Nitro Reduction‐Based RNA Control and Ultrafast Release

ABSTRACT

RNA protection and controlled release are critical for both fundamental research and therapeutic applications, yet the development of simple, efficient, and reversible post‐synthetic RNA modification strategies remains a significant challenge. Here, we introduce a straightforward approach based on ribose 2′‐hydroxyl acylation with a nitro‐functionalized carbamate, which acts as a redox‐responsive center. This modification selectively inhibits native RNA function while remaining inert to endogenous biogenic reductants and common reducing agents used in biological assays. Upon treatment with low millimolar concentrations of the diboron‐bipyridine reducing pair, RNA function is rapidly restored on a minute timescale. This methodology is broadly applicable across diverse RNA classes and functional contexts, including synthetic RNA oligomers, fluorogenic RNA aptamers, single guide RNAs in CRISPR–Cas9 gene‐editing systems, and mRNAs in living cells for translation control and RNAi‐mediated gene silencing. These results demonstrate the general utility of this approach as a chemically controllable functional switch, providing a versatile toolkit for temporal regulation of RNA activity in both research and biotechnological applications.

A chemical RNA caging–uncaging strategy based on nitro reduction is described. Nitroaryl groups introduced via acylation temporarily block the functions of diverse RNAs in vitro and in living cells. Treatment with a diboronic acid–bipyridine mixture rapidly reduces the nitro groups, releasing RNAs and restoring their activity. This approach is applicable to RNAs of any length, including short synthetic RNA oligomer, short hairpin RNA, RNA aptamer, single guide RNA, and mRNA.

1. Introduction

RNA is a versatile and essential class of biological macromolecules that plays central roles in protein translation [1, 2], regulation of gene expression [3, 4, 5], and diverse aspects of cellular function [6]. Its structural and functional plasticity underlies broad opportunities in both fundamental research and clinical applications. The development of RNA‐based therapeutics—including small interfering RNA (siRNA), antisense oligonucleotide (ASO), microRNA, messenger RNA (mRNA), CRISPR‐associated single guide RNA (sgRNA) [7, 8, 9, 10]—has advanced rapidly, with multiple cases of approval by the U.S. Food and Drug Administration (FDA) [11]. To enhance stability, reduce immunogenicity, and improve pharmacological performance, extensive efforts have been devoted to chemical modification of RNA, targeting the nucleobase (e.g., pseudouridine) [12, 13], the ribose moiety (e.g., 2’‐methoxy, or 2’‐OMe) [14] and the phosphate backbone (e.g., phosphorothioate linkage) [15, 16].

In the context of biomolecular chemical modification, most established strategies rely on stable, irreversible transformations that either enhance molecular stability or efficacy or enable probing and labeling of specific biological interactions [17, 18, 19]. In contrast, reversible modification introduces functionally disruptive moieties that act as temporary “locks”, attenuating native activity. Upon exposure to exogenous or endogenous stimuli, selective bond cleavage unlocks the biomolecule and restores its functionality [20, 21, 22, 23]. This stimulus‐responsive reversibility expands opportunities for precision medicine and has been broadly applied in peptide‐ and protein‐based therapeutics [24, 25]. In comparison, analogous strategies for RNA remain underdeveloped. The discovery of highly responsive, bioorthogonal, and RNA‐compatible reversible chemistries is therefore of substantial importance for advancing nucleic acid–based therapeutic modalities.

Although photo‐responsive [26], enzyme‐triggered [27] and click chemistry‐induced [28] strategies for reversible RNA protection have been reported, their broader implementation has been constrained by the need for synthetic elaboration to modified monomers and their applicability primarily to short RNA oligomers [29]. In contrast, post‐synthetic RNA modification enables installation of functional groups without complex phosphoramidite preparation, offering clear advantages in cost, scalability, and accessibility. Leveraging the high nucleophilicity of the ribose 2′‐hydroxyl (2’‐OH) group [30, 31, 32], N‐acyl imidazole–mediated 2’‐OH acylation has emerged as a powerful platform for RNA probing and engineering, including structural mapping [33, 34, 35] and proximity labeling [36, 37, 38]. Such acylation markedly attenuates RNA function by sterically perturbing secondary structure formation and protein recognition relative to the native state. These features render 2′‐O‐acylation an attractive scaffold for the development of stimulus‐responsive, control–release strategies.

Pioneering work by Kool and co‐workers introduced cleavable RNA acylation systems reversible either through phosphine‐mediated Staudinger reduction [39] or blue‐light‐induced Norrish type II photodeprotection [40], representing a significant advance in post‐synthetic reversible RNA acylation. However, these approaches exhibit notable limitations: photoresponsive acylation does not achieve complete substrate conversion, whereas phosphine‐triggered deprotection proceeds relatively slowly and relies on toxic P(III) reagents, restricting applications in living systems. More recently, the Zhu group reported thiol‐responsive reversible RNA acylation applicable to diverse RNA substrates, demonstrating its practical utility [41, 42]. This strategy exploits endogenous thiols, such as L‐glutathione (GSH), enabling spontaneous deprotection under conditions of elevated intracellular GSH levels. Despite these advances, the development of an orthogonal reversible acylation platform featuring rapid installation and deprotection kinetics, triggered by exogenous stimuli and unaffected by endogenous metabolites or biomolecules, remains an unmet challenge for RNA manipulation in biological contexts.

The nitro group, characterized by a high oxidation state of nitrogen, is readily reduced to the corresponding aniline when attached to aromatic systems, a transformation widely exploited in prodrug design [43, 44]. Nitroaromatic caging motifs responsive to nitroreductase (NTR)—an enzyme upregulated in hypoxic tumor cells but minimally active under normoxic conditions—have been extensively investigated for selective small‐molecule activation [45, 46, 47]. Despite promising preclinical results, clinical translation of such systems has thus far remained limited [48].

In parallel, a new class of diboron–bipyridine reducing system has emerged as a simple, efficient and chemoselective platform for nitroaromatic reduction under mild conditions [49, 50]. Notably, the Chu group demonstrated that water‐soluble tetrahydroxydiboron (THDB, A) in combination with 4,4′‐bipyridine (BIPY, B) enables effective prodrug activation in cultured cells and mice models in vivo, highlighting its biocompatibility and functional robustness in biological environments [51]. Inspired by the reductive conversion of para‐nitroaryl groups to anilines, we proposed coupling of this transformation with a self‐immolative linker architecture bearing a benzyl leaving group (Figure 1A). Reduction of the nitro group to the corresponding amine at the para position is expected to initiate a 1,6‐elimination cascade, driven by entropic gain associated with carbon dioxide release from a carbonate linkage. The rapid kinetics of this decarboxylative fragmentation suggest its suitability as a trigger for an ultrafast, post‐synthetic, control–release RNA modification platform.

FIGURE 1.

Chemical discovery of p‐nitrobenzyl imidazole‐2‐amide carbamate (PIC) and concept validation of nitro‐reduction responsive RNA acylation strategy. (A) Schematic design and proposed mechanism of RNA 2’‐OH acylation with nitrobenzyl‐based acylating agents and reductive release by tetrahydroxydiboron (THDB) and 4,4’‐bipyridine (BIPY). Structure of leaving group moieties in 1, 2, and PIC is symbolized by imidazole substituted with R group. (B) Performance screening of nitro‐based acylation agent with different leaving group on model 18‐mer FAM‐RNA. (C) Synthetic approach of PIC preparation by chloroformate 3 and imidazole‐2‐amide 4. (D) Screening of chloroformate‐imidazole ratio impact on PIC acylation performance on model 18‐mer RNA with 5’ 6‐FAM (fluorescein) fluorescent tag. Conversion in (B) and (D) was evaluated from Matrix‐Assisted Laser Desorption/Ionization—Time‐Of‐Flight (MALDI‐TOF) mass spectrometry. (E) Heat map of 18‐mer FAM‐RNA modification number (by integration of MALDI peak area and with largest peak normalized to 100%) when treated with 20, 50, and 100 mM PIC and (F) gel electrophoresis visualization of acylated FAM‐RNA. All acylation experiment was conducted in buffer‐free, RNase‐free water at room temperature (25°C) for 120 min, with final x(DMSO) = 20%.

2. Results and Discussion

2.1. Nitro Reduction‐Based RNA Acylation and Release

To validate the proposed control‐release strategy, we first designed a redox‐responsive acylating agent and evaluated suitable leaving groups for installing a carbonate linkage onto RNA oligomers. The reagent architecture comprises three functional elements: (i) a redox‐responsive nitro group, (ii) a self‐immolative benzyl alcohol scaffold capable of 1,6‐elimination, and (iii) a leaving group‐bearing carbamate to enable RNA acylation. Guided by precedent leaving group design, we synthesized candidate reagents bearing imidazole and 2‐chloroimidazole as leaving groups (1 and 2, respectively; Figure 1A), each prepared as 500 mM stock solutions in anhydrous DMSO. An 18‐nucleotide (nt) fluorescein (FAM)‐labeled RNA oligonucleotide was selected as a model substrate to assess acylation efficiency. However, treatment with 100 mM of 1 or 2 at room temperature resulted in minimal conversion, as determined by Matrix‐Assisted Laser Desorption/Ionization—Time‐Of‐Flight (MALDI‐TOF) mass spectrometry (Figures 1B and S1). Substantial precipitation was observed upon addition of the acylating agents to the aqueous RNA solution, suggesting poor solubility and consequently low effective concentrations in contact with the oligonucleotide.

To address this limitation, we adopted a water‐soluble imidazole‐based leaving‐group strategy [52] by incorporating a tertiary amine side chain to enhance hydrophilicity. This hydrophilic imidazole was subsequently incorporated into the para‐nitrobenzyl carbamate scaffold to generate para‐nitrobenzyl imidazole‐2‐amide carbamate (PIC) (Figure 1C). Addition of PIC to aqueous RNA solutions produced markedly less precipitation and enabled near‐complete conversion of the RNA substrate (Figure 1B). Notably, the increased polarity and electrophilicity of PIC rendered it incompatible with reverse‐phase chromatography following synthesis. We therefore optimized the preparation by tuning the stoichiometry between imidazole‐2‐amide (4) and para‐nitrobenzyl chloroformate (3) (Figure 1C). Unexpectedly, very limited acylation occurred when the reagents were combined in an approximate 1:1 ratio with slight excess of chloroformate 3 during synthesis, whereas use of 20%–100% excess imidazole 4 afforded full conversion. Optimal results were obtained with 1.6 equiv. of imidazole 4, yielding a median acylation number of +6 on the 18‐mer RNA at 100 mM reagent concentration and room temperature (Figures 1D and S2). This optimized batch was subsequently used for further studies. Concentration‐dependent RNA modification was then evaluated by treating the oligonucleotide with 20, 50, and 100 mM PIC, resulting in median modification levels of +1, +3, and +6, respectively, as determined by MALDI‐TOF analysis (Figures 1E and S3). These trends were corroborated by 20% denaturing polyacrylamide gel electrophoresis (PAGE), which showed progressive mobility shifts consistent with increasing degrees of acylation (Figure 1F).

Following purification of the acylated 18‐mer model RNA, we evaluated reductive deprotection under biologically relevant conditions using both enzymatic and chemical approaches. Treatment with recombinant Escherichia coli nitroreductase (NTR), even at 500 µg/mL in the presence of 5 mM reduced nicotinamide adenine dinucleotide (NADH) in 1×PBS, resulted in incomplete reduction after 24 h at 37°C (Figure S4A). We next investigated whether a lower degree of acylation could improve release efficiency. A previously reported acylation strategy using loop‐inducing complementary helper DNA [53, 54] was applied to direct PIC modification to an exposed 3‐nt loop site in a 19‐mer model RNA. This reaction generated distinct species with +0 to +3 acyl modifications. Subsequent treatment with NTR and NADH partially cleaved the PIC modifications, although complete recovery was not achieved (Figure S4B). The limited enzymatic turnover is likely due to steric shielding of the nitroaryl moiety by the bulky RNA structure, suggesting that redesign of self‐immolative systems with optimized length, orientation, and rigidity may be required for this purpose. In contrast, exposure to a chemical reducing system comprising 4 mM A and 1 mM B, premixed in a 4:1 stoichiometric ratio as previously described [51], generated a highly reactive intermediate [49] (Figure 2A) and led to rapid, quantitative restoration of the native RNA within 20 min (Figure 2B). Complete conversion was confirmed by MALDI‐TOF mass spectrometry (Figure 2C), demonstrating efficient RNA release at low millimolar concentrations of small‐molecule reductants on a remarkably short timescale while maintaining the simplicity of PIC reagent design.

FIGURE 2.

Ultra‐fast RNA reduction‐elimination when treated with reducing agent. (A) Structure and stoichiometry of THDB (A) and BIPY (B) and chemical structure of proposed intermediate. (B) Illustration of in vitro RNA reduction when treated with 4 mM THDB and 1 mM BIPY pre‐mix. (C) MALDI‐TOF of 100 mM acylated RNA before and after reduction with 4 mM THDB and 1 mM BIPY for 20 min in 1×PBS, 37°C. Red and blue numbers stand for modification numbers on one single RNA molecule (calculated by mass difference, each modification account for around +180). (D) and (E) Heat map regarding reduction kinetics of 100 mM PIC‐acylated FAM‐RNA at 1, 5, and 20 min (by integration of MALDI peak area and with largest peak normalized to 100%) when treated with mixture of (D) 4 mM A and 1 mM B or (E) 2 mM A and 0.5 mM B. (F) Observed rate constant with respect to concentration of modification on RNA at 4/1 mM and 2/0.5 mM A/B, respectively. Fundamental assumptions, complete derivation and rate function were discussed in the supporting information. (G) Reducing condition (agent type and concentration) and (H) denaturing polyacrylamide electrophoresis analysis demonstrate bioorthogonality of acyl group reduction of the PIC FAM‐RNA after incubation for 20 min in 1×PBS, 37°C. Conc., concentration. NC, unmodified FAM‐RNA; PC, 100 mM PIC acylated FAM‐RNA. All other labels symbolize different reductant treatment.

The reaction kinetics were further quantitatively validated by flash‐cooling the reaction mixture during reduction, followed by precipitation to obtain time‐dependent snapshots of the reaction. These samples were analyzed by peak integration using MALDI‐TOF mass spectrometry, confirming ultrafast reduction at the original reaction concentration (Figures 2D and S5) and even after a two‐fold dilution of the small‐molecule reagents to 2 mM A and 0.5 mM B (Figures 2E and S6). The reaction dynamics were further visualized by PAGE analysis and compared with previously reported phosphine‐mediated release under the same effective reductant concentration [39], highlighting the advantage of this new method in terms of rapid reaction kinetics at low millimolar concentrations (Figure S7). Under the equilibrium approximation, kinetic data were derived from the concentration of PIC modification, calculated as the product of the modified RNA concentration, 10 µM, and the average number of PIC modifications determined by MALDI‐TOF peak integration. Based on pseudo‐first‐order kinetic analysis (Tables S5 and S6), the observed rate constants were determined to be 3.5 × 10^{− 3} s^{− 1} at 1.0 mM B and 1.8 × 10^{− 3} s^{− 1} at 0.5 mM B (Figure 2F). These results are consistent with the equilibrium approximation and pseudo‐first‐order assumption, thereby providing a quantitative description of the reaction kinetics.

To assess orthogonality and stability, PIC‐modified RNA was incubated with A or B individually, as well as with biologically relevant reductants—including dihydronicotinamide derivatives, thiols, and ascorbate—under identical conditions (except for B concentration) for 20 min (Figure 2G). No detectable deprotection was observed with any single component or endogenous reducing agent. Only the combined A/B system effected complete restoration of the native RNA (Figure 2H). These results establish PIC‐mediated RNA acylation as a bioorthogonal modification platform that is selectively and rapidly reversed by the A/B reducing pair, while remaining inert to endogenous metabolites and common biochemical assay reagents.

2.2. In Vitro RNA Functional Control‐Recovery

Having established that PIC acylation of a model 39‐nt RNA disrupts Watson–Crick base pairing—as evidenced by melting temperature analysis—and that this effect is fully reversible upon reductive triggering (Figure S8), we next evaluated functional control and recovery in a structured RNA system. Because PIC installation perturbs both intra‐ and intermolecular interactions, we anticipated that RNA function would be suppressed upon acylation and restored only after treatment with the A/B reducing system. Fluorogenic RNA aptamers provide a convenient platform to assess this hypothesis. These structured RNAs bind cognate small‐molecule dyes and, upon proper folding, induce strong fluorescence. Steric perturbation of the aptamer scaffold is expected to disrupt folding and abolish fluorescence, whereas reductive removal of the acyl groups should restore native structure and signal (Figure 3A). We selected the 104‐nt green fluorogenic aptamer F30‐Pepper as a model system (Figure 3B), which, upon binding of defined structural scaffold with HBC530, emits green fluorescence (Figure 3C) [55, 56]. Following in vitro transcription, the aptamer was treated with 20, 50, or 100 mM PIC (DMSO alone as control), purified by spin column, quantified, and incubated in 1×folding buffer containing 5 µM HBC530. Fluorescence imaging showed a clear concentration‐dependent decrease in fluorescence intensity upon acylation in water (Figure 3D). However, when the aptamer was properly folded with additional HBC530 in folding buffer, fluorescence inhibition was negligible even at 100 mM PIC (Figure S9), suggesting decreased acylation on less accessible secondary RNA structural scaffold and agreeing with previous studies [30, 35]. Reductive treatment of the modified aptamer with 4 mM A and 1 mM B led to time‐dependent fluorescence recovery, as visualized at 1, 5, 20, and 60 min (Figure 3E). Quantitative analysis using a microplate reader, based on full emission spectra (Figure S10), confirmed both the concentration‐dependent suppression of fluorescence (Figure 3F) and the rapid kinetics of functional restoration (Figure 3G). At 100 mM PIC, fluorescence was reduced to <10% of the native signal, whereas exposure to the A/B system restored full fluorescence intensity within 5 min. These results demonstrate efficient, reversible shutdown and ultrafast recovery of RNA function through PIC‐mediated acylation and chemically triggered deprotection.

FIGURE 3.

F30‐pepper aptamer fluorescence control. (A) Demonstration of RNA control‐release system application on fluorogenic RNA aptamers. (B) Sequence and structural information of 104‐nt F30‐pepper aptamer. (C) Chemical structure of pepper aptamer ligand, HBC530. (D,E) Visualization of equal amount F30‐pepper aptamer fluorescence with respect to (D) PIC concentration and (E) reduction time course on gel scanner, illustrating clear concentration‐dependent inhibition by acylation and rapid diboron‐bipyridine reduction‐mediated recovery. Data are representatives of three replicate experiments. (F,G) Quantification of samples in (D) and (E) respectively by fluorescence emission spectra on microplate reader from 525 to 545 nm. Acylation of (D) and (F) was conducted in buffer‐free water at room temperature (25°C) for 120 min. Reduction of (E) and (G) was conducted with 4/1 mM THDB and BIPY in 1×PBS, 37°C. Ctrl: HBC530 only. DMSO: mock treatment of 20% DMSO. Modified, 50 mM PIC acylated aptamer. Data are presented as the mean ± SEM of n = 3 independent experiments.

As a foundational gene‐editing platform, CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) technology, adapted from bacterial adaptive immune systems, has been extensively engineered and deployed across diverse biological settings [57, 58]. In the canonical system, the endonuclease Cas9 is directed by sgRNA, which comprises a Cas9‐binding scaffold and a DNA‐targeting spacer sequence. Through Watson–Crick base pairing between the spacer and the complementary genomic DNA strand, Cas9 is positioned to introduce site‐specific double‐strand breaks via its nuclease‐active catalytic domains. We reasoned that installation of the control–release modification onto sgRNA would disrupt both Cas9 binding and target DNA hybridization through steric interference from acylation, thereby suppressing gene‐editing activity. Subsequent treatment with the A/B reducing system should restore sgRNA structure and function in situ. To test this hypothesis, we selected a 4 kilobase (kb) double‐stranded DNA substrate as the target and designed a corresponding 131‐nt sgRNA based on the Neisseria meningitidis Cas9 (Nme1Cas9) platform (Figure 4A). Cas9 protein was preincubated with either native or PIC‐modified sgRNA prior to exposure to the DNA substrate. Experimental groups included an untreated DNA negative control (NC) and groups 1–6 corresponding to native sgRNA, PIC‐modified sgRNA, and modified sgRNA subjected to varying concentrations of reducing agents (Figure 4B). Cleavage efficiency was quantified by agarose gel electrophoresis, comparing the intensity of integral DNA to the combined intensities of truncated products (Figure 4C). Direct treatment of the reaction mixture with A and B for 20 min resulted in a concentration‐dependent restoration of nuclease activity. At the highest reducing‐agent concentration (group 6), editing efficiency was fully recovered and statistically indistinguishable from that of native sgRNA (group 1). In contrast, treatment with A or B alone failed to reinstate sgRNA function (Figure S11). Collectively, these results demonstrate that A/B reduction‐mediated reversible PIC acylation is highly compatible with the CRISPR–Cas9 system, enabling temporally controlled and rapidly activatable gene‐editing activity. This strategy holds promise as a chemically gated toolkit for precision genome engineering.

FIGURE 4.

CRISPR‐Cas9 system functional control by sgRNA modification and release. (A) Schematic illustration of downstream effect of original and modified sgRNA on Cas9 enzyme binding and activity. (B) Outline of different entries components of sgRNA function experiment in (C) and (D). All experiments were treated one‐pot, and the reduction was incubated for 20 min at room temperature (25°C). Modified sgRNA was prepared in advance by reaction with 50 mM PIC acylating agent (10% DMSO) in water for 60 min, purified by spin column and then proceeded with subsequent CRISPR experiments. (C) Gel electrophoresis visualization showing the reductant concentration‐dependent trend of target dsDNA cleavage (with reference to target and product fragment DNA length). Data are representatives of three replicate experiments. (D) Quantification of relative percentage of target cleavage based on band intensity percentage of gel electrophoresis. While modification on sgRNA exhibit significant deactivation of the CRISPR system, 4 mM A/1 mM B released sgRNA recovers cleavage capability to comparable level with positive control group. Data are presented as the mean ± SEM of n = 3 independent experiments and p‐values: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, not significant.

2.3. In Cellulo mRNA Function Recovery

Having established proof‐of‐concept on short‐ to medium‐length (<200 nt) RNAs in vitro, we next extended the strategy to longer transcripts. Messenger RNAs encoding fluorescent proteins, such as enhanced green fluorescent protein (EGFP), with typically around 1000 nt in length, offer convenient readouts for visualization and quantification both in vitro and in cellulo (via lipid‐mediated transfection). We therefore selected EGFP mRNA (mEGFP) as a model substrate to evaluate inhibition–recovery of long‐RNA function. Translation of mRNA requires coordinated intermolecular interactions, including binding of poly(A)‐binding protein (PABP) to the poly(A) tail and assembly of the eIF4F initiation complex [2, 59]. We anticipated that PIC‐mediated acylation would sterically disrupt these interactions and suppress translation, with restoration achievable upon treatment with the A/B reducing pair (Figure 5A). Initial treatment of mEGFP with 50 mM PIC led to poor recovery during purification (<30% yield), likely reflecting substantial changes in RNA hydrophobicity that compromised spin‐column retention. Because 20 mM PIC afforded incomplete modification (Figure 1E), we selected 35 mM PIC for subsequent in vitro studies.

FIGURE 5.

mRNA in vitro and in cellulo control‐recovery. (A) Mechanism of translation initiation and inhibition for unmodified (or released) and modified mRNA. (B) Intensity of EGFP mRNA (mEGFP) in vitro translation. For unmodified (mock treated with 7% DMSO only), modified (acylated with 35 mM PIC, 7% DMSO) and released (reduction with 4 mM A and 1 mM B for 20 min, 37°C) groups, mRNA was purified with spin column, and an equal amount of respective mRNA was injected into in vitro translation system and incubated at room temperature (25°C) for 120 min. (C) Confocal microscopy of HeLa cells transfected with unmodified (mock‐treated) or PIC modified (35 mM) mEGFP for 6 h and treated with blank treatment (1×PBS) or mixture of 4 mM A and 1 mM B in 1×PBS for 120 min, then further cultured for 16 h. Scale bar: 50 um. Images are representative of three independent experiments. (D) Flow cytometry histogram demonstration and (E) quantification of median cell fluorescence of unmodified, modified and in cellulo released mEGFP signal. NC, cells without any transfection. PC, cells transfected with unmodified mEGFP. 25, 30, 35, cells transfected with mEGFP modified with 25, 30, 35 mM PIC. SM, cells released with reducing small molecule treatment after transfection of the respective PIC‐modified mEGFP. Data are presented as the mean ± SEM of n = 3 independent experiments and p‐values: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, not significant.

Following reductive treatment and repurification, equal amounts of native, acylated, or released mEGFP were subjected to a wheat germ extract translation system, and EGFP fluorescence was quantified at 530 nm. As expected, acylation reduced translation to <10% of native levels. Reductive treatment with 4 mM A and 1 mM B partially restored fluorescence to ∼55% of the original signal (Figure 5B). To assess RNA integrity, unmodified, modified, and released mEGFP samples were analyzed on 1% agarose gels. No smearing or additional bands were observed (Figure S12), indicating the absence of detectable degradation during either acylation or reduction. The incomplete functional recovery may arise from imperfect refolding of the long transcript after acylation or from residual modification persisting on a subset of sites. Nonetheless, fluorescence in the released group was significantly increased relative to the acylated group (ANOVA, p < 0.0001), supporting further evaluation in cellular systems. For in cellulo studies, HeLa cells were transfected with native or PIC‐modified mEGFP. Six hours post‐transfection, cells were treated with 4 mM A and 1 mM B for 120 min and then incubated for an additional 16 h to allow protein expression. Initially, A and B stocks were prepared in DMSO, but this resulted in minimal recovery—likely due to the mildly oxidative nature of DMSO interfering with the diboron reducing system. Switching to aqueous stock solutions resolved this issue (Figure S13). Under optimized conditions, confocal microscopy revealed strong suppression of EGFP expression in the modified group, with clear restoration upon treatment with the reducing small molecules under cellular conditions (Figure 5C). Further experiments demonstrated that PIC‐modified mRNA exhibits stability comparable to, if not greater than, that of a benchmark small‐molecule‐modified mRNA (Figure S14) [31]. For quantitative analysis, cells were transfected with unmodified or 25, 30, or 35 mM PIC‐modified mEGFP, with or without A/B treatment, and analyzed by flow cytometry after 16 h (Figure 5D). EGFP expression decreased in a PIC concentration–dependent manner, whereas A/B treatment shifted the fluorescence distribution toward higher intensity within the EGFP‐positive population. Analysis of median GFP fluorescence (Figure 5E) confirmed substantial restoration of expression by the reducing pair. Collectively, these results demonstrate that PIC‐mediated reversible acylation can modulate the function of long mRNA transcripts and enable chemically triggered reactivation in vitro and in living cells. Although further optimization may be required to achieve higher recovery efficiencies for long RNAs, this platform establishes a foundation for temporally controlled regulation of mRNA translation.

2.4. Caged Short Hairpin RNA as Controllable RNA Interference Strategy

Encouraged by the successful control–release of RNA function in living cells, we further validated by quantitative real‐time polymerase chain reaction (qRT‐PCR) that PIC‐modified RNA oligomers exhibit reduced immunostimulatory activity during transfection compared with unmodified RNA, evidenced by weaker upregulation of RNA sensing proteins associated with innate immune response pathways (Figure S15), supporting the potential applicability of PIC‐modified RNA in cell‐based strategies. We next explored the applicability of this platform as a chemical caging strategy for gene‐specific silencing. RNA interference (RNAi) relies on cellular processing of synthetic RNA into double‐stranded duplexes that guide formation of the RNA‐induced silencing complex (RISC), ultimately leading to RNase‐mediated degradation of target mRNA (Figure 6A). Although small interfering RNAs (siRNAs) are widely used for this purpose, their duplex structure renders them relatively inert to 2′‐OH acylation [30, 60], and separate strand modification would likely compromise duplex reformation and functional recovery in cells. We therefore selected short hairpin RNA (shRNA) as a more suitable scaffold. shRNAs are single‐stranded transcripts containing complementary stem regions connected by a conserved loop sequence recognized by Dicer. Upon processing, Dicer removes the loop to generate a duplex that is incorporated into RISC, where the Argonaute (Ago) protein mediates target mRNA cleavage [61, 62, 63]. We hypothesized that PIC‐mediated acylation of shRNA would disrupt Dicer recognition and processing of the hairpin structure, thereby abolishing downstream gene silencing until reductive decaging restored its functionality (Figure 6A).

FIGURE 6.

Short hairpin RNA (shRNA) caging strategy for RNAi application. (A) Mechanism of shRNA‐mediated target gene silencing after reducing stimuli induced uncaging of PIC protection. (B) Schematic illustration of RNAi experimental design. SNU449 cells were transfected with scramble shRNA, unmodified (mock‐treated) or 50 mM PIC modified shPD‐L1 for 6 h and treated with blank treatment (1×PBS) or mixture of 4 mM diboron (A) and 1 mM bipyridine (B) in 1×PBS for 120 min, then further cultured for 40 h before lysed at 4°C with RIPA buffer (protease inhibitor added) and analyzed with Western blotting. (C) Fluorescent imaging and (D) band quantification of immunoblot for shPD‐L1 uncaging. Cells were transfected with scramble shRNA (NC), unmodified (PC) or 50 mM PIC acylated shPD‐L1 (Ac, SM) and treated with blank treatment (NC, PC, Ac) or reducing small molecule stimuli (SM). For (C), data are representatives of three replicate experiments showing similar results. For (D), bands were first normalized by beta‐actin level, and then normalized to respective PD‐L1 level of NC group. (E) Comparison of knockdown efficacy between original shRNA and cage‐released shRNA. Knockdown percentages are calculated based on relative intensity difference for each replicate, for example, $KD\%=\frac{Ac-SM}{Ac}\ast 100\%$ for released shRNA. Data are presented as the mean ± SEM of n = 3 independent experiments and p‐values: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, not significant.

Cancer remains one of the most lethal diseases worldwide, in part due to the ability of malignant cells to evade immune surveillance [64]. A central mechanism of immune evasion involves overexpression of ligands that engage T cell immune checkpoints, such as programmed death‐ligand 1 (PD‐L1) on tumor cells binding to programmed cell death protein 1 (PD‐1) on T cells, thereby suppressing antitumor immune responses [65, 66]. RNA‐based strategies for modulating immune checkpoint expression have emerged as a promising approach to restore immune activity [10, 67]. To evaluate the applicability of our chemically controllable RNA platform, we designed an shRNA targeting a sequence within the PD‐L1 open reading frame, incorporating a CUCGAG loop for Dicer recognition (Figure S16). Functional validation was performed in SNU449 cells, a grade II–III/IV hepatocellular carcinoma line with confirmed PD‐L1 expression [68], establishing a model system to assess the potential of reversible RNA acylation as a chemical caging strategy for precise, stimulus‐responsive regulation of gene expression in living cells.

Cell‐based experiments were conducted over a 48‐h timeline (Figure 6B, upper). shRNA targeting PD‐L1 (shPD‐L1) was modified with 50 mM PIC and purified to afford a quantitatively recovered caged construct with confirmed integrity by gel electrophoresis (Figure S17). SNU449 hepatocellular carcinoma cells were seeded overnight and transfected with 1 µg of PIC‐caged shPD‐L1. Scrambled shRNA served as a negative control, and unmodified shPD‐L1 as a positive control. Six hours post‐transfection, cells were treated with 4 mM A and 1 mM B for 120 min, washed, and further cultured in full medium for 40 h. Total cellular protein was then extracted and analyzed by Western blotting following separation on 12% SDS‐PAGE gels (Figure 6B, lower). Immunoblot analysis (Figure 6C) revealed a marked decrease in PD‐L1 expression upon treatment with the A/B reducing pair. Densitometric quantification (Figures 6D and S18) demonstrated significant knockdown in the decaged group relative to both the caged and scrambled controls. In contrast, minimal knockdown was observed in the presence of the caging modification. We further verified that treatment of SNU449 cells with small molecules A/B alone had no significant effect on PD‐L1 expression in the absence of shRNA transfection. (Figure S19). Comparison of relative knockdown efficiencies indicated no statistically significant difference between the decaged shRNA and the unmodified positive control across triplicates, confirming effective restoration of gene‐silencing activity (Figure 6E). These findings establish PIC‐mediated reversible acylation as a controllable RISC‐inducing strategy, enabling target mRNA cleavage only upon exposure to the THDB/BIPY reducing system and thereby providing a chemically gated on–off switch for gene silencing. Collectively, these results highlight the potential of this reversible acylation platform as a chemically controllable caging toolkit for nucleic acid–based therapeutic strategies, enabling temporal and stimulus‐specific regulation of RNA‐mediated gene silencing in living cells.

3. Conclusion

We report a chemically triggered, control–release strategy for RNA based on 2′‐OH modification with a nitro‐containing carbamate. Upon treatment with the THDB–BIPY reducing pair at low millimolar concentrations, the modification is efficiently and rapidly removed within minutes. A plausible reduction mechanism was included in the Supporting Information (Figure S20). Installation of this group sterically disrupts RNA folding and intermolecular interactions, thereby suppressing native function, while reductive cleavage restores structure and activity. This approach is broadly applicable across diverse RNA classes, including fluorogenic aptamers, single guide RNAs, short hairpin RNAs, and long mRNAs. In cellulo studies confirmed biocompatibility and selective functional recovery, while in vitro assays showed bioorthogonality, with minimal reactivity toward endogenous reductants or commonly used reducing agents. Collectively, this platform enables stimulus‐responsive, exogenously controlled regulation of RNA function, providing an orthogonal toolkit for reversible RNA modulation that complements existing strategies based on endogenous‐responsive protection, with potential applications in research and therapeutic development.

Author Contributions

Yiran Zhao: investigation, methodology, validation, data curation, writing – original draft, project administration, formal analysis. Senfeng Zhang: investigation, data curation, formal analysis, validation. Junsong Guo: resources. Tuan‐Khoa Kha: resources. Song Chen: resources. Chunyi Hu: resources, supervision, funding acquisition. Ru‐Yi Zhu: writing – review and editing, resources, supervision, conceptualization, funding acquisition, project administration.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Materials and reagents used, general procedures and methods of chemical synthesis, in vitro validation and cell experiments, supplementary Figures, NMR and mass spectrometry spectra (docx).Supporting File: anie73928‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article