Researchers led by Davide Incarnato at the University of Groningen have demonstrated that the FDA-approved cancer drug mitoxantrone works against RNA not simply by attaching to it, but by forcing it to collapse from multiple folded shapes into a single stable form. The finding, reported in late March 2026, reframes a central question in drug design: why do so many small molecules that bind RNA in a test tube fail to produce a therapeutic effect inside cells? The emerging answer is that binding alone is not enough. What matters is whether a drug can physically reorganize the RNA it targets.
Mitoxantrone Locks RNA Into One Shape
RNA molecules are not static strands. They fold back on themselves in ways that determine which genes get turned on or off, which proteins get made, and how cells respond to stress. Incarnato’s group showed that one RNA sequence can populate several shapes, each with distinct regulatory consequences. In cells, these shapes coexist as a fluctuating ensemble rather than a single fixed structure, and subtle shifts in that ensemble can flip regulatory switches.
The new work focused on mitoxantrone, a chemotherapy agent used for certain cancers, and mapped how it affects this structural landscape. The team found that mitoxantrone drives three coexisting RNA conformations into one dominant fold, effectively collapsing the ensemble into a single, drug-stabilized state. That collapse is not a cosmetic change: RNA folding has direct consequences for gene regulation, so locking the molecule into one configuration can silence some regulatory options while amplifying others.
At the molecular level, mitoxantrone behaves less like a passive passenger and more like a structural clamp. In the Groningen experiments, the drug recognized particular structural motifs and then reinforced them, suppressing alternative folds that would otherwise be accessible. This helps explain why mitoxantrone emerges as a promising template for RNA-targeting design: it does not simply recognize its target, it reshapes it in a predictable way.
This distinction between binding and reshaping is crucial for drug discovery. Many candidate compounds show strong affinity for an RNA target in biochemical assays but produce little or no effect in living cells. The Groningen study underscores that a compound must do more than sit on an RNA surface; it must reconfigure the RNA’s folding landscape to change downstream biology. In practical terms, that means screening not just for tight binding, but for molecules that measurably shift RNA conformational ensembles.
Splicing Drugs Already Prove the Point
The clearest clinical validation for the “reshape, don’t just bind” thesis comes from spinal muscular atrophy (SMA), a genetic disease caused by insufficient production of the SMN protein. Two small-molecule drugs, risdiplam and branaplam, were developed to correct defective splicing of SMN2 pre-mRNA. Both increase SMN protein levels, but they do not perform equally well in patients, and their mechanistic differences have drawn close scrutiny.
Cold Spring Harbor Laboratory researchers led by Adrian Krainer have described these SMA therapies as splicing-modifying drugs that remodel the RNA they bind. The more effectively a compound reshapes the local structure around the SMN2 splice site, the more robustly it promotes inclusion of the critical exon and restores SMN production. Subtle differences in how each drug sculpts that RNA microenvironment help explain their divergent clinical profiles.
Structural biology has filled in the atomic details. High-resolution studies show that small-molecule splicing modifiers bind SMN2 pre-mRNA at defined sites and stabilize a specific RNP configuration that favors correct splice-site recognition. Rather than simply plugging into a pocket, the drug, RNA, and associated proteins form a composite structure whose geometry dictates whether the spliceosome will assemble productively.
NMR-based work has further revealed that multiple splicing modifiers converge on a shared structural motif: a bulged adenine near the 5′ splice site. These compounds clamp around this bulge, and defined structural features correlate with drug potency. Molecules that most effectively immobilize the bulged nucleotide and reorganize nearby helices are the ones that best recruit the cellular machinery needed to correct splicing. By contrast, ligands that occupy the same general region but leave the local fold largely unchanged show weaker therapeutic effects, even when their binding affinities appear comparable in vitro.
Antiviral Evidence From G-Quadruplex Shifts
The reshaping principle also extends into virology. BRACO-19, a small molecule first explored for its ability to stabilize DNA G-quadruplexes in cancer, has been shown to induce a conformational shift in an RNA G-quadruplex found in Cucumber mosaic virus. In this plant pathogen, G-quadruplex structures in the 3′ untranslated regions of the 1a, 2b, and CP genes toggle between intermolecular and intramolecular forms, influencing how efficiently the viral genome is replicated and translated.
When BRACO-19 binds, it pushes the equilibrium toward a compact intramolecular quadruplex, effectively converting looser, intermolecular assemblies into tighter structures. This structural rearrangement interferes with the viral life cycle and reduces proliferation. Again, the critical step is not mere occupancy of the RNA, but a drug-induced rebalancing of conformations that deprives the virus of a structure it depends on.
Complementary work in human cells reinforces this view of RNA as a dynamic target. Using live-cell imaging and chemical probes, researchers reporting in Nature Communications showed that RNA G-quadruplexes are highly dynamic in vivo and that small molecules can selectively compete with or modulate these structures. The observed behavior argues against designing drugs around a single “frozen” RNA model and instead supports strategies that explicitly consider how compounds will shift structural populations under physiological conditions.
Why Binding-Only Screens Fall Short
Despite this mounting evidence, most computational pipelines for RNA-targeted drug discovery still emphasize affinity over structural impact. Docking algorithms and scoring functions are typically calibrated to predict how tightly a small molecule will attach to one chosen conformation of an RNA motif. That approach implicitly assumes that the biologically relevant structure is known and static, even though experiments repeatedly show that functional RNA exists as a fluctuating ensemble.
The RNAmigos2 platform highlights both the promise and the limitations of this paradigm. Described in a recent Nature Communications study, RNAmigos2 uses machine learning to infer preferred RNA-ligand interactions from structural data and then predict small molecules likely to bind new RNA targets. It represents a major step toward systematic, data-driven discovery of RNA binders, but its core objective is still to optimize binding, not to forecast how a ligand will remodel the RNA’s conformational landscape.
That gap matters. A candidate emerging from a binding-focused screen may latch onto an RNA motif yet leave the broader folding ensemble essentially unchanged. In such cases, downstream regulatory events may proceed as if the drug were not there. By contrast, a compound that modestly sacrifices affinity but strongly favors a regulatory conformation, such as a splice-competent state or an antiviral quadruplex, could deliver a much larger biological effect.
To close this gap, researchers are beginning to argue that computational tools must evolve from static docking engines into dynamic ensemble models. Instead of scoring a ligand solely on how well it fits a single structure, future platforms could evaluate how it shifts a distribution of conformations inferred from chemical probing, NMR, or cryo-EM data. Such models would be more computationally demanding, but they would align better with what experiments on mitoxantrone, SMA drugs, and viral G-quadruplexes are already revealing.
Designing the Next Generation of RNA Drugs
Taken together, these strands of evidence point toward a new design principle: the most effective RNA-targeting drugs will be those that deliberately reshape their targets. For chemists, that means prioritizing scaffolds that stabilize or destabilize specific structural motifs (bulges, junctions, quadruplexes, or long-range contacts), rather than simply maximizing surface complementarity.
For biologists and structural researchers, it underscores the importance of mapping RNA ensembles in their native cellular contexts. Techniques such as in-cell SHAPE probing, time-resolved NMR, and live-cell imaging will be essential for identifying which conformations are truly regulatory and which are transient bystanders. Those data, in turn, can guide both medicinal chemistry and machine-learning models toward ligands that steer RNA into therapeutically useful shapes.
The mitoxantrone study offers a concrete proof of concept: a clinically used drug can be repurposed to control RNA by collapsing its structural diversity into a single, functionally distinct state. As more examples accumulate, from splicing modifiers to antiviral quadruplex ligands, the field is converging on a simple but powerful rule. For RNA-targeting drugs, binding is only the beginning; the real therapeutic leverage lies in the ability to rewrite the folding code.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
