The Druggable Transcriptome Project: From Chemical Probes to Precision Medicines

Abstract

RNA presents abundant opportunities as a drug target, offering significant potential for small molecule medicine development. The transcriptome, comprising both coding and noncoding RNAs, is a rich area for therapeutic innovation, yet challenges persist in targeting RNA with small molecules. RNA structure can be predicted with or without experimental data, but discrepancies with the actual biological structure can impede progress. Prioritizing RNA targets supported by genetic or evolutionary evidence enhances success. Further, small molecules must demonstrate binding to RNA in cells, not solely in vitro, to validate both the target and compound. Effective small molecule binders modulate functional sites that influence RNA biology, as binding to nonfunctional sites requires recruiting effector mechanisms, for example degradation, to achieve therapeutic outcomes. Addressing these challenges is critical to unlocking RNA’s vast potential for small molecule medicines, and a strategic framework is proposed to navigate this promising field, with a focus on targeting human RNAs.

The Human Genome Project^1,2 revealed that ∼80% the human genome is transcribed and that 1–2% is translated.³ Subsequent studies have established RNA as a critical driver of human biological function and disease; however, drug discovery has historically focused on proteins. The term “undruggable” originated from the observation that only ∼15% of protein families have small molecule binders.^4,5 Expanding the druggable protein space has involved various strategies, such as fragment-based drug discovery,^6,7 activity-based protein profiling,^8,9 structure-based drug design,¹⁰⁻¹² molecular docking,^13,14 and degradation by induced proximity.¹⁵⁻¹⁷ Despite these advances, many diseases and targets remain refractory to protein-focused treatments. For instance, transcription factors like c-MYC—a key driver of cancer with poor prognostic outcomes—and intrinsically disordered proteins (IDPs) such as α-synuclein, associated with Parkinson’s disease, pose significant challenges. Their shallow or disordered pockets hinder high-affinity ligand binding due to conformational entropy losses. Yet, their encoding mRNAs often have robust structure.¹⁸

A myriad of discoveries have shown the vast repertoire of RNA in biological processes and its potential to be modulated to provide medicines (Figure 1). This perspective underscores the potential of small molecules targeting human RNA to expand the druggable space. Given that more of the human genome is transcribed than translated, targeting RNA—including noncoding (nc)RNAs and mRNAs—enables intervention in disease mechanisms beyond the reach of current protein-focused methods.¹⁹⁻²¹ For example, RNA-targeted small molecules can bind to specific pockets in mRNAs encoding “undruggable” proteins, thereby modulating their translation. Promising results have already been demonstrated with mRNAs encoding proteins such as α-synuclein^22,23 and c-MYC²⁴ as well as ncRNAs.²⁵

Figure 1.

A timeline of major discoveries in RNA, cementing this biomolecule at the forefront of biological and biomedical research.

Considerations for Selection of an RNA Target for Small Molecule Development

In a precision medicine approach, it is crucial to target biomolecules with unique expression or functional profiles in disease-associated states. Functional genomics and genetic studies have provided prioritized lists of such targets. Various human genetics studies have established RNAs that have folded structures that cause disease,^26,27 which could be disabled with small molecules. In addition, large-scale functional genomics studies on cancer genes have been systematically conducted and publicly deposited in resources such as the Cancer Dependency Map (DepMap) portal.²⁸ Each will be discussed as a means to select targets in the context of RNA-targeted small molecules, along with ways in which RNA structures can cause disease (Figure 2), and how to assess whether RNAs form structures in cells (Figures 3 and 4).

Figure 2.

A significant portion of the genome is transcribed into RNA, which has diverse and critical biological roles beyond encoding and decoding protein synthesis. RNA structures play a pivotal role in regulating the functions of both coding and noncoding RNAs. For example, RNA structures in the untranslated regions (UTRs) of mRNA can influence translational efficiency and mRNA stability. Noncoding RNAs, such as long noncoding RNAs (lncRNAs) and microRNAs, are essential in regulating various biological processes, including protein translation, epigenetic modifications, transcription, and pre-mRNA splicing. Both coding and noncoding RNAs are integral to understanding human disease biology.

Figure 3.

RNA structures play crucial roles in biology, and experimentally informed prediction methods can help identify structured RNAs. Chemical modification reagents interact differently with structured and unstructured regions in an RNA target. By identifying modification sites using sequencing technologies, chemical mapping data can be incorporated into RNA structure prediction⁵⁸ through free energy minimization algorithms. Chemical modification is treated as a pseudo free energy in these prediction algorithms, as RNA dynamics—such as base pairing at the ends of helices and structures in noncanonically paired regions—can influence chemical reactivity. This pseudo free energy approach has significantly improved the accuracy of RNA structure predictions that incorporate experimental restraints.⁵⁴ Additionally, prediction algorithms can calculate base pairing probabilities to better identify regions of RNA that are most likely to be structured. Evolutionary covariation remains the gold standard for RNA structure assignment and should be integrated into these approaches to assess the confidence of predicted RNA structures, as described.⁶⁷⁻⁷⁰

Figure 4.

Experiments are essential to determine whether an RNA target is functional and to provide evidence of RNA structure within cells. In this example, the regulatory RNA hairpin structure at the exon 10-intron junction of MAPT was modified to evaluate how structural alterations affect pre-mRNA splicing.³⁴ Increasing the thermodynamic stability of this RNA structure reduces exon 10 inclusion, thereby decreasing the production of 4R Tau. Specifically, greater thermodynamic stability of the RNA structure reduces U1 snRNP binding at the exon 10-intron junction, resulting in less exon 10 inclusion in the mature mRNA. These studies demonstrated that this RNA structure regulates pre-mRNA splicing conferred by thermodynamic stability. Furthermore, they showed that small molecules capable of binding and stabilizing this RNA structure thermodynamically could reduce exon 10 inclusion, thereby reducing the levels of toxic, aggregation-prone 4R Tau.

Genetics and RNA target selection have demonstrated that many RNAs can cause disease via their folded structures (Figures 2 and 3), and these RNAs fall into several classes.²⁹ Specific RNA structures with regulatory functions have also been identified as critical contributors to disease mechanisms. For example, frontotemporal dementia with parkinsonism linked to chromosome 17 (FTDP-17) arises from mutations in an RNA structure (a splicing regulatory element; SRE) formed at the exon 10-intron junction of microtubule-associated protein Tau (MAPT) pre-mRNA.^30,31 Exon 10 encodes a microtubule-binding domain, and its inclusion affords a toxic 4-repeat Tau protein that is aggregation-prone and drives neurodegeneration.^32,33 A series of early studies showed that specific mutations in the SRE to thermodynamically stabilize or destabilize the SRE and that thermodynamic stability correlated with less inclusion of exon 10 and hence reduced expression of 4R Tau (Figure 4).³⁴ Thus, small molecules that affect the stability of the RNA structure influence the alternative splicing event.³⁵⁻⁴¹

There are >40 diseases caused by repeat expansions, and many are among the most important historically in the field of genetics, including Huntington’s disease. Toxicity in microsatellite disorders can be traced to the corresponding RNA, which often causes disease via a toxic gain of function. For example, myotonic dystrophy type 1 (DM1) is caused by an expansion of r(CUG) repeats in the 3′ untranslated region (UTR) of the dystrophia myotonica protein kinase (DMPK) gene. This r(CUG)^exp sequesters the pre-mRNA splicing regulator muscleblind-like 1 protein (MBNL1) among other proteins, the source of widespread alternative pre-mRNA splicing defects.^42,43 The most common genetic cause of amyotrophic lateral sclerosis and frontotemporal dementia (ALS/FTD) is caused by an intronic hexanucleotide repeat expansion of r(G₄C₂) repeats in the C9orf72 gene. The RNA repeat expansion disrupts RNA metabolism and triggers aberrant translation, among other mechanisms that contribute to disease pathology.⁴⁴ Much effort has shown that therapeutic intervention against these disease-driving RNAs by small molecules and oligonucleotides is a viable strategy.⁴⁵⁻⁴⁹

Important RNA drug targets have been identified by using cancer dependency maps. The studies that form the basis of these maps systematically knocked down genes in over 1,000 cancer cell lines using CRISPR or short hairpin (sh)RNA technologies. By analyzing phenotypes from these screens, genes essential for cancer cell survival were identified, and by correlating these dependencies with genetic alterations in cancer cells, it is possible to uncover vulnerabilities specific to all cancer cells or to subtypes. The DepMap portal aggregates these data, offering public access to guide the development of precision medicines.²⁸ There are significant challenges in using cancer vulnerabilities, particularly the absence of data for ncRNAs in DepMap, complicated by the diversity of ncRNAs and their sequence overlap with mRNAs. In 2024, studies addressed this gap by developing genome-scale transcriptome screening methods for lncRNAs using Cas13d/CasRx systems. These efforts produced a size-reduced guide RNA (gRNA) library, termed Albarossa, targeting 24,171 lncRNA genes.⁵⁰ This library facilitates broad, high-throughput assessments of lncRNAs’ roles in disease, offering a powerful tool to uncover novel cancer vulnerabilities linked to ncRNAs.

Selection of RNA Structures for Small Molecule Binding

Once a disease-associated or biologically relevant RNA has been selected or identified, its druggability or ligandability must be assessed. Typically, this is first accomplished by assessing if the RNA folds into robust structures and if so the types of structures formed (Figures 3 and 4). Several methods have been developed to model RNA structure from sequence, most commonly by free energy minimization; here, RNA structures are predicted using recursive algorithms^51,52 and Turner rules for RNA structural thermodynamics.^53,54 Details on how these algorithms work have been previously articulated.⁵⁵ These predictions can be further refined with experimental restraints such as those derived from chemical modification reagents (Figure 4).⁵⁴ For instance, dimethyl sulfate (DMS) reacts with the base pairing face of adenine and cytosine while Selective 2′ Hydroxyl Acylation analyzed by Primer Extension (SHAPE)⁵⁶⁻⁵⁸ reagents react with flexible 2′ hydroxyl groups present in unpaired or structurally dynamic nucleotides.⁵⁹ By using reverse transcriptase (RT), the sites of chemical modification can be detected as RT stops or mutations.^60,61 These data can then be incorporated into RNA free energy minimization algorithms by using chemical reactivity as a free energy penalty if the reacted nucleotide is paired in the prediction, rather than assuming reacted nucleotides are always unpaired (Figure 3).

Despite the advances in RNA secondary structure prediction, more robust structural analysis is required to ensure that predicted structures align with the actual structures found in cells and that these structures are functional. Evolutionary covariation is a historic gold standard for validating RNA structure predictions. This method was employed by Woese to identify the kingdoms of life and to predict the secondary structure of ribosomes.⁶²⁻⁶⁴ When the ribosome structures were published, the predictions based on evolutionary covariation were shown to be >97% accurate.⁶⁵ Therefore, when possible, evolutionary covariation should be employed to further refine RNA structure predictions. Various approaches, such as ScanFold, can be integrated with covariation and should be carefully considered.⁶⁶ A few statistical and other approaches have been described and should be used to provide confidence that an RNA structure has experimental and/or covariation support.⁶⁷⁻⁷⁰

Transcript Diversity and Disease Relevance

An important consideration in selecting RNA targets is transcript diversity, abundance, and the role a specific transcript forms in driving disease phenotypes. There are at least two elements of transcript diversity, stemming from variation at the 5′ end, producing different UTRs, and from alternative pre-mRNA splicing, particularly in disease-relevant tissues.⁷¹⁻⁷³ Moreover, the relative abundance of each form and determining which is primarily responsible for driving disease phenotype is of import. Ideally, the RNA structure selected as a therapeutic target should be present in the disease-driving form; depending on the disease, its conservation across all transcript variants may be advantageous. For example, the androgen receptor, a protein drug target in prostate cancer, has various pre-mRNA splicing variants in clinical samples, and many confer resistance to front line protein-targeted medicines.⁷³ By carefully considering transcript diversity and structure, RNA-based drug development can be better tailored to address the root causes of disease and thus can have the largest impact on a disease-causing function. Collectively, assessing the full range of transcripts produced from a gene is crucial in a drug development program.

Identification of Small Molecules That Bind RNA Structures

If developing compounds that have a simple binding mode of action, it is crucial to identify targetable structures that are functional. Numerous studies have identified regions of RNA structure that can be targeted by small molecules to elicit functional outcomes. These include structures that interact with proteins (e.g., r(CUG)^exp-MBNL1 complex in DM1), structures in 5′ UTRs nearby a start codon (e.g., α-synuclein mRNA and its iron-response element (IRE)), structures in the 3′ UTR that stabilize the RNA, or those that affect pre-mRNA splicing (e.g., Tau pre-mRNA) (Figures 3 and 4).

A variety of approaches are used to identify small molecules that bind RNA targets (Figure 5). Rather than screening the entire RNA target, RNA fragments containing key structural elements are often screened.⁷⁴ When an RNA binds to a protein, a protein displacement assay can be conducted, as implemented for r(CUG)^exp-MBNL1 complex that causes DM1.^75,76 Caution is necessary in such assays, however, as many compounds can be identified as hits regardless of whether they bind to the protein or RNA. Secondary assays are required to confirm that the intended biomolecule is bound, as binding the unintended target can have a dramatically different activity in cells.⁷⁷

Figure 5.

Various strategies developed to identify small molecules that bind RNA. Small molecule microarrays have been employed to identify compounds that bind to RNA, including isolated RNA fragments such as mimics of the A-site of rRNA⁸⁸ and larger RNAs such as self-splicing pathogenic group I introns.⁹¹ Covalent fragment screening has been used to identify RNA-binding molecules.¹⁰⁰ This approach not only provides binding profiles but also pinpoints binding sites, which can guide the optimization of lead molecules through molecular assembly strategies. DNA-encoded libraries (DELs) have also been utilized to discover small molecules that bind RNA.^98,173 Mass spectrometry techniques, including ESI-MS⁸⁰ for detecting intact complexes and affinity mass spectrometry (ALIS), have been widely applied to identify ligands that bind RNA effectively.^83,84

Various approaches have been used to find direct RNA binders. Mass spectrometry approaches, such as electrospray ionization mass spectrometry (ESI-MS), have been used to screen small molecule libraries for RNA-binding activity.⁷⁸⁻⁸⁰ In these approaches, a direct and detectable complex is formed between the RNA and small molecule. One challenge is that the high concentrations of the analyte and ligand required for ESI-MS analysis that occurs when the droplets are concentrated in the ionization chamber sometimes lead to false positives. Affinity mass spectrometry has also been employed to identify small molecule binders to RNA. In this method, RNA is incubated with a compound, and the RNA-compound complex is separated using size-exclusion chromatography.⁸¹⁻⁸⁴ This method distinguishes molecules that bind to RNA from those that do not, however compounds with short residence times may be missed, a challenge observed in other assays. Other methods, such as surface plasmon resonance (SPR), have been used to screen for RNA binders, typically by screening small molecule fragments. As with many fragment-based screening methods, low affinity can make it difficult to detect binding events. Alternatively, nuclear magnetic resonance (NMR) spectrometry-based approaches have proven useful to identify fragments that bind RNA.⁸⁵⁻⁸⁷ Small molecule microarrays are another tool for identifying RNA-targeting compounds,⁸⁸⁻⁹² which enables screening for single targets and multiplexing.^93,94

DNA-encoded libraries (DELs) are a robust platform for identifying ligands that bind to proteins,^95,96 and their application in discovering RNA binders has gained traction. A significant challenge in this approach is interference from the DNA tag, which can form nonspecific interactions with the RNA target or the compounds, potentially obscuring selection results. To address this issue, various innovative approaches have been introduced. Patches, derived from the target RNA’s sequence, bind to the DNA tags on library molecules, eliminating false positives and increasing confidence that the identified hits are binding to the intended RNA target.⁹⁷ Recently, a “DEL Zipper” approach that uses double stranded (ds)DNA tags was developed to further reduce nonspecific interactions with the target RNA.⁹⁷ Solid-phase DELs have also been developed to minimize tag interference.⁹⁸ In this approach, small molecules are displayed on beads, with most of the molar loading comprising the small molecule and only a minor portion consisting of the DNA tag. This design reduces tag-related interactions, enabling library-versus-library screenings⁹⁸ and identifying compounds that bind to specific RNA targets.⁹⁹ DEL approaches in general could deliver many compounds, however, the compounds are typically very structurally similar because of a diminished scope of DEL-compatible chemical reactions.

As mentioned above, many fragments bind RNA with low affinity and/or short residence times. To overcome these challenges, compounds that form a covalent bond upon binding RNA have been employed. Small molecules featuring both a diazirine and an alkyne tag can be screened by incubating with an RNA target and photolyzing the compounds to facilitate covalent attachment to RNA.¹⁰⁰ By screening a diverse set of compounds and mapping their binding sites transcriptome-wide, the resulting data can be used to optimize RNA-targeting compounds. This approach offers several advantages: (i) it enables the simultaneous screening of numerous compounds; (ii) the formation of covalent adducts allows for the detection of interactions with modest affinities; and (iii) binding sites for these compounds can be precisely defined using sequencing techniques. This approach has been used to profile RNA binding small molecules in cells in an unbiased fashion.^24,101

Challenges in Small Molecule Drug Design for RNA Targets

Structure-based design is another method for developing for RNA-targeting small molecules.¹⁰² One of the key difficulties is that many of the force fields¹⁰³ used for protein-based design are not ideal for nor tailored to RNA targets. Small molecules that bind RNA often rely on electrostatic interactions with the target for binding, which may differ from the interactions involved in protein-small molecule binding. Furthermore, there is a lack of sufficient data on RNA-targeting ligands to train accurate models for structure-based RNA drug discovery.

Another challenge in RNA-targeted small molecule design is that even a minor shift in the core structure of a compound can significantly affect the positioning of its auxiliary functional groups, potentially altering its affinity for the RNA target. Many ligands also bind RNA pockets that are dynamic. Thus, the sensitivity of ligand poses due to dynamics requires careful optimization to ensure effective and high affinity binding to RNA targets. Although the dynamic nature of RNA can be leveraged to identify hit compounds,^104−107 this conformational flexibility can cause significant issues when optimizing compounds for affinity (low nM range) while still maintaining cellular activity.

Sequence-based design is also an emerging approach to identify lead compounds that bind RNA.^25,108,109 Inforna is a computational tool that combines structure-based drug design and RNA sequence/structure information to predict potential RNA-targeting compounds. Inforna leverages RNA secondary structure data to model small molecules that could bind to specific regions of an RNA and potentially modulate its function, opening new avenues for drug discovery. By utilizing computational predictions, Inforna helps identify regions of RNA that are “druggable” and assesses the feasibility of targeting those regions with small molecules. Furthermore, Inforna allows identification of both on- and off-targets early in the discovery process. For example, it can identify transcripts that have structures that can bind to the same small molecule. Here, the targetable structure might be the same in more than one transcript or it might be an alternative structure that binds the small molecule with similar or lower affinity. By using information on the on- and off-targets of compounds, more specific molecules can be identified that affect an RNA target.¹¹⁰ The tool is part of a broader trend toward expanding the druggable genome to include RNA targets, and it focuses on maximizing the potential of RNA-binding small molecules in therapeutic contexts. This approach has allowed facile identification of bioactive compounds across multiple targets and has provided multiple preclinical candidates with activity in mouse models of various diseases.^{24,49,109,111−114}

Cell-Based Target Validation Is Essential to Derisk Compounds and Confirm Binding Sites

Target engagement in cells should be performed as quickly as possible using approaches such as cross-linking, direct RNA cleavage, or inhibiting antisense oligonucleotides (ASOs) from binding and inducing cleavage by RNase H (ASO-Bind-Map).¹⁹ Demonstrating direct occupancy of an RNA target in cells for compounds identified through in vitro screening helps prioritize candidates and ensures the RNA structure targeted in vitro is relevant in cells. Direct binding site mapping is preferred, but some studies rely on competition assays with reactive RNA structure mapping reagents. These assays require high-affinity compounds to protect RNA binding sites from irreversible reactions. However, structural changes induced by small molecules can complicate data interpretation.

Heterobifunctional compounds are another valuable tool for identifying RNA targets and binding sites in cells. Such molecules can contain reactive warheads—such as electrophilic compounds,^115−117 SHAPE-derived anhydrides,¹¹⁸ or diazirines^35,100,119—that covalently attach small molecules to bound RNAs. Dubbed Chemical Cross-linking and Isolation by Pull-down (Chem-CLIP), cross-linked complexes can then be retrieved via biotin tags or click-chemistry handles, with binding sites identified through reverse transcription (RT) analysis that reveals RT stops or mutations.^119−121 Competitive binding assays confirm specificity by showing that high-affinity binders block cross-linking reactions.^122,123 Likewise, cleavage-based approaches use heterobifunctional small molecules with cleavage modules, such as modified bleomycin, to cleave RNA.¹²⁴ Cleavage-based methods provide binding site fingerprints without requiring RNA enrichment but require optimization to avoid excessive RNA degradation during treatment.

The ASO-Bind Map method identifies RNA-small molecule interactions in vitro and in cells using ASOs to target specific RNA sequences.²³ ASOs form DNA-RNA hybrids that recruit RNase H and induce cleavage of the RNA target. When a small molecule binds to the RNA, it stabilizes the structure and prevents or reduces ASO binding, inhibiting RNase H cleavage. Changes in RNA cleavage patterns reveal whether a small molecule binds as well as its effects on RNA structure. This method accelerates RNA target validation as chemical derivatization of the ligand is not required. However, the ASO-Bind Map method requires designing new ASOs for each potential off-target site, which can be labor-intensive when analyzing multiple targets. Nevertheless, this approach—alongside cross-linking and cleavage methods—is indispensable for validating target engagement, prioritizing effective compounds, and advancing RNA-targeted drug discovery.

Bioactive Binding to Small Molecules: Challenges and Opportunities

Bioactive compounds that bind RNA targets and modulate their function are opening new therapeutic possibilities. Indeed, several functional sites within RNA targets can be influenced by small molecules, of which several are highlighted below.

Micro (mi)RNAs are a class of ncRNAs that play important biological roles. These small RNAs are synthesized as precursors that are processed into mature miRNAs that regulate gene expression. Bioactive ligands can inhibit miRNA maturation by binding to sites near or within these ribonuclease processing regions, that is, these functional sites, or by binding proteins involved in maturation.^125−131 In contrast, deciphering functional sites in lncRNAs can be challenging due to their complexity (Figure 6). Many of these RNAs interact with proteins via specific sequences or structures. Although some bioactive ligands, such as those for MALAT1,^132−137 bind to unusual RNA structures that form functional shapes, RNA degraders could have broad applicability for these targets.^138−140

Figure 6.

Numerous studies have identified functional RNA structures that can be modulated by small molecule binding. For instance, the translation of a specific mRNA can be diminished by a small molecule^174,175 that binds to a structure in the 5′ untranslated region (5′ UTR) near the start codon.²³ This binding thermodynamically stabilizes the mRNA structure, thereby limiting the assembly of actively translating ribosomes on the mRNA. Small molecules can also bind RNA structures to influence pre-mRNA splicing. By preventing U1 snRNP binding, they can decrease exon inclusion in the mature mRNA, altering the resulting protein’s identity,³⁵ while compounds stabilizing RNA-protein complexes can facilitate exon inclusion.²⁰ Additionally, small molecules can inhibit the formation of RNA-protein complexes. In RNA gain-of-function diseases caused by toxic RNA-protein complexes, small molecules can displace the protein, restoring its normal function.^46,76 Other functional sites are harbored in nuclease processing sites in microRNA precursors^{112,176−178} and in highly structured regions in long noncoding RNAs.^132,135,140

Another approach to drug “undruggable” proteins is inhibiting translation of the encoding mRNAs. Small molecules that bind RNA structures near translation initiation sites can stabilize these regions, preventing the ribosome from unwinding RNA and reducing translation, as demonstrated for α-synuclein.²³ Additionally, small molecules can bind pre-mRNAs to alter splicing outcomes, offering a novel method to control gene expression.^20,35

Compounds Modulating RNA Splicing

Compounds that modulate RNA splicing offer broad therapeutic potential, either as RNA-only binders or by stabilizing RNA-protein complexes. Specific RNA structures can influence exon inclusion or exclusion during splicing, as mentioned above for the MAPT splicing regulatory element, the destabilization of which causes FTDP-17. The 4R Tau protein produced by aberrant splicing is aggregation-prone, contributing to disease pathology. Interestingly, these aggregates are also biomarkers for Alzheimer’s disease. Small molecules have been identified that bind the MAPT splicing regulatory element—both mutant (FTDP-17) and wild-type (Alzheimer’s disease) forms—and reduce exon 10 inclusion and decrease 4R Tau levels. These compounds have emerged from a variety of methodologies, including aminoglycoside screening and structure-based design. Studies using nuclear magnetic resonance (NMR) spectrometry-restrained structures, pharmacophore modeling, and compound screening have helped advance these compounds.^{36,37,40,141−144}

Other splicing modulators target RNA repeat expansions, which are often associated with intronic repeats in various disorders.¹⁴⁵ Intron retention has been observed in GC-rich repeat expansions, for example in C9orf72-associated ALS/FTD [r(G₄C₂)^exp], myotonic dystrophy type 2 (DM2; r(CCUG)^exp) and Fuchs’ endothelial corneal dystrophy (FECD; r(CUG)^exp). Studies have shown that specific proteins bind to these intronic repeat expansions, and inhibiting these interactions can facilitate splicing, enabling the spliceosome to recognize the intron for decay via the nuclear RNA exosome.^47,114

Risdiplam, a splicing modulator, represents a significant advancement in the field of RNA-targeted small molecules. Identified through phenotypic screening, risdiplam enhances inclusion of an exon in SMN2 pre-mRNA as a treatment for spinal muscular atrophy (SMA) patients.^20,21 In particular, risdiplam stabilizes RNA-protein complexes to promote SMN2 exon 7 inclusion, resulting in a more stable protein that compensates for loss of its related homologue SMN1, the cause of SMA. Branaplam, developed by Novartis, also increases SMN2 exon 7 inclusion by a similar mechanism.¹⁴⁶ Risdiplam garnered FDA approval as a first-in-class oral medication to treat patients that have SMA, transforming their lives. These studies may suggest that other compounds identified by phenotypic screening could have modes of unanticipated or even mixed modes of action that affect RNA processing for therapeutic benefit. Interestingly, risdiplam, branaplam, and related compounds affect the pre-mRNA splicing of other transcripts that is suggestive of other applications of these compounds and their series.^147,148 However, care must be taken to ensure selectivity that includes, as has been described, RNA-seq analysis.²⁰ For example, in clinical trials, branaplam lowered Huntington’s protein levels by targeting a poison exon in the huntingtin (HTT) pre-mRNA, triggering nonsense-mediated decay (NMD). These and other compounds have shown to have similar effects, and these findings highlight the potential of splicing modulators as therapeutic agents for a wide range of diseases, offering a precise way to influence RNA splicing for therapeutic benefit.^147,148

RNA: An On-target and Off-target for Small Molecules

Historically, RNA has been underappreciated as both an on-target and off-target for small molecules. Previously, we highlighted areas where RNA-targeting small molecules have been overlooked.¹⁴⁹ For instance, it is well established that protein-binding small molecule drugs, such as kinase inhibitors, can also bind to and modulate RNA in addition to their protein targets.^150,151 Similarly, topoisomerase inhibitors¹⁵⁰ have been shown to target RNA, while other compounds possess protein-binding and drug-like features that bind RNA.¹⁵² These observations underscore a broader potential: phenotypic screens, which often yield bioactive molecules with unknown modes of action, may be identifying compounds that target RNAs, thus eliciting phenotypic responses. This hypothesis is further supported by examples such as SMN2(20,21) and HTT(147,148) pre-mRNA splicing modulators, which bind to RNA-protein interfaces and stabilize these complexes.

Despite these findings, RNA is rarely considered an off-target in drug development. Current clinical and preclinical safety assessments rely heavily on safety panels for studying protein binding and inhibition of enzymatic activity as a first-line evaluation.^153−155 However, with the advent of fast and cost-effective RNA-seq technologies, it could be transformative to study compound effects on RNA levels and metabolism in exposed tissues. Collectively, this raises a provocative possibility: RNA may represent an off-target liability contributing to compound toxicity. By identifying RNAs modulated by drugs and employing techniques like Chem-CLIP to interrogate RNA-compound interactions, we could gain valuable insight into such phenomenon. Such approaches may pave the way for more effective and safer drug development strategies (Figure 8).

Figure 8.

For RNA to fully realize its potential as a target for small-molecule drugs, several foundational and translational challenges need to be addressed. Structure-based drug design for RNA presents unique difficulties due to factors such as limited structural data on RNA-small molecule complexes, insufficiently refined^101,115,119 fields for modeling,^171,172 and the need to leverage RNA dynamics to optimize compounds^106,107 and establish structure–activity relationships. Additionally, little is known about the RNA targets of small molecules within cells. Unbiased approaches, such as Chem-CLIP¹⁶⁶ and high throughput compound screening, can provide valuable insights. These methods could identify ideal RNA target types for small molecules and inform other key factors influencing RNA-drug interactions. Many existing medicines are known to bind RNA targets, and phenotypic screens have likely identified bioactive small molecules that interact with RNA. However, due to a lack of comprehensive tools and knowledge, the full spectrum of RNA-binding targets remains poorly understood. Moreover, the observation that some protein-targeting medicines also bind RNA underscores the importance of profiling RNA interactions as both on-target and off-target effects during preclinical and clinical drug development.^149−152

Augmenting the Activity of Binding Compounds: Targeted RNA Degradation

Many small molecules bind to RNA targets but fail to elicit a biological response (Figure 7). These compounds can be converted into bioactive molecules by promoting targeted RNA degradation. Several approaches have been developed to induce RNA decay, showing activity in cells and transgenic mouse models. One such strategy involves the use of direct degraders.¹²⁴ In this approach, compounds that bind to a target RNA are appended with molecules that directly facilitate RNA degradation. In an early example, an RNA-binding compound was attached to a group that upon photolysis generates reactive oxygen species that cleaved the RNA target in cells.¹⁵⁶ In another example, an analog of the DNA-cleaving agent bleomycin was appended to an RNA binder in a such a manner that disabled its ability to cleave DNA.^46,49 Conjugation of bleomycin A5 to a high affinity binder of r(CUG)^exp (DM1) afforded selective elimination of the repeat expansion with no effect on transcripts with short, nonpathological repeats. In contrast, a repeat-targeting oligonucleotides degraded all transcript with short or long r(CUG) repeats. Importantly, the small molecule-bleomycin conjugate improved DM1-associated defects in patient-derived myotubes and a transgenic mouse model, as demonstrated by RNA sequencing and phenotypic analysis.⁴⁹ Other direct degraders incorporate imidazole moieties that can deprotonate 2′-hydroxyl groups, facilitating RNA decay via intramolecular attack on the phosphodiester backbone.^157−159

Figure 7.

Strategies to induce RNA degradation by small molecules. Heterobifunctional small molecules can directly cleave RNA targets^{49,124,157−159} or mediate cleavage by recruiting a ribonuclease.¹⁶¹ Small molecules that influence pre-mRNA splicing can also promote targeted RNA degradation. In diseases caused by intronic repeat expansions, the repeats are often retained in introns due to protein binding. Small molecules that bind to the repeats can displace these proteins, leading to liberation of the retained intron and facilitating nuclear RNA exosome-mediated degradation.^47,114,179 Additionally, compounds identified through phenotypic screening can induce the inclusion of a pseudo exon containing a premature termination codon in the mature mRNA, resulting in nonsense-mediated decay (NMD) of the mRNA, effectively reducing the levels of the pathogenic transcript.^147,148

In addition to direct degraders, the recruitment of decay functions has also advanced heterobifunctional compounds in which one end binds to the target RNA while the other end binds to an RNA effector enzyme.¹⁶⁰ This strategy, known as ribonuclease targeting chimeras (RiboTACs),^111,161 has been applied to various RNAs, including cancer-causing ncRNAs¹⁶² and coding RNAs¹⁶³ and a repeat expansion in an ALS transgenic mouse brain.¹¹³ RiboTACs have also been applied to various infectious disease settings by either targeting a viral RNA structure¹¹⁸ or by having a nucleotide analogue incorporate an RNase L recruiter.¹⁶⁴ This approach has been shown to augment the activity of inactive small molecules by inducing degradation and hence a functional outcome.^24,101 There are numerous ways to expand this approach, including the recruitment of additional enzymes.¹⁶⁰

Other approaches to augment binders cellular activity include covalent compounds,¹¹⁵ the addition of molecules that can produce reactive species upon photolysis to cleave an RNA target,¹⁵⁶ or to change a sequence by conversion of guanosine to 8-oxoG.¹⁶⁵

Ways to Accelerate the Druggable Transcriptome Project

RNA-targeted small molecule medicines hold great potential for transforming human health, but their development faces challenges (Figure 8). A major hurdle is identifying the specific RNAs that small molecules interact with in cells, as understanding these interactions is crucial for minimizing risks and unlocking new therapeutic opportunities. To address this challenge, direct techniques such as Chem-CLIP to map in an unbiased way the RNAs or RNA-protein complexes that bind small molecule ligands and the sites of binding have been developed. By using these data to infer RNA structures that bind ligands, RNA targets are indeed druggable and form pockets that small molecules ligands can interact with can be elucidated. Indeed, using Chem-CLIP in conjunction with RNA secondary structure and 3D structure prediction is emerging to address these challenges.^101,119,166

In conjunction with binding site analysis, data that define how to affect the biology of an RNA target are critical, not only for developing small molecule medicines but also for interpreting phenotypic screens. For example, studies have shown that targeting “functional” sites provides a bioactive response for a binder.¹¹² Emerging data support that many, if not most, of the binding sites of RNA targets in cells do not elicit a biological response, for example inhibition of translation, or affecting the function or maturation of a ncRNAs.

Screening RNA targets, whether for direct binding to small molecules or phenotypically, often yields very low hit rates. This often confounds defining SAR from the compounds that emerge, which limits confidence to move forward to hit-to-lead optimization campaigns. RNA- focused chemical matter can help fill some of these voids. For example, enriching screening libraries in chemotypes that confer RNA binding can enhance the hit rates, providing more confidence in the data set and enabling data-driven decisions. Although there have been advances in defining RNA-focused chemical matter,^{84,87,167−169} more work in this area is needed.¹⁷⁰

Bioactivity can be conferred to inactive binding interactions by recruiting a ribonuclease to a specific RNA to effect targeted degradation.^24,101 RNA-seq and functional studies in which the ribonuclease has been knocked out have identified which bound RNA targets can also be degraded. These unbiased approaches have begun to identify factors that affect targeted degradation, although many more studies are required to fully understand them including acquiring data sets in more cells lines and recruiting other effector proteins.¹⁰¹

Once effective interactions are identified, one would like to affinity mature RNA binding compounds. A key consideration in RNA-targeted drug development, however, is that the computational force fields traditionally used for protein-targeting might not be as suitable for RNA.^103,171,172 As a result, gathering more comprehensive data on RNA-small molecule interactions is crucial to establish a solid foundation for these studies. These data will also be critical for understanding SAR of RNA-targeting ligands. Since some SAR patterns for RNA targets can be more nuanced or flat, identifying chemical entities with varied binding patterns and exploring ways to influence RNA biochemically are vital for developing effective therapies. Additionally, data on RNA binding could play a critical role in computational methods, including sequence- or structure-based design²⁵ coupled to machine learning and artificial intelligence.

While much remains to be explored, there is considerable potential in RNA-targeted small molecules to transform medicine. In drug discovery and chemical biology, not all approaches will be suitable for every system; thus, it is crucial to prioritize the most relevant targets and indications. For example, diseases such as DM1, which are driven by repeat-containing RNA transcripts, could serve as key testing grounds for novel therapeutic strategies. This RNA target-centered approach will not only advance drug development for specific diseases but also contribute to the broader understanding of molecular recognition of RNA by targeted medicines, benefiting patients in need while expanding the tools available to researchers.

The author declares the following competing financial interest(s): I am a founder of Expansion Therapeutics and Ribonaut Therapeutics.