FL0445: A Better Lipid for a Circular-RNA Delivery

In brief

A new branched ionizable lipid, FL0445, is presented as the vehicle that finally makes circular RNA deliverable in vivo. The delivery data are strong. The circular-RNA framing is not what the data support — and the claim underneath the framing is bigger than the one on the cover.

Kimura and colleagues at Nagoya University report FL0445, a multi-branched ionizable lipid formulated with DOPE into a lipid nanoparticle that outperforms MC3, SM-102 and ALC-0315 across an unusually broad cargo panel: linear mRNA, capped circular RNA at three sizes, an antisense oligonucleotide, plasmid DNA, and a model mRNA vaccine. The in vivo capstone is GLP-1-encoding capped circular RNA in obese ob/ob mice.

Breadth like this is rare. Most ionizable-lipid papers optimise against one cargo and one endpoint. Benchmarking a single formulation against three clinical-standard lipids across four modalities, with cell-level transfection mapping in Cre-reporter mice and a transcriptomic safety readout, is a real contribution regardless of how the individual claims settle.

But the paper's framing and its data pull in slightly different directions. The gap is worth naming precisely, because the claim the data actually support is the more interesting one.

The stated premise isn't supported by the paper's own measurements

The motivating argument is that circular RNA's rigid topology imposes steric constraints on encapsulation, and that lipids optimised for flexible linear mRNA therefore fail on it. FL0445's branched scaffold is proposed to solve this by generating a more flexible internal particle architecture.

The characterisation data don't show the problem. Encapsulation efficiency exceeded 90% for capped circular RNA and matched linear mRNA. Dynamic light scattering found no substantial size difference across lipid composition, payload size, or topology — and that includes SM-102, the benchmark supposedly ill-suited to rigid cargo.

If the incumbent lipid encapsulates circular RNA at over 90% and produces particles of the same size, the bottleneck the introduction invokes is not visible at the level the authors measured it.

What the data do show is that FL0445-LNP delivers everything better or comparably: 10- to 100-fold higher reporter expression in vitro with linear mRNA, more than threefold with circular RNA, superior ASO-mediated factor VII knockdown versus MC3, and plasmid DNA delivery on par with SM-102.

That is a cargo-agnostic advantage, not a topology-specific one. The honest headline is that FL0445 is a better general-purpose ionisable lipid whose advantage carries over to circular RNA — a broader and more useful claim than the one on the cover.

The structural hypothesis underneath remains untested. Cryo-TEM observations are reported qualitatively — circular-RNA formulations "tended to show more bleb-like or phase-separated structures", FL0445 particles appeared more angular — without quantification or blinding. Given that bleb morphology has been repeatedly linked to mRNA accessibility and expression, this is the observation most in need of numbers. The authors flag the absence of SAXS and SANS in their limitations, which is appropriate, but it leaves the central mechanistic premise as an interpretive frame rather than a finding.

Circularity is a small-payload, slow-route strategy

The most durable result here may be the one that cuts against the modality's usual pitch. For the 604-nucleotide NanoLuc payload, linear mRNA beat circular RNA in vitro at every measured timepoint, and after intravenous dosing produced both a higher peak and a higher two-week area under the curve.

Circular RNA only won at 200 nucleotides in vitro, and after subcutaneous administration — where it crossed over linear mRNA at roughly 30 hours and peaked at 72.

This is a genuinely useful design heuristic. The fractional stability gain from closing the loop is largest when the loop is short, and the depot kinetics of local injection give a slow-building, long-tailed expression profile somewhere to accumulate. It points circular RNA toward small secreted peptides delivered subcutaneously — precisely the GLP-1 case — and away from large protein-replacement cargo delivered systemically, which is where much of the field's enthusiasm currently sits.

The authors report this cleanly. The Highlights state it without the two conditions that make it true.

The mechanism is the weakest link, and the authors mostly say so

FL0445-LNP showed lower cellular uptake than SM-102 and ALC-0315 while achieving roughly tenfold higher expression. That dissociation is the paper's most interesting cell-biological observation, and it rests on three pieces of evidence that each carry caveats.

Methyl-β-cyclodextrin inhibition

Near-complete block of FL0445-LNP uptake is read as a distinct cholesterol-dependent entry route. But MβCD depletes plasma-membrane cholesterol globally, altering fluidity and inhibiting clathrin-mediated endocytosis as well; it is not a caveolae-specific tool. It can also extract cholesterol from the particles themselves in the medium. The authors name the single-inhibitor limitation explicitly, which is the right call.

The amiloride and chlorpromazine results

Uptake increased under both. The parsimonious reading — cargo redistributing to the remaining route — is consistent with the authors' model. A less flattering reading is that DiD, a lipophilic membrane dye that reports lipid position rather than RNA position and can transfer to serum lipoproteins, is not a reliable uptake tracer across chemically dissimilar formulations. The paper does not distinguish these.

Lysosomal colocalisation

No significant difference between formulations. This is the one direct measurement of the trafficking step the model depends on, and it does not support differential escape.

With the direct assay null, the "bypasses degradative routes, reaches ER-adjacent compartments" hypothesis rests on bulk liver and spleen RNA-seq showing lower phagocytosis-associated gene expression at three hours — a tissue-level transcriptional correlate standing in for a single-cell trafficking claim. The authors call it a working hypothesis. Readers should hold it as one.

Low inflammation and a stronger vaccine: a tension worth testing

The paper argues that FL0445's low immunostimulation explains its high transfection efficiency, citing the TLR4 → PKR → eIF2α translational-shutdown literature. It then reports that FL0445-LNP produces higher ovalbumin-specific IgG1, more antigen-specific CD8⁺ T cells, and greater antigen-specific killing than the more inflammatory SM-102.

If real, decoupling reactogenicity from immunogenicity is a significant result on its own — arguably more commercially consequential than anything else in the paper. But it sits awkwardly against the mechanistic argument in the preceding sections, where inflammatory activation is cast as the thing suppressing output.

The likely resolution is that antigen dose dominates adjuvanticity here, possibly helped by the enhanced lymph-node accumulation seen after intramuscular dosing. The paper asserts both halves and reconciles neither. A dose-matched antigen-expression comparison would settle it.

GLP-1: restraint worth crediting, comparator worth questioning

The authors repeatedly and correctly frame the ob/ob result as a functional proof of concept rather than efficacy. The glucose AUC did not reach significance (p = 0.056), body weight showed only a non-significant trend, and the positive finding is a single timepoint — 30 minutes post-challenge — among several, uncorrected for multiplicity. In a field where mouse GLP-1 data are routinely oversold, this restraint deserves saying out loud.

Two design points constrain what the experiment can show.

First, the dosing schedule conflates durability with cumulative dosing. Three doses at 0.4 mg/kg every other day, with the glucose tolerance test on day 17. The clean durability data come from a different experiment, with a different payload, after a single dose.

Second, the comparator is linear mRNA, not the standard of care. Against semaglutide, a marginal glucose shift from three RNA doses is not an efficacy argument. The modality's case has to be made on dosing interval and manufacturing rather than effect size — which this design does not yet test. The paper's own reference list indicates the GLP-1 circular-RNA space is already contested; direct comparison will come quickly.

The manufacturing footnote that isn't a footnote

The internal-cap design is elegant. Dispensing with a roughly 600-nucleotide IRES is what makes a 256-nucleotide GLP-1 construct possible at all, and small constructs are exactly where circularisation pays off.

But the synthesis route is a chemically synthesised 54-nucleotide capped oligonucleotide at 39.8% yield, splint-ligated to in vitro transcribed RNA, then PAGE-purified. For the GLP-1 construct: 28.9% ligation, 8.6% isolated yield. Those numbers compound, and denaturing PAGE does not scale.

Against IRES-based permuted-intron-exon circularisation from a single transcription reaction, this is a materially heavier process. For a modality whose pitch includes manufacturing simplicity, that belongs in the discussion rather than the supplement.

Relatedly, the low-cytokine claims for circular-RNA formulations would be strengthened by explicit quantification of residual nicked linear species and double-stranded RNA in the circular preparations — the standard confounder for circular RNA immunogenicity, and one the paper addresses for linear mRNA but not visibly for circular.

Six claims most exposed to challenge

Each pairs the load-bearing assertion with what the paper measured.

Claim 01

Rigid circular-RNA topology creates an encapsulation problem that FL0445 solves.

Contradicted by the paper's own greater-than-90% encapsulation efficiency and matched particle sizes across all lipids tested, including SM-102.

Claim 02

Branching creates internal flexibility that accommodates rigid cargo.

Asserted, with only qualitative cryo-TEM support. No SAXS or SANS, no quantification, no blinding.

Claim 03

A distinct cholesterol-dependent entry route explains the potency gain.

Single-inhibitor evidence, a null result on the one direct trafficking assay, and indirect transcriptional corroboration.

Claim 04

Capped circular RNA gives more durable expression than linear mRNA.

True only at small payload size and subcutaneous route. False for intravenous dosing at 604 nucleotides, by the paper's own two-week AUC.

Claim 05

The in vivo comparison is like-for-like.

FL0445 received composition-ratio optimisation in vivo; the benchmarks did not. Credit where due — the phospholipid comparison was symmetric, with all five ionisable lipids tested against both DOPE and DSPC.

Claim 06

The safety profile supports chronic dosing.

The tolerability study is single-dose. Anti-PEG responses, accelerated blood clearance and cumulative hepatic effects are untested — and repeat administration is the paper's own stated motivation.

What would settle it

• Genetic validation of the entry pathway — caveolin or flotillin knockouts, or an arrayed CRISPR screen of the kind already applied to MC3-LNPs — instead of pharmacological inhibition.
• Direct cytosolic-release quantification via galectin-8/9 recruitment or split-luciferase complementation, rather than lysosomal colocalisation area ratios.
• SAXS or SANS on matched linear- versus circular-payload particles, to test the flexibility hypothesis directly.
• A repeat-dose study over months, in the chronic-disease setting the paper invokes.
• A head-to-head against IRES-based circular RNA at matched protein output, with yields reported.

One practical caveat for replication: FL0445 and FL2266 are FUJIFILM materials covered by an existing patent application. Independent benchmarking will be gated by material transfer, which matters for a lipid being positioned as a benchmark-beating platform.

Bottom line

None of this diminishes what the paper does well. The cargo breadth is genuine, the payload-size dependence of circularisation is a finding the field can use tomorrow, and the authors' framing of the GLP-1 data as proof of concept rather than efficacy is the kind of restraint that should be more common. The overreach is in the packaging, not the bench work — and the claim underneath the packaging is the stronger one.

Frequently asked questions

What is a branched ionisable lipid?

An ionisable lipid is the component of a lipid nanoparticle that carries positive charge at acidic pH — letting it bind RNA during formulation and destabilise the endosomal membrane after uptake. "Branched" refers to the hydrophobic tail architecture: instead of straight chains, the tails split into multiple arms, which disrupts tight lipid packing and is thought to favour the non-bilayer phases associated with endosomal escape.

Why is circular RNA harder to deliver than linear mRNA?

The usual argument is that a covalently closed loop is topologically rigid and packs differently inside a nanoparticle. This paper's own data complicate that: encapsulation efficiency and particle size were essentially identical for linear and circular payloads across every lipid tested. Whatever advantage the new lipid confers appears to happen after encapsulation, not during it.

Does circular RNA always outlast linear mRNA?

No — and this paper is a useful corrective. Circular RNA won only at small payload size and after subcutaneous injection. At 604 nucleotides given intravenously, linear mRNA produced both a higher peak and a higher two-week cumulative expression.

Is this a viable route to a long-acting GLP-1 drug?

Not on this evidence. The glucose result is a single significant timepoint with a non-significant AUC (p = 0.056), no significant weight change, and a dosing schedule that can't separate durability from repeat administration. The authors themselves label it a proof of concept. The real test is a dose-response against an approved peptide agonist, on dosing interval rather than peak effect.

Source

Kimura S., Tsutsumi S., Nakamura N., et al. "A branched ionizable lipid nanoparticle platform for versatile in vivo delivery of nucleic acids: Validation from mRNA to capped circular RNA." Cell Biomaterials, 100555 (2026). Published online 19 August 2026.

What the next decade of small molecule–RNA research has to solve before it delivers medicines

After the Tertiary Structure

After the Tertiary Structure

What the next decade of small molecule–RNA research has to solve before it delivers medicines

---

For twenty years, the case for targeting RNA with small molecules has been made mostly in the subjunctive. Only about 2% of the human genome encodes protein, the argument goes, so the transcriptome ought to be an enormous reservoir of untapped targets — a way to reach the oncogenic transcription factors, aggregation-prone proteins, and structural scaffolds that medicinal chemistry has failed at for decades. The pitch has always been compelling. The delivery has been thin.

There is still, in 2026, exactly one unambiguous success: risdiplam, Roche and PTC's SMN2 splicing modifier for spinal muscular atrophy — orally bioavailable, brain-penetrant, and disease-modifying. Everything else is either preclinical, in early trials, or instructive failure.

That gap between promise and product is the interesting thing about this moment, because the reasons for it are finally becoming legible. Three bottlenecks have held the field back: we could not see RNA structures well enough to design against them, we did not have enough interaction data to learn from, and we could not tell whether a molecule that bound one RNA was leaving the rest of the transcriptome alone. Each is now being attacked directly, and each is yielding at a different rate. Understanding which are actually loosening — and which are not — is the best available guide to where this field goes next.

---

1. The structure bottleneck is being routed around, not solved

The orthodoxy has been that drugging RNA requires the same thing that drugging proteins requires: a three-dimensional structure with a definable pocket. Docking programs adapted for nucleic acids — AutoDock Vina, rDock, RLDOCK, NLDock — all assume one. So do the recent deep learning entrants like RNAmigos2 and RLaffinity.

The problem is that RNA tertiary structure prediction has not undergone its AlphaFold moment, and the evidence on this is now unusually clear. The CASP16 assessment of nucleic acid structure prediction, published in 2025, found that accuracy still depends almost entirely on whether a closely related experimental structure already exists; without a template, performance collapses. More strikingly, the assessors concluded that despite the surge of interest and the arrival of deep learning methods including AlphaFold3, there had been no notable improvement in nucleic acid modelling accuracy relative to earlier blind challenges. Pseudoknots, non-canonical pairs, and tertiary motifs like A-minor interactions were still not reliably recovered. CASP16 also included, for the first time, blind prediction of RNA–small molecule complexes. Results were generally poor absent templates.

This matters enormously, because the RNAs people actually want to drug — a repeat expansion in a 3′ UTR, an IRES in an oncogene's 5′ leader, a stem loop in a viral genome — are exactly the ones with no solved structure and no close homolog.

Hence the significance of a shift now visible in the literature: skip the tertiary structure entirely. The clearest recent example is SMRTnet, published in Nature Biotechnology in January 2026 by Fei, Wang, Zhang and colleagues at Tsinghua and Peking University. It takes only an RNA sequence, its secondary structure in dot-bracket notation, and a compound SMILES string, fuses a pretrained RNA language model with a chemical language model through an attention-based fusion module, and predicts a binding score. No pocket geometry required.

The results are genuinely encouraging and worth stating precisely. Roughly 0.84 auROC on held-out data from its own PDB-derived training set; 0.72 on an independent benchmark of published interactions; and, in prospective wet-lab screening against ten disease-associated targets, 40 validated binders from 190 predictions — a 21% hit rate, with binding scores that tracked validation rates across the full prediction range. For a class of targets that conventional docking barely handles, that is a real result.

But the ablation buried in the same paper is the more important finding. Substituting computationally predicted secondary structures for experimentally derived ones dropped benchmark performance from 0.72 to 0.66 — most of the way back toward noise. The method does not eliminate the structure requirement; it relocates it from the tertiary to the secondary level, where we happen to have good experimental tools. icSHAPE, SHAPE-MaP, DMS-MaPseq and their descendants can now profile RNA structure transcriptome-wide, in cells, under physiological conditions.

The strategic implication is that the field's structural investment should shift accordingly. The rate-limiting resource for the next five years is not more RNA crystallography — it is chemical probing data at scale, across cell types, disease states, and conditions. Secondary structure is what we can measure well and what current models can actually use. Tertiary structure prediction will improve eventually, but planning a discovery program around its imminent arrival is, on the CASP16 evidence, unwise.

---

2. The data famine is the real rate limiter

Here is the number that should govern expectations for machine learning in this field. PDBBind, the workhorse resource for protein–ligand affinity prediction, contains tens of thousands of measured protein–ligand interactions. The equivalent figure for RNA is on the order of one hundred. The PDB holds hundreds of thousands of protein structures and roughly a couple of thousand RNA structures.

SMRTnet's training set illustrates the squeeze: 1,061 usable structures from the entire PDB, expanded to 8,672 fragment–ligand pairs by slicing binding regions into 31-nucleotide windows. That is a clever use of what exists, but it is fundamentally an augmentation strategy applied to a small, biased sample — heavy on riboswitches, aptamers, and ribosomal RNA, because those are what crystallizes.

This is why the pretrained-language-model architecture has become near-universal in recent RNA-ligand work (SMRTnet, GerNA-Bind, RLsite, and others all share the pattern). The models are not primarily learning interaction physics from interaction data — there isn't enough. They are importing general representations of RNA sequence space and chemical space learned from hundreds of millions of unlabeled sequences, then fine-tuning on a thin layer of labels. It works better than nothing. It has a ceiling.

Breaking through requires generating interaction data on a scale the field has never attempted. The most promising signals here are industrial rather than academic. xFOREST Therapeutics has described library-versus-library screening across multiplexed panels of more than a thousand RNA structures, measuring selectivity in parallel rather than as an afterthought. Arrakis has built a platform explicitly organized around the sequence–structure–function relationship rather than around one-off campaigns. These are the datasets that will matter, and their concentration in private hands is a genuine structural problem for the field.

If I were placing one bet on what unlocks this area, it would be a precompetitive consortium producing an open, multiplexed RNA–small molecule interaction resource at PDBBind scale. Nothing else on the horizon changes the ceiling as much. Every modelling advance is currently downstream of a hundred-fold data deficit.

---

3. Selectivity is the unsolved problem, and it is the one that kills drugs

Branaplam is the case everyone in this field should be able to recite. It was developed as an SMN2 splicing modulator for SMA; an off-target effect — inducing a frameshifting pseudo-exon in the huntingtin transcript — looked like a gift, and it was redirected to Huntington's disease. The phase 2 VIBRANT-HD trial, published in Nature Medicine in early 2026, reported that branaplam did what it was supposed to do: it became the first splicing modulator to lower mutant huntingtin in patient cerebrospinal fluid. It also caused peripheral neuropathy in 18 of the 21 participants in the initial dose cohort. The trial was terminated early. The neurofilament light chain elevations reversed on discontinuation, but the drug did not survive.

The mechanism, as worked out in patient-derived neurons, appears to run through nucleolar stress and p53 activation — not through huntingtin lowering. In other words, the therapeutic effect and the toxicity came from different RNAs, and the therapeutic window between them was too narrow.

This is the field's central unsolved problem, and current computational tools barely address it. Essentially every RNA-ligand model, SMRTnet included, is trained and evaluated as a binary, pairwise classifier: given this RNA and this compound, do they interact? But the question that determines whether a molecule becomes a drug is categorically different: given this compound, what does it engage across an entire cellular transcriptome of ~200,000 structured elements, in what proportions, in which tissues?

A model with 0.84 auROC on pairwise classification tells you almost nothing about that. At transcriptome scale, even excellent per-pair specificity produces thousands of off-target engagements. And RNA makes this worse than proteins do, for reasons intrinsic to the molecule: secondary structure motifs are highly recurrent (a 2×2 internal loop flanked by GC pairs appears in innumerable transcripts), RNA is conformationally dynamic, and the same sequence can adopt different structures in different cellular contexts.

The technical responses are visible but immature. Transcriptome-wide small-molecule binding site mapping via covalent crosslinking (Chem-CLIP and successors) gives empirical occupancy maps. Multiplexed selectivity screening builds counter-selection into the primary assay rather than bolting it on. What is conspicuously missing is a computational framework that predicts selectivity profiles rather than binary interactions — and that will not exist until the training data exists. Selectivity is a negative-data problem, and negative data is precisely what curated interaction databases lack.

Expect the next generation of credible RNA-ligand models to be judged on transcriptome-scale precision, not pairwise auROC. Expect most current tools to look worse under that standard.

---

4. Binding is not function, and the field keeps conflating them

The SMRTnet authors deserve credit for saying this plainly in their own discussion: binding alone does not imply that a molecule will regulate RNA expression or produce biological activity.

Their own data illustrate the difficulty. Of 40 validated binders across ten targets, most had micromolar dissociation constants — respectable for a screening hit, distant from a drug. And the one compound carried forward to cellular work is instructive in an unintended way. Irinotecan hydrochloride trihydrate reduced MYC mRNA and protein in HeLa cells, cut proliferation, and induced apoptosis across three cancer lines. It is also a well-established topoisomerase I inhibitor and an approved chemotherapeutic. When a known cytotoxic reduces proliferation in cancer cells, target engagement at the MYC IRES is not the parsimonious explanation. The authors' luciferase reporter experiment — where the compound suppressed the wild-type IRES construct but not a fully base-paired control — is the right control and provides real evidence for structure-specific activity. But the cell-viability data cannot carry the weight often placed on it.

This is not a criticism peculiar to one paper; it is endemic. Screening natural product and metabolite libraries returns compounds that are pharmacologically loaded by construction. Attributing phenotype to a newly discovered RNA interaction requires ruling out the activity the molecule was already known for, and the field's standard evidence package rarely does.

The methodological fix is not mysterious: structure-matched RNA controls (as used above), rescue with binding-site mutants, transcriptome-wide occupancy mapping, and dose-response concordance between binding affinity and cellular effect. Programs that adopt these as standard will be distinguishable from those that don't, and the distinction will show up in the clinic.

---

5. Where the value actually accrues: modality, not just binding

The most underrated development in this field is that the binder itself is increasingly not the drug. It is a recognition element.

Ribonuclease-targeting chimeras (RiboTACs) couple a weak, non-functional RNA binder to a moiety that recruits RNase L, converting occupancy into degradation. This solves the field's most awkward problem — that micromolar binding to a structured loop usually does nothing — by making the binding event catalytic rather than occupancy-driven. It is the same logic that transformed protein degradation with PROTACs, and it dramatically relaxes the affinity requirement. Amgen's 2025 deal with Arrakis for RNA-directed degraders, reportedly $75 million upfront, signals that large pharma has priced this in.

The related insight is that RNA-targeting small molecules compete with oligonucleotide therapeutics, not with protein-targeting small molecules. ASOs and siRNAs already hit RNA with exquisite, programmable specificity. What they do not offer is oral dosing, broad biodistribution, blood-brain barrier penetration, and conventional supply chains. That is the entire value proposition — and it means the field's north star should be selective, orally bioavailable, CNS-penetrant molecules, not merely potent ones. Arrakis's DM1 program, aimed at CUG repeat expansions with an oral compound and an IND targeted for 2026, is the clearest test case. Remix's REM-422, a splicing modulator reducing MYB in adenoid cystic carcinoma and AML, tests whether the approach reaches genuinely undruggable transcription factors.

---

What to watch

Five markers will tell you whether this field is compounding or stalling:

1. A second approval. Risdiplam remains a sample size of one. A splicing modulator or repeat-expansion binder clearing phase 2 with a clean neurological safety profile would change the field's risk calculus more than any methodological advance.
2. Open interaction data at scale. Watch for a multiplexed RNA–ligand dataset released precompetitively. Its absence caps every model in the field.
3. Selectivity as a primary endpoint. Papers reporting transcriptome-wide occupancy alongside target binding — rather than pairwise auROC alone — mark the transition from method development to drug discovery.
4. Chemical probing displacing crystallography as the structural foundation, with prediction models trained on in-cell secondary structure rather than PDB-derived idealizations.
5. RiboTACs reaching the clinic. If catalytic degradation works in humans, the affinity bar drops by orders of magnitude and the addressable target space expands accordingly.

The honest summary is that computational methods have recently gotten good enough to generate credible hypotheses at scale — that is what SMRTnet and its contemporaries demonstrate, and it is not nothing. What they have not done is solve the two problems that actually determine whether RNA-targeting small molecules become a therapeutic class: getting from micromolar binder to selective drug, and proving that engagement causes the biology. Those are experimental problems, they are expensive, and no amount of architectural cleverness substitutes for the data.

The field's next decade will be decided in assay development, not model design.

---

Sources

• Fei, Y. et al. "Predicting small molecule–RNA interactions without RNA tertiary structures." Nature Biotechnology (2026). https://doi.org/10.1038/s41587-025-02942-z
• "Assessment of nucleic acid structure prediction in CASP16." Proteins / PMC12248019. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12248019/
• "Oral splicing modulator branaplam in Huntington's disease: a phase 2 randomized controlled trial." Nature Medicine (2026). https://www.nature.com/articles/s41591-025-04117-4
• Krach, F. et al. "RNA splicing modulator for Huntington's disease treatment induces peripheral neuropathy." iScience (2025). https://pmc.ncbi.nlm.nih.gov/articles/PMC12059699/
• Carvajal-Patiño, J.G. et al. "RNAmigos2: accelerated structure-based RNA virtual screening with deep graph learning." Nature Communications 16, 2799 (2025).
• Childs-Disney, J.L. et al. "Targeting RNA structures with small molecules." Nature Reviews Drug Discovery 21, 736–762 (2022).
• Arrakis Therapeutics DM1 program updates. https://arrakistx.com/pipeline/
• "RNA-targeting small molecules: a new frontier of drug discovery." Pharmaceutical Technology (2026). https://www.pharmaceutical-technology.com/features/rna-targeting-small-molecules-a-new-frontier-of-drug-discovery/

RNA’s Next Act: From Biological Messenger to Programmable Nanomachine

 

https://thernablog.blogspot.com/

Researchers around the world are learning to make RNA sense, compute, assemble, edit and organize living cells. The convergence of RNA biology and nanotechnology could reshape medicine, agriculture and synthetic biology.

For decades, RNA occupied an awkward middle ground in biology. DNA stored genetic information; proteins performed most of the cell’s chemistry; RNA carried instructions between them.

That hierarchy has steadily collapsed.

RNA is now understood as an extraordinarily versatile molecule. It can catalyse reactions, recognize metabolites, regulate genes, form intricate three-dimensional structures and reorganize itself in response to its surroundings. Some RNAs act as switches. Others serve as scaffolds, molecular guides or components of cellular machines.

These properties are drawing RNA biology into an unexpected partnership with nanotechnology.

A recent Nature feature described this transformation particularly well: RNA's ability to fold, switch and reorganize is increasingly being exploited to build nanoscale biological technologies. Researchers are no longer interested only in discovering what an RNA molecule naturally does. They are beginning to ask what RNA can be engineered to do [1].

The distinction is important. It marks a transition from RNA biology as predominantly a science of discovery towards RNA biology as an engineering discipline.

And that transition is occurring globally.

Across laboratories in the United States, Europe, China, South Korea, Australia and elsewhere, researchers are developing RNA circuits, nanostructures, editing platforms, synthetic condensates, delivery vehicles and agricultural technologies. Individually, these advances belong to different specialties. Collectively, they suggest that RNA could become one of the principal programmable materials of twenty-first-century biology.

A molecule that can carry information and become machinery

RNA possesses an unusual combination of properties.

Its sequence stores information, much like DNA. But unlike the familiar textbook depiction of messenger RNA as a simple linear strand, RNA readily folds back upon itself. Complementary regions form stems, loops, bulges, junctions and elaborate tertiary structures.

Those structures matter because shape determines function.

An RNA molecule can expose or conceal a regulatory sequence. It can recognize another RNA, recruit a protein, bind a small molecule or switch conformation after encountering a particular chemical signal.

For nanotechnologists, this combination of information, structure and dynamics is particularly attractive.

DNA nanotechnology established that nucleic acids can be programmed to self-assemble into intricate structures. RNA potentially goes further because it is naturally produced within cells and participates directly in cellular regulation.

An RNA nanostructure therefore need not remain a passive molecular sculpture.

It could become machinery.

Turning RNA into a cellular switch

One of the clearest demonstrations of programmable RNA comes from riboswitches.

Natural riboswitches alter gene expression after binding specific metabolites. Their structures change in response to a chemical cue, affecting whether downstream genetic information is expressed.

Synthetic biologists are now rewriting this principle.

At the University of Konstanz in Germany, Jörg Hartig and colleagues developed engineered riboswitches based on bacterial xanthine aptamers that respond to oxypurinol, the active metabolite of the clinically used drug allopurinol. The work demonstrated strong chemically controlled regulation of gene expression in mammalian cells [2].

The long-term attraction is control.

Gene therapies generally aim to introduce or restore biological functions, but regulating therapeutic output after treatment can be difficult. An RNA switch introduces another regulatory layer: a therapeutic construct could, in principle, be activated or modulated pharmacologically.

That would make gene therapy less like installing a permanently active programme and more like installing a tunable biological system.

Other researchers are moving beyond individual switches.

At Pohang University of Science and Technology in South Korea, Jongmin Kim and colleagues have developed programmable RNA-based systems capable of processing multiple molecular inputs to regulate endogenous gene expression [3].

The analogy with electronic logic gates is useful, although biology is considerably messier than silicon.

A conventional engineered gene might respond to one trigger. A more sophisticated RNA circuit could require several conditions to be satisfied before generating an output.

A therapeutic cell might eventually detect multiple disease-associated signals and activate a treatment only when the appropriate combination is present.

In that sense, RNA begins to resemble a molecular decision-making system.

Building structures that cells manufacture themselves

An even more striking development is RNA origami.

The approach builds on a fundamental idea from nucleic-acid nanotechnology: predictable base-pairing interactions can be used to make nucleic-acid strands fold into designed geometries.

RNA introduces an additional possibility. Because cells naturally transcribe RNA, the information needed to construct a nanostructure can itself be genetically encoded.

Rather than manufacture a nanoscale object outside a cell and then attempt to deliver the completed structure, researchers could potentially provide the instructions and allow the cell to build it.

Work led by Fei Zhang and colleagues demonstrated this principle by designing RNA molecules that co-transcriptionally self-assemble within human-cell nuclei into rings, zigzag scaffolds, lattices and mesh-like architectures [4].

The achievement matters not simply because complex shapes can be produced inside cells.

Those shapes could ultimately become functional.

An RNA scaffold might recruit selected proteins.

Another could organize enzymes.

A structure positioned near chromatin might alter regulatory interactions.

Still another could provide the architecture for an intracellular biosensor.

The cell would no longer merely express an RNA sequence. It would manufacture a designed nanoscale object.

That possibility begins to erase the boundary between synthetic biology and nanofabrication.

The rise of artificial RNA organelles

Cells are spatially organized systems. Many biochemical reactions succeed because the correct molecules are concentrated in the correct location.

Yet not every cellular compartment has a membrane.

Biomolecular condensates can form through networks of interactions among proteins and nucleic acids, producing dense, dynamic compartments that remain physically distinct from their surroundings.

Researchers are now attempting to recreate this principle synthetically.

Giacomo Fabrini, Lorenzo Di Michele, Elisa Franco, Paul Rothemund and collaborators demonstrated the co-transcriptional production of programmable RNA condensates and synthetic organelle-like structures [5].

In 2026, Shiyi Li, Yuna Kim and colleagues extended this strategy to programmable artificial RNA condensates within mammalian cells [6].

Related work has now demonstrated nano-engineered RNA organelle-like assemblies in bacteria as well [7].

The implications extend beyond constructing unusual intracellular shapes.

Imagine creating an artificial compartment that concentrates several enzymes participating in one biochemical pathway. Instead of those enzymes diffusing independently throughout the cytoplasm, an RNA architecture could bring them together.

Another compartment might recruit particular RNA-binding proteins. A third could sequester molecules whose accumulation disrupts normal cellular function.

The ambition is therefore larger than regulating individual genes.

Synthetic RNA might eventually allow researchers to redesign parts of the physical geography of the cell.

When the nanostructure becomes the medicine

RNA nanotechnology is also blurring a traditional distinction in drug delivery: the boundary between therapeutic cargo and carrier.

Conventionally, a nanoparticle transports an active molecule.

But RNA structures can themselves be biologically active.

Researchers led by Hao Yan and Yung Chang at Arizona State University demonstrated that RNA-origami nanostructures can activate antitumour immunity. In mouse models, the engineered RNA structures functioned as potent immunostimulatory materials and produced antitumour responses through mechanisms involving innate and adaptive immunity [8].

This illustrates an important principle.

For an engineered RNA therapeutic, biological behaviour might depend not only on nucleotide sequence but also on shape, size, stability, molecular interactions and subcellular destination.

Future RNA medicines could therefore be designed at several levels simultaneously:

sequence, structure, delivery, immune recognition and biological function.

That is a substantially richer engineering problem than simply asking what protein an mRNA encodes.

Delivery remains the decisive bottleneck

But even the most sophisticated RNA architecture is useless if it never reaches the appropriate biological destination.

Delivery remains one of the central constraints on RNA biotechnology.

Lipid nanoparticles demonstrated dramatically that RNA can be protected and delivered effectively. The broader challenge, however, is much harder than merely encapsulating RNA.

Different applications require RNA to reach different organs, tissues and cell populations. Once inside a cell, some RNA molecules need to remain in the cytoplasm, whereas others may need access to particular intracellular environments.

Nanoparticle engineers therefore confront a cascade of barriers: extracellular stability, biodistribution, tissue penetration, cellular uptake, endosomal escape, intracellular release and eventual degradation.

A 2026 Nature Materials review describes this shift towards precision mRNA delivery, in which administration route, particle chemistry, targeting strategies and responsive release are engineered according to the biological destination rather than assuming that one formulation can serve every application [9].

This may prove to be one of nanotechnology's largest contributions to RNA biology.

The future is unlikely to belong to one universally optimal nanoparticle.

It is more likely to involve families of delivery architectures optimized for particular tissues, cell types and therapeutic tasks.

An exquisitely engineered RNA system delivered to the wrong cells is still a failed therapy.

From reading RNA to rewriting it

At the same time, RNA itself is becoming increasingly editable.

RNA editing is particularly attractive because it provides a way of changing biological information without permanently changing genomic DNA.

One major strategy exploits adenosine deaminases acting on RNA, or ADARs, which naturally convert adenosine to inosine in double-stranded RNA contexts.

Researchers can redirect endogenous ADAR activity using engineered guide RNAs.

Yuanfan Sun and colleagues recently showed that guide RNAs designed to mimic structural features of highly edited endogenous ADAR substrates can improve RNA base editing, illustrating how RNA structure itself can be engineered to recruit cellular editing machinery more effectively [10].

Researchers are also expanding the chemistry available for programmable RNA editing.

Yuan Zhuang, Qingguo Zhu, Chengqi Yi and colleagues developed AIM, a single-strand deaminase-assisted platform capable of A-to-I, C-to-U or simultaneous A+C editing within user-defined RNA regions [11].

Intriguingly, the engineering is also moving in the opposite molecular direction.

Hyeon Woo Im, Sangsu Bae and colleagues recently repurposed engineered ADAR domains for highly precise A-to-G DNA base editing within DNA–RNA hybrids. The work is not an RNA-editing technology itself, but it demonstrates how enzymes originating from RNA biology can be redesigned for entirely new information-processing roles [12].

The broader trend is unmistakable.

RNA is no longer simply being read as the output of gene expression.

It is becoming an editable information layer between genotype and phenotype.

Artificial intelligence enters RNA biology

The enormous RNA design space makes computation increasingly important.

Predicting RNA behaviour is difficult. RNA molecules are flexible, individual sequences can populate multiple conformations, and their structures can be influenced by ions, proteins, other nucleic acids and the local cellular environment.

Machine learning is beginning to address parts of this problem.

In 2026, Philip Fradkin, Bo Wang and colleagues reported Orthrus, a foundation model trained to learn evolutionary and functional representations of mature RNA molecules. The system outperformed several genomic foundation models on RNA-property prediction tasks and could distinguish functional differences among transcript isoforms [13].

At the structural level, Sumit Tarafder and Debswapna Bhattacharya introduced RNAbpFlow, a generative approach that incorporates base-pair information into three-dimensional RNA structure generation [14].

The eventual significance of such systems may extend beyond prediction.

The more transformative objective is inverse design.

Instead of asking:

What does this RNA sequence do?

a researcher might ask:

What RNA sequence should I build to obtain this function?

Design an RNA that recognizes this metabolite.

Design a scaffold that recruits these proteins.

Design an untranslated region that produces a particular expression profile.

Design a nanostructure that assembles only under defined intracellular conditions.

AI would generate candidate molecules; experiments would determine which ones actually work.

Machine learning would not eliminate experimental RNA biology.

It could profoundly change where experimentation begins.

Agriculture becomes another RNA-engineering frontier

Many of the same principles are appearing outside medicine.

Agriculture is becoming one of the most promising—and technically difficult—arenas for RNA nanotechnology.

Double-stranded RNA can initiate RNA interference and selectively suppress genes in plants, viruses, fungi and insect pests. That sequence specificity creates opportunities for crop-protection strategies fundamentally different from conventional broad-spectrum pesticides.

But the agricultural environment is unforgiving.

RNA applied to a leaf might encounter ultraviolet radiation, rain, nucleases, waxy barriers and cell walls before reaching the cellular compartment where gene silencing must occur.

Researchers have consequently investigated layered double hydroxides, carbon-based materials, mesoporous silica, chitosan formulations, lipid carriers and other nanomaterials as potential RNA-delivery platforms.

Yet a 2026 analysis in Nature Plants highlights a crucial distinction: stabilizing dsRNA on a leaf or increasing its accumulation in the apoplast does not demonstrate effective delivery to the plant cytoplasm. For antiviral applications in particular, reliable symplastic delivery remains incompletely understood [15].

This is precisely where RNA biology and nanotechnology must converge.

Sequence optimization cannot, by itself, solve a transport problem.

An effective carrier has to be designed around the biology of the organism, tissue and intracellular destination.

And another challenge is already appearing downstream.

RNA biopesticides must eventually move through regulatory systems developed largely for older classes of crop-protection chemicals. Researchers led by Sandya Gunasekara and Neena Mitter have argued that international regulatory harmonization will be important if dsRNA-based pesticides are to move efficiently towards widespread deployment [16].

The agricultural RNA revolution therefore depends simultaneously on molecular biology, nanomaterials, ecology, formulation engineering, field performance and regulation.

The convergence matters more than any single technology

RNA switches, RNA origami, artificial condensates, editing systems, machine-learning models and nanocarriers can appear to belong to separate scientific stories.

They probably do not.

Their convergence is the more important development.

Imagine a future therapeutic system.

A targeted nanoparticle first delivers an engineered RNA construct to a specific population of cells.

Inside those cells, the RNA folds into a predetermined architecture.

The structure recognizes a disease-associated molecular signature.

An RNA circuit evaluates the signal.

Only when the appropriate conditions are satisfied does the system recruit an editing enzyme or initiate production of a therapeutic protein.

Later, the RNA degrades and the programme disappears.

No research group has yet built this complete system.

But laboratories around the world are developing many of its components.

That is why the present period in RNA biology is so consequential.

The technologies are no longer advancing only along isolated tracks.

They are beginning to intersect.

A genuinely global scientific effort

The geography of this emerging field reflects its intellectual diversity.

Researchers in Germany are designing chemically controllable RNA switches.

South Korean scientists are building molecular logic systems and developing new uses for RNA-associated editing enzymes.

US laboratories are advancing RNA origami, intracellular nanostructures and immunologically active RNA materials.

British, European and American teams are collaborating on synthetic RNA condensates and artificial organelles.

Chinese researchers are expanding the chemistry and controllability of RNA editing.

Australian researchers and international collaborators are tackling the regulatory and agricultural dimensions of RNA biotechnology.

Computational researchers across multiple institutions are attempting to solve RNA sequence–structure–function relationships with increasingly sophisticated machine-learning models.

These efforts cross disciplinary boundaries as readily as geographical ones.

A nanotechnologist increasingly needs to understand RNA folding.

An RNA biologist may need materials science.

A synthetic biologist may need machine learning.

A plant biologist may need colloid chemistry and nanocarrier engineering.

A clinician may ultimately need all of them.

The field emerging from this convergence lies somewhere between molecular biology, materials science, computation and engineering.

RNA biology is changing its central question

The most important transformation may ultimately be conceptual.

Classical molecular biology usually begins with discovery.

Researchers identify a molecule and ask:

What does it do?

That question will remain fundamental.

But RNA engineering introduces another:

What could we make it do?

The distinction is profound.

One question seeks to understand biological systems as they exist.

The other seeks to construct biological behaviour.

Over the coming decade, RNA biology will increasingly involve both.

Scientists will continue discovering new RNAs, modifications, structures and regulatory pathways. Alongside them, however, an expanding engineering discipline will attempt to build RNA molecules that sense, compute, assemble, organize, edit and deliver biological information.

Some will become therapeutics.

Others may become intracellular sensors, synthetic organelles or programmable regulatory circuits.

Some could protect crops.

Others might become components of biological manufacturing systems that are difficult to envision today.

RNA was once described primarily as the molecule carrying DNA's instructions to the protein-making machinery of the cell.

That definition increasingly seems inadequate.

The deeper transformation now underway is that RNA is becoming something much more ambitious:

a programmable material from which researchers can begin to engineer biology itself.

---

References

1. Eisenstein, M. The dark horse of biology: how RNA is becoming a nanotool maker's dream. Nature 655, S2–S5 (2026). doi:10.1038/d41586-026-02180-6.

2. Hedwig, V. et al. Engineering oxypurinol-responsive riboswitches based on bacterial xanthine aptamers for gene expression control in mammalian cell culture. Nucleic Acids Research 53, gkae1189 (2025). doi:10.1093/nar/gkae1189.

3. Kang, H., Park, D. & Kim, J. Logical regulation of endogenous gene expression using programmable, multi-input processing CRISPR guide RNAs. Nucleic Acids Research 52, 8595–8608 (2024). doi:10.1093/nar/gkae549.

4. Chang, X. et al. Designer RNA nanostructures co-transcribed and self-assembled inside human cell nuclei. Nature Communications 17, 1055 (2026). doi:10.1038/s41467-025-67817-y.

5. Fabrini, G. et al. Co-transcriptional production of programmable RNA condensates and synthetic organelles. Nature Nanotechnology 19, 1665–1673 (2024). doi:10.1038/s41565-024-01726-x.

6. Li, S. et al. Programmable artificial RNA condensates in mammalian cells. Nature Nanotechnology 21, 821–830 (2026). doi:10.1038/s41565-026-02164-7.

7. Ng, B. et al. Expression of nano-engineered RNA organelles in bacteria. Nature Communications 17, 2752 (2026). doi:10.1038/s41467-026-69336-w.

8. Qi, X. et al. RNA origami nanostructures for potent and safe anticancer immunotherapy. ACS Nano 14, 4727–4740 (2020). doi:10.1021/acsnano.0c00602.

9. Deng, H., Li, L., Zhao, C. et al. Nanotechnology-mediated precision delivery of mRNA. Nature Materials (2026). doi:10.1038/s41563-026-02623-5.

10. Sun, Y., Cao, Y., Song, Y. et al. Improved RNA base editing with guide RNAs mimicking highly edited endogenous ADAR substrates. Nature Biotechnology 44, 464–476 (2026). doi:10.1038/s41587-025-02628-6.

11. Zhuang, Y., Zhu, Q., Wu, H. et al. Single-strand deaminase-assisted editing for functional RNA manipulation. Nature Biotechnology (2026). doi:10.1038/s41587-025-02956-7.

12. Im, H. W., Jeong, B., Lee, Y. et al. Engineered ADARs enable precision A-to-G base editing of DNA. Nature Biotechnology (2026). doi:10.1038/s41587-026-03223-z.

13. Fradkin, P., Shi, R. I., Dalal, T. et al. Orthrus: toward evolutionary and functional RNA foundation models. Nature Methods 23, 935–945 (2026). doi:10.1038/s41592-026-03064-3.

14. Tarafder, S. & Bhattacharya, D. RNAbpFlow: base pair-augmented SE(3) flow matching for conditional RNA 3D structure generation. Nature Methods 23, 1349–1358 (2026). doi:10.1038/s41592-026-03128-4.

15. Sede, A. R., Moorlach, B., Galli, M. et al. The unfulfilled potential of nanocarriers for RNA delivery in antiviral crop protection. Nature Plants 12, 1166–1178 (2026). doi:10.1038/s41477-026-02323-7.

16. Gunasekara, S., Fidelman, P., Fletcher, S. et al. The future of dsRNA-based biopesticides will require global regulatory cohesion. Nature Plants 11, 664–667 (2025). doi:10.1038/s41477-025-01953-7.

The Giant RNA Polymerase of CCHFV: A New Structural Window into Viral RNA Synthesis and Antiviral Design

 

https://thernablog.blogspot.com/

Viral RNA polymerases are among the most important molecular machines in virology. They copy viral RNA, make viral transcripts, and control whether an RNA virus can successfully replicate inside a host cell. Because human cells do not use the same kind of processive RNA-dependent RNA polymerase for genome replication, these enzymes have long been attractive targets for antiviral drug discovery.

A recent study by Jia and colleagues, titled “RNA synthesis and substrate analog inhibition in the CCHFV polymerase,” provides a major structural and biochemical advance in this area. The work focuses on the L protein of Crimean-Congo hemorrhagic fever virus, or CCHFV, a tick-borne virus belonging to the Nairoviridae family. 

What makes this enzyme remarkable is its size. Nairoviridae L proteins are about 4,000 residues long, making them among the largest known viral polymerases. Until now, this enormous size came with a major mystery: why does this virus need such a large polymerase to perform a job that other RNA viruses accomplish with much smaller polymerase systems? Jia and colleagues address this by reporting structures of the full-length CCHFV L protein, including a 3.0 Å polymerase elongation complex.

The Nature study is important for two reasons. First, it gives us a clearer picture of how this giant viral enzyme organizes RNA synthesis. Second, it identifies nucleotide analogs with sofosbuvir-like ribose modifications that can specifically inhibit the CCHFV polymerase by immediate chain termination.

Why CCHFV polymerase matters

CCHFV is not an ordinary virus from a public-health perspective. It is a tick-borne biosafety level-4 pathogen, meaning it requires the highest level of laboratory containment. The virus causes Crimean-Congo hemorrhagic fever, a severe disease of major concern in endemic regions. Understanding how its polymerase works is therefore not only a structural biology question; it is also a foundation for antiviral discovery.

Like other segmented negative-sense RNA viruses, CCHFV depends on an RNA-dependent RNA polymerase, or RdRP, to copy and transcribe its RNA genome. The L protein contains multiple functional regions, including an endonuclease, the central RdRP module, and a cap-binding domain. These regions cooperate during viral transcription and replication. In segmented negative-sense RNA viruses, transcription often depends on “cap-snatching,” where the viral polymerase captures capped fragments from host RNAs and uses them as primers for viral mRNA synthesis.

The puzzle is that Nairoviridae L proteins are much larger than many related viral polymerase systems. Previous structures of segmented negative-sense RNA virus polymerases generally involved systems of about 2,000–2,500 residues, while Nairoviridae L proteins can reach 3,800–4,900 residues. The authors note that apart from an N-terminal OTU domain, the reason for this unusually large size had remained unclear.

A full-length view of a giant enzyme

To solve this problem, the researchers purified full-length CCHFV L protein and used cryo-electron microscopy to capture different structural states. They obtained apo and promoter-bound states, but the major breakthrough was the 3.0 Å elongation complex, which covered a much larger portion of the enzyme. This structure allowed the authors to define a more complete architecture of CCHFV L and to see how different regions cooperate during RNA synthesis.

One of the most interesting findings is that CCHFV L is not simply a larger version of other viral polymerases. It contains large additions and insertions in all three major functional regions. These additions reshape how the enzyme interacts with RNA. Two Nairoviridae-specific elements are especially important:

FID, or the fingers insertion domain, extends the downstream template RNA-binding path.

UPD, or the upstream product-binding domain, extends the upstream product RNA-binding path.

Together, these domains help explain why the CCHFV polymerase is so large. The extra mass is not random decoration. It appears to form additional RNA-binding paths and interaction networks that may help the enzyme handle long RNA products with sufficient processivity.

FID and UPD: two additions that change the RNA path

The study shows that FID lies near the downstream side of the RdRP active site and may help coordinate template RNA binding together with other polymerase regions. Structural analysis revealed positively charged residues in the relevant groove, consistent with a role in nucleic acid interaction. The authors propose that FID contributes not only to promoter binding but also to general downstream template RNA binding.

On the other side of the active site, UPD helps form an extended path for the upstream RNA product. The study identifies a tunnel-like route involving UPD, CBD, and mid-link regions, with positively charged residues positioned along the putative product RNA exit path. This suggests that the polymerase has evolved extra structural features to guide RNA as it emerges from the active site.

This is where the structural work becomes biologically meaningful. The researchers tested mutations in these interaction networks using a CCHFV minigenome assay. All 14 tested mutations reduced minigenome replication to varying degrees, and mutations affecting FID:RNA and UPD:RNA interactions had particularly strong effects, dropping replication below 20% of the wild-type level.

In simple terms, the extra domains are not just visible in the structure; they matter for viral RNA replication.

How the enzyme moves from initiation to elongation

The authors also propose a model for CCHFV RNA replication. In this model, the polymerase first recognizes the viral promoter and positions the 3′ end of the template RNA at the active site. As RNA synthesis progresses, the enzyme transitions into elongation. When the RNA duplex reaches roughly 10 base pairs, structural elements such as the lid and priming element move to accommodate the growing RNA duplex, and the lid helps separate template and product strands.

This model is useful because viral polymerases are not static machines. They must grip the promoter, initiate RNA synthesis, elongate the RNA chain, separate RNA strands, and eventually complete an entire replication cycle. The CCHFV L structure suggests that FID and UPD may help support processive elongation, possibly allowing the enzyme to synthesize the large L transcript of Bunyaviricetes.

The antiviral angle: sofosbuvir-like nucleotide analogs

The second major part of the study concerns nucleotide analog inhibitors. Nucleotide analogs work by mimicking natural nucleotide substrates. If a viral polymerase incorporates the analog into a growing RNA chain, the analog may disrupt further RNA synthesis.

The best-known example in this category is sofosbuvir, a nucleotide analog used to treat hepatitis C virus infection. Sofosbuvir’s active triphosphate form contains characteristic 2′-α-fluoro-2′-β-C-methyl ribose modifications that cause immediate chain termination in the hepatitis C virus polymerase.

Jia and colleagues asked whether similar chemistry could work against CCHFV RdRP. They found that nucleotide analogs carrying ribose-2′ modifications identical to sofosbuvir could be incorporated by CCHFV RdRP and then stop RNA synthesis immediately. Importantly, the same analogs were not incorporated by Lassa virus and Rift Valley fever virus polymerases in their assays, suggesting specificity for CCHFV among the tested systems.

The authors tested several analogs and found that all four base types with this ribose modification showed incorporation activity and chain-terminating behavior in the CCHFV system. Competition assays further supported the potential of these compounds, although different analogs varied in how strongly they competed with the corresponding natural nucleotides.

This does not mean that sofosbuvir itself is now a proven treatment for CCHFV infection. The study works at the enzyme and structural-biochemistry level. Drug development would still require prodrug optimization, cell culture testing, animal studies, pharmacokinetic evaluation, safety testing, and eventually clinical trials. But the work identifies a promising chemical logic: ribose 2′-α-fluoro-2′-β-C-methyl modification may be a useful starting point for anti-CCHFV nucleotide analog development.

Why this study matters for RNA biology

For RNA biologists, this work is exciting because it connects structure, mechanism, and inhibition in one system. The study does not merely show a beautiful cryo-EM structure. It links structural features to RNA-binding paths, tests their functional relevance through minigenome assays, and then uses active-site insight to explore antiviral inhibition.

It also reminds us that viral RNA polymerases are diverse. The familiar “right-hand” RdRP core is conserved, but viruses build many different accessory domains around that core. These additions can determine how the polymerase recognizes RNA, how it transitions between replication stages, how it separates strands, and how vulnerable it is to nucleotide analogs.

The CCHFV L protein is therefore more than a giant enzyme. It is a molecular example of how RNA viruses expand a conserved catalytic machine into a specialized replication platform.

Conclusions 

The new CCHFV polymerase structures help answer a long-standing question: why are Nairoviridae L proteins so large? The answer appears to lie in expanded RNA-binding architecture. Domains such as FID and UPD extend the paths of template and product RNA, helping organize the enzyme during replication. At the same time, the discovery that sofosbuvir-like nucleotide analogs can terminate CCHFV RNA synthesis provides a valuable starting point for antiviral research.

For The RNA Blog, this study is a reminder of why RNA biology remains one of the most dynamic areas of modern science. A single viral enzyme can teach us about evolution, molecular architecture, disease biology, and drug discovery. In the case of CCHFV, seeing the polymerase in action may be the first step toward learning how to stop it.

How RISC Works: Argonaute, Small RNAs, and the Logic of Gene Silencing

 

A mechanistic guide to RISC assembly, guide-strand selection, target recognition, slicing, repression, deadenylation, and turnover.

Argonaute and RISC: The Molecular Engine Behind RNA Interference

RISC components: The core RISC is a small-RNA-loaded Argonaute (AGO) protein, often associated with GW182/TNRC6 in animals. In metazoans, Dicer and its dsRNA-binding cofactors (TRBP/PACT in mammals; R2D2/Loqs in flies) form a RISC-loading complex (RLC) that hands off small-RNA duplexes to Argonaute. Argonaute contains four domains (N, PAZ, MID, PIWI) that bind the 3' end, 5' end, and body of the guide.

Small-RNA biogenesis/loading: miRNAs derive from Pol II hairpins (pri-miRNAs) processed by Drosha/DGCR8 and then Dicer into ∼22-nt duplexes. siRNAs come from long dsRNA (viral or endogenous) cleaved by Dicer. piRNAs are Dicer-independent ~24-30-nt RNAs from single-stranded precursors (e.g. in germline) loaded into Piwi-clade AGOs. After biogenesis, small-RNA duplexes are loaded into AGO (with Hsc70/Hsp90 chaperones). Guide strand selection depends on 5'-end nucleotide preference and thermodynamic asymmetry, and the passenger strand is removed by Argonaute slicing (if fully complementary) or a "slicer-independent" unwinding mechanism.

Argonaute conformational states: Crystal/cryo-EM structures show AGO as a bilobed protein (MID-PIWI lobe and N-PAZ lobe) that clamps the guide (5' end in the MID pocket, 3' end in PAZ). Loading and target-binding trigger conformational shifts: apo-AGO "open" state, guide-bound "clamped" state, and target-bound state in which the central channel widens to accommodate guide-target pairing. The N-domain helps splay duplex strands and limits 3'-target pairing (enforcing seed-based recognition).

Guide selection & passenger removal: After loading, AGO uses multiple "sensors" to choose the guide strand. Factors include 5'-terminal nucleotide identity (MID pocket preference), thermodynamic stability of ends, and Ago's slicing of the passenger if perfectly paired. In slicer-competent AGOs (e.g. human AGO2, Drosophila AGO2), the passenger strand can be cleaved (at the guide's 10-11 position) to free the guide. Non-slicing AGOs or imperfect duplexes rely on thermal destabilization plus chaperones (e.g. C3PO, La/SSB) to unwind and eject the passenger. Open questions include the exact roles of unwinding factors (C3PO, etc.) and how different AGO isoforms manage strand separation.

Target recognition: The core determinant of target binding is seed pairing: perfect complementarity to guide positions 2-7 (or 8) drives binding. Additional base-pairing 3' of the seed (supplementary pairing) strengthens binding, while central mismatches/bulges generally prevent slicing. Bulged or wobbled sites can still mediate repression if seed pairing is intact. The tolerance of mismatches and the extent of supplementary pairing vary with AGO clade and species. Outstanding questions include the full rules for non-seed interactions and how AGO conformational changes propagate mismatch signals (cf. Joseph & Osman 2012).

Catalytic cleavage: Only AGOs with an active RNase H-like "PIWI" domain can slice targets. Human AGO2 (and, to a lesser extent, AGO3) carry the catalytic DEDH tetrad required for Mg²+-dependent phosphodiester hydrolysis. Cleavage chemistry resembles RNase H: the guide-bound AGO positions the scissile phosphate near two Mg²+ ions, facilitating an SN2 attack by the 2'-OH on the adjacent phosphate. Structures (e.g. human Ago2-miRNA-target complexes) show a kink at the cleavage site induced by the so-called "glutamate finger", orienting the water nucleophile. Open issues include the detailed energetics of catalysis and how slicer-inactive AGOs function in organisms like plants (some plant AGOs have lost slicing yet still mediate silencing).

Translational repression & deadenylation: In animals, AGO-guide complexes recruit GW182/TNRC6 proteins, which in turn bind poly(A)-binding protein (PABP) and the CCR4-NOT and PAN2-PAN3 deadenylase complexes. This leads to shortening of the poly(A) tail, decapping (via DCP1/2), and mRNA decay. miRNA-bound AGO may also inhibit translation initiation (via eIF4G/eIF4A interference). Key experiments tethering GW182 to reporters demonstrate that GW182 alone can induce deadenylation and repression. In flies, loss of GW182 abolishes deadenylation but has complex effects on translational repression. Open questions include how GW182 distinguishes targets for decay vs mere repression, and how initial translation inhibition is triggered prior to mRNA decay.

RISC recycling/turnover: After target repression or cleavage, RISCs must be recycled for further rounds. Target cleavage yields 5' and 3' fragments; recent work suggests phosphorylative events promote release of cleaved products (e.g. AGO2 C-terminal serine phosphorylation accelerates target release). The "loading" AGO may remain bound to the guide for multiple cycles. Small RNAs themselves can turnover (some miRNAs are stabilized by 2'-O-methylation in plants and animals). Factors like XRN1 exonuclease clear cleaved targets. A notable factor, C3PO, degrades AGO-nicked passenger fragments to fully activate RISC. Precisely how AGOs dissociate from targets for new rounds (and how Ago itself is turned over or modified) are active research areas.

Regulatory PTMs and cofactors: AGO function is modulated by post-translational modifications. Human AGO2 is phosphorylated at several sites: for example, Y393 by EGFR (in hypoxia) reduces AGO2-Dicer binding and miRNA loading; S387 by Akt3 promotes recruitment of LIMD1/TNRC6A and DDX6 into repression complexes; a C-terminal S824-S834 cluster is hyperphosphorylated after target binding to accelerate target release. Other PTMs include AGO2 sumoylation, acetylation, ubiquitination, prolyl-4-hydroxylation, and PARylation, many of which affect stability or localization. RISC cofactors include heat-shock chaperones (Hsc70/Hsp90) required for loading duplexes, the C3PO nuclease (for passenger removal), and RNA helicases (e.g. MOV10, to disrupt RNPs). Open questions include the full map of AGO modifications in various cell states and how co-chaperones influence loading kinetics.

Experimental evidence: The RISC mechanism is supported by multiple assay types. X-ray crystallography and cryo-EM have resolved Argonaute structures in apo, guide-bound, and guide-target states (e.g. archaeal and bacterial Argonautes; eukaryotic Ago2-miRNA complexes; TNRC6-AGO complexes). In vitro cleavage assays (radioactive RNA substrates) defined the catalytic "slicer" requirements and rates. Crosslinking immunoprecipitation (CLIP) sequencing (HITS-CLIP, PAR-CLIP, CLASH) have mapped AGO binding sites transcriptome-wide, confirming seed-pairing rules and identifying non-canonical sites. Luciferase reporter assays with inserted miRNA sites have quantified repression efficiency and defined seed/supplement categories. Cryo-EM of the human RISC-loading complex (Ago2-Dicer-TRBP) has recently illuminated loading intermediates.

Open questions/controversies: Despite progress, some issues remain unresolved. The relative contributions of translational repression vs mRNA decay in different contexts is debated. The existence and mechanism of miRNA "target slicing" (beyond perfect siRNA-like sites) is still being explored. The roles of many AGO co-factors (beyond Dicer/TRBP, GW182) are still being delineated. Structural snapshots capture many states, but the dynamic transitions of AGO during target search are less understood. Finally, the diversity of AGO family members (with different activities) raises questions about their specialized functions in various species and pathways.

Argonaute	Domain architecture	Active-site motif	Slicer?	Major pathway/notes
HsAGO1	N–PAZ–MID–PIWI (858 aa)	DEDH (E remains)	No	miRNA repression
HsAGO2	N–PAZ–MID–PIWI (859 aa)	DEDH (canonical)	Yes	miRNA/siRNA (viral)
HsAGO3	N–PAZ–MID–PIWI (925 aa)	DEDH (mutant form)	Marginal¹	miRNA (some slicing)
HsAGO4	N–PAZ–MID–PIWI (859 aa)	DEDN (N instead of H)	No	miRNA
DmAGO1	N–PAZ–MID–PIWI (843 aa)	DEDH (E remains)	No	miRNA (development)
DmAGO2	N–PAZ–MID–PIWI (940 aa)	DEDD (active)	Yes	siRNA antiviral
CeRDE-1	N–PAZ–MID–PIWI (925 aa)	DEDH (active)	Yes	siRNA (RNAi)
Piwi proteins (e.g. HsHIWI2)	N–PAZ–MID–PIWI (1000+ aa, plus Gly-rich N-term)	DEDH / DEDH	Yes (piRNA)	germline piRNA silencing

RISC Composition and Assembly

The core of RISC is an Argonaute protein bound to a single-stranded "guide" RNA. In animals, GW182/TNRC6 proteins (with tandem GW/WG motifs) bind AGO and mediate repression/decay. In the RISC-loading complex, AGO is physically associated with Dicer and its dsRNA-binding partners: in mammals, Dicer binds TRBP and/or PACT; in Drosophila, Dcr-2 binds R2D2 (and Loquacious). These scaffolds bring the small-RNA duplex to AGO. Biochemically, purified human Dicer-TRBP and Dicer-PACT complexes each form stable RLCs that bind siRNA or pre-miRNA; swapping TRBP and PACT domains can alter processing specificity. Notably, RLC assembly increases Dicer's affinity for RNA and presents the duplex in a conformation competent for loading.

During loading, ATP-dependent chaperones (Hsc70/Hsp90, Hop, p23, etc.) are required to "open" AGO for duplex entry. In vitro reconstitution (purified components) shows that Hsp90beta, Hsc70 and cochaperones form a loading machine that presents AGO in a high-affinity state for the duplex. Inhibition of Hsp90 blocks RISC loading in cells, indicating this is a conserved requirement (Tomari & Zamore 2005; Tahbaz et al. 2005). After AGO binds the duplex (one strand destined as guide, the other passenger), a series of strand-separation steps ensues (see below). Only after passenger ejection is the RISC considered mature and able to bind targets.

Table 1 (below) compares selected Argonaute proteins: all share N, PAZ, MID, PIWI domains (N-box and PAZ grip the duplex; MID anchors the guide's 5'-phosphate; PIWI harbors the RNaseH fold). The presence of an active-site Asp/Glu (the "slicer tetrad") determines whether a given AGO can catalyze target cleavage. For example, human AGO2 has the canonical DEDH and is an active slicer, whereas AGO1/3/4 have substitutions (AGO3 can be activated by domain swaps). Organismal distribution varies: many animals encode multiple AGO paralogs (e.g. 4 in humans, each broadly expressed), while plants have >10 AGOs with specialized roles.

Small-RNA Biogenesis and Loading

miRNA Pathway

Animal miRNAs begin as long primary transcripts (pri-miRNAs) made by Pol II. The Microprocessor complex (Drosha + DGCR8) cleaves the pri-miRNA into a ~60-70-nt precursor hairpin (pre-miRNA). This pre-miRNA is exported to the cytoplasm and further diced by Dicer into a ~22-nt RNA duplex with 2-nt 3' overhangs. TRBP and PACT (humans) or Loquacious (flies) bind Dicer's RNase III domains and influence cleavage accuracy and strand selection. The guide strand selection is influenced by 5'-terminal nucleotide preference (AGO MID-domain often favors U or A) and by the relative thermodynamic stability of the duplex ends. The duplex (with 5'-monophosphates on both strands) is presented to AGO: in humans this usually means AGO2, while other AGOs also bind miRNAs but are non-slicing. Chaperone proteins (Hsc70/Hsp90) use ATP to transiently "open" AGO for duplex entry.

siRNA Pathway

siRNAs arise from long double-stranded RNAs (exogenous viruses, transposons, or endogenous transcripts). In Drosophila, Dicer-2 (with partner R2D2) processes long dsRNA into 21-nt siRNA duplexes. In mammals, a single Dicer can generate both miRNAs and siRNAs (e.g. from shRNA expression) with the help of TRBP/PACT. Once produced, siRNA duplexes are loaded into AGO. In flies, AGO2 is specialized for siRNAs; in mammals, AGO2 is the main slicer and can load siRNAs for RNAi. A key feature of siRNA loading is that one strand (the guide) will pair fully with targets, so AGO can cleave complementary mRNA targets.

piRNA Pathway (brief)

In metazoan germlines, piRNAs are 24-30 nt RNAs that associate with PIWI-clade Argonautes (Piwi, Aubergine, AGO3 in flies; PIWIL1-4 in mammals). piRNAs derive from single-stranded cluster transcripts (no Drosha/Dicer required). Mitochondrial endonuclease Zucchini (and Tudor-domain factors) generate primary piRNAs. Ping-pong amplification creates secondary piRNAs via slicer activity of PIWI proteins. The final piRNA-Piwi complexes mediate transposon silencing by target cleavage and transcriptional repression (H3K9 methylation). piRNA 5' ends are 2'-O-methylated by Hen1, further stabilizing them. (See Iwasaki et al. 2015 for review.)

Loading and Strand Separation

In all pathways, after duplex production the RLC loads the duplex into AGO. AGO's MID domain "senses" the 5'-phosphate of one strand to position it as the guide. The other strand (passenger) must be removed. Two mechanisms operate:

Slicer-assisted: If the passenger strand is fully complementary, AGO2 (or other slicers) will cleave it between positions 10-11 (guide numbering). This "nick" promotes rapid dissociation of the passenger fragments. This is the case for siRNAs in canonical RNAi. Matranga et al. (2005) showed that human and fly AGO2 cleaves the passenger of loaded siRNA, while miRNA duplexes (imperfect) are not cleaved.

Slicer-independent: Non-slicing AGOs (e.g. AGO1, 3, 4) or imperfect duplexes rely on the intrinsic thermodynamic bias (less stable 5' end or mismatches) to eject the passenger. At 37°C human AGO1/3/4 can eject an siRNA passenger without cleavage, suggesting a "hotter" conformational dynamics (the PAZ domain transiently releases the 3' end). Accessory factors like C3PO (a Mg²+-dependent endonuclease) can degrade nicked passenger fragments, further promoting activation. La/SSB has also been reported to bind AGO2 and assist release of cleavage products. Thus, even without slicing, AGO can effect strand separation through conformational changes and ancillary helpers.

Open questions remain about the precise kinetics of passenger removal (e.g. how general is C3PO's role?) and how AGOs discriminate guide vs passenger beyond thermodynamics.

Argonaute Structure and Guide/Target Interactions

Argonaute proteins are bilobed. The MID-PIWI lobe forms one side of the nucleic-acid channel, the PAZ-N lobe forms the other (Figure 1 in). The MID domain (Rossmann-like fold) binds the 5'-phosphate and first base of the guide by a conserved pocket. The PAZ domain (OB-fold) binds the 2-nt 3' overhang of the guide. Thus the guide is anchored at both ends. The N-terminal "N domain" lies between the lobes and helps split duplexes and prevent overextension of base-pairing at the guide's 3' end. The PIWI domain is a RNase H-like fold containing the (Asp/Glu) active site. The catalytic tetrad (Asp-Glu-Asp-His) coordinates two Mg²+ ions for phosphodiester hydrolysis. (Non-slicer AGOs have one or more mutations in this motif.)

As shown by crystal structures, AGO-guide interactions define a characteristic "seed channel". Positions 2-8 of the guide (the seed) are pre-organized by contacts to the protein (e.g. MID/PAZ interactions clamp the ends). The N-PAZ lobe covers the 3' half of the guide, preventing pairing until the seed has bound. Upon target binding, structures show the seed region bound to target, inducing a kink at position 6-7. With extensive pairing beyond position 8, the guide-target duplex can extend into the supplementary chamber of the PIWI lobe. The transition from guide-only to guide-target causes conformational shifts: in some cases the PIWI domain repositions its active site loop (the "glutamate finger") to engage the scissile phosphate.

Conformational studies (FRET, cryo-EM) indicate at least three AGO states: apo-open (RNA-free), guide-loaded (central cleft clamped), and target-bound (cleft open to accommodate duplex). The MID and PIWI domains move closer upon guide binding, completing the GW182-binding surface. These structural rearrangements enforce the target recognition rules: only targets pairing to the seed (g2-7/8) can productively bind deep in the channel. Mismatches/bulges in the seed severely weaken binding (seed is base-paired in helix). In contrast, central mismatches (guide 9-11) prevent slicing by misaligning the active site. Supplementary pairing (guide 13-17) can strengthen binding if present. Figures 2-3 in Uchiumi et al. (2016) and structures in illustrate the guide and target path.

Target Recognition Rules

AGO-guide complexes scan mRNAs for complementary sequences. The seed region (guide positions 2-7/8) is paramount: a contiguous Watson-Crick match here is usually required for stable binding. Typical miRNA target sites are classified as 6mer (nts 2-7), 7mer (2-8), or 8mer (2-8+matching A at target position 1) in 3'UTRs. Additional "3'-supplementary" pairing (guide 13-17 to target positions) can compensate for a shorter seed. Many bona fide sites tolerate a single bulge or GU wobble in the seed if flanked by perfect pairs. However, a mismatch at guide position 9/10 (the scissile phosphate) abolishes cleavage.

Genome-wide CLIP-seq experiments (e.g. AGO HITS-CLIP by Chi et al. 2009, Helwak et al. 2013) confirm that 3'UTR sites with canonical seed pairing (often with flanking AU-rich context) are enriched under AGO peaks. Non-canonical sites (seedless or centered sites) exist but are generally weaker. AGO's N-domain can sometimes tolerate small 3'-bulges of the guide, but extended bulges usually require an extra stabilizing anchor (e.g. 3' supplementary pairing) to engage the PIWI lobe. Mutational studies show that introducing bulges in the seed disrupts silencing regardless of downstream pairing.

A remaining mystery is how AGOs detect and "communicate" guide-target mismatches. Molecular dynamics studies suggest an allosteric network within AGO relays information from the seed to the catalytic site and to surface sites. For example, Joseph & Osman (2012) found that seed mismatches induce small shifts in an extensive residue network, ultimately affecting surface loops. In practice, mismatches reduce slicing efficiency and accelerate turnover, but non-slicing repression can still occur (with reduced potency).

Catalytic (Slicer) Cleavage Mechanism

When a target pairs fully to the guide (especially positions 2-12), slicing occurs (in slicing-competent AGOs). The PIWI domain's RNase H fold positions two divalent cations (Mg²+) near the guide-target junction. One metal activates a water nucleophile for in-line attack on the scissile phosphate, while the other stabilizes the leaving group. The conserved glutamate finger (a loop in PIWI) contacts the phosphate backbone to position the scissile bond at the catalytic center. Structural studies of archaeal Ago and human Ago2 (bound to guide and target) reveal the cleavage geometry: the target's phosphodiester is bent at the cleavage site and the 2'-OH of the target attacks the phosphorus, yielding 5'-phosphate and 3'-OH ends.

Biochemical kinetics show slicer cleavage is single-turnover fast (∼minutes) when complementarity is perfect, and essentially abrogated by mismatches or bulges at the cleavage site. Mutagenesis of the DEDH residues (e.g. D597A, H807A in hAGO2) completely blocks cleavage but not binding. For non-slicer AGOs (lacking the full tetrad), target binding still occurs but no phosphodiester bond breakage ensues - these RISCs rely entirely on repression/deadenylation pathways.

Recent cryo-EM data (e.g. Cell 2025 by Zhang et al.) have begun to capture the intermediate states of human AGO2 during cleavage, revealing how the active site reorganizes. The precise catalytic mechanism (e.g. transition state intermediates) likely parallels RNase H enzymes. Open questions include the pH dependence and any required proton transfers, and how AGO3 (with variant PIWI) may occasionally cleave unusual substrates.

Translational Repression and Deadenylation

In metazoans, most miRNA binding triggers repression rather than cleavage. The bridge between AGO and the repression machinery is provided by GW182/TNRC6 proteins. GW182 proteins have an N-terminal AGO-binding region (with multiple tryptophan "GW" motifs) and a C-terminal effector region that interacts with mRNA decay factors. Tethering experiments (GW182 fused to a reporter) show that GW182 alone can induce poly(A) shortening and translational silencing.

Mechanistically, AGO-GW182 complexes recruit PABP and the CCR4-CAF1-NOT deadenylase and PAN2-PAN3 complexes to the target mRNA tail. The deadenylases shorten the poly(A) tail, which leads to decapping by DCP1/2 and 5'->3' exonucleolytic decay (XRN1). GW182 also interacts with DDX6 (RCK/p54) and other decapping enhancers. In Drosophila, knocking down CCR4 or NOT1 abolishes miRNA-dependent deadenylation and decay, but residual translational repression can persist. Thus, translational repression can be mechanistically separated from deadenylation, though in cells they often occur sequentially.

Proposed models for repression include interference with cap recognition or ribosome initiation. For example, GW182-bound CCR4-NOT can inhibit eIF4A/eIF4G, blocking 43S pre-initiation complex assembly. Some data suggest miRNA-mediated repression acts primarily at initiation, while others find elongation stalls or ribosome drop-off. The field agrees that deadenylation is a major downstream effect, but the timing (repression first, decay later) is still debated.

In summary, after target binding the mature RISC can silence expression by (1) slicing (if perfect match, via PIWI), or (2) recruiting GW182 to repress translation and deadenylate/decap the mRNA. The balance of these pathways depends on AGO isoform, target context, and cell type.

RISC Recycling and Turnover

Once an mRNA is cleaved or repressed, the question arises: how is the AGO-guide complex recycled? Cleavage case: Argonaute slices the target, leaving two fragments. These fragments dissociate from AGO (AGO then remains bound to the guide). Recent evidence suggests AGO2 is actively phosphorylated after target binding to promote release: phosphorylation of its C-terminal serine cluster (S824-S834) lowers the affinity for bound mRNA, allowing AGO to turn over more rapidly. Conversely, preventing this phosphorylation leads to "sticky" RISC that holds onto targets. Thus, an AGO phosphorylation cycle accelerates RISC recycling after slicing.

Repression case: If no slicing occurred, AGO stays bound to target 3' UTR. It likely releases by thermal dissociation (since pairing is partial) or with help from RNA helicases (e.g. MOV10) and ATPases (e.g. Me31b/DDX6). Notably, TNRC6 can bind multiple RISCs to one mRNA, possibly stabilizing some interactions. Eventually, after multiple rounds of repression/decay, the RISC may dissociate or be sequestered into P-bodies.

RISC turnover: Argonaute itself is relatively stable (half-lives of many hours) but is turned over by ubiquitination (especially AGO2) under some conditions. Stress or viral infection can trigger Argonaute degradation. Small RNAs also turn over: 3' end 2'-O-methylation in plants and piRNAs protects them from exonucleases. In animals, the lack of 2'-O-methylation in AGO-loaded miRNAs may make them susceptible to tailing and trimming (via TUTases and exonucleases), particularly for aged RISC.

Finally, after mRNA decay, the guide strand itself may be released from AGO and degraded, freeing AGO to load a new duplex. The details of guide recycling are less well studied, but in vitro slicing assays show that AGO2-guide can survive multiple cleavage events.

Regulatory PTMs and Cofactors

AGO activity is finely tuned by post-translational modifications and binding partners:

Phosphorylation: As noted, AGO2 undergoes key phosphorylations. EGFR (under hypoxia) phosphorylates AGO2 at Y393, which disrupts AGO2-Dicer interaction and downregulates miRNA maturation. Serine phosphorylation of AGO2 is rich: S387 (by Akt3 kinase) triggers AGO2 binding to LIMD1 and recruitment of TNRC6A and DDX6, coupling miRNA repression to the CCR4-NOT complex. A C-terminal cluster S824-S834 is phosphorylated upon target binding, lowering mRNA affinity (as above). Other kinases (CK1alpha, GRK4) also modify AGO2 at distinct sites, altering localization (nuclear vs cytoplasmic) or miRNA loading. Overall, phosphorylation regulates when and where RISC binds targets and recruits repressors.

Other PTMs: AGO2 is SUMOylated (on Lys402) to enhance stability and localization to P-bodies. Kinetic studies show SUMOylation can switch AGO2 between translational repression vs slicing modes. Lys48-linked polyubiquitination leads to proteasomal turnover of AGO. Prolyl-4-hydroxylation (on a conserved Pro700 in human AGO2) is important for miRNA activity in tumor cells. PARP enzymes can ADP-ribosylate AGO2, antagonizing its function during stress. These modifications often respond to signaling pathways, linking RISC activity to cellular state.

Cofactors: We have already mentioned Dicer/TRBP/PACT and Hsc70/Hsp90 as loading cofactors. Other notable partners include: (1) C3PO (TREX1 complex) that degrades AGO2-nicked passenger RNAs to finalize RISC activation (Ye et al., 2011). (2) La/SSB binds AGO2 and promotes release of cleaved fragments. (3) MOV10 helicase associates with AGO2-miRNA complexes and is thought to remodel target mRNPs for degradation. (4) GW182-binding factors: LIMD1 (in complex with AKT3) and FMRP/FXR1 may scaffold repression complexes.

Open areas include: How the interplay of these modifications is orchestrated (e.g. does Akt3 phosphorylation always precede AGO2-GW182 binding?), and whether there are uncharacterized AGO partners in specialized RNP granules (e.g. germ granules, stress granules).

Key Experimental Evidence

Structural studies: High-resolution crystal structures of Argonautes (bacterial, archaeal, eukaryotic) have defined the domain architecture and guide/target path. Song et al. (2004) and Nishimasu et al. (2012) solved human AGO2 with guide RNA. More recent cryo-EM (Sheu-Gruttadauria et al. 2019, Ma 2021) captured human AGO-guide complexes at different steps. Structures of AGO with GW182 peptides (Sheu-Gruttadauria et al. 2019) reveal the tryptophan-binding pockets on PIWI. These static images, combined with single-molecule FRET, illuminate how AGO opens/closes during loading and target scanning.

Biochemical assays: Slicing activity has been assayed with radio-labeled target RNAs, establishing that only AGO2 (and AGO3 with modifications) can cut. Mutational scanning of the seed region (by Oglesbee, La Rocca, etc.) mapped the exact base-pairing requirements for repression vs cleavage. Reconstitution of RISC in vitro (Doudna lab) with purified human proteins showed that an RLC of Dicer-TRBP-Ago2 suffices for efficient loading and cleavage of complementary targets (Noland & Doudna 2013). Tethering GW182 to a reporter demonstrated that CCR4-NOT recruitment alone can silence translation without AGO (Eulalio et al. 2009).

High-throughput target identification: HITS-CLIP and PAR-CLIP of AGO proteins (Hafner et al. 2010; Chi et al. 2009) identified thousands of binding sites, refining seed-match rules. CLASH (crosslinking ligation and sequencing) captured chimeric reads of miRNA-target hybrids (Helwak et al. 2013), revealing non-canonical sites and miRNA sponges. Ribosome profiling experiments (e.g. Guo et al. 2010, Eichhorn et al. 2016) showed that mRNA decay is the dominant outcome of miRNA action, supporting the model that repression precedes deadenylation and decay.

Functional reporters: Hundreds of studies using luciferase or GFP reporters with synthetic miRNA sites (seed matches, bulged sites, etc.) have empirically measured repression efficiency. Such assays confirmed the hierarchy of site types (8mer>7mer-A1>7mer-m8>6mer) and showed that supplemental 3' pairing boosts repression of 6mers (Brennecke et al. 2005).

Each of these assays underpins the mechanistic model: structural data define the molecular contacts; in vitro assays reveal the chemistry; and genomic experiments validate the rules in cells.

Pathway	Small RNA	Size (nt)	Precursor/Processing	Key AGO effector	Mode of action
miRNA	miR, miR* duplex	~22	pri-miR –(Drosha)→ pre-miR –(Dicer)→ duplex【18†L1942-L1948】	hAGO1–4 (AGO2 slicer)	Seed pairing → translational repression & decay【33†L262-L270】
siRNA	siRNA duplex	21–23	long dsRNA –(Dicer)→ duplex (TRBP/PACT or R2D2–Dicer)【23†L263-L271】	hAGO2, DmAGO2	Full pairing → target cleavage (slicer)
piRNA	piRNA	26–31	single-strand transcript –(Zucchini + ping-pong)→ piRNA【56†L113-L121】	Piwi-clade (Aub, Piwi)	Transposon silencing by cleavage and heterochromatin

References:

Perspective: machines for RNAi. Genes and Development, 2005. https://doi.org/10.1101/gad.1284105

Origins and Mechanisms of miRNAs and siRNAs. Cell, 2009. https://pmc.ncbi.nlm.nih.gov/articles/PMC2675692/

Towards a molecular understanding of microRNA-mediated gene silencing. Nature Reviews Genetics, 2015. https://www.nature.com/articles/nrg3965

The Structure of Human Argonaute-2 in Complex with miR-20a. Cell, 2012. https://doi.org/10.1016/j.cell.2012.05.017

Structural basis for microRNA targeting. Science, 2014. https://pubmed.ncbi.nlm.nih.gov/25359968/

From guide to target: molecular insights into eukaryotic RNA-interference machinery. Nature Structural and Molecular Biology, 2015. https://www.nature.com/articles/nsmb.2931

Biological principles of microRNA-mediated regulation: shared themes amid diversity. Nature Reviews Genetics, 2008. https://www.nature.com/articles/nrg2455