Crunching the numbers. Responsibly.
Zinc-finger antiviral protein (ZAP)-mediated RNA decay (ZMD) restricts the replication of viruses containing CpG dinucleotide clusters. However, why ZAP isoforms differ in antiviral activity and how they recruit cofactors to mediate RNA decay is unclear. Therefore, we determined the ordered events of the ZMD pathway. The long ZAP isoform preferentially binds viral RNA and has distinct binding motifs compared to the short isoform. The endoribonuclease KHNYN then cleaves viral RNA at positions of ZAP binding. The 5' cleavage fragment undergoes TUT4/TUT7-mediated 3' uridylation and degradation by DIS3L2. The 3' cleavage fragment is degraded by XRN1. ZAP and TRIM25 interact with KHNYN, TUT7, DIS3L2, and XRN1 in an RNase-resistant manner. Viral infection promotes the interaction between TRIM25 with these enzymes, leading to viral RNA decay while also decreasing the abundance of cellular transcripts. Overall, the long isoform of ZAP recruits key enzymes to assemble an RNA decay complex on viral RNA.
The long ZAP isoform preferentially binds viral RNA and has distinct binding motifs compared to the short isoform.
Triplicate iCLIP in endogenous isoform-edited lines plus motif analysis; motifs derived from single-isoform backgrounds limit attribution.
The endoribonuclease KHNYN then cleaves viral RNA at positions of ZAP binding.
Cleavage termini map inside ZAP peaks and depend on ZAP, but KHNYN dependence of the mapped sites is not directly shown.
The 5′ cleavage fragment undergoes TUT4/TUT7-mediated 3′ uridylation and degradation by DIS3L2. The 3′ cleavage fragment is degraded by XRN1.
Depletion qPCR and tail sequencing converge, but pooled siRNA without rescue cannot isolate each enzyme's contribution.
Viral infection promotes the interaction between TRIM25 with these enzymes, leading to viral RNA decay while also decreasing the abundance of cellular transcripts.
Infection-enhanced interaction and co-regulated gene sets are correlative; the causal linkage implied by "leading to" is untested.
Overall, the long isoform of ZAP recruits key enzymes to assemble an RNA decay complex on viral RNA.
Pan-ZAP, RNase-resistant pulldowns in HeLa cannot establish isoform specificity or assembly on viral RNA.
Derived from the full evaluation — not a separate score.
Strengths
Rather than asserting the endonucleolytic model, the study depletes CNOT1, PARN, EXOSC3/5/7, DDX17 and DHX30 and shows none affects HIV-CpG restriction or fragment abundance, while XRN1 depletion does. This converts a long-standing either/or in the literature into a testable and partly reconciled pathway.
↳ Figures S7B–S7G, 5C; Results §XRN1 degrades the 3′ fragment
Triplicate iCLIP for both ZAP and KHNYN, long-read 5′ and 3′ RACE-seq, strand-specific p1/p2 primer-probe sets, and RNase-treated co-immunoprecipitations converge on the same pathway. Cleavage positions cluster inside independently called ZAP-binding peaks, and non-templated uridine tails appear only when ZAP is present.
↳ Figures 1I, 5A, 5G, 6B, 6D–6E
Isoform comparison uses HEK293T cells CRISPR-engineered at the endogenous ZC3HAV1 locus so that only ZAP-L or only ZAP-S is expressed, directly addressing the criticism the paper levels at prior transient-transfection studies.
↳ Figures 1F–1I; Results §Differential antiviral activity between ZAP isoforms
Figure legends disclose that input and pulldown blots are shared between Figures 3B/5E and 3C/S9I because the experiments were run together, antibodies carry catalogue numbers and RRIDs, and sequencing data and analysis code are deposited under named ArrayExpress accessions and Zenodo DOIs.
↳ Figure 3B and 3C legends; Key Resources Table; Data and code availability
Limitations
The causal attributions to XRN1, TUT4/TUT7 and DIS3L2 rest on pooled siRNA sets or SMARTpools with no rescue and no independently tested single reagent reported in the main text. These are the steps the pathway model depends on most directly.
↳ Figures 5C and 6C; STAR Methods, siRNA transfection and infection
Legends for the RT-qPCR and infectivity bar charts state that error bars represent standard deviation and that one-way ANOVA was applied to log-transformed data, but do not define what the plotted n counts. The independent experimental unit for these central comparisons is therefore not recoverable.
↳ Figure legends 1E, 1G, 2D, 5C, 6C
The closing claim that the long isoform assembles a decay complex on viral RNA outruns the supporting co-immunoprecipitations, which use a pan-ZAP antibody in HeLa cells, survive RNase treatment, and show enhanced TRIM25–enzyme interaction after wild-type HIV infection as well as CpG-recoded infection.
↳ Summary, final sentence; Figures 3A–3C, 5D–5E
Because no antibody efficiently immunoprecipitates endogenous KHNYN, the KHNYN iCLIP is performed in cells reconstituted with a KHNYN-GFP fusion, so the binding map is from a non-endogenous system. The 5′ RACE assay also carries an acknowledged 3′ bias that prevents comparing cleavage frequencies between peaks.
↳ Results §XRN1 degrades the 3′ fragment; Methods, ZAP and KHNYN iCLIP
The decisive evidence is the convergence of independent readouts on one pathway: iCLIP peaks, long-read 5′ RACE cleavage termini and 3′ RACE uridine tails all localise to the same regions of the CpG-recoded genome, and depleting the classical exonucleolytic cofactors changes nothing while depleting XRN1, TUT4/7 or DIS3L2 does. Assessment divided over how much weight the pooled-siRNA depletions in Figures 5C and 6C can carry: without rescue or a separately tested single reagent, each enzyme assignment is a strong inference rather than a secured causal step, though the RACE-seq tail data and the RNase-resistant co-immunoprecipitations substantially offset that gap. The unstated replicate unit in the central RT-qPCR legends compounds this and is priced as a reporting failure rather than a design failure. On presentation, the in-text argument is carefully hedged, but the abstract's closing assertion of isoform-specific complex assembly on viral RNA outruns pan-ZAP pulldowns that are RNase-resistant and respond to wild-type as well as CpG-recoded infection. The result is a genuine mechanistic advance whose pathway ordering is better supported than its isoform attribution.
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Sound3.7
Confidence highThe study resolves a stated uncertainty about the fate of KHNYN cleavage products by assigning XRN1 to the 3' fragment and TUT4/7 uridylation plus DIS3L2 to the 5' fragment, and reconciles the exonucleolytic and endonucleolytic models by testing components of both. The advance is bounded to one engineered model virus in two cell lines, which keeps it below the highest band.
how KHNYN cleavage products are degraded is unclear in this model
Execution is broad and orthogonal — isoform-specific CRISPR lines, triplicate iCLIP, long-read 5′/3′ RACE-seq, RNase-treated co-IPs with a PABPC1 control, and a negative screen of exonucleolytic-model cofactors — but the central downstream enzyme assignments rest on pooled siRNA depletion without rescue, and the RT-qPCR legends state standard deviation without defining the plotted n.
Error bars represent standard deviation.
The argument progresses legibly from isoform comparison through complex assembly to ordered fragment decay, ending in an explicit model figure, and the Limitations section names the single-transcript resolution problem. The abstract's closing framing asserts isoform-specific assembly of a decay complex on viral RNA, which the pan-ZAP, RNase-resistant co-IPs performed in HeLa do not isolate to the long isoform.
the long isoform of ZAP recruits key enzymes to assemble an RNA decay complex on viral RNA
The Introduction and Discussion distinguish the exonucleolytic and endonucleolytic models, acknowledge that high CpG abundance sometimes fails to sensitise viruses, and treat the TRIM25 mechanism and the ZAP-L/ZAP-S controversy as open. The Limitations section is confined to two resolution caveats about iCLIP and RACE-seq, leaving the isoform attribution of the downstream complex unbounded there.
we cannot correlate cleavage sites with specific ZAP-binding sites on an individual transcript basis
Lower confidence on Methodological Rigour, Reporting — domain match limited.
No concerns4 of 4 checks
The reported numbers and statistics are consistent with the stated results.
The authors declare no competing interests, deposit sequencing data in ArrayExpress under named accessions and code at two Zenodo DOIs, and state that cells were mycoplasma-tested. Ethics approval is not applicable to the described established human cell-line work.
Conflict of interest: declared · Preregistration: not stated · Data availability: declared · Code availability: declared
137 references in manuscript 137 of 137 checkable references found in an index
No retraction or expression of concern found in Crossref (includes Retraction Watch).
Sources: Crossref (incl. Retraction Watch) ✓
Where this paper’s evidence sits on the path from initial observation to real-world use.
The work sits at mechanism characterised under controlled conditions: cleavage mapping, fragment-fate assignment and interaction mapping are all performed in HEK293T and HeLa cells with HIV-1. No animal validation, manufacturing or clinical step is reported.
we used HIV-1 as a molecularly tractable model virus
AI-generated, human-governed. Something look off? Contact us to request a review.
Keep reading
Other papers Nabu has evaluated blind, starting with the same field.
Mannes, Zelek, Trouw · 2026 · Health biotechnology
Jayaraman, Prise, Kästele, et al. · 2026 · Basic medicine
Challenging the emerging paradigm, long-lived macrophages were found in the LP and S/M, with distinct transcriptomes and responsiveness to proinflammatory stimuli.
SupportedGoudouris, Pinto-Mariz, Mendonça, et al. · 2021 · Clinical medicine
The outcome was mild in most patients, but the Case Fatality Ratio was higher than in the general population.
Tentative