Amplified Production of a DNA Decoy Catalyzed by Intracellular MicroRNA

Abstract

DNA decoys inhibit cellular transcription factors and are expected to be among the nucleic acid drugs used to downregulate the transcription process. However, spatially controlling the on/off efficacy of DNA decoys to avoid side effects on normal cells is challenging. To reduce undesired decoy function in normal cells, we adopted catalytic hairpin assembly (CHA) to produce a DNA duplex from a hairpin DNA pair in response to a specific microRNA (miRNA). We designed the DNA hairpin pairs to form a DNA decoy that binds to nuclear factor kappa B (NF‐κB), whose overexpression is related to many diseases, including cancer. The transformation of the DNA hairpin pair to the NF‐κB DNA decoy was catalyzed by miR‐21, which is expressed in various types of cancers. Intracellular CHA progression and the inhibitory effect against NF‐κB were observed only in miR‐21 overexpressing cancer cells. The intracellular miR‐21‐catalyzed production of the NF‐κB DNA decoy has the potential to reduce side effects on normal cells, thereby strengthening the therapeutic profile of the CHA‐decoy system. The ability to customize the combination of catalytic miRNA and target transcription factors would allow our technology to serve as a “personalized drug discovery system” for a variety of challenging diseases, including cancer.

We achieved the amplified intracellular production of a DNA decoy targeting NF‐κB using a hairpin DNA circuit known as catalytic hairpin assembly (CHA). This system is catalyzed by oncogenic miR‐21 to produce the NF‐κB decoy, thereby inhibiting NF‐κB with high cancer specificity. Our approach holds great potential for efficient cancer therapy and as a personalized drug discovery system based on miRNA expression profiles.

Introduction

DNA decoys are nucleic acid therapeutics that inhibit the transcription of target genes in cells. ^[1] They mimic promoter sequences that can bind certain transcription factors, thereby competitively inhibiting the access of the transcription factors to the native consensus sequences in the nuclei, leading to the downregulation of specific gene expression. DNA decoys have been developed as drug candidates for various diseases, including atherosclerosis, ^[2] atopic dermatitis, ^[3] and cancer. ^[4] The transcription factor nuclear factor kappa B (NF‐κB) is a key regulator of immune‐responsive and apoptotic genes, playing a significant role in cellular processes, such as proliferation, survival, and apoptosis. ^[5] DNA decoys targeting NF‐κB hold therapeutic promise because of their effectiveness against various diseases, including cancer, rheumatoid arthritis, cardiovascular diseases, atopic dermatitis, and glomerulonephritis. However, despite the recent attention garnered by nucleic acid drug modalities, such as small interfering RNAs (siRNAs), antisense oligonucleotides, and mRNAs, ^[6] no study has demonstrated the clinical use of DNA decoy therapeutics. This is due to the efficacy of DNA decoys being insufficient and administering sufficient doses within the therapeutic window, because of on‐target side effects caused by poor selectivity for the cells to be treated, being difficult. Methods to address these problems include aptamer‐mediated delivery of DNA decoys ^[7] and spatial control of DNA decoys using exogenous light stimuli, which involves photocaged nucleosides whose caging group is removed and efficacy is restored by UV irradiation. ^[8] However, these methods require target receptors expressed on the target cells or exogenous light stimulation, making them challenging to use in vivo. Therefore, an effective approach is to use endogenous markers that exist in a wide range of diseases and ensure cell‐specificity.

Several molecules can be used as disease‐specific biomarkers, such as hydrogen peroxide, ^[9] oxygen, ^[10] and metal ions. ^[11] MicroRNAs (miRNAs) are one of these promising markers. MiRNAs are short noncoding RNAs that repress expression from corresponding mRNAs by hybridizing with their 3′ untranslated regions. ^[12] Many miRNAs have been identified as contributors to various diseases and disease‐related processes, including the malignant transformation of cancer, ^[13] autoimmune diseases, ^[14] and neurological diseases. ^[15] Because miRNAs are capable of Watson click‐base pair communication, numerous nucleic acid‐related techniques have been reported for their detection.

Catalytic hairpin assembly (CHA) has been used to detect miRNAs in living cells with high selectivity (Figure 1A). Pierce and co‐workers introduced this DNA nanotechnology as an enzyme‐free signal amplification method for the target oligonucleotides. ^[16] In addition to miRNAs[ ¹⁷ , ¹⁸ ] other endogenous biomolecules, such as mRNAs,[ ¹⁹ , ²⁰ ] metal ions,[ ²¹ , ²² ] proteins[ ²³ , ²⁴ ] and exosomes[ ²⁵ , ²⁶ ] have been detected or imaged using CHA, taking advantage of the property that CHA isothermally proceeds at the physiological temperature and ion concentrations, thereby providing a favorable signal‐to‐background ratio. The traditional CHA reaction requires the following three components: a catalytic strand and two kinds of hairpin DNAs. Consequently, two kinds of hairpins (HP1 and HP2) are transformed into one double‐stranded DNA via the CHA reaction. The catalytic strand opens HP1, and then the exposed sequence opens HP2. Subsequently, the exposed sequence ejects the catalytic strand from the complex, and the HP1–HP2 duplex is produced as the output product. Various types of CHA‐related circuits have been developed to broaden the applicability of this technique.[ ²⁷ , ²⁸ ] Recently, our group developed a CHA system in which a therapeutic small molecule is released in the presence of an oncogenic miRNA, which is overexpressed in cancer cells. ^[29] However, no study has yet demonstrated the direct conversion of CHA products into nucleic acid therapeutic molecules.

Figure 1.

Amplified production of NF‐κB decoy catalyzed by intracellular miR‐21. (A) The mechanism of CHA. [1] The catalytic strand opens HP1, [2] the exposed sequence opens HP2, [3] the exposed sequence displaces the catalytic strand, and [4] the HP1‐HP2 duplex is produced as the output product. (B) Amplified production of NF‐κB decoy via CHA catalyzed by intracellular miR‐21.

Here, we report a new strategy that employs CHA for the in situ production of a DNA decoy selectively targeting NF‐κB in cancer cells. The decoy activity can be temporally deactivated by incorporating each strand into separate DNA hairpins to prevent NF‐κB binding. Transfected DNA hairpins are converted into complete DNA decoys targeting NF‐κB via the CHA reaction catalyzed by endogenous miR‐21. Previous reports indicated that miR‐21 is overexpressed in various cancer cell lines, such as lung cancer A549, tens to thousands of times more than their expression levels in the normal embryonic kidney cell line, HEK‐293T. ^[30] Notably, the two DNA hairpins do not react with each other in the absence of miR‐21; hence, the production of NF‐κB DNA decoy occurs only in miR‐21‐positive cancer cells (Figure 1B). This approach enables spatial control of the drug efficacy of the DNA decoy. By altering the sequences of binding sites for transcription factor and initiator miRNA, various diseases can be treated without causing adverse effects on surrounding normal cells. Our approach is promising not only as a “DNA nanotechnology therapeutics” technology for disease‐selective therapy but also as a versatile methodology for a “personalized drug discovery system.”

Results and Discussion

Design of Hairpin DNA Pair

The CHA hairpin pair, which turns into an NF‐κB decoy in a reaction catalyzed by miR‐21, was designed using NUPACK, an online software for predicting the thermal stability of nucleic acid structures (Figure S1). ^[31] The free energy of the hairpin secondary structure is critical to the leakage level of the reaction, which is the reaction's unexpected occurrence without the initiator strand. Following the tendency that the stability of the hairpin DNA is affected by stem length, candidates with different stem lengths were designed and synthesized (Figure 2A). Thirty pairs, consisting of six variants of HP1 (12 to 17 bp stem length) and five variants of HP2 (13 to 17 bp stem length), were tested (all sequences of hairpin DNAs are listed in Table S1). The synthesized hairpin pairs were incubated with or without 0.1 eq. of miR‐21 at 37 °C for 1 h and analyzed using native polyacrylamide gel electrophoresis (PAGE; Figure 2B). As expected, the shorter stem showed a higher background reaction (the product was produced without miR‐21), and the longer stem resulted in low reactivity. Among the 30 pairs, we selected the pair of HP(14)‐1/HP(14)‐2, which showed high reaction efficiency (almost all the hairpin pair was consumed) and high selectivity (almost no product was produced in the absence of miR‐21), as the promising candidate for a cancer‐selective NF‐κB decoy producer (the red rectangle in Figure 2B).

Figure 2.

Design and optimization of hairpin DNA pairs. (A) Components of HP(n)‐1 and HP(n)‐2. (B) PAGE analysis of miR‐21 (0.1 eq.)‐catalyzed CHA using the 30 hairpin DNA pairs. The red rectangle indicates the selected pair.

Considering therapeutic use, the toehold regions of both HP(14)‐1 and HP(14)‐2 were phosphorothioated to provide durability to the hairpin DNA against degradation by DNases present in the serum, cells, and human body. The modified hairpins were renamed HP1 and HP2, respectively (Figure 3A). The CHA reaction with HP1 and HP2 was analyzed in detail (Figure 3B). Approximately 89 % of the hairpins reacted in the presence of a catalytic amount of miR‐21, while only 9 % of the hairpins reacted in the absence of miR‐21, even after 6 h. This indicates that the phosphorothioate modification did not negatively affect the efficacy or selectivity of the CHA reaction. To evaluate the suitability for use in cells, we investigated whether longer reaction times would increase the leaky reaction (Figure S2). After 12 h, only a very small amount of the hairpins reacted in the absence of miR‐21, similar to the 6 h reaction time. These results indicate that the designed HP1 and HP2 are promising CHA probes for applications not only in test tubes but also in cells. The off‐target effect on other miRNAs was evaluated using native PAGE (miRNA sequences are listed in Table S1). The selected RNAs were pre‐miR‐21, miR‐7531, miR‐122, and a modified miR‐21 with one mismatch (1MM). Pre‐miR‐21 is a 72‐nucleotide RNA encompassing the mature miR‐21 sequence. miR‐7851 is a human miRNA with a sequence that is highly homologous to miR‐21, as per the online database miRbase. miR‐122 is an abundant miRNA in the human liver, where administered nucleic acids tend to accumulate, influenced by modifications such as phosphorothioate. ^[32] HP1 and HP2 specifically reacted with miR‐21 compared with other miRNAs after 6 h, suggesting that the off‐target effect from the presence of other miRNAs had a minimal effect in human cells (Figure 3C).

Figure 3.

Evaluation of the hairpin DNA pair in test tubes. (A) The optimized sequence of the hairpin DNA pair (5′→3′). ^=phosphorothioate linkages. (B) PAGE analysis of miR‐21‐catalyzed CHA. (C) PAGE analysis of off‐target reaction. (D) Quantification of the formation of CHA product/NF‐κB complex. *An excess amount (200 eq.) of NF‐κB decoy was added.

The CHA product, composed of HP1 and HP2 modified with FAM in the loop region, was incubated with miR‐21 and recombinant NF‐κB p50 protein, and the mixture was analyzed using native PAGE (Figure 3D). Consumption of the CHA product was increased with an increase in the amount of NF‐κB p50, suggesting that the duplex formed from HP1 and HP2 contained the consensus sequence of NF‐κB. Furthermore, when adding an excess amount (200 eq.) of a binding competitor (a dumbbell‐type NF‐κB decoy, which is a circular DNA previously demonstrated to bind NF‐κB[ ³³ , ³⁴ ]), the consumption of the CHA product decreased (52 % vs. 31 %), indicating the NF‐κB binding ability of our system.

Intercellular CHA Reaction Catalyzed by Endogenous MiR‐21

We assessed the nuclease resistance of the hairpin DNA before the cell experiments. HP2 modified with FAM was incubated with Dulbecco's Modified Eagle Medium containing 10 % fetal bovine serum, and the degradation was monitored using native PAGE (Figure S3). Intact HP2 remained for a few hours, but most degraded after 1 day. Considering the high reactivity of the hairpin pair, phosphorothioate modification contributed to the stable progression of the CHA reaction. Unreacted hairpins would degrade appropriately, thereby helping to avoid unexpected side effects that could be caused by residual sequences.

To evaluate the CHA efficacy in living cells, a fluorescent resonance energy transfer (FRET) experiment was conducted (Figure 4A). HP2 modified with FAM and HP1 modified with the quencher Dabcyl were prepared. FAM and Dabcyl were placed respectively on the sequences in close proximity to ensure sufficient interaction when the two hairpins were hybridized via the CHA reaction. Due to the CHA progression catalyzed by intracellular miR‐21, the fluorescence intensity of FAM was expected to decrease by FRET. The pair of HP1 and HP2‐FAM were also tested. They were expected to hybridize via the CHA reaction, but the absence of Dabcyl was not expected to induce the quenching of FAM fluorescence by FRET. In the subsequent experiments, the pair of HP1‐Dabcyl and HP2‐FAM is referred to as Pair‐A, and the pair of HP1 and HP2‐FAM is referred to as Pair‐B (Figure 4B).

Figure 4.

miR‐21‐catalyzed CHA in human living cells. (A) Intercellular CHA reaction catalyzed by endogenous miR‐21. (B) Hairpin DNA pairs used in FRET study. Fluorescence‐activated cell sorting histograms of (C) A549, (D) HeLa, and (E) HEK‐293T cells (Pair‐A: red lines; Pair‐B: blue lines). (F) Mean fluorescence intensity in each cell line. Data from three independent experiments were averaged. Error bars represent standard deviation. *P<0.05 and **P<0.01 using unpaired Student's t‐test.

Pair‐A or Pair‐B was transfected into the following three types of human cells: lung cancer A549, cervical cancer HeLa, and embryonic kidney HEK‐293T cells, and FRET efficiency was analyzed using flowcytometry (Figures 4C–E). In A549 and HeLa cells, the fluorescence intensity of Pair‐A decreased rather than Pair‐B, suggesting that the CHA occurred in the cells. However, the fluorescent intensity was similar between Pair‐A and Pair‐B in HEK‐293T cells, indicating almost no reactivity of the hairpin DNA pair. The expression level of miR‐21 is reported to be in the following order: A549>HeLa>HEK‐293T, suggesting that the hairpin DNA pair could identify the different intracellular miR‐21 levels to catalyze CHA. We also quantified the amount of miR‐21 in each cell line and confirmed the following order of decreasing miR‐21 expression levels is consistent with previous studies (Figure S4). HeLa and A549 cells exhibited a significantly higher relative mean fluorescence intensity than HEK‐293T cells, demonstrating the promising cancer cell recognition capability of the hairpin DNA pair (Figure 4F). We further investigated an alternative transfection method, in which HP1 and HP2 were pre‐mixed prior to transfection into HeLa cells, followed by FACS analysis (Figure S5). The intracellular FRET efficiency was comparable to that observed when HP1 and HP2 were transfected separately.

To confirm the miR‐21‐catalyzed mechanism of CHA, anti‐miR‐21 was cotransfected with Pair‐A and Pair‐B into A549 cells. The fluorescent intensity was higher in cells transfected with anti‐miR‐21 compared with those transfected with anti‐miR‐122, which is a negative control for miR‐21 (Figure S6). Anti‐miR‐21 competitively prevented endogenous miR‐21 from hybridizing with the toehold region of HP1, thereby hindering CHA progression in A549 cells. This observation suggests that CHA occurred in a miR‐21‐dependent manner in living cells. Furthermore, we investigated the intracellular localization of the DNA hairpin. HP2‐FAM was transfected into A549 cells, and confocal imaging was conducted (Figure S7). The transfected hairpin DNA mainly localized in the cytosol, consistent with previous observations. ^[35]

Cancer‐Selective Inhibition of NF‐κB

The hairpin DNA pair exhibited cancer‐selective inhibition of NF‐κB activity in living cells. For comparison, we prepared SC1 and SC2, which have the same length of the toehold, stem, and loop as HP1 and HP2, respectively, but contain scrambled NF‐κB decoy sequences. In the following experiments, the HP1 and HP2 pair is referred to as NFκB‐Pair, and the SC1 and SC2 pair as SCRM‐Pair. We confirmed that SCRM‐Pair showed similar miR‐21‐catalyzed CHA efficiency as NFκB‐Pair (Figure S8). The activity of NF‐κB in living cells was evaluated using a secreted alkaline phosphatase (SEAP) reporter assay. SEAP expression is activated by NF‐κB and downregulated by the NF‐κB decoy. NFκB‐Pair decreased the SEAP signal in both A549 and HeLa cells compared with SCRM‐Pair (Figure 5AB). The NF‐κB inhibitory efficiency was approximately 40 % for both cell lines when 300 ng of NFκB‐Pair was transfected, a level similar to that observed with the NF‐κB DNA decoy (Figure S9). This finding indicates that NFκB‐Pair efficiently produced NF‐κB decoy in miR‐21‐abundant cancer cells. Conversely, almost no NF‐κB inhibitory effect of NFκB‐Pair was observed for miR‐21‐deficient HEK‐293T cells (Figure 5C), suggesting that the CHA and in situ production of the NF‐κB binding motif occurred to a negligible level. On comparing the amount of SEAP expressed from each cell line treated with NFκB‐Pair, both A549 and HeLa cells expressed significantly lower SEAP than HEK‐293T cells (Figure 5D). To further demonstrate the correlation between miR‐21 expression levels and intracellular NF‐κB DNA decoy production, additional SEAP reporter assays were performed using human breast cancer cell lines MDA‐MB‐231 and MCF7,[ ³⁶ , ³⁷ ] which are miR‐21‐abundant, and the human normal breast cell line MCF10 A, ^[38] which is miR‐21‐deficient (Figure S10). NFκB‐Pair exhibited an inhibitory effect on SEAP expression in a miR‐21‐dependent manner, further supporting the mechanism of miR‐21‐catalyzed NF‐κB DNA decoy production. Cumulatively, these findings underscore the potential of our hairpin DNA pair as promising decoy therapeutics, capable of distinguishing between cancer and normal cells.

Figure 5.

Cancer‐selective inhibition of NF‐κB in living cells. SEAP reporter assay for (A) A549, (B) HeLa, and (C) HEK‐293T cells. (D) A comparison of the effect of NFκB‐Pair between cell lines. Data from three independent experiments were averaged. Error bars represent standard deviation. ns, not significant, *P<0.05, and ****P<0.0001 using unpaired Student's t‐test.

We investigated how the inhibitory effect of NFκB‐Pair responds to varying levels of miR‐21. To downregulate miR‐21, antimiR‐21 was co‐transfected with NFκB‐Pair into A549 and HeLa cells, which resulted in a suppressed inhibitory effect only in A549 cells (Figure S11). The much higher endogenous miR‐21 expression in A549 cells was more responsive to antimiR‐21 compared to HeLa cells. Conversely, upregulation of miR‐21 levels through miR‐21 transfection affected the NFκB‐Pair’s inhibitory effect only in HeLa cell (Figure S12). This suggests that in A549 cells, the sufficient catalytic amounts of endogenous miR‐21 rendered the addition of exogenous miR‐21 insignificant for NF‐κB decoy generation. In contrast, in HeLa cells, the added miR‐21 likely enhanced CHA decoy production due to the relatively low baseline miR‐21 expression.

To investigate the mechanism of action of NFκB‐Pair, we performed an immunostaining assay for p65, which is a component protein of NF‐κB. The confocal images of A549 cells transfected with NFκB‐Pair showed that the nuclear localization of p65 was inhibited compared with that in nontreated cells (Figure 6A). Findings from the quantitative analysis support the hypothesis that NFκB‐Pair inhibits the nuclear localization of p65 through the CHA reaction catalyzed by miR‐21 in the cytosol (Figure 6B). The CHA decoy, modified with FAM, was mainly observed in the cytosol, suggesting that the localization of p65 was significantly inhibited by the CHA decoy (Figure 6C). Because miRNAs are mainly present in the cytoplasm, the produced DNA decoy would trap p65 in the cytoplasm and inhibit its transfer into the nucleus. The complex formed between DNA decoys and transcription factors has been reported to induce the formation of liquid‐like droplets in cells. ^[39] When NFκB‐Pair was transfected into A549 cells, many droplets with a diameter of approximately 1–3 μm were observed in the cytoplasm (Figure S13), suggesting that the complex formed between the CHA product and NF‐κB caused liquid–liquid phase separation at the sites where miR‐21 was abundant. Conversely, when the NF‐κB decoy was transfected into A549 cells, the droplets were observed in both the cytosol and the nucleus regardless of miR‐21 distribution (Figure S14).

Figure 6.

(A) Confocal images of A549 cells. Cyan fluorescence indicates p65. (B) The ratio of fluorescence intensity from p65 in the nucleus to that in the cytosol. Data from 16 independent experiments were averaged. Error bars represent standard deviation. NT, nontreated. **P<0.01 using unpaired Student's t‐test. (C) Confocal images of A549 cells. Green fluorescence indicates the CHA decoy.

Cancer‐Selective Cytotoxicity

NFκB‐Pair showed therapeutic effects on cancer cells. Inhibition of NF‐κB function induces cell death; ^[40] therefore, we performed cell viability assays on three human cell lines with varying miR‐21 expression levels (Figure 7). NFκB‐Pair exhibited concentration‐dependent cytotoxicity in A549 and HeLa cells but showed minimal cytotoxicity in miR‐21‐deficient HEK293T cells. These results indicate that the cytotoxicity of NFκB‐Pair is miR‐21‐abundant cancer cell‐specific, which is consistent with the SEAP reporter assay results for each cell line. We confirmed that the expression levels of NF‐κB were similar among A549, HeLa, and HEK293T cells, supporting the mechanism by which the low cytotoxicity in normal cells is attributed to the low reaction efficiency of miR‐21‐catalyzed CHA (Figure S15). The cell viability assays suggest that the intracellular miRNA‐catalyzed DNA decoy production system holds significant potential for efficient and selective cancer therapy tailored to miRNA expression profiles.

Figure 7.

Cancer cell‐selective cytotoxicity of NFκB‐Pair. Cell viability of A549, HeLa, and HEK‐293T cells. Data from three independent experiments were averaged. Error bars represent standard deviation. *P<0.05 and **P<0.01 using unpaired Student's t‐test.

Thus far, many studies have focused on elucidating the safety of nucleic acid therapeutics through chemical modifications, such as the incorporation of pseudouridine into mRNA vaccines ^[41] and various unnatural nucleobases into antisense oligonucleotides. ^[42] However, chemical modifications on nucleic acid molecules often affect their drug efficacy and pharmacokinetics, posing challenges in drug design. ^[43] Furthermore, depending on the modification, synthetic difficulty and cost could also be potential problems. Meanwhile, our strategy employs a “synthetic prodrug” design, enabling the production of active DNA decoys only in marker miRNA‐abundant cells. This approach requires only phosphorothioate modification, which has been extensively studied and used in most approved nucleic acid drugs. In addition, our system can be expanded to siRNA prodrugs by producing dsRNA via miRNA‐catalyzed CHA from hairpin RNA pairs.

Selective activation of DNA decoys in living cells has been achieved using light irradiation as an external trigger. Light improves the spatiotemporal control of molecules; however, its poor biological transparency inhibits its application in therapeutic fields. In this study, we achieved conditional control of the DNA decoy using miRNA as an internal trigger. This would render its clinical application profitable because no artificial interference is required for disease‐selective treatment after drug injection. Moreover, many different miRNAs have been identified in human‐diseased cells. Thus, the hairpin DNA assembly to induce the inhibition of a target transcription factor in response to a target miRNA has implications for the broad treatment of various types of diseases.

Although the inhibition efficiency of NF‐κB was similar between NFκB‐Pair and the NF‐κB decoy, the intracellular location where each nucleic acid agent blocked NF‐κB activity was different. Confocal images indicated that the droplets containing the CHA decoy did not enter the nucleus, whereas the droplets containing the NF‐κB decoy were observed in both the nucleus and cytosol. This difference is likely due to the localization of mature miR‐21, which primarily exists in the cytosol. NFκB‐Pair did not have NF‐κB inhibition ability (OFF‐state) before encountering miR‐21, indicating that the ON‐state decoy was mainly localized in the cytosol. By contrast, the function of the NF‐κB decoy was always in an ON state and promptly spread throughout the cells after transfection, blocking NF‐κB activity in both the cytosol and nucleus. The difference in the distribution of the droplets containing the CHA decoy and the NF‐κB decoy supports the strategy of amplified production of the DNA decoy catalyzed by intracellular miRNA.

Many DNA nanotechnologies have been developed to date, with previous studies focusing on constructing more complicated and flexible structures.[ ⁴⁴ , ⁴⁵ ] Some of these technologies contribute to therapy by detecting cancer‐related biomolecules, serving as drug‐delivery systems, or releasing small molecules based on three‐dimensional structures. For instance, detecting cancer‐specific miRNAs exemplifies DNA nanotechnology's role in cancer diagnosis, using sensitive probes in conjunction with various dynamically assembling DNA probes.[ ⁴⁶ , ⁴⁷ , ⁴⁸ ] However, few reports explored the application of DNA nanotechnology products as the primary drug molecules for treating diseases.[ ³⁵ , ⁴⁹ ] The flexibility of combining target transcription factors with catalytic miRNA for CHA enhances the potential of our strategy.

Our hairpin DNA technology could serve as a “personalized drug discovery system” for various refractory diseases that cannot be treated with traditional small molecule and antibody drugs. The combination of catalytic miRNAs and target transcription factors can be freely customized because the hairpin DNA pair is chemically synthesized using the designed sequence. The specific transcription pathway is blocked by introducing the personalized DNA hairpin pair into the heterogeneous mixture of various cells, such as organoids or immune cells interacting with tumor cells. Moreover, not only decoy DNAs but also siRNAs can be intracellularly constructed by switching the hairpin DNA pair to a hairpin RNA pair, designed to cleave a specific mRNA via the CHA product. The versatility of our system could enable the regulation of both transcription and translation steps in the central dogma of molecular biology, leading to the development of novel drugs to meet unmet medical needs.

Conclusion

We have established an amplified production system of a DNA decoy from two hairpin DNAs catalyzed using a specific miRNA through CHA, which is a DNA self‐assembly technology. The designed DNA hairpin pair is converted into a DNA decoy with high efficiency and selectivity targeting the transcription factor NF‐κB, whose improper regulation is related to various diseases. The DNA pair traps and inhibits the transcription factor NF‐κB in miR‐21‐abundant cancer cells but not in miR‐21‐deficient normal cells. Our approach is the first to allow the CHA product to be directly used as a therapeutic agent. The high customizability of the combination of catalytic miRNA and target transcription factor would enable our technology to serve as a “personalized drug discovery system” for various refractory diseases, including cancer.

Conflict of Interests

A.O. is a cofounder of TKG Therapeutics, Inc., and the company focuses on the development of nucleic acid therapeutics. An international patent application covering part of this work has been filed by The University of Tokyo (inventors K.M. and A.O.; publication number WO/2023/013329). The remaining authors declare no competing interests.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Yasuda S., Morihiro K., Koga S., Okamoto A., Angew. Chem. Int. Ed. 2025, 64, e202424421. 10.1002/anie.202424421

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.