 [S/T]Q Motifs: Activated SMG1 phosphorylates multiple serine-glutamine and threonine-glutamine ([S/T]Q) motifs (such as Ser1078, Ser1096, and Ser1116) within the disordered N-terminal and C-terminal tails of UPF1.
- Bifurcated Endonucleolytic and Exonucleolytic Degradation: Hyperphosphorylated UPF1 recruits two parallel decay machineries via their 14-3-3-like phosphopeptide-binding domains:
- SMG6-Mediated Endonucleolytic Cleavage: SMG6 uses its C-terminal PIN (PilT N-terminus) endonuclease domain to cleave the single-stranded mRNA directly in the immediate vicinity of the PTC. This internal cleavage generates an unprotected 5′ fragment (rapidly degraded 3′→5′ by the cytoplasmic RNA exosome and DIS3L) and an uncapped 3′ fragment whose 5′ monophosphate end is digested 5′→3′ by the processive exoribonuclease XRN1 while UPF1 ATPase strips bound proteins.
- SMG5–SMG7-Mediated Deadenylation and Decapping: The SMG5–SMG7 heterodimer binds phosphorylated UPF1 and recruits the CCR4–NOT deadenylase complex (CNOT7/CNOT8) alongside the DCP1–DCP2 decapping holoenzyme (PNRC2), exposing both ends of the transcript to XRN1 and exosome digestion, after which Protein Phosphatase 2A (PP2A) dephosphorylates UPF1 for recycling.
Autoregulatory Poison Exons, Ultraconserved Elements, and Antisense Therapeutics
Homeostatic Negative Feedback in SR and hnRNP Splicing Factors
When the human and mouse genomes were aligned by Gill Bejerano, David Haussler, and colleagues (2004), they discovered 481 Ultraconserved Elements (UCEs)—genomic segments ≥200 bp exhibiting 100% nucleotide identity between human, rat, and mouse (~300 million years of cumulative divergence under strong purifying selection). Strikingly, a large fraction of exonic UCEs do not encode conserved protein catalytic domains; instead, they overlap alternatively spliced poison exons inside genes encoding RNA-binding proteins, including all 12 human serine/arginine-rich (SR) splicing factors (SRSF1 through SRSF12), TRA2B, HNRNPL, PTBP1 (hnRNP I), and SNRPB.
These ultraconserved poison exons operate as post-transcriptional homeostatic thermostats (Lareau et al., 2007; Ni et al., 2007):
- Low Splicing Factor Concentration: When cellular levels of SRSF3 (SRp20) or TRA2B protein are low, the pre-mRNA of SRSF3 (or TRA2B) is spliced along the canonical productive pathway that skips the internal poison exon, generating a full-length translatable mRNA that replenishes SRSF3 protein.
- High Splicing Factor Concentration: When cellular SRSF3 protein accumulates above its physiological set-point, excess SRSF3 binds directly to exonic splicing enhancers (ESEs) inside its own ultraconserved poison exon (exon 4 of SRSF3), promoting inclusion of the poison exon into the mature mRNA. Because the poison exon contains an in-frame premature termination codon located >55 nt upstream of downstream exon–exon junctions, the poison-exon-containing transcript is degraded in the cytoplasm by NMD, down-regulating SRSF3 synthesis. Conversely, splicing repressors such as PTBP1 autoregulate by binding near their own essential coding exon (exon 11 of PTBP1) to force exon skipping, shifting the downstream reading frame to create a frameshift PTC that triggers NMD.
Cryptic Poison Exons in Human Genetic Disease and TANGO Antisense Oligonucleotides
In haploinsufficient human genetic disorders, pathogenic intronic or synonymous single-nucleotide variants can strengthen splice sites or create de novo enhancer motifs that cause aberrant inclusion of a cryptic poison exon, destroying 50% of the functional mRNA via NMD. Furthermore, several severe neurodevelopmental channelopathies and synaptopathies express endogenous developmental poison exons in the human brain:
- Dravet Syndrome (SCN1A Poison Exon 20N): Over 80% of patients with Dravet syndrome (severe myoclonic epilepsy of infancy) carry heterozygous loss-of-function mutations in SCN1A, which encodes the pore-forming α-subunit of the voltage-gated sodium channel NaV1.1 in GABAergic inhibitory interneurons. Even in wild-type alleles, SCN1A pre-mRNA naturally undergoes unproductive splicing of an alternative 64-bp poison exon (Exon 20N) in ~20–30% of transcripts, which introduces a frame-shifting PTC and destroys the transcript via NMD.
- Therapeutic Rescue via TANGO (STK-001 / Zorevunersen): Using Targeted Augmentation of Nuclear Gene Output (TANGO), steric-blocking 2′-O-methoxyethyl (2′-MOE) phosphorothioate antisense oligonucleotides (ASOs, zorevunersen / STK-001) are designed to hybridize to the splice sites of SCN1A poison exon 20N. By sterically blocking spliceosomal recognition of poison exon 20N on the remaining wild-type SCN1A pre-mRNA copy, the ASO redirects unproductive AS-NMD transcripts into productive full-length SCN1A mRNA, restoring physiological NaV1.1 channel density on inhibitory interneurons without viral gene replacement. Identical poison-exon-skipping strategies target SYNGAP1, CDKL5, and PKD1.
References
- Nagy, E., & Maquat, L. E. "A rule for termination-codon position within intron-containing genes: when nonsense affects RNA abundance." Trends in Biochemical Sciences 23, 198–199 (1998). https://doi.org/10.1016/S0968-0004(98)01208-001208-0)
- Lareau, L. F., Inada, M., Green, R. E., Wengrod, J. C., & Brenner, S. E. "Unproductive splicing of SR genes associated with highly conserved and ultraconserved DNA elements." Nature 446, 926–929 (2007). https://doi.org/10.1038/nature05676
- Kurosaki, T., Popp, M. W., & Maquat, L. E. "Quality and quantity control of gene expression by nonsense-mediated mRNA decay." Nature Reviews Molecular Cell Biology 20, 403–420 (2019). https://doi.org/10.1038/s41580-019-0126-2
- Han, Z., et al. "Antisense oligonucleotides increase Scn1a expression and reduce seizures and SUDEP occurrence in a mouse model of Dravet syndrome." Science Translational Medicine 12, eaaz6100 (2020). https://doi.org/10.1126/scitranslmed.aaz6100