“Human ribozymes as novel targets of cancer therapy”

Ribozymes are non-protein-coding functional RNAs that catalyze a variety of chemical reactions. Self-cleaving ribozymes form the most diverse known group, which includes evolutionarily conserved examples in the human genome. The most extensively studied example is a highly active ribozyme that maps to an intron of the mammalian cytoplasmic polyadenylation element binding protein 3 (CPEB3) gene. CPEBs are RNA-binding proteins that regulate mRNA polyadenylation and translation, thereby altering cell behavior. Among these, CPEB3 is strongly downregulated across multiple cancer types and is increasingly recognized as a tumor suppressor and a potential prognostic biomarker. The CPEB3 gene is uniquely cis-regulated by the self-cleaving ribozyme, with established role in episodic memory, but the role of this ribozyme in cancer and other diseases with strong CPEB3 phenotype remains to be elucidated. Given that CPEB3 acts as a tumor suppressor and that its downregulation promotes cancer progression, we hypothesize that the ribozyme regulates CPEB3 mRNA maturation in several cancers, thereby modulating CPEB3 protein levels and altering tumor phenotypes. We aim to develop drugs that target the CPEB3 and other human ribozymes as an unexplored class of non-protein-coding RNA targets.

“Readthrough of disease-causing nonsense mutations by targeted selenocysteine recoding”

Nonsense (stop) mutations comprise about 10% of disease-causing mutations, and are common in Duchenne muscular dystrophy, cystic fibrosis, and other diseases. These mutations cause early termination of translation and lead to non-functional proteins. Therapies that enable readthrough of these premature termination codons (PTCs) have been sought for many years with limited success. One existing approach is based on small readthrough-promoting molecules. Despite its ease of delivery, this approach lacks specificity as it affects termination codons in general and has suffered from high toxicity and insufficient activity at safe concentrations. Other more targeted approaches for suppressing PTC mutations include CRISPR editing, recruiting endogenous ADAR for adenosine-to-inosine editing at the PTC, recruiting the endogenous pseudouridylation machinery to edit the uridine at the PTC, and exon-skipping antisense oligonucleotides. However, these approaches typically exhibit low efficiency, are limited in applicability to only one or few mutations, and/or require difficult delivery of genes. In summary, there is an urgent unmet clinical need to develop effective, safe, and deliverable readthrough therapies.

Our solution is to exploit the endogenous mechanism for opal termination codon (UGA) recoding by selenocysteine (Sec) incorporation. This endogenous recoding mechanism is activated by a specific structured element known as a selenocysteine insertion sequence (SECIS) element. We propose a novel endogenous RNA-targeting technology that uses short oligonucleotides to recruit or assemble SECIS signals on disease transcripts, thereby reprogramming premature UGA stop codons and leading to productive full-length translation. Our rationale is based on the observations that (1) SECIS elements can stimulate selenocysteine incorporation even when far from the UGA codon or in trans, (2) some viruses recruit SECIS elements from endogenous mRNA molecules through hybridization, leading to readthrough of opal termination codons, and (3) hybridizing therapeutic oligonucleotides are effective in recruiting other endogenous machineries.

Advantages of our approach include (1) off-target effects are less likely than in gene-editing, RNA-editing, or small molecule readthrough approaches; (2) oligonucleotides designed using our approach are mostly gene-specific, and not mutation-specific as with other targeted approaches, so a single oligonucleotide could potentially be effective across a broad spectrum of nonsense mutations; and (3) the use of short (~50 nt) oligonucleotides simplifies delivery and makes the approach more directly translatable to the clinic. In summary, if successful, the solution will lead to targeted, effective, and deliverable nonsense-suppression therapies.

We will use a high-throughput platform developed in our lab to screen 12,000 short RNA oligonucleotides spanning a range of design strategies and identify those inducing efficient Sec-recoding. We will initially demonstrate the solution in cystic fibrosis, focusing on clinically important CFTR UGA nonsense mutations such as G542X and W1282X and then use the resulting design rules to evaluate extension to other recessive or X-linked loss-of-function diseases.

Aligned with the TRDNT challenge, our solution targets endogenous RNA and if successful, could provide a broadly applicable path toward treating genetic diseases caused by nonsense mutations.

“Splice-switching ASO activity at scale: a predictive platform for oligonucleotide design across diseases”

Splice-switching antisense oligonucleotides (ASOs) are a clinically validated way to treat disease by redirecting how a pre-mRNA is spliced: correcting the aberrant splicing caused by a pathogenic variant, reframing a frameshift by inducing exon skipping, or raising a gene's functional output by blocking a naturally occurring non-productive splicing event. Because they act on splicing regulatory logic that is shared across genes, and because an ASO sequence follows directly from the site it must bind, they are uniquely suited to fast, individualized therapy. Therapeutic candidates can be designed as soon as a patient's causal variant is identified, positioning splice-switching ASOs as a viable treatment option for a large and diverse set of splicing-tractable diseases, both common and ultra-rare, individualized conditions.

The central bottleneck to realizing the full potential of ASO therapeutics is design. Today, finding an effective ASO is empirical: dozens to hundreds of oligonucleotides are synthesized and tested for each new gene, an expensive screen repeated from scratch every time. Existing sequence-based models for predicting splicing such as SpliceAI and Pangolin cannot substitute, because they have no way of representing an oligonucleotide. A common workaround is to delete or mask the sequence that is input into the program in order to artificially represent ASO blocking the sequence. But this approach generates results that correlate only moderately with experimentally validated ASO binding activity.

We propose to generate a predictive platform that removes the per-gene cost of experimental ASO screening and addresses the shortcomings of current best options for in silico screening, braiding two innovations. First, a massively parallel reporter assay decouples the number of ASOs synthesized from the number of measurements obtained. Each oligonucleotide is tested on in the context of thousands of barcoded sequences across a dose series, yielding tens of thousands of quantitative, dose-resolved activity measurements at the synthesis cost of a single conventional walk. Second, an occupancy-aware splicing model represents an ASO as graded occupancy sitting on top of intact sequence, preserving the competitive context an ASO must displace. The model is built as a lightweight adapter on the open-weight Pangolin model so the regulatory logic it already learned is retained. Phase I delivers on both of these innovations by first establishing a working pilot assay, the analysis pipeline that turns its sequencing reads into per-construct dose responses. Second, it delivers a first occupancy-aware model benchmarked against the sequence-editing baseline. Trained at scale, the platform would propose a credible ASO candidate by computation rather than an empirical bench screen, generalizing across the whole class of splicing-treatable diseases. On completion, model weights and training data will be released openly.

“Endogenous Protein-Bound Transcriptome Profiling for RNA-Targeted Therapeutic Discovery”

Altered interactions between RNA and RNA-binding proteins can drive disease by impacting splicing, stability, localization, translation, and ribonucleoprotein complex assembly, often without changing RNA or protein abundance. These regulatory states may create therapeutically actionable RNA targets but are largely invisible to conventional transcriptomics and proteomics. Existing protein-centric assays require an RBP to be selected in advance, whereas RNA structure and accessibility assays provide only indirect evidence of protein occupancy.

Ribolytix proposes a label-free technology for profiling the protein-bound transcriptome, comprising two modules. After native RNA–protein contacts are preserved and ribonucleoprotein material is enriched, the first module would generate NGS-compatible protein-bound RNA fragments while maintaining RNA-bound protein solubility required for downstream phase partitioning. The second module would selectively recover protein-bound RNA fragments and deplete protein-free RNA fragments. Sequencing the retained RNA would produce a regional map of protein occupancy on coding and noncoding transcripts without engineered protein tags, metabolic RNA labeling, antibody preselection, or reliance on poly(A) selection. Paired RNA-bound proteomics would characterize broader changes in the RBPome. Protein identities at individual RNA regions would be confirmed with targeted orthogonal assays.

The technology would advance RNA-targeted therapeutics in two connected ways. First, it could discover endogenous RNA regions whose protein occupancy changes in disease, prioritizing sites for RNA-directed intervention even when conventional transcriptomics and proteomics appear unchanged. Second, the same readout could measure whether a therapeutic restores the intended interaction state and whether it causes collateral remodeling elsewhere in the transcriptome. This creates a bridge from target discovery to pharmacodynamic assessment.

The approach is designed to transfer across diseases and therapeutic modalities rather than depend on a single target. In repeat-expansion and splicing disorders, it could reveal pathogenic redistribution of regulatory proteins and response to RNA-targeted treatment. In neurological disease, cancer, and inflammatory disease, it could identify condition-specific RNA regulatory regions. For engineered mRNA medicines, it could help determine which sequence or chemistry choices promote favorable protein-association patterns, stability, and translation in defined cell types. For RNA-editing programs, it could assess intended and collateral occupancy changes alongside editing efficiency.

Phase II would establish whether the two modules can be integrated into a reproducible prototype and demonstrate the resulting regional readout in a disease-relevant model. If successful, the platform would create a reusable protein-occupancy data layer that can be integrated with transcriptomic, proteomic, sequence, structural, and functional readouts to generate testable mechanisms across disease models and therapeutic modalities. This could support broadly applicable endogenous RNA-target discovery, therapeutic design, patient-relevant pharmacology, and development of standardized reagent kits for protein-bound transcriptome profiling.

“From map to molecule: nominating druggable splice sites genome-wide”

Splice-switching small molecules are among the most clinically validated RNA-targeting therapeutics. Risdiplam, an FDA-approved drug, treats spinal muscular atrophy by redirecting a single splicing event to increase production of functional SMN protein from the SMN2 gene, turning a fatal disease into a manageable one. A similar therapeutic logic could be applied to many other disorders by modulating gene expression levels through controlled alterations in splicing decisions. Antisense oligonucleotides can achieve related outcomes in cell culture, but the delivery of synthetic nucleic acids to many organs in vivo remains a major limitation. Recently, we developed CP3, a first-in-class small-molecule splicing modulator that lowers prion protein in vivo, establishing a pharmacological strategy to target PRNP, a classically undruggable driver of prion disease. Our data further demonstrate the feasibility of reprogramming risdiplam’s splice-site selectivity to target diverse undruggable disease-modifying genes. However, the principles governing drug-induced splicing selectivity remain poorly understood: these molecules act on a small subset of splice sites, and no one can predict which sites are worth targeting or which cause off-target toxicity. We propose an interpretable engine that nominates, genome-wide, the introns and genes whose splicing can be therapeutically regulated. Our approach rests on a mechanistic decomposition: selectivity is the product of two factors. The first is binding activity (A), how well a molecule stabilizes spliceosome recruitment at a 5' splice site. A is set by the local sequence and is already partly predictable. The second is splicing kinetics (K), the rate at which the competing splice sites nearby are used. A modulator works only by strengthening a target site until it outcompetes its neighbors, so K sets whether that competition is winnable. A site with the right motif but the wrong kinetics does not respond. Existing models read local sequence alone: they capture A but miss K, and so predict the binding motif, not the selectivity. Our proposed approach captures both. It integrates three components: an interpretable model that learns the splicing grammar and exposes it as readable motifs; genome-wide measurements of splicing kinetics from CoLa-seq and nascent RNA sequencing, methods we helped develop or analyzed extensively; and experimental splicing responses to modulators, which supervise training. Our model will learn the joint A and K rules of modulability, score every intron, and return a ranked, interpretable atlas of regulatable events annotated with the mechanism behind each call. Our model nominates targets across neurodevelopmental, neurodegenerative, oncological, and rare Mendelian disease, because it learns the rules of modulability rather than the biology of any one disorder. Each nomination carries its rationale, so the output is a testable hypothesis we will pursue in later phases. By shortening the path from target to candidate, the engine turns splice-switching discovery from serendipity into prediction and extends a proven modality far beyond its current reach.

“Programmable Ribonucleases for Sequence-Specific RNA Degradation”

Nearly all approved drugs act on proteins encoded by a small fraction of the genome. This narrow focus leaves most of the transcribed genome and many disease-driving proteins that lack a druggable pocket beyond reach. We are developing a programmable enzymatic technology that catalytically destroys a user-specified RNA inside human cells. The platform is built on a human enzyme that has been re-engineered so that its RNA-cleaving activity is directed to a chosen target sequence by an interchangeable synthetic recognition module. Retargeting the technology to a new disease requires only swapping that module, making the platform a plug-and-play “suite” rather than a one-off drug.

This approach advances RNA-targeted therapeutics on four fronts. It is catalytic: a single enzyme molecule degrades many copies of its target, unlike stoichiometric oligonucleotides. It is human-derived, minimizing the immunogenicity associated with bacterial or engineered protein machines. It is engineered for stringent target selectivity, suppressing the off-target cleavage that limits other approaches. And it is delivered directly into cells as a protein, without viral vectors or lipid nanoparticles, simplifying manufacturing and lowering cost. Crucially, because the platform is built on a human enzyme and synthetic chemistry rather than a bacterial nuclease, it stakes out an unencumbered U.S. intellectual-property position in RNA therapeutics, sidestepping the globally contested CRISPR patent landscape and helping secure U.S. leadership in the next generation of programmable medicines.

Because targeting is programmed by simple sequence-recognition rules, the same scaffold can be aimed at fundamentally different classes of RNA across many diseases: regulatory microRNAs, the messenger RNAs of “undruggable” oncogenes, and toxic repeat-expansion transcripts. Preliminary experiments demonstrate catalytic, sequence-directed degradation of a disease-associated RNA in vitro. In Phase I, we define the technical concept, two mechanistically distinct disease use cases, quantitative validation metrics, and a roadblock mitigation plan to reach a working prototype in Phase II.

“Spatial RNA Reprogramming: A Delivery-Ready Platform to Correct Pathogenic RNA Mislocalization”

Many diseases arise not only from changes in RNA sequence or abundance, but also from failure of endogenous RNAs to reach the subcellular compartments where they function. This problem is especially important in neurons, where messenger and regulatory RNAs must travel long distances to support synaptic activity, axonal maintenance, and tissue repair. Pathogenic RNA mislocalization is implicated in amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD), retinal and neurodevelopmental disorders, and acute neural injury, yet no therapeutic platform directly corrects RNA position.

We propose a new spatial RNA reprogramming technology that will redirect selected endogenous RNAs to defined subcellular compartments without altering their genomic sequence. The system will combine programmable RNA recognition with compact localization effectors in a self-contained, inducer-free architecture suitable for therapeutic delivery. Changing only the guide sequence will allow the same platform to reposition different disease-relevant RNAs, creating a general framework rather than a therapy restricted to one transcript or disorder.

Published CRISPR-Transcript Organization (CRISPR-TO) studies established that endogenous RNA localization can be programmed and that RNA position can control local translation and neuronal function. Building on this foundation, the proposed new technology will eliminate the need for chemical dimerizers, reduce cargo size and complexity, and create a clinically relevant, delivery-compatible platform. Preliminary studies further support feasibility: CRISPR-TO moves endogenous RNAs over long distances in primary neurons; candidate lipid nanoparticles deliver mRNA to human iPSC-derived neurons and mouse brain; and transient mislocalization of Actb RNA in mouse retina produces persistent molecular, synaptic, and physiological effects.

ALS will serve as the initial disease use case. In human TDP-43 deficient neurons, we will test whether restoring the localization of selected endogenous RNAs can rescue neurite growth, synaptic function, or cell survival. The Phase II prototype will establish quantitative benchmarks for relocalization, specificity, safety, delivery, and functional rescue. The platform will be disseminated through public research reagents, guide-design resources, standardized protocols, and benchmarking datasets. This work introduces RNA location as a therapeutically programmable dimension alongside RNA sequence, abundance, stability, and splicing.

“Precision RNA Therapeutics via Pathogenic RNA-Triggered Condensation”

Biomolecular condensates are dynamic, membraneless compartments that arise through phase separation of RNAs and proteins. They regulate processes such as gene expression and stress responses by enriching specific RNAs and proteins while excluding others. RNA is not merely a passive constituent of these assemblies: its sequence, structure, and multivalent interactions with other RNAs and RNA-binding proteins (RBPs) can nucleate condensates, define their molecular composition, and tune their material properties and functions. Leveraging this intrinsic programmability enables the creation of synthetic, genetically encodable RNA condensates to precisely modulate RNA localization, stability, translation, and disease-relevant interaction networks. In this framework, engineered RNA scaffolds can be designed to recruit defined client molecules, including endogenous RNAs, RBPs, ribosomes, enzymes, and small molecules, thereby controlling reaction kinetics, molecular access, and interaction networks within a compartmentalized microenvironment.

In this project, we posit that RNA-driven condensation can serve as a therapeutic regulatory layer for pathogenic RNAs by enabling spatiotemporal reorganization of RNA-associated biochemistry in living systems. For endogenous RNA regulation, engineered condensation can operate as a selective “capture-and-control” mechanism. Sequestration of a target RNA into a condensate could reduce its translation, alter its localization, change exposure to modification enzymes, or shift its interaction network with RBPs and decay machinery, ultimately tuning stability and gene output. Conversely, condensates can be engineered to enhance function by co-recruiting ribosomes or translation factors, or by enriching specific enzymes to promote RNA editing, cleavage, or modification. Our approach builds on our recent discovery that RNA nanostar motifs function as modular, genetically encodable scaffolds for RNA-driven condensate formation in living mammalian cells. Through systematic screening of RNA architectures, we found that circularized RNAs derived from de novo nanostar junctions robustly self-assemble into intracellular condensates. This platform addresses a critical unmet need for generalizable RNA-based condensate engineering in mammalian cells with dynamic control and RNA-centric functionality, complementing and extending existing, largely protein-focused systems.

We will develop an endogenous pathogenic RNA–driven condensation platform based on engineered RNA nanostar scaffolds that robustly form condensates in mammalian cells and can be programmed for RNA-input–dependent assembly via a structure-switching mechanism. We will extend this beyond a single proof-of-concept by establishing general design rules for RNA-triggered condensation across multiple targets and cellular contexts. Specifically, we will systematically determine how condensate activation depends on target RNA length, binding architecture and site placement, and endogenous target abundance. We will quantify performance in vitro and in HEK293T cells through target titrations to define sensitivity and to map target-length and binding-site dependencies for each scaffold configuration. Because RNA expression profiles vary across cell types, we will further test whether endogenous RNA abundance enables cell-type-selective condensate formation in cell lines with divergent target expression. Meanwhile, we will evaluate functional regulation of endogenous RNAs using therapeutically relevant miRNA pathways (miR-21, let-7) and selected disease-linked long noncoding RNAs (circPVT1, HOTAIR, NEAT1_1) as initial testbeds.

We will extend pathogenic RNA-triggered nanostar condensates from passive sequestration to active therapeutic control. RNA-responsive scaffolds will be engineered to both capture pathogenic transcripts and recruit catalytic effectors to enable RNA editing, cleavage, or modification within condensate microenvironment. Specifically, we will enrich an ADAR deaminase domain for targeted RNA editing using orthogonal RNA–protein recruitment tags (e.g., MS2/PP7) and incorporate antisense capture domains that establish programmable mismatches at defined edit sites. In parallel, we will localize Cas13 and its cognate crRNA to condensates to drive sequence-specific degradation of recruited transcripts. We will further recruit m6A writers (METTL3/14) and erasers (FTO/ALKBH5) to modulate RNA stability and translation. We will also demonstrate orthogonal multi-condensate regulation by independently controlling distinct RNA targets in parallel within single cells.

If successful, this project will establish a new class of technologies that enable cell-type-specific, context-dependent RNA therapeutics with enhanced precision and tunability. It will provide a generalizable strategy to sense disease-associated RNA states and selectively couple target recognition to programmable functional interventions. In addition, it will deliver scalable tools to interrogate pathogenic RNA interaction networks and to support discovery of new RNA disease biomarkers.

“Rebalancing the Glucocorticoid-SAA1 Axis: A Context-Specific RNA-Targeting Platform to Resolve the Paradox of Steroid Therapy in Inflammatory Disease”

Glucocorticoids are among the most widely prescribed drug classes in the world, used by 1-3% of the general population for conditions including rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), inflammatory bowel disease (IBD), asthma, vasculitis, nephrotic syndrome, cancer-related inflammation and administered to greater than 80% of women at risk of preterm birth between 24-34 weeks gestation. Yet, glucocorticoids carry a built-in molecular paradox: they suppress inflammation through NF-κB transrepression while simultaneously inducing serum amyloid A 1 (SAA1), a pathogenic effector that amplifies inflammatory tissue injury and serves as the precursor protein for AA amyloidosis, through direct binding of the glucocorticoid receptor (GR) to a glucocorticoid response element (GRE) at position -208/-194 in the SAA1 promoter. Despite the central role of SAA1 in this paradox, no adjunctive therapeutic strategy currently exists to suppress this pathogenic glucocorticoid-induced transcriptional output while preserving the beneficial anti-inflammatory actions of glucocorticoids.

We propose a context-specific RNA-targeting platform, the GRE-Responsive RNA Attenuation Platform (GRRAP), to resolve this paradox by attenuating GRE-driven SAA1 mRNA output while preserving the NF-κB transrepression arm of glucocorticoid anti-inflammatory action. GRRAP represents a new RNA-targeting strategy that selectively attenuates a pathogenic transcriptional output of glucocorticoid signaling while preserving its beneficial anti-inflammatory functions. The platform integrates three components: (1) an isoform-specific siRNA targeting SAA1 mRNA; (2) lipid nanoparticle delivery optimized for co-administration with glucocorticoid therapy; and (3) a GRE/NF-κB element-discriminating reporter system that independently quantifies GRE- and NF-κB-driven SAA1 transcription. Together, these components establish a framework for context-specific RNA targeting of genes regulated by integrated GRE- and NF-κB-dependent transcriptional mechanisms.

The technology will be disseminated as open-access tools applicable across RA, chorioamnionitis-associated preterm birth, SLE, IBD, cancer and other immune-mediated and inflammatory diseases.

“Novel RNA editing nanomachine for generation of tRNA-derived small RNAs”

Regulatory small RNAs constitute a network of molecules within the cell that serve to fine-tune cellular homeostasis and play a critical role in adaptation to stress signals. However, the majority of RNA therapeutics focus on targeting or replacing endogenous mRNAs, and the vast space of small regulatory RNAs has been largely unexploited. tRNA-derived small RNAs (tDRs), generated from the most abundant species of RNAs in the cell (tRNAs) by endogenous nucleases, have been shown to play important roles in critical cell adaptation pathways across many different cell-types and diseases. Notably, dysregulation of various tDRs have been described in diseases such as kidney disease, ischemic heart disease, diabetes, neurological disease and various cancers. However, a significant gap in our understanding of tDR biogenesis coupled with the extensive RNA modifications on the parent full-length tRNAs that are retained in the derived tDRs (and likely play an important role in their stability and function) have led to a significant roadblock in targeting tDRs for therapy. Recent seminal studies by our group (Li and Das) and others have shown the feasibility of delivery of synthetic RNA mimics for renoprotection, pancreatic beta-cell health and cardiovascular disease. However, synthetic mimics often lack the site-specific RNA modifications, differ between individual tDRs, are expensive to screen and deliver, and may require repeated administration.

To address this critical gap, the group assembled here with complementary expertise in tDR biology, preclinical models, RNA chemistry and machine-learning approaches to RNA structure propose an innovative bioengineered ‘nanomachine’ for the generation of endogenous tDRs to target a wide range of diseases. We propose hijacking the Cas13 RNA targeting nuclease (previously used for targeted degradation of specific mRNAs) to develop a programmable RNA-editing nanomachine to generate specific tDRs from endogenous full-length tRNAs using ML-optimized guide RNAs (gRNAs). This would constitute a platform for the biogenesis of any desired tDR that retains all the native modifications and may closely augment the ‘natural’ cellular stress response that may have been disrupted in the setting of disease. To derisk the prototype development phase of the TRDNT challenge, we have shown that: a) focusing on a single tDR (derived from Asp-GTC), full-length Cas13 can be used to target tRNAs despite their complex secondary structure to generate tDRs with variable efficiencies (depending on site of cleavage and sequence of the guide-RNA being used); b) Machine learning approaches can significantly aid in choosing the appropriate gRNA that can direct Cas13 specifically to the parent tRNA of choice; and c) a bioengineered Cas13 molecule where the HEPN nuclease domains have been replaced the nuclease domains of Angiogenin (the nuclease important for tDR biogenesis) can be used for generating tDRs (from Asp-GTC).

In the Phase I (ideation phase) of the challenge, we seek to develop our bio-engineered Cas13-Ang nanomachine as a generalizable platform that can be used to generate any tDR of choice by selecting an appropriate gRNA informed by machine learning approaches. We plan to test the efficiency of this platform in generating the expected tDR while minimizing off-target cleavage or degradation while optimizing the ideal delivery vehicle for this nanomachine (LNPs or AAV-mediated gene transfer) in readiness for testing in pre-clinical models in later phases of the challenge. Our highly innovative platform technology would be the first to open the door to a previously inaccessible class of regulatory molecules important across a wide range of diseases.

“COUPLET: A Paired-Adaptor Platform for Transcript-Specific Rescue of Transcripts Bearing Premature Termination Codons”

 A premature termination codon (PTC) creates two linked defects in one RNA. Translation stops before a complete protein is produced, and the transcript is often recognized by nonsense-mediated mRNA decay (NMD) and depleted before it can support repeated translation. Recoding cannot act on RNA that has already been cleared. Stabilization preserves substrate but leaves the premature stop intact. Existing approaches generally address one defect at a time and often do so globally, through broad suppression of NMD or transcriptome-wide stimulation of stop-codon readthrough.

COUPLET is a paired-adaptor platform that selectively stabilizes and recodes the same endogenous disease transcript. The stabilization adaptor, termed the shield, is an RNase H-inert steric-blocking oligonucleotide directed to a selected exon junction complex (EJC) deposition region downstream of the PTC. It reduces NMD licensing on the target RNA without inhibiting the surveillance pathway itself. The recoding adaptor is an engineered H/ACA guide RNA that positions the PTC uridine within a pseudouridylation pocket and recruits endogenous modification machinery, increasing the probability of productive decoding at that codon.

The proposed technology is the co-design system that converts transcript architecture into a ranked pair of target windows, a defined experimental plan, and a predicted rescue profile. COUPLET integrates stop identity, local sequence, exon-junction geometry, alternative splicing, RNA structure, RNA-binding-protein occupancy, sequence uniqueness, guide geometry, and protein-level feasibility. The initial engine is an interpretable scoring model with explicit eligibility gates and separate shield, recoder, protein-rescue, and pair-compatibility terms. Its rules will be calibrated on reporter and endogenous data, locked, and tested prospectively on alleles withheld from model development. This held-out evaluation is the central platform test.

Full-length protein output depends on how much mutant RNA remains available, how often the PTC is productively decoded, and whether the resulting protein is properly made and processed. The shield preserves the RNA substrate, while the recoder increases decoding at the stop codon. These effects can compound without requiring direct biochemical interaction between the two adaptors. Because pseudouridylation and decoding remain incomplete, many transcripts should still benefit from protection against NMD. COUPLET will also test whether extending transcript lifetime gives the guide more time to modify its target, further increasing productive decoding.

COL7A1 nonsense-mutant recessive dystrophic epidermolysis bullosa (RDEB) is the lead use case. Type VII collagen rescue can be followed from RNA stabilization through full-length protein production, secretion, basement-membrane deposition, and tissue integrity. The approximately 9-kb coding sequence exceeds conventional single-AAV capacity, and skin permits local delivery and direct sampling. CFTR W1282X provides a mechanistically distinct portability test in which transcript abundance, full-length protein, maturation, localization, and chloride conductance can be measured separately.

The minimum viable prototype is intentionally narrow: one authenticated COL7A1 nonsense context, one shield, one recoder, matched single-module and paired conditions, and direct measurement of target RNA, target-site pseudouridylation, and full-length collagen VII. The pair must produce more full-length protein than recoding alone at the same recoder exposure. The intended product is a public method for determining where paired stabilization and codon-directed recoding can restore functional protein from endogenous nonsense-mutant transcripts and where another therapeutic strategy should be favored.

“Self-amplifying miRNA, for mRNA Knockdown”

Targeting specific mRNAs with microRNAs (miRNAs) is a well-established and widely pursued approach for down-regulating proteins implicated in diseases that include a broad range of cancers, viral infections, cardiac pathologies, and metabolic and neurological disorders. Current micro-RNA treatments, however, suffer from issues of low dosage of initial delivery (in the case of “naked” miRNA delivery) or undesired persistence of knockdown effect (as with AAV and lentiviral vectors that involve episomal and chromosomal integration). To overcome these issues, we propose the use of novel self-amplifying RNA constructs – “replicons” – for transient delivery and in situ synthesis of large amounts of therapeutic miRNA. More explicitly, we design RNA molecules that are genetic fusions of a virus-derived RNA-dependent-RNA-polymerase (RdRp) and a core catalytic sequence (CCS) consisting of a miRNA sequence flanked by self-cleaving ribozymes that are active only in their negative sense. When these replicons are translated in cellulo their RdRp gene products amplify them into complementary strands whose active CCSs cleave out and release a large number of functional miRNAs, with subsequent downregulation of their target mRNA and protein. This is a totally new mechanism for which we want to provide “proof-of-principle” in Phase II cell culture studies using: a particular choice of RdRp (from the Nodamura virus, which works well in a wide variety of mammalian cells); different choices of miRNA (e.g., 34a whose up-regulation enhances the activity of the tumor suppressor protein p53, and let-7 that regulates cell differentiation); and different choices of self-cleaving (e.g., Hammerhead and Tetrahymena-group-I-intron) ribozymes active in their negative sense.

We describe in Phase I how replicons of this kind are strongly replicated in human cell lines and how they result in release of functional miRNAs that knock down a designed reporter mRNA into whose 3’ untranslated region the target sequence of a microRNA has been inserted. In Phase II we will extend these experiments to additional miRNAs and mammalian cell lines, and carry out the in vitro packaging of our replicons in stoichiometrically-precise, thermodynamically stable, virus-like particles for protecting, targeting, and delivering them in the mouse models pursued to test our strategy in Phase III, where we will work with a particular choice of xenograft that will be treated with the above-described miRNA-replicons and “interrogated” with a panel of cell-state, tumor growth, and metastasis assays. The Gelbart lab (UCLA Department of Chemistry & Biochemistry) has many years of experience and accumulated expertise synthesizing virus-like particles (VLPs) from purified plant virus capsid protein and in vitro transcribed RNA, and with conjugating them to targeting antibodies that direct them to specific cells in vivo. And the Guo lab UCLA Department of Biological Chemistry) has many years of experience and accumulated expertise with determining the structure and function of miRNAs and with the design of anti-sense oligonucleotides (ASOs) with enhanced mRNA affinity and specificity.

In Phase III we will be enlisting the collaboration of a former UCLA colleague of ours, presently Chair of Pathology at the Duke University School of Medicine, who is a prostate cancer expert specializing in new programmable RNA engineering platforms for immunotherapy. Specifically, we will test the in vivo efficacy of our self-amplifying miRNA reagents delivered in in vitro reconstituted virus-like particles, in comparison with a range of currently used vectors for non-self-amplifying miRNA. Our approach is promising for its ability to deliver a low dose of miRNA to a target cell which is then amplified in situ and maintained for days to ensure a strong but transient therapeutic effect, as opposed to involving permanent chromosomal or episomal integration and incurring oncogenic risk. Further, our platform is flexible in terms of the miRNA and ribozyme sequences included, allowing for a variety of mRNA targets in cellulo and disease targets in vivo.

“An Agentic Platform for Structure Mapping and De Novo Design of Gain-of-Function Activating Antisense Oligonucleotides for Ultra-Rare Haploinsufficiencies”

 Genetic disorders, particularly ultra-rare cohorts, represent an immense category of unmet medical need, often overlooked by pharma. Many of these severe phenotypes are caused by monogenic haploinsufficiencies, where a single mutated copy of a gene cannot produce sufficient functional protein to support physiological health. Because the mutations causing these phenotypes are often highly individualized, standard small-molecule screening libraries or broad-spectrum therapeutic modalities fail entirely. Development of therapeutics for haploinsufficiencies remains constrained by the necessity for a granular, case-specific comprehension of RNA-binding proteins (RBPs), RNA secondary structure, and transcript stability. Consequently, there is a critical requirement for advanced methodologies to accurately predict these regulatory regions.

Therna Biosciences is introducing our co-scientist agent built off of the RNA-Logix™ platform an autonomous computational-to-experimental engine developed for the rapid de novo design and optimization of sequence-specific, to design programmable gain-of-function, activating Antisense Oligonucleotides (ASOs). Unlike conventional approaches that rely on screening known microRNA sites or splice motifs, our platform unlocks novel biology by reasoning over complex, three-dimensional RNA structure, RNA binding protein interactions, and global splicing dynamics across the whole human transcriptome. Rather than pursuing targeted transcript degradation, the agent specializes in identification of novel and sterically blocking structural cis-repressive elements, upstream open reading frames (uORFs), and native intronic splice silencing regions on the patient's remaining healthy allele to increase its functional protein output. Through an independent review of literature, we have successfully validated the performance of our agentic platform demonstrating that the predicted regions correspond directly to experimentally confirmed RBP regions.

The critical differentiator of the co-scientist agent is its multi-parameter integration with RNA-Logix™, Therna Biosciences's comprehensive nucleotide intelligence architecture. The engine relies on massive, proprietary datasets evaluating structural context and physical target accessibility inside living cells:

• Regulatory Insight: Trained on an extensive in-house database of over 100 distinct RBPs and trans-regulatory complexes, mapping cross-species splice grammar to pinpoint inhibitory hotspots.
• RNA Structure: Fed by a massive dataset consisting of over 5 million in-cell dimethyl sulfate (DMS) mutational profiling readouts across human cell lines, assigning strict structural exposure scores to ensure physical ASO binding feasibility.
• Splice Modulation: Leveraging a proprietary long-read dataset of 100 million full-length mRNAs across diverse cell types to accurately simulate dynamic isoform variations.

Accelerated by the RNA-Logix™ platform, our agent offers an integrated blueprint that can automate transcript curation, and structure wet-lab testing loops to take a patient's raw sequencing data and generate verified, clinically viable therapeutic options in a matter of weeks. Our RNA-Logix platform possesses the capacity to redefine therapeutic development for rare genetic disorders. By prioritizing a computational-first strategy, we can significantly accelerate critical timelines and enhance cost-efficiency for traditional oligonucleotide gene tiling. This methodology allows for the high-throughput screening of dozens of optimized candidates initially, effectively replacing the need to empirically test hundreds of designs.

“A novel technology to profile and specify 5'p/3’t-2'OMe-modified sncRNAs of individual proteins for precision RNA-targeted therapies and predictors across multiple diseases”

Many life-threatening diseases, including cancer, neurodegeneration, and autoimmune disorders, are driven by proteins that are traditionally considered undruggable, as current therapeutic strategies often fail to target these proteins. Also, primary no-response and developed resistance to therapies targeting druggable proteins have long been observed in the clinic. There is a critical need for innovative technology to decode the RNA molecules governing protein functional behavior and unlock new therapeutic avenues across multiple diseases.

Our technology addresses these needs and challenges through the discovery and targeting of a previously underexplored class of regulatory molecules: 5′-phosphorylated and 3′-terminal 2′-O-methylated small non-coding RNAs (5′p/3′t-2′OMe sncRNAs). Unlike conventional regulatory sncRNAs, these molecules directly interact with target proteins and modulate their functional activity without altering mRNA or protein abundance. To systematically identify and characterize these molecules, we developed the 5′- and 3′-Ends Adaptor-Based Assay (EABA), a novel platform that profiles and specifies 5′p/3′t-2′OMe-modified sncRNAs associated with individual proteins across diverse cellular states and disease contexts.

EABA overcomes barriers of sncRNA deep Seq through optimized adaptor design, selective ligation chemistry, targeted amplification, and streamlined analysis, enabling accurate, rapid, and cost-effective identification of 5′p/3′t-2′OMe-modified sncRNAs. Compared with conventional small RNA sequencing approaches, EABA is faster, less expensive, more sensitive, and does not require specialized bioinformatics expertise. EABA achieves this through optimized innovations: (1) a 3'-end adaptor with an App- 5'-terminus to enhance ligation efficiency and a -ddC 3'-terminus to prevent self-ligation; (2) T4 RNA Ligase 2 truncated KQ specifically selected for 3't-2'OMemodified substrates; (3) a 3'-OH group in 5'-end adaptor ligates directly to the 5'p of target sncRNAs via T4 RNA Ligase 1; and (4) a gene-specific primer for reverse transcription that is complementary to the 3'-end adaptor and forms a perfect primer pair with the 5'-end adaptor. Neither adaptors nor primers align to any sequences in human, mouse, or rat genomes, ensuring cross-species applicability.

How EABA advances RNA-targeting approaches across multiple diseases.

• Targeting difficult-to-drug proteins. By identifying the specific sncRNAs that directly regulate functional behaviors of these proteins, EABA reveals upstream RNA-level controllers that can be therapeutically engaged through precision RNA-targeting strategies (refer to our TERT example).
• Overcoming primary non-response and resistance to protein-targeting therapies and developing novel predictors for treatment response. Novel molecules revealed by EABA provide a promising adjuvant agent to reduce primary non-response and overcome acquired resistance to protein-targeting therapies across multiple cancer types (refer to our PD-L1 example).
• Cross-disease and cross-species platform applicability. EABA's design is protein-agnostic and species-agnostic. It can profile 5'p/3't-2'OMe-modified sncRNAs of any protein (wild-type or mutant states, at high or low expression levels) across any disease modality in human, mouse, and rat systems, establishing applicability beyond oncology to genetic and neurodegenerative diseases rooted in RNA dysfunction.
• Broad dissemination. EABA has been developed within a CLIA/CAP-certified laboratory and follows LDT standards. It will be disseminated as a publicly available protocol, a validated research kit, and ultimately an FDA-cleared clinical kit, ensuring that the therapeutic targets it identifies are accessible to the broadest possible research and clinical community.

“RiboHelix: A Programmable lncRNA-Inspired Platform to Restore Protein Translation”

 RNA therapeutics are transforming modern medicine through RNA vaccines, antisense oligonucleotides, RNA interference, RNA editing, and mRNA-based technologies. However, most existing strategies suppress, degrade, edit, or replace RNA transcripts rather than restore translational output from endogenous mRNAs. This leaves a major unmet need for technologies that can increase protein production when native transcripts are poorly translated due to inhibitory RNA structures, altered RNA-binding protein interactions, or stress-associated ribonucleoprotein dynamics.

We propose RiboHelix, a modular RNA-guided translational remodeling platform inspired by natural sense–antisense regulation between mRNAs and long non-coding RNAs. RiboHelix is designed to recruit endogenous RNA helicase activity to structured regions of target mRNAs, particularly within 5' untranslated regions (5′UTRs), to remodel inhibitory RNA folds, improving ribosome access and restoring translation without genomic editing, transcript degradation, or exogenous protein delivery. This strategy directly addresses a largely unexplored therapeutic space: targeting RNA structure itself, rather than only RNA sequence, abundance, or coding capacity. Our approach is supported by preliminary evidence from the endogenous antisense lncRNA Emx2os and its neighboring protein-coding transcript EMX2. In this system, a compact Emx2os-derived regulatory module increases EMX2 expression, restores EMX2 protein under stress-associated repression, and alters EMX2 5′UTR accessibility. These data support the central premise that endogenous lncRNA-like regulatory mechanisms can be reduced to compact, engineerable RNA modules and adapted into programmable tools for transcript-selective translational regulation. The proposed platform contains two modular components: a helicase-recruiting RNA domain designed to engage endogenous RNA-remodeling activity, and a programmable targeting domain designed to recognize structured 5′UTRs through sequence-specific RNA–RNA interactions. Together, these modules are intended to localize RNA remodeling activity to disease-relevant transcripts and restore translational competence in a programmable manner.

In Phase I of Targeting RNA in Disease with Novel Technologies (TRDNT) Challenge, we will establish proof of concept through a focused development plan. First, we will optimize compact RNA scaffolds for stability and endogenous helicase recruitment. Second, we will add programmable targeting domains and validate engagement of structured 5′UTRs. Third, we will determine whether optimized constructs increase protein output without altering mRNA abundance. Fourth, we will evaluate specificity, off-target effects, and toxicity. While the initial proof-of-concept will focus on the Emx2os–EMX2 system, RiboHelix is designed as a broader platform for diseases driven by translational repression, RNA structural dysfunction, or impaired protein production. Potential applications include stress-related neuropsychiatric disorders, neurodegeneration, and cancer, and can be used to regulate mRNA and isoform expression and affect structured non-coding RNAs. Once established, RiboHelix would usher in a new era of RNA-targeting technology, expanding RNA medicine beyond gene silencing toward programmable restoration of gene expression and cellular function through RNA structural remodeling.

“From Fitness Landscapes to the Clinic: A Two-Tier Platform for Engineering Catalytic RNA Therapeutics”

Many major human diseases remain beyond the reach of conventional therapeutics because they are driven by pathogenic transcripts or by proteins that have proven difficult to drug. These diseases can, in principle, be addressed upstream at the level of the RNA itself. Catalytic RNAs, particularly trans-splicing ribozymes, are uniquely suited to this task. Rather than merely silencing a transcript, they recognize a disease-associated RNA and replace, repair, or augment it with single-nucleotide specificity. Yet unlike protein therapeutics, antibodies, and genome editors, catalytic RNAs have never benefited from a systematic engineering platform. Every new therapeutic ribozyme is still built essentially from scratch, one variant at a time, with no reliable way to navigate sequence space or to predict how a candidate will perform inside a human cell. This missing engineering discipline, not the underlying biology, is the central barrier to realizing the full potential of catalytic RNA therapeutics.

The platform. We propose the first integrated engineering platform for catalytic RNAs, combining two complementary technologies, Tier 1 and Tier 2, that optimize ribozyme function across biochemical and cellular fitness landscapes, respectively. Tier 1 is a high-throughput, barcoded combinatorial selection method (RABADOCS) that maps how mutations across the entire ribozyme body affect trans-splicing activity and target specificity under defined biochemical conditions, generating quantitative sequence-function relationships from very large libraries. Tier 2 is the directed evolution of ribozymes inside living human cells, where ribozyme performance is shaped by RNA-binding proteins, helicases, surveillance pathways, and other features of the intracellular environment that no in vitro system can reproduce, and where cooperative mutations can emerge under physiological selection. Neither approach alone is sufficient: the biochemical landscape provides scale and interpretability; the cellular landscape provides physiological relevance. Their integration forms a closed engineering cycle: generate, measure, evolve, validate. Variants that perform well across both landscapes constitute the highest-confidence therapeutic leads, while disagreements between them reveal the cellular determinants of catalytic RNA performance.

Impact across diseases. The platform is deliberately target-agnostic: because the engineering approach is independent of the specific transcript being targeted, the same framework applies to any disease in which a pathogenic transcript can be defined, including cancers driven by oncogenic driver and drug-resistance mutations, aberrant splice isoforms, oncogenic fusion transcripts, and repeat-expansion disorders. Beyond trans-splicing ribozymes, the engineering approach extends to other structured therapeutic RNAs, offering a general route from a disease associated RNA target to an optimized RNA therapeutic. By transforming catalytic RNA engineering from empirical optimization into a scalable, predictive discipline and by disseminating the resulting methods, datasets, and optimized catalytic RNAs to the community, this platform aims to accelerate the development of an entire new class of RNA-targeting medicines across many human diseases.

“A Non-Invasive Ultrasound-Gated Platform to Program Endogenous RNAs for Disease Treatment”

RNA-targeting therapeutics have emerged as a powerful approach for treating diseases driven by RNA dysfunction and difficult-to-drug protein targets. Technologies such as antisense oligonucleotides, RNA interference, and programmable RNA editors have greatly expanded our ability to manipulate gene expression without permanently altering the genome. However, despite these advances, current RNA-targeting interventions remain fundamentally limited in one critical dimension: control. Once delivered, therapeutic activity is typically constitutive, systemically distributed, and difficult to regulate with precision. As a result, most RNA therapies function as static interventions applied to highly dynamic biological systems.

This limitation is particularly significant for diseases characterized by spatially restricted and temporally evolving pathology, including neurological disorders, metabolic diseases, inflammatory conditions, and degenerative diseases. In these settings, therapeutic needs change over time, across anatomical regions, and throughout disease progression. An ideal RNA therapeutic would not only target the correct RNA molecule but would also provide precise control over when, where, and for how long RNA modulation occurs.

Our preliminary studies establish the feasibility of this approach through successful demonstration of inducible RNA knockdown, RNA editing, and RNA splicing regulation in vitro, together with spatially localized, repeatable, and reversible activation in vivo across multiple tissues, including brain, eye, and liver. These findings support the development of a prototype platform capable of non-invasive and programmable control of endogenous RNA in living organisms.

Accordingly, we propose a novel RNA-targeting technology platform that introduces programmable spatiotemporal control as a fundamental capability of RNA therapeutics. The platform integrates focused ultrasound (FUS), a clinically established non-invasive energy modality, with engineered RNA-targeting electors to enable externally controlled modulation of endogenous RNA in vivo. By converting acoustic energy into localized activation of RNA-regulatory machinery, the platform allows RNA-targeting activity to be initiated, repeated, tuned, and spatially confined without permanent genomic modification. To achieve this goal, we proposed 3 objectives: (1) build a FUS responsible RNA therapeutics platform (2) tune FUS-controllable RNA Regulators with programmable characteristics (3) apply and examine for representative therapeutic cases in vivo.

Unlike conventional RNA therapies, which primarily define molecular specificity, our technology adds a second layer of therapeutic programmability: control over therapeutic timing, duration, intensity, and anatomical location. This capability enables RNA modulation to be deployed only when needed, restricted to specific tissues, and reactivated through non-invasive ultrasound stimulation. The result is a new framework in which RNA therapeutics become dynamically controllable rather than constitutively active.

Importantly, the platform is modality-agnostic and supports multiple forms of endogenous RNA regulation through a common control architecture. Initial implementations include targeted RNA degradation, RNA base editing, and RNA splicing modulation, demonstrating that diverse RNA regulatory functions can be controlled using the same externally programmable system. This architecture is readily extensible to future RNA-targeting modalities, creating a broadly applicable technology platform rather than a single therapeutic product.

As such, the proposed technology addresses a major unmet need in the RNA therapeutics field: the inability to precisely control RNA-targeting activity in space and time. By introducing programmable spatiotemporal regulation as a core design principle, this platform establishes the technological foundation for a new class of spatiotemporal RNA therapeutics that can adapt to changing disease needs while maintaining the precision and reversibility inherent to RNA-based intervention.

“An all-in-one RNA trans-splicing platform for highly efficient and programmable exon editing”

We propose to develop a compact, all-in-one platform technology for programmable RNA exon editing based on engineered spliceosome-mediated trans-splicing. The technology uses an engineered trans-splicing activator (TSA) RNA to precisely replace one or more exons within an endogenous pre-mRNA without permanently modifying genomic DNA or disrupting native gene regulation. By enhancing the efficiency of spliceosome assembly and selectively promoting the engineered trans-splicing reaction over competing endogenous cis-splicing, this approach helps address a longstanding limitation that has hindered the practical application of RNA trans-splicing. The proposed technology is designed to overcome a major bottleneck in RNA-targeted therapeutics. Although trans-splicing offers the unique ability to replace large segments of RNA— including entire exons spanning several kilobases—its low efficiency has substantially limited both research and therapeutic applications. In the proposed studies, we will first establish the design principles underlying efficient trans-splicing by systematically optimizing key features of the TSA using a quantitative fluorescence-based reporter platform and multiple complementary molecular and cellular assays. As a therapeutic proof of concept, we will target exon 1 of the human HTT transcript, in which expansion of a CAG repeat causes Huntington’s disease. Optimized TSA designs will be delivered by AAV into patient-derived cellular models to evaluate endogenous correction of the mutant HTT transcript and restoration of normal transcript and protein. Successful completion of this project will establish a broadly applicable platform for efficient, programmable RNA exon replacement. If successful, the proposed technology will provide a programmable RNA-targeting framework for highly efficient and precise exon editing that holds the promise to treat a wide range of genetic disorders, thus expanding the list of druggable targets that remain difficult or inaccessible using existing genome- and RNA-editing technologies.

“An Integrated Platform for Discovery and Therapeutic Targeting of Chromatin-Regulatory Noncoding RNAs”

The human genome encodes hundreds of thousands of noncoding RNAs (ncRNAs), yet only a small fraction has defined biological functions or are considered therapeutically tractable. While RNA therapeutics have transformed the treatment of diseases caused by aberrant protein-coding transcripts, comparable strategies for targeting regulatory ncRNAs remain largely undeveloped. This gap is particularly striking for enhancer-associated RNAs (eRNAs) and long noncoding RNAs (lncRNAs), which are increasingly recognized as critical determinants of chromatin organization and gene expression. Emerging evidence indicates that these RNAs do not merely accompany transcription but frequently function as molecular scaffolds that recruit or regulate chromatin regulatory complexes, including ATP-dependent chromatin remodelers, histone-modifying enzymes, and transcriptional coactivators. Despite their central role in controlling cell identity and disease-associated transcriptional programs, there is currently no integrated strategy for identifying which ncRNAs are functionally essential, defining their mechanisms of action, or translating these discoveries into precision RNA therapeutics.

Our recent work has identified a previously unrecognized mechanism in which enhancer RNAs recruit the SWI/SNF chromatin remodeling complex through the RNA-binding eAT-hook domain of SMARCA4 and convert the complex from an ATP-dependent chromatin remodeler into a non-catalytic transcriptional scaffold that recruits Mediator, CBP/p300, and MLL3/4. These findings suggest that noncoding RNAs regulate not only where chromatin regulatory complexes localize but also how they function. Similar RNA-dependent mechanisms are emerging for additional chromatin regulators, including the MLL1/COMPASS complex, suggesting that RNA-mediated regulation of chromatin may represent a general and therapeutically exploitable principle of gene regulation.

We propose to establish an integrated discovery platform that systematically identifies chromatin-regulatory ncRNAs, determines their structural and functional properties, and enables their therapeutic targeting. Rather than focusing on a single RNA or disease, this platform integrates complementary technologies to address each major barrier limiting the development of RNA therapeutics for regulatory ncRNAs. Functional genomic screening will identify disease essential enhancers and their associated ncRNAs. Protein-centric RNA mapping will define the ncRNAs that interact with chromatin regulatory complexes. RNA structural analyses will identify accessible regions for rational antisense oligonucleotide design. Finally, clinically translatable delivery technologies will enable efficient, repeat-dosable delivery of optimized antisense therapeutics to diseased tissues.

As an initial proof of concept, we will apply this platform to SMARCA4-deficient lung adenocarcinoma, a cancer that is critically dependent on SMARCA2-mediated enhancer regulation. We will identify enhancer RNAs required for recruitment of SMARCA2-containing SWI/SNF complexes, determine their structural features that govern RNA-protein interactions, develop optimized antisense oligonucleotides that selectively disrupt these interactions, and evaluate a novel ultrasound-enabled delivery platform for safe, efficient, and repeatable therapeutic administration. Importantly, every component of this workflow is designed to be broadly applicable to other chromatin-regulatory complexes, including MLL3/4, KAT6A/6B, Mediator, and additional RNA-dependent chromatin regulators.

The long-term goal of this project is to establish a generalizable technology platform for discovering and therapeutically targeting chromatin-regulatory ncRNAs across human disease. By integrating functional genomics, RNA structural biology, chromatin biology, antisense chemistry, and precision delivery, this platform will provide a new framework for developing RNA therapeutics directed against regulatory RNAs that have previously been inaccessible to conventional drug discovery approaches.

“Spatial Mapping and Functional Perturbation of Disease-associated Non-coding RNAs”

Autism spectrum disorders and other neurodevelopmental disorders are increasingly linked to genetic variations within the non-coding genome. Recent studies have identified pathogenic mutations affecting long non-coding RNAs (lncRNAs) and short non-coding RNAs (sncRNAs), including lnc-NR2F1, PTCHD1-AS, CHASERR, RNU4-2, and RNU2-2. Despite these discoveries, the molecular mechanisms by which these non-coding RNA (ncRNAs) control gene expression remain poorly understood because current genomic approaches lack spatial resolution and cannot directly visualize RNA-chromatin interactions within intact chromatin environment. There is also no therapeutic strategy available for such ncRNA mutations.

This proposal will develop CHIR-MERFISH (Chromatin Interacting RNA MERFISH), a spatial imaging platform that combines RNA-guided chromatin tagmentation with multiplexed error-robust fluorescence in situ hybridization to map RNA-chromatin interactions at single-cell and subnuclear resolution. The method will be benchmarked against established sequencing-based assays and validated in cultured cells and adult human brain tissue. CHIR-MERFISH will simultaneously identify ncRNAs, their chromatin binding sites, and cellular identity within intact

tissue, providing a general platform for studying chromatin-associated RNAs and evaluating how therapeutic restoration of disease-associated ncRNAs reshapes chromatin interactions.

The proposal will also establish a proof of concept for therapeutic restoration of a disease-associated non-coding RNA. Using genetically engineered human induced pluripotent stem cell derived neurons, I will generate an in vitro model of lnc-NR2F1 haploinsufficiency and characterize its effects on neuronal morphology, synapse formation, electrophysiological function, and gene expression. Using endogenous CRISPR activation and lentiviral overexpression of lnc-NR2F1, I will evaluate the effects of therapeutic rescue of lnc-NR2F1 expression by sequencing, imaging and proteomic analyses.

Together, this work will establish a new class of spatial interactome technologies for studying ncRNAs while defining the therapeutic effect by restoring the expression of a disease-associated lncRNA in human neurons. The experimental plan developed here will be broadly applicable to other pathogenic non-coding RNAs, like CHASERR and RNU4-2 implicated in neurodevelopmental disorders and provide a foundation for future RNA-targeted therapeutic strategies.

“AstRNA: A Cell-Specific RNA Editing Platform for Treating Neuroinflammation”

AstRNA is a cell-specific RNA editing platform designed to overcome three fundamental limitations of current RNA-targeted therapeutics: lack of cell-type specificity, inability to modulate protein function without altering its overall expression, and narrow applicability to genetically defined diseases. By enabling site-directed RNA editing that can be restricted to specific cell populations, AstRNA opens the door to therapies that precisely tune protein activity in the cells where dysregulation occurs, without disrupting essential functions of the same protein elsewhere

in the body.

The platform's initial application targets chronic neuroinflammation, a pathological process shared across a broad range of neurodegenerative diseases and other central nervous system conditions, including glioblastoma, ischemic stroke, and spinal cord injury. Reactive astrocytes, a major driver of chronic neuroinflammation, depend on activation of the transcription factor STAT3. Our approach modulates STAT3 activity selectively within astrocytes, reducing pathological inflammation while preserving STAT3's essential functions in neurons and other cell types — an outcome that is not achievable with current RNAi- or ADAR-based therapies, which act indiscriminately across cell types.

Because AstRNA is designed to modulate disease-relevant pathways rather than correct a single genetic mutation, it has the potential to benefit the much larger patient population affected by nongenetic, dysregulation-driven conditions. If successful, this platform could establish a new class of RNA-targeted therapeutics that are precise, cell-specific, and broadly applicable across multiple diseases that converge on shared pathological mechanisms.

“Programmable Intracellular Circular RNA Factories”

Despite remarkable advances, virtually all modern molecular medicines, including RNA therapeutics, gene-editing systems, DNA vectors, and recombinant proteins, share a common bottleneck: they must transport exogenous therapeutic molecules across multiple biological barriers to reach their intracellular sites of action. This dependence on extracellular delivery fundamentally limits treatment durability, tissue accessibility, and therapeutic precision aimed at endogenous RNA. We propose Programmable Intracellular Circular RNA Factories (PICIRF), a transformative platform that replaces extracellular RNA delivery with intracellular RNA manufacturing. PICIRF harnesses engineered, self-destructing Listeria monocytogenes as programmable bacterial drones that invade target cells, escape into the cytosol, and transiently synthesize highly stable circular RNAs (circRNAs). Following programmed bacterial lysis, circRNAs are released directly into the host-cell cytoplasm, where they function as durable therapeutic effectors capable of degrading, editing, or otherwise reprogramming endogenous mRNAs, long noncoding RNAs, and microRNA networks. Unlike conventional RNA medicines, PICIRF enables intracellular production of multiplexed RNA therapeutics while exploiting the natural cell tropism, tissue penetration, and repeat-dosing capability of L. monocytogenes. To establish a rigorous proof-of-concept, bDrones, Inc. will develop and validate a prototype that combines the programmable RNA nuclease Cas13 with guide RNAs to achieve sequence-specific degradation of endogenous RNA. Successful implementation will validate PICIRF's ability to simultaneously manufacture functional protein and regulatory RNA components inside target cells, establishing a versatile platform for diverse RNA-targeting technologies. By integrating synthetic biology, intracellular bacterial engineering, and programmable circRNA therapeutics, PICIRF establishes a fundamentally new class of programmable medicines with broad potential for treating cancer, genetic, infectious, inflammatory, and other diseases driven by dysregulated RNA biology.

“A novel epigenetic RNA platform targeted to silence endogenous RNAs from a duplicated chromosome region”

 We are witnessing remarkable progress in development of therapeutics for single-gene disorders, but the common problem of chromosomal imbalances lags far behind. Small chromosome defects are individually rare but overall impact an astounding 1/140 births, with approximately 190 recognized microdeletion and duplication syndromes. We propose to demonstrate a highly innovative approach with potential to advance translational research and development of therapeutics for a broad class of chromosomal imbalances. The technology we propose to develop has potential to provide a flexible platform which can be targeted to silence multiple RNAs from a small chromosomal region of interest. This approach can help fill a major gap in human genetics: the need to advance translational research and develop therapeutics for highly diverse chromosomal imbalances (small and large dosage disorders). While Trisomy 21 (Down syndrome) is the most common (1/700 births), a host of smaller chromosomal disorders which are individually rare but collectively more common than Down syndrome (DS). For many monogenic disorders the specific gene and mutation are known, and this is critical to understand the biology and inform development of therapeutics of any kind. In contrast, for chromosomal imbalances there is not a specific gene mutation, but the issue is too much (or too little) RNA from a specific genomic region. New approaches are needed to investigate the pathogenesis of dosage disorders, and to develop therapeutic agents that can correct RNA dosage imbalances. We propose a novel technological approach that can provide a needed research tool to dissect the biology of duplication and deletion disorders, information needed to develop more traditional therapeutic strategies. Moreover, the technology we propose has transformative potential to be developed into deliverable therapeutic agents to essentially correct the root cause of duplication disorders, excess RNA expression from a specific chromosomal region.

In a high risk but innovative study, our lab first showed proof of principle in vitro that a transgene expressing full-length XIST RNA can comprehensively silence an autosome; we demonstrated comprehensive silencing of one chromosome 21 in trisomic iPS cells from an individual with Down syndrome. However, because XIST RNA spreads and silences the whole chromosome (from which it is expressed) this precludes broader applicability of XIST for the many smaller chromosomal conditions. This limits any utility of XIST for translational research on most chromosomal imbalances, and, in addition, the 14-17kb length of XIST cDNA essentially precludes any possibility of in vivo delivery using current vehicles. The technology we propose to show is based on a much smaller RNA (a 4% fragment of a lncRNA) which lacks the ability to spread on the chromosome, but we have evidence that it can impact the expression of endogenous RNAs in a small region near the targeted insertion site. We propose to demonstrate a use case which shows the utility and versatility of this technology for small and large chromosomal dosage disorders.

“FORGE platform: Systematically optimized multimeric oligonucleotides for multi-target Engagement”

• Oligonucleotide-based RNA-targeting drugs provide almost endless novel options for treatment of multiple disorders, but today they are often poorly suited for the task in polygenic diseases, largely because they are designed to hit one target at a time.
• Multi-targeting can also be justified in rare single locus disorders to enhance potency.
• FORGE platform offers a solution: linking two or more oligonucleotide payloads (siRNA or ASOs) into a single molecular entity that silences or upregulates multiple disease-relevant genes or regulatory mechanisms simultaneously.
• Why it matters: highly optimized FORGE platform could unlock a new class of RNA medicines for diseases previously too complex or too rare to treat.
• High level of FORGE standardization enables a target gene ‘plug-in’ model that can be disseminated to multiple not-for-profit entities to create treatments for specific rare/polygenic diseases.
• FORGE application examples: Dravet syndrome, Alzheimer’s disease — and extensible well beyond.
• Team credibility: 80+ years combined oligonucleotide/CNS expertise; first to deliver ASOs to the living brain; scientific advisory board with 25+ drug approvals.

“ContextSwitch-RNA/PROOF”

RNA-targeted drug discovery can advance compounds for the wrong reason. Purified-RNA binding, ribonucleoprotein (RNP) assay activity, or cellular rescue may occur without engagement of the disease RNA or claimed mechanism. In myotonic dystrophy type 1 (DM1), expanded DMPK r(CUG)exp RNA sequesters muscleblind-like (MBNL) splicing factors and disrupts splicing, yet genuine RNA-directed activity can resemble assay interference, direct MBNL action, RNA depletion, nonspecific binding, stress, or indirect rescue. This mechanism assurance problem asks not only whether a compound responds, but which biological explanation the evidence supports. ContextSwitch-RNA/PROOF turns staged evidence into a decision-ready package: the strongest permitted mechanism claim, supporting evidence, surviving alternatives, provenance, limitations, and next action. Each case advances, is downgraded, receives one resolving experiment, or stops. The output is a mechanism resolution instrument, not a therapeutic candidate.

The platform compares predeclared intended and rival mechanisms, can report that none fits, and never lets downstream rescue replace missing engagement. A structured, typed evidence ledger records each result with context, uncertainty, quality control (QC), and provenance. A fault-aware decoder compares cross-assay patterns with mechanisms and shared failures and selects a resolving test. An independent rule-based, proof-carrying claim compiler authorizes bounded wording or refuses it; its certificate verifies evidence, rules, and provenance; it does not certify biological truth.

Novelty is the prospective RNA-specific coupling of an error-correcting-output-code-inspired, dependency-qualified mechanism code, shared-fault-aware open-set decoder, and score isolated, type-safe claim compiler. Before outcomes, the system locks rival mechanisms, response and fault patterns, no-fit rules, claim obligations, and resolving tests. Correlated assays cannot count independently; missing and conflicting results remain explicit. For the strongest fault-tolerance claim, three independently qualified assay groups must separate rivals, and loss or corruption of one cannot change the decision.

Phase II has three bounded aims. Aim 1 implements the target manifest, evidence ledger, decoder, experiment selector, independent claim checker, verifier, and adversarial suite. Aim 2 demonstrates the platform in DM1 through identical first stage testing of 96 compounds: 32 context/selectivity-aware, 32 biochemical-only, and 32 diversity controls. Follow-up is capped at 12 confirmations, six cellular candidates, two finalists, and six new analog/probe compounds. Purified-RNA/RNP kinetics are conditional; cellular temporal and structural mediation studies are outside base scope. Aim 3 freezes predictions before held-out DM1/control outcomes and conditionally tests 24 DM2 r(CCUG) compounds with the same engine.

Against identical staged evidence, the platform is compared with a frozen conventional cascade and two independent experts. DM1 success requires a related series with qualified purified RNA/RNP activity, cellular RNP response, and reproducible splicing response in two patient lines; a cellular DMPK-associated engagement claim additionally requires qualified engagement evidence. Deliverables include versioned software, schemas, rights-cleared fixtures, decision/verifier records, SOP/QC templates, and an independent-user test.