The spliceosome is a colossal ribonucleoprotein assembly, built from five small nuclear RNAs and well over a hundred proteins. It assembles anew on every single intron in every pre-messenger RNA molecule, and once it has excised the intron as a looped structure called a lariat, the ligated mRNA is released. But the excised intron-lariat stays wedged inside the RNA-based active site of the machine, forming a stable complex known as the intron-lariat spliceosome. This state is a dead end. To sustain the relentless pace of splicing in a growing cell, the intron-lariat spliceosome must be dismantled so that its proteins and small nuclear RNAs can be reused, and the trapped intron must be ejected, straightened out at its branch point, and ultimately degraded. These combined events constitute spliceosome termination, and until now the molecular details of how they unfold have been largely unknown.

A team led by researchers at the Research Institute of Molecular Pathology in Vienna, working with colleagues in the Netherlands and Spain, set out to catch the spliceosome in the act of terminating. Their strategy combined cryo-electron microscopy of spliceosomes purified directly from human cells with biochemistry, RNA sequencing and genetic experiments. By tagging the proteins DHX35 and YJU2B in human K562 cells and purifying the spliceosome complexes associated with them, the researchers obtained two distinct molecular snapshots. The first, resolved at a remarkable 2.4 angstrom resolution, captured a state they named the ILS-AQR complex. The second, at 3.2 angstroms, revealed an entirely new intermediate that they called the debranched intron spliceosome, or DIS. Together, these structures trace the sequential stages through which the spliceosome passes as it winds down its work.

The ILS-AQR structure revealed several surprises. The team identified a previously uncharacterized termination factor, YJU2B, bound to the spliceosome alongside the known termination factors TFIP11, C19L1, C19L2 and the RNA helicase DHX15. More strikingly, they found that the RNA helicase Aquarius, a protein already known for its role in spliceosome activation, was gripping the circular portion of the intron-lariat RNA in its active site. Aquarius belongs to a family of helicases that move along single-stranded RNA in a defined direction, and the structure showed it poised to pull on the intron-lariat branch point adenosine, the chemical knot that holds the lariat loop together. The researchers propose that Aquarius, guided by the RNA-binding protein PPIE, tugs the branch point out of the spliceosomal RNA network, while DHX15 simultaneously unwinds a different RNA element on the opposite side. Two RNAs pulled in opposite directions, the authors suggest, begin to unravel the intricate RNA architecture that holds the spliceosome together.

Extracting the branch point serves a second critical purpose: it exposes the knot so that it can be cut open. The enzyme responsible is DBR1, a debranching enzyme whose action is essential for intron turnover and for maintaining the balance of RNA in the cell nucleus. The new study provides the first molecular evidence that debranching happens directly on the spliceosome. Through a combination of proximity labelling, protein interaction predictions and biochemical validation, the researchers showed that DBR1 is tethered to the spliceosome through two conserved peptide segments in its tail. One of these peptides binds the termination factor C19L1, anchoring the catalytic domain of DBR1 within reach of the extracted branch point. Experiments in human cells confirmed the biological importance of this connection: DBR1 lacking the peptide could not rescue the growth defect caused by acute depletion of the endogenous enzyme, whereas the full-length protein could.

Once the branch point has been cut and the early termination factors have been displaced, the spliceosome enters the newly discovered DIS state. In this complex, the RNA network is almost completely dismantled. The U2 snRNP, the ESS2 protein and all six ILS-specific termination factors are gone, and the excised intron, now linearized, remains attached only through a short duplex between the U6 small nuclear RNA and the 5-prime splice site. RNA sequencing of the introns associated with each complex confirmed that the vast majority of DIS-bound introns had indeed been debranched, and that nearly all introns found in the ILS-AQR state were also detected in the DIS, indicating that passage through these two states is the predominant termination pathway for human introns.

The final blow is delivered by the RNA helicase DHX35, recruited to the DIS together with its cofactors GPATCH1 and WDR83. The structure shows DHX35 anchored on the spliceosome through extensive contacts with GPATCH1, which itself winds around the PRP8 protein at the heart of the machine. The U6 snRNA regions just downstream of the splice-site pairing are positioned to dock into the DHX35 active site, and the authors propose that DHX35 translocates along U6 snRNA, unwinding the remaining helical turn of the U6-intron duplex and ejecting the debranched intron for decay. In support of this model, purified DIS complexes immobilized on beads released their U5 snRNP and associated protein complexes upon addition of ATP, but only when DHX35 was catalytically competent. A mutant version of DHX35 unable to hydrolyse ATP bound the spliceosome normally yet failed to trigger disassembly, demonstrating that the helicase actively drives the final separation.

The study also illuminates how the cell deals with spliceosomes that go wrong. Splicing is not always successful, and defective complexes assembled on aberrant introns must be recognized and dismantled through a quality-control pathway. By revising an existing structure of a defective spliceosome from fission yeast, the researchers identified two additional factors, the YJU2B orthologue Saf4 and an uncharacterized protein called LENG1. LENG1, a paralogue of the essential branching factor CWC25, appears to partner with YJU2B specifically in defective complexes, where it helps guide the DHX35-GPATCH1-WDR83 machinery to the stalled spliceosome. The same helicases, DHX15 and DHX35, therefore serve both routine termination and quality control, but in regular termination they act in two sequential states, whereas in defective complexes they can be deployed simultaneously for a single-step disassembly.

The significance of this work extends beyond the mechanics of one molecular machine. Errors in splicing are implicated in a wide range of human diseases, including cancers driven by mutations in the splicing factor SF3B1, and the factors identified here, several of which rank among each other’s strongest genetic dependencies in human cell screens, may represent new points of vulnerability. By revealing how three RNA helicases, Aquarius, DHX15 and DHX35, orchestrate the gradual dismantling of the spliceosomal RNA network, coupling intron debranching and ejection to machine disassembly, the study closes a long-standing gap in our understanding of gene expression. Every protein-coding message in the human body depends on this final act of molecular housekeeping, and it has now, at last, been caught on camera.

Subject of Research: Structural mechanism of human spliceosome termination and intron-lariat debranching

Article Title: Mechanism of spliceosome termination

Article References: Boreikaite, V., Faraway, R., Vorländer, M. K., Phillips, A. W., Opitz, L., Wanke, M., Yakoub, G., Raffl, G., Fin, L., González-Prieto, R., Luijsterburg, M. S., Ameres, S. L., & Plaschka, C. (2026). Mechanism of spliceosome termination. Nature. https://doi.org/10.1038/s41586-026-11101-6

Image Credits: AI Generated

DOI: 10.1038/s41586-026-11101-6

Keywords: spliceosome, pre-mRNA splicing, cryo-electron microscopy, intron lariat, DHX35, DHX15, Aquarius, DBR1, RNA helicase, spliceosome recycling, RNA quality control, gene expression

NSP14 is one of sixteen nonstructural proteins encoded by the SARS-CoV-2 genome, and it wears two enzymatic hats that are both essential for viral replication. Its N-terminal domain functions as an exoribonuclease, proofreading freshly synthesized viral RNA and excising mismatched nucleotides, while its C-terminal domain acts as an N7-methyltransferase, adding a methyl group to the 5′ cap of viral messenger RNA so that the RNA masquerades as a cellular transcript. This cap modification allows viral RNA to evade detection by innate immune receptors and to be translated efficiently by the host machinery. Previous research had implicated NSP14 in antagonizing interferon responses, but the precise route it takes through the signaling network remained incompletely mapped, particularly with respect to the RIG-I pathway that dominates the sensing of RNA viruses.

To dissect this question, the research team employed overexpression systems and reporter assays to test what happens to interferon production when NSP14 is present. The results were unambiguous. Elevated levels of NSP14 significantly suppressed both type I and type III interferon responses, along with the expression of downstream interferon-stimulated genes, the effector proteins that execute much of the antiviral program. The suppression extended to key signaling intermediates: activation of TBK1, the kinase that phosphorylates downstream transcription factors, and activation of IRF3, the transcription factor that drives interferon-beta transcription, were both dampened. In effect, NSP14 was acting upstream of the phosphorylation events that normally convert RNA sensing into gene expression, placing its point of interference early in the RIG-I signaling axis.

A central question in such experiments is which part of a viral protein carries out a given function, and the team addressed this by testing whether NSP14’s immunosuppressive effect depended on its catalytic activities or on its interactions with other viral components. NSP14 is known to form a complex with NSP10, a cofactor that stimulates its exoribonuclease activity, and the researchers examined whether this partnership was required. It was not. The inhibitory effect on RIG-I-mediated antiviral signaling persisted independently of complex formation with NSP10. What did matter was the C-terminal N7-methyltransferase activity. When that function was disabled, the ability of NSP14 to suppress interferon production was lost, indicating that the methyltransferase domain is the business end of the immune evasion strategy, even though the mechanism does not appear to operate through cap modification of viral RNA itself.

With the functional requirement mapped to the methyltransferase domain, the investigators turned to the question of which host protein NSP14 might be targeting. Using immunoprecipitation followed by mass spectrometry, they searched for cellular interaction partners of NSP14 and identified DHX15, a DEAH-box helicase that had previously been characterized as a co-receptor of RIG-I. DHX15 belongs to the large family of DExD/C-box helicases, enzymes that remodel RNA structures and, in several cases, contribute to innate immune signaling. The finding that a viral enzyme with methyltransferase activity physically contacts DHX15 suggested a plausible bridge between the protein’s known biochemistry and its observed immunological effect.

The significance of DHX15 as a target becomes clear when its normal role is considered. Earlier studies had established that DHX15 cooperates with RIG-I to enhance signaling triggered by RNA viruses, functioning as a co-receptor that strengthens the antiviral response. In other words, DHX15 is not merely an accessory protein; it is part of the amplification system that allows the RIG-I pathway to generate a robust interferon signal. A virus that could disrupt the DHX15-RIG-I partnership would effectively turn down the volume of the alarm before it ever reaches the nucleus. The research team set out to test whether NSP14 does exactly that, and their experiments confirmed the hypothesis.

Using co-immunoprecipitation assays, the researchers demonstrated that NSP14 dampens the interaction between DHX15 and RIG-I. By binding to DHX15, the viral protein occupies a partner that would otherwise assist RIG-I in propagating the activation signal downstream to MAVS, the mitochondrial antiviral-signaling protein that serves as the platform assembling TBK1 and other signaling components. The consequences of this interference follow a coherent logic: with fewer productive DHX15-RIG-I complexes, reduced activation of TBK1 and IRF3 follows, and the transcription of interferon genes falters. The pathway from viral methyltransferase to blunted cytokine output is thus traced step by step, from protein-protein contact at the level of the sensors to diminished phosphorylation at the level of the kinases to reduced gene expression at the level of the interferon loci.

These findings carry several implications for how scientists understand the immune battlefield inside SARS-CoV-2-infected cells. First, they illustrate the economy of viral genomes, in which a single protein performs multiple jobs: the same NSP14 molecule that proofreads and caps the viral RNA also moonlights as an immune antagonist, using its methyltransferase domain for a second, noncanonical purpose. Second, they highlight DHX15 as a newly appreciated node of viral attack, expanding the list of host factors that coronaviruses manipulate and suggesting that the partnership between helicase co-receptors and RIG-I-like receptors is a vulnerability that pathogens have evolved to exploit. Third, they reinforce the picture of SARS-CoV-2 as a virus with a formidable arsenal of evasion strategies, in which early innate immune suppression buys time for viral replication before adaptive immunity can mount an effective response.

The study also opens avenues for therapeutic thinking. If NSP14’s N7-methyltransferase activity is required for suppressing interferon production, then inhibitors of that enzymatic function might do double duty, impairing viral RNA capping while simultaneously relieving the brake on innate immunity. Small molecules targeting the methyltransferase domain are already of interest in antiviral drug development, and this work adds an immunological rationale to the enzymatic one. Similarly, approaches that stabilize or mimic the DHX15-RIG-I interaction could, in principle, restore signaling amplitude in infected tissues, although such strategies remain far from clinical application. As with all mechanistic cell culture studies, important caveats apply: the experiments relied on overexpression systems, and confirming that the same interference operates during authentic SARS-CoV-2 infection at physiological protein levels will be an important next step.

What the research ultimately delivers is a clearer molecular narrative of how SARS-CoV-2 mutes the body’s first response to infection. The virus does not merely hide its RNA; it actively dismantles the cooperative architecture of the sensing machinery, using NSP14 to pry apart DHX15 and RIG-I and thereby weakening the signal before it can cascade into interferon production. Every element of that cascade, from the helicase partnership at the membrane-associated sensor level to the phosphorylation of TBK1 and IRF3 to the expression of interferon-stimulated genes, is diminished as a result. In mapping this route with biochemical precision, the team at Guizhou Medical University has not only illuminated a specific evasion tactic of a globally significant pathogen but has also underscored a broader lesson of the COVID-19 era: the proteins viruses bring with them are often multitaskers, and understanding their secondary functions may reveal the most consequential battlegrounds of all.

Subject of Research: SARS-CoV-2 NSP14-mediated evasion of RIG-I-mediated innate antiviral immunity through disruption of the DHX15-RIG-I interaction

Article Title: SARS-CoV-2 NSP14 suppresses RIG-I-mediated interferon production by dampening the interaction between DHX15 and RIG-I

Article References: Liu, H., Hu, R., Lei, X., Han, Y., & Nie, Y. (2026). SARS-CoV-2 NSP14 suppresses RIG-I-mediated interferon production by dampening the interaction between DHX15 and RIG-I. Virology Journal. https://doi.org/10.1186/s12985-026-03314-y

Image Credits: AI Generated

DOI: 10.1186/s12985-026-03314-y

Keywords: SARS-CoV-2, NSP14, RIG-I, DHX15, innate immunity, interferon, immune evasion, N7-methyltransferase, TBK1, IRF3, virology, pattern recognition receptors