. Author manuscript; available in PMC: 2026 Oct 24.

Published before final editing as: Mol Cell. 2026 Oct 20:S1097-2765(26)00702-1. doi: 10.1016/j.molcel.2026.09.037

Regulatable In Vivo Gene Expression via Adaptamers

Jack Bryant ¹, Laura Herron ¹, Yesh Doctor ¹, Finn McRae ¹, Jocelyn Wang ¹, Theodore Hirsch ¹, Andrew Portell ¹, Satheesh Kumar ¹, Prashant Mali ^1,²

PMCID: PMC13692725 NIHMSID: NIHMS2217576 PMID: 42901161

SUMMARY

Precise, reversible control of transgene expression is essential for safe and durable gene therapy, yet current inducible systems remain difficult to translate in vivo due to large size, limited induction duration, and dependence on immunogenic regulators. Here we present the adaptamer (ADAR modulatable aptamer), a compact (<120 bp) RNA switch that couples FDA-approved small-molecule-responsive aptamers with endogenous ADAR–mediated RNA editing to regulate expression. Ligand binding stabilizes a double-stranded RNA structure that recruits ADAR to convert a stop codon into a sense codon, restoring downstream protein translation. This cleavage-free, post-transcriptional mechanism enables precise, small-molecule–dependent modulation without exogenous protein machinery. We demonstrate that adaptamers are highly functional across multiple cell lines, including human T cells. In mice, AAV-delivered adaptamer-controlled FGF21 expression induced metabolic remodeling, reversing obesity and hepatic steatosis. This minimal, programmable system offers a clinically compatible approach for tunable genetic medicines and safe deployment of pleiotropic and dose-limited proteins.

Graphical Abstract

graphic file with name nihms-2217576-f0001.webp

eTOC Blurb

Bryant et al. develop adaptamers, compact RNA switches that use FDA-approved drugs to control gene expression by co-opting endogenous RNA-editing machinery rather than relying on foreign proteins. In mice, this approach safely tuned a metabolic hormone to reverse obesity, pointing to a simpler path toward programmable, drug-controlled gene therapies.

INTRODUCTION

Gene and cell therapies have emerged as powerful modalities for treating a wide range of diseases; however, they generally lack precise control over transgene expression, restricting their use to transient expression, irreversible “hit-and-run” genomic modifications, or permanent constitutive expression^1–4. Because many therapeutic proteins are beneficial only within narrow dose and temporal windows—where both insufficient expression and excessive exposure can cause pathology—this limitation substantially narrows the spectrum of treatable diseases². Moreover, redosing of both viral and non-viral delivery systems remains impractical due to immunogenicity and cost. Consequently, optimal therapies for chronic diseases require vectors that persist long term while enabling tightly regulated, inducible expression².

Numerous protein-based genetic switches have been developed to meet these criteria, most notably the Tet-On system; however, their reliance on bacterial or engineered proteins can provoke immune responses that limit their utility in vivo^5,6. Riboswitches offer an alternative approach. These synthetic RNA elements, composed of a small-molecule-binding aptamer coupled to an effector domain, avoid adaptive immune activation because they are entirely RNA-based^7–9. In mammalian systems, however, riboswitches exclusively rely on small-molecule-regulated RNA cleavage or splicing to control gene expression^7–9. A cleavage-based mechanism intrinsically limits the durability of induction, as the regulated RNA is rapidly degraded following small-molecule withdrawal, necessitating frequent dosing that may be unsafe or impractical in clinical settings¹⁰. Splicing-based mechanisms, on the other hand, rely on surrounding context to function properly making them transgene and vector dependent¹¹. In addition, highly engineered riboswitches—such as the recently reported pA regulator, which exceeds 500 base pairs—further constrain adeno-associated virus (AAV) packaging capacity and introduce complex RNA secondary structures that can impair translation and splicing efficiency⁹. Dependence on cleavage or splicing also precludes deployment in RNA-based expression platforms: cleavage-based designs would self-destruct prior to delivery, while splicing-dependent switches remain inactive due to the absence of cytosolic splicing machinery. As a result, delivery approaches such as lentiviral vectors and mRNA lipid nanoparticles are excluded. Reflecting these challenges, only a single riboswitch, K19, has been evaluated in a mouse disease model; owing to rapid RNA turnover, therapeutic efficacy required twice-daily small-molecule dosing—a regimen likely intolerable in humans¹⁰.

Adenosine deaminases acting on RNA (ADARs) have recently emerged as promising tools for programmable RNA editing^12,13. These enzymes are conserved across metazoans and catalyze the deamination of adenosine to inosine, which is interpreted as guanosine by translational machinery. In humans, three catalytically active isoforms exist: the ubiquitously expressed, nuclear ADAR1 p110; the interferon-inducible, cytoplasmic ADAR1 p150; and the nervous system-enriched, nuclear ADAR2¹⁴. Although the rules governing ADAR targeting are still being elucidated, their strong preferences for specific RNA sequences and secondary structures are well characterized^15,16. This property has been exploited to conditionally silence engineered stop codons upstream of genes of interest, enabling expression in response to target transcripts or endogenous molecules^17–20. However, most such systems require exogenous ADAR expression to achieve robust activation, limiting their translational potential, as ADAR overexpression can result in widespread off-target editing with unknown consequences^17–21.

Here, we introduce a class of genetic switches, termed adaptamers (ADAR modulatable aptamers), capable of driving therapeutically relevant transgene expression in vivo. Adaptamers consist of a small-molecule-binding aptamer fused to a short double-stranded RNA region harboring a premature stop codon. Small-molecule binding stabilizes the structured RNA, promoting recruitment of endogenous ADAR, which edits the stop codon and restores translation. These switches are compact (<120 base pairs), tunable, modular, and operate exclusively at the RNA level. While adaptamers function effectively without exogenous ADAR, we further engineered a positive feedback mechanism in which ADAR expression is conditionally induced, enhancing output in tissues with low basal ADAR activity while avoiding constitutive, off-target RNA editing.

Obesity represents a major public health crisis, affecting approximately 40% of adults over the age of 25 in the United States and contributing to severe comorbidities, including type 2 diabetes, cardiovascular disease, and metabolic dysfunction-associated steatohepatitis^22,23. Fibroblast growth factor 21 (FGF21) has emerged as a promising therapeutic alternative to GLP-1 agonists, owing to its ability to ameliorate obesity and associated metabolic disorders^24–26. Unlike GLP-1 agonists, which primarily act by modulating insulin and glucagon secretion and suppressing appetite, FGF21 promotes weight loss by increasing systemic metabolic activity^26,27. FGF21 gene therapy has also been shown to extend lifespan in mice, largely through metabolic remodeling²⁶. However, FGF21 has a short circulating half-life of approximately two hours, rendering protein-based delivery impractical²⁸. Conversely, constitutive FGF21 overexpression via gene therapy carries significant risks, including muscle atrophy and reduced bone density^29–31.

To harness the therapeutic benefits of FGF21 while minimizing these risks, we employed adaptamer-regulated FGF21 expression delivered by AAV. Because edited transcripts persist beyond the presence of the inducing small molecule, this approach enables high-level, stable yet reversible expression in vivo, functionally analogous to a continuous FGF21 infusion. This unique delivery paradigm supports rapid metabolic remodeling in a controlled and safe manner, resulting in robust weight loss and reversal of obesity-associated comorbidities.

DESIGN

Engineering small molecule inducible ADAR editing

The adaptamer system was engineered by fusing the cb32 tetracycline minimer—an aptamer with high affinity for tetracycline^7,32—to destabilized variants of a naturally occurring RNA structure known to robustly recruit endogenous ADARs³³. We hypothesized that tetracycline binding would push structural equilibrium towards a double-stranded RNA conformation, enabling conditional A-to-I editing of an in-frame stop codon by endogenous ADAR and thereby permitting downstream transgene translation (Fig. 1a). Consistent with this model, transient transfection of HEK293T cells with EGFP reporters followed by exposure to 80 μM tetracycline selectively increased expression from adaptamer-containing constructs (Fig. 1b). Tetracycline had negligible effect on a non-stop control used to determine maximal theoretical expression or a non-tetaptamer negative control used to assess impact on generic ADAR editing³³ (Fig. 1b). Drug-induced A-to-I editing at the target stop codon and quantification of transgene mRNA levels confirmed that activation was editing-dependent rather than an effect of increased mRNA stability (Fig. 1b, Fig. S1a).

To optimize inducibility and expression, we systematically varied the length of the stem above and the root below the stop codon (Fig. 1c). Transient transfection assays in HEK293T cells identified constructs achieving robust tetracycline-dependent induction and up to 39% of constitutive expression (Fig. 1c). In general, extending the stem and root increased both maximal expression and background, consistent with increased thermodynamic stability of longer double-stranded regions (Fig. 1c). Short constructs had low maximum expression while excessively long constructs had high background, limiting fold change in both cases (Fig. 1c). The V₉²¹ construct (9-base root, 21-base stem) was selected for further study as it balanced the need for both absolute protein production and fold change in downstream applications, demonstrating the highest maximum expression among constructs with a fold change >5 in both HEK293T and Neuro2a cells (Fig. 1c).

To assess functionality in primary human cells, peripheral blood mononuclear cells–derived T cells were activated, transduced with EGFP-expressing lentiviral vectors, and cultured in the presence or absence of tetracycline prior to flow cytometric analysis (Fig. 1d). V₉²¹-transduced T cells exhibited a marked increase in both EGFP-positive cells and RNA editing rates upon tetracycline treatment, confirming robust inducibility in primary human cells (Fig. 1d).

Finally, modulation of ADAR abundance in HEK293T cells via shRNA-mediated knockdown or overexpression of ADAR1 p110 revealed a strong correlation between ADAR levels and both adaptamer-driven expression and A-to-I editing at the internal stop codon (Fig. 1e and Fig. S1b, c). We observed an ~80% reduction in total ADAR1 expression, of which ADAR1 p150 only made up a small subset, and no change in ADAR2 expression indicating the effect in HEK293T cells is ADAR1 p110 driven (Fig. S1b).

RESULTS

Doxycycline Adaptamer Development

To demonstrate the versatility of the adaptamer system, we next fused our optimized ADAR-recruiting dsRNA hairpin to a recently discovered doxycycline aptamer, which undergoes a drug-induced stabilization similar to that of the tetracycline aptamer (Fig. 2a)³⁴. To optimize the doxycycline adaptamers, we varied the root and stem length as before, while also testing several transducer sequences (Fig. 2a). Designs were transfected into HEK293T cells in a pooled format, and RNA editing rates were measured by NGS in the presence and absence of doxycycline to determine induction fold and maximum editing rate (Fig. 2b). No change in editing rate was observed for the control ADAR reporter upon doxycycline addition (Fig. 2b). Individual validation of constructs with a fold change above 5.5 and maximum editing above 9.5% led us to select V₉¹⁷-Dox for further testing, given its profile similar to that of V₉²¹-Tet (Fig. 2c). V₉¹⁷-Dox displayed ADAR1-dependent expression and RNA editing rates comparable to those seen with the tetracycline adaptamer (Fig. 2d).

Fig. 2 | — (a) Schematic of doxycycline adaptamer with variable transducer sequence, stem length, and root length. (b) Results of pooled in vitro RNA editing screen in HEK293T cells with 80μM doxycycline identifying several top candidates. ADAR reporter RNA editing in blue didn’t change in response to doxycycline. (c) Individual validation of doxycycline responsive adaptamers verified their functionality. (d) Doxycycline adaptamers were responsive to changes in ADAR1 protein levels. The same shRNA cell lines were used as in tetracycline adaptamer shRNA experiment (e) Cotransfection of adaptamers demonstrated orthogonality with the tetracycline controlled adaptamer only responsive to tetracycline and the doxycycline adaptamer only responsive to doxycycline. Error bars represent mean ± S.E.M (n = 3), and individual dots represent biological replicates. Percent expression is calculated as mean fluorescence intensity or luminescence normalized to the no-stop-codon control with tetracycline.

Notably, cells cotransfected with both V₉¹⁷ (doxycycline-adaptamer-controlled sNLuc) and V₉²¹ (tetracycline-adaptamer-controlled fLuc) demonstrated the orthogonality of the two control systems despite their nearly identical small-molecule inducers (Fig. 2e). Multiplexed, small-molecule-regulatable control is highly desirable across a range of basic science, biotechnology, and therapeutic applications. Given the limited number of contemporary options, both in vitro and in vivo, we are particularly excited that the versatility of the adaptamer system readily enabled the development of both tetracycline and doxycycline regulation systems.

Biochemical Mechanism

Tetracycline and doxycycline binding to their respective aptamers has been reported to stabilize the bottom (P1) stem of the aptamer^32,34; we therefore reasoned that aptamer binding shifts the RNA equilibrium towards dsRNA conformations more compatible with ADAR recruitment, thereby increasing A-to-I conversion rates (Fig. S2a). To confirm that small-molecule binding stabilizes the entire adaptamer sequence, we performed differential scanning fluorimetry^35–37, which revealed a dose-dependent increase in melting temperature only in response to the target drug (Fig. S2b, c). Furthermore, A-to-I editing rates in a biochemical ADAR editing assay increased only in the presence of the target drug, supporting the hypothesis that small-molecule binding to the RNA directly increases ADAR editing (Fig. S2d). However, this drug-dependent effect was attenuated relative to the cellular effect, supporting the idea that the increase in ADAR editing is driven predominantly by kinetics and stability rather than by a discrete conformational change or relief of steric hindrance occurring only in the presence of drug.

Engineering ADAR Feedback Loops to Increase Expression

We observed that increasing overall tetracycline adaptamer length enhanced maximal expression but reduced fold induction due to elevated basal activity (Fig. 1c). To overcome this tradeoff, we engineered a positive feedback loop in which adaptamer-regulated exogenous ADAR expression could, in turn, amplify output while maintaining low background³⁸. This design could also extend adaptamer applicability to settings with limited endogenous ADAR activity, such as immortalized cell lines³⁹. Specifically, the system comprises an adaptamer-controlled transgene followed in series by an adaptamer-controlled ADAR, separated by self-cleaving 2A peptides to ensure independent protein function (Fig. 3a).

Fig. 3 | — (a) Schematic of an adaptamer-controlled transgene followed by a secondary adaptamer regulating exogenous ADAR expression to amplify induced output. (b) Diagram illustrating the generation of single-copy integration feedback loop cell lines to approximate in vivo ADAR expression levels. (c) Expression of feedback loop constructs (black patterning) compared to an adaptamer-only cell line. (d) Representative FACS histograms (log scale) showing induced (yellow) and uninduced (green) feedback loop cell lines versus adaptamer-only cells. (e) RNA editing levels and corresponding fluorescence. (f) Dose-response curves with EC50 values of 17.1 μM for V₉²¹ and 29.6 μM for V₉²¹–V₉¹⁸–ADAR2, along with time-course data. (g) Global RNA editing rate (AEI) in feedback loop cell lines compared to untransduced cells and single-copy ADAR2-overexpressing cells (P = 0.32, unpaired two-tailed t-test). Error bars represent mean ± S.D. (n = 2), and individual dots indicate biological replicates. Percent expression is normalized to a single copy constitutive EGFP-overexpressing cell line.

To ensure broad experimental applicability, we generated single-copy lentiviral integration cell lines harboring the engineered adaptamer system (Fig. 3b). Initial comparisons revealed that constructs utilizing ADAR1 p110 or ADAR2 produced similar protein levels; we therefore selected ADAR2 due to its smaller size (2.1 kb versus 2.8 kb) and reduced off-target editing profile (Fig. 3c)²¹. Using V₉²¹ as the primary regulatory element, we inserted secondary adaptamers (V₉¹², V₉¹⁵, or V₉¹⁸) between the EGFP transgene and ADAR2 to tune effector production while maximizing reporter output. Notably, these feedback-loop cell lines exhibited up to 9-fold higher expression and approximately threefold increased RNA editing upon tetracycline induction, without increased basal expression, relative to V₉²¹ alone (Fig. 3c–e). Additionally, induction was uniform across the cell population, with no evidence of sporadic background activation (Fig. 3d, e). Utilizing V₉²¹, which has slightly higher background expression than V₉¹⁸, as the secondary adaptamer led to autoactivation in HEK293T cells, highlighting the importance of tight control of ADAR expression (Fig. S3a). The top-performing construct, V₉²¹–V₉¹⁸–ADAR2, achieved higher expression than the parental V₉²¹ across all tested tetracycline concentrations, with only a modest increase in EC50 (29.6 μM versus 17.1 μM), indicating that activation remained sensitive and tunable (Fig. 3f). Importantly, expression was rapidly reversible within three days following tetracycline withdrawal, exhibiting only a brief delay relative to the parent construct (Fig. 3f). Transcriptome-wide RNA-seq analysis revealed no increase in global A-to-I editing rates measured by AEI⁴⁰ or differentially edited windows measured by LoDEI⁴¹ (Fig. 3g, Fig S3b). To detect additional potential off-target editing sites we utilized REDItools⁴² to quantify A-to-I editing rates at individual sites transcriptome wide and compared to editing rates in untransduced cells (Fig. S3c). Notably, we were only able to detect 89 differentially edited sites in the V₉²¹–V₉¹⁸–ADAR2 cell line compared to 6,323 differentially edited sites in a cell line constitutively overexpressing ADAR2 (Fig. S3c, d). Differential gene expression analysis only identified four differentially expressed transcripts, all long non-coding RNAs on chromosome 22 with the exception of ADARB1 (Fig S3e). Together this demonstrated that background exogenous ADAR expression remained sufficiently low to have negligible off-target effects when the system is uninduced.

In vivo gene control via Adaptamers

To evaluate the tetracycline adaptamer’s capacity for enabling tunable gene therapy in vivo, we generated AAVs expressing adaptamer-controlled firefly luciferase (fLuc). We omitted the ADAR feedback loop due to packaging constraints and the potentially higher endogenous ADAR activity in vivo, which we anticipated could enhance maximal expression³⁹. Mice were systemically injected with 1×10¹² vg of AAV9 expressing either V₉²¹-controlled or constitutive fLuc to assess expression and inducibility. Initial induction cycles with 60 mg/kg tetracycline via intraperitoneal injection—a standard murine dose—demonstrated reversible and repeatable expression increases, returning to near baseline within 48 hours (Fig. S4a, b)^7,9.

Six weeks post-injection, mice received 60 mg/kg tetracycline and were imaged at 0, 4, 8, 12, and 24 hours (Fig. 4a). To confirm RNA editing at harvest, mice were redosed at 24 hours and sacrificed at 28 hours. Induced expression was detectable within 4 hours, peaking at 12 hours (Fig. 4b). Liver lysate analysis of luminescence and RNA editing corroborated IVIS findings (Fig. S4c, d). Mice transduced with an ADAR-editable loop lacking the tetracycline aptamer showed no increase in RNA editing, confirming adaptamer-specific induction (Fig. S4e).

To rapidly optimize adaptamer designs for improved in vivo functionality, we next screened a subset of designs by injecting a pool of adaptamer variant-containing AAVs. We dosed half the animals with tetracycline, and quantified differential RNA editing rates using each adaptamer sequence as its own barcode (Fig. S4f). From this screen we selected V₉¹⁸, with a three-base shorter stem than V₉²¹ and reduced background in vitro, as a potential better-performing adaptamer (Fig. 4c). Using AAV8 expressing inducible sNLuc, we compared V₉¹⁸ and V₉²¹ in vivo, confirming lower background and similar or improved maximum expression across multiple tetracycline dosing regimens for the V₉¹⁸ design (Fig. 4d). Secreted NanoLuc expression peaked at 24 hours—slightly later than fLuc, likely due to secretion and biodistribution—allowing repeated tetracycline dosing to achieve sustained induced expression (Fig. 4b, d).

Adaptamer mediated inducible FGF21 gene therapy for reversing obesity

To demonstrate therapeutic potential, we next applied the adaptamer system to inducibly overexpress FGF21 for treatment of chronic obesity. FGF21 overexpression is known to enhance adipose metabolism, promote liver rejuvenation, and improve glucose tolerance but can dysregulate muscle and bone metabolism, highlighting the need for inducible control^26,29–31. We first optimized viral dosing by injecting mice with 5×10¹⁰, 2×10¹¹, or 1×10¹² vg of tetracycline inducible AAV8-V₉¹⁸-FGF21, selecting the midpoint based on prior AAV studies and our observed ~10% background from sNLuc experiments (Fig. S5a)^26,43. The 2×10¹¹ vg dose yielded negligible basal FGF21 yet achieved therapeutically relevant levels upon 60 mg/kg intraperitoneal tetracycline administration (Fig. S5a). Next, we tested three tetracycline regimens with six days between doses—one 60 mg/kg injection, two 30 mg/kg injections 12 hours apart, and a single 30 mg/kg injection—to determine minimal effective dosing. All regimens produced comparable FGF21 expression 24 hours post-induction, so we proceeded with the conservative 30 mg/kg dose (Fig. S5b).

To model obesity therapy, 21-week-old diet-induced obese (DIO) mice were injected with 2×10¹¹ vg of inducible AAV8-FGF21 and treated one week later with daily 30 mg/kg tetracycline to trigger metabolic reprogramming (Fig. 5a). Inducible FGF21 expression achieved sustained therapeutic levels in drugged mice while remaining low in vehicle-treated controls (Fig. 5b). Induction persisted for at least 24 hours post-tetracycline, with elevated FGF21 and RNA editing despite FGF21’s short half-life²⁸ and rapid tetracycline clearance⁷(Fig. S6a). This effect was driven by RNA editing with no increase in AAV transcript levels seen upon induction (Fig. S6b). Substantial weight loss occurred only in mice receiving both AAV-FGF21 and tetracycline (FGF21 +Tet), confirming the requirement for both gene therapy and induction (Fig. 5c, Fig. S6c). Glucose tolerance testing on Day 22 further confirmed improved glycemic control in FGF21 +Tet mice, reflecting effective metabolic reprogramming (Fig. 5d). Histological analysis showed near-complete reversal of hepatic steatosis, evidenced by elimination of lipid droplets in hepatocytes (Fig. 5e). Additionally, lipid accumulation in interscapular brown adipose tissue decreased, and inguinal white adipose tissue exhibited reduced adipocyte size (Fig. S6d, e).

Fig. 5 | — (a) Schematic of the FGF21 gene therapy strategy, where inducible FGF21 expression enhances fat metabolism, improves glucose tolerance, and reduces body weight in obese mice. (b) Serum FGF21 levels in diet-induced obese (DIO) C57BL/6 mice following injection of 2×10¹¹ vg per animal, measured by ELISA. (c) Changes in body weight expressed as a proportion of initial weight at the time of AAV injection. (d) Glucose tolerance test performed on Day 22. (e) Hematoxylin-eosin staining of liver sections showing reversal of hepatic steatosis in FGF21 + tetracycline–treated mice. Error bars represent mean ± S.E.M. (n = 4).

DISCUSSION

While gene therapy offers the potential for long-lasting treatment, current modalities suffer from a “Goldilocks problem”—constitutive expression can lead to toxicity (e.g., FGF21-induced bone loss or muscle atrophy) or desensitization, while transient vectors fail to provide durable relief. To address this, we introduced the adaptamer, a compact (<120 bp), human-compatible, small-molecule–responsive riboswitch that enables inducible and reversible control of gene expression via endogenous ADAR-mediated RNA editing. Unlike existing systems that rely on immunogenic bacterial proteins (e.g., Tet-On) or RNA cleavage, the adaptamer stabilizes a double-stranded RNA structure upon ligand binding to recruit ADAR, converting a premature stop codon into a sense codon and restoring translation. This system differs from other riboswitches in three key ways, providing unique advantages for both utility and future development. First, to our knowledge, it is currently the only mammalian riboswitch demonstrated in vivo to use a non-cleavage, non-splicing mechanism, enabling RNA-based delivery and extra-nuclear activation. This allows integration with mRNA lipid nanoparticles and lentiviral vectors, expanding delivery options and enabling cell-based screens. Second, the core adaptamer is extremely compact, facilitating efficient integration into space-constrained vectors such as AAVs carrying large transgenes, including full-length SaCas9. Unlike other riboswitches with large, GC-rich structures, particularly splicing-based designs, the adaptamer is transgene-agnostic and streamlined for vector construction⁹. Finally, the system is modular, with distinct effector and small-molecule-binding domains, which allowed substitution of the tetracycline aptamer with a doxycycline binding aptamer to generate an orthogonal riboswitch without scaffold modifications. The scarcity of orthogonal, in vivo-validated small-molecule control systems is a persistent constraint across synthetic biology and gene therapy. The ease with which the adaptamer scaffold accommodated both tetracycline and doxycycline responsiveness suggests this modularity could extend to other nonperturbing, biocompatible ligands as new aptamers are discovered, broadening the system’s utility for multiplexed or context-specific control. Expression is tunable via root and stem truncation or extension, enabling precise control of dynamic range. Leveraging A-to-I editing as the effector enabled an in vivo pooled riboswitch screen, a powerful strategy for rapid testing across tissues and contexts.

The adaptamer also advances ADAR-mediated sensing. While previous ADAR-based switches respond to endogenous molecules, this system is controlled by FDA-approved exogenous ligands and is capable of producing therapeutically relevant transgene levels without exogenous ADAR overexpression^17,18,20. Its reliance on endogenous ADAR offers additional advantages: higher brain ADAR activity may allow tissue-specific expression and liver detargeting¹⁴, while inducible ADAR p150 expression in immune contexts, such as activated T cells, could enable synergistic protein production in combination with a small molecule^44,45.

Crucially, the adaptamer demonstrates clinical relevance. In a diet-induced obesity mouse model, small-molecule-induced FGF21 expression produced therapeutic benefits without background activity, reversing obesity, glucose intolerance, and hepatic steatosis. Unlike cleavage-based RNA switches, adaptamer RNA remains stable without the small molecule, permitting sustained yet inducible expression—critical when short inducer exposure windows are required, as with tetracycline⁷. This inducible stability enhances both the safety and efficacy of FGF21 gene therapy.

In conclusion, we demonstrate that adaptamers can be rationally engineered to enhance functionality and enable multiplexing, as well as deployed in vivo to deliver therapeutic benefits e.g. through small-molecule-controlled metabolic reprogramming. While further optimization will enhance performance, the adaptamer system represents a substantial advance in synthetic gene regulation, expanding the capabilities of both riboswitches and ADAR-mediated sensors.

LIMITATIONS OF THE STUDY

Adaptamers present a powerful compact tool for gene regulation, but are not without limitations. Current adaptamers display variable behavior depending on endogenous ADAR levels, and thus some optimization will be required when expanding to contexts beyond this study - for instance, considering shorter adaptamers for high endogenous ADAR expressing cells and tissues and longer adaptamers for low endogenous ADAR expressing cells and tissues. While both dynamic range and maximal expression are sufficient for in vivo studies, especially for applications such as expressing hormones, cytokines, and other biologics where a large phenotypic shift can result from modest shifts in quantity, we believe the adaptamer can be improved through mutational screens and directed evolution of both the aptamer and dsRNA to extend to applications requiring more binary switching. Additionally, as new aptamers are identified via SELEX, they can be incorporated with minimal scaffold modifications, to further broaden adaptamer versatility and multiplexing capabilities^8,46.

RESOURCE AVAILABILITY

Lead Contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Prashant Mali (pmali@ucsd.edu).

Materials Availability

All key reagents will be made available via Addgene. Source data are provided with this paper.

Data and code availability

FASTQ files can be found through the Sequence Read Archive under BioProject PRJNA1526315. This paper does not report original code.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR METHODS

Experimental Model and Study Participant Details

Cell Culture

HEK293T cells (female) were cultured in DMEM (ThermoFisher) supplemented with 10% FBS (ThermoFisher) and 1% Antibiotic-Antimycotic (ThermoFisher) in an incubator at 37°C and 5% CO₂ atmosphere.

Neuro2a cells (male) were cultured in EMEM (ATCC) supplemented with 10% FBS (ThermoFisher) and 1% Antibiotic-Antimycotic (ThermoFisher) in an incubator at 37°C and 5% CO₂ atmosphere.

Human PBMCs (Stemcell Technologies) were cultured in RPMI 1640 (ThermoFisher) supplemented with 10% FBS (ThermoFisher), 1% Antibiotic-Antimycotic (ThermoFisher), 200 IU/mL rhIL-2 (Stemcell Technologies) and ImmunoCult^™ Human CD3/CD28 T Cell Activator (Stemcell Technologies) in an incubator at 37°C and 5% CO₂ atmosphere.

Animal Models

All animal procedures were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of the University of California, San Diego. All mice were acquired from Jackson Labs. Adult male C57BL/6J mice were injected retro-orbitally with 1e12 vg of AAV virus unless otherwise noted. 21-week-old DIO C57BL/6J mice were purchased from Jackson Labs for FGF21 study. Mice were randomly assigned to experimental groups except for DIO mice which were distributed by weight to ensure equivalent groups.

Method details

Plasmid Design/Construction

Parental plasmids were constructed by overlap extension PCR of relevant components and Gibson assembly into either pX601-AAV-CMV⁴⁸ (Addgene #61591), ePIP⁴⁹ (Addgene #172110), or a custom synthesized rAAV2 vector backbone. Adaptamer sequences were chemically synthesized as gene fragments by IDT or Twist Biosciences and inserted into backbones through Gibson assembly. See Table S1: Plasmid and adaptamer DNA sequences, related to Figures 1–5.

General Transfection Protocol

24 hours prior to transfection, cells were seeded at 25,000 cells per square cm in either a 6, 24, or 48 well plate. The following day, cells were transfected with 250ng plasmid and 0.5 μL Lipofectamine 2000 (ThermoFisher) per cm² according to manufacturer’s protocol. Media was exchanged 4 hours after transfection with the appropriate tetracycline or doxycycline concentration.

Flow cytometry

Harvest for flow cytometry occurred 24 hours after transfection for HEK293T cells, 48 hours after transfection for N2a cells, or at varying timepoints for transduced HEK293T cells. Growth media was removed and 50 μL per cm² 0.05% Trypsin (ThermoFisher) was added for a 5 minute incubation at 37°C to dissociate. Cells were resuspended in a solution of 50% culture media and 50% PBS (ThermoFisher) to a final concentration of approximately 4e5 cells per mL. 10,000 single cells per sample were analysed with a BD FACSCanto or LSRFortessa machine. Using untreated cells as a gating control, live cells were gated using forward and side scatter, single cells were gated using both side and forward scatter area versus width. Percent expression was calculated using the Mean Fluorescence Intensity of transfected cells or Median Fluorescence Intensity of transduced cells, subtracting untreated cell value as background and normalizing to no adaptamer control MFI.

RNA editing analysis

Cells were lysed 48 hours after transfection and processed using the RNeasy Mini Kit (Qiagen) with optional on column DNase digestion following manufacturer’s protocol. RNA was extracted from manually homogenized mouse tissue using TRIzol (Invitrogen) and chloroform according to the manufacturer’s protocol and process using the RNeasy Mini Kit. cDNA was generated using the Protoscript^® II First Strand cDNA Synthesis Kit (NEB). The adaptamer sequence was amplified using primers 3xFlag_F and P2A_R or Adapt_Univ_R then Sanger sequenced using the appropriate reverse primer. Sanger sequencing traces were quantified utilizing Chromas, and percent editing was calculated by dividing the height of the G peak by the sum of G and A peak heights.

ADAR Modulation

For ADAR1 knockdown, a 60% confluent well of a 6 well plate of HEK293T cells was transduced with 50 μL of either NTC or ADAR1 targeting shRNA concentrated lentivirus. Viral media was exchanged with fresh media after 24 hours and supplemented with 1 μg/mL puromycin at 48 hours to ensure a pure population. 25,000 HEK293T cells per well were seeded in a 48 well plate 24 hours prior to transfection of 250ng of EGFP expressing adaptamer constructs using 0.25μL of Lipofectamine 2000 (ThermoFisher) per well.

For ADAR1 p110 overexpression, 12,500 HEK293T cells per well were seeded in a 48 well plate 24 hours prior to transfection with 250ng of ADAR p110 or pUC19 plasmid using 0.25 μL of Lipofectamine 2000 (Thermofisher) per well. 16 hours after transfection with ADAR modulators, a gentle media exchange was performed followed by a transfection of 250ng of EGFP expressing adaptamer constructs using 0.25 μL of Lipofectamine 2000 (ThermoFisher) per well.

4 hours after EGFP plasmid transfection a complete media exchange was performed with media containing either 0 or 80 μM tetracycline or doxycycline. Flow cytometry was performed at 24 hours post media change and RNA was harvested for editing analysis at 48 hours. ADAR transcript qPCR was performed using a CFX Connect Real Time PCR Detection System (Bio-Rad) and iTaq Universal SYBR Green Supermix (Bio-Rad) with primers hADAR1-qPCR-F’ and hADAR1-qPCR-R, hADAR2-qPCR-F: and hADAR2-qPCR-R, or hADARp150-only-F and hADARp150-only-R, and normalized to B-actin expression with primers qPCR_human_B-ACTIN_F and qPCR_human_B-ACTIN_R. Human ADAR1 was targeted with shRNA TRCN0000050788 (MilliporeSigma), sequence 5’GGGCCCACTGTTATCTTCACTTTCTCGAGAAAGTGAAGATAACAGTGGG C3’ that recognizes the 3’ UTR of both ADAR1 p110 and ADAR1 p150 transcripts.

Lentiviral Production

To produce concentrated lentivirus an 80% confluent 15cm plate of HEK293T cells with fresh media was transfected with 3μg pMD2.G, 12μg of pCMV delta R8.2, and 9μg of lentiviral vector using 36μL of Lipofectamine 2000 (ThermoFisher). 48 hours after transfection media was collected at 4°C and replaced. At 72 hours the second batch of media was collected, pooled with the first batch, and centrifuged at 500xg for 5 minutes to remove carryover cells. The supernatant was then passed through a 0.45μm filter and concentrated in a 100 kDa MWCO Ultra-Centrifugal Filter (MilliporeSigma) to a final volume of ~400μL.

Primary Human PBMC Transduction

Human PBMCs (Stemcell Technologies) were thawed three days prior to transduction and seeded at 1e6 cells/mL in RPMI 1640 (ThermoFisher) supplemented with 10% FBS (ThermoFisher), 1% Antibiotic-Antimycotic (ThermoFisher), 200 IU/mL rhIL-2 (Stemcell Technologies) and ImmunoCult^™ Human CD3/CD28 T Cell Activator (Stemcell Technologies). Cells were centrifuged and reseeded in fresh media 24 hours prior to transduction. 50μL of concentrated EGFP lentivirus was mixed with 1e6 cells and polybrene added to 8μg/mL (MilliporeSigma). Cell virus mixture was spinoculated at 1,000xg and 32°C for 90 minutes, then supernatant was removed and cell pellet was resuspended and seeded in RPMI 1640 (ThermoFisher) supplemented with 10% FBS (ThermoFisher), 1% Antibiotic-Antimycotic (ThermoFisher), 200 IU/mL rhIL-2 (Stemcell Technologies) at 2e6 cells per mL. Two days after transduction, PBMCs were split, centrifuged, and resuspended in complete RPMI without activator containing either 0μM or 20μM tetracycline. Two days after addition of tetracycline, cells were analyzed via flow cytometry and percent EGFP positive cells were gated using an untransduced PBMC sample as a negative control. RNA was also extracted from cells and used to determine RNA editing rates.

Doxycycline Adaptamer In Vitro Screen

HEK293T cells were transfected with a pool of doxycycline adaptamers and 4 hours after transfection media was exchanged with fresh media containing either no drug or 80μM doxycycline. RNA was purified as previously described and adaptamers were amplified from cDNA and processed via Amplicon-EZ (Azenta Life Sciences). Primers, Adapt-Screen-F, with the N nucleotides varied for multiplexing, Adapt-Screen-R. The ratio of reads with an A-to-I edit to the total number of reads for each construct was calculated to determine RNA editing rate.

Coexpression Experiment

HEK293T cells were transfected with equal amounts of fLuc and sNLuc encoding plasmid utilizing standard transfection protocol. 4 hours after transfection media was exchanged with fresh media containing no drug, 40 μM tetracycline, 40 μM doxycycline, or both 40 μM tetracycline and doxycycline. 48 hours after transfection 2 μL of supernatant was diluted 1:50 in PBS, then 10 μL of diluted supernatant was mixed with 10 μL of complete substrate from the Nano-Glo^® Luciferase Assay System (Promega) and read on a SpectaMax iD3 plate reader (Molecular Devices) to quantify sNLuc. Supernatant was aspirated and cells were resuspended in a 1:1 mix of Bright-Glo^™ (Promega) and PBS then read on a SpectaMax iD3 plate reader (Molecular Devices) to quantify fLuc.

Differential Scanning Fluorimetry (DSF)

Adaptamer sequences were PCR amplified using T7-F and IVT-R, transcribed into RNA using the HiScribe^® T7 High Yield RNA Synthesis Kit (NEB) according to manufacturer’s protocol overnight at 37°C, and purified with the Monarch^® Spin RNA Cleanup Kit (NEB). RNA was diluted to 1 μM in DSF Buffer (1x RiboGreen^®(ThermoFisher), 50 mM Tris-HCl pH 8.0, 1 mM MgCl₂, 25 mM KCl) with varying concentrations of tetracycline or doxycycline. Samples were loaded into a 96 well qPCR plate and run on a CFX Connect Real-Time PCR Detection System melt curve. In brief, samples were heated to 65°C for 5 minutes before being slowly cooled to 18°C at a rate of 0.1°C per second. Samples were then slowly heated at a rate of 0.3°C per minute to 85°C with fluorescence captured every 0.1°C. The temperature with the most negative first derivative of fluorescence was designated the melting temperature.

Biochemical ADAR Editing

Adaptamer and transgene sequences were PCR amplified, transcribed into RNA using the HiScribe^® T7 High Yield RNA Synthesis Kit (NEB) according to manufacturer’s protocol overnight at 37°C, and purified with the Monarch^® Spin RNA Cleanup Kit (NEB). Substrate was diluted to 10nM in ADAR Reaction Buffer (15 mM Tris-HCl, 30 mM KCl, 1.5 mM MgCl₂, 3% glycerol, 10μg/mL purified total HEK293T RNA, 0.5mM DTT, and 160units/mL RNAse Inhibitor) with tetracycline or doxycycline added to 40 μM with a final reaction volume of 10 μL. Samples were heated to 65°C for 3 minutes then rapidly cooled to 37°C at which point 10% of the reaction volume of purified ADAR1 p150 protein (55 ng/μL ADAR1p150, 50mM Tris-HCl pH 8.0, 750 mM NaCl, 0.01% Triton X100, 10% glycerol, and 10 μg/mL FLAG peptide) was added. The reaction was incubated at 37°C for 10 minutes then quenched by addition of 40 μL of 95°C nuclease free water and immediately cleaned with a Monarch^® Spin RNA Cleanup Kit (NEB). cDNA was generated with a ProtoScript^® II First Strand cDNA Synthesis Kit and adaptamer sequences were amplified from cDNA using Adapt-Screen-F and Adapt-Screen-R then processed via Amplicon-EZ (Azenta Life Sciences) to quantify RNA editing rates.

Lentiviral HEK293T Feedback Loop Cell Line Generation

For each lentiviral line, one well of a 6 well plate was seeded with 100,000 HEK293T cells 24 hours prior to Lipofectamine 2000 transfection with 200 ng pMD2.G, 800 ng of pCMV delta R8.2, and 600 ng of lentiviral vector with a media exchange immediately prior to transfection. 24 hours after transfection media was collected, mixed with polybrene (MilliporeSigma) to 8 μg/mL, centrifuged at 500xg for 5 minutes to remove cells, and used to replace media on a new well of HEK293T cells seeded one day prior. The following day, lentivirus containing media was replaced with standard complete media, and two days after transfection cells were cultured in media containing 1 μg/mL puromycin (ThermoFisher) for the remainder of experiments to select for transduced cells.

RNAseq and global ADAR editing analysis

A confluent well of a 6 well plate of transduced cell line was collected and RNA extracted using the RNeasy Mini Kit (Qiagen). A NEBNext Poly(A) mRNA Magnetic Isolation Module Kit (NEB) was used to select from 1 μg per sample of total RNA and Illumina-compatible RNAseq libraries were created using an NEBNext Ultra RNA Library Prep Kit (NEB). Sequencing was performed on an Illumina NovaSeq X, with paired end 100 base pair reads. Read files were aligned with BWA-MEM2 to the hg38 Human Reference genome. For RNA editing analyses, Alu Editing Index (AEI) was determined using RNAEditingIndexer with default parameters and the builtin hg38 human genome reference, alu annotations, and SNPs⁴⁰. Differentially edited sites were determined via LoDEI using default parameters (q < 0.1) and hg38 Human Genome Annotation (2024-A) with GENCODEv44 annotations⁴¹. REDItools was also used to quantify potential off-target RNA editing with an FDR cutoff of <0.001 determined by Fisher’s Exact Test⁴². DESeq2 was used on the associated bulk RNA-Seq data to detect differentially expressed genes with cutoffs of 2 for log₂ fold change and 0.05 for adjusted p-value.

AAV Production

AAV virus was produced by triple-transfection of HEK293T cells and purified with an iodixanol gradient as previously described^50,51. Briefly, cells were seeded at 10% confluency in 15cm dishes 2 days prior to transfection, with complete media exchanges performed 2 hours before and 16 hours after transfection. Each plate was transfected with 30 μg total of transgene vector, capsid vector (pAAV2/9n or pAAV2/8), and pHelper vector in an equimass ratio utilizing linear polyethyleneimine (PEI) at a ratio of 4:1 PEI to DNA (PEI dissolved 1 mg/mL in PBS, pH balanced to 7 with HCl and NaOH). 4 days post-transfection virus was harvested from supernatant via overnight 4°C 10% polyethylene glycol (PEG) incubation and directly from freeze-thaw lysed cells. The combined virus was Benzonase (SigmaAldrich) treated at 37°C for 1 hour to digest any unencapsulated DNA and an iodixanol gradient was used to isolate virus from the supernatant. Virus was then dialyzed using 50kDa MWCO Ultra-centrifugal filters (MilliporeSigma) in a solution of PBS (ThermoFisher), 50 mM NaCl, and 0.0001% of Pluronic F68 (ThermoFisher), concentrated to a final volume of ~200 μL, and quantified via qPCR compared to a known standard (ATCC VR-1616) utilizing serial dilutions and primers targeting the ITR regions, AAV-ITR-F and AAV-ITR-R.

Blood Collection

For blood collection and serum extraction mice were anesthetized using controlled O₂/isoflurane and ~20 μL of blood was collected via a tail nick. Blood was allowed to coagulate for 15 minutes at room temperature then was centrifuged at 2,500xg for 10 minutes to separate serum. Serum was removed and frozen at −80°C for processing at a later time.

Tetracycline Injections

Tetracycline injection solutions were prepared immediately before injection as previously described⁷. Briefly, 12 mg of tetracycline hydrochloride (SigmaAldrich) was dissolved by vortexing in 0.8 mL of a 25% weight by volume solution of B-cyclodextrin (MilliporeSigma) in PBS, then diluted by addition of 1.2mL of PBS. Mice were injected intraperitoneally, alternating injection sites to minimize discomfort.

IVIS Imaging

Mice were anesthetized using controlled O₂/isoflurane and injected with 10 μL of a 15 mg/ml solution of d-Luciferin (GoldBio) in PBS per gram body weight. 5 minutes after injection mice were imaged every minute with 0.5 second exposure and 4×4 binning using a Perkin Elmer IVIS Spectrum system until 20 minutes post injection. For quantification consistency the liver was gated and the maximum observed average radiance (photons per second per cm²) was taken for each animal at each timepoint. A mouse not injected with AAV was subtracted as background and percent expression was calculated at each timepoint by dividing by the average radiance across three mice injected with AAV9 expressing constitutive firefly luciferase.

Liver Lysate Luminescence

20mg liver samples were subjected to three freeze thaw cycles, alternating between liquid nitrogen and ice, manually homogenized, and lysed with Glo Lysis buffer on ice for 10 minutes. Samples were then centrifuged at 12,000xg for 15 minutes at 4°C to isolate soluble protein fraction in the supernatant. Supernatant was diluted 1 to 10 in PBS then mixed with equal part Bright-Glo^™ (Promega) and read on a SpectaMax iD3 plate reader (Molecular Devices) to quantify fLuc luminescence.

Secreted NanoLuciferase Assay

2μL of serum was diluted 1:50 in PBS, then 10 μL of diluted serum was mixed with 10 μL of complete substrate from the Nano-Glo^® Luciferase Assay System (Promega) and read on a SpectaMax iD3 plate reader (Molecular Devices). Serum from a mouse not injected with AAV was subtracted as background and percent expression was calculated at each timepoint by dividing by the average luminescence value of three mice injected with AAV8 expressing constitutive secreted NanoLuc.

In vivo RNA editing Screen

Six mice were injected with 1e12 vg each of a pooled library of AAV-encapsulated adaptamers. Two weeks later three animals were injected with two doses of 60mg/kg tetracycline 24 hours apart. All mice were harvested 4 hours after the last dose. Targeted RNAseq of adaptamer library in the liver was used to determine A-to-I editing rates of each construct in each mouse. Amplicons were amplified from cDNA and processed via Amplicon-EZ (Azenta Life Sciences). Primers: Adapt-Screen-F, with the N nucleotides varied for multiplexing, and Adapt-Screen-R. The ratio of reads with an A-to-I edit to the total number of reads for each construct was calculated to determine RNA editing rate.

FGF21 Gene Therapy

21-week-old DIO male C57BL/6J mice were purchased from Jackson Labs and injected with 2e11vg of AAV8 virus after one week of acclimatization. Mice were fed ad libitum with a high fat diet (60% calories from Fat, Bioserv). Animals were placed in weight matched groups on the day of viral injection and cages contained mixed groups to eliminate cage based effects.

FGF21 ELISA

Serum FGF21 was quantified using Mouse FGF21 DuoSet ELISA kit (R&D) and an ELISA Starter Kit (Bethyl Laboratories) according to manufacturer protocols. Briefly, a clear bottom 96 well ELISA plate was coated overnight with 100 μL of 4 μg/mL Capture Antibody diluted in PBS per well, washed 3 times with wash buffer (50 mM Tris buffered saline, pH 8.0, 0.05% Tween20; Sigma Chemical # T9039), then incubated for two hours with 300 μL Blocking Buffer (50 mM Tris buffered saline, pH 8.0, 1% BSA; Sigma Chemical # T6789). Plates were washed again three times, then 100 μL of sample (serum 1:100 dilution) or standard diluted in Sample Diluent (Blocking buffer +0.05% Tween20) and incubated at room temperature for two hours. Plates were washed as previously described then incubated for two hours with 100 μL per well of 100 ng/mL Detection Antibody in Sample Diluent. Plates were incubated with for 20 minutes in the dark with 100 μL of Streptavin-HRP after another set of washes, then incubated for 20 minutes in the dark with 100 μL of TMB One Component HRP Microwell Substrate after a final set of washes. Reactions were stopped using 0.16 M sulfuric acid stop solution and absorbance reading taken at 450 nm with a 570 nm correction using a SpectaMax iD3 plate reader (Molecular Devices). Values were interpolated using a four-parameter logistic curve fitted to the standard values, averaging technical duplicates.

Glucose Tolerance Test

Mice were fasted 8 hours overnight prior to testing. Blood and serum were collected immediately before intraperitoneal injection of 2 g/kg dextrose (Sigma-Aldrich) dissolved in PBS. For subsequent timepoints only a small drop of blood was collected. Glucose measurements were performed using a handheld glucose monitor (Metene).

DIO Mouse Histology

DIO mice were sacrificed 24 hours after last tetracycline dose and liver, interscapular brown adipose tissue (iBAT), and inguinal white adipose tissue (iWAT) were collected, briefly washed in ice cold PBS, then fixed in Neutral Buffered Formalin, 10% (SigmaAldrich) for 24 hours. After fixation samples were Paraffin-Embedded, sectioned, hematoxylin-eosin stained, and imaged at 20x magnification by the Microscopy and Histology Core of the La Jolla Institute of Immunology.

QUANTIFICATION AND STATISTICAL ANALYSIS

General statistical analysis was performed with Graphpad Prism, REDItools, or DESeq2 with details provided in figure captions. An unpaired, parametric two-sided Student’s t-test with a Welch post-correction was used for all statistical analyses except Fig. S2 and S3. Multiple unpaired t-tests with Holm-Šídák correction were performed for Fig. S2. Statistical significance was defined as p < 0.05. Data is represented as mean ± S.D. or S.E.M. specified in figure captions.

For RNA editing and differential gene expression in Fig. S3 the in package statistical analysis was used with significance defined as q < 0.1 for LoDEI, FDR <0.001 determined by Fisher’s Exact Test for REDItools, and both log2fold change >2 and adjusted p-value <0.05 for DESeq2.

Supplementary Material

Table S1: Plasmid and adaptamer DNA sequences, related to Figures 1–5.

NIHMS2217576-supplement-1.xlsx^{(55.8KB, xlsx)}

NIHMS2217576-supplement-2.pdf^{(2.8MB, pdf)}

A detailed protocol for constructing and testing adaptamer sequences in vitro is provided in Methods S1.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and virus strains
One Shot Stbl3 Chemically Competent E. coli	Invitrogen	C737303
Biological samples
Human Peripheral Blood Mononuclear Cells	Stemcell Technologies	# 70025.2
Chemicals, peptides, and recombinant proteins
Puromycin	ThermoFisher	A1113802
Lipofectamine 2000	ThermoFisher	11668030
KAPA Hifi Hotstart Ready Mix	Roche	7958935001
OptiMEM	ThermoFisher	31985062
ProtoScript^® II First Strand cDNA Synthesis Kit	NEB	E6560S
NEBNext Ultra^™ II RNA Library Prep Kit for Illumina	NEB	E7770L
NEBNext Poly(A) mRNA Magnetic Isolation Module	NEB	E7490S
Gibson Assembly Master Mix	NEB	E2611L
iTaq Universal SYBR Green Supermix	BioRad	1725120
Polybrene	MilliporeSigma	TR-1003-G
Triton X-100	Millipore-Sigma	X100-5ML
0.05% Trypsin-EDTA	ThermoFisher	#25300062
DMEM	ThermoFisher	10566016
EMEM	ATCC	30-2003
RPMI-1640	ThermoFisher	11875093
FBS	ThermoFisher	A5256801
rhIL-2, ACF Free	Stemcell Technologies	78145
1% Antibiotic-Antimycotic	ThermoFisher	15240062
PBS	ThermoFisher	10010023
ImmunoCult^™ Human CD3/CD28 T Cell Activator	Stemcell Technologies	10971
PEI 25K	Kyfora Bio	50-255-9826
Benzonase	MilliporeSigma	E1014-25KU
Optiprep	MilliporeSigma	D1556-250ML
AAV2 Standard	ATCC	VR-1616
Tetracycline hydrochloride	MilliporeSigma	T7660-5G
Doxycycline hydrochloride	MilliporeSigma	D3447-500MG
B-cyclodextrin	MilliporeSigma	332607-25G
d-Luciferin, Potassium Salt	GoldBio	Cat#LUCK-1G
Nano-Glo^® Luciferase Assay System	Promega	N1120
Dextrose	MilliporeSigma	G7021-1KG
Neutral Buffered Formalin, 10%	MilliporeSigma	HT501128-4L
Bright-Glo^™ Luciferase Assay System	Promega	E2610
RiboGreen^®	ThermoFisher	R11491
Purified ADAR1 p150 protein	BPS Bioscience	100472-1
HiScribe^® T7 High Yield RNA Synthesis Kit	NEB	E2040S
DTT	NEB	B1222AVIAL
RNAse Inhibitor	ThermoFisher	N8080119
Glo Lysis Buffer	Promega	E2661
Trizol	ThermoFisher	15596026
Polyethylene glycol	ThermoFisher	BP233-1
Chloroform	Fisher Scientific	AAJ67241AP
BamHI-HiFi	NEB	R3136S
SOC Medium	ThermoFisher	15544034
Carbenicillin Agar Plates	Teknova	L1010
Carbenicillin LB Broth	Teknova	L8185
Critical commercial assays
Mouse FGF21 DuoSet ELISA kit	R&D	DY3057
ELISA Starter Kit	Bethyl Laboratories	E101
Deposited Data
ADAR Feedback Loop RNAseq FASTQ Files	This Paper	BioProject PRJNA1526315
Experimental models: Cell lines
HEK293T	ATCC	Cat#CRL-3216; RRID:CVCL_0063
Neuro2a	ATCC	Cat#CCL-131; RRID:CVCL_0470
Experimental models: Organisms/strains
C57BL/6J	Jackson Labs	000664
C57BL/6J DIO	Jackson Labs	380050
Oligonucleotides
3xFlag_F: 5’GACTACAAAGACCACGACGGGGAC3’	IDT	25 nmole DNA Oligo
P2A_R: 5’CGGTCCAGGATTCTCTTCGACATC3’	IDT	25 nmole DNA Oligo
Adapt_Univ_R: 5’CTCGACCAGGATGGGCAC3’	IDT	25 nmole DNA Oligo
hADAR1-qPCR-F: 5’TCCGTCTCCTGTCCAAAGAAGG3’	IDT	25 nmole DNA Oligo
hADAR1-qPCR-R 5’TTCTTGCTGGGAGCACTCACAC3’	IDT	25 nmole DNA Oligo
hADAR2-qPCR-F: 5’TACATCAGCACCTCTCCCTGTG3’	IDT	25 nmole DNA Oligo
hADAR2-qPCR-R: 5’CAGTGAGCCATGATTGCACCAC3’	IDT	25 nmole DNA Oligo
hADARp150-only-F: 5’ATGAATCCGCGGCAGGGGTA3’	IDT	25 nmole DNA Oligo
hADARp150-only-R: 5’CGGTGCTTCTGGGAGCTGC3’	IDT	25 nmole DNA Oligo
qPCR_human_B-ACTIN_F: 5’CATGTACGTTGCTATCCAGGC3’	IDT	25 nmole DNA Oligo
qPCR_human_B-ACTIN_R: 5’CTCCTTAATGTCACGCACGAT3’	IDT	25 nmole DNA Oligo
AAV-ITR-F: 5’-CGGCCTCAGTGAGCGA-3’	IDT	25 nmole DNA Oligo
AAV-ITR-R: 5’-GGAACCCCTAGTGATGGAGTT-3’	IDT	25 nmole DNA Oligo
Adapt-Screen-F: 5’ACACTCTTTCCCTACACGACGCTCTTCCGATCTANNGAATTCGCCACCATGGACTACAAAG3’	IDT	25 nmole DNA Oligo
Adapt-Screen-R: 5’GACTGGAGTTCAGACGTGTGCTCTTCCGATCTCAGGCTCGACCAGGATGGGCAC3’	IDT	25 nmole DNA Oligo
T7-F: 5’CCAAGTAATACGACTCACTATAGGACACACACTACCAGAGTTAG3’	IDT	25 nmole DNA Oligo
IVT-R: 5’CTGTGTTACCAGAGCTGGG3’	IDT	25 nmole DNA Oligo
Recombinant DNA
pAAV-EGFP	This paper	Addgene #261288
pAdaptamer-AAV_V9-21_Tet_EGFP	This paper	Addgene #261289
pAdaptamer-AAV_V9-17_Dox_EGFP	This paper	Addgene #261290
ADAR1 shRNA	MilliporeSigma	TRCN0000050788
NTC shRNA	MilliporeSigma	SHC002
pADAR1p110	This paper	Addgene #261301
pLenti_WT-EGFP	This paper	Addgene #261291
pAdaptamer-Lenti_V9-21_EGFP	This paper	Addgene #261292
pAdaptamer-Lenti_V9-21_EGFP_V9-18_ADAR2	This paper	Addgene #261293
pAAV-WT-fLuc	This paper	Addgene #261294
pAdaptamer-AAV_V9-21_Tet_fLuc	This paper	Addgene #261295
pAAV-WT-sNLuc	This paper	Addgene #261296
pAdaptamer-AAV_V9-21_Tet_sNLuc	This paper	Addgene #261297
pAdaptamer-AAV_V9-18_Tet_sNLuc	This paper	Addgene #261298
pAdaptamer-AAV_V9-17_Dox_sNLuc	This paper	Addgene #261299
pAdaptamer-AAV_V9-18_Tet_FGF21	This paper	Addgene #261300
pMD2.G	Unpublished Trono Lab	Addgene #12259
pCMV delta R8.2	Unpublished Trono Lab	Addgene #12263
pHelper vector (pAdDeltaF6)	Unpublished Wilson Lab	Addgene #112867
pAAV2/9n	Unpublished Wilson Lab	Addgene #112865
pAAV2/8	Unpublished Wilson Lab	Addgene #112864
pUC19	Norrander et al. 1983⁴⁷	Addgene #50005
Software and algorithms
REDItools2	Picardi & Pesole, 2013⁴²	https://github.com/BioinfoUNIBA/REDItools2
RNAEditingIndexer (Alu Editing Index, AEI)	Roth et al., 2019⁴⁰	https://github.com/a2iEditing/RNAEditingIndexer
LoDEI	Torkler et al., 2024⁴¹	https://github.com/rna-editing1/lodei
BWA-MEM2	Open Source	https://github.com/bwa-mem2/bwa-mem2
DESeq2	Open Source	https://github.com/thelovelab/DESeq2
Biorender	Biorender	https://app.biorender.com/
GraphPad Prism	GraphPad	v11.0.0
Chromas	Technelysium	v2.6.6
Other
RNA Purification Kit	Qiagen	74104
PCR Purification Kit	Qiagen	28104
Miniprep Kit	Qiagen	27104
Plasmid Plus DNA Maxiprep Kit	Qiagen	12963
Monarch^® Spin RNA Cleanup Kit	NEB	T2040L
60% High fat diet	BioServ	S3282
BD FACSCanto flow cytometer	BD Biosciences	N/A
BD LSRFortessa flow cytometer	BD Biosciences	N/A
Andor Zyla 5.5 sCMOS	Andor	N/A
DMi8	Leica	N/A
Zeiss AxioZ1 Slide Scanner	Ziess	N/A
CFX Connect Real-Time PCR Detection System	BioRad	N/A
SpectraMax iD3 plate reader	Molecular Devices	N/A
Perkin Elmer IVIS Spectrum imaging system	PerkinElmer	N/A
100 kDa MWCO Ultra-Centrifugal Filter	MilliporeSigma	UFC9100
50 kDa MWCO Ultra-Centrifugal Filter	MilliporeSigma	UFC9050
Handheld glucose monitor	Metene	TD-4116

Open in a new tab

Highlights.

Compact (<120bp) adaptamer riboswitches use endogenous ADAR to control translation
Aptamer replacement yields an orthogonal doxycycline responsive adaptamer
ADAR feedback loops boost expression 9-fold without increasing background
Adaptamer-controlled AAV-FGF21 reverses obesity and steatosis in mice

ACKNOWLEDGEMENTS

We thank members of the Mali lab for discussions, advice, and help with experiments. We thank Sami Nourreddine, Claire Kuykendall, Michael Tong, and Joseph Rainaldi for useful discussions and support throughout all stages of the study. We thank Kristen Jepsen, Director of IGM Genomics Center UC San Diego. We thank Zbigniew Mikulski, Director of The La Jolla Institute for Immunology Microscopy & Histology Core. This work was generously supported by NIH grants (R01HG012351, R01CA310063, R01NS131560), a CIRM Grant (DISC4-19271), a Department of Defense Grant (W81XWH-22-1-0401), and UCSD Institutional Funds. Some schematics were created using BioRender.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

DECLARATION OF INTERESTS

The authors have filed a patent based on this work. P.M. is a scientific co-founder of Shape Therapeutics, Boundless Biosciences, Navega Therapeutics, Pi Bio, and Engine Biosciences. The terms of these arrangements have been reviewed and approved by the University of California San Diego in accordance with its conflict of interest policies.

REFERENCES

1.Clackson T. (2000). Regulated gene expression systems. Gene Ther. 7, 120–125. 10.1038/sj.gt.3301120. [DOI] [PubMed] [Google Scholar]
2.Sayed N, Allawadhi P, Khurana A, Singh V, Navik U, Pasumarthi SK, Khurana I, Banothu AK, Weiskirchen R, and Bharani KK. (2022). Gene therapy: Comprehensive overview and therapeutic applications. Life Sci. 294, 120375. 10.1016/j.lfs.2022.120375. [DOI] [PubMed] [Google Scholar]
3.Lim WA, and June CH. (2017). The principles of engineering immune cells to treat cancer. Cell 168, 724–740. 10.1016/j.cell.2017.01.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Kitada T, DiAndreth B, Teague B, and Weiss R. (2018). Programming gene and engineered-cell therapies with synthetic biology. Science 359. 10.1126/science.aad1067. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Le Guiner C, Stieger K, Toromanoff A, Guilbaud M, Mendes-Madeira A, Devaux M, Guigand L, Cherel Y, Moullier P, Rolling F, et al. (2014). Transgene regulation using the tetracycline-inducible TetR-KRAB system after AAV-mediated gene transfer in rodents and nonhuman primates. PLoS One 9, e102538. 10.1371/journal.pone.0102538. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Favre D, Blouin V, Provost N, Spisek R, Porrot F, Bohl D, Marmé F, Chérel Y, Salvetti A, Hurtrel B, et al. (2002). Lack of an immune response against the tetracycline-dependent transactivator correlates with long-term doxycycline-regulated transgene expression in nonhuman primates after intramuscular injection of recombinant adeno-associated virus. J. Virol. 76, 11605–11611. 10.1128/jvi.76.22.11605-11611.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Strobel B, Düchs MJ, Blazevic D, Rechtsteiner P, Braun C, Baum-Kroker KS, Schmid B, Ciossek T, Gottschling D, Hartig JS, et al. (2020). A small-molecule-responsive riboswitch enables conditional induction of viral vector-mediated gene expression in mice. ACS Synth. Biol. 9, 1292–1305. 10.1021/acssynbio.9b00410. [DOI] [PubMed] [Google Scholar]
8.Fukunaga K, Dhamodharan V, Miyahira N, Nomura Y, Mustafina K, Oosumi Y, Takayama K, Kanai A, and Yokobayashi Y. (2023). Small-molecule aptamer for regulating RNA functions in mammalian cells and animals. J. Am. Chem. Soc. 145, 7820–7828. 10.1021/jacs.2c12332. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Luo L, Jea JD-Y, Wang Y, Chao P-W, and Yen L. (2024). Control of mammalian gene expression by modulation of polyA signal cleavage at 5’ UTR. Nat. Biotechnol. 42, 1454–1466. 10.1038/s41587-023-01989-0. [DOI] [PubMed] [Google Scholar]
10.Düchs MJ, Kratzer RF, Vieyra-Garcia P, Strobel B, Schönberger T, Groß P, Aljayyoussi G, Gupta A, Lang I, Klein H, et al. (2024). Riboswitch-controlled IL-12 gene therapy reduces hepatocellular cancer in mice. Front. Immunol. 15, 1360063. 10.3389/fimmu.2024.1360063. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Bruckhoff RW, Becker O, Steinhilber D, and Suess B. (2025). Potential and optimization of mammalian splice riboswitches for the regulation of Exon skipping-dependent gene expression and isoform switching within the ALOX5 gene. ACS Synth. Biol. 14, 804–818. 10.1021/acssynbio.4c00731. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Booth BJ, Nourreddine S, Katrekar D, Savva Y, Bose D, Long TJ, Huss DJ, and Mali P. (2023). RNA editing: Expanding the potential of RNA therapeutics. Mol. Ther. 31, 1533–1549. 10.1016/j.ymthe.2023.01.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Rainaldi J, Mali P, and Nourreddine S. (2025). Emerging clinical applications of ADAR based RNA editing. Stem Cells Transl. Med. 14, szaf016. 10.1093/stcltm/szaf016. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Zhang Y, Bryant J, Herron L, and Mali P. (2025). ADARs: pleiotropy in function, versatility in application. Nucleic Acids Res. 53. 10.1093/nar/gkaf672. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Jacobsen CS, Salvador P, Yung JF, Kragness S, Mendoza HG, Mandel G, and Beal PA. (2023). Library screening reveals sequence motifs that enable ADAR2 editing at recalcitrant sites. ACS Chem. Biol. 18, 2188–2199. 10.1021/acschembio.3c00107. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Zambrano-Mila MS, Witzenberger M, Rosenwasser Z, Uzonyi A, Nir R, Ben-Aroya S, Levanon EY, and Schwartz S. (2023). Dissecting the basis for differential substrate specificity of ADAR1 and ADAR2. Nat. Commun. 14, 8212. 10.1038/s41467-023-43633-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Jiang K, Koob J, Chen XD, Krajeski RN, Zhang Y, Volf V, Zhou W, Sgrizzi SR, Villiger L, Gootenberg JS, et al. (2023). Programmable eukaryotic protein synthesis with RNA sensors by harnessing ADAR. Nat. Biotechnol. 41, 698–707. 10.1038/s41587-022-01534-5. [DOI] [PubMed] [Google Scholar]
18.Qin P-P, Chen P-R, Tan L, Chu X, Ye B-C, and Yin B-C. (2024). Programming ADAR-recruiting hairpin RNA sensor to detect endogenous molecules. Nucleic Acids Res., gkae1146. 10.1093/nar/gkae1146. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Yang X, Woldemichael K, Guo X, Zhao S, Qian Y, and Huang ZJ. (2025). CellREADR: An ADAR-based RNA sensor-actuator device. Methods Enzymol. 710, 207–227. 10.1016/bs.mie.2024.11.027. [DOI] [PubMed] [Google Scholar]
20.Kaseniit KE, Katz N, Kolber NS, Call CC, Wengier DL, Cody WB, Sattely ES, and Gao XJ. (2023). Modular, programmable RNA sensing using ADAR editing in living cells. Nat. Biotechnol. 41, 482–487. 10.1038/s41587-022-01493-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Katrekar D, Chen G, Meluzzi D, Ganesh A, Worlikar A, Shih Y-R, Varghese S, and Mali P. (2019). In vivo RNA editing of point mutations via RNA-guided adenosine deaminases. Nat. Methods 16, 239–242. 10.1038/s41592-019-0323-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.GBD 2021 US Obesity Forecasting Collaborators (2024). National-level and state-level prevalence of overweight and obesity among children, adolescents, and adults in the USA, 1990–2021, and forecasts up to 2050. Lancet 404, 2278–2298. 10.1016/S0140-6736(24)01548-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Kivimäki M, Strandberg T, Pentti J, Nyberg ST, Frank P, Jokela M, Ervasti J, Suominen SB, Vahtera J, Sipilä PN, et al. (2022). Body-mass index and risk of obesity-related complex multimorbidity: an observational multicohort study. Lancet Diabetes Endocrinol. 10, 253–263. 10.1016/S2213-8587(22)00033-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Anderson JM, Arnold WD, Huang W, Ray A, Owendoff G, and Cao L. (2024). Long-term effects of a fat-directed FGF21 gene therapy in aged female mice. Gene Ther. 31, 95–104. 10.1038/s41434-023-00422-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Jimenez V, Sacristan V, Jambrina C, Jaen ML, Casana E, Muñoz S, Marcó S, Molas M, Garcia M, Grass I, et al. (2024). Reversion of metabolic dysfunction-associated steatohepatitis by skeletal muscle-directed FGF21 gene therapy. Mol. Ther. 32, 4285–4302. 10.1016/j.ymthe.2024.10.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Jimenez V, Jambrina C, Casana E, Sacristan V, Muñoz S, Darriba S, Rodó J, Mallol C, Garcia M, León X, et al. (2018). FGF21 gene therapy as treatment for obesity and insulin resistance. EMBO Mol. Med. 10. 10.15252/emmm.201708791. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Zheng Z, Zong Y, Ma Y, Tian Y, Pang Y, Zhang C, and Gao J. (2024). Glucagon-like peptide-1 receptor: mechanisms and advances in therapy. Signal Transduct. Target. Ther. 9, 234. 10.1038/s41392-024-01931-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Kharitonenkov A, Shiyanova TL, Koester A, Ford AM, Micanovic R, Galbreath EJ, Sandusky GE, Hammond LJ, Moyers JS, Owens RA, et al. (2005). FGF-21 as a novel metabolic regulator. J. Clin. Invest. 115, 1627–1635. 10.1172/JCI23606. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Oost LJ, Kustermann M, Armani A, Blaauw B, and Romanello V. (2019). Fibroblast growth factor 21 controls mitophagy and muscle mass. J. Cachexia Sarcopenia Muscle 10, 630–642. 10.1002/jcsm.12409. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Tang Y, and Zhang M. (2023). Fibroblast growth factor 21 and bone homeostasis. Biomed. J. 46, 100548. 10.1016/j.bj.2022.07.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Liu J, Jiang J, Li Y, Chen Q, Yang T, Lei Y, He Z, Wang X, Na Q, Lao C, et al. (2024). Effects of FGF21 overexpression in osteoporosis and bone mineral density: a two-sample, mediating Mendelian analysis. Front. Endocrinol. (Lausanne) 15, 1439255. 10.3389/fendo.2024.1439255. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Kelvin D, and Suess B. (2023). Tapping the potential of synthetic riboswitches: reviewing the versatility of the tetracycline aptamer. RNA Biol. 20, 457–468. 10.1080/15476286.2023.2234732. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Fritzell K, Xu L-D, Otrocka M, Andréasson C, and Öhman M. (2019). Sensitive ADAR editing reporter in cancer cells enables high-throughput screening of small molecule libraries. Nucleic Acids Res. 47, e22. 10.1093/nar/gky1228. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Hoetzel J, Walbrun A, Schäfer M, Wang T, Jørgensen AG, Becker O, Stamatakis K, Gunawan V, Reichardt L, Boettger L, et al. (2026). Mechanism underlying the high regulatory performance of the doxycycline riboswitch G12. Nat. Commun. 17, 7697. 10.1038/s41467-026-76256-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Blackford N, Ogunkunle VS, and Silvers R. (2025). Fluorescent dyes for differential scanning fluorimetry of RNA. Chembiochem 26, e202500560. 10.1002/cbic.202500560. [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Silvers R, Keller H, Schwalbe H, and Hengesbach M. (2015). Differential scanning fluorimetry for monitoring RNA stability. Chembiochem 16, 1109–1114. 10.1002/cbic.201500046. [DOI] [PubMed] [Google Scholar]
37.Sztuba-Solinska J, Shenoy SR, Gareiss P, Krumpe LRH, Le Grice SFJ, O’Keefe BR, and Schneekloth JS Jr, (2014). Identification of biologically active, HIV TAR RNA-binding small molecules using small molecule microarrays. J. Am. Chem. Soc. 136, 8402–8410. 10.1021/ja502754f. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Gayet RV, Ilia K, Razavi S, Tippens ND, Lalwani MA, Zhang K, Chen JX, Chen JC, Vargas-Asencio J, and Collins JJ. (2023). Autocatalytic base editing for RNA-responsive translational control. Nat. Commun. 14, 1339. 10.1038/s41467-023-36851-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Schaffer AA, Kopel E, Hendel A, Picardi E, Levanon EY, and Eisenberg E. (2020). The cell line A-to-I RNA editing catalogue. Nucleic Acids Res. 48, 5849–5858. 10.1093/nar/gkaa305. [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Roth SH, Levanon EY, and Eisenberg E. (2019). Genome-wide quantification of ADAR adenosine-to-inosine RNA editing activity. Nat. Methods 16, 1131–1138. 10.1038/s41592-019-0610-9. [DOI] [PubMed] [Google Scholar]
41.Torkler P, Sauer M, Schwartz U, Corbacioglu S, Sommer G, and Heise T. (2024). LoDEI: a robust and sensitive tool to detect transcriptome-wide differential A-to-I editing in RNA-seq data. Nat. Commun. 15, 9121. 10.1038/s41467-024-53298-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Picardi E, and Pesole G. (2013). REDItools: high-throughput RNA editing detection made easy. Bioinformatics 29, 1813–1814. 10.1093/bioinformatics/btt287. [DOI] [PubMed] [Google Scholar]
43.Davidsohn N, Pezone M, Vernet A, Graveline A, Oliver D, Slomovic S, Punthambaker S, Sun X, Liao R, Bonventre JV, et al. (2019). A single combination gene therapy treats multiple age-related diseases. Proc. Natl. Acad. Sci. U. S. A. 116, 23505–23511. 10.1073/pnas.1910073116. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.George CX, and Samuel CE. (1999). Human RNA-specific adenosine deaminase ADAR1 transcripts possess alternative exon 1 structures that initiate from different promoters, one constitutively active and the other interferon inducible. Proc. Natl. Acad. Sci. U. S. A. 96, 4621–4626. 10.1073/pnas.96.8.4621. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Laxminarayana D, Khan IU, O’Rourke KS, and Giri B. (2007). Induction of 150-kDa adenosine deaminase that acts on RNA (ADAR)-1 gene expression in normal T lymphocytes by anti-CD3-epsilon and anti-CD28. Immunology 122, 623–633. 10.1111/j.1365-2567.2007.02709.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Hou Q, Oleynikov M, Mei X, Dong L, Hagen T, and Jaffrey SR. (2025). A portable poison exon for small-molecule control of mammalian gene expression. Nat. Methods 22, 2337–2348. 10.1038/s41592-025-02860-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Norrander J, Kempe T, and Messing J. (1983). Construction of improved M13 vectors using oligodeoxynucleotide-directed mutagenesis. Gene 26, 101–106. 10.1016/0378-1119(83)90040-9. [DOI] [PubMed] [Google Scholar]
48.Ran FA, Cong L, Yan WX, Scott DA, Gootenberg JS, Kriz AJ, Zetsche B, Shalem O, Wu X, Makarova KS, et al. (2015). In vivo genome editing using Staphylococcus aureus Cas9. Nature 520, 186–191. 10.1038/nature14299. [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Ford KM, Panwala R, Chen D-H, Portell A, Palmer N, and Mali P. (2021). Peptide-tiling screens of cancer drivers reveal oncogenic protein domains and associated peptide inhibitors. Cell Syst. 12, 716–732.e7. 10.1016/j.cels.2021.05.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Crosson SM, Dib P, Smith JK, and Zolotukhin S. (2018). Helper-free production of laboratory grade AAV and purification by iodixanol density gradient centrifugation. Mol. Ther. Methods Clin. Dev. 10, 1–7. 10.1016/j.omtm.2018.05.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Moreno AM, Palmer N, Alemán F, Chen G, Pla A, Jiang N, Leong Chew W, Law M, and Mali P. (2019). Immune-orthogonal orthologues of AAV capsids and of Cas9 circumvent the immune response to the administration of gene therapy. Nat. Biomed. Eng. 3, 806–816. 10.1038/s41551-019-0431-2. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials